Citation
A Weighty matter

Material Information

Title:
A Weighty matter effects of adiposity on adult neurocognitive health
Creator:
Brannon, Sara P. H. ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 online resource (288 pages). : ;

Thesis/Dissertation Information

Degree:
Doctor of Philosophy
Degree Divisions:
Department of Health and Behavioral Sciences, CU Denver
Degree Disciplines:
Health and Behavioral Sciences
Committee Chair:
Rooks, Ronica
Committee Members:
Coussons-Read, Mary
Albeck, David
Donahoo, William

Subjects

Subjects / Keywords:
Cognition ( lcsh )
Obesity ( lcsh )
Cognition ( fast )
Obesity ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
There is a growing body of evidence suggesting there are modifiable vascular and metabolic risk factors for Alzheimer's disease (AD). Despite many plausible mechanisms by which obesity could contribute, its etiological contributions remains unclear. We therefore investigated 1) the evidence linking obesity to cognitive health, 2) whether obesity in early or mid-life is associated with cognitive change and the possible factors involved, and 3) whether dietary interventions that reduce weight improve adult cognitive function. Study 1 involved a systematic literature review of the evidence linking adiposity to adult cognitive health outcomes. This revealed evidence that dementia was associated in older adults with low weight and separately with weight loss. However being overweight or obese in midlife was associated with cognitive decline and increased risk of dementia in later life. Study 2 examined the cross-sectional association between adiposity and cognitive function in a nationally representative sample of 4515 men and women, aged 20-59 years, who completed cognitive testing as part of the Third National Health and Nutrition Examination (NHANES-III). Global obesity and central obesity both predicted a small proportion of the variance on the Serial Digit Learning Task (SDLT) and the Simple Reaction Time Task (SRTT), but not the Symbol Digit Substitution Test (SDST). Frequent physical activity (PA) modified the association of SDLT and central obesity to the point that subjects who were obese or overweight but physically active showed cognitive performance similar to that of persons of normal weight. Study 3 involved a systematic literature review of the effects of weight loss interventions on adult cognitive function. The existing evidence gives mixed results, but the majority of studies reported beneficial effects of weight loss on cognitive function, including memory. Study 4 compared the effects of 8 weeks of intermittent fasting (IF) with the effects of standard dietary restriction for weight loss, on the cognitive function and health of 26 obese adults. IF consisted of completely fasting one day and eating ad libitum the next. Although no effects on cognition were apparent at 8 weeks, at 6 months post-intervention, the IF group showed improved memory, BDNF and greater loss of trunk fat. Reduced trunk fat was associated with improved memory scores. Summary. Results are consistent with the hypotheses that obesity is associated with cognitive deficits that can be ameliorated by weight loss and/or IF. However further research is needed.
Thesis:
Thesis (Ph.D.)--University of Colorado Denver. Health and behavioral sciences
Bibliography:
Includes bibliographic references.
System Details:
System requirements: Adobe Reader.
General Note:
Department of Health and Behavioral Sciences
Statement of Responsibility:
by Sarah P. A. Brannon.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
868699889 ( OCLC )
ocn868699889

Downloads

This item has the following downloads:


Full Text
A WEIGHTY MATTER:
EFFECTS OF ADIPOSITY ON ADULT NEUROCOGNITIVE HEALTH
by
SARAH P.A. BRANNON
B.Psych (Hons), University of Newcastle, 2004
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado Denver
In partial fulfillment of the requirements for the degree of
Doctor of Philosophy
Health and Behavioral Sciences
2013


This thesis for the Doctor of Philosophy degree by
Sarah P. A. Brannon
has been approved for the
Health and Behavioral Sciences Program
by
Ronica Rooks, Chair
Mary Coussons-Read, Advisor
David Albeck
William Donahoo


Brannon, Sarah P. A. (Ph.D., Health and Behavioral Sciences)
A Weighty Matter: Effects of Adiposity on Adult Neurocognitive Health
Thesis directed by Professor Mary Coussons-Read
ABSTRACT
There is a growing body of evidence suggesting there are modifiable vascular and
metabolic risk factors for Alzheimer's disease (AD). Despite many plausible mechanisms by
which obesity could contribute, its etiological contributions remains unclear. We therefore
investigated 1) the evidence linking obesity to cognitive health, 2) whether obesity in early or
mid-life is associated with cognitive change and the possible factors involved, and 3) whether
dietary interventions that reduce weight improve adult cognitive function. Study 1 involved a
systematic literature review of the evidence linking adiposity to adult cognitive health
outcomes. This revealed evidence that dementia was associated in older adults with low weight
and separately with weight loss. However being overweight or obese in midlife was associated
with cognitive decline and increased risk of dementia in later life. Study 2 examined the cross-
sectional association between adiposity and cognitive function in a nationally representative
sample of 4515 men and women, aged 20-59 years, who completed cognitive testing as part of
the Third National Health and Nutrition Examination (NHANES-III). Global obesity and central
obesity both predicted a small proportion of the variance on the Serial Digit Learning Task (SDLT)
and the Simple Reaction Time Task (SRTT), but not the Symbol Digit Substitution Test (SDST).
Frequent physical activity (PA) modified the association of SDLT and central obesity to the point
that subjects who were obese or overweight but physically active showed cognitive
performance similar to that of persons of normal weight. Study 3 involved a systematic
literature review of the effects of weight loss interventions on adult cognitive function. The
existing evidence gives mixed results, but the majority of studies reported beneficial effects of


weight loss on cognitive function, including memory. Study 4 compared the effects of 8 weeks
of intermittent fasting (IF) with the effects of standard dietary restriction for weight loss, on the
cognitive function and health of 26 obese adults. IF consisted of completely fasting one day and
eating ad libitum the next. Although no effects on cognition were apparent at 8 weeks, at 6
months post-intervention, the IF group showed improved memory, BDNF and greater loss of
trunk fat. Reduced trunk fat was associated with improved memory scores. Summary. Results
are consistent with the hypotheses that obesity is associated with cognitive deficits that can be
ameliorated by weight loss and/or IF. Flowever further research is needed.
The form and content of this abstract are approved. I recommend its publication.
Approved: Mary Coussons-Read
IV


ACKNOWLEDGEMENTS
I would like to express my deepest gratitude to my advisor, Dr. Mary Coussons-Read for
her tireless support in the pursuit of this work. It has been an honor to work together, and I look
forward to continuing to do so in the future.
I am grateful to William Donahoo, M.D. for all of his help in developing this work and for
including me in this innovating and exciting project, the DRIFT study, on which this dissertation
is based. It would not have been possible without the DRIFT study team, including William
Donahoo, M.D., Edward Melanson Ph.D. and Dr. Wendolyn Gozansky, M.D., for including me
In addition, I would like to thank the team at the Department of Health and Behavioral
Sciences for the education, training and opportunities to learn, as well as their moral and
financial support. In particular I would like to thank Abby Fitch for all of her administrative
support.
Many thanks to the Colorado Clinical and Translational Science Institute (CCTSI) for the
excellent pre-doctoral training program in translational research they provided, complete with
opportunities for clinical experience, professional development, and financial support. In
particular I would like to thank Dr. Celia Sladek and Emily Warren for their key roles in making
this program a success.
I would also like to express my appreciation for the collaboration and assistance
provided by Dr. Mark Mattson and Dr. Bronwen Martin of the National Institute on Aging. Not
only did they analyze the Brain-Derived Neurotrophic Factor that forms an important outcome
measure for this work, but their insights on the field, and excitement about this work were
invaluable.
My thanks also go to the Lionel Murphy Foundation of Australia for their significant
financial support of this dissertation.
v


This work was partially supported by NIH grants K23 AG026784 (WSG) and R21
AT0002617 (WTD). The investigators retained full independence in the conduct of this research.
The project described was also supported by the National Center for Research
Resources, Grant TL1 RR025778, and is now at the National Center for Advancing Translational
Sciences, Grant TL1 RR025778. The content is solely the responsibility of the authors and does
not necessarily represent the official views of the NIH.
VI


TABLE OF CONTENTS
Tables.....................................................................................xii
Figures....................................................................................xiv
Abbreviations...............................................................................xv
Chapter
1. Adiposity and Dementia..................................................................1
1.1 Introduction to Cognitive Decline and Dementia.......................................1
1.1.1 Pre-clinical Stages of Alzheimer's Disease...........................................3
1.2 A Lifecourse Approach to Alzheimer's Disease.........................................6
1.2.1 The Lifecourse Approach to Chronic Disease...........................................7
1.2.2 Applying the Lifecourse Approach to Cognitive Aging..................................8
1.2.3 Evidence for Modifiable Risk Factors in Midlife.....................................10
1.3 Study 1. Weight and Cognitive Health: A Systematic Literature Review................12
1.3.1 Introduction........................................................................12
1.3.2 Methods.............................................................................16
1.3.3 Results.............................................................................18
1.3.4 Conclusions.........................................................................44
1.4 Potential Confounding Factors.......................................................55
1.4.1 Education...........................................................................56
1.4.2 Sleep Apnea.........................................................................57
1.4.3 Depression..........................................................................58
1.4.4 Poor Nutrition......................................................................59
1.4.5 Physical Inactivity.................................................................60
1.4.6 Reverse Causation...................................................................61
vii


1.5 Potential Mechanisms Linking Adiposity to Dementia Risk...........................62
1.5.1 Alzheimer's Disease Pathophysiology: A Brief Overview............................62
1.5.2 Insulin Resistance and Glucose Regulation.........................................62
1.5.3 Hypertension......................................................................63
1.5.4 Dyslipidemia......................................................................65
1.5.5 Oxidative Stress..................................................................66
1.5.6 Leptin and Leptin Resistance......................................................67
1.5.7 Insulin-Like Growth Factor-1......................................................68
1.5.8 Inflammation......................................................................70
1.5.9 Cortisol and HPA Axis Dysregulation...............................................71
1.5.10 Brain-Derived Neurotrophic Factor.................................................73
1.6 Summary...........................................................................76
2. Study 2: the NHANES-III Study.......................................................76
2.1 Introduction......................................................................76
2.2 Methods...........................................................................81
2.2.1 Study Design......................................................................81
2.2.2 Participants......................................................................82
2.2.3 Tests and Materials...............................................................82
2.2.4 Data Analysis.....................................................................87
2.3 Results...........................................................................92
2.3.1 Sample Characteristics............................................................92
2.3.2 Regression Model-building process.................................................95
2.3.3 Regression Models of the Simple Reaction Time Test................................96
viii


2.3.4 Regression Models of the Symbol Digit Substitution Test............................101
2.3.5 Regression Models of the Serial Digit Learning Task................................103
2.3.6 Sub-analysis of the Duration of Obesity............................................109
2.4 Discussion.........................................................................113
3. Interventions........................................................................119
3.1 Introduction.......................................................................119
3.2 Physical Activity Interventions....................................................119
3.2.1 Animal Studies of Physical Activity................................................120
3.2.2 Human Studies of Physical Activity.................................................121
3.2.3 Potential Mechanisms for an Effect of Physical Activity............................125
3.3 Study 3: Calorie Restriction Interventions.........................................126
3.3.1 Introduction.......................................................................126
3.3.2 Methods............................................................................128
3.3.3 Results............................................................................130
3.3.4 Discussion.........................................................................138
3.4 Dietary Composition and Frequency..................................................139
3.4.1 Quality of Diet....................................................................139
3.4.2 Ketogenic Diets....................................................................139
3.4.3 Fasting............................................................................141
3.5 Intermittent Fasting...............................................................142
IX


4. Study 4: the DRIFT Study
152
4.1 Introduction.......................................................................152
4.2 Methods............................................................................153
4.2.1 Participants.......................................................................153
4.2.2 Intervention Protocols.............................................................155
4.2.3 Materials and Tests................................................................156
4.2.4 Procedure..........................................................................162
4.2.5 Data Analysis......................................................................168
4.3 Results............................................................................170
4.3.1 Study Attrition....................................................................170
4.3.2 Sample Characteristics.............................................................171
4.3.3 Safety.............................................................................173
4.3.4 Dietary Adherence..................................................................173
4.3.5 Acute Effects of Fasting...........................................................174
4.3.6 Effects of IF on Body Weight and Composition.......................................176
4.3.7 Effects of IF on Cognitive Function................................................182
4.3.8 Effects of IF on Glucose Regulation................................................183
4.3.9 Effects of IF on BDNF..............................................................184
4.3.10 Effects of IF on FIPA Axis Function................................................185
4.3.11 Effects of IF on Pro-Inflammatory Cytokines........................................187
4.3.12 Effects of IF on Leptin............................................................187
4.4 Discussion.........................................................................190
x


5. Discussion..........................................................................197
5.1 A Lifecourse Approach to Cognitive Aging.........................................197
5.2 Central Obesity May be More Strongly Linked to Neurocognitive Health than Global
Obesity..........................................................................199
5.3 Behavior May Moderate the Association Between Obesity and Cognition..............200
5.4 Interventions that Affect Obesity May Reduce Risk of Cognitive Decline...........201
5.5 IF Produces Beneficial Effects on Memory in Obese Adults.........................202
5.6 Plausible Mediating Mechanisms Exist.............................................203
5.7 Strengths and Limitations........................................................205
5.7.1 Strengths and Limitations of Studies 1 and 3: Systematic Reviews.................205
5.7.2 Strengths and Limitations of Study 2: The NHANES-III Study.......................206
5.7.3 Strengths and Limitations of Study 4: The DRIFT Study............................207
5.8 Novel Contributions..............................................................209
5.9 Future Directions................................................................210
5.10 Summary..........................................................................211
Appendix.................................................................................212
References...............................................................................227
XI


LIST OF TABLES
Table
1.1.1. Cognitive function and cognitive impairments defined.............................6
1.2.1. Some potential exposure-outcome relations suggested by the lifecourse approach...8
1.3.1. Summary of results for the obesity and cognitive function literature review......19
1.3.2. Longitudinal studies supporting an association between overweight and dementia...21
1.3.3. Longitudinal studies not supporting an association between overweight and dementia. 27
1.3.4. Longitudnial studies not supporting an association between overweight and MCI..........33
1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
.......................................................................................36
1.3.6. Longitudinal studies not supporting an association between overweight and cognitive
decline...............................................................................41
1.3.7. Longitudinal studies measuring central obesity that supported an association with any
cognitive outcomes....................................................................46
2.1.1. Specific aims and hypotheses of Study 2: NHANES.......................................80
2.2.1. Pre-existing categorical variables already created in NHANES dataset..................87
2.2.2. New categorical variables created for this analysis...................................90
2.3.1. NHANES-III sample characteristics by BMI category.....................................94
2.3.2. Correlation between anthropometric measures...........................................95
2.3.3. Multiple Linear Regression models of the association between measures of adiposity and
SRTT..................................................................................98
2.3.4. Multiple linear regression model of the association between central obesity (WHR) and
reaction time (SRTT).................................................................100
XII


2.3.5. Multiple linear regression models of the association between measures of adiposity and
measures of cognitive function. SDLT (total score) or SDST (natural log of seconds). Each
model was run separately.............................................................103
2.3.6. Multiple linear regression model of the association between central obesity and SDLT
(total score)........................................................................108
2.3.7. SDLT scores by physical activity and obesity (mean (SE))..............................109
3.3.1. Intervention studies with beneficial effects on cognition............................132
3.3.2. Intervention studies with no effects, or adverse effects, on cognition...............136
3.5.1. Effects of IF and CR compared........................................................143
4.2.1. CNS vital signs cognitive test domains and how they are calculated...................158
4.3.1. DRIFT sample characteristics at baseline after a fed day. (mean (sd) unless otherwise
indicated)...........................................................................172
4.3.2. Effects of an acute 36h fast at baseline on all participants before randomization (n=26).
Measures were collected at 7am the following day.....................................175
4.3.3. Eight week post-intervention group differences in weight and cognitive function......180
4.3.4. Eight week post-intervention group differences in biomarkers.........................181
4.3.5. Between group differences 6 months after the end of the interventions................189
XIII


LIST OF FIGURES
Figure
1.1.1. A gradual process of pre-clinical cognitive decline typically precedes Alzheimer's disease.5
1.2.1. As people age, the cumulative effects of damaging and protective exposures may lead to
widening differences in cognitive function............................................9
1.3.1. Steps of the systematic review process for longitudinal studies.......................18
2.3.1. Interaction between central obesity and physical activity for SDLT performance.......107
2.3.2. Number of participants in each strata of duration variable...........................Ill
3.3.1. Steps of the systematic review process for longitudinal studies......................130
3.5.1. An inverse U-shaped dose response curve..............................................148
3.5.1. Repeated episodes of stress in the mild-moderate "beneficial" range may have
cumulative beneficial effects......................................................149
4.2.1. Overview of the DRIFT study design..................................................156
4.2.2. Baseline fed and baseline fasted visits compared....................................163
4.2.3. Outline of procedures on a baseline study visit.....................................165
4.3.1. Eight week change in body weight by group (kg)......................................176
4.3.2. Eight week percent weight change for each participant...............................177
4.3.3. Mean change in percent fat mass by time and group...................................178
4.3.4. Mean differences in percent trunk fat by time and group.............................178
4.3.5. Mean memory scores by time and intervention group...................................182
4.3.6. Mean insulin sensitivity (SI) by time and group.....................................184
4.3.7. Mean BDNF by time and group (pg/mL).................................................185
4.3.8. Mean leptin by time and group (ng/mL)...............................................188
XIV


LIST OF ABBREVIATIONS
AD Alzheimer's Disease
ADF Alternate-Day Fasting
BBB Blood-Brain Barrier
BDNF Brain-Derived Neurotrophic Factor
BMI Body Mass Index
CR Calorie Restriction
CRP C-Reactive Protein
DR Dietary Restriction
GC Glucocorticoid
H El Healthy Eating Index
HPA Hypothalamic-Pituitary Adrenal
KD Ketogenic Diet
IGF-1 Insulin-like Growth Factor 1
IF Intermittent Fasting
IL-6 Interleukin 6
MCI Mild Cognitive Impairment
MetS Metabolic Syndrome
NHANES National Health and Nutrition Examination Survey
NIH National Institute of Health
PA Physical Activity
T2DM Type 2 Diabetes Mellitus
TNF-a Tumor Necrosis Factor a
WC Waist Circumference
WHR Waist-Hip Ratio
xv


1. Adiposity and Dementia
1.1 Introduction to Cognitive Decline and Dementia
A growing body of epidemiological evidence suggests that obesity may increase risk of
dementia, including Alzheimer's disease, the leading cause of dementia worldwide (Alzheimer's
Association, 2012). In a population that is increasingly obese and rapidly aging, a causal link
between obesity and AD would have significant public health implications. A clear
understanding of the relationship of obesity to dementia risk is therefore important, but further
research is needed before this can be achieved. In particular, there is a need to determine the
extent to which timing, duration and extent of obesity affect risk, and the mechanisms by which
such risks may be altered.
Alzheimer's disease (AD) is the 6th leading cause of death in the United States today
(Alzheimer's Association, 2012). It is the leading cause of dementia, accounting for 60-80% of
cases (Alzheimer's Association, 2009). At present, dementia afflicts over 5.4 million Americans,
however with the aging population prevalence is likely to increase to 6.7 million by 2025 and
13.5 million by 2050 if nothing is done to prevent it (Alzheimer's Association, 2012; Herbert,
Beckett, Scherr, & Evans, 2001). Globally, the direct and indirect costs of dementia were
estimated to total US$ 604 billion in 2010, or 1% of the aggregated worldwide GDP (World
Health Organization, 2012). In the same year, costs of dementia in the United Kingdom (£23
billion) almost matched those of cancer (£12 billion), heart disease (£8 billion) and stroke (£5
billion) combined (World Health Organization, 2012). In the United States, Medicare, Medicaid
and out-of-pocket payments for dementia health care, long-term care, and hospice care were
estimated to be $200B, while caregivers, primarily family members, provided an estimated
17.4B hours of unpaid care valued at $210B.(Alzheimer's Association, 2012). Furthermore, the
physical and emotional impact of dementia care giving was estimated to result in $8.7B in
1


increased healthcare costs in the United States in 2011 (Alzheimer's Association, 2012). These
costs do not account for the devastating emotional and personal costs of dementia for
caregivers and patients.
Dementia refers to a group of conditions characterized by a decline in memory and at
least one other cognitive function that are severe enough to impair activities of daily living
(Alzheimer's Association, 2009). Affected cognitive functions include a wide variety of processes
necessary for thinking, planning and action, including functions such as memory, attention, or
executive function. Dementia can be caused by a number of different conditions, of which
dementia of the Alzheimer's type is just one. Vascular dementia (VaD), is another type of
dementia, caused by cerebrovascular incidents, which accounts for 5-15% of all cases of
dementia in the United States (Alzheimer's Association, 2012). Other types of dementia, such as
frontotemporal dementia (FTD) and dementia with lewy bodies, make up the remainder of
dementia cases.
Alzheimer's disease is unique among the 10 leading causes of death in that there are
currently no effective treatments for the underlying pathology, and no options for prevention.
Since age is a primary risk factor, the development of primary prevention strategies to delay the
onset of AD by 5 years could result in an estimated 57% reduction in the number of persons
with AD in the United States, and reduce the projected Medicare costs of AD from $627 to $344
billion (Sperling et al., 2011). However there is a clear need to better understand the factors that
contribute to the disease, particularly early risk factors and pathophysiology, before successful
interventions can be developed. This has been highlighted by the lack of efficacy of clinical trials
directed at treating AD pathophysiology in persons already clinically diagnosed with dementia.
The lack of success may reflect the difficulty of reversing the neurological damage already done
2


by this late stage of the disease, rather than the appropriateness of the target. Intervening
earlier in the disease process may have better success.
Hence we need to better understand the disease process at pre-clinical stages of the
disease (Sperling et al., 2011). At present formal diagnosis of AD can be made on autopsy, and is
confirmed by the presence of hallmark pathology neuritic (B-amyloid plaques and tau tangles in
the brain (Budson & Solomon, 2011). However although a correlation between these
pathological features and cognitive function is observed (Riley, Snowdon, Desrosiers, &
Markesbery, 2005), some persons who show these pathological markers never develop clinical
symptoms of dementia in their lifetime (Riley et al., 2005), indicating that factors other than the
presence of plaques and tangles themselves may be involved. By contrast clinical diagnosis of
AD is currently based on severity of behavioral and cognitive symptoms. No biomarkers have yet
been found to be reliable markers of the disease, so clinical diagnosis occurs once cognitive
impairments have become so severe that they interfere with activities of daily living (McKhann
et al., 2011).
1.1.1 Pre-clinical Stages of Alzheimer's Disease
The search for points of early intervention may be assisted by the growing recognition
that pre-clinical signs of the disease are apparent years, or even decades, before clinical
diagnosis (Jack et al., 2011; Sperling et al., 2011). These findings have been made possible by a
combination of factors. Advances in neuroimaging have made structural, neurochemical and
functional alterations in the brains of asymptomatic persons apparent, and evidence of
potential biomarkers of pathology in pre-clinical persons also continues to build (Jack et al.,
2011; Sperling et al., 2011). Many longitudinal studies have also contributed to the growing
awareness of subtle cognitive alterations many years before clinical diagnosis (Sperling et al.,
2011). Together, these advances, and the lack of success in treating symptomatic AD, have
3


contributed to an awareness of the need to look for modifiable pre-clinical processes during
mid-adulthood that could contribute to modification of disease pathology. Epidemiological
evidence, discussed in detail in section 1.3, suggests that midlife obesity may be one such pre-
clinical risk factor contributing to disease pathology.
What precedes clinically diagnosable dementia? Evidence of a gradual progression of
cognitive decline is now apparent, though the rate of progression varies considerably between
individuals (Sperling et al., 2011).
Alzheimer's disease is preceded by the clinically diagnosable state of Mild Cognitive
Impairment (MCI), which is now increasingly viewed as a prodromal stage of dementia (Albert et
al., 2011) (Budson & Solomon, 2011; Sperling et al., 2011). MCI can be diagnosed when declines
in memory or other cognitive functions are clinically detectable, but not severe enough to
interfere with activities of daily living (Albert et al., 2011). Many older adults do not seek
medical attention at this stage, but it is estimated that 10-20% of adults older than 65 years
have MCI (Alzheimer's Association, 2009). Prognosis for those diagnosed with MCI varies. A
small minority of individuals regain normal cognitive function, and some others maintain their
mild impairment (Bennett et al., 2002). However MCI has a high risk of progression to dementia,
particularly dementia of the Alzheimer's type. It is estimated that in the community the
conversion rate from MCI to dementia is about 5-10% per year (Etgen, Bickel, & Forstl, 2010)
Rates may be higher among clinical populations presenting with memory complaints.
Prior to diagnosable MCI, subtle cognitive declines may also become gradually apparent
in pre-clinical populations (Jack et al., 2011; Sperling et al., 2011). Cognitive decline involves
decreased function from prior levels in one or more cognitive domains. With cognitive function
defined as the capacity for abilities such as attention, memory, perception, self-volition,
language and judgment (Anderson & McConnell, 2007). A small amount of cognitive decline is a
4


normal part of aging (Budson & Solomon, 2011), and subjective complaints about memory and
other cognitive functions increase with advancing age (Newson & Kemps, 2006), however for
some the extent of decline across midlife and into old age is greater than would be expected in
normal cognitive aging.
MECHANISMS?
Figure 1.1.1. A gradual process of pre-clinical cognitive decline typically precedes
Alzheimer's disease.
5


Table 1.1.1. Cognitive function and cognitive impairments defined.
Cognitive function The capacity for thinking, planning and acting, including functions such as attention, memory, self-volition, language and judgment.
Neurocognitive health Healthy brain and cognitive function.
Cognitive impairment A generic term referring to any impairment in cognitive function/s.
Mild Cognitive Impairment (MCI) A clinical diagnosis involving impairment in one or more cognitive functions greater than would be expected for the person's age and education (Albert et al., 2011).
Dementia A group of conditions characterized by significant declines in memory and/or other cognitive functions that are severe enough to affect activities of daily living (McKhann et al., 2011).
Alzheimer's disease (AD) Dementia of the Alzheimer's Type (DAT) is the development of memory impairment and at least one other cognitive disturbance, that each cause significant impairment and represent a decline from prior functioning (APA, 2000).
Vascular dementia (VaD) The development of memory impairment and at least one other cognitive disturbance, with evidence of cerebrovascular disease (APA, 2000)
1.2 A Lifecourse Approach to Alzheimer's Disease
The emerging evidence of a long trajectory for AD pathology (Sperling et al., 2011) and
cognitive decline suggests that this may be a disease developed over a lifetime, rather than a
disease of old age (Gustafson, 2008). Hence a lifecourse approach to AD could be useful. (Kuh &
Ben-Shlomo, 2004). As already noted, significant cognitive decline with advancing age is not a
normal part of aging, and most older adults retain excellent neurocognitive function late in their
lives (Alzheimer's Association, 2012). However age is the leading risk factor for AD. Most cases
of dementia are diagnosed after the age of 65 (Alzheimer's Association, 2012; Budson &
Solomon, 2011). This raises an important question: what differentiates people who experience
healthy cognitive aging from those who experience significant cognitive decline and eventual
dementia? Genetic research indicates that for most people genes contribute to only a small
6


proportion of the difference (Alzheimer's Association, 2012). This leaves a significant role for
social and environmental exposures across the lifespan.
1.2.1 The Lifecourse Approach to Chronic Disease
The lifecourse approach (Kuh & Ben-Shlomo, 2004) is widely used to understand the
etiology of chronic diseases. As described by (Kuh & Ben-Shlomo, 2004), exposures throughout
the lifespan can influence both the incidence of chronic disease and its course. Sensitivity to risk
factors may vary across the lifespan. This can lead to different patterns of risk-outcome
relationships, such as those summarized in Table 1.2.1 below (Glymour & Manly, 2008; Power &
Hertzman, 1997; Wadsworth, 1997). These models of exposure are not mutually exclusive, but
can interact to shape health outcomes, with effects that could become increasingly apparent
with advancing age.
7


Table 1.2.1. Some potential exposure-outcome relations suggested by the lifecourse approach. Adapted from (Glymour & Manly, 2008; Power & Hertzman, 1997; Wadsworth, 1997)
Immediate risk model Short period between exposure and health outcome. Return to baseline health after removal of risk factor. E.g. Delirium in hospitalized older adults can be reversed by medication management (Gray, Lai, & larson, 1999).
Cumulative model Each exposure leads to some harm. The cumulative effect of exposures increases disease risk. Removal of exposure does not reverse harm already done. E.g. Effects of lead exposure over time impairs older adults' cognitive function (Weisskopf et al., 2004).
Latency model Exposure during a critical period of development increases risk of disease much later in life, but health effects may not be immediately apparent. E.g. Poverty in early life may provide an early exposure to stress that increases vulnerability to cognitive decline later in life (Lupien, King, Meaney, & McEwen, 2000)
Social trajectory model Exposure sets in motion a succession of adverse social events that increase vulnerability later in life. E.g. Poverty early in life may reduce educational access, which reduces education attainment and opportunities for mentally stimulating occupations, which may increase risk of cognitive decline and dementia later in life (Al Hazzouri, Haan, Whitmer, Yaffe, & Neuhaus, 2012).
1.2.2 Applying the Lifecourse Approach to Cognitive Aging
With its emphasis on understanding the timing of exposures, the lifecourse approach
could provide a useful framework to investigate factors differentiating healthy cognitive aging
from cognitive decline or dementia. Cognitive aging has not been a prominent focus for
lifecourse epidemiology (Glymour & Manly, 2008). However the lifecourse approach suggests
that differences in cognitive function between individuals of the same age may reflect a range of
8


detrimental or neuroprotective biological, psychological and social exposures experienced in
early and mid-adulthood (Stein & Moritz, 1999), and as a person ages, the cumulative effects of
these exposures could bring widening differences in cognitive function, as depicted in Figure
1.2.1 below.
Figure 1.2.1. As people age, the cumulative effects of damaging and protective exposures may
lead to widening differences in cognitive function.
As indicated in the four exposure-outcome models summarized in Table 1.2.1,
investigation of the effects of risk factors for AD will require careful attention to the timing and
duration of exposure. An exposure may be detrimental during sensitive periods of development,
with effects that do not become apparent for many years, or which only appear when combined
with another later exposure. For example it remains possible that exposure to obesity in late life
alone does not confer additional risk for dementia, but exposure to maternal obesity while in
utero, or exposure to childhood obesity during critical periods of neural development, increase
risk of dementia much later in life (latency model). It is also possible that prolonged duration of
9


obesity during early and mid-adulthood could contribute cumulative damage to the brain over
many years (cumulative model), and that the duration of exposure acquired by older adults who
become obese in their old age is insufficient to confer significant risk. It is beyond the scope of
this dissertation to address the effects of obesity across the entire lifespan. Instead, this paper
will focus on investigating factors consistent with a cumulative model. It will therefore focus on
factors that could have cumulative effects on the neural and cognitive health of adults in early
and mid-adulthood. This time period could correspond to the pre-clinical stages of AD in
affected persons and so reflect early-stage risk factors for later cognitive decline with advancing
age.
1.2.3 Evidence for Modifiable Risk Factors in Midlife
A number of factors have recently emerged that could potentially act as midlife risk
factors for AD later in life. These include Type 2 diabetes mellitus, the metabolic syndrome and
hypertension. Obesity is causally related to each of these conditions, and so has the potential to
the underlying pathological process.
Evidence indicates that Type 2 diabetes mellitus (T2DM), a chronic disease characterized
by insulin resistance and glucose dysregulation, increases the risk of MCI and dementia (Biessels
& Gispen, 2005; Whitmer, Karter, Yaffe, Quesenberry, & Selby, 2009; Yaffe et al., 2004a). It is
well-known that people with diabetes have increased risk of micro-vascular complications (e.g.
neuropathy, retinopathy, nephropathy, and macrovascular events (myocardial infarction,
stroke), and are therefore at increased risk for vascular dementia. However there also seems to
be an independent risk of developing AD. Type 2 diabetes mellitus can also have adverse effects
on general cognitive function earlier in life, in the absence of clinically diagnosed MCI or AD
(Cukierman, Gerstein, & Williamson, 2005; Elias, Elias, Sullivan, Wolf, & D'Agostino, 2005). Both
men and women with T2DM show deficits in attention, processing speed, memory, and
10


executive functioning (Biessels, ter Braak, Erkelens, & Hijman, 2001; Biessels & Gispen, 2005;
Manschot et al., 2007; Messier, 2005; Ostrosky-Solis, Mendoza, & Ardila, 2001; Yaffe et al.,
2004b). The association has been found in cross-sectional (Messier, 2005; Ryan, 2005; Strachan,
Deary, Ewing, & Frier, 1997) and prospective follow-up studies (Cukierman et al., 2005; Elias et
al., 2005; Messier, 2005). In addition, intervention studies using insulin-sensitizing agents have
had some success in reducing cognitive decline (Craft et al., 2003; Craft et al., 2012). Though
these findings are not universal, systematic reviews of prospective studies to date (Cukierman et
al., 2005), and a growing professional consensus, indicates that the consistency and strength of
the evidence is sufficient to warrant including cognitive decline as one of the potential
complications of diabetes (Plassman, Williams, Burke, Holsinger, & Benjamin, 2010).
Several studies also report that the Metabolic Syndrome (MetS) independently
increases risk of cognitive decline and incident dementia (Yaffe, 2007; Yaffe et al., 2004c).
According to the criteria of the International Diabetes Federation (IDF, 2006), the MetS is
defined as having central obesity and any two of the following: raised triglycerides (>
150 mg/dL), reduced HDL cholesterol (< 40 mg/dL), raised blood pressure (systolic BP > 130 or
diastolic BP >85 mm Fig), or raised fasting glucose (>100 mg/dL), or treatment for previous
diagnosis of any of those conditions. Since the combination of components can vary it is possible
that different components confer independent risk of cognitive decline. Consistent with this,
non-diabetic individuals with insulin resistance also show some evidence of deficits in learning
and memory (Vanhanen et al., 1997; Vanhanen et al., 2006), even after controlling for vascular
factors (Convit, Wolf, Tarshish, & de Leon, 2003).
An emerging body of evidence now suggests an association between obesity and
increased risk of cognitive decline and dementia (Cournot et al., 2006a; Gustafson, 2006;
Whitmer, 2007; Whitmer, Gunderson, Quesenberry, Zhou, & Yaffe, 2007). Obesity is a risk factor
11


forT2DM and all components of the MetS (Abbasi, Brown, Lamendola, McLaughlin, & Reaven
2002; Boyko et al., 2000; Zimmet, Boyko, Collier, & Courten, 1999)), and may play a causal role
in many of the clinical features of these conditions. Obesity already contributes to over 300,000
deaths per year in the United States alone (Boeka & Lokken, 2008), and rates of obesity have
risen dramatically across the population over the past 20 years. The majority of the population
who have experienced these weight gains have yet to reach the ages at which AD is most likely
to manifest. Hence if obesity does contribute even a small increase in risk of dementia, the
implications at a population level could be particularly significant for an aging and increasingly
obese population. Therefore the goal of this dissertation is to investigate whether obesity
affects adult neurocognitive health, as well as the potential that behavioral interventions
directed at reducing obesity or mitigating its effects could improve adult neurocognitive health.
Towards these ends the author and colleagues conducted 3 studies. Study 1 involved a
systematic review of the published empirical evidence linking weight or adiposity to adult
cognitive function. Study 2 investigated the association between adiposity and cognitive
function in the general population. Study 3 involved a systematic review of the published
empirical data on interventions for weight loss, and their effects on human cognitive function.
Finally, study 4 investigated the effects of weight loss diets on the cognitive function of obese
adults.
1.3 Study 1. Weight and Cognitive Health: A Systematic Literature Review
1.3.1 Introduction
Alzheimer's disease is the leading cause of dementia and the 6th leading cause of death
in the United States today (Alzheimer's Association, 2012). It afflicts over 5.4 million Americans
today, and prevalence is likely to increase to 6.7 million by 2025 (Herbert et al., 2001). There is
12


currently no effective treatment, but evidence for modifiable risk factors is emerging. Among
these, several recent studies have linked obesity to increased risk of cognitive decline and
dementia (Cournot et al., 2006a; Gustafson et al., 2009; Gustafson, Lissner, Bengtsson,
Bjorkelund, & Skoog, 2004; Whitmer, Gunderson, Barrett-Connor, Quesenberry, & Yaffe, 2005).
If obesity contributes even a small increase in risk of dementia, the effects across the
increasingly obese populations of Western nations could be significant. The purpose of this
review is therefore to systematically investigate the evidence from prospective studies that
obesity increases risk of cognitive decline or dementia.
Obesity is a significant excess of body fat, or adipose tissue, often defined as a Body
Mass Index (BMI, kg/m2) greater than 30. The BMI is a simple but useful measure that allows a
general estimation of adiposity; however it cannot accurately reflect the proportion of adipose
tissue carried by an individual. It is likely that adiposity, rather than weight, would mediate any
effect of obesity on neurocognitive health, for adipose tissue is not just a storage depot for fat,
but is also endocrinologically and immunologically active. Adipocytes secrete various active
metabolites that could cross the blood-brain barrier (BBB) to affect brain health. Central adipose
tissue, distributed around the trunk and including subcutaneous and omental adipose reserves,
is known to be particularly active in this regard, secreting numerous adipokines such as leptin,
and cytokines (Bastard et al., 2000; Bastard et al., 2006; Fain, 2006; Ingvartsen & Boisclair,
2001; Wellen & Hotamisligil, 2003). In addition, obesity is related to many other vascular and
metabolic factors implicated in AD pathology, including insulin resistance and hypertension.
It is important to discover the age/s at which obesity might have its greatest impact on
neurocognitive health since, according to the lifecourse approach (Kuh & Ben-Shlomo, 2004),
the age at which an exposure is encountered can have a significant effect on outcome. There
may be some evidence for modifiable exposures early in life contributing to late life dementia
13


risk. For example, some studies have investigated the association between childhood obesity
and adult cognitive function. (Lupien et al., 2000; Miller et al., 2009). It is beyond the scope of
this paper to address all the evidence across the lifespan. We will therefore focus this review on
the risk of adult obesity for risk of cognitive decline and dementia in late life, where the
exposure is most proximal to potential "pre-clinical" stages of dementia (Sperling et al., 2011).
Among adults, it remains important to account for the age at which obesity was
measured, the measure of adiposity used, and the age of cognitive outcomes (Gustafson, 2008;
Whitmer et al., 2007). These are important for two main reasons: 1) because of the effects of
aging on body composition, and 2) because of the tendency for weight loss in clinical and
subclinical dementia.
Firstly, advancing age commonly brings significant changes in body composition, even if
overall weight does not change. Muscle mass often declines, and adipose tissue often increases
with increasing age (Miller & Wolfe, 2008). For this reason, BMI can be a poor estimate of
adiposity for older adults, as it cannot differentiate fat free mass from adipose tissue. Thus
studies measuring the association between BMI in older adults and dementia cannot rule out
the possibility that someone who has a healthy BMI is actually carrying a relatively large fat
mass, having lost significant bone density and muscle mass. Similarly, studies comparing BMI
across multiple ages cannot address the potential for loss of fat-free mass with advancing age.
This may be a significant confounding factor for studies that rely on BMI alone, particularly
among older adults.
Secondly, measurement of the association between weight and cognitive function
among older adults can be complicated by the tendency of persons with dementia to lose
significant amounts of weight as part of the disease (Wirth, Bauer, & Sieber, 2007), This
observation has been reported in many observational and clinical studies of dementia (Berlinger
14


& Potter, 1991; Gao et al., 2011; Johnson, Wilkins, & Morris, 2006; Renvall, Spindler, Nichols, &
Ramsdell, 1993). Weight loss may occur shortly before dementia diagnosis (Johnson et al.,
2006), so that on clinical presentation, the person with dementia is more likely to be
underweight than overweight (Gorospe & Dave, 2007). The weight loss that occurs in dementia
could possibly occur because of neural damage to appetite and self-regulation regions of the
brain, dysregulated circadian rhythm, or dysregulated behavior more generally. As a result,
cross-sectional studies of weight and dementia, or longitudinal studies among older adults with
short follow-up periods, could give the impression that persons who are overweight or obese in
old age have lower dementia risk (Dahl, Lopponen, Isoaho, Berg, & Kivela, 2008a), no matter
what measure of adiposity was used.
While a number of other reviews have indeed taken these factors into account (Anstey,
Cherbuin, Budge, & Young, 2011; Dahl & Hassing, 2012; Gorospe & Dave, 2007; Luchsinger &
Gustafson, 2009; Naderali, Ratcliffe, & Dale, 2009; Yen, 2005), this review differs from other
recent reviews in that it is a systematic review that attempts to provide a comprehensive
snapshot of all longitudinal studies on weight, adiposity and adult cognitive health outcomes
available through MEDLINE and PsychINFO. Other reviews have limited their searches to studies
of AD, only studies with specific follow-up periods (Beydoun, Beydoun, & Wang, 2008; Dahl &
Hassing, 2012), only studies that included at least 2 cognitive domains rather than diagnoses of
dementia or global screening instruments (van den Berg, Kloppenborg, Kessels, Kappelle, &
Biessels, 2009)), used either Medline or PubMed but not both (Beydoun et al., 2008; van den
Berg et al., 2009) or were not systematic reviews (Gustafson, 2006; Luchsinger, Patel, Tang,
Schupf, & Mayeux, 2008; Naderali et al., 2009; Whitmer, 2007).
15


1.3.2 Methods
Literature Search Terms and Study Inclusion Criteria
English-language articles focused on weight or adiposity and cognition or dementia
outcomes were identified using MEDLINE and PsycINFO. The following search terms were used
in various combinations: dementia, Alzheimer's disease, cognition disorders, cognitive decline,
cognitive impairment, cognition, cognitive function, and cognitive health, as well as obesity,
overweight, weight, fat, adiposity, central obesity, visceral obesity, visceral adiposity, waist-hip
ratio, and waist circumference. All relevant articles published up until January 30th, 2013, and
retrievable by university library search or interlibrary loan were considered for this systematic
review. Reference lists of all potentially eligible articles were reviewed to ensure inclusion of all
relevant literature.
To be included in this review, the articles had to meet the following eligibility criteria:
empirical articles that are available via university libraries or interlibrary loan, written in English,
and included the weight/adiposity and dementia/cognition related search terms above within
the title, abstract, and/or keywords. Studies of weight in childhood or adolescence were
considered eligible if they included adult cognitive outcomes. Studies of change in weight or
weight outcomes in a population that began with dementia at baseline were excluded. Similarly
studies focused on health outcomes for other cognitively impaired persons with a medical or
psychiatric diagnosis known to cause cognitive impairment, such as developmental disability,
schizophrenia, bipolar disorder or traumatic brain injury, were excluded. Studies with a specific
focus on eating disordered populations were also excluded, as were other studies focused on
specific medical or psychiatric populations. Dissertations, reviews, opinions, theoretical papers
or editorials were excluded from this review.
16


Article Selection and Abstraction
A three-step process guided assessment and selection of articles. First, the study author
reviewed the titles and abstracts of all potential articles retrieved by the search terms,
identifying the set of article abstracts that potentially matched the eligibility criteria. Second,
the study author reviewed in-depth the abstracts and full articles for studies whose abstracts
passed the first review for inclusion. Information from the full articles were entered into
summary tables. From this set the author identified the set of full articles which matched the
eligibility criteria below. Finally, reference lists of eligible articles were reviewed for additional
relevant articles to potentially include. These articles were also assessed for eligibility through a
two step process.
17


Figure 1.3.1. Steps of the systematic review process for longitudinal studies.
1.3.3 Results
Number of Articles Included in Review
A total of 4320 articles were identified using the search terms. Of these articles, 4085
were excluded after a preliminary review of title and/or abstract because they were not relevant
and/or did not meet the inclusion criteria (e.g. topic was relevant but the article was an
editorial). The remaining 235 full articles were then reviewed and abstracted by the study
author. Of these 235 articles, 64 observational studies were considered eligible after more
thorough review, and were therefore included in this study. Included articles were then
categorized according to study population (persons with dementia, cognitive impairment, or a
study of cognitive function more generally. Articles were further categorized by whether their
18


results supported or opposed a link between adiposity and cognitive health outcomes, and age
at which weight and cognitive outcomes were measured. A flow chart of the sorting and
inclusion process can be seen in Figure 1.3.1.
Table 1.3.1. Summary of results for the obesity and cognitive function literature review.
FOR AN ASSOCIATION AGAINST AN ASSOCIATION
Midlife Older adults Midlife Older adults
Dementia 61% (11) 39% (7) 20% (4) 80% (16)
MCI Na na 0% 100% (5)
Cognitive Function 71% (10) 29% (4) 31% (4) 69% (9)
Central obesity 55% (6) 45% (5) 17% (2) 83% (10)
Articles Assessing Dementia Risk
Among the 38 longitudinal studies in which dementia was the principle cognitive
outcome, 18 reported evidence supporting an association between obesity and increased risk
dementia. Of these, 15 reported specific results for the diagnosis of AD. In contrast, 20 studies
reported no association between obesity and dementia. Of these, 9 reported specific results for
a diagnosis of AD. The results are shown in Table 1.3.2 and Table 1.3.3.
Supporting an association between obesity and dementia: Among the 18 studies
finding a significant association between higher weight/adiposity and increased dementia risk,
11 (61%) reported an association between midlife weight measures and increased dementia risk
(Beydoun & Beason-Held, 2008; Chiang et al., 2007; Fitzpatrick et al., 2009; Gelber et al., 2012;
Gustafson et al., 2009; Flassing et al., 2009; Kivipelto et al., 2005; Rosengren, Skoog, Gustafson,
& Wilhelmsen, 2005; Whitmer et al., 2005; Whitmer et al., 2007; Whitmer et al., 2008), while 7
(39%) reported that increased weight later in life (> 65 years) increased risk of dementia
(Buchman et al., 2005; Gustafson, Rothenberg, Blennow, Steen, & Skoog, 2003; Flayden et al.,
2006a; Kerwin et al., 2011; Luchsinger, Cheng, Tang, Schupf, & Mayeux, 2012; Luchsinger, Patel,
Tang, Schupf, & Mayeux, 2007; Xu et al., 2011). Fifteen of these studies reported outcomes for
19


AD specifically. All 18 of these studies measured BMI. Only 7 reported BMI with other measures
of adiposity such as WHR, WC or percent fat mass.
Opposing an association between obesity and dementia: Twenty longitudinal studies
reported evidence that did not support an association between obesity and dementia. Of these
studies only 4 (20%) reported adiposity in midlife, while 16 (80%) assessed adiposity in older
adults. Only 9 of these studies addressed AD specifically (refs). Among these studies, 13
measured only BMI. Only 6 reported BMI with other measures of adiposity such as WHR, WC or
percent fat mass.
20


Table 1.3.2. Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF MIDLIFE ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Beydoun & Beason- Held, 2008) 2,322 >20 median 23.4 BMI, WC Dx AD, NINCDS-ADRDA criteria. Midlife (30, 40 or 45 years): - Men: being underweight (BMI 30, WC >80th percentile) increased AD risk (HR = 6.57, 95% Cl: 1.96, 22.02). Women who lost weight (BMI change <10th percentile) between ages 30 and 45 years were also at increased risk (HR = 2.02, 95% Cl: 1.06, 3.85). Weight gain among men (BMI change >90th percentile) between age 30 and 50 years increased AD risk (HR = 3.70, 95% Cl: 1.43, 9.56).
(Chiang et al., 2007) 157 demented cases 628 matched controls Age 30 and older (1982- 1992) 8-20 years Nested case- control study BMI (Chinese criteria) Dx AD, VaD, Chinese version of DSM-IV A J-shaped relationship was observed between BMI and dementia. Compared to BMI 20.5-22.9 odds (OR) for developing dementia was: 1.84 (1.02-3.33) for BMI <20.5, 1.87 (1.08-3.23) for BMI 23.0-25.4, and 2.44 (1.39-4.28) for BMI >/=25 Similar findings were observed for AD and VaD.
NJ


Table 1.3.2 . Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF MIDLIFE ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Fitzpatric k et al., 2009) 2798 74.7 (but baseline BMI based on self- report of weight at age 50) 5.4 mean (but much more when compared to self- reported BMI at age 50) BMI AD dx: NINCDS-ADRDA criteria; VaD dx: California Alzheimer's Disease and Treatment Centers criteria Midlife obesity increased risk of dementia compared to BMI 20-25 healthy weight group (HR 1.39; 95% Cl 1.03- 1.87). Reversed in assessments of late-life BMI: - Underweight persons (BMI 20) had an increased risk of dementia (HR 1.62; 1.02-2.64), - being overweight (BMI25-30) was not associated with risk of dementia (0.92; 0.72-1.18) - being obese reduced the risk of dementia (0.63; 0.44- 0.91).
(Gelber et al., 2012) 3468 52 25-28 BMI Dementia dx: DSM-lll-R criteria; AD dx: NINCDS- ADRDA criteria; VaD dx: California Alzheimer's Disease and Treatment Centers criteria Compared to BMI <22.6, being overweight or obese (BMI > 25.0) was associated with greater risk of dementia (OR = 1.87, 95% Cl = 1.26-2.77)
(Gustafso n et al., 2009) 1462 38-60 32 BMI, WC, WHR Dementia dx: DSM-lll-R criteria; AD dx: NINCDS- ADRDA criteria; VaD dx: NINDS-AIREN criteria Logistic models showed that a midlife WHR greater than 0.80 more than doubled dementia risk (OR 2.22, 95% Cl: 1.00-4.94, p=0.049). Cox models showed no association.
(Hassing et al., 2009) 1152 45-65 years (mean 52.5) up to 40 BMI AD dx: NINCDS-ADRDA criteria; VaD dx: NINDS- AIREN criteria Overweight in midlife had an elevated risk of - dementia (OR=1.59; 95% Cl: 1.21-2.07), - AD (OR=1.71; 95% Cl: 1.24-2.35), - VaD (OR=1.55; 95% Cl: 0.98-2.47).
NJ
NJ


Table 1.3.2 . Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF MIDLIFE ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Kivipelto et al., 2005) 1449 50.6 Mean 21 BMI Dx dementia, AD: DSM-IV criteria; AD dx: NINCDS-ADRDA criteria Midlife obesity was associated with late life dementia after adjusting for vascular factors (OR = 1.9, 95% Cl: 1.0- 4.6).
(Rosengre n et al., 2005) 7402 47-55 25-28 BMI Dx Dementia, AD: Death register and hospital discharge diagnoses J-shaped curve. BMI less than 20 in midlife was associated with increased risk of primary hospital diagnosis of dementia in late life. BMI of >22.5 in midlife was associated with increased risk of a primary hospital diagnosis of dementia.
(Whitmer et al., 2005) 10,276 40-45 21-39 BMI, skinfold thickness Dementia dx: ICD-9 codes Compared with healthy weight: - Obesity in midlife (BMI >= 30) increased risk of dementia 74% (HR 1.74, 95% Cl 1.34 to 2.26). - Overweight in midlife (BMI 25.0-29.9) increased risk of dementia 35% (1.35,1.14 to 1.60). The highest quintile of skinfold thickness had a 72% greater risk of dementia than the lowest quintile (1.72, 1.36 to 2.18, and 1.59, 1.24 to 2.04).
(Whitmer et al., 2007) 10,136 40-45 26-42 BMI Review of medical records from Neurology visits Dx VaD, AD Obesity in midlife increased risk of AD (adjHR=3.10, 95% Cl 2.19-4.38), and risk of VaD (adjHR=5.01, 95% Cl 2.98- 8.43) Overweight in midlife increased risk of AD (adj HR=2.09, 95% Cl 1.69-2.60) and VaD (HR=1.95, 95% Cl 1.29-2.96 for VaD).
NJ
UJ


Table 1.3.2 . Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF MIDLIFE ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Whitmer et al., 2008) 6583 40-45 26-42? BMI, Sagittal Abdominal Diameter (SAD) ICD-9 codes Persons in the highest quintile of SAD in midilfe had increased risk of dementia (HR, 2.72; 95% Cl, 2.33-3.33) compared to persons in the lowest quintile. Those with high SAD (>25 cm) but healthy BMI had an increased risk (HR, 1.89; 95% Cl, 0.98-3.81) vs. those with low SAD (<25 cm) and healthy BMI. Persons who were both obese and with high SAD had the highest risk of dementia (HR, 3.60; 95% Cl, 2.85- 4.55).
NJ
4^


Table 1.3.2 . Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Buchman et al., 2005) 820 As above up to 10 (mean 5.5) BMI ADdx: NINCDS-ADRDA criteria A 1-point annual decline in BMI was associated with increased risk of AD compared with persons with no change in BMI (HR 0.730; 95% Cl 0.625 to 0.852).
(Gustafso n et al., 2003) 392 70 18 BMI Dementia dx: DSM-lll-R criteria; AD dx: NINCDS- ADRDA criteria; VaD dx: NINDS-AIREN criteria. For every 1.0 increase in BMI at age 70 years, AD risk increased by 36% (Hazard Ratio, 95% Cl). These associations were not found in men.
(Hayden et al., 2006b) 3264 >65 3.2 BMI AD dx using NINCDS- ADRDA criteria; VaD dx using NINDS-AIREN criteria Obesity increased the risk of AD in females (adjHR 2.23, 95% Cl 1.09-4.30) but not males.
(Kerwin et al., 2011) 7163 65 to 80 up to 8 average 4.4 WHR, BMI 3MSE Dementia Dx Central obesity was what mattered. Women with a WHR of >0.80 and a BMI of 20.0 24.9 had a greater risk of cognitive impairment and probable dementia than more- obese women or women with a WHR less than 0.80. However women with a WHR less than 0.80 and a BMI of 20.0 to 24.9 kg/m2 had poorer scores on cognitive assessments.
(Luchsinge r et al., 2007) 893 BMI, 907 WC, 709 weight >65 5 BMI, WC, weight Dementia dx: DSM-IV criteria; AD dx: NINCDS- ADRDA criteria In persons <76 years the association between BMI and dementia resembled a U shape, meaning that both low BMI and high BMI increased risk of dementia. In those >76 years, dementia risk decreased with higher BMI. In persons <76 years, the highest quartile of WC correlated to dementia and AD risk. Weight loss was related to a higher risk of dementia and dementia associated with stroke DAS). Weight gain was related to a higher DAS risk only.
NJ
Ui


Table 1.3.2 . Longitudinal studies supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Luchsinge r et al., 2012) 1459 >65 7 BMI, WHR, WC AD dx NINCDS-ADRDA criteria Only higher WHR was related to higher AD risk, with the highest quartile of WHR increasing risk 2.5 times compared to the lowest quartile (HR = 2.5; 95% Cl = 1.3 4.7).
(Xu et al., 2011) 8,534 43.4 mean ~30 BMI MMSE, CERAD, Memory in Reality test Dx dementia, VaD, AD Higher midlife BMI was associated with an increased risk of dementia (OR 1.08, 95% Cl 1.03-1.14)
NJ


Table 1.3.3 . Longitudinal studies not supporting an association between overweight and dementia .
STUDIES OF MIDLIFE ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Chen et al., 2010) 286 20s & 40s ~50yr case- control study BMI MMSE dx AD: NINCDS-ADRDA criteria; Dx VaD: NINDS- AIREN criteria Men and women with low BMI in midlife had increased risk of AD (OR = 2.62-3.97, 95% Cl) and VaD (OR = 6.23- 11.11) compared with those with healthy BMI. High BMI in midlife was associated with increased VaD risk (OR = 15.29 and 10.32) among women.
(Rosengre n et al., 2005) 7402 47-55 25-28 BMI Dementia dx: Death register and hospital discharge diagnoses J-shaped curve. BMI less than 20 in midlife was associated with increased risk of primary hospital diagnosis of dementia in late life. BMI of >22.5 in midlife was associated with increased risk of a primary hospital diagnosis of dementia.
(Stewart et al., 2005) 1890 46-68 32 Weight Dementia dx: DSM-III-R15 Criteria; AD dx: NINCDS-ADRDA criteria; dxVaD: California Alzheimer Disease and Treatment Centers criteria Groups that developed and did not develop dementia did not differ with respect to baseline weight or change in weight from mid to late life.
(Strand et al., 2013) 48,793 35-50 31-35 BMI Dx dementia, AD: Norwegian Cause of Death Registry Low BMI (<20 vs. BMI 20-25) in midlife was associated with increased risk of dementia death (HR = 1.76, 95% Cl 1.15-2.68).
NJ


Table 1.3.3. Longitudinal studies not supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Abelian van Kan et al., 2012) 647 >75 7 Total fat mass (DXA) SPMSQ Total fat mass was not significantly associated with dementia risk.
(Atti et al., 2008) 646 >75 9 BMI MMSE; Dementia dx DSM-III criteria Persons with a BMI of >25.0 had a lower risk of dementia than those with a BMI of 20.0 to 24.9 (HR = 0.75, 95% (Cl) = 0.59-0.96),
(Barrett- Connor, Edelstein, Corey- Bloom, & Wiederholt, 1996) 299 50-79 16-21 Baseline 1972-74; Dementia diagnosis visit 1990- 93 Weight Dx AD neurological test scores, physical examination Men who developed AD weighed slightly more at baseline (1972-74) than men who remained cognitively intact ( P = .03); baseline weights did not differ in women by dementia status ( P = .54). Both men and women who were later diagnosed with AD had decreasing weight measured across the three time points ( P < .001 for men and P < .003 for women), whereas men and women who were diagnosed as cognitively intact had no significant change in their weights
(Dahl et al., 2008a; Dahl, Lopponen, Isoaho, Berg, & Kivela, 2008b) 605 65-92 8 BMI Dementia dx DSM-IV criteria. Women with high BMI scores had a lower dementia risk (HR = 0.90, 95% Cl = 0.84-0.96). Trend for men with high BMI scores to have a lower dementia risk, (HR = 0.95, 95% Cl = 0.84-1.07).
(Forti et al., 2010) 749 >65 3-5 Metabolic Syndrome (MetS) WC MMSE Dx VaD, AD In participants aged 75 and older, abdominal obesity was associated with a lower risk of overall dementia (HR = 0.53, 95% Cl = 0.28-0.98).
NJ
00


Table 1.3.3. Longitudinal studies not supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Gao et al., 2011) 1,331 >65 mean 6.4 BMI dx dementia using ICD-10 and DSM-IIIR Greater decline in BMI was associated with greater risk of dementia or MCI (p = .02 for dementia, p = .04 for MCI). BMI in participants with incident dementia, MCI, and normal cognition did not differ 9-12 years before diagnosis. Six years before diagnosis, participants with incident dementia or MCI had significantly lower BMI than participants with normal cognition.
(Han et al., 2009) 721 60-85 2 BMI, WHR, WC, PBF CERAD-Korean version Dementia dx using DSM-IV criteria The change in cognitive function in the elderly was associated with the baseline assessment of BMI, WC, and % body fat. Men who were obese (WHR, BMI) at baseline and subsequently increased weight had improved cognitive function. Women with high WHR at baseline and a subsequent decrease in adiposity had increased risk of cognitive decline. Women with normal WC at baseline and subsequent increased adiposity also had increased risk of cognitive decline.
(Hughes, Boren stein, Schofield, Wu, & Larson, 2009) 1,478 mean 71.8 mean 7.8 BMI dxVaD, AD, DSM-IV criteria Higher baseline BMI was significantly associated with a reduced risk of AD (adj HR] = 0.56, 95% [Cl] = 0.33-0.97). Slower rate of decline in BMI was associated with a reduced risk of dementia (HR = 0.37, 95% Cl = 0.14-0.98), with the association stronger for those who were overweight or obese (HR=0.18, 95% CI=0.05-0.58) compared to normal or underweight (HR=1.00, 95% Cl=0.18-5.66) at baseline.
NJ
<£>


Table 1.3.3. Longitudinal studies not supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Johnson et al., 2006) 449 65-95 Mean 6 Weight AD dx Clinical Dementia Rating Scale Among older adults, rate of weight loss doubled about 1 year before dementia diagnosis among individuals who developed AD.
(Knopman, Edland, Cha, Petersen, & Rocca, 2007) 481 At time of onset, 6.1% under age 70 years, 75.6% ages 70 to 89 years, and 18.2% age 90 years or older Case- Control ~30 years Weight DSM-IV criteria applied to medical records There were no differences in weight between cases and controls 21 to 30 years prior to the onset of dementia. However starting from 11 20 years prior to the index year, women with dementia had lower weight than controls, and the difference increased over time. There was a trend for increasing risk of dementia with decreasing weight in women 9 to 10 years before the index year (p = 0.001).
(Nourhashe mi et al., 2003) 3646 >65 8 BMI MMSE Dx Dementia, AD by DSM-lll-R criteria The risk of dementia was highest for those with a BMI <21 (RR=1.48, Cl=95%: 1.08-2.04) and those with a BMI of 21- 22 (RR-1.072, Cl=95%: 0.759-1.514) compared with BMI between 23 -26. However relationship disappeared after excluding people who developed dementia early in the study.
(Ogunniyi et al., 2011) 1559 >65 5.97 BMI CERAD, Clinician Home-based Interview, and Cambridge Examination for Mental Disorders A significantly greater decline in BMI was found in those with either incident dementia (p < 0.001) or incident MCI (p < 0.001) compared to healthy subjects.
(Power et al., 2011) 12,047 65-84 (mean 72.1) 9.7 BMI, WC, WHR ICD-9 & ICD-10 dx codes in Western Australia Data Linkage System Overweight men and those with WHR > 0.9 have lower risk of dementia than men with normal weight and with WHR < 0.9. Higher adiposity was not associated with incident dementia.
UJ
o


Table 1.3.3. Longitudinal studies not supporting an association between overweight and dementia.
STUDIES OF ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Raffaitin et al., 2009) 2049 74 4 WC MMSE, Benton Visual Retention Test, Isaac's Set Test Dx VaD, AD Participants with high WC had decreased risk for all-cause dementia (HR = 0.86, 95% Cl: 0.64-1.17), AD (HR=0.63, 95% Cl: 0.43-0.94) and VaD (HR=0.82, 95% Cl: 0.41-1.66)
(West & Haan, 2009) 1,351 60-101 (mean 69.9) 8(5.6) BMI, WC 3MSE and DelRec dx dementia by DSM- III Compared with the lowest BMI category, overweight participants had a 48% decreased rate of dementia or being cognitively impaired not dementia (CIND) (adj [HR] = 0.52, 95% [Cl]: 0.30-0.91) and obese participants had a 61% decreased rate of dementia/CIND (HR = 0.39, 95% Cl: 0.20-0.78).
By contrast, the middle and high tertiles of WC were associated with higher rates of dementia/CIND compared with the low tertile, (adj HR = 1.8, 95% Cl: 1.1-3.1, and adj HR = 1.9, 95% Cl: 0.91-3.8 respectively).
(Xiong, Plassman, Helms, & Steffens, 2006) 166 65.81 12 BMI Dementia dx: TICS-m and clinical assessments Cognitive change was not significantly different between twins discordant for BMI.
UJ


Articles Assessing Mild Cognitive Impairment
Among the 5 longitudinal studies in which MCI was the principle cognitive outcome,
none reported evidence supporting an association between obesity and increased risk MCI. In
contrast, all 5 reported no association between obesity and MCI. Furthermore, all 5 studies
(100%) were assessed among older adults, not midlife. Results are displayed in Table 1.3.4
below.
32


Table 1.3.4. Longitudnial studies not supporting an association between overweight and MC 1.
OLDER ADULTS
Author (date) N Baseline Age Follow- up (years) Weight Assessment MCI Assessment Results
(Gao et al., 2011) 1331 >65 mean 75.8 6.4 mean BMI CERAD, Clinician Home-based Interview, and Cambridge Examination for Mental Disorders BMI in participants with incident MCI, did not differ 12 or 9 years before diagnosis, but 6 years before diagnosis, participants with incident MCI had significantly lower BMI than participants with normal cognition (p = .006). Participants with incident MCI had greater decline in BMI than those without (P = .04).
(Kerwin et al., 2011) 7163 65-80 4.4 mean BMI, WHR 3MSE Association between BMI and risk of Cl was modified by body fat distribution (WHR). Older women with a lower WHR had greater risk of Cl at both low and high BMI categories. Underweight women with a WHR less than 0.80 had a greater risk than those with higher BMI. In normal-weight to overweight women (BMI 20.0-29.9), central adiposity (WHR>0.80) is associated with greater risk of cognitive impairment and probable dementia than in women with higher BMI.
(Newman et al., 2009) 1677 77-102 Median 85 13 WC, BMI Modified MMSE, DSST, CES-D, Cl based on <80 on 3 MS Greater weight was not associated with higher rates of cognitive impairment. There was no significant association between WC and cognitive impairment.
(Ogunniyi et al., 2011) 1559 >65 5.97 mean BMI CERAD, Clinician Home-based Interview, Cambridge Examination for Mental Disorders A significantly greater decline in BMI was found in those later diagnosed with MCI (p < 0.001) compared to normal subjects.
UJ
UJ


Table 1.3.4. Longitudnial studies not supporting an association between overweight and MC 1.
OLDER ADULTS
Author (date) N Baseline Age Follow- up (years) Weight Assessment MCI Assessment Results
(Sachs- Ericsson, Sawyer, Corsentino, Collins, & Blazer, 2010) 2840 71.1(5.3) 4 times over 10 years BMI SPMSQ Lower BMI was a predictor of Cl for those with the APOE-e4 allele. For non-APOE 4 allele carriers, BMI was unrelated to Cl.
UJ
4^


Articles Assessing Cognitive Function
Among the 26 articles assessing cognitive function outcomes other than MCI or
dementia, one article reported both evidence that could be used to support and evidence that
could be used to oppose an association (Kanaya et al., 2009), so that in total 14 studies were
included in the table supporting an association between obesity and dementia, while 13 articles
were included in the table opposing this association. The articles assessing the association
between weight or adiposity and cognitive function tested a variety of cognitive domains,
including memory, executive function, and attention, which are recorded in Table 1.3.5 and
Table 1.3.6 below.
Supporting an association between obesity and cognitive function: Fourteen
longitudinal studies reported evidence for an association between obesity and declines in
various cognitive functions. Of these, 10 (71%) found that obesity in midlife increased risk of
cognitive decline, while 4 (29%) reported that obesity in older adults increased risk of cognitive
decline. Of the studies supporting a link, all 14 used BMI as an estimate of adiposity, only 4
combined BMI with waist circumference or waist-hip ratio (WHR).
Opposing an association between obesity and cognitive function: Thirteen longitudinal
studies reported no evidence of an association between any measure of adiposity and cognitive
decline. Four (31%) of these studies examined adiposity in midlife and 9 (69%) were among
older adults. Among the studies not supporting an association, 11 used BMI to estimate obesity,
while six used BMI combined with other measures such as WFIR.
35


Table 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
MIDLIFE
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Risk ratio Hazard ratio, Odds ratio
(Albanese et ah, 2012) 2083 Began at 15, 20, 26, 36, 43, and 53 years; 38 BMI Animal naming, National Adult Reading Test, word list recall, letter cancellation Midlife BMI gain from youth was inversely associated with memory in midlife. BMI gain from youth to age 53 years in men was independently associated with better memory. Average BMI was approx. 27 at age 53. Both underweight and obese women at 53 years had significantly lower memory scores, cross-sectionally.
(Cournot et al., 2006a) 2223 32-62 5 BMI Word-list learning, Digit-Symbol Substitution Test, WAIS, selective attention test, delayed free recall test Cross-sectionally, BMI was associated with lower cognitive scores. A higher BMI at baseline was also associated with a higher decline in word list learning (delayed recall) at follow-up. No significant association was found between changes in BMI and cognitive function.
(Dahl et al., 2010) 781 25-63 (mean 41.6) 16 BMI Information Synonyms, Analogies, Figure Logic, Kohs Block Design, Card Rotations, Digit Span (Forward and Backward), Thurstone's Picture Memory, Names and Faces, (Immediate and Delayed), Digit Symbol, Figure Identification Higher midlife BMI scores preceded lower general cognitive ability and steeper cognitive decline in both men and women.
(Dahl et al., 2012) 657 <50 (mean 39.9) 25 BMI Test battery representing four domains: verbal, spatial/fluid, memory, and perceptual speed Being overweight or obese in midlife was associated with cognitive decline later in life. Weight decline across midlife rather than low weight in late midlife per se was associated with cognitive decline.
UJ
O')


Table 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
MIDLIFE
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Risk ratio Hazard ratio, Odds ratio
(Debette et ah, 2011) 1352 54 +/- 9 6.3 +/-1.1 BMI, WHR logical memory delayed recall, visual reproductions delayed- recall (VR-d), and Trail-Making Test B-A (TrB-A) Midlife obesity was associated with an increased rate of progression of decline in executive function a decade later. Large WHR in midlife was associated with marked decline in total brain volume.
(Hassing, Dahl, Pedersen, & Johansson, 2010) 417 50-60 30 BMI long-term memory, short-term memory, speed, verbal and spatial ability Midlife overweight is related to lower overall cognitive function in old age.
(Laitala et al., 2011) 2606 twins Midlife Old age BMI Validated telephone interview Midlife overweight increased the risk for mild impairment of cognitive function. Weight gain more than 1.7 kg/m and loss more than 2 kg/m within an average of 5.6 years were associated with lower cognitive performance independently of BMI.
(Sabia, Kivimaki, Shipley, Marmot, & Singh- Manoux, 2009) 10,308 35-55 21 BMI MMSE, AH4-I, inductive reasoning, phonemic fluency Long-term obesity and long-term underweight in adulthood are associated with lower cognitive scores in late midlife.
UJ
"sj


Table 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
MIDLIFE
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Risk ratio Hazard ratio, Odds ratio
(Singh- Manoux et al., 2012) 6401 39-63 (mean 48.9-50.0 10 BMI Memory, reasoning, semantic, and phonemic fluency In the metabolically abnormal group, the 10 year decline on the global cognitive score was faster among obese than among healthy weight individuals. In the metabolically normal group, the 10-year decline in the global cognitive score was similar in the normal weight, overweight, and obese groups.
(Wolf et al., 2007) 1814 40-69 8-12 BMI WHR Trails B, Visual Reproductions- Immediate and Delayed Recall Verbal memory (immediate and delayed recall) Midlife WHR in highest quartile was significantly related to poorer performance on executive function & visuomotor skills. Obesity was not related to verbal memory (immediate or delayed recall). Visuomotor skills but not memory were related to WHR.
UJ
00


Table 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Risk ratio Hazard ratio, Odds ratio
(Brubache r, Monsch, & Stahelin, 2004) 531 59.4(7.9) 10 BMI CERAD-NAB battery This study demonstrates that for both BMI loss and BMI gain there is a worsening of cognition. Odds ratio is greater at the extremes of weight loss or weight gain, up to OR=4 for BMI loss of 0.8kg/m2/yr and up to OR = 8 for BMI gain of 0.8kg/m2/yr.
(Elias, Elias, Sullivan, Wolf, & D'Agostin o, 2003) 1423 55-88 18 BMI Kaplan-Albert Neuropsychological Test Battery Obese men performed at a level of 0.44 s.d. below the level of non-obese men for total test score (P<0.0001). Obese vs. non-obese women showed no differences in total test score.
(Gunstad, Lhotsky, Wendell, Ferrucci, & Zonderma n, 2010) 1703 19-93 (Mean 55.5) Average 3.1 visits, average 2.0 years between visits BMI, WHR, WC MMSE, BIMC, WAIS-R, CVLT, letter fluency, card rotation Obesity was associated with poorer performance in a variety of cognitive domains, including global screening measures, memory, and verbal fluency tasks. Obesity was associated with better performance on tests of attention and visuospatial ability. An obesity by age interaction emerged in some domains, including memory, attention, executive function.
UJ
<£>


Table 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline.
OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Risk ratio Hazard ratio, Odds ratio
(Kanaya et al., 2009) 3054 70-79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneo us and visceral fat by CT 3MS In men higher levels of all adiposity measures were associated with worsening cognitive function in men. In women there was no association between adiposity and cognitive change.
4^
O


Table 1.3.6. Longitudinal studies not supporting an association between overweight and cognitive decline
MIDLIFE
Author (date) N Baseline Age Follow-up (years) Weight Measures Cognitive Measures Resu Its
(Arntzen, Schirmer, Wilsgaard, & Mathiesen, 2011) 5033 59 mean 7 BMI 12 word memory, Digit- Symbol Coding, Tapping test The authors found no consistent association between BMI and cognitive test results.
(Knopman, Mosley, Catellier, & Coker, 2009) 1130 59.6 +/- 4.3 12-16? BMI Delayed Word Recall (DWR) Test, the Digit Symbol Substitution (DSS) Test, and the Word Fluency (WF) Test Baseline BMI was not a risk factor for cognitive decline
(Lo, Pachana, Byrne, Sachdev, & Woodman, 2012) 334 40-79 (Mean 58.72) Mean 7.45 BMI, WC, WHR MMSE, Auditory Delayed Index, Visual Delayed Index, and Working Memory Index from WMS). Processing Speed Index from WAIS. No significant associations were found between BMI, WC, or WHR and any cognitive domains at follow-up. Both weight gain and loss were associated with poor Visual Delayed Index performance at follow- up compared with stable weight.
(Thilers, Macdonald, Nilsson, & Herlitz, 2010) 1480 40-65 10 BMI 3MSE Spanish and English Verbal Learning Test (SEVLT) Accelerated postmenopausal cognitive decline is restricted to women with normal BMI.


Table 1.3.6. Longitudinal studies not supporting an association between overweight and cognitive decline.
OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Resu Its
(Bagger, Tanko, Alexandersen, Qin, & Christiansen, 2004) 5607 63.8 mean 7.3 Weight, DXA body composition, TFM, TLM, CFM Short Blessed test Higher baseline weight was associated with lower cognitive decline. Women with the worst cognitive performance at follow-up were the ones who lost the most body weight and had the lowest central fat mass (CFM).
(Deschamps et al., 2002) 169 69 to 89 Mean 75.4 5 BMI ADL dependency, IADL dependency, MMSE No overweight patients (BMI >27) declined in cognitive function. Subjects with a BMI 23-27 had 3.6 times lower chance cognitive decline in the subsequent 5 y (OR=0.28, 95%, Cl 0.09-0.90) than subjects with BMI less than 23.
(Driscoll et al., 2011) 2283 65-79 Up to 5 Mean 3 BMI, WHR, WC 3 MS, PM A, Letter fluency, BVRT, CVLT, digital span, CRT, No association between weight and cognition in women who remained stable or gained weight. Cognition was not related to changes in WC. Weight loss was associated with cognitive decline, independent of initial BMI.
(Han et al., 2009) 721 60-85 2.13 BMI, WHR, WC, and % body fat CERAD-Korean version For men obese at baseline, increasing adiposity (BMI, WHR, WC) was associated with improved cognitive function. For women obese at baseline, both an increase or decrease in WHR or WC were associated with cognitive decline.
(Kanaya et al., 2009) 3054 70-79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3 MS There was no association between adiposity and cognitive change in women. Higher levels of all adiposity measures were associated with worsening cognitive function in men.
4^
NJ


Table 1.3.6. Longitudinal studies not supporting an association between overweight and cognitive decline.
OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Measu res Cognitive Measures Resu Its
(Raffaitin et a 12011) 7087 65+ 4 WC MMSE, Isaacs Set Test (verbal fluency), Benton Visual Retention Test (BVRT, visual working memory) No significant change in cognition in those who started with a high WC.
(Regal & Heatherington, 2012) 94 71-86 1 BMI MMSE, Montreal Cognitive Assessment, Frontal Assessment Battery, and Addenbrooke Cognitive Assessment Underweight group had the lowest cognitive scores of the four groups in all 12 comparisons. The overweight group had the highest cognitive scores in nine of 12 comparisons.
(Sturman et al., 2008) 3885 65+ Up to 10 Mean 6.4 BMI MMSE, East Boston Tests of Immediate Memory and Delayed Recall and the Symbol Digit Modalities Test Among older adults, higher BMI was not predictive of cognitive decline.
(Xiong et al., 2006) 166 pairs of twins discordant for BMI 67-77 12 BMI >30 kg/m Modified Telephone Interview for Cognitive Status (TICS-m) Cognitive change was not significantly different between members of pairs discordant for BMI.
4^
UJ


Articles Measuring Central Obesity
Looking across the different studies examining cognitive function, MCI or dementia
diagnosis outcomes, a total of 22 studies investigated the association between central obesity
and any of these cognitive health outcomes. The results of these studies can be seen in Table
1.3.7 and Table 1.3.8.
Supporting an association between central obesity and cognitive health: A total of 11
studies reported an association between central obesity and cognitive outcomes. Six (55%) of
these studies measured central obesity in midlife, while 5 (45%) measured central obesity in
older adults.
Opposing an association between central obesity and cognitive health: In contrast, 12
studies reported no association between central obesity and cognitive outcomes. Only two
(17%) of these studies reported the association with midlife central obesity, while 10 (83%)
measured central obesity in older adults.
1.3.4 Conclusions
The existing empirical literature on weight and cognitive function reveals a mixture of
results that support an association between increased weight and cognitive decline or dementia,
and studies that do not. Among the studies that do not support this association, many report
that low weight or weight loss increases risk. Some others report that higher weight may even
be protective, decreasing risk of dementia or cognitive decline. However overall, it would
appear that the mixed findings are at least partly due to differences between midlife and late
life measures of adiposity. The majority of negative findings measured baseline adiposity among
older adults (>65 years), while the majority of positive findings were measured adiposity among
adults in midlife. Research has indicated that weight loss is a common feature of dementia, and
may precede dementia diagnosis by 6 years or more (Gao et al., 2011). It is therefore possible
44


that the baseline weight among older adults already reflected some of these changes. Upon
comparison of the studies of older adult weight that were for or against the association there
did not appear to be a systematic difference in follow-up time, but there may be other
systematic differences not accounted for in this study.
45


Table 1.3.7. Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes.
STUDIES OF MIDLIFE CENTRAL ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Beydoun & Beason- Held, 2008) 2,322 >20 23.4 years BMI, WC Dx AD, NINCDS-ADRDA criteria Among men, being underweight (BMI /=30) at age 30, 40, or 45 years and jointly centrally obese (waist circumference >/= 80th percentile) at age 30, 35, or 50 years increased AD risk (HR = 6.57, 95% Cl: 1.96, 22.02). Women who lost weight (BMI change <10th percentile) between ages 30 and 45 years were also at increased risk (HR = 2.02, 95% Cl: 1.06, 3.85). Weight gain among men (BMI change >90th percentile) between age 30 and 50 years increased AD risk (HR = 3.70, 95% Cl: 1.43,9.56).
(Debette et al., 2011) 1352 54 +/-9 6.3 +/-1.1 BMI, WHR logical memory delayed recall, visual reproductions delayed- recall (VR-d), and Trail-Making Test B-A (TrB-A) Midlife obesity was associated with an increased rate of progression of decline in executive function a decade later. Large WHR in midlife was associated with marked decline in total brain volume.
(Gustafson et al., 2009) 1462 38-60 32 BMI, WC, WHR Dementia dx: DSM-lll-R criteria; AD dx: NINCDS-ADRDA criteria; VaD dx: NINDS-AIREN criteria While Cox models showed no association between baseline anthropometric factors and dementia risk, logistic models showed that a midlife WHR greater than 0.80 increased risk for dementia approximately twofold (odds ratio 2.22, 95% confidence interval 1.00-4.94, p=0.049) among surviving participants.
4^
O')


Table 1.3.7. Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes.
STUDIES OF MIDLIFE CENTRAL ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Whitmer et al., 2005) 10,276 40- 45 21-39 BMI, skinfold thickness Dementia dx: ICD-9 codes Compared with healthy weight: - Obesity in midlife (BMI >= 30) increased risk of dementia 74% (HR 1.74, 95% Cl 1.34 to 2.26). - Overweight in midlife (BMI 25.0-29.9) increased risk of dementia 35% (1.35, 1.14 to 1.60). The highest quintile of skinfold thickness had a 72% greater risk of dementia than the lowest quintile (1.72,1.36 to 2.18, and 1.59,1.24 to 2.04).
(Whitmer et al., 2008) 6583 40-45 ~26-42 BMI, Sagittal Abdominal Diameter (SAD) ICD-9 codes Persons in the highest quintile of SAD in midlife had increased risk of dementia (HR, 2.72; 95% Cl, 2.33-3.33) compared to persons in the lowest quintile. Those with high SAD (>25 cm) but healthy BMI had an increased risk (HR, 1.89; 95% Cl, 0.98-3.81) vs. those with low SAD (<25 cm) and healthy BMI. Persons who were both obese and with high SAD had the highest risk of dementia (HR, 3.60; 95% Cl, 2.85-4.55).
(Wolf et al., 2007) 1814 40-69 8-12 BMI WHR Trails B, Visual Reproductions- Immediateand Delayed Recall, Verbal memory (immediate and delayed recall) Midlife WHR in highest quartile was significantly related to poorer performance on executive function & visuomotor skills. Obesity was not related to verbal memory (immediate or delayed recall). Visuomotor skills but not memory were related to WHR.


Table 1.3.7. Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes.
STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Gunstad et al., 2010) 1703 19-93 (Mean 55.5) Average 3.1 visits, BMI, WHR, WC MMSE, BIMC, WAIS-R, CVLT, letter fluency, card rotation Obesity was associated with poorer performance in a variety of cognitive domains, including global screening measures, memory, and verbal fluency tasks. Obesity was associated with better performance on tests of attention and visuospatial ability. An obesity by age interaction emerged in some domains, including memory, attention, executive function.
(Kanaya et al., 2009) 3054 70-79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3 MS In men higher levels of all adiposity measures were associated with worsening cognitive function in men. In women there was no association between adiposity and cognitive change.
(Kerwin et al., 2011) 7163 65 to 80 up to 8 average 4.4 WHR, BMI 3MSE Dementia Dx Central obesity was what mattered. Women with a WHR of >0.80 and a BMI of 20.0 24.9 had a greater risk of cognitive impairment and probable dementia than more-obese women or women with a WHR less than 0.80. However women with a WHR less than 0.80 and a BMI of 20.0 to 24.9 kg/m2 had poorer scores on cognitive assessments.
4^
00


Table 1.3.7. Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes.
STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Luchsinger et al., 2007) 893 BMI, 907 WC, 709 weight >65 5 BMI, WC, weight Dementia dx: DSM-IV criteria; AD dx: NINCDS-ADRDA criteria In persons <76 years the association between BMI and dementia resembled a U shape, meaning that both low BMI and high BMI increased risk of dementia. In those >76 years, dementia risk decreased with higher BMI. In persons <76 years, the highest quartile of WC correlated to dementia and AD risk. Weight loss was related to a higher risk of dementia and dementia associated with stroke DAS). Weight gain was related to a higher DAS risk only.
(Luchsinger et al., 2012) 1459 >65 7 BMI, WHR, WC AD dx NINCDS-ADRDA criteria Only higher WHR was related to higher AD risk, with the highest quartile of WHR increasing risk 2.5 times compared to the lowest quartile (HR = 2.5; 95% Cl = 1.3 4.7).


Table 1.3.8 . Longitudinal studies measuring central obesity that did not support a link between obesity and cognitive outcomes.
STUDIES OF MIDLIFE CENTRAL ADIPOSITY
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Lo et al., 2012) 334 40-79 (Mean 58.72) Mean 7.45 Weight, WC, WHR MMSE, Auditory Delayed Index, Visual Delayed Index, and Working Memory Index from WMS), Processing Speed Index from WAIS No significant associations were found between BMI, WC, or WHR and any cognitive domains at follow-up. Both weight gain and loss were associated with poor Visual Delayed Index performance at follow-up compared with stable weight.
(Newman et al., 2009) 1677 77-102 Median 85 13 WC, BMI Modified MMSE, DSST, CES-D, Cl based on <80 on 3MS Greater weight was not associated with higher rates of cognitive impairment. There was no significant association between WC and cognitive impairment.
un
o


Table 1.3.8 . Longitudinal studies measuring central obesity that did not support a link between obesity and cognitive outcomes.
STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Bagger et al., 2004) 5607 63.8 mean 7.3 Weight, DXA body composition, TFM, TLM, CFM Short Blessed test Higher baseline weight was associated with lower cognitive decline. Women with the worst cognitive performance at follow-up were the ones who lost the most body weight and had the lowest central fat mass (CFM).
(Driscoll et al., 2011) 2283 65-79 Up to 5 Mean 3 BMI, WHR, WC 3MS, PMA, Letter fluency, BVRT, CVLT, digital span, CRT, No association between weight and cognition in women who remained stable or gained weight. Cognition was not related to changes in WC. Weight loss was associated with cognitive decline, independent of initial BMI.
(Forti et al., 2010) 749 >65 3-5 Metabolic Syndrome (MetS) WC MMSE Dx VaD, AD In participants aged 75 and older, abdominal obesity was associated with a lower risk of overall dementia (HR = 0.53, 95% Cl = 0.28-0.98).
(Han et al., 2009) 721 60-85 2 BMI, WHR, WC, PBF CERAD-Korean version Dementia dx using DSM-IV criteria For men obese at baseline, increasing adiposity (BMI, WHR, WC) was associated with improved cognitive function. For women obese at baseline, both an increase or decrease in WHR or WC were associated with cognitive decline
(Hughes et al., 2009) 1,478 Japanese Americans of the Kame Project mean age of 71.8 years and were dementia- free at baseline (1992- 1994) Biennially for 8 years (mean 7.8) BMI, WC, WHR Cognitive Abilities Screening Instrument and CERAD dx using DSM-IV criteria Higher baseline BMI was significantly associated with a reduced risk of AD ([HR] = 0.56, 95% [Cl] = 0.33-0.97) in the fully adjusted model. Slower rate of decline in BMI was associated with a reduced risk of dementia (HR = 0.37, 95% Cl = 0.14-0.98), with the association stronger for those who were overweight or obese (HR=0.18, 95% CI=0.05-0.58) compared to normal or underweight (HR=1.00, 95% CI=0.18-5.66) at baseline.


Table 1.3.8 . Longitudinal studies measuring central obesity that did not support a link between obesity and cognitive outcomes.
STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS
Author (date) N Baseline Age Follow-up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio
(Kanaya et al., 2009) 3054 70-79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3 MS There was no association between adiposity and cognitive change in women. Higher levels of all adiposity measures were associated with worsening cognitive function in men.
(Power et al., 2011) 12,047 65-84 (mean 72.1) 9.7 BMI, WC, WHR ICD-9 & ICD-10 dx codes in Western Australia Data Linkage System Overweight men and those with WHR > 0.9 have lower risk of dementia than men with normal weight and with WHR< 0.9. Higher adiposity was not associated with incident dementia.
(Raffaitin et al., 2009) 2049 74 4 WC MMSE, Benton Visual Retention Test, Isaac's Set Test Dx VaD, AD Participants with high WC had decreased risk for all- cause dementia (HR = 0.86, 95% Cl: 0.64-1.17), AD (HR=0.63, 95% Cl: 0.43-0.94) and VaD (HR=0.82, 95% Cl: 0.41-1.66)
(Raffaitin et al., 2011) 7087 65+ 4 WC MMSE, Isaacs Set Test (verbal fluency), Benton Visual Retention Test (BVRT, visual working memory) No significant change in cognition in those who started with a high WC.
(West & Haan, 2009) 1,351 60-101 (mean 69.9) 8(5.6) BMI and WC 3MSE and DelRec dx dementia by DSM-III Compared with the lowest BMI category, overweight participants had a 48% decreased rate of dementia or being cognitively impaired not dementia (CIND) (adj [HR] = 0.52, 95% [Cl]: 0.30-0.91) and obese participants had a 61% decreased rate of dementia/CIND (HR = 0.39, 95% Cl: 0.20-0.78). By contrast, the middle and high tertiles of WC were associated with higher rates of dementia/CIND compared with the low tertile, (adj HR = 1.8, 95% Cl: 1.1-3.1, and adj HR = 1.9, 95% Cl: 0.91-3.8 respectively).
un
NJ


For risk of incident dementia, there appears to be an increased risk associated with
midlife obesity. More studies assessed incident dementia than any other category. Among
these studies, diagnosis of dementia was generally made along similar criteria, such as DSM-IV
or NINCDS-ADRDA criteria. It is worth noting that many of these studies specifically examined the
diagnosis of AD, as opposed to simply all-cause dementia, including 15 of the studies supporting
an association between overweight and dementia. Therefore the obesity-dementia link is not
restricted to VaD, although this too was apparent.
Relatively little research has been conducted on the association between adiposity and
MCI. Importantly, no research studies have examined midlife adiposity in relation to incident
MCI. This may explain why no studies supported an association between overweight and MCI.
Clearly further research assessing midlife adiposity and incident MCI is needed. This may be
particularly important because MCI often represents a prodromal, stage of dementia one
which is clinically identifiable yet early enough to be amenable to effective intervention. In any
case, primary prevention of MCI may be just as valuable as (or equivalent to) prevention of
dementia. Studies of incident dementia could add this to the available outcome measures to
increase the useful information.
The current literature on cognitive function suggests that being overweight in midlife
may increase risk of declines in a broad range of cognitive functions. Positive results were not
restricted to memory, nor any other cognitive domain, though memory was certainly
represented in both groups. Among studies of baseline adiposity in older adults, weight was
related to cognitive function but it was more common to find an association between low
weight and cognitive decline.
Despite these findings, the current literature on obesity and cognitive health is limited
by its reliance on BMI as the principle measure of adiposity. The majority of studies relied on
53


BMI alone, without inclusion of measures of fat distribution. This may have been a particular
problem for baseline weight among older adults, who tend to lose lean mass as they age,
making BMI a poor reflection of adiposity. It is possible that the reliance on BMI has contributed
to the mixed findings. To investigate whether central obesity is more closely associated with
cognitive decline or dementia than BMI we examined the studies of central obesity separately.
Across all categories of cognitive outcome the number of studies for and against an association
were approximately equal. Evidence for an association included a roughly equal mix of studies
using midlife vs. older adult adiposity measures at baseline. However the vast majority of
negative studies (83%) were among older adults at baseline. Therefore central obesity among
older adults did not seem to increase risk of cognitive decline or dementia. By contrast, central
obesity in midlife tended to do so.
In conclusion, the evidence to date suggests that midlife adiposity, whether global or
central, may increase risk of cognitive decline and dementia later in life. As this is drawn from
observational studies there is clearly a need for further research to investigate whether
adiposity plays a causal role, or whether the association is simply an artifact of other factors
such as poor nutrition or sedentary lifestyle. Inclusion of potential mechanisms that could
mediate an effect of obesity on neurocognitive health could add valuable information to our
understanding here. In addition weight loss studies may be useful, but should be conducted
with caution and appropriate safety measures given the evidence linking weight loss to
increased risk of cognitive decline and dementia. A systematic investigation of the effects of
weight loss on cognitive function is warranted before proceeding in this direction.
Limitations of Current Research
While the convenience of BMI makes it an understandable first choice measure of
adiposity, the evidence linking obesity to cognitive health would be strengthened by the use of
54


more accurate measures of adiposity. This is particularly important among older adults, where
BMI can be a poor reflection of body fat.
While many of these studies provide valuable information about real-world associations
in community-dwelling adults, the observational study designs cannot be used to determine
causality. Even among well-controlled longitudinal studies it remains possible that some other
factor related to obesity is in fact responsible for the cognitive decline.
Most of the studies controlled for socio-demographic factors such as education or SES,
and many controlled for related health conditions such as diabetes or cardiovascular risk factors,
including smoking, but few controlled for health-related behaviors linked to obesity, such as
quality of diet, physical activity, social support or stress. Furthermore, few studies of the
association between weight or adiposity and cognitive health outcomes have directly measured
mechanisms that could potentially mediate a link between obesity and risk of dementia. We
therefore discuss some of the potential confounding factors and mediating mechanisms that
could be implicated in an association between obesity and cognitive function.
1.4 Potential Confounding Factors
In the previous section we reported epidemiological evidence of a link between midlife obesity
and cognitive decline or risk of dementia later in life. However it remains possible that the
apparent link is due to other factors also related to obesity, rather than obesity per se. Many of
the observational studies described above statistically adjusted for some of these factors,
however none accounted for all of them. Such potential confounding factors could include
education (Evans et al., 2003; Evans et al., 1997; Fitzpatrick et al., 2004; Kukull et al., 2002),
depression (Green et al., 2003a), sleep apnea (Bedard, Montplaisir, Malo, Richer, & Rouleau,
1993; Grigg-Damberger & Ralls, 2012), physical inactivity (Plassman et al., 2010) and poor
55


nutrition (Solfrizzi, Panza, & Capurso, 2003). Reverse causation is also possible the potential
that cognitive impairment or neural damage precedes or even contributes to obesity. The
following sections give an overview of evidence regarding these different factors.
1.4.1 Education
Lower levels of education are associated with increased risk of dementia, even after
controlling for socio-economic status (Evans et al., 2003; Evans et al., 1997; Fitzpatrick et al.,
2004; Kukull et al., 2002; Stern, 2002)). The association is not simply for all-cause dementia, nor
just VaD. For example Evans et al. (1997) calculated that the risk of AD was reduced by 17% for
every additional year of educational attainment. In a meta-analysis of 10 studies available
before 2005, Caamano-lsorna, Corral, Montes-Martinez, and Takkouche (2006) found that
compared to persons with high education level, persons with low education had a 59%
increased risk of all dementias (95% Cl: 1.26-2.01), with the risk of non-AD dementias was 1.32
(95% Cl: 0.92-1.88), and the risk for AD of 1.80 (95% Cl: 1.43-2.27). Similarly Meng and D'Arcy
(2012) analyzed 133 articles with 437,477 subjects published before 2011 and found that low
education increased risk of AD, VaD and unspecified dementia. The pooled incidence studies
gave an odd ratio of 1.88 (95%CI 1.51-2.34). McDowell, Xi, Lindsay, and Tierney (2007),
examined the compounding effect of socioeconomic and health factors associated with higher
education and determined that these could reduce but not remove the association of higher
education and lower risk of dementia.
The finding that lower education increases risk of dementia cannot yet be fully
explained. According to one theory, education may build "cognitive reserve" that helps to delay
the detection of cognitive decline (Stern, 2002, 2009). The cognitive reserve hypothesis is also
used to explain how some people with pathological signs of AD plaques and tangles never
develop symptoms within their lifetime (Roe, Xiong, Miller, & Morris, 2007). In an alternative
56


hypothesis for the link with education, it is possible that people with early cognitive deficits or
early stages of AD pathology may be less inclined to seek out mental stimulation many years
before dementia symptoms appear. It is also possible that people who have low education may
share a common set of independent risk factors. Education might be associated with good
lifestyle choices that reduce dementia risk, such as avoiding smoking, maintaining a healthy
weight, or avoiding diabetes (Knopman, 2008).
The association between education and obesity is complex, likely confounded by many
other factors. In addition, as overweight and obesity have become increasingly widespread
across the population in recent decades, the association has likely changed. There may have
been an increased risk of obesity with less educated individuals in the past, however more
contemporary analyses have found that the differentials may vary by sex and race (Yu, 2013)
found the majority of the differences between education and obesity disappeared after careful
adjustments, and only white college graduates were found to be less obese than white high
school graduates.
1.4.2 Sleep Apnea
Obstructive sleep apnea (OSA) is a disorder of sleep caused by frequent collapse of the
pharyngeal airway during sleep causing airways obstruction (Rosenberg & Doghramji, 2009). It is
usually diagnosed when the number of disordered breathing events per hour is >5 (Rosenberg &
Doghramji, 2009). In the general population approximately 3-7% in men and 2-5% in women
have OSA (Lurie, 2011). However prevalence rises significantly with increasing weight (Pillar &
Shehadeh, 2008). Among obese patients who present for bariatric surgery incidence of OSA can
be as high as 78% (Lopez, Stefan, Schulman, & Byers, 2008). Incidence also increases with age
(Lam, Sharma, & Lam, 2010). Up to 20% of adults over 70 years of age have OSA (Onen & Onen,
2010). Ancoli-lsrael, Klauber, Butters, Parker, and Kripke (1991) reported that in a sample of 235
57


nursing home patients with median age 83.5 (women) and 79.7 (men), 96% had some
degree of dementia and 70% had symptoms of sleep apnea.
Obstructive sleep apnea can cause neurocognitive damage across the lifespan (Bedard
et al., 1993; Grigg-Damberger & Ralls, 2012). It can also exacerbate cognitive dysfunction in the
elderly (Cooke et al., 2009; Kim, Lee, Lee, Jhoo, & Woo, 2011). Studies of the effects of CPAP
treatment for OSA on neurocognitive outcomes have produced mixed results (e.g. (Cooke et al.,
2009); compare (Ancoli-lsrael et al., 1991).
The association of OSA with cognitive impairment and AD in elderly patients does not
answer the question of causation. This is an area of active research (Roth, 2012). However given
the link between OSA and obesity it is a potential confounding factor that ought to be addressed
in future studies of the effects of obesity and cognitive function where possible.
1.4.3 Depression
Depressive symptoms are often associated with the development of AD and may appear
years before clinical AD diagnosis (Green et al., 2003b). However it is difficult to determine
whether these symptoms represent early manifestations of dementia, or whether depression
itself is a risk factor for AD. Major depression can lead to difficulty concentrating, memory
impairment, and difficulty making decisions (Taylor Tavares et al., 2007), which are also features
of AD (Sinz, Zamarian, Benke, Wenning, & Delazer, 2008; Zamarian, Sinz, Bonatti, Gamboz, &
Delazer, 2008). Furthermore persons with a history of depression are 2.5 times more likely to
develop AD than those who did not, and persons who experienced depression before age 60
have been found to be more than 4 times more likely to develop AD (Geerlings, Den Heijer,
Koudstaal, Hofman, & Breteler, 2008). The nature of the association between depression and
obesity is less clear. Depression and obesity often co-occur (Appelhans et al., 2012; Luppino et
al., 2010; Zhao et al., 2011), however some studies find no association (Hach et al., 2007;
58


Rivenes, Harvey, & Mykletun, 2009). In a study of 2,439 adults from the NHANES data, (Zhao et
al., 2011) demonstrated an incidence of major depressive symptoms in 1% of the general
population vs. 3% in the obese group and an incidence of moderate to severe depressive
symptoms as 2.4% in the general population and 6.7% in the obese group.
While the relationship between depression and obesity needs further clarification,
potential depression should be accounted for in studies of neurocognitive health and dementia
risk.
1.4.4 Poor Nutrition
It is also possible that persons who are overweight or obese in midlife have had poor
nutritional intake over many years. Nutritional deficiencies, rather than adiposity, could be the
cause for cognitive deficits. Over the past decade, a large number of studies have been
conducted to investigate the association between dietary composition and risk of dementia, and
this continues to be an active area of research.
Nutritional deficiencies can also cause impaired cognitive function more generally
(Solfrizzi et al., 2003). Deficiencies in essential micro-nutrients such as antioxidants (Vitamins C,
E, carotenes, etc.) and B vitamins have been associated with cognitive impairment (Del Parigi,
Panza, Capurso, & Solfrizzi, 2006). In addition, in some cross-sectional studies higher intake of
healthy foods was associated with better cognition, while intake of refined sugars, high
cholesterol and trans fats were associated with poorer cognitive performance (Lee et al., 2001;
Requejo et al., 2003). Higher intake of mono-unsaturated and polyunsaturated fatty acids has
been linked to better cognitive function and lower rates of cognitive decline. (Morris et al.,
2003; Solfrizzi et al., 2005; Solfrizzi et al., 2003).
There is also some evidence to suggest that nutritional factors could influence risk of
dementia. From a whole food perspective, there is evidence that persons who adhere to the
Mediterranean diet have lower risk of dementia (Scarmeas, Stern, Mayeux, & Luchsinger, 2006).
59


The strength of these findings was reflected in a recent NIH consensus report on preventing
cognitive decline and dementia, which reported that the Mediterranean diet is associated with
decreased risk of cognitive decline and dementia (Plassman et al., 2010) I. Observational studies
also link anti-oxidants such as Vitamin E and C to reduced risk of dementia (Engelhart et al.,
2002; Morris, 2005; Morris et al., 2002), however the NIH report found that there is currently
insufficient evidence to determine if this is true. Overall, the NIH report (Plassman et al., 2010)
concluded that the strength of evidence is currently insufficient to claim that specific nutrients
decrease risk of dementia. Clearly there is much more work to be done in this area, however
dietary composition should be considered in a thorough investigation of the association
between obesity and neurocognitive health outcomes.
1.4.5 Physical Inactivity
Persons who are overweight or obese are also less likely to be physically active than the
general population (Church et al., 2011; Tudor-Locke, Brashear, Johnson, & Katzmarzyk, 2010). If
physical inactivity impacts risk of cognitive decline or dementia then the association between
weight and dementia risk could be due to the adverse effects of physical activity, rather than to
obesity per se.
There is a growing body of evidence supporting an important role for physical activity in
promoting neural and cognitive health. Many observational studies report an association
between physical activity and lower risk of cognitive decline or dementia (Middleton & Yaffe,
2009). A recent meta-analysis (Skog, date) of prospective cohort studies found that 10 of 11
studies reported reduced risk of cognitive decline or dementia among adults who regularly
engaged in physical activity in midlife. While the epidemiological evidence is promising, the NIH
consensus report on preventing cognitive decline and dementia, found that there is currently
insufficient evidence from randomized controlled trials (RCTs) to conclude that physical activity
60


protects against cognitive decline and dementia (Plassman et al., 2010). Similarly Snowden and
colleagues (Snowden et al., 2011) found that there was not enough available data from quality
intervention studies to determine whether physical activity interventions improve cognitive
function in older adults. Nonetheless, the association seen in observational studies and small
clinical trials suggests that physical activity should be considered in assessments of the
association between obesity and dementia.
1.4.6 Reverse Causation
It remains possible that poor cognitive function can contribute to the development of
obesity, rather than the reverse. This would be consistent with the findings of Cournot et al
(Cournot et al., 2006b) that obesity in midlife was associated with increased risk of cognitive
decline later in life, but change in weight over time (including weight gain) was not associated
with increased risk of cognitive decline. Halkjaer, Holst, and Sprensen (2003) demonstrated that
intelligence test score was inversely related to the risk of developing obesity in a longitudinal
study of 1709 men (median age 19) over more than two decades.
Experimental studies are needed before the question of causation can be adequately
addressed. Interventions that reduce midlife obesity and measure incident dementia in older
age would be ideal. Such a study would be particularly difficult to implement given the large
sample size and long follow-up periods required. Experimental studies of more short-term
effects of weight loss on neurocognitive outcomes and on potential mechanisms implicated in
AD pathology would be more immediately practicable. The short-term effects of weight loss
interventions on adult neurocognitive health will be discussed in section 3.
61


1.5 Potential Mechanisms Linking Adiposity to Dementia Risk
1.5.1 Alzheimer's Disease Pathophysiology: A Brief Overview
While the pathophysiology of AD is not yet fully understood, many useful advances have
been made in this field in recent decades. The prevailing theory of AD pathophysiology holds
that the (B-amyloid-42 plaques which characterize the disease directly contribute to neuronal
death, though the possibility that (B-amyloid is instead a by-product of the real underlying
problem cannot yet be ruled out (Budson & Solomon, 2011; Resende, Ferreiro, Pereira, &
Oliveira, 2008). Beta-amlyoid is produced from the Amyloid Precursor Protein (APP), and is
found in the healthy human brain, though its function is not known. The reasons for (B-amyloid
accumulation are still under investigation. However some of the factors that could serve as
physiological mechanisms mediating an effect of obesity on neurocognitive health have already
been linked to increased (3-amyloid burden (Craft, 2005; Zhao et al., 2003).
1.5.2 Insulin Resistance and Glucose Regulation
Glucose dysregulation is one factor which could mediate an effect of obesity on risk of
AD. Impaired glucose regulation and insulin resistance are common features of obesity, and
insulin resistance is a central feature of both the metabolic syndrome (MetS) and Type 2
diabetes mellitus (T2DM). As already noted, persons with T2DM, are at increased risk of
cognitive decline or dementia, including AD (Biessels & Gispen, 2005; Elias et al., 2005; Yaffe et
al., 2004a). However, non-diabetic individuals with insulin resistance also show some evidence
of deficits in learning and memory (Vanhanen et al., 1997; Vanhanen et al., 2006), even after
controlling for vascular factors (Convit et al., 2003). In animal studies, a dose-response
relationship can be seen, with worse memory performance and smaller hippocampal volumes
recorded for animals with worse glycemic control, independent of age and overall cognitive
function (Convit et al., 2003).
62


Insulin resistance and hyperglycemia in the peripheral circulation can have significant
effects on neural glucose availability and on insulin levels in the brain. Growing evidence
indicates that peripheral insulin resistance and hyperglycemia can decrease the transport of
glucose across the blood-brain barrier (BBB) (McNay, Fries, & Gold, 2000), resulting in
hypoglycemia in the brain. Since neurons depend on glucose for energy the effect may be
devastating, particularly for brain regions frequently activated. Peripheral insulin resistance and
hyperinsulinemia also decrease insulin transport across the BBB, leading to insulin deficiency in
the brain (Baura et al., 1996; Kaiyala, Prigeon, Kahn, Woods, & Schwartz, 2000). Insulin
receptors can be found in particularly high concentrations in the hippocampus, a brain region
involved in learning and memory (Bingham et al., 2002; Jacobson & Sapolsky, 1991). Insulin may
affect memory through direct receptor-mediated effects (Craft, 2007).
Thus the brain of a person with peripheral insulin resistance, hyperglycemia and
hyperinsulinemia may become hypoglycemic and lacking the insulin normally involved in
memory function. Furthermore, raised serum insulin levels have been shown to increase IB-
amyloid formation in the brain and decrease its clearance to the periphery (Craft, 2005; Flo et
al., 2004; Marambaud, Zhao, & Davies, 2005; Zhao, Tuominen, & Kinnunen, 2004). Consistent
with these findings, patients with AD have high rates of glucose dysregulation (Craft, 2007) and
post-mortem reveals decreased insulin in the brains of AD patients relative to controls (Cole &
Frautschy, 2007; Rivera et al., 2005; Steen et al., 2005)
1.5.3 Hypertension
Hypertension is normally defined as a systolic blood pressure of 140mmHg or higher or
a diastolic pressure of 90mmHg or higher (Carretero & Oparil, 2000). In the general population,
prevalence of hypertension in adults is estimated at 24% in the USA (Burt et al., 1995), and 26%
worldwide (Kearney et al., 2005). Reviews indicate that those who are obese are more likely
63


than non-obese individuals to be hypertensive (Koebnick et al., 2012; McAuley et al., 2012;
McAuley et al., 2009). There are several potential reasons for this. Persons who are obese tend
to have higher levels of insulin and insulin stimulates the sympathetic nervous system,
stimulating the kidneys to retain sodium (Krieger & Landsberg, 1988), Obesity also increases
leptin levels, and leptin has been shown to increase blood pressure (Landsberg, 2001),). Obesity
can also contribute to increased aldosterone production (Ahmed, Fisher, Stevanovic, &
Hollenberg, 2005; Sarzani, Salvi, Dessi-Fulgheri, & Rappelli, 2008) and is associated with a high
incidence of obstructive sleep apnoea (Garni et al., 2004), which can increase sympathetic
nervous system output.
Hypertension has been linked to increased risk of MCI (Reitz et al, 2010) and dementia
(Kivipelto et al., 2001; Launer et al., 2000). For example, Kivipelto et al. (2001) found that people
with raised systolic blood pressure (SBP) (>160 mm Hg) in midlife had a more than two-fold
higher risk of AD in later life, compared to persons with normal SBP (OR 2.3, 95% Cl: 1.0 5.5). In
this study diastolic blood pressure (DBP) in midlife had no significant effect on the risk of AD. In
contrast Launer et al. (2000) found an association for both SBP and DBP in midlife with incident
dementia. Incidence of dementia among men with hypertension in midlife whose hypertension
was never treated increased almost five-fold in those with SBP 160 mm Hg and higher compared
with SBP of 110 to 139 mm Hg (OR 4.8, 95% Cl: 2.0-11.0). Risk of dementia was also increased
for DBP of 90-94 mm Hg (OR 3.8, 95% Cl: 1.6-8.7), and for DBP of 95 mmHg and over (OR 4.3,
95% Cl: 1.7-10.8), compared to those with DBP of 80 to 89 mm Hg. By contrast, men whose
hypertension was treated showed no increased risk for dementia. Consistent with this, Forette
et al. (1998) demonstrated that lowering of SBP in elderly patients significantly reduced the
incidence of dementia. At a structural level, Petrovitch et al. (2000) showed that midlife
64


hypertension was associated with neurofibrillary tangles (for DBP > 95 mm Hg), amyloid plaques
and low brain weight at autopsy (for SBP > 160 mm Hg).
In contrast to these studies of midlife hypertension increases risk, some studies have
indicated that hypotension late in life is associated with increased dementia risk. For example
Guo, Viitanen, Fratiglioni, and Winblad (1996) noted that people with moderate or severe
dementia were more likely persons without dementia to have hypotension, either systolic or
diastolic. Similarly Qiu, Winblad, and Fratiglioni (2009) found low blood pressure to be
associated with a more than two-fold increased risk of dementia or AD specifically, and Morris
et al. (2000) showed the same relationship with AD.
These findings may point to a non-linear, inverse U-shaped relationship between blood
pressure and cognitive health. In support of this, Razay, Williams, King, Smith, and Wilcock
(2009) found that among persons with AD rate of cognitive decline was increased in the groups
with either high or low DBP. Glynn et al. (1999) found a similar U shaped relationship with both
diastolic blood pressure and systolic blood pressure. As with the studies described above, their
results also suggest that midlife hypertension is a stronger risk factor for dementia than
hypertension later in life.
1.5.4 Dyslipidemia
Dyslipidemia, or abnormally elevated lipids in the bloodstream, is another common
feature of obesity (Nguyen, Magno, Lane, Hinojosa, & Lane, 2008). Dyslipidemia has been
associated with increased risk of dementia. High total cholesterol in midlife has been associated
with an increased risk for AD(Anstey, Lipnicki, & Low, 2008; Kivipelto et al., 2002; Solomon et al.,
2007b). By contrast, high total cholesterol late in life was not associated with MCI (Reitz et al.,
2008) nor with dementia (Anstey et al., 2008; Reitz, Luchsinger, Tang, Manly, & Mayeux, 2005;
Solomon et al., 2007a). However neither high total cholesterol nor high triglycerides late in life
65


were associated with reduced risk of dementia (Mielke et al., 2005). Interestingly, Solomon et al.
(2007a) noted that a moderate fall in TC levels between midlife and late-life was associated with
a more severe cognitive impairment.
As with other intervention trials for symptomatic dementia, various studies report that
use of statins in treatment of persons with established AD did not produce improvements in
symptoms (Feldman et al., 2010; McGuinness et al., 2010; Sano et al., 2011) nor did they
produce improvements in subjects who were 65 years of age or older (Arvanitakis et al., 2008;
Rea et al., 2005; Zandi et al., 2005). By contrast other studies examining the potential for statins
to prevent dementia have demonstrated a positive association (Cramer, Haan, Galea, Langa, &
Kalbfleisch, 2008; Haag, Hofman, Koudstaal, Strieker, & Breteler, 2009; Jick, Zornberg, Jick,
Seshadri, & Drachman, 2000; Wolozin, Kellman, Ruosseau, Celesia, & Siegel, 2000; Wolozin et
al., 2007), though the NIH consensus report on preventing cognitive decline and dementia found
that there is no consistent association and still insufficient evidence to claim that statin use
reduces risk of dementia (Plassman et al., 2010).
1.5.5 Oxidative Stress
Cellular damage by free radicals, such as reactive oxygen species, may be a universal
feature of the aging process (Calabrese et al., 2010; Knight, 2000; Nunomura et al., 2012) .
Affecting proteins, lipids and nucleic acids this oxidative damage, or oxidative stress, is normally
met by anti-oxidant defenses (Texel & Mattson, 2011). When these defenses are inadequate,
overwhelmed or exhausted tissue damage occurs (Friguet, 2002; Squier, 2001; Venkateshappa,
Harish, Mahadevan, Srinivas Bharath, & Shankar, 2012). The role of oxidative stress in AD
continues to be investigated as it may provide a trigger for some of the initial neural damage
(Foley & White, 2002; Sultana & Butterfield, 2008).
66


1.5.6 Leptin and Leptin Resistance
Leptin is a hormone secreted by adipocytes, Leptin levels are proportional to adiposity
in normal individuals (Ingvartsen & Boisclair, 2001; Popovic & Duntas, 2005). Although obese
individuals often have high leptin levels, many lack a normal response to leptin administration,
suggesting leptin resistance (Jung and Kim, 2013). This may occur for a variety of reasons,
including defective signaling pathways (Sahu, 2003), fructose consumption (Scarpace & Zhang,
2009; Shapiro et al., 2008; Shapiro, Turner, Gao, Cheng, & Scarpace, 2011), high sucrose and fat
diets (Vasselli, Scarpace, Harris, & Banks, 2013) or as a response to hyperleptinemia (Knight,
Hannan, Greenberg, & Friedman, 2010).
Leptin's actions include regulation of appetite, stimulation of thermogenesis,
enhancement of fatty acid oxidation, decreasing glucose, and reduction of body weight and fat
(Yadav, Kataria, Saini, & Yadav, 2012). Consistent with its regulatory roles, leptin receptors can
be found in the cerebral cortex, cerebellum, brainstem, basal ganglia, and hippocampus (Harvey,
2003) . Leptin also has effects on immunity (Ingvartsen & Boisclair, 2001). Leptin deficiency is
rare, but leads to severe obesity, hyperphagia, hypogonadism, hyperinsulinemia,
hypercholesterolemia and impaired immune function, all of which reverse with leptin
administration (Farooqi & O'Rahilly, 2009).
Leptin appears to have additional effects on the brain beyond the regulation of appetite
or weight. In particular, it has been implicated in AD pathology and symptoms. Leptin has been
shown to reduce (3-amyloid levels in vitro and brain levels of (3-amyloid in vivo (Fewlass et al.,
2004) . Levels have been shown to be decreased in AD, with serum levels inversely proportional
to AD severity (Greco, Sarkar, Johnston, & Tezapsidis, 2009; Holden et al., 2009). In a study of
2,871 older adults, Holden and colleagues (2009), followed over a period of four years,
demonstrated that those with a higher initial leptin level had lower rates of cognitive decline,
67


independent of other comorbidities or body fat (OR=0.66 95% Cl: 0.48-0.91). Interestingly, Zeki
Al Hazzouri and colleagues (Zeki Al Hazzouri, Haan, Whitmer, Yaffe, & Neuhaus, 2012)
demonstrated that a raised leptin level in elderly women of normal weight but not in
overweight or obese women was significantly associated with a lower risk of dementia or MCI.
They found similar results in a subsequent study of both men and women (Zeki Al Hazzouri,
Stone, Haan, & Yaffe, 2013).
1.5.7 Insulin-Like Growth Factor-1
Insulin-like growth factor (IGF-1) is a hormone involved in the regulation of cell
proliferation, cell differentiation, promotion of metabolism (nutrient transport, energy storage,
gene transcription and protein synthesis) and programmed cell death in adults (Feldman,
Sullivan, Kim, & Russell, 1997; Lauterio, 1992; Pavelic, Matijevic, & Knezevic, 2007). Most
circulating IGF-1 is produced by hepatocytes, however many other cell types, including neurons
(D'Ercole, Ye, Calikoglu, & Gutierrez-Ospina, 1996) also produce IGF-1, with paracrine or
autocrine effects (Croci et al., 2011; Frystyk, 2010; Laron, 2001). Receptors (IGF1R) can be found
throughout the body, and have widespread distribution in the brain (D'Ercole et al., 1996). IGF-1
also binds to insulin receptors, though with a lower affinity than to IGF1R. The relationship
between IGF-1 bioactivity with insulin levels, insulin resistance and the metabolic syndrome is
not straight-forward. There is an inverse U relationship between increasing insulin resistance
and IGF-1 activity. Increasing the number of elements of the metabolic syndrome correlates
with a peak of IGF-1 activity with three elements but falls thereafter (Brugts et al., 2010).
The function and activity of this hormone can be affected by a number of factors.
Activity of IGF-1 is modulated by growth hormone, IGF binding proteins, and IGF receptor
resistance (Connor et al, 2008). It is also affected by nutritional factors (Runchey et al., 2012)
and physical activity (Bermon, Ferrari, Bernard, Altare, & Dolisi, 1999; Rojas Vega, Knicker,
68


Hollmann, Bloch, & Struder, 2010). Levels of IGF-1 tend to decrease with age (Laron, 2002;
Muller et al., 1993; Sonntag, Ramsey, & Carter, 2005; Toogood, O'Neill, & Shalet, 1996).
Interestingly, some mutations in genes for the insulin/IGF-1 pathway have been linked to
increased longevity (Bonafe et al., 2003; Flachsbart et al., 2009; Kojima et al., 2004; Pawlikowska
et al., 2009) (Soerensen et al., 2010; Tazearslan, Huang, Barzilai, & Suh, 2011). It is also
associated with adiposity. For example, in a cross-sectional study of over 6,000 people Parekh
and colleagues (Parekh et al., 2010) found a strong positive correlation of serum IGF-1 levels and
adiposity (BMI or waist circumference), and a strong negative correlation with age. Consistent
with this, Gapstur and colleagues (Gapstur et al., 2004) showed a positive association between
weight and IGF-1 in a 9 year longitudinal study of 1418 adults aged 20 to 34 at enrolment. This is
not found in all studies, for Nam and colleagues (Nam et al., 1997) found that obese subjects
had similar levels of total IGF-1 compared to the controls, though they did have higher free IGF-
1 levels. The association with weight may not be linear, for Gram et al (Gram et al., 2006)
showed lower levels of IGF-1 with both low weight and high weight participants compared to
mid-weight participants in a study of 2139 women.
Several lines of evidence point to a neuroprotective role for IGF-1, and consequently
falling IGF-1 levels with age may reduce some of that protection. Firstly, IGF-1 normally provides
neuroprotection by increasing neuronal survival (Carro & Torres-Aleman, 2004). It can also
promote (3-amyloid clearance from the brain (Carro & Torres-Aleman, 2004; Freude, Schilbach,
& Schubert, 2009). There is increasing evidence that cerebral insulin and IGF-1 resistance are
major factors in AD (de la Monte, 2012; Talbot et al., 2012).
Human Studies Relating to IGF-1 and Cognition
In a cross-sectional study of 22 subjects aged between 65 and 86 years of age by
(Rollero et al., 1998) serum IGF-1 levels correlated positively with MMSE scores. Angelini and
69


colleagues (Angelini et al., 2009) found a similar association in a study of 75 hypertensive
patients over 65 years of age. Cognition was measured with MMSE, Cambridge cognitive
examination (CAMDEX-R), and the frontal assessment battery (FAB). In a 2 year prospective
study by Kalmijn and colleagues (Kalmijn, Janssen, Pols, Lamberts, & Breteler, 2000), a higher
IGF-1 level was associated with less cognitive decline among 186 healthy subjects aged 55 to 80
years of age. Dik and colleagues (Dik, Deeg, Visser, & Jonker, 2003) demonstrated that lower
IGF-1 levels were associated with poorer information processing speed and a faster decline over
three years among 1318 subjects aged 68 88 years. In a study of 17 healthy subjects between
the age of 66 and 77, Aleman and colleagues (Aleman et al., 2000) found that higher serum IGF-
1 levels were associated with better scores in mental processing speed.
Raising IGF-1 levels to treat AD symptoms has shown mixed results. Alvarez et al
(Alvarez et al., 2009a) demonstrated that the neurotrophic agent, cerebrolysyn, which improves
serum IGF-1 levels, also improved global function, disabilities and behavior in 207 late-onset AD
patients over a 24 week trial. However Sevigny and colleagues (Sevigny et al., 2008) stimulated
production of IGF-1 in a randomized trial of 563 patients with mild to moderate AD over 12
months. No improvement in the treatment group over the controls was observed.
While the research to date suggests that IGF=1 may play a role in AD pathophysiology
much more research is needed to determine whether the association is causal, or whether
interventions that increase IGF-1 can have neuroprotective effects.
1.5.8 Inflammation
A strong body of evidence shows that systemic low grade inflammation is a common
feature of obesity (Bastard et al., 2006; Black, 2002; Das, 2002; Ford, 2003; Wellen &
Hotamisligil, 2003), particularly central obesity (Craft, 2007; Fried, Bunkin, & Greenberg, 1998).
Epidemiological evidence demonstrates that inflammation is present during cognitive decline
70


and AD, however it is not clear whether the inflammation is a driving force behind neurological
damage, an innocent bystander, or part of a repair process (Bruunsgaard & Pedersen, 2003;
Bruunsgaard, Skinhoj, Pedersen, Schroll, & Pedersen, 2000). The relationship between the MetS
and risk of cognitive decline may be moderated by inflammatory cytokines such as IL-6, though
further research is needed (Yaffe et al., 2004b; Yaffe et al., 2004c). Chronic low-grade
inflammation, indicated by biomarkers such as Interleukin (IL)-6 and Tumor Necrosis Factor
(TNF)-a becomes increasingly common with age (Bruunsgaard et al., 2000; McGeer, Klegeris, &
McGeer, 2005). For example TNF-a increases in the cerebrospinal fluid (CSF) with brain aging,
and elevations have been reported in MCI and AD (Carro, Trejo, Gomez-lsla, LeRoith, & Torres-
Aleman, 2002; Tarkowski, Andreasen, Tarkowski, & Blennow, 2003). Furthermore TNF-a in the
CSF has been shown to antagonize the beneficial increases in (3-amyloid clearance that are
induced by IGF-1 (Alvarez et al., 2009b; Carro et al., 2002). Free IGF-1 correlates negatively with
serum TNF-a (Alvarez, Cacabelos, Sanpedro, Garcia-Fantini, & Aleixandre, 2007; Alvarez et al.,
2009b), so elevated TNF-a may contribute to increased (3-amyloid load (Craft, 2007). While
these findings are suggestive of a link between inflammation and neurocognitive function more
research is needed to determine the consistency, magnitude and meaning of the associations.
1.5.9 Cortisol and HPA Axis Dysregulation
Adipose tissue, and particularly central adipose tissue, secretes the glucocorticoid
"stress" hormone cortisol (Bjorntorp & Rosmond, 2000; Lottenberg et al., 1998) (Pasquali et al.,
1993). However it is now widely believed that in obese individuals cortisol secretion is increased
relative to healthy weight controls, but cortisol clearance is also increased, leading to normal or
low plasma cortisol concentrations (Morton, Ramage, & Seckl, 2004; Roberge et al., 2007;
Salehi, Ferenczi, & Zumoff, 2005). However obesity has been associated with dysregulation of
the normal variability of the cortisol diurnal rhythm (Bjorntorp & Rosmond, 2000).
71


Exposure to glucocorticoids (GCs) such as cortisol can have both direct effects on the
brain (McEwen, 2000, 2008; McEwen, Magarinos, & Reagan, 2002). Glucocorticoids readily cross
the BBB to act directly on the brain (McEwen, 2000). The effects of these GCs on neurocognitive
function depend on the magnitude and duration of exposure(McEwen, 1998, 2004). While mild-
moderate exposure can enhance attention and memory, more intense or severe exposure
impairs them both (Diamond, Park, & Woodson, 2004; Karlamangla, Singer, Chodosh, McEwen,
& Seeman, 2005; McEwen, 1998, 2000; McGaugh & Roozendaal, 2002). Furthermore, elevated
GCs are associated with clinical AD symptoms and pathology. For example, patients with early
AD may exhibit significantly higher total plasma cortisol than controls (Peskind, Wilkinson,
Petrie, Schellenberg, & Raskind, 2001), and higher plasma cortisol predicts more rapid cognitive
decline and decreased hippocampal volume (Lupien, Buss, Schramek, Maheu, & Pruessner,
2005a; Lupien et al., 2005b; Peskind et al., 2001; Rasmuson, Nasman, Carlstrom, & Olsson,
2002). Among those without dementia, older adults with significant increases in cortisol over 4
years show deficits in explicit memory, selective attention and an average 14% decrease of
hippocampal volume on MRI compared to others with no change or reductions in cortisol
(McEwen, 2000). Similarly older women with the highest cortisol levels had the lowest memory
scores, and increasing concentrations over a 2.5 year follow-up were associated with cognitive
decline (Seeman, McEwen, Singer, Albert, & Rowe, 1997). Building on findings such as these, a
glucocorticoid hypothesis of brain aging (Landfield, Blalock, Chen, & Porter, 2007) proposes that
repeated stress produces cumulative damage to the brain across the lifespan.
Glucocorticoids can also indirectly affect neurocognitive health via their effects on
immune function (McEwen, 1997, 1998; Munck& Naray-Fejes-Toth, 1994), insulin resistance
and glucose regulation (Black, 2002; Kyrou & Tsigos, 2007; Wellen & Hotamisligil, 2003, 2005),
72


and brain-derived neurotrophic factor (Taliaz et al., 2011; Tapia-Arancibia, Rage, Givalois, &
Arancibia, 2004).
1.5.10 Brain-Derived Neurotrophic Factor
Brain-derived neurotrophic factor (BDNF) is a neurotrophin that shows neuroprotective
effects in adults. It is involved learning and memory (Duan, Lee, Guo, & Mattson, 2001b; Lee,
Duan, Long, Ingram, & Mattson, 2000; Lu, Christian, & Lu, 2008) in as well as neurogenesis,
synaptic plasticity, and neurotransmitter synthesis (Diogenes, Assaife-Lopes, Pinto-Duarte,
Ribeiro, & Sebastiao, 2007; Lu, 2003; Mattson, Duan, & Guo, 2003). It is likely that BDNF is
involved in the protective cellular repair response after damage (Begliuomini et al., 2008;
Mattson, 2005), and it has been found to protect neurons in experimental models of AD and
Parkinson's disease (Duan, Guo, & Mattson, 2001a). Increased levels of BDNF are therefore
beneficial, but it is difficult to determine whether they reflect good health or the presence of
damage that needs repair.
The relationship between obesity and BDNF in humans is not yet clear. In animals, BDNF
deficiency leads to hyperphagia, obesity and insulin resistance (Duan, 2003), while central
infusion of BDNF in rats induces weight loss due to appetite suppression (Pelleymounter, Cullen,
& Wellman, 1995). However the effect of obesity on BDNF is not clear. In humans cross-
sectional studies of BDNF and weight give mixed. Some studies find decreased serum BDNF in
obese adults relative to controls (Krabbe et al., 2007), yet others have found higher serum BDNF
(Suwa et al., 2006). The differences may be due to the populations studied, or to potential
confounding factors, as BDNF is affected by gender, age, stress and inflammation among other
things (Makar et al., 2008); Makar et al, 2007)
While animal studies show significant effects of central (i.e. brain, CSF) BDNF on brain
function, there remains some question as to how well BDNF levels in peripheral circulation
73


reflect those in the central nervous system. Few studies have directly addressed this question
but there is some evidence that BDNF does cross the BBB. Poduslo & Curran (Poduslo & Curran,
1996) showed that methionine-BDNF can cross the BBB, though they did not test passage of the
natural form of BDNF. In a separate study, Pan and colleagues (Pan, Banks, Fasold, Bluth, &
Kastin, 1998) used radiolabelled BDNF in mice and showed that BDNF injected intravenously
could be found in the cerebral cortex parenchyma, suggesting passage across the BBB.
Conversely, after cerebroventricular injection, radiolabelled BDNF became detectable in blood
at a rate similar to that seen for re-absorption of cerebrospinal fluid (CSF). Despite these two
studies, the correlations between central and peripheral levels in other studies are mixed. Some
researchers have found correlation between serum & cortical BDNF in rats ((Hellweg,
Ziegenhorn, Heuser, & Deuschle, 2008; Ziegenhorn et al., 2007). Others have found that serum
BDNF correlates positively with cortical BDNF levels in newborn rats, but not in adults (Karege et
al., 2002). One study in humans attempted to measure what they hypothesized was BDNF
output from the brain using jugular to arterial concentration difference of BDNF in direct,
internal jugular vein sampling (Krabbe et al, 2008). However the validity of this methodology is
questionable (Lambert, 2008).
Animal studies of BDNF report an important role for BDNF in normal learning and
memory (Duan et al., 2001a; Lu et al., 2008; Mattson, 2000; Mu, Li, Yao, & Zhou, 1999). Mice
that over-express the BDNF receptor (TrkB) show evidence of enhanced learning and memory
(Koponen et al., 2004). Intra-hippocampal administration of BDNF facilitates short-term memory
(Alonso et al., 2002), and infusion of anti-BDNF antibodies causes amnesia for spatial learning
tasks in rats (Mu et al., 1999). Anti-BDNF antibodies block long-term potentiation (LTP) and
impair long-term memory. Transgenic mice deficient in BDNF show deficits in LTP (Croll et al.,
1999).
74


Evidence in humans also indicates an association between BDNF and memory, using
measures of BDNF in the peripheral circulation. Interestingly the direction of the association
appears to vary with severity of cognitive impairment. Yasutake et al (Yasutake, Kuroda,
Yanagawa, Okamura, & Yoneda, 2006a) compared healthy controls to patients with either AD or
VaD, matching these patients for age, gender and severity of dementia. Serum BDNF was
significantly lower in AD patients than in VaD patients and controls respectively. Despite this,
serum BDNF did not correlate with scores on the MMSE or the Functional Assessment Rating
Test (FAST) in patients with AD. Similarly Laske et al (Laske et al., 2007; Laske et al., 2006) found
that patients with established AD showed the lowest serum BDNF, but BDNF was slightly higher
in patients with early stage AD, relative to both normal controls and patients with AD. In this
study serum BDNF correlated significantly with MMSE (MCI r=0,855; p=0,001; AD r=0,396;
p=0,010). In another study, Zhang and colleagues (2008) found that patients with amnestic MCI
had significantly lower circulating BDNF than healthy controls, and that BDNF correlated
positively with scores on tests of delayed recall. Genotype for the BDNF gene did not differ
between aMCI patients and normal controls, nor did genotype correspond to serum BDNF
concentration, indicating that other factors strongly affect BDNF expression. In another study,
patients with AD showed lower BDNF than controls in the hippocampus and temporal cortex on
autopsy (Yasutake, Kuroda, Yanagawa, Okamura, & Yoneda, 2006b).
In the non-clinical population, BDNF also appears to correlate with cognitive function.
For example Komulainen et al (Komulainen et al., 2008), found that decreased BDNF was
associated with impaired global cognitive function (CERAD test battery) in women but not men,
and with specific impairments in memory but not executive function. In women a one standard
deviation decrease in BDNF was associated with 50-60% decreased memory scores, and
increased the probability of low MMSE. Effects remained after controlling for age, education,
75


depression, impaired glucose metabolism, cardiovascular disease, antihypertensive medication
lipid lowering medication, use of sex hormones, smoking, alcohol consumption, storing time of
plasma in the freezer and platelet count.
1.6 Summary
Obesity affects a number of factors currently under investigation for their role in
dementia pathophysiology, including glucose regulation, inflammation, leptin, IGF-1 and BDNF
(Lupien et al., 2005a). Observational studies suggest that midlife obesity can increase risk of
cognitive decline or dementia, yet relatively few observational or intervention studies have
concurrently investigated the potential mediating role of these mechanistic factors. Most
studies have controlled for education, socio-economic status, gender, age and diabetes status,
few to date have accounted for the potential role that physical activity, quality of diet or other
behaviors linked to obesity could have on the association between obesity and dementia. Flence
more research is needed before it can be concluded that obesity plays a causal role in cognitive
decline and dementia, and more research is needed to understand the mechanisms that may
mediate the effect. Ultimately the best information will be from large, well-controlled
randomized controlled trials. Flowever observational studies that account for frequently
unmeasured variables such as physical activity or diet can also contribute useful information.
2. Study 2: the NHANES-III Study
2.1 Introduction
Despite a growing body of epidemiological evidence suggesting that midlife obesity
increases risk of cognitive decline and dementia later in life (Gustafson, 2008; van den Berg et
al., 2008), much remains unknown about the association between obesity and cognitive
function in early and mid-adulthood. To better understand the nature of this relationship it will
76


be important to determine not on the direction of association, but also to investigate whether
duration of obesity affects the association, whether factors such as quality of diet and physical
activity moderate the association, and which physiological mechanisms could mediate the
effect. It is also important to determine whether distribution of body fat plays a role, for the
adipokines secreted by central adipose tissue may confer additional risk of neurocognitive
damage.
The primary aim of this study was to determine whether obesity is associated with
reduced cognitive function in early and mid-adulthood in the general population. Given the
evidence for an association between midlife obesity and cognitive decline ((Gorospe & Dave,
2007; Whitmer et al., 2008), and evidence for prolonged and gradual pre-clinical stages of AD
pathology that predate symptoms by decades (Sperling et al., 2011)it is possible that an
association between obesity and cognitive function is already apparent in midlife. This has been
supported by at least one study (Cournot et al., 2006a), but warrants further investigation.
Assessing dementia risk requires extremely long follow-up times, and it was not be possible to
determine whether such an association reflected an early stage of dementia pathology. Nor was
it possible to determine causal direction. However it is possible to assess whether results are
consistent with what would be expected from a causal association between obesity and later
cognitive decline.
Many studies of the association between obesity and cognitive function have used BMI
as the sole measure of adiposity. Yet at best, BMI is a rough indicator of adiposity, potentially
masking differences in body composition and fat distribution. The mechanisms that may link
obesity to dementia are more closely related to central adiposity than to global obesity, and
studies that differentiated between global and central obesity have tended to show stronger
associations for the latter (Cereda, Sansone, Meola, & Malavazos, 2007; West & Haan, 2009;
77


Whitmer et al., 2008). The present study therefore investigated the possibility that central
obesity, as measured by waist-hip ratio (WHR) is more closely related to cognitive function than
global obesity.
Duration of exposure to obesity may be important, since the cumulative effects of small
but regular damage could be significant over decades. A cumulative exposure-outcome model of
the lifecourse approach to chronic disease etiology (Kuh & Ben-Shlomo, 2004) would predict
that prolonged duration of exposure could contribute cumulative effects over time. This study
therefore explored whether longer duration of obesity was associated with worse cognitive
performance.
If obesity affects cognitive function, then behaviors that affect obesity may affect the
association between obesity and cognitive function. Differences in these behaviors could be
what really drives the association with obesity, rather than adiposity itself. Alternatively, health-
related behaviors may moderate the effects of obesity on neurocognitive health by affecting the
physiological mechanisms that increase risk of cognitive decline, such as cardiovascular health,
immune function, or levels of neuroprotective factors such as IGF-1 and BDNF. While quality of
diet and frequency of PA could affect rates of obesity itself, such that people with good diet and
regular PA are less likely to be obese, it is also possible that these factors may have independent
effects on cognitive function. Two possibilities present themselves. The association between
obesity and cognitive function described elsewhere may simply be an artifact of an effect of
poor quality diet and sedentary lifestyle on neurocognitive health. Alternatively, obesity could
have an independent effect that can be moderated by good quality diet and regular PA. For
example a person who is obese but physically active may have better cognitive function than
others who are sedentary. This study therefore investigated whether the association between
obesity and cognitive function in midlife was moderated by these health-related behaviors.
78


Since there is a promising body of evidence supporting a role for physical activity in promoting
neurocognitive health (Fratiglioni, Paillard-Borg, & Winblad, 2004; Smith et al., 2010b; Snowden
et al., 2011; Stranahan, Zhou, Martin, & Maudsley, 2009b), while the literature on diet is more
mixed (Faxen-lrving, Basun, & Cederholm, 2005; Gibson & Green, 2002; Plassman et al., 2010),
we hypothesized that PA, but not quality of diet, would moderate the association between
obesity and cognitive function in adults.
The present analysis was therefore undertaken to 1) determine whether obesity is
associated with cognitive function in the general US population in midlife, and 2) explore factors
that may moderate the association, including duration of obesity and health-related behaviors
such as diet and PA. The specific aims and hypotheses can be seen in Table 2.1.1.
79


Table 2.1.1. Specific aims and hypotheses of Study 2: NHANES.
Specific aim Hypothesis
SA.l. To determine whether obesity is HI. Obesity will be associated with worse
associated with worse cognitive function in midlife. cognitive function..
SA.2. To investigate whether central H2. The association will be stronger for central
obesity in midlife has a stronger obesity (WHR) than for measures of global
association with cognitive function than global obesity. obesity (BMI, % fat mass).
SA.3. To investigate whether duration H3. Obese persons who report they were obese
of obesity affects the association with 10 years ago will have worse cognitive function
cognitive function. than persons who have been obese less than 10
years.
SA.4. To determine whether quality of H4. The association between obesity and
diet and physical activity moderate the cognitive function will remain after adjusting for
association between adiposity and quality of diet and frequency of physical activity.
cognitive function. Physical activity, but not quality of diet, will
moderate the association between adiposity and
cognitive function.
Data from the Third National Health and Nutrition Examination Survey (NHANES-III)
provides a useful opportunity to investigate whether central obesity or global obesity is
associated with minor cognitive deficits in midlife, as well as the potential for health-related
behaviors to moderate any such association. Although new NHANES surveys have been
conducted since 1994, NHANES-III is the only NHANES study to date to include measures of
cognitive function. This secondary data analysis was declared exempt by the Colorado
Institutional Review Board. COMIRB Protocol 10-1154
80


2.2 Methods
2.2.1 Study Design
The Third National Health and Nutrition Examination Survey (NHANES-III) was a cross-
sectional epidemiological study of the civilian, non-institutionalized population of the United
States. A detailed description of NHANES-III recruitment, sampling and study methods can be
found online in the Plan and Operations Manual (NCHS, 1994; U.S. DHHS, 1996). A brief
summary of participants, methods and measured variables is provided here.
NHANES-III used a stratified, multi-stage probability sampling design to represent the
civilian population in 50 states, based on the 1980 census. Thirteen large counties were chosen
and grouped into 34 strata, with 89 sampling locations chosen and further segmented into city
and suburban blocks. The detailed approach has been previously described (Miller, 1973);
McDowell, 1981). Over the 6 years of the study, 39,695 people were selected for inclusion,
33,994 were interviewed in their homes and were invited to the mobile examination center for
medical examinations and follow-up tests. Of these, 78% (30,818) attended the mobile
examination centers and 493 others were given examinations in their homes. These home
examinations were included in order to gain data from very young children and the very elderly
who were unable to visit the mobile examination center. Home examinations included only a
subset of the components used at the mobile examination centers.
All participants were interviewed in their homes and given questionnaires to complete.
All interviewed participants were invited to attend the mobile examination center for medical
exam and further tests. At the mobile examination centers data collection began with a
household interview and several questionnaires. These were followed by a medical examination,
including the collection of blood and urine specimens. Other tests, such as hearing, vision and
the cognitive tests, followed. Staff administering the tests included nurses and clinicians for the
81


medical and clinical examinations and trained interviewers/research assistants for other tests as
appropriate. Standard operating procedures for each test are detailed in the NHANES-III 1988-
94 Reference Manuals and Reports (1996).
2.2.2 Participants
Cognitive tests were administered to a sub-sample of adults aged 20-59 years.
Assignment to the tests was based on participant number all participants who had an odd
participant number were assigned to the cognitive tests. This produced a sub-sample of 5138
non-institutionalized civilian men and women aged 20-59 years who completed the
computerized cognitive test battery. There were no medical or safety exclusions for the
cognitive testing component of NHANES-III. However for the purposes of our study, participants
were excluded if they did not complete both cognitive tests of interest (SDST and SDLT,
described below). Participants were also excluded if they were pregnant, or had evidence of a
condition known to affect cognitive function, including vitamin B12 deficiency (<174pg/mL),
hypothyroidism (TSH <10 ull/mL), reported a past medical diagnosis of stroke, or reported
recent alcohol use (>1 drink in preceding 3hrs). Since physical activity was a key independent
variable participants were also excluded if they reported significant difficulty walking / of a
mile. The final sample contained 4515 men and women aged 20 59 years.
2.2.3 Tests and Materials
Demographics and Medical History
All participants were asked to provide information on their age, gender, education,
occupation, income, ethnicity and medical history as part of an in-depth structured interview.
Data were recorded at both the NHANES home assessments and at the mobile examination
center. The poverty-income ratio (PIR) was computed as a ratio of family income to a composite
82


variable comprised of poverty threshold, the age of the family reference person, and the
calendar year in which the family was interviewed.
Assessment of Cognitive Function
Three cognitive tests were administered using the computerized Neurobehavioral
Evaluation System (NES). The Neurobehavioral Testing Manual details materials and procedures
used for tests of cognitive function (DHHS, 1986). However, only two were of interest for the
purposes of this study. The simple reaction time test (SRTT) was omitted as speed of tapping
was unlikely to be relevant to this study. A summary of the procedures and general results of
the cognitive tests have been reported by Krieg et al. (2001). The tests were conducted in a
quiet audiometry room of the mobile examination center trailer. Lighting, temperature,
environment and test administration were standardized and distractions minimized. Tests were
administered on a Compaq 286 Deskpro portable computer with a keyboard overlay to hide
unnecessary keys and a joystick. Participants were given the option of taking the tests in English
or Spanish. Tests were performed in a fixed order.
The Simple Reaction Time Test (SRTT) is a basic test of motor response speed to a visual
stimulus. (Baker, Chrzan, Park, & Saunders, 1985; Krieg et al., 2001; Letz, 1989, 1990).
Respondants rested the index finger of their preferred hand on a push button and were asked to
push the button as quickly as possible when they saw a solid square (4x4 cm) in the center of
the computer screen. When the button was pushed the square disappeared from the screen,
and reappeared between 2.5 5 seconds later. Each respondent was presented with a total of
50 trials. Responses latency from the time of the square appearing to the time the button was
pushed was recorded for each trial and averaged across all trials to produce a summary
response (Krieg et al., 2001).
83


The Symbol Digit Substitution Test (SDST) is a computerized adaptation of the Digit
Symbol Substitution Subtest of the WAIS-R (Wechsler, 1981), and is a test of visual motor speed
and coding speed (Krieg et al., 2001; Pavlik, Hyman, & Doody, 2004). Nine symbols were
presented, paired with 9 digits. Participants were presented with a grid that paired 9 different
symbols with the numbers 1 9. A similar grid, but with symbols in scrambled order and with
the spaces for the digits left blank, was presented at the bottom of the screen (Krieg et al.,
2001). Respondents were asked to press a numbered key to match the symbols that were
presented in scrambled order lower as quickly as possible. The test contained a practice and 4
trials. The time required to enter each digit and the number of errors were recorded. Responses
are in seconds, and higher scores represent worse performance.
Serial Digit Learning Test (SDLT) is reported by the NHANES-III test manual and by Krieg
et al. (2001) as a test of learning and memory, but is elsewhere considered a test of attention,
concentration or working memory e.g. (Suhr, Stewart, & France, 2004). Participants were asked
to learn a series of numbers that were presented slowly (0.6s with 0.6s in-between), one at a
time, on the screen. After the numbers were presented, participants used the numbered keys to
enter the sequence of numbers they remembered. The practice set involved 4 numbers.
Subsequent trials involved a sequence 8 numbers. The test ended when participants recalled
two consecutive number sets. Up to 8 trials were presented, each of which repeated the same
sequence of numbers. Total score was based on number of incorrect digits in each trial, and the
number of trials needed for accurate recall of all 8 numbers. SDLT scores ranged from 0-16 and
higher scores represent worse SDLT performance.
Depression
A sub-sample of participants aged 18 39 were given additional questions on history of
depression and mania symptoms. After being asked questions about whether they had ever
84


experienced periods of depressive symptoms they were asked "Are you in one of these spells of
feeling low or disinterested and having some of these other problems now?"
Anthropometric Measures
Weight, height, waist circumference, and buttock circumference were measured at the
NHANES-III mobile examination center (MEC). Waist circumference and buttock circumference
were used to calculate the waist-hip ratio (WHR). Bioelectric Impedance Analysis (BIA) was also
conducted at the mobile examination center. In BIA, a small and painless electrical current is
passed through the body and electrical impedance (ohms) and reactance (ohms) recorded. Body
composition was calculated using reactance and impedance ((Lukaski, Johnson, Bolonchuk, &
Lykken, 1985), creating an estimate of percent fat mass (PFM).
In addition to the objective anthropomorphic measures, participants also provided
some information on self-reported weight history. For the purpose of this analysis, responses to
the question: "Flow much did you weigh 10 years ago?" were used as an estimate of duration of
obesity for participants aged 30-59 years. Responses were given as weight in lbs.
Laboratory Measures
Blood was drawn at the NHANES-III mobile examination center for analysis of glucose,
insulin, glycated hemoglobin (FlbAlc), triglycerides, cholesterol (HDL, LDL, total), thyroid
stimulating hormone (TSH), vitamin B12, C-reactive protein (CRP), and insulin-like growth factor
(IGF)-l. A full description of laboratory procedures used for analysis is provided in the
Laboratory examination file (U.S. Department of Health and Fluman Services, 1996). Glycated
hemoglobin measurements for NFIANES III were performed by the Diabetes Diagnostic
Laboratory at the University of Missouri Columbia using the Diamat Analyzer System (Bio-Rad
Laboratories, Flercules, CA). Insulin-like growth factor 1 was analyzed in 2002 using a sub-set of
stored samples (U.S. Department of Health and Fluman Services, 2003).
85


Full Text

PAGE 1

A WEIGHTY MATTER: EFFECTS OF ADIPOSITY ON ADULT NEUROCOGNITIVE HEALTH by SARAH P.A. BRANNON B.Psych (Hons), University of Newcastle, 2004 A thesis submitted to the Faculty of the Graduate School of the University of Colorado Denver In partial f ulfillment of the requirements for the degree of Doctor of Philosophy Health and Behavioral Sciences 2013

PAGE 2

ii This thesis for the Doctor of Philosophy degree by Sarah P. A. Brannon h as been approved for the Health and Behavioral Sciences Program b y Ronica Rooks, Chair Mary Coussons Read, Advisor David Albeck William Donahoo May 2 2013

PAGE 3

iii Brannon, Sarah P. A. ( Ph.D., Health and Behavioral Sciences ) A Weighty Matter: Effects of Adiposity on Adult Neurocognitive Health Thesis directed by Professor Mar y Coussons Read ABSTRACT There is a growing body of evidence suggesting there are modifiable vascular and which obesity could contribute, its etiological contribution s remains unclear. We therefore investigated 1) the evidence linking obesity to cognitive health, 2) whether obesity in early or mid life is associated with cognitive change and the possible factors involved, and 3) whether dietary interventions that reduc e weight improve adult cognitive function. Study 1 involved a systematic literature review of the evidence linking adiposity to adult cognitive health outcomes. This revealed evidence that dementia was associated in older adults with low weight and separa tely with weight loss. However being overweight or obese in midlife was associated with cognitive decline and increased risk of dementia in later life. Study 2 examined the cross sectional association between adiposity and cognitive function in a nationall y representative sample of 4515 men and women, aged 20 59 years, who completed cognitive testing as part of the Third National Health and Nutrition Examination (NHANES III). Global obesity and central obesity both predicted a small proportion of the varia nce on the Ser ial Digit Learning Task (SDLT) and the Simple Reaction Time Task (SRTT), but not the Symbol Digit Substitution Test (SDST). Frequent physical activity (PA) modified the association of SDLT and central obesity to the point that subjects who w ere obese or overweight but physically active showed cognitive performance similar to that of persons of normal weight. Study 3 involved a systematic literature review of the effects of weight loss interventions on adult cognitive function The existing ev idence gives mixed results, but the majority of studies reported beneficial effects of

PAGE 4

iv weight loss on cognitive function, including memory. Study 4 compared the effects of 8 weeks of intermittent fasting (IF) with the effects of standard dietary restrictio n for weight loss, on the cognitive function and health of 26 obese adults. IF consisted of completely fasting one day and eating ad libitum the next. Although no effects on cognition were apparent at 8 weeks, at 6 months post intervention, the IF group sh owed improved memory, BDNF and greater loss of trunk fat. Reduced trunk fat was associated with improved memory scores. Summary. Results are consistent with the hypotheses that obesity is associated with cognitive deficits that can be ameliorated by weight loss and/or IF. However further research is needed. The form and content of this abstract are approved. I recommend its publication. Approved: Mary Coussons Read

PAGE 5

v ACKNOWLEDGEMENTS I would like to express my deepest gratitude to my advisor, Dr. Mary Cou ssons Read for her tireless support in the pursuit of this work. It has been an honor to work together, and I look forward to continuing to do so in the future. I am grateful to William Donahoo, M.D. for all of his help in developing this work and for inc luding me in this innovating and exciting project, the DRIFT study, on which this dissertation is based. It would not have been possible without the DRIFT study team, including William Donahoo, M.D., Edward Melanson Ph.D. and Dr. Wendolyn Gozansky, M.D., f or including me In addition, I would like to thank the team at the Department of Health and Behavioral Sciences for the education, training and opportunities to learn, as well as their moral and financial support. In particular I would like to thank Abby Fitch for all of her administrative support. Many thanks to the Colorado Clinical and Translational Science Institute (CCTSI) for the excellent pre doctoral training program in translational research they provided, complete with opportunities for clinical experience, professional development, and financial support. In particular I would like to thank Dr. Celia Sladek and Emily Warren for their key roles in making this program a success. I would also like to express my appreciation for the collaboration an d assistance provided by Dr. Mark Mattson and Dr. Bronwen Martin of the National Institute on Aging. Not only did they analyze the Brain Derived Neurotrophic Factor that forms an important outcome measure for this work, but their insights on the field, and excitement about this work were invaluable. My thanks also go to the Lionel Murphy Foundation of Australia for their significant financial support of this dissertation.

PAGE 6

vi This work was partially supported by NIH grants K23 AG026784 (WSG) and R21 AT0002617 ( WTD). The investigators retained full independence in the conduct of this research. The project described was also supported by the National Center for Research Resources, Grant TL1 RR025778 and is now at the National Center for Advancing Translational Sc iences, Grant TL1 RR025778 The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

PAGE 7

vii TABLE OF CONTENTS T ables ................................ ................................ ................................ ................................ ............ xii F igures ................................ ................................ ................................ ................................ ........... xiv A bbreviations ................................ ................................ ................................ ................................ xv Chapter 1. Adiposity and Dementia ................................ ................................ ................................ ........... 1 1.1 Introduction to Cognitive Decline and Dementia ................................ ............................... 1 1.1.1 Pre ................................ ................................ .......... 3 1.2 ................................ ................................ ... 6 1.2.1 The Lifecourse Approach to Chronic Disease ................................ ................................ ..... 7 1.2.2 Applying the Lifecourse Approach to Cognitive Aging ................................ ........................ 8 1.2.3 Evidence for Modifiable Risk Factors in Midlife ................................ ............................... 10 1. 3 Study 1. Weight and Cognitive Health: A Systematic Literature Review .......................... 12 1.3.1 Introduction ................................ ................................ ................................ ...................... 12 1.3.2 Methods ................................ ................................ ................................ ............................ 16 1.3.3 Results ................................ ................................ ................................ ............................... 18 1.3.4 Conclusions ................................ ................................ ................................ ....................... 44 1.4 Potential Confounding Factors ................................ ................................ ......................... 55 1.4.1 Education ................................ ................................ ................................ .......................... 56 1.4.2 Sleep Apnea ................................ ................................ ................................ ...................... 57 1.4.3 Depression ................................ ................................ ................................ ........................ 58 1.4.4 Poor Nutrition ................................ ................................ ................................ ................... 59 1.4.5 Physical Inactivity ................................ ................................ ................................ .............. 60 1.4.6 Reverse Causation ................................ ................................ ................................ ............. 61

PAGE 8

viii 1.5 Potential Mechanisms Linking Adiposity to Dementia Risk ................................ .............. 62 1.5.1 ................................ ................. 62 1.5.2 Insulin Resistance and Glucose Regulation ................................ ................................ ....... 62 1.5.3 Hypertension ................................ ................................ ................................ ..................... 63 1.5.4 Dyslipidemia ................................ ................................ ................................ ...................... 65 1.5.5 Oxidative Stress ................................ ................................ ................................ ................ 66 1.5.6 Leptin and Leptin Resistance ................................ ................................ ............................ 67 1.5.7 Insulin Like G rowth Factor 1 ................................ ................................ ............................. 68 1.5.8 Inflammation ................................ ................................ ................................ .................... 70 1.5.9 Cortisol and HPA Axis Dysregulation ................................ ................................ ................ 71 1.5.10 Brain Derived Neurotrophic Factor ................................ ................................ .................. 73 1.6 Summary ................................ ................................ ................................ ........................... 76 2. Study 2: the NHANES III Study ................................ ................................ ............................... 76 2.1 Introduction ................................ ................................ ................................ ...................... 76 2.2 Methods ................................ ................................ ................................ ............................ 81 2.2.1 Study Design ................................ ................................ ................................ ..................... 81 2.2.2 Participants ................................ ................................ ................................ ....................... 82 2.2.3 Tests and Materials ................................ ................................ ................................ ........... 82 2.2.4 Data Analysis ................................ ................................ ................................ ..................... 87 2.3 Results ................................ ................................ ................................ ............................... 92 2.3.1 Sample Characteristics ................................ ................................ ................................ ...... 92 2.3.2 Regression Model building process ................................ ................................ .................. 95 2.3.3 Regression Models of the Simple Reaction Time Test ................................ ...................... 96

PAGE 9

ix 2.3.4 Regression Models of the Symbol Digit Substitution Test ................................ .............. 101 2.3.5 Regression Models of the Serial Digit Learning Task ................................ ...................... 103 2.3.6 Sub analysis of the Duration of Obesity ................................ ................................ ......... 109 2.4 Discussion ................................ ................................ ................................ ....................... 113 3. Interventions ................................ ................................ ................................ ........................ 119 3.1 Introduction ................................ ................................ ................................ .................... 119 3.2 Physical Activity Interventions ................................ ................................ ........................ 119 3.2.1 Animal Studies of Physical Activity ................................ ................................ ................. 120 3.2.2 Human Studies of Physical Activity ................................ ................................ ................. 121 3.2.3 Potential Mechanisms for an Effect of Physical Activity ................................ ................. 125 3.3 Study 3: Calorie Restriction Interventions ................................ ................................ ...... 126 3.3.1 Introduction ................................ ................................ ................................ .................... 126 3.3.2 Methods ................................ ................................ ................................ .......................... 128 3.3.3 Results ................................ ................................ ................................ ............................. 130 3.3.4 Discussion ................................ ................................ ................................ ....................... 138 3.4 Dietary Composition and Frequency ................................ ................................ .............. 139 3.4.1 Quality of Diet ................................ ................................ ................................ ................. 139 3.4.2 Ketogenic Diets ................................ ................................ ................................ ............... 139 3.4.3 Fasting ................................ ................................ ................................ ............................. 141 3.5 Intermittent Fasting ................................ ................................ ................................ ........ 1 42

PAGE 10

x 4. Study 4: the DRIFT Study ................................ ................................ ................................ ...... 152 4.1 Introduction ................................ ................................ ................................ .................... 152 4.2 Methods ................................ ................................ ................................ .......................... 153 4.2.1 Participants ................................ ................................ ................................ ..................... 153 4.2.2 Intervention Protocols ................................ ................................ ................................ .... 155 4.2.3 Materials and Tests ................................ ................................ ................................ ......... 156 4.2.4 Procedure ................................ ................................ ................................ ........................ 162 4.2.5 Data Analysis ................................ ................................ ................................ ................... 168 4.3 Results ................................ ................................ ................................ ............................. 170 4.3.1 Study Attrition ................................ ................................ ................................ ................. 170 4.3.2 Sample Characteristics ................................ ................................ ................................ .... 171 4.3.3 Safety ................................ ................................ ................................ .............................. 173 4.3.4 Dietary Adherence ................................ ................................ ................................ .......... 173 4.3.5 Acute Effects of Fasting ................................ ................................ ................................ .. 174 4.3.6 Effects of IF on Body We ight and Composition ................................ .............................. 176 4.3.7 Effects of IF on Cognitive Function ................................ ................................ ................. 182 4.3.8 Effects of IF on Glucose Regulation ................................ ................................ ................ 183 4.3.9 Effects of IF on BDNF ................................ ................................ ................................ ...... 184 4.3.10 Effects of IF on HPA Axis Function ................................ ................................ .................. 185 4.3.11 Effec ts of IF on Pro Inflammatory Cytokines ................................ ................................ .. 187 4.3.12 Effects of IF on Leptin ................................ ................................ ................................ ..... 187 4.4 Discussion ................................ ................................ ................................ ....................... 190

PAGE 11

xi 5. Discussion ................................ ................................ ................................ ............................. 197 5.1 A Lifecourse Approach to Cognitive Aging ................................ ................................ ...... 197 5.2 Central Obesity May be More Strongly Linked to Neurocognitive Health than Global Obesity. ................................ ................................ ................................ ........................... 199 5.3 Behavior May Moderate the Association Between Obesity and Cognition ................... 200 5 .4 Interventions that Affect Obesity May Reduce Risk of Cognitive Decline ...................... 201 5.5 IF Produces Beneficial Effects on Memory in Obese Adults ................................ ........... 202 5.6 Plausible Mediating Mechanisms Exist ................................ ................................ ........... 203 5.7 Strengths and Limitations ................................ ................................ ............................... 205 5.7.1 Strengths and Limitations of Studies 1 a nd 3: Systematic Reviews ................................ 205 5.7.2 Strengths and Limitations of Study 2: The NHANES III Study ................................ ......... 206 5.7.3 Strengths and Limitations of S tudy 4: The DRIFT Study ................................ ................. 207 5.8 Novel Contributions ................................ ................................ ................................ ........ 209 5.9 Future Directions ................................ ................................ ................................ ............ 210 5.10 Summary ................................ ................................ ................................ ......................... 211 A p p e n d i x ................................ ................................ ................................ ................................ ...... 212 R e f e r e n c e s ................................ ................................ ................................ ................................ ... 2 2 7

PAGE 12

xii LIST OF TABLES Table 1.1.1. Cognitive function an d cognitive impairments defined ................................ ........................ 6 1.2.1. Some potential exposure outcome relations sugge sted by the lifecourse approach .......... 8 1.3.1. Summary of results for the obesity and cogni tive function literature review ................... 19 1.3.2. Longitudinal studies supporting an association bet ween overweight and dementia ........ 21 1.3.3. Longitudinal studies not supporting an association between overweight and dementia. 27 1.3.4. Longitudnial studies not supporting an associa tion between overweight and MCI .......... 33 1.3.5. Longitudinal studies supporting an association between overweight and cognitive decline. ................................ ................................ ................................ ................................ ............ 36 1.3. 6 Longitudinal studies not supporting an association between overweight and cognitive decline. ................................ ................................ ................................ ............................... 41 1.3.7. Longitudinal studies measuring cen tral obesity that supported an associat ion with any cognitive outcomes ................................ ................................ ................................ ............ 46 2.1.1. Specific aims and hypotheses of Study 2: NHANES. ................................ ........................... 80 2.2.1. Pre existing categorical variables al ready created in NHANES dataset .............................. 87 2.2.2. New categorical vari ables created for this analysis ................................ ............................ 90 2.3.1. NHANES III sample characteristics by BMI category. ................................ ......................... 94 2.3.2. Correlation between anthropometric measures ................................ ................................ 95 2.3.3. Multiple Linear Regression models of the association between measures of adiposity and SRTT ................................ ................................ ................................ ................................ .... 98 2.3.4. Multiple linear regression model of the association between central obesity (WHR) and reaction time (SRT T) ................................ ................................ ................................ ......... 100

PAGE 13

xiii 2.3.5. Multiple linear regression models of th e association between measures of adiposity and measures of cognitive function. SDLT (total score) or SDST (natural log of seconds) Each model was run separately ................................ ................................ ................................ 103 2.3.6. Multiple linear reg ression model of the association between central obesity and SDLT (total score) ................................ ................................ ................................ ...................... 108 2.3.7. SDLT scores by physical a ctivity and obesity (mean (SE)) ................................ ................. 109 3.3.1. Intervention studies with beneficial effects on cognition. ................................ ............... 132 3.3.2. Intervention studies with no effects, o r adverse effects, on cognition ............................ 136 3.5.1. Effects of IF and CR compared ................................ ................................ .......................... 143 4.2.1. CNS vital signs cognitive test doma ins and how they are calculated ............................... 158 4.3.1. DRIFT sample characteristics at baseline after a fed day. (mean (sd) unless otherwise indicated ) ................................ ................................ ................................ .......................... 172 4.3.2. Effects of an acute 36h fast at baseline on all participants before randomization (n=26). Measures were col lected at 7am the following day ................................ ........................ 175 4.3.3. Eight week post intervention group differences i n weight and cognitive function ......... 180 4.3.4. Eight week post intervention group differences in biomarkers. ................................ ...... 181 4.3.5. Between group differences 6 months aft er the end of the interventions ....................... 189

PAGE 14

xiv LIST OF FIGURES Figure 1.1. 1. A gradual process of pre 5 1.2.1. As people age, the cumulative effects of damaging and protective exposures may lead to wi dening differences in cognitive function. ................................ ................................ ........ 9 1.3.1. Steps of the systematic review p rocess for longitudinal studies ................................ ........ 18 2.3.1. Interac tion between central obesity and physica l activity for SDLT performance ............ 107 2.3.2. Number of participants in e ach strata of duration var iable ................................ ............. 111 3.3.1. Steps of the systematic review p rocess for longitudinal studies ................................ ...... 130 3.5.1. An inverse U shaped dose response curve. ................................ ................................ ...... 148 3.5.1. Repeated episodes of stress in the mild e cumulative beneficial effects ................................ ................................ .......................... 149 4.2.1. Ove rview of the DRIFT study design ................................ ................................ ................. 156 4.2.2. Baseline fed and baseline fasted visits compared ................................ ............................ 163 4.2.3. Outline of proce dures on a baseline study visit ................................ ................................ 165 4.3.1. Eight week chan ge in body weight by group (kg) ................................ ............................. 176 4.3.2. Eight week percent wei ght change for each participant ................................ .................. 177 4.3.3. Mean change in per cent fat mass by time and group ................................ ...................... 178 4.3.4. Mean differences in perc ent trunk fat by time and group ................................ ............... 178 4.3.5. Mean memory scores by time and intervention group ................................ .................... 182 4.3.6. Mean insulin sen sitivity (SI) by time and group ................................ ................................ 184 4.3.7. Mean BDNF by time and group (pg/mL) ................................ ................................ ........... 185 4.3.8. Mean leptin by time and group (ng/mL). ................................ ................................ .......... 188

PAGE 15

xv LIST OF ABBREVIATIONS AD isease A DF Alternate D ay Fasting BBB Blood Brain Barrier BDNF Brain Derived Neurotrophic Factor BMI Body Mass Index CR Calorie Restriction CRP C Reactive Protein DR Dietary Restriction GC Glucocorticoid HEI Healthy Eating Index HPA Hypothalamic Pituitary Adrenal KD Ketogenic D iet IGF 1 Insulin l ike Growth Factor 1 IF Intermittent Fasting IL 6 Interleukin 6 MCI Mild C ognitive I mpairment MetS Metabolic Syndrome NHANES National Health and Nutrition Examination Survey NIH National I nstitute of H ealth P A Physical Activity T2DM Type 2 Diabetes M ellitus TNF WC Waist C ircumference WHR Waist Hip Ratio

PAGE 16

1 1. Adiposity and Dementia 1.1 I ntroduction to Cognitive Decline and Dementia A growing body of epidemiological evidence suggests that obesity may increase risk of dementia ( Association, 2012 ) In a popula tion that is increasingly obe se and rapidly aging a causal link between obesity and AD would have significant public health implications. A clear understanding of the relationship of obesity to dementia risk is therefore important, but further research is needed before this can be a chieved. In particular, there is a need to determine the extent to which timing, duration and extent of obesity affect risk, and the mechanisms by which such risks may be altered. th leading cause of death in the United Sta tes today ( ) It is the leading cause of dementia, accounting for 60 80% of cases ( ) At present dementia afflicts over 5.4 million Americans, however with the aging population prevalence is likely to increase to 6.7 million by 2025 and 13.5 million by 2050 if nothi ng is done to prevent it ( ; Herbert, Beckett, Scherr, & Evans, 200 1 ) Globally, the direct and indirect costs of dementia were estimated to total US$ 604 billion in 2010, or 1% of the aggregated worldwide GDP ( World Health Organization, 2012 ) I n the same year, costs of dementia in the United Kingdom (£23 billion) almost matched those of cancer (£12 billion), heart disease (£8 billion) and stroke (£5 billion) combined ( World Health Organization, 2012 ) In the United States, Medicare, Medicaid and out of pocket payments for dementia health care, long term care, and hospice care were estimated to be $200B, while caregivers, primarily family members, provided an estimated 17.4 B hours of unpaid care valued at $210B ( ) Furthermore, the p hysical and emotional impact of dementia care giving was estimated to result in $8.7B in

PAGE 17

2 incr eased healthcare costs in the United States in 2011 ( ) These costs do not account for the devastating emotional and personal costs of dementia for caregi vers and patients. Dementia refers to a group of conditions characterized by a decline in memory and at least one other cognitive function that are severe enough to impair activities of daily living ( ) Affected cognitive functions include a wide variety of processes necessary for thinking, planning and action, including functions such as memory, attention, or executive function. Dementia can be caused by a number of different conditions, of which dementia of the Vascular dementia (VaD), is another type of dementia, caused by cerebrovascular incidents, which accounts for 5 15 % of all cases o f dementia in the United States ( ) Other types of dementia such as frontotemporal dementia (FTD) and deme ntia with lewy bodies make up the remainder o f dementia cases. currently no effective treatments for the underlying pathology, and no options for prevention. Since age is a primary risk factor, the development of pr imary prevention strategies to delay the onset of AD by 5 years could result in an estimated 57% reduction in the number of persons with AD in the United States, and reduce the projected Medicare costs of AD from $627 to $344 billion ( Sperling et al., 2011 ) However there is a clear need to better understand the factors that contribute to the disease, particularly early risk factors and pathophysiology, before su ccessful interventions can be developed. This has been highlighted by the lack of efficacy of clinical trials directed at treating AD pathophysiology in persons already cli nically diagnosed with dementia The lack of success may reflect the difficulty of r eversing the neurological damage already done

PAGE 18

3 by this late stage of the disease, rather than the appropriateness of the target. Intervening earlier in the disease process may have better success. Hence we need to better understand the disease process at pre clinical stages of the disease ( Sperling et al., 2011 ) At present formal diagnosis of AD can be made on autopsy, and is confirmed by the presence of hallma rk p athology amyloid plaques and tau tangles in the brain ( Budson & Solomon, 2011 ) However although a correlation between these pathological features and cognitive function is observed ( Riley, Snowdon, Desrosiers, & Marke sbery, 2005 ) some persons who show these pathological markers never develop clinical symptoms of dementia in their lifetime ( Riley et al., 2005 ) indicating tha t factors other than the presence of plaques and tangles themselves may be involved. By contrast clinical diagnosis of AD is currently based on severity of behavioral and cognitive symptoms. No biomarkers have yet been found to be reliable markers of the disease, so clinical diagnosis occurs once cognitive impairments have become so severe that they interfere with activities of daily living ( McKhann et al., 2011 ) 1.1.1 Pre clinical S tages isease The search for points of early intervention may be assisted by the growing recognition that pre clinical signs of the disease are apparent years, or even decades, before clinical diagnosis ( Jack et al., 2011 ; Sperling et al., 2011 ) These findings have been made possible by a combination of factors. Advances in neuroi maging have made structural, neurochemical and functional alterations in the brains of asymptomatic persons apparent and evidence of potential biomarkers of pathology in pre clinical persons also continues to build ( Jack et al., 2011 ; Sperling et al., 2011 ) Many longitudinal studies have also contributed to the growing awareness of subtle cognitive alterations many years before clinical diagnosis ( Sperling et al., 2011 ) Together, these advances, and the lack of success in treating symptomatic AD, have

PAGE 19

4 contributed to an awareness of the need to look for modifiable pre clinical processes during mid adulthood that could contribute to modification of disease pathology. Epidemiological evidence, discussed in detail in section 1 .3, suggests that midlife obesity may be one s uch pre clinical risk factor contribu ting to disease pathology What precedes clinically diagnosable dementia? Evidence of a gradual progression of cognitive decline is now apparent, though the rate of progression varies considerably between individuals ( Sperling et al., 2011 ) Impairment (MCI), which is now increasingly viewed as a prodromal stage of dementi a ( Albert et al., 2011 ) ( Budson & Solomo n, 2011 ; Sperling et al., 2011 ) MCI can be diagnosed when declines in memory or other cognitive functions are clinically detectable, but not severe enough to interfere with activities of daily living ( Albert et al., 2011 ) Many older adults do not seek medical attention at this stage, but it is estimated that 10 20% of adults older than 65 years have MCI ( ) Prognosis for those diagnosed with MCI varies. A small minority of individuals regain normal cognitive function, and s ome others maintain their mild impairment ( Bennett et al., 2002 ) However MCI has a high risk of progression to dementia, conversion rate from MCI to dementia is abou t 5 10% per year ( Etgen, Bickel, & Forstl, 2010 ) Rates may be higher among clinical populations pr esenting with memory complaints Prior to diagnosable MCI, subtle cognitive declines may also become gradually apparent in pre clinical popula tions ( Jack et al., 2011 ; Sperling et al., 2011 ) Cognitive decline involves decreased function from prior le vels in one or more cognitive domains. With cognitive function defined as the capacity for abilities such as attention, memory, perception, self volition, language and judgment ( Anderson & McConnell, 2007 ) A small amount of cognitive decline is a

PAGE 20

5 normal part of aging ( Budson & Solomon, 2011 ) and subjective complaints about memory and other cognitive functi ons increase with adv ancing age ( Newson & Kemps, 2006 ) however for some the extent of decline across midlife and into old age is greater than would be exp ected in normal cognitive ag ing Figure 1.1 1 A gradual process of pre clinical cognitive decline typically precedes

PAGE 21

6 Table 1.1 1 Cognitive function and cogni tive impairments defined Cognitive function The capacity for thinking, planning and acting, including functions such as attention, memory, self volition, language and judgment. Neurocognitive health Healthy brain and cognitive function. Cognitive impai rment A generic term referring to any impairment in cognitive function/s. Mild Cognitive Impairment (MCI) A clinical diagnosis involving impairment in one or more cognitive education ( Albert et al., 2011 ) Dementia A group of conditions characterized by significant declines in memory and/or other cognitive functions that are severe enough to affect activities of daily living ( McKhann et al., 2011 ) memory impairment and at least one other cognitive disturbance, that each cause significant impairment and represent a decline from prior functioning ( APA, 200 0 ) Vascular dementia (VaD) The development of memory impairment and at least one other cognitive disturbance, with evidence of cerebrovascular disease ( APA, 2000 ) 1.2 A L The emerging evidence of a long trajectory for AD pat hology ( Sperling et al., 2011 ) and cognitive decline suggests that this may be a disease developed over a lifetime, rather than a disease of old age ( Gustafson, 2008 ) Hence a lifecourse approach to AD could be useful. ( Kuh & Be n Shlomo, 2004 ) As already noted, significant cognitive decline with advancing age is not a normal part of aging, and most older adults retain excellent neurocognitive function late in their lives ( ) However age is the leading risk factor for AD. Most cases of dementia are diagnosed after the age of 65 ( ; Budson & Solomon, 2011 ) This raises an important question: what differentiates people who experience healthy cognitive aging from those who experience significant c ognitive decline and eventual dementia? Genetic research indicates that for most people genes contribute to only a small

PAGE 22

7 proportion of the difference ( ) This leaves a significant role for social and environmental exposures across the lifespan. 1.2.1 The Lifecourse Approach to Chronic D isease The lifecourse approach ( Kuh & Ben Shlomo, 2004 ) is widely used to understand the etiology of chronic diseases. As described by ( Kuh & Ben Shlomo, 2004 ) exposures throughout the lifespan can influence both the incidence of chronic disease and its course. Sensitivity to risk factors may vary across the lifespan. This can lead to different patterns of risk outcome relationships, such as those summarized in Table 1.2 1 below ( Glymour & Manly, 2008 ; Power & Hertzman, 1997 ; Wadsworth, 1997 ) These models of exposure are not mutually exclusive, but can interact to shape health outcomes, with effects that could become increasingly apparent with advancing age.

PAGE 23

8 Table 1.2 1 So me potential exposure outcome relations suggested by the lifecourse approach. Adapted from ( Glymour & Manly, 2008 ; Power & Hertzman, 1997 ; Wadsworth, 1997 ) Immediate risk model Short period between exposure and health outcome. Return to baseline health after removal of risk factor. E.g. Delirium in hospitali zed older adults can be reversed by medication management ( Gray, Lai, & larson, 1999 ) Cumulative model Each exposure le ads to some harm. The cumulative effect of exposures increases disease risk. Removal of exposure does not reverse harm already done. cognitive function ( Weisskopf et al., 2004 ) Latency model Exposure during a critical period of development increases risk of disease much later in life, but health effects may not be immediately apparent. E.g. Pove rty in early life may provide an early exposure to stress that increases vulnerability to cognitive decline later in life ( Lupien, King, Meaney, & McEwen, 2000 ) Social trajectory model Exposure sets in motion a succession of adverse social events that increase vulnerability later in life. E.g. Poverty early in life may reduce educational access, which reduces education attainment and opportunities for mentally s timulating occupations, which may increase risk of cognitive decline and dementia later in life ( Al Hazzouri, Haan, Whitmer, Yaffe, & Neuhaus, 2012 ) 1.2.2 Applying the Lifecourse Approach to Cognitive A ging With its emphasis on understanding the timing of exposures, the lifecourse approach could provide a useful framework to investigate factors differentiating healthy cognitive aging from cognitive decline or dement ia. Cognitive aging has not been a prominent focus for lifecourse epidemiology ( Glymour & Manly, 2008 ) However the lifecourse approach suggests that differences in cognitive function between individuals of the same age may reflect a range of

PAGE 24

9 detrimental or neuroprotective biological, psychological and social exposures experience d in early and mid adulthood ( Stein & Moritz, 1999 ) and as a person ages, the cumulative effects of these exposures could bring widening differences in cognitive function, as depicted in Figure 1.2 1 below. Figure 1.2 1 As people age, the cumulative effects of damaging and protective exposures may lead to widening diff erences in cognitive function. As indicated in the four exposure outcome models summarized in Table 1.2 1 investigation of the effects of risk factors for AD will require careful attention to the timing and duration of exposure. An exposure may be detrimental during sensitive periods of development, with effects that do not become apparent for many years, or which only appear when combined with another later exposure. For example it remains possible that exposure to obesity in late life alone does not confer additional risk for dementia, but exposure to maternal obesity while in utero or exposure to childhoo d obesity during critical periods of neural development, increase risk of dementia much later in life (latency model). It is also possible that prolonged duration of

PAGE 25

10 obesity during early and mid adulthood could contribute cumulative damage to the brain ove r many years (cumulative model), and that the duration of exposure acquired by older adults who become obese in their old age is insufficient to confer significant risk. It is beyond the scope of this dissertation to address the effects of obesity across the entire lifespan. Instead, this paper will focus on investigating factors consistent with a cumulative model. It will therefore focus on factors that could have cumulative effects on the neural and cognitive health of adults in early and mid adulthood. This time period could correspond to the pre clinical stages of AD in affected persons and so reflect early stage risk factors for later cognitive decline with advancing age. 1.2.3 Evidence for Modifiable Risk Factors in M idlife A number of factors have recent ly emerged that could potentially act as midlife risk factors for AD later in life. These include Type 2 diabetes mellitus, the metabolic syndrome and hypertension. Obesity is causally related to each of these conditions, and so has the potential to the un derlying pathological process. E vidence indicates that Type 2 diabetes mellitus (T2DM), a chronic disease characterized by insulin resistance and glucose dysregulation, increases the risk of MCI and dementia ( Biessels & Gispen, 2005 ; Whitmer, Karter, Yaffe, Quesenberry, & Selby, 2009 ; Yaffe et al., 2004a ) It is well known that people with diabetes have increased risk of micro vascular complications (e.g. neuropathy, retinopathy, nephr opathy, and macrovascular events (myocardial infarction, stroke), and are therefore at increased risk for vascular dementia. How ever there also seems to be an independent risk of developing AD. Type 2 diabetes mellitus can also have adverse effects on general cognitive function earlier in life, in the absence of clinically diagnosed MCI or AD ( Cukierman, Gerstein, & Williamson, 2005 ; Elias, Elias, Sullivan, Wolf, & D'Agostino, 2005 ) Both men and women with T2DM show deficits in attention processing speed, memory, and

PAGE 26

11 executive functioning ( Biessels, ter Braak, Erkelens, & Hijman, 2001 ; Biessels & Gispen, 2005 ; Manschot et al., 2007 ; Messier, 2005 ; Ostrosky Solis, Mendoza, & Ardila, 2001 ; Yaffe et al., 2004b ) The association has been found in cross sectional ( Messier, 2005 ; Ryan, 2005 ; Strachan, Deary, Ewing, & Frier, 1997 ) and prospective follow up studies ( Cukierman et al., 2005 ; Elias et al., 2005 ; Messier, 2005 ) In addition, intervention studies using insulin sensitizing agents have had some success in reduc ing cognitive decline ( Craft et al., 2003 ; Craft et al., 2012 ) Though these findings are not universal, systematic reviews of prospective studies to date ( Cukierman et al., 2005 ) and a growing professional consensus, indicates that the consistency and strength of the evidence is sufficient to warrant including cognitive decline as one of the potential complications of diabetes ( Plassman, Williams, Burke, Holsinger, & Benjamin, 2010 ) Several studie s also report that the Metabolic Syndrome (MetS) independently increases risk of cognitive decline and incident dementia ( Yaffe, 2007 ; Yaffe et al., 2004c ) According to the criteria of the Intern ational Diabetes Federation (IDF 2006), the MetS is defined as having central obesity and any two of the following: raised triglycerides (> 150 mg/dL), reduced HDL cholest erol (< 40 mg/dL), raised blood pressure (systolic BP > 130 or diastolic BP >85 mm Hg), or raised fasting glucose (>100 mg/dL), or treatment for previous diagnosis of any of those conditions. Since the combination of components can vary it is possible that different components confer independent risk of cognitive decline. Consistent with this, non diabetic individuals with insulin resistance also show some evidence of deficits in learning and memory ( Vanhanen et al., 1997 ; Vanhanen et al., 2006 ) even after controlling for vascular factors ( Convit, Wolf, Tarshish, & de Leon, 2003 ) An emerging body of evidence now suggests an association between obesity and increased risk of cognitive decline and dementia ( Cournot et al., 2006a ; Gustafson, 2006 ; Whitmer, 2007 ; Whitmer, Gund erson, Quesenberry, Zhou, & Yaffe, 2007 ) Obesity is a risk factor

PAGE 27

12 for T2DM and all components of the MetS ( Abbasi, Brown, Lamendola, McLaughlin, & Reaven, 2002 ; Boyko et al., 2000 ; Zimmet, Boyko, Collier, & Courten, 1999 ) ), and may play a causal role in many of the clinic al features of these conditions Obesity al ready contributes to over 300,000 deaths per year in the United States alone ( Boeka & Lokken, 2008 ) and rates of obesity have risen dramatically across the po pulatio n over the past 20 years. T he majority of the population who have experienced these weight gains have yet to reach the ages at which AD is most likely to manifest Hence if obesity does contribute even a small increase in risk of dementia, the implication s at a population level could be particularly significant for an aging and increasingly obese population. Therefore the goal of this dissertation is to investigate whether obesity affects adult neurocognitive health, as well as the potential that behaviora l interventions directed at reducing obesity or mitigating its effects could improve adult neurocognitive health. Towards these ends the author and colleagues conducted 3 studies. Study 1 involved a systematic review of the published empirical evidence lin king weight or adiposity to adult cognitive function. Study 2 investigated the association between adiposity and cognitive function in the general population. Study 3 involved a systematic review of the published empirical data on interventions for weight loss, and their effects on human cognitive function. Finally, study 4 investigated the effects of weight loss diets on the cognitive function of obese adults. 1.3 S tudy 1. W eight and C ognitive H ealth : A S ystematic L iterature Review 1.3.1 I ntroduction isease is the leading cause of dementia and the 6 th leading cause of death in the United States today ( ) It afflicts over 5.4 million Americans today, an d prevalence is likely to increase to 6.7 million by 2025 ( Herbert et al., 2001 ) Ther e is

PAGE 28

13 currently no effective treatment, but evidence for modifiable risk factors is emerging. Among these, several recent studies have linked obesity to increased risk of cognitive decline and dementia ( Cournot et al., 2006a ; Gustafson et al., 2009 ; Gustafson, Lissner, Bengtsson, Bjorkelund, & Skoog, 2004 ; Whitmer, Gunderson, Barrett Connor, Quesenberry, & Yaffe, 2005 ) If obesity contributes even a small increase in risk of dementia, the effects across the increasingly obese populations of Western nations coul d be significant. The purpose of this review is therefore to systematically investigate the evidence from prospective studies that obesity increases risk of cognitive decline or dementia. Obesity is a significant excess of body fat, or adipose tissue, oft en defined as a Body Mass Index (BMI, kg/m 2 ) greater than 30. The BMI is a simple but useful measure that allows a general estimation of adiposity; however it cannot accurately reflect the proportion of adipose tissue carried by an individual. It is likely that adiposity, rather than weight, would mediate any effect of obesity on neurocognitive health, for adipose tissue is not just a sto rage depot for fat, but is also endocrinologically and immunologically active. Adipocytes secrete various active metaboli tes that could cross the blood brain barrier (BBB) to affect brain health. Central adipose tissue, distributed around the trunk and including subcutaneous and omental adipose reserves, is known to be par ticularly active in this regard secreting nume rous a dipokines such as leptin, and cytokines ( Bastard et al., 2000 ; Bastard et al., 2006 ; Fain, 2006 ; Ingvartsen & Boisclair, 2001 ; Wellen & Hotamisligil, 2003 ) In addition, obesity is related to many other vasc ular and metabolic factors implicated in AD pathology, including insulin resistance and hypertension It is important to discover the age/s at which obesity might have its greatest impact on neurocognitive health since, according to the lifecourse approa ch ( Kuh & Ben Shlomo, 2004 ) the age at which an exposure is encountered can have a significant effect on outcome. There may be some evidence for modifiable exposures early in life contributing to late life dementia

PAGE 29

14 risk. For example, some studies have investigated the association between childh ood obesity and adult cognitive function. ( Lupien et al., 2000 ; Miller et al., 2009 ) It is beyond the scope of this paper to address all the evidence across the lifespan. We will therefore focus this review on the risk of adult obesity for risk of cognitive decline and dementia in late life, where the dementia ( Sperling et al., 2011 ) Among adults, it remains important to account for the age at which obesity was measured, the measure of adiposity used, and the age of cognitive outcomes ( Gustafson, 2008 ; Whitmer et al., 2007 ) These are important for two main re asons: 1) because of the effects of aging on body composition, and 2) because of the tendency for weight loss in clinical and subclinical dementia. Firstly, a dvancing age commonly brings significant changes in body composition, even if overall weight does not change. Muscle mass often declines, and adipose tissue often increases with increasing age ( Miller & Wolfe, 2008 ) For this reason, BMI can be a poor estimate of adiposity for older adults, as it cannot differentiate fat free mass from ad ipose tissue. Thus studies measuring the association between BMI in older adults and dementia cannot rule out the possibility that someone who has a healthy BMI is actually carrying a relatively large fat mass, having lost significant bone density and musc le mass. Similarly, studies comparing BMI across multiple ages cannot address the potential for loss of fat free mass with advancing age. This may be a significant confounding factor for studies that rely on BMI alone, particularly among older adults. Seco ndly, measurement of the association between weight and cognitive function among older adults can be complicated by the tendency of persons with dementia to lose significant amounts of weight as part of the disease ( Wirth, Bauer, & Sieber, 2007 ) This observation has been reported in many observational and clinical studies of dementia ( Berlinger

PAGE 30

15 & Potter, 1991 ; Gao et al., 2011 ; Johnson, Wilkins, & Morris, 2006 ; Renvall, Spind ler, Nichols, & Ramsdell, 1993 ) Weight loss may occur shortly before dementia diagnosis ( Johnson et al., 2006 ) so that on clinical presentation, the person wi th dementia is more likely to be underweight than overweight ( Gorospe & Dave, 2007 ) The weight loss that occurs in dementia could possibly occur because of neur al damage to appetite and self regulation regi ons of the brain dysregulated circadian rhythm or dysre gulated behavior more generally. As a result, cross sectional studies of weight and dementia, or longitudinal studies among older adults with short follo w up periods, could give the impression that persons who are overweight or obese in old age have lower dementia risk ( Dahl, Lopponen, Isoaho, Berg, & Kivela, 2008a ) no matter what measure of adiposity was used. While a number of other reviews have indeed taken these factors into account ( Anstey, Cherbuin, Budge, & Young, 2011 ; Dahl & Hassing, 2012 ; Gorospe & Dave, 2007 ; Luchsinger & Gustafson, 2009 ; Naderali, Ratcliffe, & Dale, 2009 ; Yen, 2005 ) this review differs from other recent reviews in that it is a systematic review that attempts to provide a comprehensive snapsho t of all longitudinal studies on weight, adiposity and adult cognitive health outcomes available through MEDLINE and PsychINFO. Other reviews have limited their searches to studies of AD only studies with specific follow up periods ( Beydoun, Beydoun, & Wang, 2008 ; Dahl & Hassing, 2012 ) only studies that included at least 2 cognitive domains rather than diagnoses of dementia or global screening instruments ( van den Berg, Kloppenborg, Kessels, Kappelle, & Biessels, 2009 ) ), used either Medline or PubMed but not both ( Beydoun et al., 2008 ; van den Berg et al., 2009 ) or were not systematic reviews ( Gustafson, 2006 ; Luchsinger, Patel, Tang, Schupf, & Mayeux, 2008 ; Naderali e t al., 2009 ; Whitmer, 2007 )

PAGE 31

16 1.3.2 M ethods Literature S earch T erms and S tudy I nclusion C riteria English language articles focused on weight or adiposity and cognition or dementia outcomes were identified us ing MEDLINE and PsycINFO. The following search terms were used cognitive impairment, cognition, cognitive function, and cognitive health, as well as obesity, o verweight, weight, fat, adiposity, central obesity, visceral obesity, visceral adiposity, waist hip ratio, and waist circumference. All relevant articles published up until January 30 th 2013, and retrievable by university library search or interlibrary loan were considered for this systematic review. Reference lists of all potentially eligible articles were reviewed to ensure inclusion of all relevant literature. To be included in this review, the articles had to meet the following eligibility criteria : empirical articles that are available via university libraries or interlibrary loan, written in English, and included the weight/adiposity and dementia/cognition related search terms above within the title, abstract, and/or keywords. Studies of weight i n childhood or adolescence were considered eligible if they included adult cognitive outcomes. Studies of change in weight or weight outcomes in a population that began with dementia at baseline were excluded. Similarly studies focused on health outcomes f or other cognitively impaired persons with a medical or psychiatric diagnosis known to cause cognitive impairment, such as developmental disability, schizophrenia, bipolar disorder or traumatic brain injury, were excluded. Studies with a specific focus on eating disordered populations were also excluded, as were other studies focused on specific medical or psychiatric populations. Dissertations, reviews, opinions theoretical papers or editorials were excluded from this review.

PAGE 32

17 Article S election and A bstra ction A three step process guided assessment and selection of articles. First, the study author reviewed the titles and abstracts of all potential articles retrieved by the search terms, identifying the set of article abstracts that potentially matched th e eligibility criteria. Second, the study author reviewed in depth the abstracts and full articles for studies whose abstracts passed the first review for inclusion. Information from the full articles were entered into summary tables. From this set the author identified the set of full articles which matched the eligibility criteria below. Finally, reference lists of eligible articles were reviewed for additional relevant articles to potentially include. These articles were also assessed for eligibility through a two step process.

PAGE 33

18 Figure 1.3 1 Steps of the systematic review process for longitudinal studies 1.3.3 R esults Number of A rticles I ncluded in R eview A total of 4320 articles were identified using t he search terms. Of these articles, 4085 were excluded after a preliminary review of title and/or abstract because they were not relevant and/or did not meet the inclusion criteria (e.g. topic was relevant but the article was an editorial). The remaining 235 full articles were then reviewed and abstracted by the study author. Of these 235 articles 64 observational studies were considered eligible after more thorough review, and were therefore included in this study. Included articles were then categor ized according to study population (persons with dementia, cognitive impairment, or a study of cognitive function more generally. Articles were further categorized by whether their

PAGE 34

19 results supported or opposed a link between adiposity and cognitive healt h outcomes, and age at which weight and cognitive outcomes were measured. A flow chart of the sorting and inclusion process can be seen in Figure 1.3 1 Table 1.3 1 Summary of results for the obesity and cognitive function literature review FOR AN ASSOCIATION AGAINST AN ASSOCIATION Midlife Older adults Midlife Older adults Dementia 61% (11) 39% (7) 20% (4) 80% (16) MCI Na na 0% 100% (5) Cognitive Function 7 1% (10) 29% (4) 31% (4) 69% (9) Central obesity 55% (6) 45% (5) 17% (2) 83% (10) Articles A ssessing D ementia R isk Among the 38 longitudinal studies in which dementia was the principle cognitive outcome, 18 reported evidence supporting an association be tween obesity and increased risk dementia. Of these, 15 reported specific results for the diagnosis of AD. In contrast 20 studies reported no association between obesity and dementia. Of these, 9 reported specific results for a diagnosis of AD. The re sults are shown in Table 1.3 2 and Table 1.3 3 Supporting an association between obesity and dementia : Among the 18 studies finding a significant association between high er weight/adiposity and increased dementia risk, 11 (61%) reported an association between midlife weight measures and increased dementia risk ( Beydoun & Beason Held, 2008 ; Chiang et al., 2007 ; Fitzpatrick et al., 2009 ; Gelber et al., 2012 ; Gustafson et al., 2009 ; Hassing et al., 2009 ; Kivipelto et al., 2005 ; Rosengren, Skoog, Gustafson, & Wilhelmsen, 2005 ; Whitmer et al., 2005 ; Whitmer et al., 2007 ; Whitmer et al., 2008 ) while 7 (39%) reported that increased weight ( Buchman et al., 2005 ; Gustafson, Rothenberg, Blennow, Steen, & Skoog, 2003 ; Hayden et al., 2006a ; Kerwin et al., 2011 ; Luchsinger, Cheng, Tang, Schupf, & Mayeux, 2012 ; Luchsinger, Patel, Tang, Schupf, & Mayeux, 2007 ; Xu et al., 2011 ) Fifteen of these studies report ed outcomes for

PAGE 35

20 AD specifically All 18 of these studies measured BMI. Only 7 reported BMI with other measures of adiposity such as WHR, WC or percent fat mass. Opposing an association between obesity and dementia: Twenty longitudinal studies reported evidence that did not support an associ ation between obesity and dementia. Of these studies only 4 (20%) report ed adiposity in midlife, while 16 (80%) ass essed adiposity in older adults Only 9 of these studies addressed AD specifically (refs). Among t hese studies, 13 measured only BMI. Only 6 reported BMI with other measures of adiposity such as WHR, WC or percent fat mass.

PAGE 36

21 Table 1.3 2 Longitudinal studies supporting an association between overweight and dementia STUDIES OF MIDLIFE ADIPOSIT Y Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Beydoun & Beason Held, 2008 ) 2,322 median 23.4 BMI, WC Dx AD, NINCDS ADRDA criteria. Midlife (30, 40 or 45 years): Men: being underweight (BMI 90th percentile) between age 30 a nd 50 years increased AD risk (HR = 3.70, 95% CI: 1.43, 9.56). ( Chiang et al., 2007 ) 157 demented cases 628 matched controls Age 30 and older (1982 1992) 8 20 ye ars Nested case control study BMI (Chinese criteria) Dx AD, VaD, Chinese version of DSM IV A J shaped relationship was observed between BMI and dementia. Compared to BMI 20.5 22.9 odds (OR) for developing dementia was: 1.84 (1.02 3.33) for BMI <20. 5, 1.87 (1.08 3.23) for BMI 23.0 25.4, and 2.44 (1.39 4.28) for BMI >/=25 Similar findings were observed for AD and VaD.

PAGE 37

22 Table 1.3 2 Longitudinal studies supporting an association between overweight and dementia STUDIES OF MIDLIFE ADIPOSIT Y Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Fitzpatric k et al., 2009 ) 2798 74 .7 (but baseline BMI based on self report of weight at age 50) 5.4 mean (but much more when compared to self reported BMI at age 50) BMI AD dx: NINCDS ADRDA criteria; VaD dx: Disease and Treatment Centers criteria Midlife obesity inc reased risk of dementia compared to BMI 20 25 healthy weight group ( HR 1.39; 95% CI 1.03 1.87). Reversed in assessments of late life BMI: Underweight persons (BMI 20) had an increased risk of dementia (HR 1.62; 1.02 2.64), being overweight (BMI25 3 0) was not associated with risk of dementia (0.92; 0.72 1.18) being obese reduced the risk of dementia (0.63; 0.44 0.91). ( Gelber et al., 2012 ) 3468 52 25 28 BMI Dementia dx: DSM III R criteria; AD dx: NINCDS ADRDA criteria; VaD dx: Disease and Treatment Centers criteria Compared to BMI <22.6, being overweight or obese (BMI > 25.0) was associated with greater risk of dementia (OR = 1.87, 95% CI = 1.26 2.77) ( Gustafso n et al., 2009 ) 1462 38 60 32 BMI, WC, WHR Dementia dx: DSM III R criteria; AD dx: NINCDS ADRDA criteria; VaD dx: NINDS AIREN cr iteria Logistic models showed that a midlife WHR greater than 0.80 more than doubled dementia risk (OR 2.22, 95% CI: 1.00 4.94, p=0.049). Cox models showed no association. ( Hassing et al., 2009 ) 1152 45 65 years (mean 52.5) up to 40 BMI AD dx: NINCDS ADRDA criteria; VaD dx: NINDS AIREN criteria Overweight in midlife had an elevated risk of dementia (OR=1.59; 95% CI: 1.21 2.07), AD (OR=1.71; 95% CI: 1.24 2.35), VaD (OR=1.55; 95% CI: 0.98 2.47).

PAGE 38

23 Table 1.3 2 Longitudinal studies supporting an association between overweight and dementia STUDIES OF MIDLIFE ADIPOSIT Y Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Kivipelto et al., 2005 ) 1449 50.6 Mean 21 BMI Dx dementia, AD: DSM IV criteria; AD dx: NINCDS ADRDA criteria Midlife obesity was associated with late life dementia after adjusting for vascular factors (OR = 1.9, 95% CI: 1.0 4.6). ( Rosengre n et al., 2005 ) 7402 47 5 5 25 28 BMI Dx Dementia, AD: Death register and hospital discharge diagnoses J shaped curve. BMI less than 20 in midlife was associated with increased risk of primary hospital diagnosis of dementia in late life. with increased risk of a primary hospital diagnosis of dementia. ( Whitmer et al., 2005 ) 10,276 40 45 21 39 BMI, skinfold thickness Dementia dx: ICD 9 co des Compared with healthy weight: Obesity in midlife (BMI >= 30) increased risk of dementia 74% (HR 1.74, 95% CI 1.34 to 2.26). Overweight in midlife (BMI 25.0 29.9) increased risk of dementia 35% (1.35, 1.14 to 1.60). The highest quintile of skinf old thickness had a 72% greater risk of dementia than the lowest quintile (1.72, 1.36 to 2.18, and 1.59, 1.24 to 2.04). ( Whitmer et al., 2007 ) 10,136 40 45 26 42 BMI Review of medical records from Neurology visits Dx VaD, AD Obesity in midlife increased risk of AD (adjHR=3.10, 95% CI 2.19 4.38), and risk of VaD (adjHR=5.01, 95% CI 2.98 8.43) Overweight in midlife increased risk of AD (adj HR=2.09, 95% CI 1.69 2.60) and VaD (HR=1.95, 95% CI 1.29 2.96 for VaD).

PAGE 39

24 Table 1.3 2 Longitudinal studies supporting an association between overweight and dementia STUDIES OF MIDLIFE ADIPOSIT Y Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Whitmer et al., 2008 ) 6583 40 45 26 42? BMI, Sagittal Abdominal Diameter (SAD) ICD 9 codes Person s in the highest quintile of SAD in midilfe had increased risk of dementia (HR, 2.72; 95% CI, 2.33 3.33) compared to persons in the lowest quintile. Those with high SAD (>25 cm) but healthy BMI had an increased risk (HR, 1.89; 95% CI, 0.98 3.81) vs. thos e with low SAD (<25 cm) and healthy BMI. Persons who were both obese and with high SAD had the highest risk of dementia (HR, 3.60; 95% CI, 2.85 4.55).

PAGE 40

25 Table 1 .3.2 Longitudinal studies supporting an association between overweight and dementia. STUDI ES OF ADIPOSITY IN OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Buchman et al., 2005 ) 820 As above up to 10 (mean 5.5) BMI AD dx: NINCDS ADRDA criteria A 1 point annual decline in BMI was associated with increased risk of AD compared with persons with no change in BMI (HR 0.730 ; 95% CI 0.625 to 0.852). ( Gustafso n et al., 2003 ) 392 70 18 BMI Dementia dx: DSM III R criteria; AD dx: NINCDS ADRDA criteria; VaD dx: NINDS AIREN criteria. For every 1.0 increase in BMI at age 70 yea rs, AD risk increased by 36% (Hazard Ratio, 95% CI). These associations were not found in men. ( Hayden et al., 2006b ) 3264 3.2 BMI AD dx using NINCDS ADRDA criteria; VaD dx using NINDS AIREN criteria Obesity increased the risk of AD in females (adjHR 2.23, 95% CI 1.09 4.30) but not males. ( Kerwin et al., 2011 ) 7163 65 to 80 up to 8 average 4.4 WHR, BMI 3MSE Dementia Dx Central obesity was what mattered. Women with a WHR 24.9 had a greater risk of cognitive impairment and probable dementia than more obe se women or women with a WHR less than 0.80. However women with a WHR less than 0.80 and a BMI of 20.0 to 24.9 kg/m2 had poorer scores on cognitive assessments. ( Luchsinge r et al., 2007 ) 893 BMI, 907 WC, 709 weight 5 BMI, WC, weight Dementia dx: DSM IV criteria; AD dx: NINCDS ADRDA criteria In persons <76 years the association between BMI and dementia resembled a U shape, meaning that both low BMI and high BMI increased risk of dementia. BMI. In persons <76 years, the highest quartile of WC correlated to dementia and AD risk. Weight loss was related to a higher risk of dementia and dementia assoc iated with stroke DAS). Weight gain was related to a higher DAS risk only.

PAGE 41

26 Table 1 .3.2 Longitudinal studies supporting an association between overweight and dementia. STUDI ES OF ADIPOSITY IN OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Luchsinge r et al., 2012 ) 1459 7 BMI, WHR, WC AD dx NINCDS ADRDA criteria On ly higher WHR was related to higher AD risk, with the highest quartile of WHR increasing risk 2.5 times compared to the lowest quartile (HR = 2.5; 95% CI = 1.3 4.7). ( Xu et al., 2011 ) 8,534 43.4 mean ~30 BMI MMSE, CERAD, Memory in Reality test Dx dementia, VaD, AD Higher midlife BMI was associated with an increased risk of dementia (OR 1.08, 95% CI 1.03 1.14)

PAGE 42

27 Table 1.3 3 Longitudinal studies not supporting an a ssociation between overweight and dementia. STUDIES OF MIDLIFE ADIPOSITY Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Chen et al., 2010 ) 286 20s & 40s ~50yr case control study BMI MMSE dx AD: NINCDS ADRDA criteria; Dx VaD: NINDS AIREN criteria Men and women with low BMI in midlife had inc reased risk of AD (OR = 2.62 3.97, 95% CI ) and VaD (OR = 6.23 11.11) compared with those with healthy BMI. High BMI in midlife was associated with increased VaD risk (OR = 15.29 and 10.32) among women. ( Rosengre n et al., 2005 ) 7402 47 55 25 28 BMI Dementia dx: Death register and hospital discharge diagnoses J shaped curve. BMI less than 20 in midlife was associated with increased risk of primary hospi tal diagnosis of dementia in late life. of a primary hospital diagnosis of dementia. ( Stewart et al., 2 005 ) 1890 46 68 32 Weight Dementia dx: DSM III R15 Criteria; AD dx: NINCDS ADRDA criteria; dx VaD: California Alzheimer Disease and Treatment Centers criteria Groups that developed and did not develop dementia did not differ with respect to baseli ne weight or change in weight from mid to late life. ( Strand et al., 2013 ) 48,793 35 50 31 35 BMI Dx dementia, AD: Norwegian Cause of Death Registry Low BMI (<20 vs. BMI 20 25) in midlife was associated with increased risk of dementia death (HR = 1.76, 95% CI 1.15 2.68).

PAGE 43

28 Table 1 .3.3 Longitudinal studies not supporting an association between overweight and dementia. STUDIES OF ADIPOSITY IN OLDER ADULTS Aut hor (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Abellan van Kan et al., 2012 ) 647 7 Total fat mass (DXA) SPMSQ Total fat mass was not significantly associated with dementia risk. ( Atti et al., 2008 ) 646 9 BMI MMSE; Dementia dx DSM II I criteria than those with a BMI of 20.0 to 24.9 (HR = 0.75, 95% (CI) = 0.59 0.96), ( Barrett Conno r, Edelstein, Corey Bloom, & Wiederholt, 1996 ) 299 50 79 16 21 Baseline 1972 74; Dementia diagnosis visit 1990 93 Weight Dx AD neurological test scores, physical examination Men who developed AD weighed slightly more at baseline (1972 74) than men who remained cognitively intact ( P = .03); baseline weights did not differ in women by dementia status ( P = .54). Both men and women who were later diagnosed with AD had decreasing weight measured across the three time points ( P < .001 for men and P < .003 for women), whereas men and women who were diagnosed as cognitively intact had no significant change in their weights ( Dahl et al., 2008a ; Dahl, Lopponen, Isoaho, Berg, & Kivela, 2008b ) 605 65 92 8 BMI Dementia dx DSM IV criteria. Women with high BMI scores had a lower dementia risk (HR = 0.90, 95% Cl = 0.84 0.96). Trend for men with high BMI scores to have a lower dementia risk, (HR = 0.95, 95% Cl = 0.84 1.07). ( Forti et al., 2010 ) 749 3 5 Metabolic Syndrome (MetS) WC MMSE Dx VaD, AD In participants aged 75 and older, a bdominal obesity was associated with a lower risk of overall dementia (HR = 0.53, 95% CI = 0.28 0.98).

PAGE 44

29 Table 1 .3.3 Longitudinal studies not supporting an association between overweight and dementia. STUDIES OF ADIPOSITY IN OLDER ADULTS Aut hor (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Gao et al., 2011 ) 1,331 mean 6.4 BMI dx dementia using ICD 10 and DSM IIIR Greater decline in BMI was associated with greater risk of dementia or MCI (p = .02 for dementia, p = .04 for MCI). BMI in participants with incident dementia, MCI, and normal cognition did not differ 9 12 years before diagnosis. Six years before diagnosis, participants with incident dementia or MCI had significantly lower BMI than participants with normal cognition. ( Han et al., 2009 ) 721 60 85 2 BMI, WHR, WC, PBF CERAD Korean version Dementia dx using DSM IV criteria The change in cognitive function in the elderly was associated with the baseline assessment of BMI, WC, and % body fat. Men who were obese (WHR, BMI) at baseline and subsequentl y increased weight had improved cognitive function. Women with high WHR at baseline and a subsequent decrease in adiposity had increased risk of cognitive decline. Women with normal WC at baseline and subsequent increased adiposity also had increased ris k of cognitive decline. ( Hughes, Borenstein, Schofield, Wu, & Larson, 2009 ) 1,478 mean 71.8 mean 7.8 BMI dx VaD, AD, DSM IV criteria Higher baseline BMI was sign ificantly associated with a reduced risk of AD (adj HR] = 0.56, 95% [CI] = 0.33 0.97). Slower rate of decline in BMI was associated with a reduced risk of dementia (HR = 0.37, 95% CI = 0.14 0.98), with the association stronger for those who were overweig ht or obese (HR=0.18, 95% CI=0.05 0.58) compared to normal or underweight (HR=1.00, 95% CI=0.18 5.66) at baseline.

PAGE 45

30 Table 1 .3.3 Longitudinal studies not supporting an association between overweight and dementia. STUDIES OF ADIPOSITY IN OLDER ADULTS Aut hor (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Johnson et al., 2006 ) 449 65 95 Mean 6 Weigh t AD dx Clinical Dementia Rating Scale Among older adults, rate of weight loss doubled about 1 year before dementia diagnosis among individuals who developed AD. ( Knopman, Edland, Cha, Petersen, & Rocca, 2007 ) 481 At time of onset, 6.1% under age 70 years, 75.6% ages 70 to 89 years, and 18.2% age 90 years or older Case Control ~30 years Weight DSM IV criteria applied to medical records There were no differen ces in weight between cases and controls 21 to 30 years prior to the onset of dementia. However starting from 11 20 years prior to the index year, women with dementia had lower weight than controls, and the difference increased over time. There was a trend for increasing risk of dementia with decreasing weight in women 9 to 10 years before the index year (p = 0.001). ( Nourhashe mi et al., 2003 ) 3646 8 BMI MMSE Dx Dementia, AD by DSM III R criteria The risk of dementia was highest for those with a BMI <21 (RR=1.48, CI=95%: 1.08 2.04) and those with a BMI of 21 22 (RR 1.072, CI=95%: 0.759 1.514) compared with BMI between 23 26. However relationship di sappeared after excluding people who developed dementia early in the study. ( Ogunniyi et al., 2011 ) 1559 5.97 BMI CERAD, Clinician Home based Interview, and Cambridge Examination for Mental Disorders A significantly greater decline in BMI was found in those with either incident dementia (p < 0.001) or incident MCI (p < 0.001) compared to healthy subjects ( Power et al., 2011 ) 12,047 65 84 (mean 72.1) 9.7 BMI, WC, WHR ICD 9 & ICD 10 dx codes in Western Australia Data Linkage System Overweight men and those with risk of dementia than men with normal weight and with WHR < 0.9. Higher adiposity was not associated with incident dementia.

PAGE 46

31 Table 1 .3.3 Longitudinal studies not supporting an association between overweight and dementia. STUDIES OF ADIPOSITY IN OLDER ADULTS Aut hor (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Raffaitin et al., 2009 ) 2049 74 4 WC MMSE, Benton Visual Set Test Dx VaD, AD Participants with high WC had decreased risk for all cause dementia (HR = 0.86, 95% CI: 0.64 1.17), AD (HR=0.63, 95% CI: 0.43 0.94) and VaD (HR=0.82, 95% CI: 0.4 1 1.66) ( West & Haan, 2009 ) 1,351 60 101 (mean 69.9) 8 ( 5.6) BMI, WC 3MSE and DelRec dx dementia by DSM III Compared with the lowest BMI category, overweight p articipants had a 48% decreased rate of dementia or being cognitively impaired not dementia (CIND) (adj [HR] = 0.52, 95% [CI]: 0.30 0.91) and obese participants had a 61% decreased rate of dementia/CIND (HR = 0.39, 95% CI: 0.20 0.78). By contrast, the mi ddle and high tertiles of WC were associated with higher rates of dementia/CIND compared with the low tertile, (adj HR = 1.8, 95% CI: 1.1 3.1, and adj HR = 1.9, 95% CI: 0.91 3.8 respectively). ( Xiong, Plassman, Helms, & Steffens, 2006 ) 166 65.81 12 BMI Dementia dx: TICS m and clinical assessments Cognitive change was not significantly different between twins discordant for BMI.

PAGE 47

32 Articles A ssessing Mild Cogniti ve Impairment Among the 5 longitudinal studies in which MCI was the principle cognitive outcome, none reported evidence supporting an association between obesity and increased risk MCI. In contrast all 5 reported no association between obesity and MCI. Furthermore, all 5 studies (100%) were assessed among older adults, not midlife. Results are displayed in Table 1.3 4 below.

PAGE 48

33 Table 1.3 4 Lon gitudnial studies no t supporting an association between overweight and MCI. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Assessment MCI Assessment Results ( Gao et al., 2011 ) 1331 75.8 6.4 mean BMI CERAD, Clinician Home based Interview, and Cambridge Examination for Mental Disorders BMI in participants with incident MCI, did not differ 12 or 9 years before diagnosis, but 6 years before diagnosis, partici pants with incident MCI had significantly lower BMI than participants with normal cognition (p = .006). Participants with incident MCI had greater decline in BMI than those without (P = .04). ( Kerwin et al., 2011 ) 7163 65 80 4.4 mean BMI, WHR 3MSE Association between BMI and risk of CI was modified by body fat distribution (WHR). Older women with a lower WHR had greater risk of CI at both low and high BMI categ ories. Underweight women with a WHR less than 0.80 had a greater risk than those with higher BMI. In normal weight to overweight women (BMI 20.0 29.9), central adiposity (WHR>0.80) is associated with greater risk of cognitive impairment and probable dement ia than in women with higher BMI. ( Newman et al., 2009 ) 1677 77 102 Median 85 13 WC, BMI Modified MMSE, DSST, CES D, CI based on <80 on 3MS Greater weight was not associated with higher rates of cognitive impairment. There was no significant association between WC and cognitive impairment. ( Ogunniyi et al., 2011 ) 155 9 5.97 mean BMI CERAD, Clinician Home based Interview, Cambridge Examination for Mental Disorders A significantly greater decline in BMI was found in those later diagnosed with MCI (p < 0.001) compared to normal subjects.

PAGE 49

34 Table 1.3 4 Lon gitudnial studies no t supporting an association between overweight and MCI. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Assessment MCI Assessment Results ( Sachs Ericsson, Sawyer, Corsentino, Collins, & Blazer, 2010 ) 2840 71.1(5.3) 4 times over 10 years BMI SPMSQ Lower BMI was a predictor of CI for those with the APOE allele. For non APOE 4 allele carriers, BMI was unrelated to CI.

PAGE 50

35 Articles A ssessing C ognitive F unction Among the 26 articles assessing cognitive function outcomes other than MCI or dementia one article reported both evidence that could be used to support and evidence that could be used to oppose an association ( Kanaya et al., 2009 ) so that in total 14 studies were included in the table supporting an associatio n between obesit y and dementia, while 13 articles were included in the table opposing this association. The articles assessing the association between weight or adiposity and cognitive function tested a variety of cognitive domains, including memory, executive function, and attention, which are recorded in Table 1.3 5 and Table 1.3 6 below Supporting an association between obesity and cognitive function : Fourteen longitudi nal studies re ported evidence for an association between obesity and declines in various cognitive functions Of these, 10 (71%) found that obesity in midlife increased risk of cognitive decline, while 4 (29%) reported that obesity in older adults increased risk of cog nitive decline. Of the studies supporting a link, all 14 used BMI as an estimate of adiposity, only 4 combined BMI with waist circumfere nce or waist hip ratio (WHR). Opposing an association between obesity and cognitive function : Thirteen longitudinal s tudies reported no evidence of an association between any measure of adiposity and cognitive decline. Four (31%) of these studies examined adiposity in midlife and 9 (69%) were among older ad ults Among the studies not supporting an association, 11 used B MI to estimate obesity, while six used BMI combined with other measures such as WHR.

PAGE 51

36 Table 1.3 5 Longitudinal studies supporting an association between overweight and cognitive decline. MIDLIFE Au thor (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Risk ratio Hazard ratio, Odds ratio ( Albanese et al., 2012 ) 2083 Began at 15, 20, 26, 36, 43, a nd 53 years; 38 BMI Animal naming, National Adult Reading Test, word list recall, letter cancellation Midlife BMI gain from youth was inversely associated with memory in midlife. BMI gain from youth to age 53 years in men was independently associated with better memory. Average BMI was approx. 27 at age 53. Both underweight and obese women at 53 years had significantly lower memory scores, cross sectionally. ( Cour not et al., 2006a ) 2223 32 62 5 BMI Word list learning, Digit Symbol Substitution Test, WAIS, selective attention test, delayed free recall test Cross sectionally, BMI was associated with lower cognitive scores. A higher BMI at baseline was also assoc iated with a higher decline in word list learning (delayed recall) at follow up. No significant association was found between changes in BMI and cognitive function. ( Dahl et al., 2010 ) 781 25 63 (mean 41.6) 16 BMI Information Synonyms, Analogies, Figure Logic, Kohs Block Design, Card Rotations, Digit Span (Forward and Backward), Thurstone's Picture Memory, Names and Faces, (Immediate and Delayed), Digit Symbol, Fi gure Identification Higher midlife BMI scores preceded lower general cognitive ability and steeper cognitive decline in both men and women. ( Dahl et al., 2012 ) 657 (mean 39.9) 25 BMI Test battery representing four domains: verbal, spatial/fluid, memory, and perceptual sp eed Being overweight or obese in midlife was associated with cognitive decline later in life. Weight decline across midlife rather than low weight in late midlife per se was associated with cognitive decline

PAGE 52

37 Table 1.3 5 Longitudinal studies supporting an association between overweight and cognitive decline. MIDLIFE Au thor (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Risk ratio Hazard ratio, Odds ratio ( Debette et al., 2011 ) 1352 54 +/ 9 6.3 +/ 1.1 BMI, WHR logical memory delayed recall, visual reproductions delayed recall (VR d), and Trail Making Test B A (TrB A) Midlife obesity was associated with an increased rate of progression of decline in executive function a decade later. Large WHR in midlife was associated with marked decline in total brain volume. ( Hass ing, Dahl, Pedersen, & Johansson, 2010 ) 417 50 60 30 BMI long term memory, short term memory, speed, verbal and spatial ability Midlife overweight is related to lower overall cognitive function in old age. ( Laitala et al., 2011 ) 2606 twins Midlife Old age BMI Validated telephone interview Midlife overweight increased the risk for mild impairment of cognitive function. Weight gain more than 1.7 kg/m and loss more than 2 kg/m within an average of 5.6 years were associated with lower cognitive performance independently of BMI. ( Sabia, Kivimaki, Shipley, Marmot, & Singh Ma noux, 2009 ) 10,308 35 55 21 BMI MMSE, AH4 I, inductive reasoning, phonemic fluency Long term obesity and long term underweight in adulthood are associated with lower cognitive scores in late midlife.

PAGE 53

38 Table 1.3 5 Longitudinal studies supporting an association between overweight and cognitive decline. MIDLIFE Au thor (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Risk ratio Hazard ratio, Odds ratio ( Singh Manoux et al., 2012 ) 6401 39 63 (mean 48.9 50.0 10 BMI Memory, reasoning, semantic, and phonemic fluency In the metabolically abnormal group, the 10 year decline on the global cognitive score was fast er among obese than among healthy weight individuals. In the metabolically normal group, the 10 year decline in the global cognitive score was similar in the normal weight, overweight, and obese groups. ( Wolf et al., 2007 ) 1814 40 69 8 12 BMI WHR Trails B, Visual Reproductions Immediate and Delayed Recall Verbal memory (immediate and delayed recall) Midlife WHR in highest quartile was significantly related to poo rer performance on executive function & visuomotor skills. Obesity was not related to verbal memory (immediate or delayed recall). Visuomotor skills but not memory were related to WHR.

PAGE 54

39 Table 1 .3.5 Longitudinal studies supporting an association bet ween overweight and cognitive decline. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Risk ratio Hazard ratio, Odds ratio ( Brubache r, Monsch, & Stahelin, 2004 ) 531 59.4(7.9) 10 BMI CERAD NAB battery This study demonstrates that for both BMI loss and BMI gain there is a worsening of cognition. Odds ratio is greater at the extremes of weight loss or weight gain, up to OR=4 for BMI loss of 0.8kg/m2/yr and up to OR = 8 for BMI gain of 0.8kg/m2/yr. ( Elias, Elias, Sullivan, Wolf, & D'Agostin o, 2003 ) 1423 55 88 18 BMI Kaplan Albert Neuropsychological Test Battery Obese men performed at a level of 0.44 s.d. below the level of non obese men for total test score (P<0.0001). Obese vs. non obese women showed no differences in total test score. ( Gunstad, Lhotsky, Wendell, Ferrucci, & Zonderma n, 2010 ) 1703 19 93 (Mean 55.5) Average 3.1 visits, average 2.0 years between visits BMI, WHR, WC MMSE, BIMC, WAIS R, CVLT, letter fluency, card ro tation Obesity was associated with poorer performance in a variety of cognitive domains, including global screening measures, memory, and verbal fluency tasks. Obesity was associated with better performance on tests of attention and visuospatial ability. An obesity by age interaction emerged in some domains, including memory, attention, executive function.

PAGE 55

40 Table 1 .3.5 Longitudinal studies supporting an association bet ween overweight and cognitive decline. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Risk ratio Hazard ratio, Odds ratio ( Kanaya et al., 2009 ) 3054 70 79 Up to 8 BMI, WC, sagi ttal diameter, total fat by DXA Subcutaneo us and visceral fat by CT 3MS In men higher levels of all adiposity measures were associated with worsening cognitive function in men. In women there was no association between adiposity and cognitive change.

PAGE 56

41 Table 1.3 6 Longitudinal studies not supporting an association between overweight and cognitive decline. MIDLIFE Author (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Results ( Arntzen, Schirmer, Wilsgaard, & Mathiesen, 2011 ) 5033 59 mean 7 BMI 12 word memory, Digit Symbol Coding, Tapping test The authors found no consistent associati on between BMI and cognitive test results. ( Knopman, Mosley, Catellier, & Coker, 2009 ) 1130 59.6 +/ 4.3 12 16? BMI Delayed Word Recall (DWR) Test, the Digit Symbol Substitution (DSS) Test, and the Word Fluency (WF) Test Baseline BMI was not a risk factor for cognitive decline ( Lo, Pachana, Byrne, Sachdev, & Woodman, 2012 ) 334 40 79 (Mean 58.72) Mean 7.45 BMI, WC, WHR MMSE, Auditory Delayed Index, Visual Delayed Index, and Working Memory Index from WMS). Processing Speed Index from WAIS. No significant associations were found between BMI, WC, or WHR and any cognitive do mains at follow up. Both weight gain and loss were associated with poor Visual Delayed Index performance at follow up compared with stable weight. ( Thilers, Macdo nald, Nilsson, & Herlitz, 2010 ) 1480 40 65 10 BMI 3MSE Spanish and English Verbal Learning Test (SEVLT) Accelerated postmenopausal cognitive decline is restricted to women with normal BMI.

PAGE 57

42 Table 1 .3.6 Longitudinal studies not supporting an associat ion between overweight and cognitive decline. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Results ( Bagger, Tan ko, Alexandersen, Qin, & Christiansen, 2004 ) 5607 63.8 mean 7.3 Weight, DXA body composition, TFM, TLM, CFM Short Blessed test Higher baseline weight was associated with lower cognitive decline. Women with the worst cognitive performance at follow up we re the ones who lost the most body weight and had the lowest central fat mass (CFM). ( Deschamps et al., 2002 ) 169 69 to 89 Mean 75.4 5 BMI ADL dependency, IADL dependency, MMSE No overweight patients (BMI >27) declined in cognitive function. Subjects with a BMI 23 27 had 3.6 times lower chance cognitive decline in the subsequent 5 y (OR=0.28, 95%, CI 0.09 0.90) than subjects with BMI less than 23. ( Driscoll et al., 2011 ) 2283 65 79 Up to 5 Mean 3 BMI, WHR, WC 3MS, PMA, Letter fluency, BVRT, CVLT, digital span, CRT, No association between weight and cognition in women who remained stable or gained weight. Cognition was not related to changes in WC. Weight loss was associated with cognitive decline, independent of initial BMI. ( Han et al., 2009 ) 721 60 85 2.13 BMI, WHR, WC, and % body fat CERAD Korean version For men obese at baseline, increasing adiposity (BMI, WHR, WC) was associated with improved cognitive function. For women obese at baseline, both an increase or d ecrease in WHR or WC were associated with cognitive decline. ( Kanaya et al., 2009 ) 3054 70 79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3MS There was no association between adiposity and cognitive change in women. Higher levels of all adiposity measures were associated with worsening cognitive function in men.

PAGE 58

43 Table 1 .3.6 Longitudinal studies not supporting an associat ion between overweight and cognitive decline. OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Measures Cognitive Measures Results ( Raffaitin et al., 2011 ) 7087 65+ 4 WC MMSE, Isaacs Set Test (verbal fluency), Benton Visual Retention Test (BVRT, visual working memory) No significant change in cognition in those who started with a high WC. ( Regal & Heatherington, 2012 ) 94 71 86 1 BMI MMSE, Montreal Cognitive Assessment, Frontal Assessment Battery, and Addenbrooke Cognitive Assessment Underweight grou p had the lowest cognitive scores of the four groups in all 12 comparisons. The overweight group had the highest cognitive scores in nine of 12 comparisons. ( Stur man et al., 2008 ) 3885 65+ Up to 10 Mean 6.4 BMI MMSE, East Boston Tests of Immediate Memory and Delayed Recall and the Symbol Digit Modalities Test Among older adults, higher BMI was not predictive of cognitive decline. ( Xiong et al., 2006 ) 166 pairs of twins discordant for BMI 67 77 12 BMI >30 kg/m Modified Telephone Interview for Cognitive Status (TICS m) Cognitive change was not significantly different betw een members of pairs discordant for BMI.

PAGE 59

44 Articles M easuring C entral O besity Looking across the different studies examining cognitive function, MCI or dementia diagnosis outcomes a total of 22 studies investigated the association between central obesity and any of these cognitive health outcomes. The results of these studies can be seen in Table 1.3 7 and Table 1.3 8 Supporting an association between central obesity and cognitive health : A total of 11 studies reported an association between central obesity and cognitive outcomes. Six (55%) of these studies measured central obesity in midlife, while 5 (45%) measured central obesity in older adults. Opposing an associatio n between central obesity and cognitive health : In contrast, 12 studies reported no association between central obesity and cognitive outcomes. Only two (17%) of these studies reported the association with midlife central obesity, while 10 (83%) measure d central obesity in older adults. 1.3.4 C onclusions The existing empirical literature on weight and cognitive function reveals a mixture of results that support an association between increased weight and cognitive decline or dementia, and studies that do not. Among the studies that do not support this association, many report that low weight or weight loss increases risk. Some others report that higher weight may even be protective, decreasing risk of dementia or cognitive decline. However o verall, it would app ear that the mixed findings are at least partly due to differences between midlife and late life measures of adiposity. The majority of negative findings measured baseline adiposity among re measured adiposity among adults in midlife. Research has indicated that weight loss is a common feature of dementia, and may precede dementia diagnosis by 6 years or more ( Gao et al., 2011 ) It is therefore possible

PAGE 60

45 that the baseline weight among older adults already reflected some of these changes. Upon comparison of the studies of older adult weight that were for or against the association there did not appear to b e a systematic difference in follow up time, but there may be other systematic differences not accounted for in this study.

PAGE 61

46 Table 1.3 7 Longitudinal studies measuring central obesity that supported an ass ociation with any cognitive outcomes. STUDIES OF MIDLIFE CENTRAL ADIPOSITY Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Beydoun & Beason Held, 2008 ) 2,322 23.4 years BMI, WC Dx AD, NINCDS ADRDA criteria Among men, being underweight (BMI /=30) at age 30, 40, or 45 years and jointly centrally obese (waist circumference >/= 80th percentile) at age 30, 35, or 50 years increased AD risk (HR = 6.57, 95% CI: 1.96, 22.02). Women who lost weight (BMI change <10th percentile) between ages 30 and 45 years were also at increased risk ( HR = 2.02, 95% CI: 1.06, 3.85). Weight gain among men (BMI change >90th percentile) between age 30 and 50 years increased AD risk (HR = 3.70, 95% CI: 1.43, 9.56). ( Debette et al., 2011 ) 1352 54 +/ 9 6.3 +/ 1.1 BMI, WHR logical memory delayed recall, visual reproductions delayed recall (VR d), and Trail Making Test B A (TrB A) Midlife obesity was associated with an increased rate of progression of decline in ex ecutive function a decade later. Large WHR in midlife was associated with marked decline in total brain volume. ( Gustafson et al., 2009 ) 1462 38 60 32 BMI, WC WHR Dementia dx: DSM III R criteria; AD dx: NINCDS ADRDA criteria; VaD dx: NINDS AIREN criteria While Cox models showed no association between baseline anthropometric factors and dementia risk, logistic models showed that a midlife WHR greater than 0.80 increased risk for dementia approximately twofold (odds ratio 2.22, 95% confidence interval 1.00 4.94, p=0.049) among surviving participants.

PAGE 62

47 Table 1 .3.7 Longitudinal studies measuring central obesity that supported an association with any cognitive o utcomes. STUDIES OF MIDLIFE CENTRAL ADIPOSITY Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Whitmer et al., 2005 ) 10,276 40 45 21 39 BMI, skinfold thickness Dementia dx: ICD 9 codes Compared with healthy weight: Obesity in midlife (BMI >= 30) increased risk of dementia 74% (HR 1.74, 95% CI 1.34 to 2.26). Overweight in midlife (BM I 25.0 29.9) increased risk of dementia 35% (1.35, 1.14 to 1.60). The highest quintile of skinfold thickness had a 72% greater risk of dementia than the lowest quintile (1.72, 1.36 to 2.18, and 1.59, 1.24 to 2.04). ( Whitmer et al., 2008 ) 6583 40 45 ~26 42 BMI, Sagittal Abdominal Diameter (SAD) ICD 9 codes Persons in the highest quintile of SAD in midlife had increased risk of dementia (HR, 2.72; 95% CI, 2.33 3. 33) compared to persons in the lowest quintile. Those with high SAD (>25 cm) but healthy BMI had an increased risk (HR, 1.89; 95% CI, 0.98 3.81) vs. those with low SAD (<25 cm) and healthy BMI. Persons who were both obese and with high SAD had the highes t risk of dementia (HR, 3.60; 95% CI, 2.85 4.55). ( Wolf et al., 2007 ) 1814 40 69 8 12 BMI WHR Trails B, Visual Reproductions Immediate and Delayed Recall, Verbal memory (immediate and delayed recall) Midlife WHR in highest quartile was significantly related to poorer performance on executive function & visuomotor skills. Obesity was not related to verbal memory (immediate or delayed recall). Visuomotor skills bu t not memory were related to WHR.

PAGE 63

48 Table 1 .3.7 Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes. STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS Author (date) N Baseline Age Follow up (years) Wei ght Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Gunstad et al., 2010 ) 1703 19 93 (Mean 55.5) Average 3.1 visits, BMI, WHR, WC MMSE, BI MC, WAIS R, CVLT, letter fluency, card rotation Obesity was associated with poorer performance in a variety of cognitive domains, including global screening measures, memory, and verbal fluency tasks. Obesity was associated with better performance on test s of attention and visuospatial ability. An obesity by age interaction emerged in some domains, including memory, attention, executive function. ( Kanaya et al., 2 009 ) 3054 70 79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3MS In men higher levels of all adiposity measures were associated with worsening cognitive function in men. In women there was no association betwe en adiposity and cognitive change. ( Kerwin et al., 2011 ) 7163 65 to 80 up to 8 average 4.4 WHR, BMI 3MSE Dementia Dx Central obesity was what mattered. Women wit of 20.0 24.9 had a greater risk of cognitive impairment and probable dementia than more obese women or women with a WHR less than 0.80. However women with a WHR less than 0.80 and a BMI of 20.0 to 24.9 kg/m2 had poorer scores on cognitive assessments.

PAGE 64

49 Table 1 .3.7 Longitudinal studies measuring central obesity that supported an association with any cognitive outcomes. STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS Author (date) N Basel ine Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Luchsinger et al., 2007 ) 893 BMI, 907 WC, 709 weight 5 BMI, WC, weight Dementia dx: DSM IV criteria; AD dx: NINCDS ADRDA criteria In persons <76 years the association between BMI and dementia resembled a U shape, meaning that both low BMI and high BMI increased risk of dementia. tia risk decreased with higher BMI. In persons <76 years, the highest quartile of WC correlated to dementia and AD risk. Weight loss was related to a higher risk of dementia and dementia associated with stroke DAS). Weight gain was related to a higher DAS risk only. ( Luchsinger et al., 2012 ) 1459 7 BMI, WHR, WC AD dx NINCDS ADRDA criteria Only higher WHR was related to higher AD risk, with the highest qu artile of WHR increasing risk 2.5 times compared to the lowest quartile (HR = 2.5; 95% CI = 1.3 4.7).

PAGE 65

50 Table 1.3 8 Longitudinal studies measuring central obesity that did not support a link between obes ity and cognitive outcomes. STUDIES OF MIDLIFE CENTRAL ADIPOSITY Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Lo et al., 2012 ) 334 40 79 (Mean 58.72) Mean 7.45 Weight, WC, WHR MMSE, Auditory Delayed Index, Visual Delayed Index, and Working Memory Index from WMS), Processing Speed Index from WAIS No significant associations were found betwe en BMI, WC, or WHR and any cognitive domains at follow up. Both weight gain and loss were associated with poor Visual Delayed Index performance at follow up compared with stable weight. ( Newman et al., 2009 ) 1677 77 102 Median 85 13 WC, BMI Modified MMSE, DSST, CES D, CI based on <80 on 3MS Greater weight was not associated with higher rates of cognitive impairment. There was no significant association between WC and cognitive impairment.

PAGE 66

51 Table 1.3.8 Longitudinal studies measuring central obesity that did not support a link between obesity and cognitive outcomes. STUDIES OF CENTRAL ADIPOSITY IN OLDER ADULTS Author (date) N Baseline Age Follow up (years) Wei ght Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Bagger et al., 2004 ) 5607 63.8 mean 7.3 Weight, DXA body composition, TFM, TLM, CFM Short Blessed test Higher baseline weight was associated with lower cognitive decline. Women with the worst cognitive performance at follow up were the ones who lost the most body weight and had the lowest central fat mass (CFM). ( Driscoll et al., 2011 ) 2283 65 79 Up to 5 Mean 3 BMI, WHR, WC 3MS, PMA, Letter fluency, BVRT, CVLT, digital span, CRT, No association between weight and cognition in women who remained s table or gained weight. Cognition was not related to changes in WC. Weight loss was associated with cognitive decline, independent of initial BMI. ( Forti et al., 201 0 ) 749 3 5 Metabolic Syndrome (MetS) WC MMSE Dx VaD, AD In participants aged 75 and older, abdominal obesity was associated with a lower risk of overall dementia (HR = 0.53, 95% CI = 0.28 0.98). ( Han et al., 2009 ) 721 60 85 2 BMI, WHR, WC, PBF CERAD Korean version Dementia dx using DSM IV criteria For men obese at baseline, increasing adiposity (BMI, WHR, WC) was associated with improved cognitive function. For women obese at baseline, both an increase or decrease in WHR or WC were associated with cognitive decline ( Hughes et al., 2009 ) 1,478 Japanese Americans of t he Kame Project mean age of 71.8 years and were dementia free at baseline (1992 1994) Biennially for 8 years (mean 7.8) BMI, WC, WHR Cognitive Abilities Screening Instrument and CERAD dx using DSM IV criteria Higher baseline BMI was significantly associate d with a reduced risk of AD ([HR] = 0.56, 95% [CI] = 0.33 0.97) in the fully adjusted model. Slower rate of decline in BMI was associated with a reduced risk of dementia (HR = 0.37, 95% CI = 0.14 0.98), with the association stronger for those who were over weight or obese (HR=0.18, 95% CI=0.05 0.58) compared to normal or underweight (HR=1.00, 95% CI=0.18 5.66) at baseline.

PAGE 67

52 Table 1 .3.8 Longitudinal studies measuring central obesity that did not support a link between obesity and cognitive outcomes. STUDIE S OF CENTRAL ADIPOSITY IN OLDER ADULTS Author (date) N Baseline Age Follow up (years) Weight Assessment Dementia Assessment Risk ratio Hazard ratio, Odds ratio ( Kanaya et al., 2009 ) 3054 70 79 Up to 8 BMI, WC, sagittal diameter, total fat by DXA Subcutaneous and visceral fat by CT 3MS There was no association between adiposity and cognitive change in women. Higher levels of all adiposity measures were associated with worsening cognitive function in men. ( Power et al., 2011 ) 12,047 65 84 (mean 72.1) 9.7 BMI, WC, WHR ICD 9 & ICD 10 dx codes in Western Australia Data Linkag e System lower risk of dementia than men with normal weight and with WHR < 0.9. Higher adiposity was not associated with incident dementia. ( Raffaitin et al., 2009 ) 2049 74 4 WC MMSE, Benton Visual Set Test Dx VaD, AD Participants with high WC had decreased risk for all cause dementia (HR = 0.86, 95% CI: 0.64 1.17), AD (HR=0.63, 95% CI: 0.4 3 0.94) and VaD (HR=0.82, 95% CI: 0.41 1.66) ( Raffaitin et al., 2011 ) 7087 65+ 4 WC MMSE, Isaacs Set Test (verbal fluency), Benton Visual Retention Test (BVR T, visual working memory) No significant change in cognition in those who started with a high WC. ( West & Haan, 2009 ) 1,351 60 101 (mean 69.9) 8 ( 5.6) BMI and WC 3MSE and DelRec dx dementia by DSM III Compared with the lowest BMI category, overweight participants had a 48% decreased rate of dementia or being cognitively impaired not dementia (CIND) (adj [HR] = 0.52, 95% [CI]: 0.30 0.91) and obese participants had a 61% decreased rate of dementia/CIND (HR = 0.39, 95% CI: 0.20 0.78). By contrast, the middle and high tertiles of WC were associated with higher rates of dementia/CIND compared with the low tertile, (adj HR = 1.8, 95% CI: 1.1 3.1, and adj HR = 1.9, 95% CI: 0.91 3.8 respectively).

PAGE 68

53 For risk of incident dementia, there appears to be an increased risk associated with midlife obesity. More studies assessed incident dementia than any other category. Among these studies, diagnosis of dementia was generally m ade along similar criteria, such as DSM IV or NINCDS ADRDA criteria. It is worth noting that many of these studies specifically examined the diagnosis of AD, as oppos ed to simply all cause dementia, including 15 of the studies supporting an association bet ween overweight and dementia. Therefore the obesity dementia link is not restricted to VaD, although this too was apparent. Relatively little research has been conducted on the association between adiposity and MCI. Importantly, no research studies have ex amined midlife adiposity in relation to incident MCI This may explain why no studies supported an association between overweight and MCI. Clearly further research assessing midlife adiposity and incident MCI is needed. This may be particularly importa nt because MCI often represents a prodromal, stage of dementia one which is clinically identifiable yet early enough to be amenable to effective intervention In any case, primary prevention of MCI may be just as valuable as (or equivalent to) prevention of dementia. Studies of incident dementia could add this to the available outcome measures to increase the useful information. The current literature on cognitive function suggests that being overweight in midlife may increase risk of declines in a broad range of cognitive functions. Positive results were not restricted to memory, nor any other cognitive domain, though memory was certainly represented in both groups. Among studies of baseline adiposity in older adults, weight was related to cognitive func tion but it was more common to find an association between low weight and cognitive decline. Despite these findings, the current literature on obesity and cognitive health is limited by its reliance on BMI as the principle measure of adiposity. The majori ty of studies relied on

PAGE 69

54 BMI alone, without inclusion of measures of fat distribution. This may have been a particular problem for baseline weight among older adults, who tend to lose lean mass as they age, making BMI a poor reflection of adiposity. It is p ossible that the reliance on BMI has contributed to the mixed findings. To investigate whether central obesity is more closely associated with cognitive decline or dementia than BMI we examined the studies of central obesity separately. Across all categor ies of cognitive outcome the number of studies for and against an association were approximately equal. Evidence for an association included a roughly equal mix of studies using midlife vs. older adult adiposity measures at baseline. H owever the vast majo rity of negative studies (83%) were among older adults at baseline. Therefore central obesity among older adults did not seem to increase risk of cognitive decline or dementia. By contrast, central obesity in midlife tended to do so. In conclusion, the ev idence to date suggests that midlife adiposity, whether global or central, may increase risk of cognitive decline and dementia later in life. As this is drawn from observational studies there is clearly a need for further research to investigate whether ad iposity plays a causal role, or whether the association is simply an artifact of other factors such as poor nutrition or sedentary lifestyle. Inclusion of potential mechanisms that could mediate an effect of obesity on neurocognitive health could add valua ble information to our understanding here. In addition w eight loss studies may be useful, but should be conducted with caution and appropriate safety measures given the evidence linking weight loss to increased risk of cognitive decline and dementia. A sys tematic investigation of the effects of weight loss on cognitive function is warranted before proceeding in this direction. Limit ations of C urrent R esearch While the convenience of BMI makes it an understandable first choice measure of adiposity, the evid ence linking obesity to cognitive health would be strengthened by the use of

PAGE 70

55 more accurate measures of adiposity. This is particularly important among older adults, where BMI can be a poor reflection of body fat. While many of these studies provide valuab le information about real world associations in community dwelling adults, the observational study designs cannot be used to determine causality. Even among well controlled longitudinal studies it remains possible that some other factor related to obesity is in fact responsible for the cognitive decline. Most of the studies controlled for socio demographic factors such as education or SES, and many controlled for related health conditions such as diabetes or cardiovascular risk factors, including smoking, but few controlled for health related behaviors linked to obesity, such as quality of diet, physical activity, social support or stress. Furthermore, few studies of the association between weight or adiposity and cognitive health outcomes have directly m easured mechanisms that could potentially mediate a link between obesity and risk of dementia. We therefore discuss some of the potential confounding factors and mediating mechanisms that could be implicated in an association between obesity and cognitive function. 1.4 P otential Confounding Factors In the previous section we reported epidemiological evidence of a link between midlife obesity and cognitive decline or risk of dementia later in life. However it remains possible that the apparent link is due to other factors also related to obesity, rather than obesity per se. Many of the observational studies described above statistically adjusted for some of these factors, however none accounted for all of them. Such potential confounding factors could include education ( Evans et al., 2003 ; Evans et al., 1997 ; Fitzp atrick et al., 2004 ; Kukull et al., 2002 ) depression ( Green et al., 2003a ) sleep apnea ( Bdard, Montplaisir, Malo, Richer, & Rouleau, 1993 ; Grigg Damberger & Ralls, 2012 ) physical inactivity ( Plas sman et al., 2010 ) and poor

PAGE 71

56 nutrition ( Solfrizzi, Panza, & Capur so, 2003 ) Reverse causation is also possible the potential that cognitive impairment or neural damage precedes or even contributes to obesity. The following sections give an overview of evidence regarding these different factors. 1.4.1 Education Lower leve ls of education are associated with increased risk of dementia, even after controlling for socio economic status ( Evans et al., 2003 ; Evans et al., 1997 ; Fitzpatrick et al., 2004 ; Kukull et al., 2002 ; Stern, 20 02 ) ). The association is not simply for all cause dementia, nor just VaD. For example Evans et al. (1997 ) calculated that the risk of AD was reduced by 17% for every additional year of edu cational attainment. In a meta analysis of 10 studies available before 2005, Caamao Isorna, Corral, Montes Martnez, and Takkouche (2006 ) found that compared to persons with high education level, persons with low educat ion had a 59% increased risk of all dementias (95% CI: 1.26 2.01), with the risk of non AD dementias was 1.32 (95% CI: 0.92 1.88), and the risk for AD of 1.80 (95% CI: 1.43 2.27). Similarly (2012 ) analyzed 133 articles with 437,477 subjects published before 2011 and found that low education increa sed risk of AD, VaD and unspecified dementia. The pooled incidence studies gave an odd ratio of 1.88 (95%CI 1.51 2.34). McDowell, Xi, Lindsay, and Tierney (2007 ) examined the compounding effect of socioeconomic and health factors associated with higher education and determined that these could reduce but not remove the association of higher education and lower risk of d ementia. The finding that lower education increases risk of dementia cannot yet be fully the detection of cognitive decline ( Stern, 2002 2009 ) The cognitive reserve hypothesis is also used to explain how some people with pathological signs of AD plaques and tangles ne ver develop symptoms within their lifetime ( Roe, Xiong, Miller, & Morris, 2007 ) In an alternative

PAGE 72

57 hypothesis for the link with education, i t is possible that people with early cognitive deficits or early stages of AD pathology may be less inclined to seek out mental stimulation ma ny years before dementia symptoms appear. It is also possible that people who have low education may share a common set of independent risk factors. E ducation might be associated with good lifestyle choices that reduce dementia risk, such as avoiding smoki ng, maintaining a healthy weight, or avoiding diabetes ( Knopman, 2008 ) The association between education and obesity is complex, likely confounded by many other factors. In addition, as overweight and obesity have become increasingly widespread across the population in recent decades, the association has likely changed. There may have been an increased risk of obesity with less educated individuals in the past h owever more contemporary analyses have found that the differentials may vary by sex and race ( Yu, 2013) found the majority of the differences between education and obesity disappeared after careful adjustments, and only white college graduates were found t o be less obese than white high school graduates. 1.4.2 Sleep A pnea Obstructive sleep apnea (OSA) is a disorder of sleep caused by frequent collapse of the pharyngeal airway during sleep causing airways obstruction ( Rosenberg & Doghramji, 2009 ) It is ( Rosenberg & Doghramji, 2009 ) In the general population approximately 3 7% in men and 2 5% in women have OSA ( Lurie, 2011 ) However prevalence rises significantly with increasing weight ( Pillar & Shehadeh, 2008 ) Among obese patients who present for bariatric surgery inciden ce of OSA can be as high as 78% ( Lopez, Stefan, Schulman, & Byers, 2008 ) Incidence also increases wi th age ( Lam, Sharma, & Lam, 2010 ) Up to 20% of adults ove r 70 years of age have OSA ( Onen & Onen, 2010 ) Ancoli Israel, Klauber, Butters, Parker, and Kripke (1991 ) reported that in a sample of 235

PAGE 73

58 nursing home patients with med ian age 83.5 (women) and 79.7 (men), 96% had some degree of dementia and 70% had symptoms of sleep apnea Obstructive sleep apnea can cause neurocognitive damage across the lifespan ( Bdard et al., 1993 ; Grigg Damberger & Ralls, 2012 ) It can also exacerbate cognitive dysfunction in the elderly ( Cooke et al., 2009 ; Kim, Lee, Lee, Jhoo, & Woo, 2011 ) Studies of the effects of CPAP treatment for OSA on neurocognitive outcomes have produced mixed results (e.g. ( Cooke et al., 2009 ) ; compare ( Ancoli Israel et al., 1991 ) The association of OSA with cognitive impairment and AD in elderly patients does not answer the question of causation. This is an area of active research ( Roth, 2012 ) However given the link between OSA and obesity it is a potential confounding factor that ought to be addressed in future studies of the effects of obesity and cognitive function where possible. 1.4.3 Depression Depressive symptoms are o ften associated with the development of AD and may appear years before clinical AD diagnosis ( Green et al., 2003b ) However it is difficult to determine whether these symptoms represent early manifestations of dementia, or whether depression itself is a risk factor for AD. Major depression can lead to difficulty concentrating, memory impairment, and difficulty making decisions ( Taylor Tavares et al., 2007 ) which are also features of AD ( Sinz, Zamarian, Benke, Wenning, & Delazer, 2008 ; Zamarian, Sinz, Bonatti, Gamboz, & Delazer, 2008 ) Furthermore persons with a history of depression are 2.5 times more likely to develop AD than those who did not, and persons who experienced depression before age 60 have been found to be more than 4 times more likely to develop AD ( Geerl ings, Den Heijer, Koudstaal, Hofman, & Breteler, 2008 ) The nature of the association between depression and obesity is less clear. Depression and obesity often co occur ( Appelhans et al., 2012 ; Luppino et al., 2010 ; Zhao et al., 2011 ) however some studies find no association ( Hach et al., 2007 ;

PAGE 74

59 Rivenes, Harvey, & Mykletun, 2009 ) In a study of 2,439 adults from the NHANES data, ( Zhao et al., 2011 ) demonstrated an incidence of major depressive symptoms in 1% of the general population vs. 3% in the obese group and an incidence of moderate to severe depressive symptoms as 2.4% in the general population and 6.7% in the obese group. While the relationship between depression and obesity needs further clarification, potential depression should be accounted for in studies of neurocognitive health and dementia risk. 1.4.4 Poor N utrition It is also possible that person s who are overweight or obese in midlife have had poor nutritional intake over many years. Nutritional deficiencies, rather than adiposity, could be the cause for cognitive deficits. Over the past decade, a large number of studies have been conducted to i nvestigate the association between dietary composition and risk of dementia, and this continues to be an active area of research. Nutritional deficiencies can also cause impaired cognitive function more generally ( Solfrizzi et al., 2003 ) Deficiencies in essential micro nutrients such as antioxidants (Vitamins C, E, carotenes, etc.) and B vitamins have been associated with cognitive impairment ( Del Parigi, Panza, Capurso, & Solfrizzi, 2006 ) In addition, in some cross sectional studies higher intake of healthy foods was associated with better cognition, while intake of refined su gars, high cholesterol and trans fats were associated with poorer cognitive performance ( Lee et al., 2001 ; Requejo et al., 2003 ) Higher intake of mono unsaturated and polyunsaturated fatty acids has been linked to better cognitive function and lower rates of cognitive decline. ( Morris et al., 2003 ; Solfrizzi et al., 2005 ; Solfrizzi et al., 2003 ) There is also some evidence to suggest that nutritional factors could influence risk of dementia. From a whole food perspective, there is evidence that persons who adhere to the Mediterranean diet have lower risk of dementia ( Scarmeas, Stern, Mayeux, & Luchsinger, 2006 )

PAGE 75

60 The strength of these findings was reflected in a recent NIH consensus report on pre venting cognitive decline and dementia, which reported that the Mediterranean diet is associated with decreased risk of cognitive decline and dementia ( Plassman et al., 2010 ) l Observational studies also link anti oxidants such as Vitamin E a nd C to reduced risk of dementia ( Engelhart et al., 2002 ; Morris, 2005 ; Morris et al., 2002 ) however the NIH report found that there is currently insufficient evidence to determine if this is true. Overall, the NIH report ( Plassman et al., 2010 ) concluded that the strength of evidence is currentl y insufficient to claim that specific nutrients decrease risk of dementia. Clearly there is much more work to be done in this area, however dietary composition should be considered in a thorough investigation of the association between obesity and neuroco gnitive health outcomes. 1.4.5 Physical I nactivity Persons who are overweight or obese are also less likely to be physically active than the general population ( Church et al., 2011 ; Tudor Locke, Brashear, Johnson, & Katzmarzyk, 2010 ) If physical inactivity impacts risk of cognitive decline or dementia then the association between weight and dementia risk could be due to the adverse effects of physical activity, rather than to obesity per se. There is a growing body of evidence supporting an important role for physical act ivity in promoting neural and cognitive health. Many observational studies report an association between physical activity and lower risk of cognitive decline or dementia ( Middleton & Yaffe, 2009 ) A recent meta analysis (Skog, date ) of prospective cohort studies found that 10 of 11 studies reported reduced risk of cognitive decline or dementia among adults who regularly engaged in physical activity in midlife. While the epidemiological evidence is promising, the NIH consensus report on preventing cognitive decline and dementia, found that there is currently insufficient evidence from randomized controlled trials (RCTs) to conclude that physical activity

PAGE 76

61 protects against cognitive decline and dementia ( Plassman et al., 2010 ) Similarly Snowden and colleagues ( Snowden et al., 2011 ) found that there was not en ough available data from quality intervention studies to determine whether physical activity interventions improve cognitive function in older adults. Nonetheless, the association seen in observational studies and small clinical trials suggests that physic al activity should be considered in assessments of the association between obesity and dementia. 1.4.6 Reverse C ausation It remains possible that poor cognitive function can contribute to the development of obesity, rather than the reverse. This would be consis tent with the findings of Cournot et al ( Cournot et al., 2006b ) that obesity in midlife was associated with increased risk of cognitive decline later in life, bu t change in weight over time (including weight gain) was not associated with increased risk of cognitive decline. Halkjr, Holst, and Srensen (2003 ) demonstrated that intelligence test score was inversely related to the risk of developing obesity in a longitudinal study of 1709 men (med ian age 19) over more than two decades. Experimental studies are needed before the question of causation can be adequately addressed. Interventions that reduce midlife obesity and measure incident dementia in older age would be ideal. Such a study would be particularly difficult to implement given the large sample size and long follow up periods required. Experimental studies of more short term effects of weight loss on neurocognitive outcomes and on potential mechanisms implicated in AD pathology would be more immediately practicable. The short term effects of weight loss interventions on adult neurocognitive healt h will be discussed in section 3

PAGE 77

62 1.5 P otential M echanisms L inking Adiposity to Dementia Risk 1.5.1 isease P athophysiology: A B rief O vervi ew While the pathophysiology of AD is not yet fully understood, many useful advances have been made in this field in recent decades. The prevailing theory of A D pathophysiology holds amyloid 42 plaques which characterize the disease directly contribute to neuronal amyloid is instead a by product of the real underlying problem cannot yet be ruled out ( Budson & Solomon, 2011 ; Resende, Ferreiro, Pereira, & Oliveira, 2008 ) Beta amlyoid is produced from the Amyloid Precursor P rotei n (APP), and is amyloid accumulation are still under investigation. However some of the factors that could serve as physiological mechanisms mediating an effect of obesi ty on neurocognitive health have already amyloid burden ( Craft, 2005 ; Zhao et al., 2003 ) 1.5.2 Insulin R esistance and G lucose R egulation Glucose dysregulation is one factor which could mediate an effect of obesity on risk of AD. Impaired glucose regulation and insulin resistance are common features of obesity, and insulin resistance is a centra l feature of both the metabolic syndrome (MetS) and Type 2 diabetes mellitus (T2DM). As already noted, persons with T2DM, are at increased risk of cognitive decline or dementia, including AD ( Biessels & Gispen, 2005 ; Elias et al., 2005 ; Yaffe et al., 2004a ) However non diabetic individuals with insulin resistance also show some evidence of deficits in learning and memory ( Vanhanen et al., 1997 ; Vanhanen et al., 2006 ) even after controlling for vascular factors ( Convit et al., 2003 ) In animal studies, a dose response relationship can be seen, with worse memory performance and smaller hippocampal volumes recorded for animals with worse glycemic control, independent of age and overall cognitive function ( Convit et al., 2003 )

PAGE 78

63 Insulin r esistance and hyperglycemia in the peripheral circulation can have significant effects on neural glucose availability and on insulin levels in the brain. Growing evidence indicates that peripheral insulin resistance and hyperglycemia can decrease the trans port of glucose across the blood brain barrier (BBB) ( McNay, Fries, & Gold, 2000 ) resulting in hypoglycemia in the brain. Since neurons depend on glucose for ener gy the effect may be devastating, particularly for brain regions frequently activated. Peripheral insulin resistance and hyperinsulinemia also decrease insulin transport across the BBB, leading to insulin deficiency in the brain ( Baura et al., 1996 ; Kaiyala, Prigeon, Kahn, Woo ds, & Schwartz, 2000 ) Insulin receptors can be found in particularly high concentrations in the hippocampus, a brain region involved in learning and memory ( Bing ham et al., 2002 ; Jacobson & Sapolsky, 1991 ) Insulin may affect memory through direct receptor mediated effects ( Craft, 2007 ) Thus the brain of a person with peripheral insulin resistance, hyperglycemia and hyperinsulinemia may become hypoglycemic and lacking the insulin normally involved in amyloid formation in the brain and decrease its clearance to the periphery ( Craft, 2005 ; Ho et al., 2004 ; Marambaud, Zhao, & Davies, 2005 ; Zhao, T uominen, & Kinnunen, 2004 ) Consistent with these findings, patients with AD have high rates of glucose dysregulation ( Craft, 2007 ) and post mortem reveals decreased insulin in the brains of AD patients relative to controls ( Co le & Frautschy, 2007 ; Rivera et al., 2005 ; Steen et al., 2005 ) 1.5.3 Hypertension Hypertension is normally defined as a systolic blood pressure of 140mmHg or higher or a diastolic pressure of 90mmHg or higher ( Carretero & Oparil, 2000 ) In the general population, prevalence of hypertension in adults is estimated at 24% in the USA ( Burt et al., 1995 ) and 26% worldwide ( Kearney et al., 2005 ) Reviews indicate t hat t hose who are obese are more likely

PAGE 79

64 than non obese individuals to be hypertensive ( Koebnick et al., 2012 ; McAuley et al., 2012 ; McAuley et al., 2009 ) There are several potential reasons for this. Persons who are obese tend to have higher levels of insulin and insulin stimulates the sympathetic nervous sy stem, stimulating the kidneys to retain sodium ( Krie ger & Landsberg, 1988 ) Obesity also increases leptin levels, and leptin has been shown to increase blood pressure ( Landsberg, 2001 ) ,). Obesity can also contribute to increased aldosterone production ( Ahmed, Fisher, Stevanovic, & Hollenberg, 2005 ; Sarzani, Salvi, Dess Fulgheri, & Rappelli 2008 ) and is associated with a high incidence of obstructive sleep apnoea ( Gami et al., 2004 ) which can increase sympathetic nerv ous system output. Hypertension has been linked to increased risk of MCI (Reitz et al, 2010) and dementia ( Kivipelto et al., 2001 ; Launer et al., 2000 ) For example, Kivipelto et al. (2001 ) found that people w fold higher risk of AD in later life, compared to persons with normal SBP (OR 2.3, 95% CI: 1.0 5.5). In this study diastolic blood pressure (DBP) in midlife had no signi ficant effect on the risk of AD. In contrast Launer et al. (2000 ) found an association for both SBP and DBP in midlife with incident dementia. Incidence of de mentia among men with hypertension in midlife whose hypertension was never treated increased almost five fold in those with SBP 160 mm Hg and higher compared with SBP of 110 to 139 mm Hg (OR 4.8, 95% CI: 2.0 11.0). Risk of dementia was also increased for D BP of 90 94 mm Hg (OR 3.8, 95% CI: 1.6 8.7), and for DBP of 95 mmHg and over (OR 4.3, 95% CI: 1.7 10.8), compared to those with DBP of 80 to 89 mm Hg. By contrast, men whose hypertension was treated showed no increased risk for dementia. Consistent with t his, Forette et al. (1998 ) demonstrated that lowering of SBP in elderly patients significantly reduced the incidence of dementia. At a structu ral level, Petrovitch et al. (2000 ) showed that midlif e

PAGE 80

65 In contrast to these studies of midlife hyp er tension increases risk, some studies have indicated that hy p o tension late in life is associated with increased dementia risk. For example Guo, Viitanen, Fratiglioni, and Winblad (1996 ) noted that people with moderate or severe dementia were more likely persons without dementia to have hypotension, either systolic or diastolic. Similarly Qiu, Winblad, and Fratiglioni (2009 ) found low blood pressure to be associated with a more than two fold increased risk of dementia or AD specifically, and Morris et al. (2000 ) showed the same relationship with AD. These findings may point to a non linear, inverse U shaped relationship between blood pressure and cognitive health. In support of t his, Razay, Williams, King, Smith, and Wilcock (2009 ) found that among perso ns with AD rate of cognitive decline was increased in the groups with either high or low DBP. Glynn et al. (1999 ) found a similar U shaped relationship with both diastolic blood pressure and systolic blood pressure. As with the studies described above, their results also suggest that midlife hypertension is a stronger r isk factor for dementia than hypertension later in life. 1.5.4 Dyslipidemia Dyslipidemia, or abnormally elevated lipids in the bloodstream, is another common feature of obesity ( Nguyen, Magno, Lane, Hinojosa, & Lane, 2008 ) Dyslipidemia has been associated with increased risk of dementia. High total cholesterol in midlife has been associa ted with an increased risk for AD ( Anstey, Lipnicki, & Low, 2008 ; Kivipelto et al., 2002 ; Solomon et al., 2007b ) By contrast, high total cholesterol late in life was not associated with MCI ( Reitz et al., 2008 ) nor with dementia ( Anstey et al., 2008 ; Reitz, Luchsinger, Tang, Manly, & Mayeux, 2005 ; Solomon et al., 2007a ) However neither high total cholesterol nor high triglycerides late in life

PAGE 81

66 were associated with reduced risk of demen tia ( Mielke et al., 2005 ) Interestingly, Solomon et al. (2007a ) noted that a moderate fall in TC levels between midlife and late life was associated with a more severe cognitive impairment. As with other intervention trials for symptomatic dementia, various studies report that use of statins in treatment of persons with established AD did not produce improvements in symptoms ( Feldman et al., 2010 ; McGuinness et al., 2010 ; Sano et al., 2011 ) nor did they produce improvements in subjects who were 65 years of age or older ( Arvanitakis et al., 2008 ; Rea et al., 2005 ; Zandi et al., 2005 ) By contrast other studies examining the potential for statins to prevent dementia have demonstrated a positive association ( Cramer, Haan, Galea, Langa, & Kalbfleisc h, 2008 ; Haag, Hofman, Koudstaal, Stricker, & Breteler, 2009 ; Jick, Zornberg, Jick, Seshadri, & Drachman, 2000 ; Wolozin, Kellman, Ruosseau, Celesia, & Siegel, 2000 ; Wolozin et al., 2007 ) though the NIH consensus report on preventing cognitive decline and dementia found that there is no consistent assoc iation and still insufficient evidence to claim that statin use reduces risk of dementia ( Plassman et al., 2010 ) 1.5.5 Oxidative S tress Cellular damage by free radicals, such as reactive oxygen species, may be a universal feature of the aging proc ess ( Calabrese et al., 2010 ; Knight, 2000 ; Nunomura et a l., 2012 ) Affecting proteins, lipids and nucleic acids this oxidative damage, or oxidative stress, is normally met by anti oxidant defenses ( Texel & Mattson, 2011 ) When these defenses are inadequate, overwhelmed or exhausted t issue damage occurs ( Friguet, 2002 ; Squier, 2001 ; Ve nkateshappa, Harish, Mahadevan, Srinivas Bharath, & Shankar, 2012 ) The role of oxidative stress in AD continues to be investigated as it may provide a trigger for some of the initial neural damage ( Foley & White, 2002 ; Sultana & Butterfield, 2008 )

PAGE 82

67 1.5.6 Leptin and L eptin R esistance Leptin is a hormone secreted by adipocytes, Leptin levels are proportional to adiposi ty in normal individuals ( Ingvartsen & Boisclair, 2001 ; Popovic & Duntas, 2005 ) Although obese individua ls often have high leptin levels, many lack a normal response to leptin administration, suggesting leptin resistance (Jung and Kim, 2013). This may occur for a variety of reasons, including defective signaling pathways ( Sahu, 2003 ) fructose consumption ( Scarpace & Zhang, 2009 ; Shapiro et al., 2008 ; Shapiro, Tmer, Gao, Cheng, & Scarpace, 2011 ) high sucrose and fat diets ( Vasselli, Scarpace, Harris, & Banks, 2013 ) or as a response to hyperleptinemia ( Knight, Hannan, Greenberg, & Friedman, 2010 ) regul ation of appetite, stimulation of thermogenesis, enhancement of fatty acid oxidation, decreasing glucose and reduction of body weight and fat ( Yadav, Kataria, Saini, & Yadav, 2012 ) Consistent with its regulatory roles, leptin receptors can be found in the cerebral cortex, ce rebellum, brainstem, basal ganglia, and hippocampus ( Harvey, 2003 ) Leptin also has effects on immunity ( Ingvartsen & Boisclair, 2001 ) Leptin deficiency is rare, but leads to severe obesity, hyperphagia, hypogonadism, hype rinsulinemia, hypercholesterol emia and impaired immune function, all of which reverse with leptin administration ( Farooqi & O'Rahilly, 2009 ) Leptin appears to have additional effects on the brain beyond the regulation of appetite or weight. In particular, it has been implicated in AD pathology and symptoms. Leptin has been s amyloid in vivo ( Fewlass et al., 2004 ) Levels have been shown to be decreased in AD, with serum l evels inversely proportional to AD severity ( Greco, Sarkar, Johnston, & Tezapsidis, 2009 ; Holden et al., 2009 ) In a study of 2,871 older adults, Holden and colleagues (2009), followed over a period of four years, demonstrated that those with a higher initial leptin level had lower rates of cognitive decline,

PAGE 83

68 independent of other comorbidities or body fat (OR=0.6 6 95% CI: 0.48 0.91). Interestingly, Zeki Al Hazzouri and colleagues ( Zeki Al Hazzouri, Haan, Whitmer, Yaffe, & Neuhaus, 2012 ) demonstrated that a raised leptin level in elderly women of normal weight but not in overweight or obese women was significantly associated with a lower risk of dementia or MCI. They found similar results in a subsequent study of both men and women ( Zeki Al Hazzouri, Stone, Haan, & Yaffe, 2013 ) 1.5.7 Insulin Like Growth F actor 1 Insulin like growth factor (IGF 1) is a hormone involved in the regulation of cell proliferation, cell differentiation, promotion of metabolism (nutrient transport, en ergy storage, gene transcription and protein synthesis) and programmed cell death in adults ( Feldman, Sullivan, Kim, & Russell, 1997 ; Lauterio, 1992 ; Pavelic, Matijevic, & Knezevic, 2007 ) Most circulating IGF 1 is produced by hepatocytes, however many other cell types, including neurons ( D'Ercole, Ye, Calikoglu, & Gutierrez Ospina, 1996 ) also produce IGF 1, with paracrine or autocrine effects ( Croci et al., 2011 ; Frystyk, 2010 ; Laron, 2001 ) Receptors (IGF1R) can be found throughout the body, and have widespread distributi on in the brain ( D'Ercole et al., 1996 ) IGF 1 also binds to insulin receptors, though with a lower affinity than to IGF1R. The relationship between IGF 1 bioac tivity with insulin levels, insulin resistance and the metabolic syndrome is not straight forward. There is an inverse U relationship between increasing insulin resistance and IGF 1 activity. Increasing the number of elements of the metabolic syndrome corr elates with a peak of IGF 1 activity with three elements but falls thereafter ( Brugts et al., 2010 ) The function and activity of this hormone can be affected by a number of factors. Activity of IGF 1 is modulated by growth hormone, IGF binding proteins, and IGF receptor resistance (Connor et al, 2008). It is also affected by nutritional factors ( Runchey et al., 2012 ) and physical activity ( Ber mon, Ferrari, Bernard, Altare, & Dolisi, 1999 ; Rojas Vega, Knicker,

PAGE 84

69 Hollmann, Bloch, & Struder, 2010 ) Levels of IGF 1 tend to decrease with age ( Laron, 2002 ; Muller et al., 1993 ; Sonntag, Ramsey, & Carter, 2005 ; Toogood, O'Neill, & Shalet, 1996 ) Interestingly, some mutations in genes for the insulin/IGF 1 pathway have been linked to increased longevity ( Bonafe et al., 2003 ; Flachsbart et al., 2009 ; Kojima et al., 2004 ; Pawlikowska et al., 2009 ) ( Soerensen et al., 2010 ; Tazearslan, Huang, Barzilai, & Suh, 2011 ) It is also associated with adiposity. Fo r example, in a cross sectional study o f over 6,000 people Parekh and colleagues ( Parekh et al., 2010 ) found a strong positive correlation of serum IGF 1 levels a nd adiposity (BMI or waist circumference), and a strong negative correlation with age. Con sistent with this, Gapstur and colleagues ( Gapstur et al., 2004 ) showed a positive association between weight and IGF 1 in a 9 year longitudinal study of 1418 adults aged 20 to 34 at enrolment. This is not foun d in all studies, for Nam and colleagues ( Nam et al., 1997 ) found that obese subjects had similar levels of total IGF 1 compared to the controls, though they did have higher free IGF 1 levels. The association with weight may not be linear, for Gram et al ( Gram et al., 2006 ) showed lower levels of IGF 1 with both low weight and high weight participants compared to mid weight participants in a study of 2139 women. Several lines of evidence poi nt to a neuroprotective role for IGF 1, and consequently falling IGF 1 levels with age may reduce some of that protection. Firstly, IGF 1 normally provides neuroprotection by increasing neuronal survival ( Carro & Torres Aleman, 2004 ) It can also promot amyloid clearance from the brain ( Carro & Torres Aleman, 2004 ; Freude, Schilbach, & Schubert, 2009 ) Ther e is increasing evidence that cerebral insulin and IGF 1 resistance are major factors in AD ( de la Monte, 2012 ; Talbot et al., 2012 ) Human S tudies R elating to IGF 1 and C ognition In a cross sectional study of 22 subjects aged between 65 and 86 years of age by ( Roll ero et al., 1998 ) ser um IGF 1 levels correlated positively with MMSE scores. Angelini and

PAGE 85

70 colleagues ( Angelini et al., 2009 ) found a similar association in a s tudy of 75 hypertensive patients over 65 years of age. Cognition was measured with MMSE, Cambridge cognitive examination (CAMDEX R), and the frontal assessment battery (FAB). In a 2 year prospective study by Kalmijn and colleagues ( Kalmijn, Janssen, Pols, Lamberts, & Breteler, 2000 ) a higher IGF 1 leve l was associated with less cognitive decline among 186 healthy subjects aged 55 to 80 years of age. Dik and colleagues ( Dik, Deeg, Visser, & Jonker, 2003 ) demonstrat ed that lower IGF 1 levels were associated with poorer information processing speed and a faster decline over three years among 1318 subjects aged 68 88 years. In a study of 17 healthy subjects between the age of 66 and 77, Aleman and colleagues ( Aleman et al., 2000 ) found that higher serum IGF 1 levels were associated with better scores in mental processing speed. Raising IGF 1 levels to treat AD symptoms has shown mixed results. Alvarez et al ( Alvarez et al., 2009a ) demonstrated that the neurotrophic agent, cerebrolysyn, which improves serum IGF 1 levels, also improved global function, disabilities and behavior in 207 late onset AD patients over a 24 w eek trial. However Sevigny and colleagues ( Sevigny et al., 2008 ) stimulated production of IGF 1 in a randomized trial of 563 patients with mild to moderate A D over 12 months. No improvement in the treatment group over the controls was observed. While the research to date suggests that IGF=1 may play a role in AD pathophysiology much more research is needed to determine whether the association is causal, or whe ther interventions that increase IGF 1 can have neuroprotective effects. 1.5.8 Inflammation A strong body of evidence shows that systemic low grade inflammation is a common feature of obesity ( Bastard et al., 2006 ; Black, 2002 ; Das, 2002 ; Ford, 2003 ; Wellen & Hotamisligil, 2003 ) particularly central obesity ( Craft, 2007 ; Fried, Bun kin, & Greenberg, 1998 ) Epidemiological evidence demonstrates that inflammation is present during cognitive decline

PAGE 86

71 and AD, however it is not clear whether the inflammation is a driving force behind neurological damage, an innocent bystander, or part of a repair process ( Bruunsgaard & Pedersen, 2003 ; Bruunsgaard, Skinhoj, Pedersen, Schroll, & Pedersen, 20 00 ) The relationship between the MetS and risk of cognitive decline may be moderated by inflammatory cytokines such as IL 6, though further research is needed ( Ya ffe et al., 2004b ; Yaffe et al., 2004c ) Chronic low grade inflammation, indicated by biomarkers such as Interleukin (IL) 6 and Tumor Necrosis Factor (TNF) ( Bru unsgaard et al., 2000 ; McGeer, Klegeris, & McGeer, 2005 ) For example TNF with brain aging, and elevations have been reported in MCI and AD ( Carro, Trejo, Gomez Isla, LeRoith, & Torres Aleman, 2002 ; Tarkowski, Andreasen, Tarkowski, & Blennow, 2003 ) Furthermore TNF amyloid clearance that are induced by IGF 1 ( Alvarez et al., 2009b ; Carro et al., 2002 ) F ree IGF 1 correlates negatively with serum TNF ( lvarez, Cacabelos, Sanpedro, Garca Fantini, & Aleixandre, 2007 ; Alvarez et al., 2009b ) so elevated TNF amyloid load ( Cr aft, 2007 ) While these findings are suggestive of a link between inflammation and neurocognitive function more research is needed to determine the consistency, magnitude and meaning of the associations. 1.5.9 Cortisol and HPA A xis D ysregulation Adipose tissu e, and particularly central adipose tissue, secretes the glucocorticoid ( Bjorntorp & Rosmond, 2000 ; Lottenberg et al., 1998 ) ( Pasquali et al., 1993 ) However it is now widely believed that in obese individuals cortisol secretion is increa sed relative to healthy weight controls, but cortisol clearance is also increased, leading to normal or low plasma cortisol concentrations ( Morton, Ramage, & Seckl, 2004 ; Roberge et al., 2007 ; Salehi, Ferenczi, & Zumoff, 2005 ) However obesity has been associated with dysregulation of the normal variability of t he cortisol diurnal rhythm ( Bjorntorp & Rosmond, 2000 )

PAGE 87

72 Exposure to glucocorticoids (GCs) such as cortisol can have both direct effects on the brain ( McEwen, 2000 2008 ; McEwen, Magarinos, & Reagan, 2002 ) Glucocorticoids readily cross the BBB to act directly on the brain ( McEwen, 2000 ) The effects of these GCs on neurocognitive function depend on the magnitude and duration of exposure ( McEwen, 1998 2004 ) While mild moderate exposure can enhance attention and memory, more intense or severe exposure impairs them both ( Diamond, Park, & Woodson, 2004 ; Karlamangla, Singer, Chodosh, McEwen, & Seeman, 2005 ; McEwen, 1998 2000 ; McGaugh & Roozendaal, 2002 ) Furthermore, elevated GCs are associated with clinical AD symptoms and pathology. For example, patients with early AD may exhibit s ignificantly higher total plasma cortisol than controls ( Peskind, Wilkinson, Petrie, Schellenberg, & Raskind, 2001 ) and higher plasma cortisol predicts more rapid cognitive decline and decreased hippocampal volume ( Lupien, Buss, Schramek, Maheu, & Pruessner, 2005a ; Lupien et al., 2005b ; Peskind et al., 2001 ; Rasmuson, Nasman, Carlstrom, & Olsson, 2002 ) Among those without dementia, older adults with significant increases in cortisol over 4 years show deficits in explicit memory, selective attention and an average 14% d ecrease of hippocampal volume on MRI compared to others with no change or reductions in cortisol ( McEwen, 2000 ) Similarly older women with the highest cortisol l evels had the lowest memory scores, and increasing concentrations over a 2.5 year follow up were associated with cognitive decline ( Seeman, McEwen, Singer, Albert, & Rowe, 1997 ) Building on findings such as these, a glucocorticoid hypothesis of brain aging ( Landfield, Blalock, Chen, & Porter, 2007 ) proposes that repeated stress produces cumulative damage to the brain across the lifespan. Glucocorticoids can also indirectly affect neurocogni tive health via their effects on immune function ( McEwen, 1997 1998 ; Munck & Naray Fejes Toth, 1994 ) insulin resistance and glucose regulation ( Black, 2002 ; Kyrou & Tsigos, 2007 ; Wellen & Hotamisligil, 2003 2005 )

PAGE 88

73 and brain derived neurotrophic factor ( Taliaz et al., 2011 ; Tapia Arancibia, Rage, Givalois, & Arancibia, 2004 ) 1.5.10 Brain D erived N eurotrophic F actor Brain derived neurotrophic factor (BDNF) is a neurotrophin th at shows neuroprotective effects in adults. It is involved learning and memory ( Duan, Lee, Guo, & Mattson, 2001b ; Lee Duan, Long, Ingram, & Mattson, 2000 ; Lu, Christian, & Lu, 2008 ) in as well as neurogenesis, synaptic plasticity, and neurotransmitter synthesis ( Diogenes, Assaife Lopes, Pinto Duarte, Ribeiro, & Sebastiao, 2007 ; Lu, 2003 ; Mattson, Duan, & Guo, 2003 ) It is likel y that BDNF is involved in the protective cellular repair response after damage ( Begliuomini et al., 2008 ; Mattson, 2005 ) and it has been found to protect neurons in experimental models of AD and ( Duan, Guo, & Mattson, 2001a ) Increased levels of BDNF are therefore beneficial, but it is difficult to determine whether they reflect good health or the presence of damage that needs repair. The relationship b etween obesity and BDNF in humans is not yet clear. In animals, BDNF deficiency leads to hyperphagia, obesity and insulin resistance ( Duan, 2003 ) while central infusion of BDNF in rats induces weight loss due to appetite suppression ( Pelleymounter, Cullen, & Wellman, 1995 ) However the effect of ob esity on BDNF is not clear. In humans cross sectional studies of BDNF and weight give mixed. Some studies find decreased serum BDNF in obese adults relative to controls ( Krabbe et al., 2007 ) yet others have found higher serum BDNF ( Suwa et al., 2006 ) The differences may be due to the populations studied, or to potential confounding factors, as BDNF is affected by gender, age, stress and inflammation among other things ( Makar et al., 2008 ) ; Makar et a l, 2007) While animal studies s how significant effects of central (i.e. brain, CSF) BDNF on brain function, there remains some question as to how well BDNF levels in peripheral circulation

PAGE 89

74 reflect those in the central nervous system. Few studies have directly addressed this question, bu t there is some evidence that BDNF does cross the BBB. Poduslo & Curran ( Poduslo & Curran, 1996 ) showed that methionine BDNF can cross the BBB, though they did not test passage of the natural form of BDNF. In a separate study, Pan and colleagues ( Pan, Banks, Fasold, Bluth, & Kastin, 1998 ) used radiolabelled BDNF in mice and showed that BDNF injected intravenou sly could be found in the cerebral cortex parenchyma, suggesting passage across the BBB. Conversely, after cerebroventricular injection, radiolabelled BDNF became detectable in blood at a rate similar to that seen for re absorption of cerebrospinal fluid ( CSF). Despite these two studies, the correlations between central and peripheral levels in other studies are mixed. Some researchers have found correlation between serum & cortical BDNF in rats ( ( Hellweg, Ziegenhorn, Heuser, & Deuschle, 2008 ; Ziegenhorn et al., 2007 ) Others have found that serum BDNF correlates positively with cortical BDNF levels in newborn rats, but not in adults ( Karege et al., 2002 ) One study in humans attempted to measure what they hypothesized was BDNF output from the brain using jugular to ar terial concentration difference of BDNF in direct, internal jugular vein sampling (Krabbe et al, 2008). However the validity of this methodology is questionable (Lambert, 2008). Animal studies of BDNF report an important role for BDNF in normal learning a nd memory ( Duan et al., 2001a ; Lu et al., 2008 ; Mattson, 2000 ; Mu, Li, Yao, & Zhou, 1999 ) Mice that over express the BDNF receptor (TrkB) show evidence of enhanced learning and memory ( Koponen et al., 2004 ) Intra hippocampal administration of BDNF facilitates short term memory ( Alonso et al., 2002 ) and infusion of anti BDNF antibodies causes amnesia for spatial learning tasks in rats ( Mu et al., 1999 ) Anti BDNF antibodies block long term potentiation (LTP) and impair long term memory. Transgenic mice deficient in BDNF show deficits in LTP ( Croll et al., 1999 )

PAGE 90

75 Evidence in humans also indicates an association between BDNF and memory, using measures of BDN F in the peripheral circulation. Interestingly the direction of the association appears to vary with severity of cognitive impairment. Yasutake et al ( Yasutake, K uroda, Yanagawa, Okamura, & Yoneda, 2006a ) compared healthy controls to patients with either AD or VaD, matching these patients for age, gender and severity of dementia. Serum BDNF was significantly lower in AD patients than in VaD patients and controls respectively. Despite this, serum BDNF did not correlate with scores on the MMSE or the Functional Assessment Rating Test (FAST) in patients with AD. Similarly Laske et al ( Laske et al., 2007 ; Laske et al., 2006 ) found that patients with established AD showed the lowest serum BDNF, but BDNF was slightly higher in patients with early stage AD, relative to both normal controls and patients with AD. In this study serum BDNF correlated significantly with MMSE (MCI r=0,855; p=0,001; AD r=0,396; p=0,010). In another study, Zhang and colleagues (2008) found that patients with amnestic MCI had significantly lower circulatin g BDNF than healthy controls, and that BDNF correlated positively with scores on tests of delayed recall. Genotype for the BDNF gene did not differ between aMCI patients and normal controls, nor did genotype correspond to serum BDNF concentration, indicati ng that other factors strongly affect BDNF expression. In another study, patients with AD showed lower BDNF than controls in the hippocampus and temporal cortex on autopsy ( Yasutake, Kuroda, Yanagawa, Okamura, & Yoneda, 2006b ) In the non clinical population, BDNF also appears to correlate wi th cognitive function. For example Komulainen et al ( Komulainen et al., 2008 ) found that decreased BDNF was associated with impaired global cognitive functio n (CERAD test battery) in women but not men, and with specific impairments in memory but not executive function. In women a one standard deviation decrease in BDNF was associated with 50 60% decreased memory scores, and increased the probability of low MMS E. Effects remained after controlling for age, education,

PAGE 91

76 depression, impaired glucose metabolism, cardiovascular disease, antihypertensive medication, lipid lowering medication, use of sex hormones, smoking, alcohol consumption, storing time of plasma in the freezer and platelet count. 1.6 S ummary Obesity affects a number of factors currently under investigation for their role in dementia pathophysiology, including glucose regulation, inflammation, leptin, IGF 1 and BDNF ( Lupien et al., 2005a ) Observational studies suggest that midlife obesity can increase risk of cognitive decline or dementia, yet relatively few observational or intervention studies have concurrently investigated the potential mediating role of these me chanistic factors. Most studies have controlled for education socio economic status, gender, age and diabetes status, few to date have accounted for the potential role that physical activity, quality of diet or other behaviors linked to obesity could have on the association between obesity and dementia. Hence more research is needed before it can be concluded that obesity plays a causal role in cognitive decline and dementia, and more research is needed to understand the mechanisms that may mediate the eff ect. Ultimately the best information will be from large, well controlled randomized controlled trials. However observational studies that account for frequently unmeasured variables such as physical activity or diet can also contribute useful information. 2. S tudy 2 : the NHANES III S tudy 2.1 I ntroduction Despite a growing body of epidemiological evidence suggesting that midlife obesity increases risk of cognitive decline and dementia later in life ( Gustafson, 2008 ; van den Berg et al., 2008 ) much remains unknown about the association between obesity and cognitive function in early and mid adulthood. To better understand the nature of this relationship it will

PAGE 92

77 be important to determine not on the direction of association but also to investigate whether duration of obesity affects the association, whether factors such as quality of diet and physical activity mod erate the association and which physiological mechanisms could mediate the effect. It is also important to determine whether distribution of body fat plays a role, for the adipokines secreted by central adipose tissue may confer additional risk of neuroco gnitive damage. The primary aim of this study was to determine whether obesity is associated with reduced cognitive function in early and mid adulthood in the general population. Given the evidence for an association between midlife obesity and cognitive decline ( ( Gorospe & Dave, 2007 ; Whitmer et al., 2008 ) and evidence for prolonged and gradual pre clinical s tages of AD pathology that predate symptoms by decades ( Sperling et al., 2011 ) it is possible that an association between obesity and cognitive function is alrea dy apparent in midlife. This has been supported by at least one study ( Cournot et al., 2006a ) but warrants further investigation. Assessing dementia risk requires extremely long follow up times, and i t was not be possible to determine whether such an association reflect ed an early stage of dementia pathology. Nor was it possible to determine causal direction. However it is possible to assess whether results are con sistent with what would be expected from a causal association between obesity and later cognitive decline. Many studies of the association between obesity and cognitive function have used BMI a s the sole measure of adiposity Yet at best, BMI is a rough i ndicator of adiposity, potentially masking differences in body composition and fat distribution. The mechanisms that may link obesity to dementia are more closely related to central adiposity than to global obesity, and studies that differentiated between global and central obesity have tended to show stronger associations for the latter ( Cereda, Sansone, Meola, & Malavazos, 2007 ; West & Haan, 2009 ;

PAGE 93

78 Whitmer et al., 2008 ) The present study therefore investigated the possibility that central obesity, as measured by waist hip ratio (WHR) is more closely related to cogn itive function than global obesity. Duration of exposure to obesity may be important, since the cumulative effects of small but regular damage could be significant over decades. A cumulative exposure outcome model of the lifecourse approach to chronic dis ease etiology ( Kuh & Ben Shlomo, 2004 ) w ould predict that prolonged duration of exposure could contribute cumulative effects over time. This study therefore explored whether longer duration of obesity was associated with worse cognitive performance. If obesity affects cognitive function then be haviors that affect obesity may affect t he association between obesity and cogn itive function. Differences in these behaviors could be what really drives the association with obesity, rather than adiposity itself. Alternatively, health related behaviors ma y moderate the effects of obesity on neurocognitive health by affecting the physiological mechanisms that increase risk of cognitive decline, such as cardiovascular health, immune function, or levels of neuroprotective factors such as IGF 1 and BDNF. While quality of diet and frequency of PA could affect rates of obesity itself, s uch that people with good diet and regular PA are less likely to be obe s e it is also possible that these factors may have independent effects on cognitive function Two possibili ties present themselves. The association between obesity and cognitive function described elsewhere may simply be an artifact of an effect of poor quality diet and sedentary lifestyle on neurocognitive health. Alternatively, obesity could have an independe nt effect that can be moderated by good quality diet and regular PA. For example a person who is obese but physically active may have better cognitive function than others who are sedentary. This study therefore investigated whether the association between obesity and cognitive function in midlife was moderated by these health related behaviors.

PAGE 94

79 Since the re is a promising body of evidence supporting a role for physical activity in promoting neurocognitive health ( Fratiglioni, Paillard Borg, & Winblad, 2004 ; Smith et al., 2010b ; Snowden et al., 2011 ; Stranahan, Zhou, Martin, & Maudsley, 2009b ) whi le the literature on diet is more mixed ( Faxn Irving, Basun, & Cederholm, 2005 ; Gibson & Green, 2002 ; Plassman et al., 2010 ) we hypothesized that PA, but not quality of diet, would moderate the association betw een obesity and cognitive function in adults. The present analysis was therefore undertaken to 1) determine whether obesity is associated with cognitive function in the general US population in midlife, and 2) explore factors that may moderate the associa tion, including duration of obesity and health related behaviors such as diet and PA. The specific aims and hypotheses can be seen in Table 2.1 1

PAGE 95

80 Table 2.1 1 Spec ific aims and hypotheses of Study 2: NHANES Specific aim Hypothesis SA.1. To determine whether obesity is associated with worse cognitive function in midlife. H1. Obesity will be associated with worse cognitive function.. SA.2. To investigate whethe r central obesity in midlife has a stronger association with cognitive function than global obesity. H2. The association will be stronger for central obesity (WHR) than for measures of global obesity (BMI, % fat mass). SA.3. To investigate whether duratio n of obesity affects the association with cognitive function. H3. Obese persons who report they were obese 10 years ago will have worse cognitive function than persons who have been obese less than 10 years. SA.4. To determine whether quality of diet an d physical activity moderate the association between adiposity and cognitive function. H4. The association between obesity and cognitive function will remain after adjusting for quality of diet and frequency of physical activity. Physical activity, but no t quality of diet, will moderate the association between adiposity and cognitive function. Data from the Third National Health and Nutrition Examination Survey (NHANES III) provides a useful opportunity to investigate whether central obesity or global o besity is associated with minor cognitive deficits in midlife, as well as the potential for health related behaviors to moderate any such association. Although new NHANES surveys have been conducted since 1994, NHANES III is the only NHANES study to date t o include measures of cognitive function. This secondary data analysis was declared exempt by the Colorado Institutional Review Board. COMIRB Protocol 10 1154

PAGE 96

81 2.2 M ethods 2.2.1 Study D esign The Third National Health and Nutrition Examination Survey (NHANES III) was a cross sectional epidemiological study of the civilian, non institutionalized population of the United States. A detailed description of NHANES III recruitment, sampling and study methods can be found online in the Plan and Operations Manual (NCHS, 1994; U.S. DHHS, 1996). A brief summary of participants, methods and measured variables is provided here. NHANES III used a stratified, multi stage probability sampling design to represent the civilian population in 50 states, based on the 1980 census. Thirtee n large counties were chosen and grouped into 34 strata, with 89 sampling locations chosen and further segmented into city and suburban blocks. The detailed approach has been previously described ( Miller, 1973 ) ; McDowell, 1981). Over the 6 years of the study, 39,695 people were sel ected for inclusion, 33,994 were interviewed in their homes and were invited to the mobile examination center for medical examinations and follow up tests. Of these, 78% (30,818) attended the mobile examination centers and 493 others were given examination s in their homes. These home examinations were included in order to gain data from very young children and the very elderly who were unable to visit the mobile examination center. Home examinations included only a subset of the components used at the mobil e examination centers. All participants were interviewed in their homes and given questionnaires to complete. All interviewed participants were invited to attend the mobile examination center for medical exam and further tests. At the mobile examination ce nters data collection began with a household interview and several questionnaires. These were followed by a medical examination, including the collection of blood and urine specimens. Other tests, such as hearing, vision and the cognitive tests, followed. Staff administering the tests included nurses and clinicians for the

PAGE 97

82 medical and clinical examinations and trained interviewers/research assistants for other tests as appropriate. Standard operating procedures for each test are detailed in the NHANES III 1 988 94 Reference Manuals and Reports (1996). 2.2.2 Participants Cognitive tests were administered to a sub sample of adults aged 20 59 years. Assignment to the tests was based on participant number all participants who had an odd participant number were assig ned to the cognitive tests. This produced a sub sample of 5138 non institutionalized civilian men and women aged 20 59 years who completed the computerized cognitive test battery. There were no medical or safety exclusions for the cognitive testing compone nt of NHANES III. However for the purposes of our study, participants were excluded if they did not complete both cognitive tests of interest (SDST and SDLT, described below). Participants were also excluded if they were pregnant, or had evidence of a cond ition known to affect cognitive function, including vitamin B12 deficiency (<174pg/mL), hypothyroidism (TSH <10 uU/mL), reported a past medical diagnosis of stroke, or reported recent alcohol use (>1 drink in preceding 3hrs). Since physical activity was a key independent variable participants were also excluded if they reported significant difficulty walking of a mile. The final sample contained 4515 men and women aged 20 59 years. 2.2.3 Tests and Materials Demographics and Medical H istory All participants were asked to provide information on their age, gender, education, occupation, income, ethnicity and medical history as part of an in depth structured interview. Data were recorded at both the NHANES home assessments and at the mobile examination center. T he poverty income ratio (PIR) was computed as a ratio of family income to a composite

PAGE 98

83 variable comprised of poverty threshold, the age of the family reference person, and the calendar year in which the family was interviewed. Assessment of C ognitive F unct ion Three cognitive tests were administered using the computerized Neurobehavioral Evaluation System (NES). The Neurobehavioral Testing Manual details materials and procedures used for tests of cognitive function (DHHS, 1986). However only two were of int erest for the purposes of this study. The simple reaction time test (SRTT) was omitted as speed of tapping was unlikely to be relevant to this study. A summary of the procedures and general results of the cogni tive tests have been reported by Krieg et al. (2001 ) The tests were conducted in a quiet audiometry room of the mobile examination center trailer. Lighting, temperature, environment and test administration were standardized and distractions minimized. Tests were administered on a Compaq 286 Deskpro portable computer with a keyboard overlay to hide unnecessary k eys and a joystick. Participants were given the option of taking the tests in English or Spanish. Tests were performed in a fixed order. The Simple Reaction Time Test ( SRTT ) is a basic test of motor response speed to a visual stimulus. ( Baker, Chrzan, Park, & Saunders, 1985 ; Krieg et al., 2001 ; Letz, 1989 1990 ) Respondants rested the index finger of their preferred hand on a push button and were asked to push the button as quickly as possible when they saw a solid square (4 x 4 cm) in the center of the computer screen When the button was pushed the square disappeared from the screen, and reappeared between 2.5 5 seconds later. Each respondent was presented with a total of 50 trials. Responses latency from the time of the square appearing to the time the button wa s pushed was recorded for each trial and averaged across all trials to produce a summary response ( Krieg et al., 2001 )

PAGE 99

84 The Symbol Digit Substitution Test (SDST) is a computerized adaptation of the Digit Symbol Substitution Subtest of the WAIS R ( Wechsler, 1981 ) and i s a test of visual motor speed and coding speed ( Krieg et al., 2001 ; Pavlik, Hyman, & Doody, 2004 ) Nine symbols were presented, paired with 9 digits. Participants were presented with a grid that paired 9 different symbols with the numbers 1 9. A similar grid, but with symbols in scrambled order and with the spaces for the digits left blank, was presen ted at the bottom of the screen ( Krieg et al., 2001 ) Respondents were asked to press a numbered key to match the symbols that were presented in scr ambled order lower as quickly as possible The test contained a practice and 4 trials. The time required to enter each digit and the number of errors were recorded. Responses are in se conds, and higher scores represent worse performance. Se rial Digit Learning Test (SDLT) is reported by the NHANES III test manual and by Krieg et al. (2001 ) as a test of learning and memory, but is elsewhere considered a test of attention, concentration or working memory e.g. ( Suhr, Stewart, & France, 2004 ) Participants were asked to learn a series of numbers that were presented slowly (0.6s with 0.6s in between) one at a time, on the screen. After the numbers were presented, participants used the numbered keys to enter the sequence of numbe rs they remembered. The practice set involved 4 numbers. Subsequent trials involved a sequence 8 numbers. The test ended when participants recalled two consecutive number sets. Up to 8 trials were presented, each of which repeated the same sequence of numb ers. Total score was based on number of incorrect digits in each trial, and the number of trials needed for accurate recall of all 8 numbers. SDLT scores ranged from 0 16 and higher scores represent worse SDLT performance. Depression A sub sample of par ticipants aged 18 39 were given additional questions on history of depression and mania symptoms. After being asked questions about whether they had ever

PAGE 100

85 feelin Anthropometric M easures Weight, height, waist circumference, and buttock circumference were measured at the NHANES III mobile examination center (MEC). Waist circumference and buttock ci rcumference were used to calculate the waist hip ratio (WHR). Bioelectric Impedance Analysis (BIA) was also conducted at the mobile examination center. In BIA, a small and painless electrical current is passed through the body and electrical impedance (ohm s) and reactance (ohms) recorded. Body composition was calculated using reactance and impedance ( ( Lukaski, Johnson, Bolonchuk, & Lykken, 1985 ) creating an estimate of percent fat mass (PFM). In addition to the objective anthropomorphic measures, p articipants also provided some information on self report ed weight history. For the purpose of this analysis, responses to were used as an estimate of duration of obesity for participants aged 30 59 years. Responses were given as weight in lbs Laboratory M easures Blood was drawn at the NHANES III mobile examination center for analysis of glucose, insulin, glycated hemoglobin (Hb A1c), triglycerides, cholesterol (HDL, LDL, total), thyroid stimulating hormone (TSH), vitamin B12, C reactive protein (CRP) and insulin like growth factor (IGF) 1. A full description of laboratory procedures used for analysis is provided in the Laborat ory examination file (U.S. Department of Health and Human Services, 1996). Glycated hemoglobin measurements for NHANES III were performed by the Diabetes Diagnostic Laboratory at the University of Missouri Columbia using the Diamat Analyzer System (Bio R ad Laboratories, Hercules, CA). Insulin like growth factor 1 was analyzed in 2002 using a sub set of stored samples (U.S. Department of Health and Human Services, 2003).

PAGE 101

86 Self Reported P hysical A ctivity Self reported physical activity was recorded as part of the adult household questionnaire. Interviewers recorded responses on Likert type scales to 6 questions on physical past month, how often did you jog or run variety of different physical activity options, including: ride a bicycle, swim, do aerobics, go dancing, garden or do yard work, lift weights, or participants could specify other activities. Each activ ity was assigned an intensity rating (e.g. Jog or run intensity rating [mets]). Healthy Eating Index Participants responded to an oral food frequency questionnaire in which they indicated how often they ate various types of food. Researchers at the Nationa l Center for Health Statistics compiled dietary responses into a Healthy Eating Index (HEI) for each participant (DHHS, 1996). The HEI is the sum of 10 dietary sub scales, weighted equally, each of which ranges from 0 10. The HEI is a measure of the qualit y of dietary composition, rather than caloric intake. High HEI scores indicate dietary intakes close to the recommended ranges or amounts. Social S upport The Adult Household Questionnaire asked participants about their frequency of social contact with fri ends, family, neighbors, clubs and organizations or religious organizations. Smoking S tatus Participants were asked a wide range of questions about their use of tobacco products. Based on their responses participants were classified as current smokers if they endorsed questions related to current cigarette, cigar or pipe use, former smokers if they endorsed questions related to former (but not current) cigarette, cigar or pipe use, and never smokers if they indicated they had never made significant use of these tobacco products.

PAGE 102

87 2.2.4 Data Analysis Due to the complex, multi stage survey design used in NHANES III (described above), data were analyzed using SPSS 21 Complex Samples (IBM). The standardized weighting and estimation methodology for NHANES III are ava ilable for download on the NCSH website. Strata (SDPSTRA6), cluster (SDPPSU6), and sample weights (WTPFCNS6) specific to the cognitive subsample were applied. Table 2.2 1 Pre existing categorical variables already created in NHANES dataset. Variable Gender Male~ Female Ethnicity Non Hispanic white~ Non Hispanic Black Mexican American Other Depression now Yes No~ Currently experiencing depressed symptoms Not currently experiencing. Note. ~ Refere nce group in regression analyses. Calculation of New V ariables To facilitate analysis, calculation of new categorical variables was performed where necessary. Categories of BMI were created according to the World Health Organization criteria (WHO, 2006) as can be seen in Table 2.2 2 overweight if BMI was 25 29.9, and a healthy weight if BMI was between 18.5 and 24.9. A dichotomous variable for central obesity (yes/no) was also created using criteria used by the World Health Organization in their definition of the metabolic syndrome (WHO, 1998). Central

PAGE 103

88 Duration of obesity was estimated using partic estimated BMI 10 years ago. This was done with the assumption that for most perso ns aged 30 59 years height would not have changed much over the course of 10 years. Participants younger than 30 at the time of the NHANES study were excluded from this sub analysis since they may not have reached their adult height 10 years prior. The aut hor then created a categorical variable to classify participants into 3 groups: 1) persons who were obese at both time points, 2) participants who were not obese at time 1, but were obese at time 2, and 3) participants who were not obese at either time poi nt. Obesity was assessed using BMI alone, as it was not possible to estimate WHR 10 years ago from the data available. Categorical variables for physical activity, social support and smoking status were also created. The estimate of self reported frequenc y of physical activity was recorded as follows. reported frequency of leisure time physical activity and the activity intensity rating encoded in the NHANES dataset (mets) we categorized participants as either sedentary (engagi ng in activity of at least 3 mets <4 times per month), moderately active (activity of at least 3 mets 5 19 times per month) or active (activity of at least 3 mets >20 times per month). The dichotomous variable for s ocial support was categorized on the basi s of responses to two questions on frequency of contact with family or friends. Persons who reported they were not married and had <1x per week on the phone with family of friends were Smoking status was calculated based on responses to multiple questions on tobacco use. Participants who reported that they smoke cigarettes, pipe, cigars or use chewing tobacco or snuff now were classified as current smokers

PAGE 104

89 Dummy variables were created for all categorical variables. Reference groups are indicated in results tables below. Cognitive measures, a ge, education and all biomarkers (HbA1c, CRP, IGF 1, BP, HDL) were entered as continuous variables, consistent with other analyses of similar data.

PAGE 105

90 Table 2.2 2 New categorical variables created for this analysis. Variable Definition Central obesity Yes No~ Men Men WHR < 0.9, Women WHR < 0.85 BMI categories Healthy weight ~ Overweight Obese BMI 18.5 24.9 BMI 25 29.9 Obesity duration 10+ years <10 years Not obese ~ Obese at interview and 10 years before Obese at time i nterview only Not obese at 1 or 2 Physical activity Sedentary ~ Moderately active Active 4 19 times per month Smoking status Current smoker Former smoker Never smoked~ Use cigarettes, pipe, cigars, chewin g tobacco or snuff now Not current, but have ever smoked 100+ cigarettes, 20+ pipes etc. Not current or former. Social support Little Some~ Not married, < 1 phone contact with family or friends per week. Married or greater than 1 contact with family or friends per week. Note. ~ Reference group in regression analyses. A pproximately 75% of participants meeting inclusion criteria for this study did not report their income, preventing the calculation of the poverty income ratio. This variable wa s

PAGE 106

91 therefore excluded from analysis. Only 476 of participants with responses to depression questions ( depressed or not ) met our inclusion criteria. To avoid loss of statistical power, a separate sub sample analysis was run for persons with depression data. Regression D iagnostics Before conducting regression modeling, the distribution of each variable was examined visually using frequency histograms and scatter plots were used in the assessment of outliers. Seven part icipants with extreme WHR (e.g. 2.0) greater than 3 standard deviations (sd) from the mean were excluded from analyses due to their potential to unduly affect regression results. Regression diagnostics were conducted to check for linearity, normality and m ulticollinearity. To check that the distribution of the dependent variables was normal, with a constant variance, for each combination of values of the independent variables and covariates, scatter plots of predicted versus observed values and residuals pl ots (studentized, studentized deleted) were assessed. Results of the symbol digit substitution test (SDST) were positively skewed and scores required a natural log transformation to achieve better approximation of a normal distribution. Pre model screening of scatter plots was conducted to determine the linear association between each covariate or independent variable and cognitive outcome measure. Checks for multicollinearity were made using tolerance scores, and correlation coefficients used to check biva riate correlations between key variables. Regression M odel B uilding Separate multiple regression analyses were used to test the associations between each of the 2 cognitive variables of interest (SDST or SDLT) and the 3 different measures of adiposity (B MI, WHR or % fat mass) to determine if central obesity or global obesity differed in their association with cognitive function. Regression model building followed the purposeful

PAGE 107

92 hierarchical model building approach described by Kleinbaum et al (2008). Firs t, sociodemographic variables (age, gender, ethnicity, education) were entered as a set. Second, measures of adiposity were entered separately to allow for comparison of models: BMI alone, WHR alone or percent fat mass alone. Significance of the parameter estimate, and the contribution to proportion of variance explained, were evaluated. Keeping these variables in the model, the health related behavior (PA, diet, smoking, social support) variables were each added to the model, to determine main effect. Next interaction terms between behavioral variables and the relevant adiposity variable for that model (e.g. PA x WHR) were entered to assess potential moderation of the association between adiposity and cognitive score. Finally, the model was adjusted for bi omarkers that are potential mediators of an association between obesity and cognitive function, including CRP, HbA1c, blood pressure, and cholesterol. 2.3 R esults 2.3.1 Sample C haracteristics The final sample included 4515 adults aged 20 59 years (mean=36.95 yea rs ). As seen in Table 2.3.1 50.5% were women, and the sample included 77.2% non Hispanic whites, 11% African Americans, 5.2% Mexican Americans and 6.5% other ethnicities. The average level of education in the sample was 12.86 years (sd = 0.08). Obese par ticipants (BMI>30) represented 22.5% of the sample, while 31.7% were overweight (BMI 25 29.9) and 45.9% were in the healthy weight range (BMI 18.5 24.9) (WHO, 2006). Consistent with the hypothesis that central obesity is distinct from overall obesity, more than half of the sample (53.8%) met criteria for central obesity (WHO, 1998) based on waist hip ratio (WHR), including 14.9% in the healthy BMI range, and 22.2% who were overweight according to the BMI. Mean percent fat mass and WHR for each BMI categ ory can be seen in Table2.3.1 Correlations between BMI, percent fat mass and waist hip ratio are displayed in Table 2.3 2 Increasing age was associated with higher waist

PAGE 108

93 hip ratio (p<0.000), higher BMI (p<0.000) and higher proportion of fat mass (p<0.000). Being female was associated with greater adiposity (p<0.000). Higher education was associated with less adiposity in each of these measures (p<0.000). As previously described, participant SES (poverty income rat io) was not assessed due to high non response rates for the income variable and the potential for non response bias. Quality of diet, as estimated by the Healthy Eating Index (HEI) score was 62.39 across all participants. Higher scores indicate better qua lity diet. As shown in Figure 2.3.1 mean HEI scores were similar for the healthy weight, overweight and obese groups. The correlation between quality of diet and frequency physical activity was low (R 2 = 0.049). Current smokers composed 35.5% of the sample former smokers 20.9% and never smokers made up 43.6% of the sample. Among the sample of participants who completed the cognitive tests, 476 participants aged 20 39 completed the Diagnostic Interview Schedule that assessed depression. Of these 129 perso ns reported that they were experiencing a spell of feeling low now.

PAGE 109

94 Table 2.3 1 NHANES III sample characteristics by BMI category. Characteristic Total Healt hy Weight Overweight Obese Sample size (n) 4515 1842 1463 1210 Age (years) 37.05 34.63 38.74 39.59 Education (years) 12.85 13.13 12.79 12.35 Gender (%) Male (%) Female (%) 50.2 49.8 20.9 24.9 19.5 12.2 9.8 12.8 Ethnicity (%) Non Hispanic White (%) No n Hispanic Black (%) Mexican American (%) Other (%) 76.9 11.1 5.3 6.7 36.5 4.3 1.8 3.0 24.3 3.4 2.0 2.0 16.1 3.3 1.5 1.7 Adiposity BMI (continuous) 26.67 22.21 27.11 35.12 % Fat mass 30.14 25.90 30.44 38.45 WHR (continuous) 0.89 0.85 0.92 0.94 C entrally obese (%) 54.8% 14.9% 22.2% 17.7% Cognitive function Ln (SRTT) ^ 3.08 3.04 3.11 3.14 SDLT^ 4.36 3.77 4.68 5.11 Ln(SDST) (s) ^ 3.08 3.04 3.11 3.12 Health related behaviors Physical activity Active (%) Some activity (%) Sedenta ry (%) 56.6 22.9 20.5 29.0 9.1 7.7 17.9 6.7 7.0 9.7 7.1 5.8 Healthy Eating Index 62.39 62.83 62.86 61.12 Smoking status Never smoked (%) Former smoker (%) Current smoker (%) 43.4 21.3 35.3 19.6 8.9 17.2 13.9 6.8 11.0 9.9 5.5 7.1 Biomarkers HbA1c ( %) 5.23 5.07 5.26 5.54 CRP (mg/dL) 0.37 0.31 0.34 0.54 IGF 1 (ng/mL) 286.84 300.72 285.11 260.04 Systolic blood pressure (mmHg) 117.87 113.70 119.33 124.29 HDL cholesterol (mmol/L) 1.30 1.42 1.23 1.14 Depressive symptoms now (n) 24.2 14.1 3.3 6. 8 Notes: ^ Higher scores indicate worse performance. Notes: ^ Higher scores indica te worse performance. Ln(SDST) is t he natural l og transformation of SDST; CRP C reactive protein. BMI Body Mass Index; HbA1c Glycated hemoglobin A1c; IGF 1 Insulin like growth factor 1; HDL High Density Lipoprotein; SDLT Serial Digit Learning Task; SDST Symbol Digit Substitution Test.

PAGE 110

95 Based on self reported frequency and intensity of leisure time physical activity 56.4% of the sample were active (20 or more times a month), 22.9% were moder ately active (4 19 times a month), and 20.7% met study criteria for being sedentary (<4 times per month). Correlations between self reported physical activity and measures of adiposity were low, as shown in Table 2.3 2 Table 2.3 2 Correlation between anthropometric measures. Body Mass Index Waist Hip Ratio Percent fat mass Body Mass Index 1 0.412 0.645 Waist Hip Ratio 0.412 1 0.019 Percent fat mass 0.645 0.019 1 Physical activity 0.119 0.084 0.183 Healthy Eating Index 0.013 0.057 0.025 Smoking status 0.071 0.159 0.173 2.3.2 Regression Model building process Regression analyses were run separately for the SRTT SDLT and SDST. The regression model build ing process involved the following 5 steps for each of these models. Model 1. In the first step all the demographic variables were entered together. These variables were retained in the model irrespective of whether they were significant predictors of cog nitive function. Model 2. Next, a measure of adiposity was added to the model, and its association with the outcome variable assessed. If there was no association the model building process ended at this step for that particular measure of adiposity. If i t was significant the model building process continued on to model 3. Model 3. Next, the behavioral variables were added to the model to create a main effects model. Model 4 Whether or not any of the main effects of those behavioral variables were sign ificant predictors of cognitive function, the interaction terms were also added to the model.

PAGE 111

96 Model 5. Finally, the potential mediating biomarkers were added to the model. 2.3.3 Regression M odels of the Simpl e Reaction Time Test The SRTT is a test of simple m otor response speed. Higher scores on the SRTT test represent longer time (seconds) to task completion, and hence worse performance. The distribution of the variable was skewed, so was adjusted using a natural l og transformation All results below refer to the log transformed values. SRTT Model 1: Demographic V ariables As has been reported elsewhere with this dataset (Krigg et al, 2001), after log transformation to increase approximation towards normality, the SRTT (ln(SRTT)) scores were significantly ass ociated with demographic factors. Increasing age was associated with worse SRTT performance, (p<0.001 ). Lower education was related to worse performance on the SRTT (p<0.000). Gender was also a significant predictor of SRTT score, with women performing the test approximately 0.05 ms faster than men. African Americans and Mexican Americans performed slightly slower on the SRTT than non Hispanic whites. However it was not possible to conduct planned adjustments for socio economic status using the poverty inco me ratio due to large amounts of missing data I t is likely that this is an artifact attributable to other unmeasured variables such as socio economic status. All subsequent regression analyses were adjusted for age, sex, education and ethnicity, which we re entered in the first step of the regression model building process. The model with all four of these demographic variables entered simultaneously accounted for 9.1 % of the variance in SRTT score. Women had faster reaction times than men

PAGE 112

97 SRTT Model 2: A diposity Measures To determine whether increased adiposity was associated with increased reaction time, regression analyses were performed for 3 separate models with different measures of adiposity. Model 2a assessed the association between BMI and SRTT Model 2 b assessed the association between percent body fat and SRTT Model 2c assessed the association between WHR and SRTT The results of the three models are presented in Table 3 .3.3.. All models are adjusted for age, education, gender and ethnicity. Model 2a (BMI): As can be seen in Table 3 .3.3., higher BMI was associated with higher SRTT scores, indicating poorer performance. Controlling for demographic variables, a 1 point increase in BMI increased SRTT score by 0.001 (SE=0.001, p=0.040). The mode l including BMI accounted for only 9.3% of the variance in SRTT adding only 0.2% to the demographics only model. Model 2 b (%fat): In contrast to BMI, higher proportion of body fat was not associated with increased SRTT scores ( 01 p=0. 173 ). Model 2c (Central obesity): Increasing WHR was associated with higher SRTT scores, indic SE= 0.050, p=0.034 ). The range for WHR is between 0.7 1.5, therefore for every 0.1 point increase in W HR, the SRTT score increased by 0. 011 points The regression m odel with WHR accounted for 9.2 % of the variance in SRTT score, adding only 0.1% to the demographics only model.

PAGE 113

98 Table 2.3 3 Multiple Linear R egression models of the association between measures of adiposity and SRTT. ^ Variable BMI % Fat Mass WHR P P P Education .010 .001 .000 .009 .002 .000 .009 .001 .000 Age .001 .000 .006 .001 .000 .005 .001 .000 .047 Gender Male~ Female .062 .008 .000 .056 .008 .000 .073 .010 .000 Ethnicity Non Hispanic Wh ite~ Non Hispanic Black Hispanic Other .045 .036 .022 .008 .009 .017 .000 .000 .188 .045 .036 .020 .008 .009 .017 .000 .000 .232 .050 .036 .027 .007 .009 .018 .000 .000 .137 Adiposity measure .001 .001 .040 .001 .001 .173 .110 .050 .034 Model R 2 0.093 0.092 0.095 Note: ^ Natural log of milliseconds. Higher scores indicate worse performance. Each model was run separately. ~ Coded as the reference group for the dummy variable. Other scores within this variable are relative to th is group. SRTT Model 3: Is the association Between O besity and SRTT M oderated by H ealth B ehaviors? Health related behaviors were added as a block to determine whether they moderated the association between adiposity and SRTT performance. This was conduct ed for the BMI and WHR models only, since % fat mass was not associated with SRTT performance. SRTT Model 3a (BMI): After adjusting for quality of diet, smoking status, social support and physical activity BMI was no longer significantly associated with S RTT score (p=0.075). Neither quality of diet nor smoking status was associated with SRTT scores. Being sedentary was associated with HEI score 0.022, SE=0.007, p=0.005). Having little social support was also associated with worse SRTT performance. The h ealth b ehavior model accounted for only 9.9 % of the variance in SRTT. SRTT Model 3 c (central obesity): After adjusting for health related behaviors, WHR 0.102, SE=0.05 p=0.0 47), though the model accounted for only 10.2 % of the variance in SRTT. As with BMI, quality of diet (HEI) was not

PAGE 114

99 associated with SRTT 0.000, SE=0.00, p=0.212 ) Nor was smoking status associated with SRTT performance. However being sedentary 0.0.025, SE=0.0.008, p=0.00 2) or doing some activity 0.012, SE=0.008, p=0.032) was associated with worse performance relative to active participants. Having little social support was not associated with SRTT performance. SRTT Model 4: Interaction M odels of H ealth B ehaviors : To assess the potential for moderating effects of health related behaviors on the association between adiposity and SRTT performance, interaction terms between each of the behaviors and measures of adiposity were entered into the respective models. Each inter action term was entered separately and together with the other terms. To facilitate interpretation of interaction terms, interaction models were run using categorical variables for the measures of adiposity that is interaction models were run for catego rical measures of BMI and central obesity. SRTT Model 4a (BMI): No interactions between BMI categories (healthy, overweight, and obese ) were significant predictors in the model. Similarly, no interaction terms were statistically significant for Model 3 ( Central obesity) Interaction terms for health behavior were therefore excluded from the ongoing model building process.

PAGE 115

100 Table 2.3 4 Multiple linear regression model of the association between central obesi ty (WHR) and reaction time (SRTT ).^ Central obesity P Age .001 .000 .053 Education .009 .002 .000 Gender Male~ Female .067 .009 .000 Ethnicity Non Hispanic White~ Non Hispanic Black Hispanic Other .048 029 .024 .008 .010 .019 .0 00 .005 .208 WHR .102 .050 .047 Healthy Eating Index Physical activity Sedentary Moderately active Active~ .022 .018 .007 .008 .005 .028 Social support Little Some .031 .014 .036 Model R2 = 9.6 Notes: ^Natural log transformed. ~ Refe rence group. SRTT Model 5: Potential F actors M ediating the A ssociation Between A diposity and SRTT To test the hypothesis that the association between adiposity and cognitive function is mediated by physiological mechanisms potentially linked to cognitive function and adiposity, biomarkers were next added to the model. Building on the behavioral model ( containing no interaction term s) biomarkers were added to the model to determine whether the associations between adiposity and SRTT was mediated by physio logical factors. Biomarkers added included HbA1c, CRP, IGF 1, systolic blood pressure, and HDL cholesterol. N one of the variables were associated with change in SRTT score, for either adiposity model (BMI or WHR), nor did their inclusion, or removal, appre ciably alter the direction or magnitude of the associations already found between SRTT and other variables.

PAGE 116

101 2.3.4 Regression M odels of the Symbol Digit Substitution Test The SDS T is a test of visuomotor speed and coding speed ( Krigg et al, 2001; Wechsler, 1981 ). Higher scores on the SDST test represent longer time (seconds) to task completion, and hence worse performance. SDST Model 1: Demographic V ariables As has been reported elsewhere with this dataset (Krigg et al, 2001), after log transformation to increas e approximation towards normality, the SDST (ln(SDST)) scores were significantly associated with demographic factors. Increasing age was associated with worse cognitive test performance, with age independently accounting for 19% of the variance in SDST (p< 0.000). Lower education was related to worse performance on the SDST, accounting for 23% of the variance in SDST (p<0.000). Gender was also a significant predictor of SDST score. As can be seen in Table 2.3 5 SDST test scores differed by ethnicity, but since it was not possible to conduct planned adjustments for socio economic status using the poverty income ratio due to large amounts of missing data it remains likely that this is an artifact attributable to other unmeasured variables such as socio economic status. All subsequent regression analyses were adjusted for age, sex, education and ethnicity, which were entered in the first step of the regression modeling. The model with all four of these demographic variab les entered simultaneously accounted for 46.6% of the variance in SDST score. SDST Model 2: Is A diposity R elated to SDST P erformance? To t est the hypothesis that higher adiposity, particularly central adiposity, is related to cognitive function in midlife measures of adiposity were added to the demographic factor model described above. T hree separate models with different measures of adiposity were run to determine whether central obesity (WHR) showed a different association with cognitive function than gl obal obesity.

PAGE 117

102 SDST Model 2a (BMI): The inclusion of BMI in the model had no significant effect on variance explained, nor was the coefficient significantly associated with SDST test performance. SDST Model 2 b (%fat): The inclusion of percent fat mass ha d no significant effect on variance in SDST explained, nor was the coefficient significantly associated with SDST test performance. SDST Model 2c (Central obesity): As with other measures of adiposity, WHR was not significantly associated with SDST test p erformance, and its inclusion in the model did not alter variance explained. SDST Model 3. Do B ehavioral F actors M oderate the A ssociation with O besity? Regression analyses were planned to investigate the potential moderating effect of health related behavi ors on the association between obesity and cognitive function in midlife. However s ince performance on the SDST was not associated with any measure of adiposity potential moderators of the effect are not reported.

PAGE 118

103 Table 2.3 5 Multiple linear regression models of the association between measures of adiposity and measures of cognitive function. SDLT (total score) or SDST (natural log of seconds). Each model was run separately. Notes: ^Higher scores indica te worse p dummy variable. Other scores within this variable are relative to this group. 2.3.5 Regression M odels of the Serial Digit Learning Task The SDLT requires participants to learn and recall a sequence o f 8 numbers. Total score was based on number of incorrect digits in each trial, and the number of trials needed for accurate recall of all 8 numbers. SDLT scores ranged from 0 16 and higher scores represent worse SDLT performance. SDLT Model 1: Demograph ic V ariables As has been reported elsewhere with this sample (Krigg et al, 2001), performance on the SDLT was significantly associated with demographic factors. Lower education was related to

PAGE 119

104 poorer performance, accounting for 16.8% of the variance in SDLT (p<0.000). While age accounted for only 3.2% of the variance in SDLT with the sample size this was statistically significant (p<0.000). Gender was not a significant predictor of SDLT, but was retained in the model due to its potential for association with independent variables to be added in subsequent steps. As can be seen in Table 2.3 5 SDLT scores were related to ethnicity, but due to missing data in variables such as the poverty income ratio it was not possibl e to adjust for socio economic status. All subsequent regression analyses were adjusted for age, sex, education and ethnicity. Together the demographic variables accounted for 24.4% of the variance in SDLT. SDLT Model 2: Is There an A ssociation B etween A d iposity and SDLT P erformance? To determine whether central obesity or global obesity have different associations with cognitive function, three separate models were run, each with a different measure of adiposity. Model 1 assessed the association between B MI and SDLT. Model 2 assessed the association between percent body fat and SDLT. Model 3 assessed the association between WHR and SDLT. The results of the three models are presented in Figure C.3.3.. All models are adjusted for age, education, gender and ethnicity. SDLT Model 2a (BMI): As can be seen in Figure C.3.3., higher BMI was associated with higher SDLT scores, indicating poorer performance. Controlling for demographic variables, a 1 point increase in BMI increased SDLT score by 0.029 (SE=0.013, p =0.035). The model including BMI accounted for 24.9% of the variance in SDLT. SDLT Model 2 b (%fat): Higher proportion of body fat was associated with increased SDLT scores, such that each 1% increase in fat mass was associated with a 0.035 point increase in SDLT score (SE=0.013, p=0.013). The total model accounted for 24.8% of the variance in SDLT. SDLT Model 2c (Central obesity): Increasing WHR was associated with higher SDLT scores,

PAGE 120

105 between 0.7 1.5, therefore for every 0.1 point increase in WHR, the SDLT score increased by 0.556 points (SE=0.19, p=0.005). The regression model with WHR accounted for 25.1% of the variance in SDLT score. Do H ealth B ehaviors M oderate the A ssociation B etween A diposity and SDLT? To test the hypothesis that health related behaviors moderate the association between adiposity and cognitive function, h ierarchical regression model building was conducted to determine whether the association between SDLT performance and adiposity was moderated by interactions with modifiable health related behaviors. SDLT Model 3. Main E ffects M odels of H ealth R elated B ehaviors and SDLT SDLT Model 3a (BMI): After adjusting for quality of diet, smoking status, social support and physical activity BMI was no longer associated with SDLT score (p=0.081). By contrast, 0.017 SE=0.006, p=0.008). Former smokers showed better SDLT performance than persons who had 0.888, SE=0.231, p=0.000). No significant association was found for current smokers. Having little social support was not associated with SDLT perform ance. Physical activity was not associated with SDLT performance. The health behavior model accounted for 25.8% of the variance in SDLT. SDLT Model 3b (%fat): In contrast to BMI, percent fat remained a significant predictor of SDLT performance after adjus ting for health The model accounted for 26.0% of the variance in SDLT score. SDLT Model 3 c (central obesity): After adjusting for health related behaviors, WHR remained a significant predictor of SDLT score accounted for 26.5% of the variance in SDLT. As with BMI, better quality of diet (higher HEI) was 0.016, SE=0.006, p=0.005). Former smokers

PAGE 121

106 showed better SDLT performanc 0.858, SE=0.237, p=0.001). No significant association was found for current smokers. Having little social support was not associated with SDLT performance. Physical activity was not associated with SDLT performance. SDLT Model 4: Interactions Between H ealth R elated B ehaviors and SDLT To assess the potential for moderating effects of health related behaviors on the association between adiposity and SDLT performance, interaction terms between each of the behaviors and measures of adiposity were entered into the respective models. Each interaction term was entered separately and together with the other terms. To facilitate interpretation of interaction terms, interaction models were run using categorical variables for t he measures of adiposity. Interaction models were therefore run only for categorical measures of BMI and central obesity. SDLT Model 4a (BMI): No interactions between BMI categories (healthy, overweight, and obese ) were significant predictors in the model SDLT Model 4c (Central obesity): Interactions between quality of diet, smoking status or social support did not contribute significantly to the model, nor did they significantly change the significance of other variables when removed. By contrast, the i nteraction between central obesity and physical activity was significant (p=0.05). The interaction terms that were not statistically significant were removed from the model, as was the main effect of social support, which remained non significant. The phys ical activity interaction model accounted for 25.8% of the variance in SDLT, and is presented in Table 2.3 6 As in the main effects model, better quality of diet remained associated with better (lower) SDLT scores 0.017, SE=0.006, p=0.009). Similarly, former smokers continued to show better SDLT performance than persons who had 0.859, SE=0.226, p=0.000).

PAGE 122

107 Figure 2.3 1 Interaction between centr al obesity and physical activity for SDLT performance.

PAGE 123

108 Table 2.3 6 Multiple l inear regression model of the association between central obesity and SDLT (total score). Central obesity Age Gender Male F emale Ethnicity Non Hispanic White Non Hispanic Black Hispanic Other Education Central obesity Healthy Eating Index Smoking status Never smoked Former smoker Current smoker Physical activity Sedentary Moderately active Active Interaction PA x centr al obesity Centrally obese & sedentary Centrally obese & moderate Centrally obese & active P 0.086 0.847 0.000** 0.054 0.176 0.759 1.78 2.369 2.412 0.195 0.328 0.454 0.000** 0.000** 0.000** 0.544 0.037 0.000** 0.017 0.006 0.000** 0.859 0.341 0.226 0.211 0.000** 0.112 0.050* 0.714 2.063 0.044 0.971 3.331 0.002 Model R2 = 0.258 Note: ^ Higher scores indicate worse performance. Statistically significa nt at p<0.05. ** Statistically significant at p<0.001. Closer investigation of the interaction between central obesity and physical activity, shown in Figure D.3.1, indicated that among participants who were not centrally obese, those who were sedentary moderately active or active had similar SDLT scores, though being active was associated with slightly better (lower) SDLT performance. However among participants who were centrally obese the picture was different. Persons who were centrally obese and se dentary showed the worst overall SDLT scores. By contrast persons who were centrally obese but

PAGE 124

109 moderately active or active had SDLT scores similar to non obese participants. Mean SDLT scores and standard errors for these groups are depicted in Table 2.3 7 Table 2.3 7 SDLT scores by physical activity and obesity (mean (SE)). Sedentary Moderately active Active Centrally obese 6.45 (0.388) 4.37 (0.225) 4.58 (1.76) Not centrally obese 4.52 (0.314) 3.89 (0.364) 3.42 (0.181) SDLT Model 5. Potential F actors M ediating the A ssociation B etween A diposity and SDLT To test the hypothesis that the association between adiposity and cognitive function is mediated by physiol ogical mechanisms potentially linked to cognitive function and adiposity, biomarkers were next added to the model. Building on the behavioral model containing the interaction term, biomarkers were added to the model to determine whether the associations be tween behavior and cognition was mediated by physiological factors they can affect. Biomarkers added included HbA1c, CRP, IGF 1, systolic blood pressure, and HDL cholesterol. In each adiposity model none of the variables were associated with change in SDLT score, nor did their inclusion, or removal, appreciably alter the direction or magnitude of the associations already found between SDLT and other variables. 2.3.6 Sub analysis of the D uration of O besity The following sub analysis was performed to explore whet her duration of exposure to obesity has a significant association with cognitive performance in midlife. It is possible that the chief difference between the results found by observational studies of midlife overweight and cognitive decline, compared to st udies of overweight in older adults, is due to the difference in duration of exposure to the harmful effects of obesity. If so, then persons who are obese for longer periods of time during early and mid adulthood may show worse cognitive performance

PAGE 125

110 than p ersons of the same weight who were exposed to a shorter duration of obesity. Although NHANES III was a cross sectional study, it did include some retrospective self report data on weight 10 years prior to the interview, as described in the methods and data analysis sections above. Sub Sample S election and C haracteristics To estimate the effects of different lengths of exposure to obesity, participants were stratified into four groups. Group 1 contained persons who, based on their estimated BMI 10 years ago fluctuated between those times, but plausible that they remained obese during the entire 10 year period. Group 2 included persons who were not obese 10 years prior to t he interview, but had subsequently gained weight and become obese. Group 3 included participants who reported obese weights 10 years ago, but were not obese at the time of the interview, representing probable weight loss at some point in the preceding 10 y ears. Group 4 included participants who were not obese at either time point. Participants aged 20 29 at the time of the NHANES III interview were excluded from the sub sample because 10 years earlier they would not necessarily have reached their full adul t height or weight. After accounting for this, and for missing data in the self report variable, the sample available for this sub analysis incl uded 2363 participants. As shown in Figure 2.3 2 12.6% were obese at both time points (group 1), 14.8% became obese within the past 10 years (group 2), only 2% were obese 10 years ago but subsequently lost weight (group 3), and 70.4% were not obese at either time (group 4). As would be expected, there was a strong correlati on between this variable and BMI at time of intervie w (R= 0.788, p<0.000).

PAGE 126

111 Figure 2.3 2 Number of participants in each strata of duration variable. Duration of Obesity and Regression M odels of the SRTT The regression model building process matched that used for the full sample and described abov e. For the duration sub sample, results of SRTT Model 1 demographics only model (including age, education, gender and ethnicity) were similar to those of the larg er NHANES sample. Each of these demographic variables was significantly associated with SRTT scores, and the model accounted for 10% of the variance in SRTT. However SRTT Model 2 showed that, in contrast to the full sample, BMI was not a significant predi ctor of SRTT in the sub sample. This may be due to the systematic removal of participants aged 20 29 years from the sub sample, or it may be due to the loss of power with the smaller sample size. The BMI model accounted for 10.2% of the variance in SRTT which is actually slightly more than the BMI model was able to account for in the full sample, suggesting that the issue may have been one related to power and sample size, particularly given the very small magnitude of effect for SRTT seen in the full s ample model. Running a separate model collinearity) produced similar results. The different groups did not show significantly different

PAGE 127

112 scores on the SRTT. This remained t small group of people who reported weight loss, to check whether the small cell size for this group was having an undue effect. Duration of Obesity and Regression M odels of the SDST When including du ration of obesity, the SDST model 1 ( the demographics only model ) produced results similar to those of the larger NHANES sample described previously for SDST. All of the demographic variables were significantly associated with SDST performance, and the mo del accounted for 43% of the variance. As was found in Model 2 for the full sample for SDST, Model 2 for the sub sample with duration of obesity showed that BMI was not a significant predictor of SDST in the sub sample The model with BMI added did not a ccount for any more variance than the demographics alone produced similar results. Differences in duration were not significantly associated with SDST score. This re people who reported weight loss. Duration of Obesity and Regression M odels of the SDLT Re sults of the demographics only M odel 1 for the SDLT were similar to those of the larger NHANES sample. All demographic variables except gender were significant predictors of SDLT. The model accounted for 27.8% of the variance in SDLT. However in contrast to the results obtained from the full sample, BMI was not a significant predictor of SD LT in Model 2 for this sub sample This may be due to a loss of power, or to the systematic removal of participants aged 20 29 years from the sub sample. After adding BMI, the model only accounted for 27.9% of the variance in SDLT. Running a separate model

PAGE 128

113 results. No level of the duration of obesity variable was associated with SDLT performance. This remained the case after recoding the duration variable to remove the group of people who lost weight over the last 10 years and were no longer considered obese. 2.4 D iscussion Consistent with previous research suggesting a link between midlife obesity and cognitive decline later in life, greater adiposity in early and mid adu lthood was associated with more difficulty learning and recalling a list of 8 numbers presented repeatedly, after adjusting for age, education, gender and ethnicity. Significantly, the association was most apparent in a test of working memory and attention /concentration the SDLT cognitive test ( ( Pavlik et al., 2004 ; Suhr et al., 2004 ) Some evidence of a small but statistically significant association between adiposity (BMI, WHR) and simple reaction time was also apparent, however the association disappeared after adding health behaviors to these models. T he Symbol Digit Substitution Task (SDST), wh ich tested the speed with which participants matched a single digit with a symbol presented on the screen and taps attention and psychomotor speed ( Pavlik et al., 2004 ; Suhr et al., 2004 ; Wechsler, 1981 ) was not associated with markers of adiposity. This difference suggests that the finding is more than just a spurious result. While learning and memory are not the only cognitive functions that have been linked to excess body weight in epidemiological research ( Wolf et al., 2007 ) they are more commonly implicated than any other domain. Furthermore, animal models of interventions that affect weight and metabolism, such as physical activity, calorie restriction and intermittent fasting demonstrate improvements in learning and memory ( Mattson et al., 2003 ) Central obesity (WHR) was more closely related to performance on both the SRTT and the SDLT. While both global obesity and cent ral obesity were associated with SDLT test performance, the magnitude of the association between SDLT score and central obesity (WHR)

PAGE 129

114 was greater than that found for either BMI or percent fat mass. Based on these regression results, when holding other thi ngs constant, a man wi th a waist circumference of 40.5 inches and buttock circumference of 45 inches (WHR=0.90) would be predicted to score 0.56 points worse on the SDLT than a man with a waist circumference just half an inch smaller (40. 0 inches, WHR=0.8 9). By contrast, a n 180lb man with 26% fat mass would score just 0.035 points worse than a similar man carrying 0.45lbs less fat (26% fat mass). Thus the difference of half an inch in waist circumference is linked to greater differences in SDLT than the ga ining of almost half a pound of fat mass. This difference between the measures of adiposity reinforces the value of adding alternate measures of obesity to future research studies. As previously described, studies using BMI as the sole measure of obesity w hen studying the relationship to cognitive outcomes miss this distinction and with it valuable information on the effects of adiposity on neurocognitive health. It is also consistent with epidemiological studies described in Study 1 found that midlife ce ntral obesity was more often linked to cognitive decline and dementia than late life central obesity. The reasons for a link in midlife certainly warrant further investigation. Intervention studies with the potential to draw causal inferences would be part icularly valuable in this regard. Different measures of adiposity also showed different associations with health behaviors. Percentage fat mass and WHR remained significant predictors of SDLT performance after controlling for health related behaviors, but this was not true for BMI. Quality of diet and being a former smoker attenuated the association between BMI and SDLT performance. Although the magnitude of the associations, and the differences between the measures of adiposity, were small they are potent ially significant The association was found in adults aged 20 59 years, after controlling for age. This is consistent with longitudinal studies that found an association between obesity in midlife and later cognitive decline ( Cournot et al., 2006a ;

PAGE 130

115 Whitmer et al., 2005 ; Whitmer et al., 2007 ) Furthermore, unlike research with older adults, the cross sectional association in midlife is unlikely to be confounded by undiagnosed dementia. Although it is not possible to conclude from these cross sectional data that a gain in waist circumference would lead to worse SDLT scores, a causal role for obesity in cognitive decline could have significant cumulative effects over a prolonged period of time. This would also be consistent with the prior research reporting an association between central obesity in midlife and cogniti ve decline or dementia later in life ( West & Haan, 2009 ) The association between duration of obesity and cognitive function could not be determined from the data o btained. This was at least partly due to the sample and power available, or possibly inaccuracies of 10 year recall in weight. In the sub sample of participants aged 30 59 who had self report data on their weight 10 years ago, and who met all other inclusi on criteria, the regular adiposity models did not show the significant effects that were found in the full sample. It is therefore not surprising that no association was observed between groups likely exposed to differing duration of obesity. However this null finding also suggests that if there is an effect of duration of obesity the effect size is not likely to be dramatically large in this age group. Further research on this is needed, and future longitudinal studies would do well to include repeated mea sures of fluctuations in weight over time to provide a more accurate estimate of the effects of duration of obesity on cognitive function in midlife and beyond. It should be noted that health related behaviors had some independent associations with SDLT p erformance. As previously noted, it is possible that poor quality diet or sedentary lifestyle, are the real culprit in the association between obesity and cognitive decline, rather than the physiological effects of obesity itself. However in this study we did not find support for the idea that quality of diet was worse among persons who were obese. Quality of diet,

PAGE 131

116 measured by the Healthy Eating Index (HEI : range 1 100, mean of 62.39 in this sample ), did not differ significantly between BMI groups, and th e correlation with this and other measures of adiposity was low. Despite this, quality of diet was significantly associated with SDLT performance independent of weight T he magnitude of the improvement in SDLT score was small. The improvement in SDLT ass ociated with a 1 point improvement in HEI ( 0.017 SDLT points) was almost half of that found for BMI or percent fat, and more than 3 times smaller than that associated with WHR. No interaction between HEI and any measure of adiposity was found. So while qu ality of diet may have independent associations with cognition, it did not appear to moderate the association with obesity in this study. Smoking status was also related to cognitive performance. Interestingly, being a former smoker was significantly assoc iated with better SDLT score relative to persons who never smoked. The reason for this finding is unclear, however it could reflect a composition effect persons who succeed in quitting smoking may differ from those who have not quit, or from those who ne ver started smoking. Longitudinal or intervention studies with baseline pre smoking measures of cognitive function and repeated measures comparisons would be useful future approaches to tell the story. Physical activity is another health behavior that aff ects obesity. Unlike the other health related behaviors measured PA moderated the association between central obesity and SDLT performance. Among persons who were not centrally obese, more regular physical activity was associated with better SDLT scores. As predicted, persons who were centrally obese but active had better cognitive performance than persons who were centrally obese but sedentary, and being moderately active was associated with the best cognitive scores. Hence there was a significant interac tion between PA and adiposity on SDLT performance. This is consistent with the reports of beneficial effects of physical activity on cognitive function made elsewhere ( Baker

PAGE 132

117 et al., 2010 ; Buchman et al., 2012 ; Erickson et al., 2010 ; Laurin, Verreault, Lindsay, MacP herson, & Rockwood, 2001 ; Voss et al., 2013 ) It is possible that the benefits of physical activity on cognition are independent of weight or weight loss, making this a potential candidate for interventio ns to promote neurocognitive health across the population. Further research is needed to determine the efficacy of interventions in promoting neurocognitive health however, for this cross sectional association could be an artifact of composition effects (e .g. persons who are obese and active may represent a rather unique segment of the population), or other factors not measured here. In this study we did not find sufficient evidence to indicate a role for HbA1c CRP, IGF 1, systolic blood pressure or HDL c holesterol in mediating the association between adiposity and cognitive function. Adding any of these potential mechanisms to the model either individually or in combination did not have a significant effect on the association, nor were any of these measu res significant predictors of cognitive function. This contrasts with some other studies finding an association between glucose regulation, inflammation, IGF 1, hypertension and dyslipidemia and cognitive function ( Carro, Trejo, Busiguina, & Torres Aleman, 2001 ) However results of the prior research have been mixed. Many possible explanations for the null findings here are possible. They may be at least partially d ue to the limitations on sample size that some of these variables created, as some were only measure d in a sub sample (e.g. IGF 1). Alternatively it may be that the action of these factors is not independent of adiposity, particularly central adiposity, so that a model already adjusting for adiposity will not show an effect without interaction terms. Addition of such interaction terms to this study would impose serious sample size limitations and the difficulty of interpreting interactions between continuou s variables should not be overlooked.

PAGE 133

118 While numerous longitudinal studies of the association between adiposity and cognitive decline and dementia have been conducted, relatively few of these have assessed central obesity, and even fewer have assessed the m oderating role of health behaviors or potential mediation of the association by physiological factors shown to affect cognitive function. These factors should be included in future longitudinal studies of cognitive aging N ew intervention studies to assess the effects of weight loss, or change in obesity related behaviors, should also incorporate these factors into their designs. Though limited by a cross sectional study design, self report measures of behavior and a small number of cognitive tests, the re sults of this study are consistent with emerging evidence that modifiable factors in mid adulthood are associated with cognitive function. Obesity related deficits could be apparent decades before dementia would be likely to occur. While the population in this study could not be considered pre clinical, the finding of an association between obesity and cognitive function in midlife is worth noting. Not only was weight associated with cognitive test performance, but physical activity moderated the associatio n between central obesity and minor cognitive deficits in the general population in early and mid adulthood. Further research is needed to determine whether weight loss interventions, or other interventions that affect the health risks of obesity can sign ificantly improve cognitive fu nction in persons who are obese

PAGE 134

119 3. I nterventions 3.1 I ntroduction Taken together the observational studies suggest a link between obesity and risk of cognitive decline or dementia. Consistent with this, intervention studies pres ent evidence that reducing weight can have beneficial effects on cognitive function If obesity contributes to cognitive decline and dementia, then interventions that reduce obesity, or moderate the health risks that it imposes, could im prove cognitive func tion in obese adults. Such interventions c ould have significant public health implications, providing both a means for improving neuro cognitive health in the short term and supporting healthy cognitive aging among the increasingly overweight and obese popu lation. Interventions could also provide a useful opportunity to explore causal mechanisms relevant to the understanding of dementia etiology. Increasing physical activity and altering dietary intake are well known approaches to weight management. These a ffect not only daily energy balance, but also modulate some of the mechanisms implicated in the pathology of obesity and dementia The purpose of this paper is to discuss the evidence that physical activity and dietary interventions can affect the risk of cognitive decline and dementia among adults who are overweight or obese. 3.2 P hysical Activity Interventions Regular p hysical activity (PA) provides well recognized reductions in risk for cardiovascular disease, Type 2 diabete s mellitus, metabolic syndrome, o steoporosis and depression ( Bassey, 2000 ; Byberg, Zethelius, McKeigue, & Lithell, 2001 ; Wannamethee, Shaper, & Alberti, 2000 ) It may also decrease stress and depression ( Brown et al., 2010 ; Brown, Varghese, & McEwen, 2004 ; Gamble, Ormerod, & Frenneaux, 2008 ; Strawbridge, Deleger, Roberts, & Kaplan, 2002 ; Trivedi, Greer, Grannemann, Chambliss, & Jordan, 2006 ) A mong older

PAGE 135

120 adults PA can help to maintain strength and physical function and hence independence and the benefits of an engaged lifestyle (Aufderm, 2013). O lder adults who regularly participate in endurance, balance and resistance training can experience benefits that include improved muscle mass, arterial compliance, energy, metabolism, cardiovascular fitness, muscle strength, overall functional capacity ( Lemura, von Duvillard, & Mookerjee, 2000 ) In addition to the physical benefits, exercise may also increase self confidence ( Lindwall & Hassmen, 2006 ) 3.2.1 Animal S tudies of P hysical A ctivity Numerous studies from rats, mice and non hum an primates show that PA can have a significant impact on cognitive function and brain health. Regular PA is widely shown to improve learning and memory ( Alaei, Moloudi, Sarkaki, Azizi Malekabadi, & Hanninen, 2007 ; Cotman & Berchtold, 2002 ; Molteni et al., 2004 ; Ploughman, 2008 ; Radak et al., 2001 ; van Praag, Shubert, Zhao, & Gage, 2005 ) For example Khabour ( Khabour, Alzoubi, Alomari, & Alzubi, 2010 ) showed that 6 weeks of voluntary exercise significantly increased short term, medium term and long term spatial memory formation and increased BDNF in hippoc ampus twofold in male Wistar rats. PA can also lower the risk of cognitive impairment ( Cotman & Berchtold, 2002 ) and reduce the effects of brain aging ( Laurin et al., 2001 ) Furthermore, PA can enhance resilience against damage caused by neurodegenerative diseases ( Tillerso n, Caudle, Reveron, & Miller, 2003 ; Wu et al., 2011 ) and may mediate recovery of function after brain injury ( Bo hannon, 1993 ; Grealy, Johnson, & Rushton, 1999 ; Naylor et al., 2008 ) ( Gobbo & O'Mara, 2005 ) found that PA before administration of Kainic acid (causing neuronal death by apoptosis and necrosis) improved functional performance on the Morris water maze and object exploration tasks an d increased BDNF in the dentate gyrus relative to controls Similarly Wu ( Wu et al., 2011 ) found that 4 weeks of PA prior to injection of LPS (which induces loss of dopaminergic neurons and decreases BDNF in the substantia nigra of a mouse model of

PAGE 136

121 completely prevented LPS induced loss of neurons and restored BDNF signaling. Blocking BDNF using receptor TrkB antagonist removed the PA induced prot ection against LPS induced neuronal loss. Thus animal studies suggest that PA can have significant neuroprotective effects. 3.2.2 Human S tudies of P hysical A ctivity Observational S tudies Most observational studies of PA among humans use self reported PA often retrospective reports, while f ew use objective measures of PA (Spencer & Karzeski, ref) This inevitable introduces unwanted error in their results, which should be treated with appropriate caution. With this in mind we consider the evidence for an associa tion between PA and neurocognitive health. Association with risk of dementia : Numerous o bservational studies report that persons who engage in regular PA show decreased risk of dementia, including AD. For example, Scarmeas and colleagues ( Scarmeas, Levy, Tang, Manly, & Stern, 2001 ) found that leisure time activity was associate d with decreased risk of dementia in older adults. Similarly, Laurin and colleagues ( Laurin et al., 2001 ) reported that higher levels of PA among high functioning older adults ( > 65) were associated with lower rates of cognitive decline and dementia, including AD 5 years later. Consistent with this, Abbot et al ( Abbott et al., 2004 ) found an association between distances walked each day and probability of developing AD up to 8 years later among men aged 71 93 years While the preceding studies used self report measures of PA, similar results were reported by Buchman (2012), who used actigraphs to provide objective measures of PA to assess incidence of AD. After an average follow up of about 4 years, PA at baseline was associated with decreased risk of AD (HR 0.477, 95% CI: 0.273 0.832), and rate of global

PAGE 137

122 cognitive decline, even after controlling for self reported physical, social, or cognitive activities, depressive symptoms, motor function, chronic health conditions and APOE genotype. S ome ob servational studies find no evidence of a link between PA and dementia risk ( Fabrigoule et al., 1995 ) however several systematic reviews and meta analyses of prospective studies in on PA including the NIH consensus report on preventing cognitive decline and de mentia ( Plassman et al., 2010 ) have concluded that observational studies suggest that increased physical activity is associated with decreased dementia risk. This include s evidence that PA in midlife is associated with significantly reduced risk of MCI or dementia later in life (Skog, 2011) Association With C ognitive F unction or R isk of C ognitive D ecline : As with the studies on dementia, a mounting body of evidence links regular PA t o lower rates of cognitive decline ( Fratiglioni et al., 2004 ) A number of cross sectional studies show better neurocognitive function in persons who are physically active ( Smith et al., 2010b ) Prospective studies show similar results. In a study of 5925 community dwelling women Yaffe et al ( Yaffe, Barnes, Nevitt, Lui, & Covinsky, 2001 ) found that women who were more active experienced less cognitive decline over the following 6 8 years, after adjusting for age, education, health status, depression, stroke, diabetes, hypertension, smoking, and estrogen use Richards, Hardy and Wadsworth ( Richards, Hardy, & Wadsworth, 2003 ) found a similar association in midlfe. Among 1919 adults aged 36 at baseline and followed up to ages 43 53 PA was associated with better memory performance. While it is likely that many different aspects of PA co ntribute to the association, Barnes et al ( Barnes, Yaffe, Satari ano, & Tager, 2003 ) found that aerobic fitness, assessed by V02 peak predicted higher cognitive performance 6 years later. Other studies have found no association between PA and cognitive function. For example Holtsch et al (1999) found

PAGE 138

123 no association b etween PA and cognitive function after a 6 year follow up with in adults aged 55 86 years. Intervention S tudies Studies of the effects of PA interventions on human cognition or dementia risk have likewise produced mixed results possibly reflecting the w ide range of interventions, populations and cognitive outcome measures used ( Kramer, Colcombe, McAuley, Scalf, & Erickson, 2005 ) The variety is easily observed on inspection of the research in this area. For example Liu Liu Ambrose et al. (2010 ) found that resistance training improved executive function by more than 10%.after conducting a 12mo RCT among community dwelling women aged 65 75 years. ( Baker et al., 2010 ) reported sex specific benefi ts of a 6 month aerobic PA intervention on executive function, fasting insulin, cortisol and BDNF (improved in women but n ot men) after a well controlled RCT randomizing 33 adults aged 33 85 years to either 6 months of supervised aerobic PA or 6 months supervised stretching control. In another study, Ruscheweyh et al ( Ruscheweyh et al., 2011 ) found that increased aerobic PA was associated with increased memory increased gray matter volume in the prefrontal and cingulate cortex, and a trend for increased BDNF among 62 healthy older adults. By contrast, Erickson and colleagues ( Erickson et al., 2011 ) found that aerobic exercise training did not alter spatial memory relative to stretching co ntrol, but did increase the size of the anterior hippocampus as well as circulating BDNF. Given the mounting evidence from observational studies, and the varied intervention studies, a number of recent systematic reviews and meta analyses have investigate d the evidence that PA interventions could protect cognitive function and prevent dementia. In a review of aerobic PA interventions, Smith ( Smith et al., 2010b ) re ported that aerobic PA was associated with modest improvements in memory relative to controls with effects of similar magnitude across the 16 studies. However intensity and duration of PA did not appear to

PAGE 139

124 moderate the effect. Another meta analyses conclud ed that RCTs show a clear effect of aerobic PA on varied cognitive function ( Kramer et al., 2005 ) A m eta analysis by Heyn, Abreu, and Ottenbacher (2004 ) found an effect of PA training on the cognitive function of cognitively impaired older adults and people with dementia and these results were echoed in another more recent review concluding that RCTs among persons with MCI or dementia show that 6 12 months of PA improve cognitive scores relative to sedentary controls (Skog, 2011). Despite the results of these reviews, others with more stringent entry criteria and quality ratings have drawn different conclusions. A C ochrane review of the effects of PA and enhanced fitness on cognitive function in perso n s without known cognitive impairment (Aufderm et al, 2008) found that the data are currently insufficient to draw conclusions in this area. A systematic review by Snow den et al ( Snowden et al., 2011 ) drew a similar conclusion, determining that no studies showed really good quality overall. Limitations of the studies included s mall sample size, non randomized and quasi experimental study designs, and widespread absence of intention to treat analyses. The recent NIH consensus report on preventing cognitive decline and dementia reported that observational studies suggest a link be tween reduced PA and cognitive decline, but found only 1 intervention study that met the quality standards for inclusion. This study ( Lautenschlager & Almeida, 2006 ) 2008 ) showed that among 138 adults aged >50 years with subjective memory problems but not dementia a home based 6 month PA intervention produced modest bu t significant improvements in scores on the Alzheimer Disease Assessment Scale Cognitive Subscale (ADAS Cog) 12 months later Differences between the results of observational studies and intervention studies in this area may reflect not only the difficulty of constructing and funding large, well controlled PA trials, but also the fact that observational studies have typically examined much longer time spans than intervention studies ( Snowden et al., 2011 ) Consistent with links between PA in

PAGE 140

125 midlife and cognitive function in later lif e, it may be that the benefits of PA are gradually cumulative over time, and that regular exposure over a prolonged period of time is an important. Alternatively it is possible that the association between PA and cognitive function is simply an artifact of underlying differences between people who are likely to exercise and people who do not. However, as with obesity, there are plausible mechanisms that could mediate an effect of PA on neurocognitive health. 3.2.3 Potential M echanisms for an E ffect of P hysical A ctivity Animal research has provided valuable insights into the actions of PA on the brain (Aufderm et al, 2005). Physical activity can promote long term potentiation (LTP), a physiological process believed to be the cellular correlate of learning and memo ry in the hippocampus ( O'Callaghan, Ohle, & Kelly, 2007 ) It can also produce numerous neuroprotective changes, including enhance d neurogenesis ( Aberg, Perlmann, Olson, & Brene, 2008 ; van Praag, Chr istie, Sejnowski, & Gage, 1999 ; van Praag, Kempermann, & Gage, 1999 ) and increased synaptic plasticity ( Fa rmer et al., 2004 ; O'Callaghan et al., 2007 ) In a mouse model of AD (TgCRND8), 5 months of voluntary PA enhanced learning and memory and decreased amyloid burden in the frontal cortex and hippocam pus, possibly by altering the processing of the amyloid precursor protein ( Adlard, Perreau, Pop, & Cotman, 2005 ) Physical activity can also affect factors thought to be involved in neuronal health and amyloid clearance. For example episodes of PA can increase IGF 1 ( Carro et al., 2001 ; Kohman, DeYoung, Bhattacharya, Peterson, & Rhodes, 2012 ) which in turn can amyloid clearance ( Carro et al., 2002 ) In addition, exercise can increase BDNF in brain areas such as the hippocampus ( Vaynman & Gomez Pinilla, 2006 ; Vaynman, Yi ng, Yin, & Gomez Pinilla, 2006 ) As previously described, BDNF is involved in synaptic plasticity, neurogenesis and neuronal survival ( Cotman & Berchtold, 2002 ; Mattson, Maudsley, & Martin, 2004b ; Shirayama, Chen, Nakagawa,

PAGE 141

126 Russell, & Duman, 2002 ; Zhang & Par dridge, 2006 ) A number of animal studies suggest that the beneficial effects of PA on cognitive function and neural health may be mediated by BDNF ( Gobbo & O'Mar a, 2005 ; Khabour et al., 2010 ) and that these effects are blocked by blocking the BDNF receptor ( Dishman et al., 2006 ; Griesbach, Hovda, Molteni, Wu, & Gomez Pinilla, 2004 ; Vaynman & Gomez Pinilla, 2006 ) Physical activity can also beneficially affect other pathways implicated in neurocognitive health, including a Fleshner, Maier, Lyons and Raskind, 2011) that could be an important counter balance to the HPA axis dysregulation sometimes reported with advancing age ( Kramer, Erickson, & Colcombe, 2006 ) Rese arch among humans indicates that PA can have effects not only on cardiovascular fitness and BDNF ( Baker et al., 2010 ; Erickson et al., 2011 ; McAuley, Kramer, & Colcombe, 2004 ) but also brain structure and function ( Colco mbe & Kramer, 2003 ; Colcombe et al., 2005 ) For example an a erobic PA interventions as short as 6 months has been shown to increase brain volume in areas such areas as the anterior cingulate, middle fr ontal gyrus, and superior temporal lobe ( Colcombe et al., 2004 ) and improve functional p erformance ( Colcombe et al., 2005 ) 3.3 S tudy 3: C alorie Restriction In terventions 3.3.1 Introduction Caloric restriction (CR) is the most widely studied form of dietary restriction in animal models, and corresponds most closely to the dieting often used for weight loss among humans. Caloric restriction can increase media n lifespan in rodents, flie s, yeast and non human primates ( Heilbronn & Ravussin, 2003 ) and has many beneficial effects on markers of aging and vulnerability to age related disease ( Berner & Stern, 2004 ; Mager et al., 2006 ; Mattson et al., 2003 ; Roth, Ingram, & Lane, 2001 ; ROTH, LANE, & INGRAM, 2005 ) Behaviorally, CR increases

PAGE 142

127 exploration and physical activity in rodents ( Martin et al., 2007b ) Furthermore, m ice maintained on 40% CR from time of weaning do not exhibit the deficits in motor coordination and spatial learning that are seen in control mice fed ad libitum ( Ingram et al., 2001 ) Similarly, life long CR prevents age related deficits in performance on learning and memory tasks such as the radial arm maze and Morris water maze ( Martin et al., 2007b ; Stewart, Mitchell, & Kalant, 1989 ) Cognitive and behavioral effects have also been reported when CR was initiated at midlife ( Means, Higgins, & F ernandez, 1993 ) At a neural level, CR reduces age related deficits in long term potentiation (LTP), a process believed to be the cellular correlate of learning and memory in the hippocampus. CR has also been shown to improve synaptic plasticity and de ndritic branching ( Mattson, Maudsle y, & Martin, 2004a ; Stranahan et al., 2009a ) as well as to increase resistance to degeneration after excitotoxic injury in a rat model relevant to the pathogenesis of epilepsy and AD ( Umberger, McFall, & Mattson, 1999 ; Duan et al., 2001a ; Zhu, Guo, & Mattson, 1999 ) This neuroprotection corresponded with preserved learning and memory. Neurons in brain regions involved in learning and memory, such as the hippocampus and prefrontal cortex, are affected in AD (Ray et al. 1998) and are among the a reas protected against injury and degeneration by CR While CR has clear neuroprotective effects in animals, its effects on human cognitive function are less clear. Much of the evidence in animal studies has come from CR interventions that began shortly a fter weaning or in early adulthood, in rodents that were not previously overweight. As such they neither illust rate the effects of weight loss and it is difficult to determine whether effects result from exposure during critical periods in early in life. The effects of weight loss by calorie restriction on human cognitive function are not yet fully understood. As previously discussed, epidemiological evidence suggests that overweight and

PAGE 143

128 obesity in midlife increase risk of cognitive decline and dementia la ter in life but numerous studies also link weight loss to cognitive decline and dementia. This finding was more frequent among older adults and may reflect early dementia pathophysiology in these observational studies. Nonetheless, the effects of weight loss interventions on human cognition warrant further investigation. 3.3.2 Methods Literature S earch T erms and S tudy I nclusion C riteria English language articles focused on weight or adiposity and cognition or dementia outcomes were identified using MEDLINE and PsycINFO. The following search terms were used cognitive impairment, cognition, cognitive function, and cognitive health, as well as obesity, overweight, weig ht, fat, adiposity, central obesity, visceral obesity, visceral adiposity, waist hip ratio, and waist circumference. All relevant articles published up until January 30 th 2013 and retrievable by university library search or interlibrary loan were conside red for this systematic review. Reference lists of all potentially eligible articles were reviewed to ensure inclusion of all relevant literature. To be included in this review, the articles had to meet the following eligibility criteria: empirical inter vention studies assessing adult cognitive function that are available via university libraries or interlibrary loan, written in English, and included the weight/adiposity and dementia/cognition related search terms above within the title, abstract, and/or keywords. Interventions in childhood or adolescence were considered only if they included adult cognitive outcomes. Interventions to change the weight of a population that already had dementia at baseline were excluded. Similarly interventions to improve the health outcomes for other cognitively impaired persons with an existing medical or psychiatric diagnosis known to cause

PAGE 144

129 cognitive impairment, such as developmental disability, schizophrenia, bipolar disorder or traumatic brain injury, were excluded. S tudies with a specific focus on eating disordered populations were also excluded, as were other studies focused on specific medical or psychiatric populations. Dissertations, reviews, opinions theoretical papers or editorials were excluded from this revie w. Art icle Selection and A bstraction A three step process guided assessment and selection of articles. First, the study author reviewed the titles and abstracts of all potential articles retrieved by the search terms, identifying the set of article abstr acts that potentially matched the eligibility criteria. Second, the study author reviewed in depth the abstracts and full articles for studies whose abstracts passed the first review for inclusion. Information from the full articles was entered into summ ary tables. From this set the author identified the set of full articles which matched the eligibility criteria below. Finally, reference lists of eligible articles were reviewed for additional relevant articles to potentially include. These articles wer e also assessed for eligibility through a two step process.

PAGE 145

130 Figure 3.3 1 Steps of the systematic review process for longitudinal studies 3.3.3 Results Number of A rticles I ncluded in R eview A total of 4320 articles were identified using the search terms. Of these articles 4085 were excluded after a preliminary review of title and/or abstract bec ause they were not relevant or did not meet the inclusion criteria (e.g. topic was relevant but the article was a n editorial). The remaining 235 full articles were then reviewed and abstracted by the study author. Of these articles 16 intervention studies were considered eligible after more thorough review, and were therefore included in this study. Included art icles were then categorized by whether

PAGE 146

131 their results showed beneficial effects of weight loss interventions on adult cognitive function, no effect, or detrimental effects. A flow chart of the sorting and inclusion process can be seen in Figure 3.3 1 Be neficial E ffects Among the 16 intervention studies included in this review, 10 showed beneficial effects of weight loss diets on cognitive function. Six of these studies showed beneficial effects on memory recall (Gu nstad, 2011; Halyburton, 2007; Kretsch, 1997; Krikorian, 2012; Smith, 2010; Witte, 2009). One of these interventions reported the results of bariatric surgery (Gunstad, 2011. The other interventions ranged from 6 weeks to 4 months and most involved 30 50% caloric restriction. Other c ognitive domains that showed beneficial changes after interventions included working memory (Brinkworth, 2009), reaction time (Buffenstein, 1999; Kretsch, 1997), speed of processing (Brinkworth, 2009; Halyburton, 2007; Siervo, 2 012 ; Smith, 2010 ), accuracy (Bryan, 2000; Buffenstein, 1999), and executive function (Siervo, 2012 ; Smith, 2010 ) No E ffects or De trimental E ffects Of the 16 intervention studies exploring the effects of dieting for weight loss on adult cognitive functi on, 6 reported either no effect on cognitive function or adverse effects. These studies tended to have smaller sample sizes than the studies showing beneficial effects, tended to be of shorter duration and may have involved a smaller degree of calorie rest riction. The interventions employed a similar range of cognitive tests to the other studies that showed benefits, including tests of short term and long term memory.

PAGE 147

132 Table 3.3 1 Intervention studies with b eneficial effects on cognition First author (year) Participants Intervention Cognitive measures Weight measures Results ( Brinkwort h, Buckley, Noakes, Clifton & Wilson, 2009 ) N = 106, Age: 24 64 Mean 50.0 [0.8]. Other: central obesity + at least 1 MetS factor. Duration: 1 year Randomized: yes Groups: 1) very low carb high fat (LC) 2) a high carb low fat (LF) diet for 1 year. Digit span backward (DSB) inspection time Weight Each group achieved similar weight loss, with no significant difference between the diets Working memory improved by week 8 and remained stable at 1 year (P < .001). Speed of processing improved at week 8 but rebounded to origin al levels by week 52. Significant inverse association between change in working memory and change in fasting insulin. ( Bryan & Tiggemann, 2001 ) N= 63 Age: 30 50 Ot her: overweight women Duration: 12 weeks Groups: 1) 42 Ss weight reduction diet (15% fat, 2) 21 usual diet controls Digit Symbol Coding subtest of the of WAIS III,Trail Making Test Part A & B, Stroop test, Self Ordered Pointing Task, Digit Span Backwa rds subtest of WAIS III, Rey Auditory Verbal Learning Test [RAVLT], verbal ability. BMI Weight Being on the diet made participants less susceptible to interference effects and more accurate in their recall. No other measure of cognitive performance was af fected by being on the diet. ( Buffenstei n, Karklin, & Driver, 2000 ) N= 9 Age: 20 36 Other: overweight university students, women Duration: 4 weeks Randomized: no Groups: R estrict food & beverage intake to 800 kcal/day. 2 different visual hand eye coordination reaction tests BMI Motor performance reaction time and accuracy improved. Urinary ketones not associated with deterioration in cognitive performance. Mood, concentration, tempera ture sensitivity, appetite, and sleep quality using visual analogue scales, were not significantly altered.

PAGE 148

133 Table 3.3 1 Intervention studies with b eneficial effects on cognition First author (year) Participants Intervention Cognitive measures Weight measures Results ( Gunstad et al., 2011 ) N= 150 Age: 20 70 Other: o bese Duration: na Randomized: No Groups: 1) Bariatric Surgery 2) control Integneuro test battery. Weight Surgery patients had improved memory performance at 12 week follow up whereas obese controls actually declined. ( Halyburto n et al., 2007 ) N= 93 Age: 24 64 Other: overweigh t or obese Duration: 8 weeks Randomized: yes Groups: 1) 30% deficit LCHF 2) 30% deficit HCLF Digit span backwards (DSB) and inspectio n time (IT) tests. Weight; checked at baseline and every 2 weeks LCHF diet group had greater weight loss than the HCLF diet group. DSB test scores increased in both groups(P<0.001 for time effect),. The difference between the groups was not significant (P= 0.67). Speed of processing improved in both treatment groups during the intervention but with a significant effect of diet; the HCLF diet promoted greater improvements in speed of processing than did the LCHF diet. ( Kretsch, Green, Fong, Elliman, & Johnson, 1997 ) N= 25 Age: 23 42 Other: healthy, obese premenop ausal women Duration: 21 weeks Randomized: Groups: 1) 50% caloric restriction for 15 weeks. Bakan v igilance task, word recall task, simple reaction time,2 finger tapping task, Ericksen effect. BMI, body composition (TOBEC) Dieting women lost 12.3 += 5.5 kg of body weight. CR significantly improved word recall by 24%. CR significantly slowed simple r eaction time. This did not readily reverse upon restoration of sufficient calories to maintain body weight. No effect for obese women on sustained attention and finger tapping.

PAGE 149

134 Table 3.3 1 Intervention studies with b eneficial effects on cognition First author (year) Participants Intervention Cognitive measures Weight measures Results ( Krikorian et al., 2012 ) N= 23 Age: mean 70.1 Other: older adults with MCI Duration: 6 weeks Randomized: yes Groups: 1) 50% deficit high carb. (50% of calories) 2) very low carb diet Verbal memory performance, Trail Making Test Part B, Verbal Paired Associate Learning Test (V PAL), Geriatric Depression Scale (GDS) Weight, waist circumferen ce, fasting glucose, fasting insulin, urinary ketones Very low carbohydrate diet led to improved verbal memory as well as reductions in weight waist circumference, improved fasting insulin & glucose. Ketone levels positively correlated with memory performance. ( Siervo et al., 2012 ) N= 50 Age: Other : obese Duration: Until subjects lost 8 10% body weight Randomized: No Groups: Mini Mental State Examination (MMSE), Short Portable Mental Status questionnaire (SPMSQ), Trail Making Test (TMT) A & B Weight, height, WHR FM, FFM, MMSE and TMT B scores improved significantly after weight loss in older obese participants, The middle aged group showed improvement in speed processing (TMT A and B). ( Smith et al., 2 010a ) N= 124 Age: Other: high blood pressure, overweight to obese Duration: 4 months Randomized: yes Groups: 1) DASH diet alone 2) DASH + behavioral weight management (WM) including exercise and CR 3) Control (usual diet) Trail Making Test B & A, Verbal Paire d Associates, Stroop Interference Test, Controlled Oral Word Assoc. Test, Verbal fluency Test, Digit Span, Ruff 2 and 7 Test, Digit Symbol Substitution Test BMI, DASH diet combined with a behavioral weight management program (DASH + WM) showed greater im provements in executive function memory learning and psychomotor speed, compared with the usual diet control. Neurocognitive improvements appeared to be mediated by increased aerobic fitness and weight loss.

PAGE 150

135 Table 3.3 1 Intervention studies with b eneficial effects on cognition First author (year) Participants Intervention Cognitive measures Weight measures Results ( Witte, Fobker, Gellner, Knecht, & Floel, 2009 ) N= 50 Age: mean 60.5 Other: 50 normal to overweight elderly subjects Duration: 3 months Randomized: Groups: 1) CR 30% reduction 2) 20% increase UFAs 3) C ontrol German Rey Auditory Verbal Learning Task (AVLT), Trail Making Tests (TMT) A & B, forward and backward Digit Span Weight, height, BMI, WHR, Significant increase in verbal memory scores after caloric restriction diet (mean 20%) correlated with decrea sed fasting insulin and high CRP. Significant weight loss in the CR group. BDNF unchanged.

PAGE 151

136 Table 3.3 2 Intervention s tudies with no effects, or adverse effects on cognition. First author (year) Part icipants Intervention Cognitive measures Weight measures Results ( Cheatham et al., 2009 ) N= 42 Age: 35 5 Other: overweight Duration: Randomized: yes Groups: 1) High glycemic load (HG) & 10% CR 2) High glycemic load (HG) & 30% CR 3) low glycemic load (LG) & 10% CR 4) low glycemic load (LG) & 10% CR S imple reaction time, vigilance, learning, short term memory and attention, and language based logical reasoning. BMI, weight No significant change over time or vs. weight change in any cognitive performance values. ( Choma, Sforzo, & Keller, 1998 ) N= 29 Age: college age Othe r: wrestlers Duration: Randomized: Groups: 1) Rapid weight loss (RWL) 2) Control (maintained normal body weight and diet) Letter cancellation, a test of visual attention and visuomotor skills, digit symbol and digit span, tests of attention and short te rm memory, Trail Making A and B, story recall. Weight, After RWL, wrestlers scored significantly lower on digit span and story recall tests, than controls. RWL did not affect performance on tasks demanding attention, visual acuity, or visuomotor skills. ( Green, Elliman, & Kretsch, 2005 ) N= 56 Age: 20 45 Other: overweight women Duration: Randomized: yes Groups: 1) commercially available weight loss group; 2) diet without any group support using any CR, LC, or LF diet; 3) non dieting controls Bakan Vigilance task, Simplre reaction time, 2 finger tapping performance, verbal recall, mental rotation task. Weight, BMI, % body fat Both groups lost roughly the same body mass. Unsupported dieters lost more body fat. No differences in performance between groups. Unsupported dieting was associated with impaired cognitive function in the early stages. ( Guldstrand et al., 2003 ) N= 8 Age: 26 55 Other: severely obese non diabetic Duration: na Randomized: No Groups: Vertical banded gastroplasty (VBG), Perceptual maze test (PMT) Weight After weight loss subjects used a more speed rather than accuracy, preferring cognitive strategy.

PAGE 152

137 Table 3.3 2 Intervention s tudies with no effects, or adverse effects on cognition. First author (year) Part icipants Intervention Cognitive measures Weight measures Results ( Martin et al., 2007b ) N= 48 Age: 25 49 Other: overweight Duration: 6 months Randomized: yes Groups: 1) control (weight maintenance) 2) CR (25% restriction) 3) CR + structured exercise (12.5% restriction + 12.5% increased energy expenditure 4) low cal diet (890 kcal/d until 15% weight loss, then weight maintenance). Rey Auditory and Verbal Learning Test (R AVLT), Auditory Consonant Trigram (ACT), Benton Visual Retention Test (BVRT), Performance Test II (CPT II) Weight No consistent pattern of verbal memory, visual retention memory, or attention & concentration deficits. ( Wing, Vazquez, & Ryan, 1995 ) N= 21 Age: Other: overweight women Duration: 28 days Randomized: yes Groups: 1) ketogenic liquid formula 2) non ketogenic liquid formula very low energy diet Trail making task Weight Weight losses were comparable on the two diets (mean = 8.1 kg). Attention was not affected by diet. Ketogenic diet adversely affected the trail making task.

PAGE 153

138 3.3.4 Discussion The existing empirical literature on calorie re striction interventions or dieting for weight loss in humans (n=16 studies) reveals a mixture of results that show a beneficial effect of CR or weight loss dieting on adult cognitive function, and other studies that find no effect or adverse effects. Howev er more studies found beneficial effects (n=10) than found null or adverse effects (n=6). Those reporting beneficial effects tended to have larger sample sizes, longer durations and involved more energy restriction. Significant beneficial effects on memor y were demonstrated for 6 interventions. Although one of these interventions involved bariatric surgery (Gunstad, 2010), the other 5 interventions involved calorie restriction of 30% 50% energy restriction and ranged in duration from 6 weeks 4 months ( Halyburton et al., 2007 ; Kretsch et al., 1997 ; Krikoria n et al., 2012 ; Smith et al., 2010a ; Witte et al., 2009 ) These are surprisingly short interventions to show significant effects One of these studies wa s among older adults with MCI ( Krikorian et al., 2012 ) and showed significant effects on memory after just 6 weeks. This is a high risk group for the developmen t of dementia. The potential for targeted intervention using weight loss to promote neurocognitive health among this high risk group should not be overlooked. In conclusion, while the results of dietary weight loss interventions show mixed effects on cogn itive function there is evidence to suggest that they can have cognitive benefits, perhaps particularly for memory. Relative to observational studies, these interventions were of short duration (max 12 months) and had short follow up periods. Further resea rch is needed to determine the long term effects of calorie restricted weight loss diets on human cognitive function.

PAGE 154

139 3.4 D ietary Composition and Frequency 3.4.1 Quality of D iet Dietary composition could also have differential effects on obesity related factors suc h as insulin sensitivity, dyslipidemia, or BDNF, with consequent effects on neurocognitive health Consistent with this, d ifferent weight loss diets can have different effects on cognition and some effects can be independent of weight. The effects of vari ous dietary components on cognitive decline and dementia risk have been extensively reviewed elsewhere ( Plassman et al., 2010 ) and a full review of this large body of literature is beyond the scope of this dissertation. However a few examples can highlight some of the results that have been found for differences in macronutrient content. In a study comparing isocaloric low carbohydrate or standard dietary restriction diets, ) found that dieters on the standard caloric restricted diet perfor med better on memory based tasks but worse on measures of vigilance and attention. In another study, a comparison of two 30% calorie restricted diets in overweight and obese women (BMI 26 43, mean age 50.2 years) found that although a low carbohydrate ket ogenic diet produced less weight loss than an isocaloric high carbohydrate low fat diet (6.6 vs. 8.0% body weight) both diets improved speed of processing and working memory ( Halyburton et al., 2007 ) Since effects remained significant after controlling for weight lost, factors other than absolute weight loss may have been in play. 3.4.2 Ketogenic D iets The ketogenic provides a distinctive example of the effects of macronutrient differences on neurocognitive health. The ketogenic diet (KD) is a low carbohydrate, high fat diet intended to induce and sustain a state of ketosis in the body by minimizing somatic glucose ( Hallbook, Ji, Maudsley, & Martin, 2012 ; Zupec Kania & Spellman, 2008 ) Though it can be highly restrictive ( Miranda, Turner, & Magrath, 2012 ) and may lead to some reductions in caloric intake

PAGE 155

140 ( Cullingford, 2004 ) the KD is not considered form of caloric restriction. The person on a KD usually c onsumes the majority of their calories from fat, as well as 1g of protein per kilogram of body weight and around 5 10g of carbohydrates daily ( Kossoff, 2004 ) This increases circulating free fatty acids and promotes fatty acid oxidatio n, leading to a state of ketosis ( Henderson, 2008 ) While the anti convulsant properties of KD are w e ll recognized, and their use in treating seizures widespread, evidence now al so suggests that the KD is neuroprotective and that the mechanisms are similar to those of caloric restriction. ( Maalouf, Rho, & Mattson, 2009 ) Effects on cogni tive function, including memory performance, have al so been reported. For example, i n patients with AD, the KD has been found to enhance cognitive activity ( Reger et al., 2004, in Maalouf et al, 2009). Consistent with this, KDs have been found to reduce be ta amyloid burden in mice ( Van dA et al., 2005 ) Cognitive benefits can be observed even if the KD is begun in older age, as has been demonstrated in rats ( Xu et al., 2 010 ) The KD may have neuroprotective effects for diseases in which of oxidative damage is implicated. KD may improve mitochondrial function while also lowering production of reactive oxygen species ( Hallbook et al., 2012 ; Maalouf et al., 2009 ) The change in macronutrient content may also alter insulin signaling, with beneficial effects consistent with t he insulin hypothesis previously described ( ( Craft & Stennis Watson, 2004 ; Henderson, 2008 ) These potential improvements in metabolic efficiency, insulin signalling could make it an attractive way to treat AD and other neurological diseases of aging where oxidative stress is implicated ( Henderson, 2008 ) Despite the mounting evidence for anticonvulsant and n europrotective effects of the ketogenic diet, can also have adverse effects that make widespread use in the population unlikely. These can include hunger, gastrointestinal effects, nephrolithiasis, hyperlipidemia and

PAGE 156

141 slowed growth ( Hartman & Vining, 2007 ; Kang, Chung, Kim, & Kim, 2004 ; Kwiterovich, Vining, Pyzik, Skolasky, & Freeman, 2003 ; Mosek, Natour, Neufeld, Shiff, & Vaisman, 2009 ) Deficits in learning and memory, and impaired bran growth have a lso been reported ( Zhao, Stafstrom, Fu, Hu, & Holmes, 2004 ) It is possible this was due to malnutrition rather than the macronutrient content ( Cunnane & Likhodii, 2004 ) In addition to these po tential adverse effects the ketogenic diet may be considered less palatable than a diet containing more carbohydrates. For these reasons a more safe and acceptable alternative is being sought. Since ketone bodies might mediate the neuroprotective effects of KDs, fasting and intermittent fasting may provide viable alternatives. 3.4.3 Fasting Fasting is commonly used for weight reduction purposes or for religious reasons; however its effect on cognitive function in humans has been poorly studied. Over time, fast ing reduces the amount of glucose readily available as an energy source in the brain ( Owen et al., 1967 ) Under complete starvation the brain will eventually obtain its energy from ketone bodies, which have been shown to have neuroprotective effects ( Maalouf et al., 2009 ) Fasting has other effects on the body which might affect cognition indirectly. These include reduced insulin sensitivity ( Duska, Andel, Kubena, & Macdonald, 2005 ; Johnston et al., 2006 ; Newman & Brodows, 1983 ) changes in lipid profiles (Bayer et al, 1997), and reduction of serum leptin levels ( Ahima et al., 1996 ) Though little research into the cognitive effects of fa sting has been conducted, some studies of brief fasting show mild reductions in cognitive func tion. For example Pollitt and colleagues ( Pollitt, Cueto, & Jacoby, 1998 ) demo nstrated a reduction in stimulus discrimination, an increased in errors, and slower memory recall in children fasting overnight and during the morning compared with children who fasted overnight only. They also showed that missing

PAGE 157

142 breakfast correlated wit h adverse effects on children's late morning problem solving performance ( Pollitt, Lewis, Garza, & Shulman, 1982 ) Doniger and colleagues ( Doniger, Simon, & Zivotofsky, 2006 ) showed that fasting was associated with cross domain deficits for tasks req uiring perception of spatial relations in subjects completing a 12 to 16 hour fast. Other authors showed fasting caused heterogeneous and domain specific changes ( Lo tfi, Madani, Tazi, Boumahmaza, & Talbi, 2010 ; Tian et al., 2011 ) By contrast others found no detrimental effects in 10 days of fasting in the obese ( Liebermeister & Schroter, 1983 ) 3.5 I ntermittent Fasting While it is often assumed that the beneficial effects of calorie restriction (CR) are due to weight loss, or the fact that calorie restricted animals weig h proportionally less than their ad libitum fed counterparts, this is not necessarily the case. Not only do human studies indicate that different diets have different effects on cognition, but animal studies demonstrate that altering the frequency of food intake may produce similar or greater neurocognitive effects than CR. Intermittent fasting (IF) is a form of dietary restriction in which organisms alternate between periods of complete fasting and ad libitum feeding. The most commonly researched form of IF involves alternating fed and fasted days, or alternate day fasting Like CR, IF produces significant improvements in lifespan and animal health ( Goodrick, Ingram, Reynolds, Freeman, & Cider, 2009 ) ( Anson et al., 2003a ; Azab, Khabour, Al Omari, Alzubi, & Alzoubi, 2009 ; Lee et al., 2006b ; Sogawa & Kubo, 2000 ; Sohal & Weindruch, 1996 ) and many of the effects of IF are similar to those of prolonged CR ( Mattson, 2005 ) However the effects of IF may be independent of weight loss, as was demonstrated in an elegant study by Anson and colleagues ( Anson et al., 2003b ) in which rodents on an IF diet showed greater health benefits than pair fed CR controls. Furthermore, the beneficial effects of

PAGE 158

143 IF may be greater than those found in CR. For example Duan et al. (2001a ) demonstrated that a 15% caloric deficit with IF produces greater resistance to excitotoxic damage to hippocampal neurons, and larger increases in other markers of neurocognitive health than 30% CR. Table 3.5 1 Effects of IF and CR compared. Parameter CR IF Learning & m emory Improved Improved (animal models) BDNF Increase Increase Synaptic plasticity Increase Increase Neurogenesis No change Increase HSP70 Increase Increase Body weight Decrease Decrease or no change Body fat Decrease Decrease Body tempera ture Decrease Decrease Blood pressure Decrease Decrease Heart rate Decrease Decrease Blood glucose Decrease Decrease Blood insulin Decrease Decrease Insulin sensitivity Increase Increase IGF 1 levels Decrease Increase IL 6 Decrease Decrease TNF D ecrease Decrease Hydroxybutyrate Unchanged Increase HDL Increase Increase Homocysteine Decrease Decrease Adapted from Mattson, Duan and Guo, 2003. Animal S tudies of Intermittent F asting In animal models IF does appear to produce additional health benefits beyond those seen after CR. As can be seen in Table 3.5 1 both diets decrease fasting glucose and insulin, as well as heart rate and blood pressure in animal models ( Anson et al., 2003b ; Wan, Camandola, &

PAGE 159

144 Mattson, 2003 ) However the magnitude of improvement in insulin sensitivity can be greater in IF ( Anson et al., 2003b ) Intermittent fasting also differs from CR by increasing circulating insulin like growth factor (IGF) 1, which is decreased in animals on a CR diet ( Anson et al., 2003b ) Interestingly, IGF amyloid clearance from the brain, with significant ( Carro et al., 2002 ; Craft, 2007 ) Another difference between IF and CR are their effects on the neuroprotective Brain Derived Neurotrophic Factor (BDNF), with larger increases af ter IF than are seen in CR ( Duan et al., 2001b ) Animal studies show that IF improves learning and memory in ways similar to CR, but with additional neurocognitive benefits ( Mattson et al., 2003 ) As with CR, improvements in learning and memory are seen with corresponding improvements in neural health, including increased synaptic plasticity and increased neural resistance to excitotox ic insult and age related neurodegeneration ( ; Duan et al., 2001a ; Yu & Chung, 2001 ) Yet unlike CR, IF also increases ne urogenesis in the brains of adult rats and mice, particularly in the hippocampus, an area well known for its role in learning and memory, and among the first brain regions affected in AD pathology ( Duan et al., 2001b ; Gage, 2000 ; Lee et al., 2000 ; Lee, Duan, & Mattson, 2002 ) It is perhaps relevant that neurogenesis is typically impaired in mice that have low BDNF ( Lee et al., 2002 ) Human S tudies of I ntermittent F asting T he effects of IF on human cognition have not yet been tested, but several human experiments with IF to date suggest that general health benefits parallel those seen in animal models. Normal weight men and women aged 20 55 years on an IF regimen for 3 weeks decreased their body weight by approximately 2k g and insulin sensitivity improved for

PAGE 160

145 men, though not women ( Heilbronn, Smith, Martin, Anton, & Ravussin, 2005 ) and showed no change in fasting glucose, but did sho w decreases in fasting insulin Likewise Halberg et al. (2005 ) found that normal weight men on an IF diet for 2 weeks who were instructed to fully compensate for their fasting day lost no weight but still showed improved insulin mediated glucose uptake. Johnson et al (2008) placed obese adults with asthma on a modified IF regimen, in which they ate <20% of normal daily intake on their restricted day, and ad libitum the next. Nine of the te n participants adhered to the diet and lost an average of 8% of their body weight over the 8 week period. In this sample IF diet improved asthma related symptoms, and decreased inflammation (serum TNF studies show incre ased BDNF after IF, they also show that BDNF can be elevated in response to potential neuronal threats such as inflammation. It is possible that these asthma patients started with high BDNF as part of their asthma, which was normalized alongside the asthma and inflammation after the intervention. In all other respects results in humans are consistent with those seen in animal models of IF. It is therefore possible that neurological and cognitive effects of IF would also be comparable. Since beneficial effec ts of IF on health appear to be independent of weight loss, research into the effects of IF on human cognition could potentially provide a means of improving cognitive health of obese adults, even if weight loss is not achieved. IF may also provide some va luable insights into potential mechanisms linking obesity and cognitive function. Adverse E ffects of I ntermittent F asting Human and animal studies of IF have widely report reported beneficial effects of IF on health. The brief duration of fasting typical ly used (~1 day) may contribute to a lower likelihood of the adverse effects than can be found in prolonged fasting. However it remains possible that

PAGE 161

146 IF could have adverse effects, particularly on perceived hunger, mood, hydration and nutritional adequacy. Among studies of IF in humans to date, o ne minor unfavorable change was reported by Heilbronn et al, ( Heilbronn et al., 2005 ) who found an unfavorable rise in glucose among women but not men after 3 weeks of IF. Natural instances of variants of I F may also provide some insight into its potential effects on health. Ramadan fasting may be one such example. Ramadan is a month of the Muslim calendar during which Muslims abstain from eating or drinking between sunrise and sunset. It is therefore a for m of intermittent, though not alternate day, fasting. During Ramadan subjective alertness and mood have been shown to decrease ( Roky, Iraki, HajKhlifa, Lakhdar Ghazal & Hakkou, 2000 ) as well as psychomotor performance ( Roky, Houti, Moussamih, Qotbi, & Aadil, 2004 ) and an increase in daytime sleepiness has been reported ( Roky et al., 2003 ) However Ramadan fasting typically produces significant changes to sleeping patterns and duration, for all food must be consumed after sunset and be fore sunrise, and many practitioners will stay up late to fit in their normal daily caloric intake during these hours ( Roky et al., 2000 ) In addition, drinking dur ing daylight hours is also restricted, so practitioners may experience dehydration. Hence the detrimental effects of Ramadan that have been reported may be due to sleep deprivation and/or dehydration ( Suhr, Patterson, Austin, & Heffner, 2010 ) Mechanisms of I ntermittent F asting While the distinctive mechanisms of IF are not fully understood, some researchers theorize that IF specific effects are due to activation of t he physiological repair mechanisms that are part of a normal response to stress ( Martin, Mattson, & Maudsley, 2006 ; Mattson & Calabrese, 2010 ; Mattson et al., 2003 ) Fasting is psychologically and physiologically stressful (Mamczar, 2000; Mattson et al, 2000). Physiologically, reduced availability of glucose is a threat to neuronal integrity ( Buckner, Snyder, Sanders, Raichle, & Morris, 2000 ; Vaishnavi et al., 2010 )

PAGE 162

147 so energy deficits lead to counter regulatory mechanisms to ameliorate the threat of hypoglycemia. Short term fasting therefore leads to an increase in the aptly named glucocorticoid (GC) stress hormon es cortisol in humans, corticosterone in rodents which mobilize glucose from storage ( Hanniman, Lambert, Inoue, Gonzalez, & Sinal, 2006 ; Lee et al., 2006a ) once a single episode of fasting is a stressful, IF therefore involves repeated exposure to a mild metabolic stressor (glucose deprivation), a stress response (GC secretion), and the opportunity for recovery in between exposures. Consistent with this theory, animal models indicate that the repeated exposure to short term fasting that occurs in IF produces metabolically stressful energy deprivation and increases GC production ( Mager et al., 2006 ; Varady & Hellerstein, 2007 ; Yu & Chung, 2001 ) Furthermore, rodents treated with a competitive inhibitor of glycol sis (2 deoxy d glucose) show physiological effects similar to those of IF (Mamczar, 2005). These findings suggest that IF does in fact produce repeated mild metabolic stress in animals. Hormesis is a term used by tox icologists to refer broadly to any biphasic dose response in which a low dose produces beneficial effects while a high dose produces harmful effects ( Calabrese et al., 2007 ; Mattson, 2008 ) A wide variety of stressors can have hormetic effects, including exposure to heavy metals, pesticides, antibiotics, chemotherapeutic agents, ethanol, chloroform, hypergravit y, cold, ionizing radiation and energy deficit by fasting or physical activity ( Calabrese, 2008a ; Calabrese & Cook, 2006 ) The same stress dose response is seen in many psychological functions, and is referred to as the Yerkes Dodson law in psychology ( Calabrese, 2008 b ) Thus exposure to many different stressors leads to the fundamental bi phasic inverse U shaped curve ( Calabrese & Baldwin, 2001 2003 ; Depke et al., 2008 ) shown in Figure 3.5.1 that is so familiar in stress research ( Mattson & Cheng, 2006 ; Rattan, 2008 )

PAGE 163

1 48 Figure 3.5 1 An inverse U shaped dose response curve. While a large body of literature demonstrates that severe stress can have detrimental effects on health ( Chrousos & Gold, 1998 ) a growing body of evidence suggests that repeated exposure to mild stressors could have bene ficial effects. While this remains controversial and under ongoing investigation, beneficial effects have been reported for a wide variety of different physiological stressors, including physical activity, heat shock, irradiation, pro oxidants, hypergravi ty and curcumin, can have general health benefits similar to those found with IF fasting ( Calabrese & Baldwin, 2001 ) These effects include anti aging and life prolonging effects ( Calabrese & Cook, 2006 ; Rattan, 2008 ) cells (Bates, 2008), improved immune regulation (Bauer, 2001), and enhanced metabolic efficiency ( Roth, Ingram, & Lane, 1999 ; Sohal & Weindruch, 1996 ; Weindruch & Walford, 1988 ) Organisms exposed to small dose s of these stressors appear to adapt, compensate, and show protection against future exposures ( Calabrese & Baldwin, 2001 ) Interestingly, mild exposure to one stressor may confer resilience against other different (heterotypic) stressors, suggesting a common mechanism ( Masoro, 2000 ) It has therefore been proposed that repeated exposure to mild stressors can

PAGE 164

149 progressively condition an organism towards resilience against stress ( Masoro, 2000 ; Mattson & Calabrese, 2010 ) Figure 3.5 2 Repeated episodes of stress in the mild ge may have cumulative beneficial effects. Intermittent fasting may increase resilience to stress by triggering protective repair mechanisms normally activated in response to stress ( ; Duan, Guo, Jiang, Ware, & Mattson, 2003b ; Duan & Mattson, 1999 ; Martin et al., 2006 ; Maswood et al., 2004 ) Insulin like growth factor (IGF) 1 may be one example. This protective growth mechanism is upregulated after IF and after physical activity but not after CR ( Anson et al., 2003b ; Duan et al., 2001b ; Mattson et al., 2003 ) The increased BDNF found after IF may be another example. This neurotrophin shows neuroprotective effects, and is involved in neurogenesis, synaptic plasticity, and neurotransmitter synthesis ( Diogenes et al., 2007 ; Mattson et al., 2003 ) Neural BDNF production increases in response to cellular stressors that could cause neuronal damage, such a s hypoglycemia, trauma, ischemia, or seizures, and can protect neurons against death ( Mattson et al., 2003 ) BDNF also appears to protect neurons in experimental

PAGE 165

150 disease ( D uan et al., 2001a ) Upregulation of BDNF after IF may therefore be one case in which mild stress leads to beneficial increases in repair. T he release of GC stress hormones such as cortisol may also be part of the stress response to IF. As mentioned abov e, low doses of cortisol can have beneficial effects on the bra in and cognitive function ( Lupien et al., 2005b ; McEwen, 2000 ) These hormones are released by the Hypothalamic Pituitary Adrenal (HPA) axis, a primary stress response system, which may be required as part of the adaptive response to IF. This is suggested by the effects of IF on a transgenic mouse mod el of AD the APP mutant mouse. The APP mutant mouse model of AD shows not only increased amyloid deposition in the brain but also has abnormal glucose regulation and GC responses to restraint stress ( Pedersen, Culmsee, Ziegler, Herman, & Mattson, 1999 ) When placed on an IF dietary regimen they are apparently unable to mount a counter regulatory response to the fasting, instead becoming severely hypoglycemic on their fasted days and dying within 2 3 weeks ( Pedersen et al., 1999 ) This is consistent with the view that the stress response plays an important role in the e ffect of IF ( Martin et al., 2006 ) In other rodent strains IF appears to improve general stress resistance ( Mager et al., 2006 ; Varady & Hellerstein, 2007 ; Yu & Chung, 2001 ) Intermittent fasting has also been shown to downregulate glucocorticoid receptors, while maintaining mineralocorticoid receptors ( Lee et al., 2000 ) However the eff ects on general HPA axis diurnal rhythm are unknown, as the sample collection methods required in animal models are stressful, making it difficult to obtain repeated measures across the day. Thus it may be that the stress response is an important part of t he beneficial adaptation to IF, though this needs further testing. The effect of IF on human stress response and HPA axis function has not yet been adequately tested. The only human study to measure cortisol during a modified intermittent fasting paradigm collected samples just once a day and did not control time of day for sample

PAGE 166

151 collection ( Stote et al., 2007 ) Since cortisol follows a strong diurnal rhythm this may have confounded the results. The evidence from animal studies suggests that it is worth investigating not only whether IF can benefit human learning and memory, but also the relationship of any such alterations to changes in resilience to stress. This is particularly important because of the body of literature demonstrating that stress can have significant detrimental effects on human cognition.

PAGE 167

152 4. S tudy 4 : the D RIFT S tudy 4.1 I ntroduction Given recent epidemiological evidence linking obesity to cognitive de cline and dementia described above, interventions that have the potential to reduce obesity and improve cognitive function could be valuable. A nimal studies show significant neuroprote ctive effects for IF including protection against age related declines in learning and memory, improved synaptic plasticity and neural health, increased neurogenesis, increased brain derived neurotrophic factor, resilience against neurotoxic insult, and increased resistance to neurodegenerative diseases ( Martin et al., 2007a ; Mattson et al., 2003 ; Mattson & Wan, 2005 ) ( Anson et al., 2003b ; Martin et al., 2006 ) Intermittent fasting is a dietary restriction regimen in which organisms alte rnate between periods of complete fasting and periods of ad libitum feeding. However the effects of IF on human cognition have not yet been tested. While many of the effects of IF are similar to those found in calorie restricted (CR) diets, Animal studies also show other significant health benefits for IF, including improved insulin sensitivity ( Mattson & Wan, 2005 ) This effect has been replicated in studies of I F in healthy weight ( Halberg et al., 2005 ) and obese humans ( Johnson et al., 2007 ) It is possible that the neurocognitive benefits seen in animals are also replicable in humans, but this remains to be tested If IF affects human neurocognitive health it will also be imp ortant to understand its mechanisms of ac tion Results from animal and human studies show that IF can affect some of the mechanisms that could link obesity to AD, including glucose regulation and insulin resistance IGF 1 leptin BDNF, inflammation, and glucocorticoids ( Martin et al., 2006 ; Mattson, 2005 ; Mattson & Calabrese, 2010 ; Mattson et al., 2003 ; Mattson & Wan, 2005 ) I nterventions that begin ( Heilbronn et al., 2005 ; Johnson et al., 2007 ; Sperling et al., 2011 ) may be particularly

PAGE 168

153 useful in preventing cognitive decline. While weight loss by calorie restricted diets can be a useful way to study these mechanisms, evidence indicates that IF can have unique effects suggesting that some different m echanisms are involved. For example IF has been shown to increase neurogenesis and IGF 1 in animals while CR does not ( Mattson, 2005 ; Mattson et al., 2003 ) In addition evidence suggests that glucose regulation and BDNF are better in IF fed animals relative to pair fed calorie restricted controls ( Anson et al., 2003b ) In humans, beneficial effects of IF on glucose regulation were apparent in healthy young men who were instructed to maintain a stable weight ( Halberg et al., 2005 ) suggesting that some effects on health could be independent of weight loss Intermittent fasting could therefore have neuroprotective effects and may provide a useful opportunity to study the mechanisms linking obesity and cognitive function in midlife To test this we conducted a randomized controlled trial to investigate 1) the safety and efficacy of IF for weight loss among obese adults, 2) the effects of IF on cognitive function among obes e adults, 3) explore the potential mechanisms by which IF might affect neurocognitive health outcomes. We hypothesized that IF would have similar effects to those previously described in animals, including cognitive improvements specific to learning and me mory, increased BDNF, improved insulin sensitivity, reduced inflammation and increased resilience against stress, and explored the possibility that effects on memory function may be mediated by these 4.2 M ethods 4.2.1 Participants Tw enty six obese (BMI 30 45 kg/m 2 ) but otherwise healthy community volunteers, recruited through community flyers and advertisements, took part in the study. Participants were eligible to be included if they were obese but otherwise healthy, had no evidence of depression (CESD, Radloff (1977 ) showed no evidence of binge eating disorder (Questionnaire

PAGE 169

154 of E ating and Weight Patterns: Spitzer, Yanovski, and Marcus (1993 ) English was their first language and they were able to read and write to a 6 th grade level. Participants were recruited using IRB approved advertisements placed around the campus of the University of Colorado Denver and IRB approved email advertisements through the University list serv. Full informed consent was sought and documented before study participation. This study was reviewed and approved by the Colorado Institution al Review Board COMIRB Protocol 06 0383. Participants were paid up to $80 0 for taking part in the study. This included $50 for the test fast, $75 for each of the baseline study visits, $350 for completion of the one week in patient study at the beginning of the weight loss period, $150 for completion of visits at 8 weeks, and $100 for the 6 month follow up visit. Sample S ize and Power C alculations Sample size for the DRIFT study was chosen to provide the power to detect statistically significant changes i n weight between the intervention and control groups. The minimum clinically meaningful difference in weight was determined to be 1 kg. Sample size to detect this difference was calculated using 1) pilot data from the PI (WTD) of an 8 week CR protocol ( weight loss of 6.0 + 2.8 kg) and pilot data from an 8 week IF protocol with n=2, (weight loss of 11.25 + 3.3 kg) as well as 2) extrapolations from other studies of general fasting ( Jackson et al., 1971 ; Runcie & Thomson, 1970 ) Johnstone et al, 2002; Oh, Kim and Choe, 2002) and allowing for ~20% compensation (per day) on a feast day (Johnstone et al, 2002 ), To detect an anticipated 15kg weight loss in the IF group, with 30% variance (15 + 4.5 kg) a sample size of 7 per group was required to 90% power with an of 0.025. In order to account for potential dropout 15 per group were recruited. It was determined unlikely to detect statistically significant differences in

PAGE 170

155 insulin sensitivity, cort isol or other physiological markers so these were considered exploratory aims. 4.2.2 Intervention P rotocols The DRIFT study used an experimental randomized controlled design. A flow diagram of the study design can be seen in Figure 4.2 1 In brief, after baseline assessment of metabolic, behavioral and cognitive function after both a fed day and after a short term fast, participants were randomized to either 8 weeks of a standard calorie restricted diet (SDR), or to 8 we eks of intermittent (alternate day) fasting (IF). During the study, all participants were admitted to the Clinical Translational Research Center (CTRC) at the University of Colorado Hospital on 6 separate occasions for monitoring. These visits included: 1 ) a test fast to test the safety and tolerability of fasting for each potential participant, 2) baseline fed visit, 3) baseline fasting visit, 4) week 1 visit, 5) week 8 visit, 6) 6 mo nth follow up. A more detailed description of the procedures used on the se visits is given below.

PAGE 171

156 Figure 4.2 1 Overview of the DRIFT study design. 4.2.3 Materials and T ests Assessment of C ognitive F unction Cognitive function was assessed by performance on a computerized test batte ry administered at 7am each day of in patient visits. The CNS Vital Signs test battery (Psychology Software Tools, Inc. Pittsburgh, PA) was administered on a laptop PC and administered in a quiet private hospital room free of distractions or interruptions. The CNS Vital Signs test was developed for repeated measures testing purposes and aims to minimize practice effects. Test retest reliability i s 0.67 0.85 (CNS Vital Signs ). Each test administration takes approximately 20

PAGE 172

157 minutes. Participants responded on the standard computer keyboard using pre specified keys. The CNS Vital Sign s test battery is comprised of 7 subtests administered in a fixed order Verbal Memory Test (VBM ): This test is an adaptation of the Rey Auditory Verbal Learning Test (Taylor, 1959; Rey 1964). Respondents are required to remember 15 words and recognize them in a field of 15 distracters. The test is repeated at the end of the test battery. Low scores indicate verbal memory impairment. Visual Memory (VIM): This test is based on th e Rey Visual Design learning test. Respondents are asked to remember 15 geometric figures and recognize them in a field of 15 distracters. The test is repeated at the end of the battery. Low scores indicate memory impairment. Finger tapping test (FTT): De veloped by Mitrushina et al, (1999), this task measures motor speed and fine motor control. Respondents tap a key as quickly as they can in 3 rounds. Low scores indicate motor slowing. Symbol Digit Coding (SDC): Symbol digit coding is a 2 minute test tha t measure of psychomotor speed and visuo motor coordination. The test is very sensitive to aging, and errors may be due to impulsive responding, misperception or confusion. Stroop Test (ST): T he Stroop task (Stroop, 1935) is comprised of three types of tr ials: a) color words (red, green, yellow), b) neutral words to match the color words in length and frequency of occurrence: intent, lot, ship, advice, cross, debate (Battig & Montague, 1969), and c) a string of asterisks with lengths varying to match the l engths of the color words. The stimuli were presented in one of three colors (red, green or yellow) centered on a black background. It measures processing speed, cognitive flexibility and cognitive inhibition. Prolonged reaction times may indicate cognitiv e slowing, while errors may indicate impulsiveness or disinhibition.

PAGE 173

158 Shifting Attention Test (SAT) : In this test of executive function respondents need to adjust their responses to randomly changing rules. The best scores have many correct responses, few e rrors and a short reaction time. Continuous Performance Test (CPT): The CPT is a measure of vigilance or sustained attention or attention over time ( Rosvold et al, 1956). It is sensitive to CNS dysfunction in general, and is not specific to any particular condition (Riccio & Reynolds, 2001). The scores of these tests were used to automatically generate cognitive domain scores as outlined in Table 4.2 1 These domains were memory, attention, cognitive flexibility, c ognitive speed and reaction time Table 4.2 1 CNS vital s igns cognitive test domains and how they are calculated. Memory VBM and VIM total correct, immediate and delayed, responses and omissions. Attenti on^ SDC errors + ST errors + SAT errors + CPT omission errors + CPT com errors. Cognitive flexibility SAT correct SAT errors ST errors. Psychomotor speed Right taps + left taps + SDC correct. Reaction time^ Average [ST complex RT + Stroop RT] ^ Lo wer scores indicate better performance. Assessment of S tress Both psychological and physiological measures of stress were assessed Psychological stress : To assess psychological stress pa r ticipants were asked to rate their perceived stress and their m ood in two questionnaires, which were administered at approximately 7am of each study visit, directly before the cognitive test battery. The 10 item Perceived Stress Scale (PSS: Cohen, Kamarck, and Mermelstein (1983 ) ) is a self report questionnaire in which resp ondents rate on a 5 point Likert type scale ranging from

PAGE 174

159 In the last week, how often have you felt that you were unable to control the important things in yo was added late to the study and was administered to a subsample of participants. Th e Profile of Mood States (POMS ) is a self report inventory in which participants rate the 30 item POMS Brief form. Items correspond to 6 identifiable mood states including tension anxiety, depression dejection, anger hostility vigor activity, fatigue inertia, and confusion bewilderment. Participants respond on a 5 point Likert create a) a total mo od score, and b) a subscale for tension/anxiety. Physiological stress : Physiological measures of stress were based on two different measures of cortisol. Basal diurnal cortisol During in patient visits, participants gave saliva samples at 6:00am, 6:30a m, 7:00am, and every 2 hours thereafter while they were awake. These samples were analyzed by standard laboratory assays (Salimetrics) to assess basal cortisol. Basal cortisol was used to calculate peak morning cortisol (standardized as half an hour after waking, i.e. at 6:30am of each study visit), evening nadir (standardized as cortisol at 21:30pm on each study visit). To assess cortisol rhythm, these values were used to calculate daily cortisol decline which was calculated as the difference between the morning peak and the evening nadir. Cortisol decline was not available for the 6 month follow up, as participants did not spend 24h at the in patient visit.

PAGE 175

160 Stimulated stress response At each visit participants completed a 90% V0 2 max exercise stress test which acted as a stimulated HPA axis stressor. This exercise stress test avoids issues (e.g., mental stress tests) because the same relative exercise intensi ty results in similar magnitude of change in HPA axis activity across different fitness levels. Hence responses can be compared between participants with widely different absolute maximal aerobic capacities. In the 90% V0 2 max stress test participants walk ed on a treadmill for 5 minutes to warm up. This was followed by an increasing grade and speed until they were working at 90% of V0 2 max (maximum assessed at baseline) for 10 minutes. Blood was drawn and saliva samples collected for measurement of salivary cortisol at 0, +15, +25, +35, +45, +55, +65 and +75 minutes while the participant was seated and resting quietly. Cortisol values were used to calculate area under the curve (AUC) as an estimate of cortisol response to the exercise stress test. Assessmen t of D epressive S ymptoms Symptoms of depression were assessed at baseline and post intervention using the Center for Epidemiological Studies Depression scale (CESD: ( Radloff, 1977 ) ). The CESD is a 20 item self report scale intended to measure depressive symptoms in the general population. Participants answer on a 4 point Likert type scal e how often they feel depressive symptoms such as tearful, lonely, sad or inadequate. Reliability in the general population has been assessed (r=0.85, patient samples 0.90, ( Radloff, 1977 ) ). Scores were summed according to CESD scoring instructions Total CESD score was recorded for baseline and post intervention. Assessment of D isorde red E ating B ehaviors The Questionnaire of Eating and Weight Patterns (QEWP: ( Spitzer et al., 1993 ) ) is a screening test for eating disorders and was administered at baseline to screen out potential participants with eating disorders, such as binge eating disorder. I t was also administered after

PAGE 176

161 the 8 week intervention to determine whether the IF intervention increased disordered eating patterns. Eating disorder status was coded as a dichotomous variable (yes/no). Assessment of A diposity ht (kg) were measured by standard methods. P ercentage of body fat and percent trunk fat were determined using a Dual X ray Anthropometry (DXA) scan at baseline, week 8 and at 6 month follow up. Assessment of I nsulin S ensitivity (SI) Insulin sensitivity wa s assessed on the final day of each study visit using a Frequently Sampled Intravenous Glucose Tolerance Test (FSIVGTT). In this test, two intravenous lines were placed into ante cubital veins in each arm. A 20% glucose solution was infused gradually over the span of 15 minutes into one arm. Blood was drawn from the other arm for analysis of glucose and insulin at regular intervals, giving 30 time points over 180 minutes after the full glucose load was delivered. Insulin sensitivity was calculated using Min Mod Millenium ( Pacini & Bergman, 1986 ) a computer package which can be used to estimate insulin sensitivity from th e dynamics of glucose insulin collected by FSIVGTT. Insulin sensitivity was measured on the final day of each study visit. Laboratory M easures Blood collected from participants during in patient hospital visits was analyzed for electrolytes, glucose, insul in, free fatty acids, leptin, ghrelin, ketones (3 OH butyrate and acetoacetate), and triglycerides using standard laboratory procedures in the CTRC core laboratory. A 24 hour urine collection was also collected for analysis of nitrogen, cortisol, epinephri ne, and norepinephrine. Markers of inflammation, including serum interleukin 6 (IL 6), C reactive protein (CRP) and plasma Tumor Necrosis Factor Enzyme Linked ImmunoSorbent Assay (ELISA, R&D systems ). For analysis of BDNF, plasma

PAGE 177

162 samples were sent to collaborators Mark P. Mattson and Bronwen Martin at the National Institutes on A ging because of their experience with this difficult assay. As a result BDNF was assessed using the blood drawn at 7am of each study visit. 4.2.4 Procedure Screening Inclusion and exclusion criteria were assessed at a) phone screening, and b) in person screening by a professional research assistant and a physician. In person s creening involved a comprehensive medical examination to ensure participants met selection criteria and that they were in good health. This visit included a medical examination, laboratory m easures, the QEWP. CESD and anthropometric measures. Participants were excluded if they reported >10lb weight change in the last 6 months, had presence or history of a chronic disease known to affect appetite, food intake or metabolism, including hypothyro idism and cancer, used medications known to affect appetite, were pregnant or currently lactating, were a current smoker, had a history of cardiovascular disease, renal disease, hepatic disease, seizures, migraine, or disorder of the gastrointestinal tract Test F ast The test fast was conducted in order to determine safety and tolerability of fasting for each participant. Participants reported to the CTRC at the University of Colorado Hospital at 7pm and were admitted as in patients. They ate no food for t he remainder of the evening, and were asked to go to bed at 10pm. Participants were woken at 6:30am and gave a saliva sample, collected in a small tube. This was followed by measurement of resting metabolic rate (RMR), another saliva sample, echocardiogram ( ECG ) saliva sample, and the VAS & POMS questionnaires. After another saliva sample, blood was drawn at 7am (for glucose, insulin, electrolytes, triglycerides, free fatty acids, ketones, complete blood count, ghrelin, and leptin),

PAGE 178

163 followed by the cogniti ve computer battery. Participants fasted for the rest of the day but were allowed free access to water and non caloric beverages, including caffeinated beverages. They remained in the hospital for monitoring for the rest of the day and slept overnight agai n at the CTRC. The same series of tests were conducted before the participant was given breakfast and discharged home. After the test fast, only participants who reported that they tolerated the fast well and who did not have adverse changes in safety data were continued in the study Baseline V isits Figure 4.2 2 Baseline fed and baseline fasted visits compared. Participants attended two separate baseline visits one fed and one fasted. Prior to each ba seline visit, all participants ate a 5 day pre study weight maintenance diet, the dietary characteristics of which are described above, with meals provided for home by the CTRC metabolic kitchen. Participants were asked to refrain from vigorous physical ac tivity in the 24h

PAGE 179

164 before each visit. As shown in Figure 4.2 2 procedures for fed and fasted baseline visits were identical, with the exception that on the fed visit participants were fed breakfast, lunch, dinner a nd a snack at standardized times. On a fasted visit participants were provided no food but had ad libitum access to water, non caloric beverages and sugar free chewing gum. On both visits, each morning blood collection at 7am was in the fasted state after an overnight fast Baseline visits were completed in random order (fed fasted vs. fasted fed) and were separated by 4 weeks to allow menstrual cycle consistency for women. On the night before the study participants checked at the CTRC at 7pm, completed ea ting questionnaires for the last week, and went to bed at 10pm. Participants stayed for 2 nights and 1 days. Figure 4.2 2 gives an overview of the timing of tests conducted during baseline visits. On day 1 of ea ch baseline study participants were woken at 6:00am and immediately gave a saliva sample before getting out of bed. They gave another saliva sample at 6:30am after resting quietly in bed while resting metabolic rate (RMR) was assessed. After a 6:30 ECG for safety monitoring, a third saliva sample was collected at 7:00am. A nurse then drew blood, which was later used to assay glucose, insulin, pro inflammatory cytokines and BDNF Participants were then permitted to get up. They next completed the profile of mood states questionnaire (POMS), the perceived stress scale (PSS), and rated their hunger satiety on visual analogue scales (VAS). Following the questionnaires, participants completed the computerized cognitive test battery while seated in a quiet private hospital room The cognitive test battery took approximately 20 minutes to complete. After completing the cognitive tests, participants were free to spend the remainder of the day as they chose, but were asked to remain within the hospital and wear an act ivity monitor. Vital signs were checked every 2 hours, and blood was drawn at 10:30am and at 2pm. Participants went to bed at 10pm.

PAGE 180

165 Figure 4.2 3 Outline of procedures on a baseline study visit. On day 2 of each baseline visit procedures for the morning were the same as day 1 until completion of the cognitive tests. A fter completing the cognitive test battery participants underwent a frequently sampled intravenous glucose tolerance test (FSIVGTT) to dete rmine insulin sensitivity. They were then fed a breakfast that provided approximately 20% of daily caloric needs, and rested fo r 15 minutes before completing the 90% VO 2 max stress test. Participants were then discharged from the hospital. Randomization to I ntervention G roup Sex stratified randomization to either an intermittent fasting (IF) group or a standard dietary restriction (SDR) control group took place 1 2 weeks prior to commencing the dietary intervention. Study D iets In order to closely monitor provided by the CTRC metabolic kitchen. Dietary composition contained a macronutrient

PAGE 181

166 content of 55% carbohydrates, 15% protein and 30% fat. To minimize the risk of micronutrient deficiency parti cipants were given a multivitamin with iron to take every second day (the fed day for the IF group). Daily calorie distribution was 20% at breakfast, 30% at lunch, 40% at dinner and 10% for a snack. Baseline caloric intake for a weight maintenance diet was calculated to meet free living energy requirements based on the formula [(372 + 23.9 X FFM) x 1.5]. The accuracy of this estimate was tested during pre baseline weight maintenance diets and adjusted to ensure weight maintenance. The calories required to m aintain weight were used as the baseline caloric intake. Participants on the IF diet were given a standard rotating menu for a weight maintenance diet, as well as 5 7 optional additional food modules for each meal (200 kcal each, macronutrient content sim ilar to the rest of the diet) to allow the IF participant ad libitum access to food on their fed day. Provision of optional modules was adjusted as necessary throughout the intervention. Participants were given permission to eat as much as they wished from this food, but were not actively encouraged to eat all food provided. On fasted days participants were asked to consume only water and non caloric beverages. Participants in the control group on the SDR diet received a standard rotating menu with a 400kca l/day deficit. All participants were instructed to maintain their usual levels of physical activity without change, however physical activity was not measured. In order to minimize the risk of micronutrient deficiency, throughout the study both groups were given a multiple vitamin with iron (2 every other day, on feeding days, for the IF group in order to maximize absorption of fat soluble vitamins ). Week 1 V isit To begin the dietary intervention, all participants were admitted to the CTRC at the beginning of their first week and remained in patients for the entire week. For the first 4 days

PAGE 182

167 they were given a day pass to go to work, but were asked to return to sleep at the CTRC to allow for continued safety monitoring. Metabolic testing similar to that condu cted for the baseline visits began on day 5 of week 1. That is, on the 5 th day (a fasting day for the IF group and a fed day for the SDR group) of the week, participants underwent a testing procedure identical to that already described for day 1 of the bas eline visit. On the 6 th day of week 1 (a fed day for both IF and SDR groups) the testing procedure was again the same. The 7 th day of week 1 matched the procedures already described for day 2 of the baseline visit. As a result day 2 of the baseline fed vis it and day 7 of the week 1 visit provide directly comparable measures, assessing both groups after a fed day. After completing the 90% V02max stress test on the morning of the 7 th day participants were discharged, and given meals and instructions to comple te the remaining 7 weeks of intervention as closely monitored outpatients. They were instructed to maintain their usual level of physical in/activity. Intervention P eriod For the following 7 weeks all food for both groups was provided by the CTRC metabolic kitchen, so participants collected pre prepared meals from the metabolic kitchen twice a week. In the IF group food collection took place after both a fed day and after a fasted day, and urine was collected for ketone measurement after the fasted day to m easure dietary adherence. Weight was measured for all participants on each visit as another measure of adherence. For example, since the SDR group was expected to lose 6kg measured weight loss of <1kg/2wks was considered non compliant, and participants wer e given options to assist adherence to the study diet. Once a week participants also completed a weekly eating questionnaire and POMS. Week 8 V isit During week 8 participants followed their respective diets. During the week a separate visit for final wei ght measurement and a repeat DXA were scheduled, giving the final post

PAGE 183

168 intervention weight. Participants were also admitted for an in patient stay at the CTRC for days 5, 6 and 7 of week 8 to match the week 1 procedures described above. As a result day 7 o f the week 8 visit provides measures directly comparable t o day 7 of week 1. These match day 2 of the baseline fed visit, as each assess es outcomes after a fed day. At week 8 participants completed the questionnaires of eating disordered behavior (QEWP) an d depression symptoms (CESD) comparable to those assessed at baseline. On day 7 participants ended their dietary intervention and were discharged with instructions to maintain a balanced low fat diet as per the American Heart Association guidelines, but we re free to choose their diet and levels of activity. 6 month F ollow U p After 6 months participants were asked to return for a follow up visit. They were scheduled for weight measurement and another DXA, as well as an overnight in patient visit at the CTRC Pre visit dietary intake was not standardized. The testing conducted on the morning of the 6 month follow up visit resembled day 2 of the baseline visit s Participants also completed the CESD, QEWP and a standard 7 day dietary recall diary. They were the n fed a snack and discharged from the study. 4.2.5 Data A nalysis To control for baseline cognitive function and physiological function, repeated measures analysis of variance (ANOVA) of the effects of intervention group (IF or SDR) and time (baseline, week 8 an d 6 month follow up) were used for statistical comparisons in each cognitive domain (primary outcome). Between group differences at each time point were also calculated. Data were screened for outliers using frequency histograms, scatterplots and assessmen t. Outliers further than 3 standard deviations from the mean were assessed and analysis run both with and

PAGE 184

169 without these variables to assess their influence on results Normality was assessed for all cognitive domain scores. Cognitive domain scores were ap proximately normally distributed. Secondary analyses examined effects of intervention group and time on adiposity ( BMI, weight, and DXA determined percentage of body fat and percentage trunk fat), glucose regulation (SI, glucose, insulin), HPA axis functi on (plasma cortisol morning peak, diurnal decline and cortisol response to exercise stress ), pro inflammatory cytokines ( serum CRP, IL 6 and plasma TNF ) and neuro trophic (plasma BDNF) activity, using similarly structured ANOVAs. For secondary outcomes an d independent variables, normality, equality of variance and linearity of relationship to the cognitive outcomes assessed by residuals plots and scatterplots as part of regression diagnostics to prepare for regression with cognitive outcomes. Pairwise comp arisons were performed using t tests when appropriate. Two sided tests were used for all the comparisons, with a p value of 0.05 or less considered statistically significant and a p value of 0.01 or less considered highly statistically significant. All ana lyses were conducted with SPSS version 19 ( IBM ). For each variable measured, two sets of change scores were calculated. First, scores after the 8 week intervention, taken at 7 :00 am on a fed day, were subtracted from values corresponding to 7:00am after a baseline fed day. Similarly, 6 month change from baseline was calculated as the difference between scores at 7am on the 6 month follow up visit, subtracted from scores at 7:00am after a baseline fed day. While raw scores were used in analysis of variance, the change scores were used to assess correlation coefficients and in separate linear regression s to assess the association between change in cognitive domain (e.g. memory) and change in secondary outcome measures, including adiposity, diurnal cortisol, s timulated cortisol response (AUC), insulin sensitivity, inflammatory markers (TNF small sample size, principal covariates statistically considered for inclusion in the model included

PAGE 185

170 age, education and change in measures of adiposity. While gender may influence outcome variables, such as change in HPA axis function (e.g. see ( Baker et al., 2010 ) ), in our sample of 26, 20 participants were female, reducing the power of adjustments for gender. Diurnal cortisol was calculated using two measures. Peak morning cortisol, assessed at 6:30am of each study visit day, was used as the primary outcome for cortisol because it was available for each study visit, including the 6 month follow up visit. Cortisol diurnal decline was also calculated by subtracting evening cortisol nadir, taken at 9:30pm, from morning peak half an hour after waking (6:30am) for study days were evening nadir was available. Stimulated cortisol res ponse to the 90%V02max exercise stress test was calculated using cortisol output at each time point to generate area under the curve (AUC). 4.3 R esults 4.3.1 Study A ttrition A total of 46 potential participants were screened for eligibility at the University of Col orado Hospital. Of these, 8 did not qualify for study entry (7 for medical reasons, including hyperthyroidism, uncontrolled hypertension and untreated bipolar disorder, and inability to manage a weight stable diet). The distribution of age, gender, ethnic ity and education were similar to those in the remaining sample. Thirty eight remaining participants were considered eligible However 5 declined to participate (for reasons of family or not enough time) and 2 withdrew shortly after enrolment (moving from Denver, new job responsibilities) Age, gender, education and ethnicity were similar to those found in the remaining sample. Hence only 31 completed the test fast, while 29 completed at least one test fast. Twenty nine participants were randomized to the intervention. Of these 2 withdrew (did not like IVs, scheduling issues) and one person was dropped from the study (poor IV access, never started intervention) No systematic differences in age, gender,

PAGE 186

171 education ethnicity or BMI were detectable between th e people who withdrew and those who chose to continue. A total of 26 participants completed the intervention to its conclusion at 8 weeks. Of these, 14 were randomized the IF group, while 12 were randomized to the SDR group. Of the completers 19 returned for follow up at 6 months. In addition to attrition some variables suffered loss of data at different time points. The reasons for the loss of data varied, but included technician error, laboratory error, or the inability to contact study personnel respo nsible for the data, as described above in the methods data available for that measure at that time point if the full sample sizes was not available. 4.3.2 Sample C haracteristics Twenty six obese (mean BMI=37.1, sd=5.3) adults aged 23 55 years completed the DRIFT study. Of these, 20 were women. Two participants self identified as African American, 6 as Hispanic and 1 as other ethnicity. For one participant English wa s his second language, but he had completed a Ph.D. in English and was fluent in English. This participant completed the cognitive tests in Spanish. As can be seen in Table 4.3 1 some minor differences between the groups were apparent at baseline. Both African American participants were randomized to the IF dietary intervention group and more participants in the IF group reported being physically active at baseline than persons randomized to the SDR group. In addit ion the SDR group showed faster cognitive speed at baseline (p=0.049). Otherwise baseline characteristics of the two groups did not differ significantly.

PAGE 187

172 Table 4.3 1 DRIFT sample characteristics at baseline after a fed day (mean (sd) unless otherwise indicated). Characteristic Total IF Group SDR Group p Sample size (n) 26 14 12 Age (years) 40.5 (9.0) 38.7 ( 9.7 ) 42.6 ( 7.9 ) Education (n) Completed college Some college or assoc. Completed HS or equiv. 26 15 9 2 14 9 5 0 12 6 4 2 Gender (n) Male Female 6 20 3 11 3 9 Ethnicity (n) Non Hispanic White Hispanic African American Other 17 6 2 1 8 3 2 0 9 3 0 1 Depression (CESD) 15.8 ( 4.3 ) 15.7 ( 5.6 ) 15.8 ( 2.3 ) 0.945 Eating Disorders (QEWP) 0 0 0 Self reported phys ical activity Active Some activity Sedentary 21 11 3 7 12 9 0 3 9 2 3 4 Adiposity BMI 37.1 ( 5.3 ) 35.6 ( 4.2 ) 38.8 (6.0) 0.116 Fat mass (%) 41.8 ( 5.7 ) 40.5 ( 6.2 ) 43.3 ( 4.8 ) 0.213 Trunk fat (%) 43.7 ( 5.7 ) 42.6 ( 6.7 ) 45.0 ( 4.2 ) 0.288 HPA axis funct ion 0.66 ( .25 ) 0.63 ( .27 ) 0.70 ( .25 ) 0.657 0.60 ( .29 ) 0.55 ( .34 ) 0.66 ( .23 ) 0.488 90% V02max cortisol AUC (n=22) 31 ( 24 ) 35 ( 31 ) 27 ( 13 ) 0.493 Cognitive function Memory 97 ( 7.5 ) 98 ( 5.8 ) 95 ( 9.2 ) 0.447 Attention^ 4.5 ( 3.7 ) 4.2 ( 3.9 ) 4.9 ( 3.7 ) 0.632 Reaction time^ 620 ( 71 ) 615 ( 70) 624 ( 74 ) 0.760 Cognitive flexibility 56 ( 8.5 ) 57 ( 9.8 ) 55 ( 6.9 ) 0.453 Cognitive speed 184 ( 22.5 ) 192 ( 25.5 ) 175 ( 15.0 ) 0.049* Biomarkers Insulin sensitivity (SI) 1.62 ( 1.1 ) 1.90 ( 1.2 ) 1.29 ( .96 ) 0.205 Glucose (mg/dL) 91.13 (7.82) 92.27 (7.16) 90.09 (8.21) 0.505 Insulin ( IU/mL ) 18.71 (6.58) 15.58 (4.78) 21.83 (6.82) 0.521 BDNF (pg/mL ) 20253 ( 5197 ) 19130 ( 5569 ) 21469 ( 4689 ) 0.270 Leptin (ng/mL) 33.49 (13.62) 33.12 (12.38) 33.87 (15.31) 0.896 Note: Significant at p=0.05. ^ Lower scores represent better function.

PAGE 188

173 4.3.3 Safety A safety monitor (DSMP) followed the DRIFT study closely. Four minor adverse events were r eported to the safety officer (headaches). One person experienced DVT during the course of the study, but it was determined that this was not related to study involvement. One person had cholycystectomy but not gallstones, which was also unlikely to be ca used by study involvement. 4.3.4 Dietary A dherence All participants were asked to remain as in patients in the CTRC unit in the first week of the intervention to ensure adherence to the diet and to monitor for safety. After their discharge, adherence over the ensuing 7 weeks of the intervention was monitored on a bi weekly basis in a number of ways. All meals were provided to both groups, and participants were asked to return uneaten food, which was weighed as a measure of compliance with the diet. All partici pants were weighed at these vi sits, and urine samples were also collected from participants in the IF group after a fasting day to test for ketones Based on the expected weight loss of 6kg in the CR group and 6 11kg in the IF group a weight loss < 1kg/2we eks was considered non compliant for either group. Participants also completed a weekly eating questionnaire, which was checked for evidence of dietary non compliance. Participants who were non compliant were encouraged to work collaboratively with the stu dy team to develop strategies that would increase the tolerability of the diet, and/or asked to work with the study team to develop minor adjustments to the diet to optimize adherence. While two participants did not demonstrate the anticipated weight loss by the end of the study period (both IF, stated reason was a felt need to consume all the food provided by the metabolic kitchen, which was made available in excess of caloric needs for the IF group) no participants were considered non compliant, and no p articipants were

PAGE 189

174 dropped from the study for reasons of non adherence to the diet. In addition no participants withdrew from the intervention because of difficulty tolerating either the diet. 4.3.5 Acute E ffects of F asting Acute effects of a single short term (36 h) fast on obese adults were assessed with the full sample, before randomization (n=26). It was not possible to determine whether participants found fasting psychologically stressful due to the large number of missing questionnaires for the baseline visits (technician error, POMS and PSS not administered or misplaced) However short term fasting did produce mild but significant reductions in glucose consistent with mild metabolic stress. As can be seen in Table 4.3 2 below, serum glucose and insulin were significantly lower after a day of fasting than after a fed day (p<0.001) but insulin sensitivity decreased (p=0.010) Cortisol morning peak was also significantly lower after a fasted day than after a fed day (p=0 .002). Short term fasting did not have any significant effects on any cognitive domains Most other physiological markers were unchanged by acute fast ing To determine whether regular exposure to short term fasting produces habituation to the effects of fa sting, the effect of a matched 36 hour fasting day on in week 8 (day 5) was assessed for participants in the IF group only. Fasting response was not assessed for the SDR group, as they did not undergo fasting at 8 weeks. Within group comparison of the base line and week 8 fasts for the 14 participants in the IF group indicated that the decrease in glucose after a fasting day was the same at both times (mean change=0.000, p=1.00). However the morning peak cortisol after a fast was significantly higher after 8 w eeks of the IF intervention than that

PAGE 190

175 Table 4.3 2 Effects of an acute 36h fast at baseline on all participants before randomizat ion (n=26). Measures were collected at 7am the following day. Characteristic After a fed day Mean (sd) After 36h fast Mean (sd) p Physiological stress Glucose ( mg/dL) (n=23) 91.1 ( 7.6 ) 84.2 ( 5.3 ) .000 ** Insulin ( IU/mL ) (n=23) 18.9 ( 6.7 ) 12.0 ( 4 .5 ) .000 ** Cortisol morning peak ( ) (n=14) .66 ( .25 ) .46 ( .20 ) .002 ** Psychological stress Perceived stress (PSS) Insufficient n Insufficient n N/A Tension/anxiety (POMS) Insufficient n Insufficient n N/A Mood score (POMS) Insufficient n Ins ufficient n N/A Cognitive f unction Memory (n=25) 96.6 ( 7.5 ) 97.0 ( 11.1 ) .792 Attention ^ (n=24) 4.5 ( 3.7 ) 7.8 ( 10.5 ) .074 Reaction time ^ (n=25) 620 ( 70.7 ) 609 ( 58.2 ) .318 Cognitive flexibility (n=25) 56 ( 8.5 ) 53 ( 15.4 ) .111 Cognitive speed (n=25) 184 ( 22.5 ) 181 ( 19.7 ) .411 Biomarkers Insulin sensitivity (SI) (n=24) 1.62 ( 1.15 ) 1.24 ( 1.09 ) .010* BDNF ( pg/mL ) 20126 ( 5270 ) 20649 ( 6054 ) .728 HPA axis function Cortisol decline ( ) (n=14) .597 ( .29 ) .472 ( .25 ) .306 90% V02max cortisol AUC (n=22) 31.14 ( 24.4 ) 29.50 ( 27.1 ) .566 Notes: ^ lower scores indicate better performance statistically significant, p<0.05. ** highly statistically significant, p<0.01. All 26 participants participated in both baseline fed and baseline fasted visits. Where data were missing for specific variables the actual sample size used in the calculation is indicated above.

PAGE 191

176 4.3.6 Effects of IF on B ody W eight and C omposition Figure 4.3 1 Eight week change in body weight by group (kg). After 8 weeks of dietary restriction both the IF and SDR groups lost significant amounts of weight (p<0.05), losing an average of 6.5% (sd=5.0%), and 5.75% ( sd=2.5%) of body weight respectively. As seen in Table 4.3 3 below weeks was not statistically significant. However d ata screening revealed that although all oth er participants lost weight by 8 weeks, one participant in the IF group gained weight by the end of the intervention (6.95kg), and another in the same group maintained her baseline weight. This is depicted in Figure 4.3 1 Both participants were African American. After excluding these two participants from the comparison of weight change, mean weight loss in the IF group was 8.0%

PAGE 192

177 (sd=3.41), though the difference in weight loss between the two groups was not statisti cally significance (p=0.082). Since the participant who gained weight did not return for 6 month follow up this person was automatically excluded from analysis of variance and regression across the three key study times. The person who maintained her weigh t was retained in the analyses. Figure 4.3 2 Eight week percent weight change for each participant. Analysis of variance indicated that both groups decreased their proportion of fat mass over time (p=0. 01), with a trend for between group differences (p=0.056). As shown in Figure 4.3 3 p articipants in the IF group continue d to los e weight after the intervention, while t he SDR group did not However the interactio n did not reach sta tistical significance (p=0.071). A 6 month between group differe nce in weight lost is apparent ( Table 4.3 5 ).

PAGE 193

178 Figure 4.3 3 Mean change in perc ent fat mass by time and group. Figure 4.3 4 Mean differences in percent trunk fat by time and group. While both groups lost trunk fat during the 8 week intervention analysis of variance revealed a significant interaction between intervention group and time (p=0.008) A s shown in Figure 4.3 4 participants in the IF group continued to lose trunk fat after the end of the intervention while the SDR group did n ot The 6 month change in trunk fat showed a strong

PAGE 194

179 correlation with 6 month change in BMI (0.65, p=0.002) and fat mass (R=0.79, p=0.000). As noted in Table 4.3 3 change scores used in correlations and regressions were calculated by subtracting 6 month weight from baseline weight. As a result, positive change score s in analyses represent a better weight outcome (i.e. more weight lost).

PAGE 195

180 Table 4.3 3 Eight week post intervention group differences in weight and cognitive function. Characteristic IF Group mean (sd) SDR Group mean (sd) p Sample size~ 14 12 Adiposity Change in weight (%) 6.50 (5.03) 5.75 (2.56) 0.65 BMI 33.43 ( 4.70 ) 36.75 ( 5.80 ) 0.12 Change in B MI 2.29 (1.73) 2.25 (1.22) 0.95 Fat mass (%) 39.57(6.51) 42.42(5.07) 0.22 Change in % fat mass 1.07 (1.27) 1.00 (1.13) 0.63 Trunk fat (%) 41.29(7.08) 44.00 (4.57) 0.25 Change in % trunk fat 10.07 (7.74) 8.58 (5.66) 0.59 Cognitive function Memory 98.00 (12.25) 95.67(10.52) 0.61 Change in memory .00 (11.17) .33 (8.08) 0.93 Attention ^ 5.50 (3.59) 6.08(5.66) 0.75 Change in attention 1.62 (4.61) 1.63 (5.26) 0.99 Reaction time ^ 620.36 (59.51) 632.08 (85.49) 0.69 Change in reaction t ime 12.15 (79.78) 7.29 (64.89) 0.87 Cognitive flexibility 57.93 (7.44) 57.58 (7.59) 0.91 Change in cognitive flexibility .38 (9.75) 2.50 (9.59) 0.46 Cognitive speed 193.00 (19.86) 174.83 (20.23) 0.03 Change in cognitive speed 0.4615 (14.08) 0.333 (11.86) 0.84 Depression (CESD) 8.15 (6.44) n=13 5.91 (3.91) n=11 0.32 Change in depression 7.54 (5.61) 9.82 (3.28) Notes: ^ lower scores indicate better performance. *Statistically significant at p<0.05. Change scores calculated as change from b aseline (baseline 8 weeks). ~All 26 participants participated in the baseline and 8 week study visits. Where data were missing for specific variables the actual sample size used in the calculation is indicated above. Sample sizes are only provided where the sample differed from the full sample.

PAGE 196

181 Table 4.3 4 Eight week post intervention group differences in biomarkers. Characteristic IF Group mean (sd) SDR Group mean (sd) p Sample size~ 14 12 HPA axis function 0.59 (.36) n=13 0.65 (.28) n=10 0.66 Change in cortisol morning peak .03 (0.45) n=8 .09 (.50) n=8 0.64 0.56 (0.33) n=12 0.64 (0.317) n=10 0.55 0.0 8 (0.16) 0.08 (0.10) n=6 0.059 90% V02max cortisol AUC 28.56 (26.84) n=9 15.00 (7.07) n=9 0.16 Change in 90% V02max cort AUC 1.44 (21.83) n=9 13.56 (12.48) n=9 0.09 Biomarkers Insulin sensitivity (SI) 2.04 (1.42) n=12 1.55 (.756) n=10 0.34 C hange in Insulin Sensitivity .17 (.54) n=12 .21 (.67) n=10 0.88 Glucose (mg/dL) 93.08 (5.52) 96.50 (8.63) 0.260 Change in glucose 1.22 (6.30) n=9 6.42 (8.61) 0.14 Insulin ( ) 15.42 (5.99) 17.17 (7.09) 0.521 Change in insulin .80 (5.49) 4.67 (6.67) 0.05 BDNF (pg/mL) 18127.79 (7410.36) 20343.75 (5825.80) 0.411 Change in BDNF 1177.08 (9522.81) 1125.5 (6020.05) 0.99 Leptin (ng/mL) 27.23 (13.48) 25.13 (12.89) 0.700 Change in leptin 6.28 (7.26) n=10 8.73 (6.10) 0.40 Notes: ^ lower scores indicate better performance. *Statistically significant at p<0.05. Change scores calculated as change from baseline (baseline 8 weeks). ~All 26 participants participated in t he baseline and 8 week study visits. Where data were missing for specific variables the actual sample size used in the calculation is indicated above. Sample sizes are only provided where the sample differed from the full sample.

PAGE 197

182 4.3.7 Effects of IF on C ogn itive F unction To test the hypothesis that IF improves memory function in obese humans, effects of IF on cognitive function were measured. A fter the 8 week interventions neither group showed significant changes from baseline in any cognitive domains Ho wever by the 6 month follow up the IF group showed significantly greater improvements in memory relative to baseline performance than the SDR group and analysis of variance indicated a significant main effect of group (p =0.013 ). No other cognitive domains showed significant change at the 6 month follow up. The changes in memory performance are shown in Figure 4.3 5 For the CNS Vital Signs cognitive test battery, higher memory scores represent better performance. Figure 4.3 5 Mean memory scores by time and intervention group. The effects of adjusting for weight or adiposity were assessed to determine whether changes in cognitive function were mediated by changes in weight or fat distribution. The effect of intervention group on 6 month change in memory remained significant after controlling for 6

PAGE 198

183 month change in BMI (p=0.013) or after controlling for 6 month change in percentage fat mass (p=0.021). However after c ontrolling for 6 month change in percentage trunk fat the between group differences in memory were no longer statistically significant (p=0.088). As described above, there was an interaction between group and time for trunk fat, such that participants in t he IF group continued to lose trunk fat after the intervention, while the SDR group did not. Linear regression showed that change in trunk fat was a significant predictor of change in 8). 4.3.8 Effects of IF on G lucose R egulation Effects of IF on glucose regulation were explored to investigate whether change in glucose regulation might mediate the effect of IF on memory. As can be seen in Figure 4.3 6 analysis of variance showed a trend for improved insulin sensitivity in the IF group at 6 months that was not apparent for the SDR controls. The trend did not reach statistical significance even though the two groups were different at 6 months. Insulin sensitivity at 6 months was significantly associated with 6 month change in insulin (R= 0.676, p=0.011, n=13), but not associated with change in glucose, or changes in weight. Simple linear regression models showed that there was no association between 6 month change in memory and change in insulin sensitivity, glucose, or insulin.

PAGE 199

184 Figure 4.3 6 M ean insulin sensitivity (SI) by time and group. 4.3.9 Effects of IF on BDNF Effects of IF on BDNF were explored to i nvestigate whether change in BDNF might mediate the effect of IF on memory. Analysis of variance indicated a significant time by group interaction effect for BDNF. As can be seen in Figure 4.3 7 BDNF did not cha nge significantly for either group by the end of the 8 week intervention. However participants in the IF group showed increases in BDNF at 6 month follow up (p=0.07), while the SDR group tended towards a decrease in BDNF. In separate models the interaction remained significant after controlling for 6 month change in BMI (p=0.010), or 6 month change in percent fat mass (p=0.011). However the interaction was no longer significant after controlling for change in percentage trunk fat (p=0.056). Controlling for change in insulin sensitivity increased the significance of the interaction (p=0.004), and remained significant after controlling for change in glucose (p=0.040).

PAGE 200

185 Increased BDNF was associated with reductions in glucose, but the association was no longer s ignificant after controlling for intervention group. The timing and direction of the changes in BDNF reflected those found for memory. However no evidence was available that BDNF mediated the effect of IF on memory function. Simple linear regression model s showed that there was no association between 6 month change in memory and change in BDNF. Figure 4.3 7 Mean BDNF by time and group (pg/mL). 4.3.10 Effects of IF on HPA A xis F unction There were insufficien t data to test whether HPA axis function mediate the 6 month change in memory. Due to circumstances beyond our control, some cortisol data could not be retrieved for statistical analysis. With the sub set of data available, only a relatively small number o cross all of the three time points. Repeated measures analysis of variance therefore included only 5 participants from the IF group and 4

PAGE 201

186 participants from the SDR group, and did not show statistical significance. B etw een group comparisons at 8 weeks showed no differences in peak morning cortisol or change in peak cortisol at 8 weeks. Between group comparison of change in stimulated cortisol output in response to exercise stress was no t significant (p=0.092 ). These res ults did not differ after excluding participants who gained or maintained weight. However a fter excluding participants in the IF group who either gained or maintained their weight (n=2), the IF group showed significant increases in cortisol decline (morni ng peak evening nadir) after the 8 week interven tion, relative to the SDR group (p=0.039, n=13). When the IF participants who did not lose weight were included the comparison approached statistical significance (p=0.053, n=15). Cortisol decline at 8 weeks predicted memory scores the following but regression of changes from baseline did not reach significance (n=12) Six month changes in cortisol decline could not be assessed since full 24h cortisol was not collected at 6 month follow up. There were no significant between group differences in other measures of HPA axis function at 6 month follow up. However 6 month c hange in peak morning cortisol from baseline to 6 months was strongly correlated with 6 month change in glucose ( R=0.914, p=0.000, n=7), though not with change in insulin or insulin sensitivity. Although correlation coefficients between peak morning cortisol and change in percent trunk fat and percent fat mass were high (0.51 and 0.62 respectively) they did not reach statistical significance in a sample size of 7. With very few participants with cognitive and cortisol data at both baseline and 6 months (n=5), change in peak cortisol was not a significant predictor of memory (p=0.099). However there was a trend for an association between higher morning cortisol and better same day memory scores at the 6 month visit ( adj R 2 0.464, p=0.06 1 n= 17 ).

PAGE 202

187 4.3.11 Effects of IF on Pro I nflammatory C ytokines There were insufficient data to test whether changes in inflammatory markers had the potential to partially mediate the 6 month change in memory. Data for inflammatory markers were unreliable due to systematic inter assay differences between the results obtained by different laboratory personnel, making interpretation of these data difficult. The data for TNF suggest that both groups showed increased TNF a at 8 weeks, with no significant between group differences. At 6 month follow up TNF in the IF group appears to decline towards baseline, while the SDR group continued to rise. However calculated values systematically increased across different assays, and later study time poi nts were more likely to be done on later assays significant variability in the assays for CRP and IL 6 rendered these data uninterpretable. 4.3.12 Effects of IF on L eptin Effects of IF on leptin were explored to investigate whether change in leptin might partia lly mediate the effect of IF on memory. As shown in Figure 4.3 8 t here were no significant between group differences in leptin at any time point. However both groups showed a highly significant decrease in leptin a t 8 weeks (p=0.000). Simple linear regression showed that change in trunk fat and change in fat mass were significant predictors of this 8 week change in leptin 0.476, SE=0.192, p=0.025 respectively) As with other studies, consistent gender differences in leptin levels were observed. Women had significantly higher levels of leptin than men. Change in leptin did not predict change in memory scores. However leptin at 8 weeks did predict memory at 8 w eeks, accounting for 17.1% of the variance in memory score at that 0.455, SE=0.168, p=0.026).. Similarly leptin at baseline predicted memory at baseline

PAGE 203

188 0.353, SE=0.107, p=0.000). Leptin at 6 months did not predict memory at that time. Figure 4.3 8 Mean leptin by time and group (n g/mL).

PAGE 204

189 Table 4.3 5 Between group differences 6 months after the end of the interve ntions. Characteristic IF Group Mean (sd) SDR Group Mean (sd) p Sample size~ n = 10 n = 10 BMI 31.70 ( 2.79 ) 38.50 ( 6.80 ) .023* Change in BMI 2.00 (2.21) 0.70 (1.57) 0.147 Trunk fat (%) 38.60 (.95) 45.70 (4.90) 0.441 Change in % trunk fat 2.40 (1.71) 0.00 (1.8) 0.008* HPA axis function 0.50 (0.24) n=9 .484700 (0.25) n=10 0.671 Change in cortisol morning peak 0.23 (0.32) n=5 0.07 (0.07) n=4 0.376 90% V02max cortisol AUC 21.40 (15.32) n=5 15.20 (6.8 3) n=5 0.367 Change in 90% V02max 8.4 (16.23) n=5 10.40 (11.71) n=5 0.829 Cognitive function Memory 104.75 (5.26) n=8 93.80 (8.07) n=10 0.450 Change in memory 7.29 (8.32) n=7 0.9 (8.79) n=10 0.006* Attention^ 3.25 (3.85) n=8 6.00 (3.43) n=9 0.648 Change in attention .43 (3.65) n=7 .38 (4.03) n=8 0.694 Reaction time^ 592.13 (55.85) n=8 602.44 (65.85) n=9 0.841 Change in reaction time 7.00 (72.78) n=7 18.89 (46.83) n=9 0.697 Cognitive flexibility 63.00 (7.84) n=8 51.20 (15.84) n=10 0.299 Cha nge in cognitive flexibility 1.43 (5.50) n=7 3.20 (17.86) n=10 0.520 Cognitive speed 193.63 (22.14) n=8 179.56 (17.47) n=9 0.418 Change in cognitive speed 9.43 (9.03) n=7 4.00 (9.03) n=9 0.021 Biomarkers Insulin sensitivity (SI) 3.04 (3.84) n=7 1 .31 (.60) n=9 0.046* Change in insulin sensitivity 0.97 (2.27) n=7 0.55 (0.76) n=9 0.218 Glucose (mg/dL) 85.63 (5.85) 93.30 (6.33) 0.018* Change in glucose 2.17 (7.08) n=6 2.90 (4.28) n=10 0.093 Insulin ( ) 12.86 (5.38) 16.80 (4.27) 0.121 Change in insulin 1.17 (2.87) n=7 5.70 (6.07) n=10 0.130 BDNF (pg/mL) 26401.11 (6170.05) n=9 18498.60 (5514.60) n=10 0.613 Change in BDNF 6970.62 (6790.56) n=8 3332.90 (7313.31) n=10 0.007* Leptin (ng/mL) 27.85 (15.16) n=8 33.23 (18.17) n=10 0.512 Change in leptin 5.47 (16.53) n=7 1.21 (4.90) n=10 0.449 Notes: ^ lower scores indicate better performance. statistically significant at p<0.05; ** statistically significant at p<0.001. Change scores are calculated as change from baseline (baseline 6 months). ~20 participants returned for the 6 month follow up visits. Where data were missing for specific variables the actual sample size used in the calculation for that variable is indicated above.

PAGE 205

190 4.4 D iscussion As hypothesized, IF produced significantly greater effects on the cognitive function of obese adults than standard dietary restriction. The observed changes in cognitive function were specific to improvements in memory, a finding consistent with the effect s of IF in animals (Anson et al, 2003 ). Interestingly, t hese effects on memory were not apparent at the end of the 8 week intervention Instead memory effects appeared 6 months after the interventions concluded. It is therefore significant that by 6 months the IF group showed significantly greater reductions in trunk fat than the SDR group. This reduction in trunk fat partially mediated the change in memory performance observed at 6 months. B oth dietary interventions significantly reduced global adiposity, as indicated by changes in BMI weight and fat mass; however between group differences were not statistically significant at 8 weeks or 6 months and were not related to change in memory scores. While these findings do not prove that central obesity was th e cause of cognitive deficits in study participants, t he results are consistent with the hypothesis that factors specifically related to central adiposity affect human memory function. This is consistent with e pidemiological evidence of an association betw een central obesity in midlife and later cognitive decline ( Cereda, Sacchi, & Malavazos, 2009 ; West & Haan, 2009 ; Whitmer et al., 2008 ) and argues for more widespread inclusion of measures of central obesity in future epidemiological and intervention studies. Several plausible mechanisms link central obesity to co gnitive decline and dementia. Excess central adipose tissue can contribute to glucose dysregulation, low grade systemic inflammation (Barzilay et al, 2001), and HPA axis dysregulation ( Bjorntorp & Rosmond, 2000 ) each of which may play a causal role in the p ( Craft, 2005 2007 ; De Leon et al., 1997 ; Lupien et al., 1999 ) However as previously discussed, many research studies to date have used BMI as their sole measure of adiposity when testing for an

PAGE 206

191 association between obesity a nd cognitive function. However BMI gives a poor reflection of body composition, and cannot be used assess any specific risks attributable to central obesity. T his study did not collect data that would determine why participants in the IF group but not t he SDR group, continued to lose weight after the conclusion of the intervention. However there are a number of possible explanations Many possible factors could explain these differences, but no additional measures were added to the 6 month follow up that could be used to give clear answers. All measures at the 6 month follow up mirrored those of the preceding visits. For example it would have been useful to collect data on participant behavior in the intervening time, including dietary patterns and rates of physical activity. Among the possible explanations for the differences, it is possible that participants in the IF group continued to engage in some form of regular fasting. No systematic data were collected on post intervention dietary practices, howev er the experience of IF and the behavior change skills acquired during the intervention period would have been quite different to those of the SDR group. Anecdotally, some people find IF more simple and easy to follow than calorie restriction. Different po st intervention dietary behavior would account for the additional weight loss in the IF group observed at the 6 month follow up, but it would not fully explain the between group differences in fat distribution, for t he IF group lost significantly more trun k fat than the SDR group. Another possibility for the difference in fat distribution between the groups at 6 months is that IF produced different neuroendocrine changes than SDR. It may have had differential effects in impro ving metabolic efficiency, or i mproving regulat ion of a wide variety of hormones involved in energy balance. In addition, IF can be considered a form of repeated mild metabolic stress. The hormesis hypothesis ( Calabrese & Baldwin, 2001 ; Mattson & Calabrese, 2010 ) suggests that this mild stress can be beneficial, and co ntributes to the beneficial effects of IF by

PAGE 207

192 improving function of stress response systems and resilience against stress ( Martin et al., 2006 ) According to this hypothesis short term fasting is psychologically and metabolically stressful Consistent with this, we found that acute (36h) fasting induced mild decrease s in glucose. As expected, the fast ing induced change in glucose was significantly correlated with changes in cortisol, however the direction of the association was opposite to that hypothesized. The mild reductions in glucose observed after fasting did not appear to provoke a counter regul atory increase in cortisol in this sample. Instead reductions in glucose were associated with decreased morning cortisol peak. The reason for this counter intuitive response to fasting at baseline is not clear. It may reflect a dysregulated baseline state, for the cortisol response to a fast changed after 8 weeks. Eight week changes in fasting response could only be assessed within the IF group, since they alone underwent fasting at that time response to a short ter m 36h fast remained the same as baseline suggesting that a short term fast continued to represent a mild metabolic stressor throughout the study. However after 8 than baseline fasting response (p=0.001) even though peak morning cortisol after a fed day remained unchanged. Hence participants mounted a stronger cortisol response to an acute fast after 8 weeks of IF. A strong cortisol response to fasting may be an a daptive, healthy counter regulatory response ( Bergendahl, Iranmanesh, Mulligan, & Vel dhuis, 2000 ) In animal research the APP mutant m ouse model of AD, which shows HPA axis abnormalities and excess amyloid deposition in the brain, was apparently unable to mount an adaptive response to fasting when placed on an IF diet, and died within w eeks of starting the diet ( Pedersen et al., 1999 ) A healthy adaptation to fasting may therefore be an important to the effects of IF, and t his finding in humans is consistent wi th an intervention related change in the HPA axi s response to fasting. Since the effect could not be compared to the control group it is not possible to determine if it

PAGE 208

193 was specific to the IF intervention. This study is the first to assess the effects of I F on human HPA axis function using rigorous standardized measures. Although only a partial dataset was finally available for most measures of HPA axis function, trends within the available data were consistent with intervention related changes in HPA axis function. These included a tendency for response to the exercise stress test to increase for the IF but not SDR group, greater increase in diurnal cortisol decline after the 8 week intervention in the IF group ( p=0.039, n=13) that occurred before significa nt between group differences in trunk fat were apparent. Since both groups lost a similar amount of weight and trunk fat by 8 weeks the difference s are unlikely to be attributable to reductions in adiposity alone Hence it may provide some evidence to sup port the idea that altered neuroendocrine effects contributed to the ongoing IF group. The results of this study were also consistent with existing evidence surrounding mechanisms that could mediate an effect of obesity on neurocognitive health. As hypothe sized, IF produced significantly greater improvements in BDNF than the SDR control, and these differences remained afte r adjusting for weight change, percent fat mass change or change in BMI. However controlling for change in trunk fat attenuated the inter vention related effect on BDNF and halved the effect size, though the effect still approached significance (p=0.056) Interestingly, change in BDNF was not associated with change in memory, although both were improved in the IF group at 6 month follow up, and both effects appeared to be partially mediated by changes in trunk fat. Participants varied in whether they improved on memory or BDNF, and these did not co vary together. It is unlikely that changes in plasma BDNF contributed causally to improvements in memory in this study. Other human research has indicated lo w BDNF in persons who are obese or who have dementia ( Cole & Frautschy, 2007 ; Vaynman & Gomez Pinilla, 2006 ) and higher levels of BDNF are associated with better learning and memory in animals (neural BDNF; e.g. ( Diogenes et al., 2007 ) and humans (plasma BDNF;

PAGE 209

194 ( Alonso et al., 2002 ; Komulainen et al., 2008 ) Howev er BDNF is thought to be a neuroprotective factor that is upregulated in response to potential neuronal stress, injury, or inflammation ( Yasutake et al., 2006a ) Johnson et al. (2007 ) found that modified IF led to decreased BDNF in obese persons with asthma, and that these reductions corresponded to reductions in infla mmation and asthma symptoms. It is possible that part of our sample entered the study with heightened BDNF in response to inflammation or other stressors, and that BDNF decrea sed for this sub group as inflammation decreased. Since the pro inflammatory cyto kine data collected as part of this study were not reliable, it is not pos sible to test this possibility, and future research should explore this further. It does, however, highlight an issue with BDNF testing and interpretation. As currently conceptualize d, high BDNF can represent either the good health that comes with a neuroprotective factor, or the poor health that requires an adaptive protective response. Other factors are therefore needed to aide the interpretation of BDNF results. In this study, inc reases in BDNF were related to reductions in glucose Since the brain relies on glucose for fuel, BDNF may have been mobilized as part of a protective response against the threat of glucose deprivation. Only the IF group had recurring periods of glucose de privation. The exposure to repeated glucose deprivation may have shaped the between group differences in BDNF, however further research with larger sample sizes is needed to parse out these effects from those attributable to intervention group or change in trunk fat. Extensive mediation modeling of the effects of changes in BDNF, glucose regulation, inflammatory cytokines or HPA axis function on memory were beyond the power of this exploratory study. However trends in the data are consistent with beneficia l effects of IF on central adiposity, glucose regulation, HPA axis function and trophic factors. Each of these factors is independently implicated in the pathophysiology of age related cognitive decline and dementia. Given the animal literature supporting significant neuroprotective effects for IF,

PAGE 210

195 including slower cognitive aging, the effects of IF on these potential mediating mechanisms should be explored further. In this first test of the effects of intermittent fasting on human cognitive function it is worth noting that IF did not have detrimental effects on any cognitive domain measured, including memory, attention, cognitive flexibility, cognitive speed and reaction time. Similarly, supervised weight loss by standard dietary restriction did not result in cognitive deficits. Before contemplating weight loss as a potential solution any increased risk of cognitive decline and dementia in obese adults, it is important to establish the safety and efficacy of weight loss. The direction of the effect is impor tant, particularly in light of the association between weight loss and adverse cognitive outcomes in some older adults ( Stewart et al., 2005 ; Wirth et al., 2007 ) adverse effects of Ramadan fasting ( Roky et al., 2003 ; Roky et al ., 2004 ; Roky et al., 2000 ) and some findings of worse cognitive function after unsupported dieting, these results are potentially important ( Green & Rogers, 1995 1998 ; Green, Rogers, Elliman, & Gatenby, 1994 ) Results of this study suggest that it is safe to tes t the effects of IF more widely on the human population, and that participants in such studies may experience significant health benefits that endure beyond the conclusion of the study. Follow up research in populations at increased risk of cognitive decli ne, including persons with Mild Cognitive Impairment (MCI), Type 2 diabetes mellitus or the Metabolic syndrome, or persons with HPA axis dysregulation, should be conducted to determine whether IF may reduce risk of cognitive decline. Such research should i nclude sample sizes sufficient to provide the power to detect significant changes in HPA axis function, to determine whether beneficial effects on memory related to increased resilience against stress. Although IF is arguably stressful, both metabolically and psychologically, it produced greater improvements in memory and BDNF for obese adults than weight loss alone. The diet

PAGE 211

196 produced significant changes in potentially mediating factors, including BDNF and leptin, and t rends for concurrent improvements in insulin sensitivity and HPA a xis function were also apparent. These results are consistent with animal research on calorie restriction and intermittent fasting, which finds specific improvements in learning and memory ( Duan, 2003 ; Duan et al., 2003a ; Martin et al., 2006 ) Although IF is likely to be difficult to tra nslate into regular community practice, it can serve as a useful tool to explore mechanisms that may improve neurocognitive function across the lifespan. Such research may provide the foundation for more widely acceptable interventions to promote cognitive health. In addition, it is worth exploring the frequency and duration of fasting required to produce beneficial effects. Taken together, the results of this exploratory study suggest that IF involves repeated exposure to mild metabolic stress, with effect s that are consistent with the hypothesis that IF can improve memory and increase BDNF in obese adults, and may contributes to more healthy HPA axis function. As the first published test of the effects of IF on human cognitive function, and the first to im plement controlled measures of HPA axis function, this exploratory study paves the way for future research into the potential neuroprotective effects of intermittent fasting.

PAGE 212

197 5. D iscussion st feared risks of growing older ( Anderson & McConnell, 2007 ) However it is not yet clear what differentiates those who experience serious cognitive decline from those who show more healthy cognitive aging. pathological and functional changes decades before dementia diagnosis h as led to a search for factors that can affect neural and cognitive function in midlife or even earlier ( Sperling et al., 2011 ) The lifecourse approach to chro nic disease etiology ( Kuh & Ben Shlomo, 20 04 ) suggests that prolonged exposure over many years from even small neurotoxic insults could contribute to cumulative damage that could contribute to more rapid cognitive aging. 5.1 A L ifecourse A pproach to C ognitive A ging A number of important pieces of i nformation are needed to piece together the jigsaw nce of factors that can damage human neurocognitive health at any age It is now considered likely that AD is a lifespan disease, not simp ly a disease of old age ( Gustafson, 2008 ) Factors that can protect human neurocognitive health are also particularly needed. Whether or not they directly affe ct AD etiology, factors that contribute to increased cognitive reserve ( Stern, 2002 2009 ) may go a long way to buffering the brain against the ravages of A D. Some utopsy studies ( Roe et al., 2007 ) indicate that some people can have AD pathology without ever showing symptoms during their lifetime. For this reason alone it is worth exploring factors that promote good cognitive health. The existing evidence from e pidem iological studies discussed in S tudy 1 indicates that weight loss, particularly among older adults, is associated with increased risk of cognitive decline and dementia. Obesity is well known to increase risk of cardiovascular disease, diabetes and can cer, some of the leading causes of death in the nation. The risk that weight loss interventions

PAGE 213

198 to protect against these diseases might inadvertently increase risk of dementia is troubling. It is possible that the weight loss observed in these observationa l studies does not play a direct causal role in the onset of cognitive decline and dementia, but this remains to be established. In either case, weight loss interventions in midlife could be a valuable approach to avoiding potential neurocognitive harm lat er in life, while concurrently reducing risk of other chronic diseases and improving quality of life. Consistent with a lifecourse approach to cognitive aging, S tudy 1 demonstrated empirical evidence from longitudinal studies linking obesity in midlife to greater risk of cognitive decline or dementia later in life. Significantly, the majority of studies that did not produce evidence for such an association were among older adults, a time in life when BMI is a particularly poor indicator of body composition and when early stages of undetected dementia may already be wreaking havoc on energy balance. While relatively few (e.g. Cournot et al. (2006a ) of these longitud inal studies reported cognitive function at midlife when baseline measures of weight were taken, Study 2, the NHANES III study, produced evidence that increased weight in midlife is associated with worse cognitive function in at least two functional areas : working memory and reaction time. Animal research and the evidence from human PA and CR /weight loss interventions (Study 3) indicates that interventions that affect weight or adiposity in midlife can have neuroprotective effects that correspond to func tional improvements in domains such as memory. Consistent with this study 4 ( the DRIFT study ) produced evidence of beneficial effects on both weight and memory function in young and middle aged obese adults. Taken together, the evidence from these studie s is consistent with what would be predicted by a cumulative model within a lifecourse approach to cognitive aging. A cumulative model would predict that the cumulative effects of even small exposures over time can

PAGE 214

199 contribute to increased disease risk, an d that removal of the exposure does not reverse the harm already done. Study 2 reported small but significant differences between the cognitive function of obese versus non obese persons in early and mid adulthood. Study 1 found existing epidemiological e vidence for a link between increased weight in midlife and increased risk of cognitive decline and dementia later in life. It is po ssible that exposure to obesity has cumulative effects on the br ain over many years. If true, one would expect neural damage and cognitive function to worsen with increased duration of obesity. The effects of duration of obesity have not been widely investigated, and attempts to incorporate this into Study 2 were hampered by loss of sample size and power. T his remains an area i n need of further investigation. 5.2 Central O besity M ay be M ore S trongly L inked to N eurocognitive Health than G lobal O besity. Plausible mechanisms make central obesity a likely culprit in any causal association between obesity and neurocognitive damage. Centr al obesity provides a number of potential mechanisms that could affect cognitive function in the short term or over many years. These include effects on glucose regulation and insulin resistance, inflammation, HPA axis regulation, leptin and BDNF. Thus cen tral obesity, rather than global obesity, may be the real concern for cognitive aging. However to date most research into the association between obesity and cognitive function has relied on BMI as the sole marker of adiposity. Study 1 demonstrated that th e majority of research to date has relied on BMI as the sole measure of adiposity. This may have contributed to the mixture of findings in this area, and should be addressed in future research by the additional assessment of body composition and/or distrib ution. Study 2, the NHANES III study, demonstrated that a measure as simple as WHR is sometimes more likely to show significant association with cognitive measures than BMI, and that the strength of the association can be greater. Categories of obesity me asured by BMI were

PAGE 215

200 not closely associated with percent fat mass or central obesity measured. In addition, the magnitude of the association of each measure of adiposity with cognitive function differed. Consistent with prior research where central obesity w as assessed, the magnitude of the relationship between WHR and attention/memory score was much greater than that found with BMI or percent fat mass. This is not consistent with a spurious finding. Similarly the DRIFT study demonstrated that central obes ity may be more closely linked to a number of mechanisms that potentially mediate effects of obesity on the brain, including leptin, BDNF and HPA axis function. Consistent with the hypothesis that obesity affects cognitive function, the NHANES III study s howed a small but significant relationship between increased global or central obesity and worse performance on a test of attention/memory. Though several other cross sectional studies have found no association, or even the reverse, they have typically bee n among older adults, who may already be experiencing the weight loss common in dementia. Dementia is typically diagnosed in persons aged >65 years. In the NHANES III study, increasing weight and worse cognitive performance were related in adults aged 20 5 9 years. The small magnitude of the association would be expected if this were in fact related to cognitive decline over many years. However causal inferences are beyond the scope of this cross sectional study. While it is possible that these 5.3 B ehavior M a y M oderate the Association Between O besity and Cognit ion It is possible that the association between obesity and cognitive function is merely an artifact of some other unmeasured factor. Most studies of obesity have controlled for education, age, socio eco nomic status, ethnicity, or gender, and many for diabetes or cardiovascular risk factors. Fewer have assessed the effect of health related behav iors known to affect obesity as either potential confounding factors or potential moderators of the association.

PAGE 216

201 However growing evidence indicates that physical activity and quality of diet could affect neurocognitive health. For this reason, study 2 investigated the potential moderating effects of quality of diet, physical activity, smoking status and social suppo rt on the association between obesity and adult cognitive function. Quality of diet, social support and smoking status showed independent associations with cognitive function in at least one test. However only physical activity showed evidence of an inter action between central obesity and cognitive function. Persons who were obese but active showed better cognitive performance that those who were obese and sedentary. This could be a composition effect, and much more research is needed to determine the effe cts of PA interventions on cognitive function. However some other intervention studies do find support for a causal role for PA on cognition. 5.4 Interventions that A ffect O besity M ay R educe R isk of C ognitive D ecline It is difficult to experimentally test the effects of increasing weight on humans. However the effects of weight loss may shed some light on the effects of obesity on cognitive function. Interventions that can affect obesity, either by reducing it, or by mediating changes in mechanisms that affect the brain, have potential utility in this regards. E vidence from animal studies demonstrates that calorie restriction leads to significant neuroprotective effects including preserved memory function with advancing aging, increased synaptic plasticity and neural health, and protection against excitotoxic insults or neurodegenerative disease. In humans the evidence is more mixed. Study 3 showed that, based on the current empirical literature, d ieting for weight loss can produce worse cognitive performanc e. H owever more of the larger, well designed studies appeared to support beneficial effects. It is important to understand the effects of weight loss on human cognition, not only because it may inform safety practices around weight loss interventions but also because it may provide insight into the

PAGE 217

202 direction of causation in the apparent relationship between obesity and cognition suggested in epidemiological studies. In the DRIFT study, 8 weeks of calorie restricted dieting produced significant weight loss (app rox. 6%) but did not produce changes in any cognitive domain at 8 weeks. Effects on memory were not apparent until 6 months later, and corresponded to decreased central adiposity and increased BDNF. T his finding is consistent with a direct causal relation ship between obesity and cognitive function in midlife for health adults. 5.5 IF Produces Beneficial Effects on M emory in O bese A dults Much more research is needed before solid conclusions can be drawn from the IF study regarding neuroprotective effects. At p resent the results do suggest that IF is well tolerated and safe for obese adults in midlife. However the effects on older adults, or adults with existing neurocognitive vulnerabilities remains to be tested. Animal models indicate that IF can have neuropro tective effects against neurodegenerative disease, neurotoxic insult or age related cognitive declines. However these effects have yet to be tested in humans. At least one study of caloric restriction in older adults with MC I supports a beneficial effect ( Krikorian et al., 2012 ) but the neuroendocrine effects of IF may be distinct and are as yet unknown. Furthermore it is not possible to extrapolate from the res ults of this 8 week intervention to the longer term effects that might occur with a public health intervention. With those caveats in mind, the IF dietary regimen used in study 4 holds good promise for older adults and clinical populations, such as person s with Type 2 diabe tes mellitus. IF proved to be safe, well tolerated and required only a relatively brief intervention (8 weeks) to produce longer term cognitive changes. This warrants further investigation, particularly among persons at high risk for cog nitive decline and dementia. Whether or not it eventually proves an acceptable public health intervention, this interesting paradigm of repeated mild metabolic

PAGE 218

203 stress, with some effects distinct from those seen in CR (such as increased neurogenesis, IGF 1, and potentially greater increases in BDNF), should provide an interesting and useful opportunity for better understanding the mechanisms that link human obesity to neurocognitive health outcomes. 5.6 Plausible Mediating Mechanisms Exist As previously discuss ed, numerous plausible mechanisms could mediate an effect of obesity on human neurocognitive health. While study 2 did not find evidence that IGF 1, glucose regulation or inflammation affected the association between central obesity and SDLT or SRTT, the r e have been many other studies that find evidence of mediating roles. In Study 4, the DRIFT study intervention the direction fo r change in insulin sensitivity, glucose insulin, BDNF and HPA axis activity were all consistent with the hypothesis of their be neficial effects on neurocognitive health. IF produced significant improvements in memory at 6 months follow up that were not apparent for the SDR control group. These changes were mediated by greater loss of trunk fat for the IF group that occurred after the end of the 8 w eek intervention. By contrast memory changes w ere not associated with change in global obesity (BMI, % fat mass). This finding is consistent with existing evidence that central obesity is particularly important for cognitive function. The reasons for the additional weight loss in the IF group are not known. It is possible that the IF group were more likely than the SDR group to continue some form of dietary restriction after the end of the intervention, though this was no t measured Follow up studies should investigate this possibility, as well as the possibility that the IF intervention had differential effects on neuroendocrine factors that influenced fat distribution. Much more work needs to be done to test the tenets of the theory of h ormesis ( the idea that repeated exposure to mild stress can have cumulative health benefits ) In particular the idea that repeated stress/stimulation leads to increased resilience against future stressors of the

PAGE 219

204 same or different type will require extensiv e testing. As noted previously, an extensive body of literature already documents many detrimental effects of repeated stress, and it would be insufficient to simply state that all of these stressors were above the threshold for detrimental effects. Clearl y the hypothesis needs more specification. However it is an interesting theory, and if adopted widely could result in a paradigm shift in many fields involving adaptive systems, including immunology, neuroscience and stress research. The results of the DR IFT study are consistent with the hypothesis that repeated mild metabolic stress can lead to greater improvements in memory function than weight loss alone. However this exploratory study, with its small sample size and missing data, can at best provide pr eliminary results to stimulate future research. In this study it was not possible to analyze data on pro inflammatory cytokines, psychological stress or mood. Furthermore, results of the sub set of data available on HPA axis function should be interpreted with caution given the small sample size. However the HPA axis measures showed trends for increased variability of the diurnal HPA axis rhythm, and increased capacity for response to a novel stressor in the IF group that are consistent with expectations th at IF would lead to improved HPA axis function and resilience against stress. Since GCs produced by the HPA axis, and GCs produced by central adipocytes, are known to have significant effects on the hippocampal neurons, learning and memory, the ability to improve HPA axis function and reduce central obesity after just 8 weeks of intervention would be valuable. This was the first test of the effects of IF on human cognition and the finding of a memory specific effect is consistent with research in animals ( Maalouf et al., 2009 ) Importantly, IF did not lead to the cognitive deficits sometimes found in Ramadan fasting ( Roky et al., 2003 ; Roky et al., 2004 ; Roky et al., 2000 ) Indeed IF did not have detrimental effects on cogniti ve performance though it was metabolically stressful throughout the study, as indicated by

PAGE 220

205 decreased serum glucose. Unfortunately enough data were missing from psychological questionnaires that it was not possible to determine whether the diet was also ps ychologically stressful, but it is reasonable to assume that fasting involved some stress for most participants. The DRIFT study has demonstrated the capacity of IF to improve memory and increase circulating BDNF in obese adults after 8 weeks of interve ntion and a delay of 6 months. The magnitude of the change in memory by 6 months was small, but important. 5.7 S trengths and Limitations Though suggestive of an association between obesity and cognitive function that can be affected by repeated exposure to mild metabolic stressors, the studies reported here had several limitations that prevent definitive conclusions. 5.7.1 Strengths and Limitations of Stud ies 1 and 3 : Systematic Reviews The systematic literature reviews conducted on the association between obesit y and cognitive decline and dementia (Study 1) and the effects of weight loss interventions on adult neurocognitive health (Study 3) provided an excellent overview of the existing empirical literature in these areas. However they were limited to the MEDLIN E and PsychINFO databases, and as such do not contain all of the available literature on these topics. Furthermore, the value of these reviews is limited to the quality of the studies that are available. Most of these studies controlled for potential confo unding factors such as education, ethnicity, age, and gender, and many controlled for vascular and metabolic factors, other potential confounding factors such as obstructive sleep apnea were not addressed. This study aimed to encompass all of the availabl e literature, and so deliberately did not select for the studies with the best quality designs and execution, which naturally have a significant effect on the validity of their results. A next step in this review process will be to review other databases (Cinahl, Embase, Google Scholar) and

PAGE 221

206 apply quality ratings to all the available studies. This will give a better estimation of the strength of the available evidence, where the current review focuses principally on the quantity. 5.7.2 Strengths and Limitations of Study 2 : The NHANES III S tudy The cross sectional design of the NHANES III study prohibited causal inferences or extensive statistical modeling of the association between obesity and cognitive function. While the NHANES III study is the only NHANES stu dy to date to include cognitive outcome measures, the measures used were limited. The SDLT, for example, ranged from 0 16, and had obvious floor and ceiling effects that may have contributed to the small amount of variance explained by regression models. Similar low estimations of variance explained have been reported by other researchers using the same dataset for other studies ( Pavlik et al., 2004 ; Suhr et al., 2004 ) A more full neuropsychological test battery would provide significantly more information. Though useful in this exploratory study, many other variables available in the NHANES III st udy were far from the ideal. For example bioelectric impedance analysis (BIA) is not the gold standard in measurement of body composition, though our calculated values for fat mass were correlated with WHR and BMI. In addition s elf report is not the most r eliable measure of physical activity and follow up studies would benefit from objective measurement by actigraphs. Furthermore, some potentially relevant mediating, moderating or confounding factors were not measured as part of the NHANES III study. While it would have been good to test stress and HPA axis related hypotheses in the general population, the NHANES III study was not designed to assess psychological or physiological stress. No measures of perceived stress or glucocorticoid secretion were avail able. Similarly, measures of some potential confounding factors, such as obstructive sleep apnea, were not included. Given the age of the NHANES III dataset, with data collected between 1988 94, i t is also possible that societal factors may have changed, and that results would be different if collected

PAGE 222

207 today. For example familiarity and comfort with computer use would not have been as widespread in 1988 94 as it is today. Lack of familiarity with computers could have caused some people to perform more poo rly, confounding the results. Rates of obesity in the population are now higher than they were in 1988 1994. If the association between weight and cognitive function reported here was an epiphenomenon of other differences in the population that were also related to obesity, then more widespread obesity would tend to attenuate the strength of the association. Comparison of our results with data collected more recently would provide useful insights into these possibilities. The most recent NHANES study has a pparently included some cognitive measures, so further exploration of this question may soon become a possibility. In spite of these limitations the availability of behavioral and anthropometric and cognitive measures together in a large nationally repres entative sample made it possible to assess the hypotheses. While other studies have shown an association between adiposity and cognitive decline or dementia, to our knowledge no other studies have concurrently assessed the potential for behavioral factors to moderate this association. The sample size of the NHANES III study was a real strength, allowing the power to control for multiple covariates or confounding factors and making regression model building possible. 5.7.3 Strengths and Limitations of Study 4 : Th e DRIFT S tudy This exploratory R21 study was originally powered to detect between group differences in weight loss. While the sample size is appropriate for this purpose it did not provide the power needed for statistical tests of change in many of the se condary outcome variables, such as insulin sensitivity. The small sample size also prohibited regression model building or structural equation modeling that could have provided an understanding of the causal relationships between observed changes. A furthe r loss of data through internal study errors

PAGE 223

208 such as incomplete tests, missing samples, study attrition at the 6 month follow up, and the inability to retrieve s ome data further compounded these issue s The potential to explain the differences seen at 6 m onth follow up was limited by the absence of any measures of participant behavior in the intervening 6 month period. It is therefore impossible to determine whether some participants continued the IF dietary regimen beyond the study period, whether some pa rticipants engaged in more physical activity and so forth. In addition to these variables, f uture investigations of IF should include measures of physical activity, sleep patterns, self image and self confidence, self efficacy for behavior change, and sys tematic collection of feedback on the lived experience of being on the study diet Qualitative information on factors that influenced adherence to the study die t would strengthen future research and practical application. Furthermore, the potential to t ranslate the results of this study into community practice is limited. For safety reasons this R21 pilot study employed highly restrictive exclusion criteria that may have led to a sample that was not representative of the general population. In addition, much of the recruiting was done on an academic university campus, making it likely that the sample was more educated than the general population. The invasive and intensive nature of the study also made selection bias likely. In addition, IF has limited f ace validity or ecological validity as a sustainable intervention for prevention of cognitive decline. In the general population adherence to alternate day fasting principles may be difficult. Eating is a highly social practice and carries many cultural an d emotional meanings. Rather than using this study to promote widespread adoption of IF for cognitive health this study can be used as an unique opportunity to investigate the distinctive mechanisms that may underpin the effects of IF. The idea that repea ted exposure to stress can have cumulative

PAGE 224

209 beneficial effects is not widely accepted, and has not been widely tested. The DRIFT study provides evidence that should encourage further investigation, as IF may provide clues to mechanisms that promote neuroc ognitive health. A better understanding of these mechanisms may help to lay the foundation for future interventions to promote neurocognitive health. In spite of its limitations, the DRIFT study successfully implemented a rigorously controlled interventio n protocol that employed optimal measures of the outcome variables of interest. The use of DXA scans for body composition, FSIVGTT for insulin sensitivity, and the 90% V02max stress test are good examples of this. Furthermore a large number of useful varia bles were measured, making this a rich source of information on the effects of IF on the health of obese adults. 5.8 N ovel C ontributions To date this is the first published study of IF to include a thorough investigation of effects of IF on HPA axis function in humans. As previously described, one other study of IF in humans measured cortisol at a single time of day, but did not control for different time of day for the two study groups, thus confounding the results ( Stote et al., 2007 ) Given the trends for improved HPA axis observed in this study, the DRIFT study can provide important preliminary data to inform the development of larger study of these effects. This is also the first study to assess the effects of IF on human cognitive performance. The cognitive test battery used was designed for repeated measures, had good reliability and validity, and was practical for implementation in the study visits. Hence this exploratory study is uniquely placed to investigate the theoretical effects of this weight loss diet on human cognition and potential hormetic mechanisms. Given the evidence of significant neuroprotective effects of IF from animal research, investigation o f the effects of IF on human cognitive function may be important to our understanding of cognitive aging.

PAGE 225

210 Furthermore Study 2, using the NHANES III data provides a novel assessment of the association between obesity and cognitive function in the general population. Unlike prior studies showing an association between obesity and cognitive decline or dementia late in life, the NHANES study provided evidence of an association in early and mid adulthood, and assessed potential for health behaviors to moderate the association. 5.9 F uture Directions In spite of the wealth of information on the systemic effects of obesity on health, relatively little is known about the effects of obesity on brain health or cognitive function. The present exploratory study provides further evidence of a link between obesity and cognitive function, and highlights the particular importance of including measures of central obesity in future studies. Future epidemiological research in this area should also include repeated measures analy sis with rigorous assessment of cognitive function baseline. Weight loss intervention studies with larg er and more diverse samples will also be an important source of information. Assessment of the impact of repeated mild stress on the association between obesity and cognitive function was a particularly novel contribution of the present study. The idea that repeated stress can be beneficial rather than detrimental requires much more investigation. D irectly comparing the effects of different stressors woul d be a useful sta rt to a better understanding whether a common mechanism does indeed underpin effects of repeated stress on health. Intervention over a longer duration may help to clarify whether it was the intervention or post intervention behaviors that contributed to 6 month between group differences in the DRIFT study. The potential for IF to improve cognitive function in persons with diabetes and/or MCI would also be a logical next step. These studies should be sure to include measures of physical ac tivity, sleep, social interactions at meals, and psychological stress

PAGE 226

211 to assess the impact of the intervention on these factors, and their potential role in outcomes. Finally, future studies should investigate the frequency and duration of fasting required for an effect since alternate day fasting is unlikely to become a widespread public health practice. 5.10 S ummary In conclusion, the findings from the present study provide evidence that increased weight is re lated to poorer performance on an attention/learning task in a large, nationally representative sample of adults aged 20 59 years. Consistent with prior research, the finding was particularly strong for central obesity, highlighting the importance of inclu ding measures of central obesity in future research. As predicted, the association between central obesity and attention/memory scores was moderated by frequency of physical activity, but not other health related behaviors. Regular physical activity and in termittent fasting can both be considered examples of repeated stress, or hormesis, and emerging evidence suggests that repeated mild stress can have beneficial effects on allostatic load and cognitive aging. Consistent with this hypothesis, intermittent f asting produced significant improvements in memory, BDNF and trunk fat of obese adults, as well as trends for improvement in HPA axis function. These findings provide an important extension of the existing literature on hormesis and the effects of stress o n cognitive health. Further comprehensive testing of these effects in early and mid adulthood may provide important insights into the cognitive deficits that can predate the development of dementia.

PAGE 227

212 APPENDIX This report serves as an opportunity to reflec t on the clinical observations made and the experience that I gained through my involvement in the Colorado Clinical and Translational Science Institute (CCTSI) Pre doctoral training program at the University of Colorado Denver. This training program provi ded the opportunity for students from non clinical doctoral training programs to gain skills and insights that would equip them to better translate research in their field into real world practice. Didactic learning, seminars and discussions provided only part of the learning experience the most valuable aspect was the opportunity for direct exposure to the realities of patient experiences, patient care and clinical practice. Clinical placements were research. With my interests in promoting observations of memory clinics and neuropsychological testing services were a logical choice of placement to complement my con current didactic learning and the research accomplished and described above. Furthermore, since my work focused on the potential that overweight and obesity may play a causal role in increasing risk of age related cognitive decline or dementia, exposure to weight management services was also a logical choice. As a result I completed three distinct clinical placements through the CCTSI training program, in addition to extra curricular n relevant clinical trials at the University of Colorado Denver. Placements were at: 1) the Weight Management Clinic at Kaiser Permanente Colorado, 2) the Clinical Neuropsychology unit at National Jewish Health Colorado, and 3) the newly established Memory Clinic at Kaiser Permanente Colorado. These placements were supervised and/or arranged by my clinical mentor, Dr. William Donahoo, M.D.

PAGE 228

213 Since I learned much that was practical, useful and interesting from these experiences the focus of this report gives an overview of just some of what I learned, and a reflection of how this has shaped my understanding of the field, and current and future research. Clinical Placement 1: Kaiser Permanente Weight Management Clinic Supervised by endocrinologist Dr. William Donahoo, M.D., my placement at the Kaiser Permanente weight management involved accompanying Dr. Donahoo in the weight management clinic to a) group visits and b) individual patient consultations for weight management. All patients were insured by Kaiser Permanente Colorado Weight management groups At the time of my clinical placement, the group visits were a very new addition to the weight management services offered locally by KP Colorado. As explained by Dr. Donahoo, the benefit of group visits for pati ents was that they involved a lower copay and shorter waiting period than individual visits. In addition, the rationale for group visits may have been drawn from research such as that describing how behavior modification group visits and medication use tog ether were the most effective treatment for weight loss. The weight management group visits at this clinic in KP Colorado were made up of patients seeking information and options for weight loss, particularly those seeking medications to assist weight los s. Group participants did not know one another, and the need for confidentiality was described before the start of the group. They were not intended as therapeutic groups or support groups. At the time of my placement the groups were offered as a single se ssion, chiefly providing information about different weight management options.

PAGE 229

214 There was an expectation among staff and providers that they would introduce group follow up visits in the future. The group sessions were run by a nurse and the endocrinolog ist, Dr. Donahoo. As part of the visit, all patients received confidential weight and blood pressure check conducted individually, that was included in their medical records, and through this and other individual conversations the nurse did a lot to quickl y build rapport with the patients. The remainder of the group session was chiefly comprised of information about different weight loss options, and the opportunity to ask questions and discuss the options together as a group. Patients were also offered the opportunity to ask the doctor questions specific to their own medical needs, at the end of the group in a confidential setting. Surprisingly, only a few patients tended to take up this offer. I learned much from these groups. First, I found it informativ e to listen and learn about the different weight management options that are available. Prior to this I was not familiar with any of the medications that can be used in weight loss efforts. I became aware of a number of weight loss drugs, including Orlista t (Xenical), which reduces fat absorption by acting on an enzyme responsible for cleaving fat. Though it can be effective in producing 5 7% weight loss in 3 6 months for some people, there are medical contraindications and side effects that may limit its w idespread use. By contrast, Sibutramine (Meridia), has serotonergic action and was originally developed to treat depression, but was found to act on brain regions responsible for hunger/satiety. It can have additive, or interactive effects with a number of other drugs, such as SSRIs, pseudoephedrine, or the over the had interested most patients enough to attend the group session, even though the cost of the nce. The apparent widespread interest in this

PAGE 230

215 drug was itself interesting. Where were patients learning of this option? What was driving the demand for this medication? Clearly many patients in the groups found medications more appealing than behavior cha nge options for weight loss. Behavior change options for weight loss were also discussed in the group visit. The options discussed were those available and insured through KP Colorado. These included 1) a free online program to develop individually tai lored weight loss strategies, 2) a phone support program that provided 1:1 customized weight loss programs, 3) a free CD to help people identify unhealthful foods in their kitchens, 4) dietician visits for persons with additional medical issues beyond over weight or obesity, such as diabetes, 5) discounts at Weight Watchers, and 6) Optifast, a medically supervised fasting program for persons with BMI >30. The value of group dynamics in this setting became apparent in a number of ways. As already described, these were not therapeutic groups. However patients were able to share and discuss some of their own experiences with different weight loss options. For example during the discussion of Orlistat, the group was asked if anyone had personal experience using this drug. A few patients raised their hands, and reported significant issues with oily stools. Even more interesting was the discussion between two patients who described their observations of the extreme and enduring physical and psychiatric/behavioral e ffects of the drug Phentermine on friends who had used it. This drug is not supported by the KP physicians or clinic, and so was not included in the information provided to all patients. Individual weight loss consultations

PAGE 231

216 After the morning group visits I attended, I was able to observe individual patient appointments in the weight management clinic. These provided useful insights into the running of clinical weight loss practices, as well as the kinds of issues faced by patients who are obese. From a pr actice standpoint, each patient was weighed by a nurse before seeing the doctor. The physician had access to the full medical record for each patient, allowing him to quickly access any relevant medical history that may be affecting weight. The consultatio ns were typically fairly structured, and the types of questions and information that patients sought tended to be similar. For example patients were likely to be aware of the benefits of weight loss by physical activity and dietary restriction, and instead wanted information about medications or bariatric surgery. This was accommodated, with information on the latter readily displayed using pre prepared information boards. Each patient left with a letter outlining the things discussed and their various opti ons. They were also usually also told they could contact the physician with questions and for the results of any tests that were ordered. These provided fairly systematic approaches to addressing weight management in patients. I also learned that candidate s for bariatric surgery were encouraged to try other options first, and were carefully vetted and counseled before surgery was indicated. Some of the issues faced by patients, and their barriers to weight loss, can be illustrated by some of the following cases. Patient case study #1 likely reflecting my own personal expectations and norms for weight of a man his age. This itself was an interesting surprise, as were the various other patien ts who attended the clinic who did not appear overweight to my eye. Patient 1 reported struggling with his weight for 15 20 years

PAGE 232

217 without much success in weight loss. He was clearly concerned about his current weight. As a smoker, he reported being ready t o quit, but was concerned about the potential for weight gain if he did so. Interestingly, while he arrived at the clinic officially asking for a prescription for the drug Meridia, further discussion revealed that he was really interested in bariatric surg ery. Both he and his wife had applied. His wife had been accepted, but he had been denied. Altogether this patient provided interesting insights into the variety of motivations that can drive motivation for choosing or avoiding different weight loss option s. Patient case study #2 was a multiple significant health concerns, including cardiovascular problems, Type 2 diabetes mellitus, thyroid problems and back problems, and used oxygen supplied by a nasal tube. Despite this she was still working as a nurse. She reported being quite thin until after having her third child, when she gained quite a lot of weight, then later gained further weight when she injured her back. The back problems continue to ma ke it difficult to walk or exercise, and as she has gained weight physical exertion has become increasingly difficult. With adult children and grandchildren now living with her she continues to find many barriers to self care. Again, this patient provided many insights into the barriers to a healthy weight that some people face. Clinical Placement 2: National Jewish Health Adult Neuropsychology Clinical Service The Adult Neuropsychology Clinical Service at National Jewish Health provides neuropsychologica l evaluation for many different conditions. The results of the neuropsychological tests are used to complement information from medical history, laboratory work and imaging such as MRI or CT scans. My placement at the clinical neuropsychology unit, which w as overseen by Dr. E. Kozora, a highly experienced a clinical neuropsychologist, provided

PAGE 233

218 very useful insights into the nature of truly comprehensive neuropsychological assessment for potential dementia. I attended a number of all day testing sessions desi gned to assess patients about the flow of events in conducting these testing sessions, the personnel and training required, the nature of the tests themselves, an d the way that [this sample of] patients responded to taking the test battery. Patients were usually referred by another healthcare provider. The neuropsychological testing was usually covered by insurance, though some patients could choose to pay out of pocket if they chose to do so. The testing was collaboration between the clinical neuropsychologist responsible, and a specially trained technician, usually with a bachelors in psychology or more. The file of tests and scoring sheets was compiled by the te chnician in advance at the beginning of the day. Patients were scheduled for a very long day of testing, often 9am 4pm, with a 45 minute break. During this time they saw the clinical neuropsychologist, who would take a history, and completed a variety of paper and pencil neuropsychological tests administered by the technician. The results of the tests were reviewed by the clinical neuropsychologist, who also prepared an in depth report. Patients were encouraged to bring a friend or family member, at their discretion. These were particularly helpful in the assessment of dementia, as the patient themselves may not be able to recall much of their own history, or may not have much insight into the changes that had occurred in their daily function. The neurops ychological test battery used during the assessment of dementia included a range of widely accepted tests. Tests included Trails A and B, the Dementia Rating Scale II, the Brief Visuo spatial test revised, a test of logical memory, the Boston naming test, and the

PAGE 234

219 Wechsler Abbreviated Scale of Intelligence. As a result, across the visit patients could expect to do writing, drawing and answer some questions verbally. Interestingly, the day also included a test of effort, to gauge the extent to which patients were trying their best. I was interested to that the testing day, though long, was conducted in a gentle, low stress manner. Technicians were attentive to how tired a person was becoming as the day progressed, and responded accordingly. Patient case study #3 provides a snapshot of the testing exper iences I observed. An 86 year husband and medical provider had raised concerns about her memory and function, and she was happy to go along with the assessments b y the doctor/s and neuropsychologist. She was articulate, well dressed, had good comprehension of requests and directions. Her hearing and eyesight were good. Her education comprised of high school education followed by a few short courses over the years. both in good health. Her mother had AD for about 10yrs before death. Both the patient and her husband were lean and physically active on a regular basis for many years. The patient had be en in a serious car accident 50 years ago, after which she lost consciousness for 10days. She had amnesia for the event but reports returning to full function with no personality change or other notable changes. This patient was unclear on the reasons for attending the testing session, but was very happy to go through with the testing, and appeared to try her best at the tests. She got lost trying to return from the bathroom at a bathroom break. During the testing she struggled with Trails B, and many of t he tasks on the Dementia Rating Scale, including differences,

PAGE 235

220 similarities and verbal recall. In the Wechsler Abbreviated Scale of Intelligence she had some difficulty with block design, and some others. Overall, the patient showed a trend for reduced fun ction in her activities of daily living, and her performance on many of the cognitive tests was below what would be expected for someone of her education and age. Her pattern of performance was consistent with the possible effects of obesity and lifestyle factors on the risk of dementia it was worth noting that this patient had been physically active for most of her life, and had never been overweight or obese, nor did she have significant health concerns beyond a history of a traumatic brain injury 50 years ago. As I continue research into the effects of obesity and lifestyle factors on healthy cognitive aging it is worth remembering that these factors are neither necessary, nor likely to be sufficient, to cause In summary, my clinical observations during this placement provided many insights into noting that these assessments form part of the d iagnostic process for patients who receive them. Furthermore, many patients never receive such formal and rigorous testing, and their clinical diagnosis rests on their history and clinical judgment. Clinical Placement #3: Kaiser Permanente Memory Clinic The Memory Clinic at Kaiser Permanente (KP) Colorado was newly established only weeks before I began my clinical placement there. The team working in this clinic included specially trained nurses, general practitioners with special interest and training in geriatrics, social workers, administrators, and geriatric psychiatrists. The clinic rotated around different KP

PAGE 236

221 sites in order to maximize accessibility for patients hence it did not have its own dedicated clinic space, but rather used available free sp ace within various clinic locations. The clinic provided a particularly interesting opportunity to learn due to the manner in which it was integrated into the overall care provided through the HMO. Patients were referred from internal KP physicians, and the referral and medical records were accessible to the Memory Clinic team in advance of the appointments they had within the clinic. The Memory Clinic team truly worked as a team meeting together at the beginning of the day to discuss the cases for the day. This ensured that the wisdom of more experienced and highly trained members of the team was shared efficiently, while ensuring that the clinical judgments and actions of those more new to the field were appropriately supervised. The clinic was stru ctured to provide a sequence of at least two visits for all patients who were referred an initial assessment and then a follow up for discussion of results and diagnoses. Ongoing care could also be provided to those how needed it. The initial assessment visit included collecting medical history, a physical examination that covered relevant neurological aspects, and a sequence of neuropsychological tests. The latter included the Dementia Rating Scale II and Trails B, among other possible tests. With patien t permission, an accompanying family member or friend could provide additional information during the medical history and also in a separate interview with a social worker. Where history or medical examination suggested a need for additional medical tests (such as an MRI or blood tests) these were also ordered through the KP system and results were available to the team before the next Memory Clinic visit. Another advantage of the integrated health records was the ability to view patient medications that ha d the potential to affect memory or cognitive performance, and suggest changes before further testing was completed. The presence of prescribing physicians

PAGE 237

222 ensured that those patients who needed medication changes, or perhaps had medical conditions which r equired the introduction of new medications to help stabilize the health of a patient, could be treated immediately, with direct communication with the regular primary care physician. I was impressed by the efficiency and integration of these services. The diagnostic process appeared to provide a thorough, whole health, perspective. Diagnosis was made by the team at the team meeting. Patient follow up visits were made once all of the relevant tests were assessed and the medical history reviewed as a whole Where necessary, diagnoses were given to a patient in a sensitive manner by the person who had seen them, and who had coordinated their overall assessment. At the same visit, a consultation with a social worker was also included for those patients who we re given a diagnosis of dementia. During this visit they were offered resources for understanding and coping with dementia, and referrals to other services. The time spent with the social workers seemed a highly valuable component of the services provided. It was pleasing to see that patients were not simply left with a diagnosis and uncertainty about what to do next. They were left with practical help, ongoing contacts, and the expectation of continued holistic care. During my placement I was able to obs erve many different consultations, both assessment, giving of diagnoses, and also follow up visits for those who had previously received interesting to note the seemingly large proportion of patients attending the clinic who had diabetes. This was consistent with what my literature review had shown regarding the increased risk of cognitive impairment with Type 2 diabetes mellitus can. I therefore asked a physician at the c linic about their subjective impression of the proportion of patients referred for memory

PAGE 238

223 problems who had diabetes. He thought that most patients had diabetes, certainly more than 50%. Anecdotally I also noted that a lot of patients were overweight, thoug h this should come as no surprise given the high rates of overweight and obesity in the general population. Similarly, relatively few patients were physically active on a regular basis, but this too is quite common in the general population, and particular ly among older adults. I found it very interesting to see how many different patients with subjective memory complaints presented for clinical assessment. This varied widely. Some clearly presented with very limited function and comprehension of their surrounds, while others appeared quite functional, but testing revealed significant deficits. Many patients had very little insight into the problems reported by their caregiver, spouse or friend who accompanied them. Since this was an insured population, I expect that the patients attending the clinic were more likely to be at a higher level of function than the general community, consistent with subjective observations I ine, where I also volunteered for more than a year. That being the case, the difficulties faced by persons experiencing significant memory decline and dementia should not be underestimated. People who are unaware of their problems, or even resistant to the idea that they have problems, are likely to be underserved unless represented by a caring and capable family memory or friend. Furthermore their care options are constrained by their resources and the services available in their community. Where they can be supported by an integrated healthcare system, reversible causes of cognitive deficits such as some medications not appropriate for older adults can be detected and deal with to maximize healthy function. Where dementia diagnosis is clear and gradual decline is inevitable, comprehensive medical and social support make a significant

PAGE 239

224 In summary my time at the KP memory clinic provided very interesting and very informative insights into the background and liv ed experience of a wide variety of people with memory problems. It also provided an opportunity to learn about the full diagnostic process for dementia and similar disorders that has proven invaluable as a scholar and as a professional. Summary of clinica l observations and relevance to research Clinical placement at the KP weight management clinic provided the opportunity to learn more about the challenges to managing a healthy weight faced by persons who are overweight and obese. It was also a chance to learn about the different treatment options for weight management. With my training in psychology, public health and behavioral science my prior exposure had principally been to behavioral lifestyle change options. It was therefore useful to learn about ad ditional options such as medications, bariatric surgery and specialized fasting or weight management programs. My dissertation research focuses on the effects of overweight and obesity on cognitive function in adults. It was clear that there are many fact ors related to obesity that could contribute to cognitive issues. Depression, social isolation, pain from past injuries or poor health and functional limitations caused by obesity itself could contribute to the cognitive deficits that have recently reporte d in persons who are overweight or obese. The extent to which these factors are causally related to risk of dementia should also be investigated. Placement at the clinical neuropsychology unit at National Jewish Health provided quite a different chance t o learn about the intricacies of rigorous cognitive testing. This testing by highly trained professionals may be the gold standard of cognitive testing, and when combined with other sources of information such as medical history, laboratory tests and medic al imaging,

PAGE 240

225 can provide a thorough assessment and diagnostic procedure. However many patients do not have access to such services, and many do not have the health or stamina to undergo a full day of testing. This clinical placement therefore provided a val uable opportunity to inform my the tests used, functions measured, and providing an excellent opportunity to talk with a number of people with specialized expertis e in neuropsychological assessment. My dissertation research builds on an understanding of a need to create opportunities for earlier diagnosis and treatment for cognitive decline. Will this involve more sensitive neuropsychological tests, novel biomarker s or careful imaging techniques? Perhaps all of the above are needed to describe the function and underlying pathology accurately. In any case, a thorough understanding of cognitive and functional assessment is vital to pursing research in this area. Witho ut it research in this field will not be valid or reliable. Finally, my experience at the KP Memory Clinic provided many valuable insights into the ways that people present with memory problems, as well as the background and history that may accompany co gnitive decline. The goal of both my dissertation research and my ongoing research in this field is that my work will have real world applications to the clinic and community. Thus learning more about the lives of community dwelling adults experiencing th ese problems in daily life was invaluable. Furthermore, clinical applications are made difficult without knowledge of what clinical practice really looks like, and who practices it. Of course clinical practice varies from place to place, but it was invalu able to observe an integrated, holistic approach to assessment of cognitive decline that was efficient, professional and caring. My subjective observations of high rates of obesity, diabetes and related issues consistent with the literature I encountered during my dissertation research and the findings of my own studies,

PAGE 241

226 thus making the problems seem more real and less theoretical. A better understanding of the realities of clinical practice shall inform my future research will seek to provide practical, r ealistic interventions to reduce risk of cognitive decline and dementia, and primary, secondary and tertiary prevention efforts to promote healthy cognitive aging. Taken together, these clinical observation opportunities have been highly valuable for inf orming the depth of my dissertation research, as well as laying the groundwork for further translational research in this area.

PAGE 242

227 REFERENCES Abbasi, F., Brown, B. W., Lamendola, C., McLaughlin, T., & Reaven, G. M. (2002). Relationship between obesity, insulin resistance, and coronary heart disease risk. Journal of the American College of Cardiology, 40 (5), 937 943. Abbott, R. D., White, L. R., Ross, G. W., Masaki, K. H., Curb, J. D., & Petrovitch, H. (2004). Walking and dementia in phy sically capable elderly men. JAMA, 292 (12), 1447 1453. doi: 10.1001/jama.292.12.1447 Abellan van Kan, G., Rolland, Y., Gillette Guyonnet, S., Gardette, V., Annweiler, C., Beauchet, O., . Vellas, B. (2012). Gait speed, body composition, and dementia. Th e EPIDOS Toulouse cohort. Journals of Gerontology Series A Biological Sciences & Medical Sciences, 67 (4), 425 432. doi: 10.1093/gerona/glr177 Aberg, E., Perlmann, T., Olson, L., & Brene, S. (2008). Running increases neurogenesis without retinoic acid recep tor activation in the adult mouse dentate gyrus. Hippocampus, 18 (8), 785 792. Adlard, P. A., Perreau, V. M., Pop, V., & Cotman, C. W. (2005). Voluntary exercise decreases amyloid load in a transgenic model of Alzheimer's disease. Journal of Neuroscience, 25 (17), 4217 4221. Ahima, R. S., Prabakaran, D., Mantzoros, C., Qu, D., Lowell, B., Maratos Flier, E., & Flier, J. S. (1996). Role of leptin in the neuroendocrine response to fasting. Nature, 382 (6588), 250 252. doi: 10.1038/382250a0 Ahmed, S. B., Fisher, N. D., Stevanovic, R., & Hollenberg, N. K. (2005). Body mass index and angiotensin dependent control of the renal circulation in healthy humans. Hypertension, 46 (6), 1316 1320. Al Hazzouri, A. Z., Haan, M. N., Whitmer, R. A., Yaffe, K., & Neuhaus, J. (20 12). Central obesity, leptin and cognitive decline: The Sacramento area Latino study on aging. Dementia and Geriatric Cognitive Disorders, 33 (6), 400 409. doi: 10.1159/000339957 Alaei, H., Moloudi, R., Sarkaki, A. R., Azizi Malekabadi, H., & Hanninen, O. ( 2007). Daily running promotes spatial learning and memory in rats. Pathophysiology, 14 (2), 105 108. Albanese, E., Hardy, R., Wills, A., Kuh, D., Guralnik, J., & Richards, M. (2012). No association between gain in body mass index across the life course and midlife cognitive function and cognitive reserve -the 1946 British Birth Cohort study. Alzheimers Dement, 8 (6), 470 482. doi: 10.1016/j.jalz.2011.09.228 Albert, M. S., DeKosky, S. T., Dickson, D., Dubois, B., Feldman, H. H., Fox, N. C., . Phelps, C. H (2011). The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7 (3), 270 279. doi : 10.1016/j.jalz.2011.03.008

PAGE 243

228 Aleman, A., de Vries, W. R., de Haan, E. H., Verhaar, H. J., Samson, M. M., & Koppeschaar, H. P. (2000). Age sensitive cognitive function, growth hormone and insulin like growth factor 1 plasma levels in healthy older men. Neur opsychobiology, 41 (2), 73 78. doi: 26636 Alonso, M., Vianna, M. R., Depino, A. M., Mello e Souza, T., Pereira, P., Szapiro, G., . Medina, J. H. (2002). BDNF triggered events in the rat hippocampus are required for both short and long term memory forma tion. Hippocampus, 12 (4), 551 560. doi: 10.1002/hipo.10035 lvarez, A., Cacabelos, R., Sanpedro, C., Garca Fantini, M., & Aleixandre, M. (2007). Serum TNF alpha levels are increased and correlate negatively with free IGF I in Alzheimer disease. Neurobiolo gy of Aging, 28 (4), 533 536. Alvarez, X. A., Sampedro, C., Cacabelos, R., Linares, C., Aleixandre, M., Garcia Fantini, M., & Moessler, H. (2009a). Reduced TNF alpha and increased IGF I levels in the serum of Alzheimer's disease patients treated with the n eurotrophic agent cerebrolysin. International Journal of Neuropsychopharmacology, 12 (7), 867 872. doi: 10.1017/s1461145709990101 Fantini, M., & Moessler, H. (2009b). Reduced T NF a and increased IGF I levels in the serum of Alzheimers Dement, 5 (3), 234 270. doi: 10.1016/j.ja lz.2009.03.001 Alzheimers Dement, 8 (2), 131 168. doi: 10.1016/j.jalz.2012.02.001 Ancoli Israel, S., Klauber, M. R., Butters, N., Parker, L., & Kripke, D. (1991). Dementia in insti tutionalized elderly: relation to sleep apnea. Journal of the American Geriatrics Society, 39 (3), 258. Anderson, L. A., & McConnell, S. R. (2007). The healthy brain and our aging population: translating science to public health practice. Alzheimers Dement 3 (2 Suppl), S1 2. doi: 10.1016/j.jalz.2007.01.017 Angelini, A., Bendini, C., Neviani, F., Bergamini, L., Manni, B., Trenti, T., . Neri, M. (2009). Insulin like growth factor 1 (IGF 1): relation with cognitive functioning and neuroimaging marker of br ain damage in a sample of hypertensive elderly subjects. Archives of Gerontology and Geriatrics, 49 Suppl 1 5 12. doi: 10.1016/j.archger.2009.09.006 Anson, R. M., Guo, Z., de Cabo, R., Iyun, T., Rios, M., Hagepanos, A., . Mattson, M. P. (2003a). Inter mittent fasting dissociates beneficial effects of dietary restriction on glucose metabolism and neuronal resistance to injury from calorie intake. Proceedings of the National Academy of Sciences, 100 (10), 6216 6220.

PAGE 244

229 Anson, R. M., Guo, Z., de Cabo, R., Iyu n, T., Rios, M., Hagepanos, A., . Mattson, M. P. (2003b). Intermittent fasting dissociates beneficial effects of dietary restriction on glucose metabolism and neuronal resistance to injury from calorie intake. Proceedings of the National Academy of Sci ences of the United States of America, 100 (10), 6216 6220. doi: 10.1073/pnas.1035720100 Anstey, K. J., Cherbuin, N., Budge, M., & Young, J. (2011). Body mass index in midlife and late life as a risk factor for dementia: a meta analysis of prospective studi es. Obesity Reviews, 12 (5), e426 437. doi: 10.1111/j.1467 789X.2010.00825.x Anstey, K. J., Lipnicki, D. M., & Low, L. F. (2008). Cholesterol as a risk factor for dementia and cognitive decline: a systematic review of prospective studies with meta analysis. American Journal of Geriatric Psych, 16 (5), 343 354. APA. (2000). Diagnostic and statistical manual of mental disorders (4th ed.). Washington. Appelhans, B. M., Whited, M. C., Schneider, K. L., Ma, Y., Oleski, J. L., Merriam, P. A., . Ockene, I. S. ( 2012). Depression severity, diet quality, and physical activity in women with obesity and depression. Journal of the Academy of Nutrition and Dietetics, 112 (5), 693 698. Arntzen, K., Schirmer, H., Wilsgaard, T., & Mathiesen, E. (2011). Impact of cardiovas cular risk factors on cognitive function: The Tromso study. European Journal of Neurology, 18 (5), 737 743. doi: 10.1111/j.1468 1331.2010.03263.x Arvanitakis, Z., Schneider, J., Wilson, R., Bienias, J., Kelly, J., Evans, D., & Bennett, D. (2008). Statins, i ncident Alzheimer disease, change in cognitive function, and neuropathology. Neurology, 70 (19 Part 2), 1795 1802. Atti, A. R., Palmer, K., Volpato, S., Winblad, B., De Ronchi, D., & Fratiglioni, L. (2008). Late life body mass index and dementia incidence: Nine year follow up data from the Kungsholmen Project. Journal of the American Geriatrics Society, 56 (1), 111 116. doi: 10.1111/j.1532 5415.2007.01458.x Azab, M., Khabour, O. F., Al Omari, L., Alzubi, M. A., & Alzoubi, K. (2009). Effect of every other day fasting on spontaneous chromosomal damage in rat's bone marrow cells. Journal of Toxicology and Environmental Health, Part A, 72 (5), 295 300. Bagger, Y. Z., Tanko, L. B., Alexandersen, P., Qin, G., & Christiansen, C. (2004). The implications of body fat mass and fat distribution for cognitive function in elderly women. Obesity Research, 12 (9), 1519 1526. Baker, L. D., Frank, L. L., Foster Schubert, K., Green, P. S., Wilkinson, C. W., McTiernan, A., . Craft, S. (2010). Aerobic exercise improves cognit ion for older adults with glucose intolerance, a risk factor for Alzheimer's disease. Journal of Alzheimer's Disease, 22 (2), 569 579. doi: 10.3233/JAD 2010 100768

PAGE 245

230 Baker, S. J., Chrzan, G. J., Park, C. N., & Saunders, J. H. (1985). Validation of human behav ioral tests using ethanol as a CNS depressant model. Neurobehavioral Toxicology and Teratology, 7 (3), 257 261. Barnes, D. E., Yaffe, K., Satariano, W. A., & Tager, I. B. (2003). A longitudinal study of cardiorespiratory fitness and cognitive function in h ealthy older adults. Journal of the American Geriatrics Society, 51 (4), 459 465. Barrett Connor, E., Edelstein, S. L., Corey Bloom, J., & Wiederholt, W. C. (1996). Weight loss precedes dementia in community dwelling older adults. Journal of the American G eriatrics Society, 44 (10), 1147 1152. Bassey, E. J. (2000). The benefits of exercise for the health of older people. Reviews in Clinical Gerontology, 10 (01), 17 31. Bastard, J. P., Jardel, C., Bruckert, E., Blondy, P., Capeau, J., Laville, M., . Hain que, B. (2000). Elevated levels of interleukin 6 are reduced in serum and subcutaneous adipose tissue of obese women after weight loss. Journal of Clinical Endocrinology and Metabolism, 85 (9), 3338 3342. Bastard, J. P., Maachi, M., Lagathu, C., Kim, M. J. Caron, M., Vidal, H., . Feve, B. (2006). Recent advances in the relationship between obesity, inflammation, and insulin resistance. European Cytokine Network, 17 (1), 4 12. Baura, G. D., Foster, D. M., Kaiyala, K., Porte, D., Kahn, S. E., & Schwartz, M. W. (1996). Insulin transport from plasma into the central nervous system is inhibited by dexamethasone in dogs. Diabetes, 45 (1), 86 90. Bdard, M. A., Montplaisir, J., Malo, J., Richer, F., & Rouleau, I. (1993). Persistent neuropsychological deficits and vigilance impairment in sleep apnea syndrome after treatment with continuous positive airways pressure (CPAP). Journal of Clinical and Experimental Neuropsychology, 15 (2), 330 341. Begliuomini, S., Lenzi, E., Ninni, F., Casarosa, E., Merlini, S., Pluc hino, N., . Genazzani, A. R. (2008). Plasma brain derived neurotrophic factor daily variations in men: correlation with cortisol circadian rhythm. Journal of Endocrinology, 197 (2), 429 435. doi: 10.1677/joe 07 0376 Bennett, D. A., Wilson, R. S., Schnei der, J. A., Evans, D. A., Beckett, L. A., Aggarwal, N. T., . Bach, J. (2002). Natural history of mild cognitive impairment in older persons. Neurology, 59 (2), 198 205. doi: 10.1212/wnl.59.2.198 Bergendahl, M., Iranmanesh, A., Mulligan, T., & Veldhuis, J. (2000). Impact of Age on Cortisol Secretory Dynamics Basally and as Driven by Nutrient Withdrawal Stress*. Journal of Clinical Endocrinology and Metabolism, 85 (6), 2203 2214. Berlinger, W. G., & Potter, J. F. (1991). Low body mass index in demented out patients. Journal of the American Geriatrics Society, 39 (10), 973 978.

PAGE 246

231 Bermon, S., Ferrari, P., Bernard, P., Altare, S., & Dolisi, C. (1999). Responses of total and free insulin like growth factor I and insulin like growth factor binding protein 3 after r esistance exercise and training in elderly subjects. Acta Physiologica Scandinavica, 165 (1), 51 56. Berner, Y. N., & Stern, F. (2004). Energy restriction controls aging through neuroendocrine signal transduction. Ageing Res Rev, 3 (2), 189 198. doi: 10.101 6/j.arr.2003.10.004 Beydoun, M. A., & Beason Held, L. L. (2008). Does hypertension interact with body weight to impact cognitive function in the elderly?: Emerging evidence. American Journal of Hypertension, 21 (6), 603. doi: 10.1038/ajh.2008.181 Beydoun, M A., Beydoun, H. A., & Wang, Y. (2008). Obesity and central obesity as risk factors for incident dementia and its subtypes: a systematic review and meta analysis.[Erratum appears in Obes Rev. 2008 May;9(3):267]. Obesity Reviews, 9 (3), 204 218. doi: 10.111 1/j.1467 789X.2008.00473.x Biessels, G. J., ter Braak, E. W. M. T., Erkelens, D. W., & Hijman, R. (2001). Cognitive function in patients with type 2 diabetes mellitus. Neuroscience Research Communications, 28 (1), 11 22. doi: 10.1002/1520 6769(200101/02)28: 1<11::AID NRC2>3.0.CO;2 N Biessels, G. J., & Gispen, W. H. (2005). The impact of diabetes on cognition: what can be learned from rodent models? Neurobiology of Aging, 26 Suppl 1 36 41. doi: 10.1016/j.neurobiolaging.2005.08.015 Bingham, E. M., Hopkins, D., Smith, D., Pernet, A., Hallett, W., Reed, L., . Amiel, S. A. (2002). The role of insulin in human brain glucose metabolism: an 18fluoro deoxyglucose positron emission tomography study. Diabetes, 51 (12), 3384 3390. Bjorntorp, P., & Rosmond, R. (2000). Obesity and cortisol. Nutrition, 16 (10), 924 936. Black, P. H. (2002). Stress and the inflammatory response: a review of neurogenic inflammation. Brain, Behavior, and Immunity, 16 (6), 622 653. Boeka, A. G., & Lokken, K. L. (2008). Neuropsychological per formance of a clinical sample of extremely obese individuals. Archives of Clinical Neuropsychology, 23 (4), 467 474. doi: 10.1016/j.acn.2008.03.003 Bohannon, R. W. (1993). Physical rehabilitation in neurologic diseases. Current Opinion in Neurology, 6 (5), 7 65 772. Bonafe, M., Barbieri, M., Marchegiani, F., Olivieri, F., Ragno, E., Giampieri, C., . Paolisso, G. (2003). Polymorphic variants of insulin like growth factor I (IGF I) receptor and phosphoinositide 3 kinase genes affect IGF I plasma levels and human longevity: cues for an evolutionarily conserved mechanism of life span control. Journal of Clinical Endocrinology and Metabolism, 88 (7), 3299 3304.

PAGE 247

232 Boyko, E. J., de Courten, M., Zimmet, P. Z., Chitson, P., Tuomilehto, J., & Alberti, K. (2000). Featu res of the metabolic syndrome predict higher risk of diabetes and impaired glucose tolerance: a prospective study in Mauritius. Diabetes Care, 23 (9), 1242 1248. Brinkworth, G. D., Buckley, J. D., Noakes, M., Clifton, P. M., & Wilson, C. J. (2009). Long te rm effects of a very low carbohydrate diet and a low fat diet on mood and cognitive function. Archives of Internal Medicine, 169 (20), 1873 1880. doi: 10.1001/archinternmed.2009.329 Brown, A. D., McMorris, C. A., Longman, R. S., Leigh, R., Hill, M. D., Frie denreich, C. M., & Poulin, M. J. (2010). Effects of cardiorespiratory fitness and cerebral blood flow on cognitive outcomes in older women. Neurobiology of Aging, 31 (12), 2047 2057. doi: 10.1016/j.neurobiolaging.2008.11.002 Brown, E. S., Varghese, F. P., & McEwen, B. S. (2004). Association of depression with medical illness: does cortisol play a role? Biological Psychiatry, 55 (1), 1 9. Brubacher, D., Monsch, A., & Stahelin, H. (2004). Weight change and cognitive performance. International Journal of Obesit y, 28 (9), 1163 1167. doi: 10.1038/sj.ijo.0802721 brain damage and improves behavioral outcome following excitotoxic and metabolic insults. Annals of Neurology, 45 (1), 8 15. Brugts, M. P., van Duijn, C. M., H ofland, L. J., Witteman, J. C., Lamberts, S. W., & Janssen, J. A. (2010). Igf I bioactivity in an elderly population: relation to insulin sensitivity, insulin levels, and the metabolic syndrome. Diabetes, 59 (2), 505 508. doi: 10.2337/db09 0583 Bruunsgaard, H., & Pedersen, B. K. (2003). Age related inflammatory cytokines and disease. Immunology and Allergy Clinics of North America, 23 (1), 15 39. Bruunsgaard, H., Skinhoj, P., Pedersen, A. N., Schroll, M., & Pedersen, B. K. (2000). Ageing, tumour necrosis fac tor alpha (TNF alpha) and atherosclerosis. Clinical and Experimental Immunology, 121 (2), 255 260. Bryan, J., & Tiggemann, M. (2001). The effect of weight loss dieting on cognitive performance and psychological well being in overweight women. Appetite, 36 ( 2), 147 156. doi: 10.1006/appe.2000.0389 Buchman, A. S., Boyle, P. A., Yu, L., Shah, R. C., Wilson, R. S., & Bennett, D. A. (2012). Total daily physical activity and the risk of AD and cognitive decline in older adults. Neurology, 78 (17), 1323 1329. doi: 1 0.1212/WNL.0b013e3182535d35 Buchman, A. S., Wilson, R. S., Bienias, J. L., Shah, R. C., Evans, D. A., & Bennett, D. A. (2005). Change in body mass index and risk of incident Alzheimer disease. Neurology, 65 (6), 892 897.

PAGE 248

233 Buckner, R. L., Snyder, A. Z., Sand ers, A. L., Raichle, M. E., & Morris, J. C. (2000). Functional brain imaging of young, nondemented, and demented older adults. Journal of Cognitive Neuroscience, 12 (Supplement 2), 24 34. Budson, A. E., & Solomon, P. R. (2011). Memory Loss: A Practical Gui de for Clinicians Philadelphia: Elsevier Inc. Buffenstein, R., Karklin, A., & Driver, H. S. (2000). Beneficial physiological and performance responses to a month of restricted energy intake in healthy overweight women. Physiology and Behavior, 68 (4), 439 444. Burt, V. L., Whelton, P., Roccella, E. J., Brown, C., Cutler, J. A., Higgins, M., . Labarthe, D. (1995). Prevalence of hypertension in the US adult population results from the Third National Health and Nutrition Examination Survey, 1988 1991. Hyp ertension, 25 (3), 305 313. Byberg, L., Zethelius, B., McKeigue, P. M., & Lithell, H. O. (2001). Changes in physical activity are associated with changes in metabolic cardiovascular risk factors. Diabetologia, 44 (12), 2134 2139. Caamao Isorna, F., Corral M., Montes Martnez, A., & Takkouche, B. (2006). Education and dementia: a meta analytic study. Neuroepidemiology, 26 (4), 226 232. Calabrese, E. J. (2008a). Neuroscience and hormesis: overview and general findings. Critical Reviews in Toxicology, 38 (4), 249 252. doi: 10.1080/10408440801981957 Calabrese, E. J. (2008b). Stress biology and hormesis: the Yerkes Dodson law in psychology -a special case of the hormesis dose response. Critical Reviews in Toxicology, 38 (5), 453 462. doi: 10.1080/1040844080200400 7 Calabrese, E. J., Bachmann, K. A., Bailer, A. J., Bolger, P. M., Borak, J., Cai, L., . Mattson, M. P. (2007). Biological stress response terminology: Integrating the concepts of adaptive response and preconditioning stress within a hormetic dose resp onse framework. Toxicology and Applied Pharmacology, 222 (1), 122 128. Calabrese, E. J., & Baldwin, L. A. (2001). Hormesis: U shaped dose responses and their centrality in toxicology. Trends in Pharmacological Sciences, 22 (6), 285 291. Calabrese, E. J., & Baldwin, L. A. (2003). Hormesis: the dose response revolution. Annual Review of Pharmacology and Toxicology, 43 (1), 175 197. Calabrese, E. J., & Cook, R. (2006). The Importance of Hormesis to Public Health. Environmental Health Perspectives doi: 10.1289 /ehp.8606 Calabrese, V., Cornelius, C., Mancuso, C., Lentile, R., Stella, A. M., & Butterfield, D. A. (2010). Redox homeostasis and cellular stress response in aging and neurodegeneration. Methods in Molecular Biology, 610 285 308. doi: 10.1007/978 1 6032 7 029 8_17

PAGE 249

234 Carretero, O. A., & Oparil, S. (2000). Essential hypertension part I: definition and etiology. Circulation, 101 (3), 329 335. Carro, E., & Torres Aleman, I. (2004). Insulin like growth factor I and Alzheimer's disease: therapeutic prospects? Exp ert Review of Neurotherapeutics, 4 (1), 79 86. doi: 10.1586/14737175.4.1.79 Carro, E., Trejo, J., Gomez Isla, T., LeRoith, D., & Torres Aleman, I. (2002). Serum insulin like growth factor I regulates brain amyloid Nature Medicine, 8 (12), 1390 1397 Carro, E., Trejo, J. L., Busiguina, S., & Torres Aleman, I. (2001). Circulating insulin like growth factor I mediates the protective effects of physical exercise against brain insults of different etiology and anatomy. Journal of Neuroscience, 21 (15), 5 678 5684. Cereda, E., Sacchi, M. C., & Malavazos, A. E. (2009). Central obesity and increased risk of dementia more than three decades later. Neurology, 72 (11), 1030 1031; author reply 1031. doi: 10.1212/01.wnl.0000343499.72241.ea Cereda, E., Sansone, V., Meola, G., & Malavazos, A. E. (2007). Increased visceral adipose tissue rather than BMI as a risk factor for dementia. Age and Ageing, 36 (5), 488 491. doi: 10.1093/ageing/afm096 Cheatham, R. A., Roberts, S. B., Das, S. K., Gilhooly, C. H., Golden, J. K., Hyatt, R., . Lieberman, H. R. (2009). Long term effects of provided low and high glycemic load low energy diets on mood and cognition. Physiology and Behavior, 98 (3), 374 379. doi: 10.1016/j.physbeh.2009.06.015 Chen, Y. C., Chen, T. F., Yip, P. K., Hu, C. Y., Chu, Y. M., & Chen, J. H. (2010). Body mass index (BMI) at an early age and the risk of dementia. Archives of Gerontology and Geriatrics, 50 Suppl 1 S48 52. doi: 10.1016/S0167 4943(10)70013 3 Chiang, C. J., Yip, P. K., Wu, S. C., Lu, C. S., Liou, C. W., Liu, H. C., . Chen, C. J. (2007). Midlife risk factors for subtypes of dementia: a nested case control study in Taiwan. American Journal of Geriatric Psychiatry, 15 (9), 762 771. Choma, C. W., Sforzo, G. A., & Keller, B. A. (1998). Impact of rap id weight loss on cognitive function in collegiate wrestlers. Medicine and Science in Sports and Exercise, 30 (5), 746 749. doi: 10.1097/00005768 199805000 00016 Chrousos, G. P., & Gold, P. W. (1998). A healthy body in a healthy mind and vice versa the dama Journal of Clinical Endocrinology and Metabolism, 83 (6), 1842 1845. Church, T. S., Thomas, D. M., Tudor Locke, C., Katzmarzyk, P. T., Earnest, C. P., Rodarte, R. Q., . Bouchard, C. (2011). Trends over 5 decades i n US occupation related physical activity and their associations with obesity. PloS One, 6 (5), e19657.

PAGE 250

235 Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior 385 396. Colcombe, S., & Kramer, A. F. (2003). Fitness effects on the cognitive function of older adults: a meta analytic study. Psychological Science, 14 (2), 125 130. Colcombe, S. J., Kramer, A. F., Erickson, K. I., Scalf, P., McAuley, E., Cohen, N. J., . Elavsky, S. (2004) Cardiovascular fitness, cortical plasticity, and aging. Proceedings of the National Academy of Sciences of the United States of America, 101 (9), 3316 3321. Colcombe, S. J., Wadwha, R., Kramer, A. F., McAuley, E., Scalf, P. E., Alvarado, M., & Kim, J. (2 005). Cardiovascular fitness training improves cortical recruitment and working memory in older adults: Evidence from a longitudinal fMRI study. Paper presented at the Proceedings of the Ann Meeting Cogn Neurosci Soc. Cole, G. M., & Frautschy, S. A. (2007) The role of insulin and neurotrophic factor signaling in brain aging and Alzheimer's Disease. Experimental Gerontology, 42 (1 2), 10 21. doi: 10.1016/j.exger.2006.08.009 Convit, A., Wolf, O. T., Tarshish, C., & de Leon, M. J. (2003). Reduced glucose toler ance is associated with poor memory performance and hippocampal atrophy among normal elderly. Proceedings of the National Academy of Sciences of the United States of America, 100 (4), 2019 2022. doi: 10.1073/pnas.0336073100 Cooke, J. R., Ayalon, L., Palmer, B. W., Loredo, J. S., Corey Bloom, J., Natarajan, L., . Ancoli Israel, S. (2009). Sustained use of CPAP slows deterioration of cognition, sleep, and mood in patients with Alzheimer's disease and obstructive sleep apnea: a preliminary study. Journal of clinical sleep medicine: JCSM: official publication of the American Academy of Sleep Medicine, 5 (4), 305. Cotman, C. W., & Berchtold, N. C. (2002). Exercise: a behavioral intervention to enhance brain health and plasticity. Trends in Neurosciences, 25 (6) 295 301. Cournot, M., Marquie, J., Ansiau, D., Martinaud, C., Fonds, H., Ferrieres, J., & Ruidavets, J. (2006a). Relation between body mass index and cognitive function in healthy middle aged men and women. Neurology, 67 (7), 1208 1214. doi: 10.1212/01.w nl.0000238082.13860.50 Cournot, M., Marquie, J. C., Ansiau, D., Martinaud, C., Fonds, H., Ferrieres, J., & Ruidavets, J. B. (2006b). Relation between body mass index and cognitive function in healthy middle aged men and women. Neurology, 67 (7), 1208 1214. Craft, S. (2005). Insulin resistance syndrome and Alzheimer's disease: age and obesity related effects on memory, amyloid, and inflammation. Neurobiology of Aging, 26 Suppl 1 65 69. doi: 10.1016/j.neurobiolaging.2005.08.021 Craft, S. (2007). Insulin res istance and Alzheimer's disease pathogenesis: potential mechanisms and implications for treatment. Curr Alzheimer Res, 4 (2), 147 152.

PAGE 251

236 Craft, S., Asthana, S., Cook, D. G., Baker, L. D., Cherrier, M., Purganan, K., . Krohn, A. J. (2003). Insulin dose re sponse effects on memory and plasma amyloid precursor protein in Alzheimer's disease: interactions with apolipoprotein E genotype. Psychoneuroendocrinology, 28 (6), 809 822. Craft, S., Baker, L. D., Montine, T. J., Minoshima, S., Watson, G. S., Claxton, A. . Gerton, B. (2012). Intranasal insulin therapy for Alzheimer disease and amnestic mild cognitive impairment: a pilot clinical trial. Archives of Neurology, 69 (1), 29 38. doi: 10.1001/archneurol.2011.233 Craft, S., & Stennis Watson, G. (2004). Insuli n and neurodegenerative disease: shared and specific mechanisms. The Lancet Neurology, 3 (3), 169 178. doi: 10.1016/s1474 4422(04)00681 7 Cramer, C., Haan, M., Galea, S., Langa, K., & Kalbfleisch, J. (2008). Use of statins and incidence of dementia and cogn itive impairment without dementia in a cohort study. Neurology, 71 (5), 344 350. Croci, L., Barili, V., Chia, D., Massimino, L., van Vugt, R., Masserdotti, G., . Consalez, G. G. (2011). Local insulin like growth factor I expression is essential for Pur kinje neuron survival at birth. Cell Death and Differentiation, 18 (1), 48 59. doi: 10.1038/cdd.2010.78 Croll, S., Suri, C., Compton, D., Simmons, M., Yancopoulos, G., Lindsay, R., . Scharfman, H. (1999). Brain derived neurotrophic factor transgenic mic e exhibit passive avoidance deficits, increased seizure severity and< i> in vitro hyperexcitability in the hippocampus and entorhinal cortex. Neuroscience, 93 (4), 1491 1506. Cukierman, T., Gerstein, H. C., & Williamson, J. D. (2005). Cognitive decline and dementia in diabetes -systematic overview of prospective observational studies. Diabetologia, 48 (12), 2460 2469. doi: 10.1007/s00125 005 0023 4 Cullingford, T. E. (2004). The ketogenic diet; fatty acids, fatty acid activated receptors and neurological disorders. Prostaglandins, Leukotrienes, and Essential Fatty Acids, 70 (3), 253 264. doi: 10.1016/j.plefa.2003.09.008 Cunnane, S. C., & Likhodii, S. S. (2004). Claims to identify detrimental effects of the ketogenic diet (KD) on cognitive function in rats. Pediatric Research, 56 (4), 663 664. D'Ercole, A. J., Ye, P., Calikoglu, A. S., & Gutierrez Ospina, G. (1996). The role of the insulin like growth factors in the central nervous system. Molecular Neurobiology, 13 (3), 227 255. doi: 10.1007/bf02740625 i, K. E., Watts, K. L., Kanarek, R. B., & Taylor, H. A. (2009). Low carbohydrate weight loss diets. Effects on cognition and mood. Appetite, 52 (1), 96 103. Dahl, A., Hassing, L. B., Fransson, E., Berg, S., Gatz, M., Reynolds, C. A., & Pedersen, N. L. (201 0). Being overweight in midlife is associated with lower cognitive ability and steeper

PAGE 252

237 cognitive decline in late life. The Journals of Gerontology: Series A: Biological Sciences and Medical Sciences, 65A (1), 57 62. doi: 10.1093/gerona/glp035 Dahl, A. K., & Hassing, L. B. (2012). Obesity and Cognitive Aging. Epidemiologic Reviews doi: 10.1093/epirev/mxs002 Dahl, A. K., Hassing, L. B., Fransson, E. I., Gatz, M., Reynolds, C. A., & Pedersen, N. L. (2012). Body mass index across midlife and cognitive change in late life. International Journal of Obesity (2005) doi: 10.1038/ijo.2012.37 Dahl, A. K., Lopponen, M., Isoaho, R., Berg, S., & Kivela, S. L. (2008a). Overweight and obesity in old age are not associated with greater dementia risk. Journal of the American Geriatrics Society, 56 (12), 2261 2266. doi: 10.1111/j.1532 5415.2008.01958.x Dahl, A. K., Lopponen, M., Isoaho, R., Berg, S., & Kivela, S. L. (2008b). Overweight and obesity in old age are not associated with greater dementia risk. Journal of the American Geriatrics Society, 56 (12), 2261 2266. doi: 10.1111/j.1532 5415.2008.01958.x Das, U. N. (2002). Obesity, metabolic syndrome X, and inflammation. Nutrition, 18 (5), 430 432. de la Monte, S. M. (2012). Brain insulin resistance and deficiency as therapeutic targets in Alzheimer's disease. Curr Alzheimer Res, 9 (1), 35 66. De Leon, M., McRae, T., Rusinek, H., Convit, A., De Santi, S., Tarshish, C., . Orentreich, N. (1997). Cortisol reduces hippocampal glucose metabolism in normal elderly, but not in Alzhei Journal of Clinical Endocrinology and Metabolism, 82 (10), 3251 3259. Debette, S., Seshadri, S., Beiser, A., Au, R., Himali, J., Palumbo, C., . DeCarli, C. (2011). Midlife vascular risk factor exposure accelerates structural brain aging and cognitive decline. Neurology, 77 (5), 461 468. doi: 10.1212/WNL.0b013e318227b227 Del Parigi, A., Panza, F., Capurso, C., & Solfrizzi, V. (2006). Nutritional factors, cognitive decline, and dementia. Brain Research Bulletin, 69 (1), 1 19. Depke, M., Fusc h, G., Domanska, G., Geffers, R., Volker, U., Schuett, C., & Kiank, C. (2008). Hypermetabolic syndrome as a consequence of repeated psychological stress in mice. Endocrinology, 149 (6), 2714 2723. doi: 10.1210/en.2008 0038 Deschamps, V., Astier, X., Ferry, M., Rainfray, M., Emeriau, J. P., & Barberger Gateau, P. (2002). Nutritional status of healthy elderly persons living in Dordogne, France, and relation with mortality and cognitive or functional decline. European Journal of Clinical Nutrition, 56 (4), 305 3 12. Diamond, D. M., Park, C. R., & Woodson, J. C. (2004). Stress generates emotional memories and retrograde amnesia by inducing an endogenous form of hippocampal LTP. Hippocampus, 14 (3), 281 291. doi: 10.1002/hipo.10186

PAGE 253

238 Dik, M., Deeg, D. J., Visser, M., & Jonker, C. (2003). Early life physical activity and cognition at old age. Journal of Clinical and Experimental Neuropsychology, 25 (5), 643 653. doi: 10.1076/jcen.25.5.643.14583 Diogenes, M. J., Assaife Lopes, N., Pinto Duarte, A., Ribeiro, J. A., & Sebas tiao, A. M. (2007). Influence of age on BDNF modulation of hippocampal synaptic transmission: interplay with adenosine A2A receptors. Hippocampus, 17 (7), 577 585. doi: 10.1002/hipo.20294 Dishman, R. K., Berthoud, H. R., Booth, F. W., Cotman, C. W., Edgerto n, V. R., Fleshner, M. R., . Hillman, C. H. (2006). Neurobiology of exercise. Obesity, 14 (3), 345 356. Doniger, G. M., Simon, E. S., & Zivotofsky, A. Z. (2006). Comprehensive computerized assessment of cognitive sequelae of a complete 12 16 hour fast. Behavioral Neuroscience, 120 (4), 804 816. doi: 10.1037/0735 7044.120.4.804 Driscoll, I., Espeland, M. A., Wassertheil Smoller, S., Gaussoin, S. A., Ding, J., Granek, I. A., . Resnick, S. M. (2011). Weight change and cognitive function: findings from t he Women's Health Initiative Study of Cognitive Aging. Obesity (Silver Spring), 19 (8), 1595 1600. doi: 10.1038/oby.2011.23 Duan, W. (2003). Reversal of Behavioral and Metabolic Abnormalities, and Insulin Resistance Syndrome, by Dietary Restriction in Mice Deficient in Brain Derived Neurotrophic Factor. Endocrinology, 144 (6), 2446 2453. doi: 10.1210/en.2002 0113 Duan, W., Guo, Z., Jiang, H., Ware, M., Li, X. J., & Mattson, M. P. (2003a). Dietary restriction normalizes glucose metabolism and BDNF levels, slow s disease progression, and increases survival in huntingtin mutant mice. Proceedings of the National Academy of Sciences, 100 (5), 2911 2916. Duan, W., Guo, Z., Jiang, H., Ware, M., & Mattson, M. P. (2003b). Reversal of behavioral and metabolic abnormaliti es, and insulin resistance syndrome, by dietary restriction in mice deficient in brain derived neurotrophic factor. Endocrinology, 144 (6), 2446 2453. excitoprotecti ve effect of dietary restriction in mice. Journal of Neurochemistry, 76 (2), 619 626. Duan, W., Lee, J., Guo, Z., & Mattson, M. P. (2001b). Dietary restriction stimulates BDNF production in the brain and thereby protects neurons against excitotoxic injury. Journal of Molecular Neuroscience, 16 (1), 1 12. improve behavioral outcome and reduce degeneration of dopaminergic neurons in models of Parkinson's disease. Journal of Neuroscience Research, 57 (2), 195 206. Duska, F., Andel, M., Kubena, A., & Macdonald, I. A. (2005). Effects of acute starvation on insulin resistance in obese patients with and without type 2 diabetes mellitus. Clinical Nutrition, 24 (6), 1056 1064. doi : 10.1016/j.clnu.2005.08.008

PAGE 254

239 Elias, M., Elias, P., Sullivan, L., Wolf, P., & D'Agostino, R. (2003). Lower cognitive function in the presence of obesity and hypertension: The Framingham heart study. International Journal of Obesity, 27 (2), 260 268. doi: 10. 1038/sj.ijo.802225 Elias, M. F., Elias, P. K., Sullivan, L. M., Wolf, P. A., & D'Agostino, R. B. (2005). Obesity, diabetes and cognitive deficit: The Framingham Heart Study. Neurobiology of Aging, 26 (Suppl1), S11 S16. doi: 10.1016/j.neurobiolaging.2005.08. 019 Engelhart, M. J., Geerlings, M. I., Ruitenberg, A., van Swieten, J. C., Hofman, A., Witteman, J. C., & Breteler, M. M. (2002). Dietary intake of antioxidants and risk of Alzheimer disease. JAMA: the journal of the American Medical Association, 287 (24), 3223 3229. Erickson, K. I., Raji, C. A., Lopez, O. L., Becker, J. T., Rosano, C., Newman, A. B., . Kuller, L. H. (2010). Physical activity predicts gray matter volume in late adulthood: the Cardiovascular Health Study. Neurology, 75 (16), 1415 1422. d oi: 10.1212/WNL.0b013e3181f88359 Erickson, K. I., Voss, M. W., Prakash, R. S., Basak, C., Szabo, A., Chaddock, L., . Kramer, A. F. (2011). Exercise training increases size of hippocampus and improves memory. Proceedings of the National Academy of Scien ces of the United States of America, 108 (7), 3017 3022. doi: 10.1073/pnas.1015950108 Etgen, T., Bickel, H., & Forstl, H. (2010). Metabolic and endocrine factors in mild cognitive impairment. Ageing Res Rev, 9 (3), 280 288. doi: 10.1016/j.arr.2010.01.003 Eva ns, D., Bennett, D., Wilson, R., Bienias, J., Morris, M. C., & Scherr, P. (2003). Incidence of Alzheimer disease in a biracial urban community: Relation to apolipoprotein E allele status. Archives of Neurology, 60 (2), 185 189. Evans, D., Herbert, L. F., B eckett, L. A., Scherr, P., Albert, M. S., & Chown, M. J. (1997). Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons. Archives of Neurology, 56 (11), 1399 1405. Fabrigoule, C. Letenneur, L., Dartigues, J. F., Zarrouk, M., Commenges, D., & Barberger Gateau, P. (1995). Social and leisure activities and risk of dementia: a prospective longitudinal study. Journal of the American Geriatrics Society, 43 (5), 485 490. Fain, J. N. (20 06). Release of interleukins and other inflammatory cytokines by human adipose tissue is enhanced in obesity and primarily due to the nonfat cells. Vitamins and Hormones, 74 443 477. doi: 10.1016/s0083 6729(06)74018 3 Farmer, J., Zhao, X., van Praag, H., Wodtke, K., Gage, F. H., & Christie, B. R. (2004). Effects of voluntary exercise on synaptic plasticity and gene expression in the dentate gyrus of adult male Sprague Dawley rats in vivo. Neuroscience, 124 (1), 71 79. Farooqi, I. S., & O'Rahilly, S. (2009) Leptin: a pivotal regulator of human energy homeostasis. American Journal of Clinical Nutrition, 89 (3), 980S 984S. doi: 10.3945/ajcn.2008.26788C

PAGE 255

240 Faxn Irving, G., Basun, H., & Cederholm, T. (2005). Nutritional and cognitive relationships and long term mo rtality in patients with various dementia disorders. Age and Ageing, 34 (2), 136 141. Feldman, E. L., Sullivan, K. A., Kim, B., & Russell, J. W. (1997). Insulin like growth factors regulate neuronal differentiation and survival. Neurobiology of Disease, 4 ( 3), 201 214. Feldman, H., Doody, R., Kivipelto, M., Sparks, D., Waters, D., Jones, R., . DeMicco, D. (2010). Randomized controlled trial of atorvastatin in mild to moderate Alzheimer disease LEADe. Neurology, 74 (12), 956 964. Fewlass, D. C., Noboa, K ., Pi Sunyer, F. X., Johnston, J. M., Yan, S. D., & Tezapsidis, N. (2004). Obesity related leptin regulates Alzheimer's Abeta. FASEB Journal, 18 (15), 1870 1878. doi: 10.1096/fj.04 2572com Fitzpatrick, A., Kuller, L. H., Ives, D. G., Lopez, O. L., Jagust, W ., & Breitener, J. C. (2004). Incidence and prevalence of dementia in theCardiovascular Health Study. Journal of the American Geriatrics Society, 52 (2), 195 204. Fitzpatrick, A. L., Kuller, L. H., Lopez, O. L., Diehr, P., O'Meara, E. S., Longstreth, W., & Luchsinger, J. A. (2009). Midlife and late life obesity and the risk of dementia: Cardiovascular Health Study. Archives of Neurology, 66 (3), 336 342. doi: 10.1001/archneurol.2008.582 Flachsbart, F., Caliebe, A., Kleindorp, R., Blanche, H., von Eller Ebe rstein, H., Nikolaus, S., . Nebel, A. (2009). Association of FOXO3A variation with human longevity confirmed in German centenarians. Proceedings of the National Academy of Sciences of the United States of America, 106 (8), 2700 2705. doi: 10.1073/pnas.0 809594106 Foley, D. J., & White, L. R. (2002). Dietary intake of antioxidants and risk of Alzheimer disease. JAMA: the journal of the American Medical Association, 287 (24), 3261 3263. Ford, E. S. (2003). The metabolic syndrome and C reactive protein, fibr inogen, and leukocyte count: findings from the Third National Health and Nutrition Examination Survey. Atherosclerosis, 168 (2), 351 358. Forette, F., Seux, M. L., Staessen, J. A., Thijs, L., Birkenhger, W. H., Babarskiene, M. R., . Girerd, X. (1998). Prevention of dementia in randomised double blind placebo controlled Systolic Hypertension in Europe (Syst Eur) trial. The Lancet, 352 (9137), 1347 1351. Forti, P., Pisacane, N., Rietti, E., Lucicesare, A., Olivelli, V., Mariani, E., . Ravaglia, G. (2 010). Metabolic syndrome and risk of dementia in older adults. Journal of the American Geriatrics Society, 58 (3), 487 492. doi: 10.1111/j.1532 5415.2010.02731.x Fratiglioni, L., Paillard Borg, S., & Winblad, B. (2004). An active and socially integrated lif estyle in late life might protect against dementia. Lancet Neurology, 3 (6), 343 353. doi: Doi 10.1016/S1474 4422(04)00767 7

PAGE 256

241 Freude, S., Schilbach, K., & Schubert, M. (2009). The role of IGF 1 receptor and insulin receptor signaling for the pathogenesis of Alzheimer's disease: from model organisms to human disease. Curr Alzheimer Res, 6 (3), 213 223. Fried, S. K., Bunkin, D. A., & Greenberg, A. S. (1998). Omental and subcutaneous adipose tissues of obese subjects release interleukin 6: depot difference and r egulation by glucocorticoid. Journal of Clinical Endocrinology and Metabolism, 83 (3), 847 850. Friguet, B. (2002). Protein repair and degradation during aging. ScientificWorldJournal, 2 248 254. doi: 10.1100/tsw.2002.98 Frystyk, J. (2010). Exercise and t he growth hormone insulin like growth factor axis. Medicine and Science in Sports and Exercise, 42 (1), 58 66. doi: 10.1249/MSS.0b013e3181b07d2d Gage, F. H. (2000). Mammalian neural stem cells. Science, 287 (5457), 1433 1438. Gamble, J. H. P., Ormerod, J. O M., & Frenneaux, M. P. (2008). Exercise can be effective therapy for depression. Practitioner, 252 (1710), 19 20. Gami, A. S., Pressman, G., Caples, S. M., Kanagala, R., Gard, J. J., Davison, D. E., . Somers, V. K. (2004). Association of atrial fibri llation and obstructive sleep apnea. Circulation, 110 (4), 364 367. Gao, S., Nguyen, J. T., Hendrie, H. C., Unverzagt, F. W., Hake, A., Smith Gamble, V., & Hall, K. (2011). Accelerated weight loss and incident dementia in an elderly African American cohort Journal of the American Geriatrics Society, 59 (1), 18 25. doi: 10.1111/j.1532 5415.2010.03169.x Gapstur, S. M., Kopp, P., Chiu, B. C., Gann, P. H., Colangelo, L. A., & Liu, K. (2004). Longitudinal associations of age, anthropometric and lifestyle factors with serum total insulin like growth factor I and IGF binding protein 3 levels in Black and White men: the CARDIA Male Hormone Study. Cancer Epidemiology, Biomarkers and Prevention, 13 (12), 2208 2216. Geerlings, M., Den Heijer, T., Koudstaal, P., Hofman, A., & Breteler, M. (2008). History of depression, depressive symptoms, and medial temporal lobe atrophy and the risk of Alzheimer disease. Neurology, 70 (15), 1258 1264. Gelber, R. P., Petrovitch, H., Masaki, K. H., Abbott, R. D., Ross, G. W., Launer, L. J., & White, L. R. (2012). Lifestyle and the risk of dementia in Japanese American men. Journal of the American Geriatrics Society, 60 (1), 118 123. doi: 10.1111/j.1532 5415.2011.03768.x Gibson, E. L., & Green, M. W. (2002). Nutritional influences on cognit ive function: mechanisms of susceptibility. Nutrition Research Reviews, 15 (1), 169. Glymour, M. M., & Manly, J. (2008). Lifecourse Social Conditions and Racial and Ethnic Patterns of Cognitive Aging. Neuropsychology Review, 18 (3), 223 254. doi: 10.1007/s1 1065 008 9064 z

PAGE 257

242 Glynn, R. J., Beckett, L. A., Hebert, L. E., Morris, M. C., Scherr, P. A., & Evans, D. A. (1999). Current and remote blood pressure and cognitive decline. JAMA: the journal of the American Medical Association, 281 (5), 438 445. Gobbo, O. L. & O'Mara, S. M. (2005). Exercise, but not environmental enrichment, improves learning after kainic acid induced hippocampal neurodegeneration in association with an increase in brain derived neurotrophic factor. Behavioural Brain Research, 159 (1), 21 26. doi: 10.1016/j.bbr.2004.09.021 Goodrick, C., Ingram, D., Reynolds, M., Freeman, J., & Cider, N. (2009). Effects of intermittent feeding upon growth and life span in rats. Gerontology, 28 (4), 233 241. Gorospe, E. C., & Dave, J. K. (2007). The risk of deme ntia with increased body mass index. Age and Ageing, 36 (1), 23 29. Gram, I. T., Norat, T., Rinaldi, S., Dossus, L., Lukanova, A., Tehard, B., . Kaaks, R. (2006). Body mass index, waist circumference and waist hip ratio and serum levels of IGF I and IG FBP 3 in European women. International Journal of Obesity (2005), 30 (11), 1623 1631. doi: 10.1038/sj.ijo.0803324 Gray, S. L., Lai, K. V., & larson, E. (1999). Drug induced cognition disorders in the elderly: Incidence, prevention and management. Drug Safet y, 21 101 122. Grealy, M. A., Johnson, D. A., & Rushton, S. K. (1999). Improving cognitive function after brain injury: the use of exercise and virtual reality. Archives of Physical Medicine and Rehabilitation, 80 (6), 661 667. Greco, S. J., Sarkar, S. Johnston, J. M., & Tezapsidis, N. (2009). Leptin regulates tau phosphorylation and amyloid through AMPK in neuronal cells. Biochemical and Biophysical Research Communications, 380 (1), 98 104. doi: 10.1016/j.bbrc.2009.01.041 Green, M. W., Elliman, N. A., & Kretsch, M. J. (2005). Weight loss strategies, stress, and cognitive function: Supervised versus unsupervised dieting. Psychoneuroendocrinology, 30 (9), 908 918. doi: 10.1016/j.psyneuen.2005.05.005 Green, M. W., Jones, A. D., Smith, I. D., Cobain, M. R., Williams, J. M., Healy, H., . Durlach, P. J. (2003a). Impairments in working memory associated with naturalistic dieting in women: no relationship between task performance and urinary 5 HIAA levels. Appetite, 40 (2), 145 153. Green, M. W., & Rogers, P. J. (1995). Impaired cognitive functioning during spontaneous dieting. Psychological Medicine, 25 (5), 1003 1010. Green, M. W., & Rogers, P. J. (1998). Impairments in working memory associated with spontaneous dieting behaviour. Psychological Medicine, 28 ( 5), 1063 1070.

PAGE 258

243 Green, M. W., Rogers, P. J., Elliman, N. A., & Gatenby, S. J. (1994). Impairment of cognitive performance associated with dieting and high levels of dietary restraint. Physiology and Behavior, 55 (3), 447 452. doi: 10.1016/0031 9384%2894%299 0099 X Green, R. C., Cupples, L. A., Kurz, A., Auerbach, S., Go, R., Sadovnick, D., . Edeki, T. (2003b). Depression as a risk factor for Alzheimer disease: the MIRAGE Study. Archives of Neurology, 60 (5), 753. Griesbach, G. S., Hovda, D., Molteni, R., Wu, A., & Gomez Pinilla, F. (2004). Voluntary exercise following traumatic brain injury: brain derived neurotrophic factor upregulation and recovery of function. Neuroscience, 125 (1), 129 140. Grigg Damberger, M., & Ralls, F. (2012). Cognitive dysfunction and obstructive sleep apnea: from cradle to tomb. Current Opinion in Pulmonary Medicine, 18 (6), 580 587. Guldstrand, M., Ahren, B., Wredling, R., Backman, L., Lins, P. E., & Adamson, U. (2003). Alteration of the counterregulatory responses to insulin ind uced hypoglycemia and of cognitive function after massive weight reduction in severely obese subjects. Metabolism: Clinical and Experimental, 52 (7), 900 907. Gunstad, J., Lhotsky, A., Wendell, C. R., Ferrucci, L., & Zonderman, A. B. (2010). Longitudinal e xamination of obesity and cognitive function: Results from the Baltimore Longitudinal Study of Aging. Neuroepidemiology, 34 (4), 222 229. doi: 10.1159/000297742 Gunstad, J., Strain, G., Devlin, M. J., Wing, R., Cohen, R. A., Paul, R. H., . Mitchell, J. E. (2011). Improved memory function 12 weeks after bariatric surgery. Surgery for Obesity and Related Diseases, 7 (4), 465 472. doi: 10.1016/j.soard.2010.09.015 Guo, Z., Viitanen, M., Fratiglioni, L., & Winblad, B. (1996). Low blood pressure and dementia in elderly people: the Kungsholmen project. BMJ, 312 (7034), 805 808. Gustafson, D. (2006). Adiposity indices and dementia. Lancet Neurology, 5 (8), 713 720. doi: 10.1016/s1474 4422(06)70526 9 Gustafson, D. (2008). A life course of adiposity and dementia. Eur opean Journal of Pharmacology, 585 (1), 163 175. doi: 10.1016/j.ejphar.2008.01.052 Gustafson, D., Backman, K., Waern, M., Ostling, S., Guo, X., Zandi, P., . Skoog, I. (2009). Adiposity indicators and dementia over 32 years in Sweden. Neurology, 73 (19), 1559 1566. doi: 10.1212/WNL.0b013e3181c0d4b6 Gustafson, D., Lissner, L., Bengtsson, C., Bjorkelund, C., & Skoog, I. (2004). A 24 year follow up of body mass index and cerebral atrophy.[Summary for patients in Neurology. 2004 Nov 23;63(10):E19 20; PMID: 155 57485]. Neurology, 63 (10), 1876 1881. Gustafson, D., Rothenberg, E., Blennow, K., Steen, B., & Skoog, I. (2003). An 18 year follow up of overweight and risk of Alzheimer disease. Archives of Internal Medicine, 163 (13), 1524 1528.

PAGE 259

244 Haag, M. D., Hofman, A., Koudstaal, P. J., Stricker, B. H., & Breteler, M. M. (2009). Statins are associated with a reduced risk of Alzheimer disease regardless of lipophilicity. The Rotterdam Study. Journal of Neurology, Neurosurgery and Psychiatry, 80 (1), 13 17. Hach, I., Ruhl U. E., Klose, M., Klotsche, J., Kirch, W., & Jacobi, F. (2007). Obesity and the risk for mental disorders in a representative German adult sample. The European Journal of Public Health, 17 (3), 297 305. Halberg, N., Henriksen, M., Sderhamn, N., Stallkne cht, B., Ploug, T., Schjerling, P., & Dela, F. (2005). Effect of intermittent fasting and refeeding on insulin action in healthy men. Journal of Applied Physiology, 99 (6), 2128 2136. Halkjr, J., Holst, C., & Srensen, T. I. (2003). Intelligence test scor e and educational level in relation to BMI changes and obesity. Obesity Research, 11 (10), 1238 1245. Hallbook, T., Ji, S., Maudsley, S., & Martin, B. (2012). The effects of the ketogenic diet on behavior and cognition. Epilepsy Research, 100 (3), 304 309. doi: 10.1016/j.eplepsyres.2011.04.017 Halyburton, A. K., Brinkworth, G. D., Wilson, C. J., Noakes, M., Buckley, J. D., Keogh, J. B., & Clifton, P. M. (2007). Low and high carbohydrate weight loss diets have similar effects on mood but not cognitive perfor mance. American Journal of Clinical Nutrition, 86 (3), 580 587. Han, C., Jo, S. A., Seo, J. A., Kim, B. G., Kim, N. H., Jo, I., . Park, K. W. (2009). Adiposity parameters and cognitive function in the elderly: application of "Jolly Fat" hypothesis to c ognition. Archives of Gerontology and Geriatrics, 49 (2), e133 138. doi: 10.1016/j.archger.2008.11.005 Hanniman, E. A., Lambert, G., Inoue, Y., Gonzalez, F. J., & Sinal, C. J. (2006). Apolipoprotein A IV is regulated by nutritional and metabolic stress: inv olvement of glucocorticoids, HNF and PGC Journal of Lipid Research, 47 (11), 2503 2514. Hartman, A. L., & Vining, E. P. G. (2007). Clinical aspects of the ketogenic diet. Epilepsia, 48 (1), 31 42. Harvey, J. (2003). Leptin: a multifaceted hormone i n the central nervous system. Molecular Neurobiology, 28 (3), 245 258. doi: 10.1385/mn:28:3:245 Hassing, L. B., Dahl, A. K., Pedersen, N. L., & Johansson, B. (2010). Overweight in midlife is related to lower cognitive function 30 years later: A prospective study with longitudinal assessments. Dementia and Geriatric Cognitive Disorders, 29 (6), 543 552. doi: 10.1159/000314874 Hassing, L. B., Dahl, A. K., Thorvaldsson, V., Berg, S., Gatz, M., Pedersen, N. L., & Johansson, B. (2009). Overweight in midlife and ri sk of dementia: a 40 year follow up study. International Journal of Obesity, 33 (8), 893 898. doi: 10.1038/ijo.2009.104

PAGE 260

245 Hayden, K. M., Zandi, P. P., Lyketsos, C. G., Khachaturian, A. S., Bastian, L. A., Charoonruk, G., . Welsh Bohmer, K. A. (2006a). Vas cular Risk Factors for Incident Alzheimer Disease and Vascular Dementia: The Cache County Study. Alzheimer Disease and Associated Disorders, 20 (2), 93 100. doi: 10.1097/01.wad.0000213814.43047.86 Hayden, K. M., Zandi, P. P., Lyketsos, C. G., Khachaturian, A. S., Bastian, L. A., Charoonruk, G., . Cache County, I. (2006b). Vascular risk factors for incident Alzheimer disease and vascular dementia: the Cache County study. Alzheimer Disease and Associated Disorders, 20 (2), 93 100. Heilbronn, L. K., & Ravus sin, E. (2003). Calorie restriction and aging: review of the literature and implications for studies in humans. American Journal of Clinical Nutrition, 78 (3), 361 369. Heilbronn, L. K., Smith, S. R., Martin, C. K., Anton, S. D., & Ravussin, E. (2005). Alt ernate day fasting in nonobese subjects: effects on body weight, body composition, and energy metabolism. American Journal of Clinical Nutrition, 81 (1), 69 73. Hellweg, R., Ziegenhorn, A., Heuser, I., & Deuschle, M. (2008). Serum concentrations of nerve g rowth factor and brain derived neurotrophic factor in depressed patients before and after antidepressant treatment. Pharmacopsychiatry, 41 (02), 66 71. Neurotherapeutics, 5 4 70 480. Herbert, L. F., Beckett, L. A., Scherr, P. A., & Evans, D. (2001). Annual Incidence of Alzheimer Disease in the United States Projected to the Years ooo Through o5o. Annual Incidence of Alzheimer Disease in the United States Projected to the Years ooo Through o5o, 15 (4), 169 173. Heyn, P., Abreu, B. C., & Ottenbacher, K. J. (2004). The effects of exercise training on elderly persons with cognitive impairment and dementia: a meta analysis. Archives of Physical Medicine and Rehabilitation, 85 (10), 1694 1704. Ho, L., Qin, W., Pompl, P. N., Xiang, Z., Wang, J., Zhao, Z., . Mobbs, C. V. (2004). Diet induced disease. The FASEB journal, 18 (7), 902 904. Holden, K. F. Lindquist, K., Tylavsky, F. A., Rosano, C., Harris, T. B., Yaffe, K., & Health, A. B. C. s. (2009). Serum leptin level and cognition in the elderly: Findings from the Health ABC Study. Neurobiology of Aging, 30 (9), 1483 1489. doi: 10.1016/j.neurobiolagin g.2007.11.024 Hughes, T., Borenstein, A., Schofield, E., Wu, Y., & Larson, E. (2009). Association between late life body mass index and dementia: The Kame Project. Neurology, 72 (20), 1741 1746. doi: 10.1212/WNL.0b013e3181a60a58

PAGE 261

246 Ingram, D. K., Chefer, S., M atochik, J., Moscrip, T. D., Weed, J., Roth, G. S., . Lane, M. A. (2001). Aging and caloric restriction in nonhuman primates: behavioral and in vivo brain imaging studies. Annals of the New York Academy of Sciences, 928 316 326. Ingvartsen, K. L., & Boisclair, Y. R. (2001). Leptin and the regulation of food intake, energy homeostasis and immunity with special focus on periparturient ruminants. Domestic Animal Endocrinology, 21 (4), 215 250. Jack, C. R., Jr., Albert, M. S., Knopman, D. S., McKhann, G. M., Sperling, R. A., Carrillo, M. C., . Phelps, C. H. (2011). Introduction to the recommendations from the National Institute on Aging Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7 (3), 257 262 doi: 10.1016/j.jalz.2011.03.004 Jackson, R., Moloney, M., Lowy, C., Wright, A., Hartog, M., Pilkington, T., & Fraser, T. (1971). Differences between metabolic responses to fasting in obese diabetic and obese nondiabetic subjects. Diabetes, 20 (4), 214 227 Jacobson, L., & Sapolsky, R. (1991). The role of the hippocampus in feedback regulation of the hypothalamic pituitary adrenocortical axis. Endocrine Reviews, 12 (2), 118 134. Jick, H., Zornberg, G. L., Jick, S. S., Seshadri, S., & Drachman, D. A. (2000) Statins and the risk of dementia. The Lancet, 356 (9242), 1627 1631. Johnson, D. K., Wilkins, C. H., & Morris, J. C. (2006). Accelerated Weight Loss May Precede Diagnosis in Alzheimer Disease. Archives of Neurology, 63 (9), 1312 1317. doi: 10.1001/archneu r.63.9.1312 Johnson, J. B., Summer, W., Cutler, R. G., Martin, B., Hyun, D. H., Dixit, V. D., . Mattson, M. P. (2007). Alternate day calorie restriction improves clinical findings and reduces markers of oxidative stress and inflammation in overweight a dults with moderate asthma. Free Radical Biology and Medicine, 42 (5), 665 674. doi: 10.1016/j.freeradbiomed.2006.12.005 Johnston, C. S., Tjonn, S. L., Swan, P. D., White, A., Hutchins, H., & Sears, B. (2006). Ketogenic low carbohydrate diets have no metabo lic advantage over nonketogenic low carbohydrate diets. The American journal of clinical nutrition, 83 (5), 1055 1061. Kaiyala, K. J., Prigeon, R. L., Kahn, S. E., Woods, S. C., & Schwartz, M. W. (2000). Obesity induced by a high fat diet is associated wit h reduced brain insulin transport in dogs. Diabetes, 49 (9), 1525 1533. Kalmijn, S., Janssen, J. A., Pols, H. A., Lamberts, S. W., & Breteler, M. M. (2000). A prospective study on circulating insulin like growth factor I (IGF I), IGF binding proteins, and cognitive function in the elderly. Journal of Clinical Endocrinology and Metabolism, 85 (12), 4551 4555. Kanaya, A. M., Lindquist, K., Harris, T. B., Launer, L., Rosano, C., Satterfield, S., . Health, A. B. C. S. (2009). Total and regional adiposity an d cognitive change in older adults: The Health,

PAGE 262

247 Aging and Body Composition (ABC) study. Archives of Neurology, 66 (3), 329 335. doi: 10.1001/archneurol.2008.570 Kang, H. C., Chung, D. E., Kim, D. W., & Kim, H. D. (2004). Early and late onset complications of the ketogenic diet for intractable epilepsy. Epilepsia, 45 (9), 1116 1123. Karege, F., Perret, G., Bondolfi, G., Schwald, M., Bertschy, G., & Aubry, J. M. (2002). Decreased serum brain derived neurotrophic factor levels in major depressed patients. Psyc hiatry Research, 109 (2), 143 148. Karlamangla, A. S., Singer, B. H., Chodosh, J., McEwen, B. S., & Seeman, T. E. (2005). Urinary cortisol excretion as a predictor of incident cognitive impairment. Neurobiology of Aging, 26 Suppl 1 80 84. doi: 10.1016/j.n eurobiolaging.2005.09.037 Kearney, P. M., Whelton, M., Reynolds, K., Muntner, P., Whelton, P. K., & He, J. (2005). Global burden of hypertension: analysis of worldwide data. Lancet, 365 (9455), 217 223. Kerwin, D. R., Gaussoin, S. A., Chlebowski, R. T., Ku ller, L. H., Vitolins, M., Coker, L. H., . Espeland, M. A. (2011). Interaction between body mass index and central adiposity and risk of incident cognitive impairment and dementia: Results from the Women's Health Initiative Memory Study. Journal of the American Geriatrics Society, 59 (1), 107 112. doi: 10.1111/j.1532 5415.2010.03219.x Khabour, O. F., Alzoubi, K. H., Alomari, M. A., & Alzubi, M. A. (2010). Changes in spatial memory and BDNF expression to concurrent dietary restriction and voluntary exerci se. Hippocampus, 20 (5), 637 645. doi: 10.1002/hipo.20657 Kim, S. J., Lee, J. H., Lee, D. Y., Jhoo, J. H., & Woo, J. I. (2011). Neurocognitive dysfunction associated with sleep quality and sleep apnea in patients with mild cognitive impairment. American Jou rnal of Geriatric Psych, 19 (4), 374 381. Kivipelto, M., Helkala, E. L., T., Hallikainen, M., Alhainen, K., . and high midlife systolic blood pressure are independent risk factors for late life Alzheimer dis ease. Annals of Internal Medicine, 137 (3), 149 155. Kivipelto, M., Helkala, E. L., Laakso, M. P., Hnninen, T., Hallikainen, M., Alhainen, K., . Nissinen, A. (2001). Midlife vascular risk factors and Alzheimer's disease in later life: longitudinal, po pulation based study. BMJ, 322 (7300), 1447 1451. Kivipelto, M., Ngandu, T., Fratiglioni, L., Viitanen, M., Kareholt, I., Winblad, B., . Nissinen, A. (2005). Obesity and vascular risk factors at midlife and the risk of dementia and Alzheimer disease. A rchives of Neurology, 62 (10), 1556 1560. doi: 10.1001/archneur.62.10.1556 Knight, J. A. (2000). The biochemistry of aging. Advances in Clinical Chemistry, 35 1 62.

PAGE 263

248 Knight, Z. A., Hannan, K. S., Greenberg, M. L., & Friedman, J. M. (2010). Hyperleptinemia is required for the development of leptin resistance. PloS One, 5 (6), e11376. Knopman, D. S. (2008). Go to the head of the class to avoid vascular dementia and skip diabetes and obesity. Neurology, 71 (14), 1046 1047. doi: 10.1212/01.wnl.0000326899.25052.8 2 Knopman, D. S., Edland, S. D., Cha, R. H., Petersen, R. C., & Rocca, W. A. (2007). Incident dementia in women is preceded by weight loss by at least a decade. Neurology, 69 (8), 739 746. Knopman, D. S., Mosley, T. H., Catellier, D. J., & Coker, L. H. (20 09). Atherosclerosis Risk in Communities Study Brain, M. R. I. Study Fourteen year longitudinal study of vascular risk factors, APOE genotype, and cognition: the ARIC MRI Study. Alzheimers Dement, 5 (3), 207 214. doi: 10.1016/j.jalz.2009.01.027 Koebnick, C. Smith, N., Huang, K., Martinez, M. P., Clancy, H. A., & Kushi, L. H. (2012). The prevalence of obesity and obesity related health conditions in a large, multiethnic cohort of young adults in California. Annals of Epidemiology Kohman, R. A., DeYoung, E. K., Bhattacharya, T. K., Peterson, L. N., & Rhodes, J. S. (2012). Wheel running attenuates microglia proliferation and increases expression of a proneurogenic phenotype in the hippocampus of aged mice. Brain, Behavior, and Immunity, 26 (5), 803 810. doi: 1 0.1016/j.bbi.2011.10.006 Kojima, T., Kamei, H., Aizu, T., Arai, Y., Takayama, M., Nakazawa, S., . Hirose, N. (2004). Association analysis between longevity in the Japanese population and polymorphic variants of genes involved in insulin and insulin lik e growth factor 1 signaling pathways. Experimental Gerontology, 39 (11 12), 1595 1598. doi: 10.1016/j.exger.2004.05.007 Komulainen, P., Pedersen, M., Hanninen, T., Bruunsgaard, H., Lakka, T. A., Kivipelto, M., . Rauramaa, R. (2008). BDNF is a novel mark er of cognitive function in ageing women: the DR's EXTRA Study. Neurobiology of Learning and Memory, 90 (4), 596 603. doi: 10.1016/j.nlm.2008.07.014 Koponen, E., Voikar, V., Riekki, R., Saarelainen, T., Rauramaa, T., Rauvala, H., . Castren, E. (2004). T ransgenic mice overexpressing the full length neurotrophin receptor trkB exhibit increased activation of the trkB PLCgamma pathway, reduced anxiety, and facilitated learning. Molecular and Cellular Neurosciences, 26 (1), 166 181. doi: 10.1016/j.mcn.2004.01. 006 Kossoff, E. H. (2004). More fat and fewer seizures: dietary therapies for epilepsy. Lancet Neurology, 3 (7), 415 420. Krabbe, K. S., Nielsen, A. R., Krogh Madsen, R., Plomgaard, P., Rasmussen, P., Erikstrup, C., . Pedersen, B. K. (2007). Brain deri ved neurotrophic factor (BDNF) and type 2 diabetes. Diabetologia, 50 (2), 431 438.

PAGE 264

249 Kramer, A. F., Colcombe, S. J., McAuley, E., Scalf, P. E., & Erickson, K. I. (2005). Fitness, aging and neurocognitive function. Neurobiology of Aging, 26 Suppl 1 124 127. doi: 10.1016/j.neurobiolaging.2005.09.009 Kramer, A. F., Erickson, K. I., & Colcombe, S. J. (2006). Exercise, cognition, and the aging brain. Journal of Applied Physiology, 101 (4), 1237 1242. doi: 10.1152/japplphysiol.00500.2006 Kretsch, M. J., Green, M. W ., Fong, A. K., Elliman, N. A., & Johnson, H. L. (1997). Cognitive effects of a long term weight reducing diet. International Journal of Obesity & Related Metabolic Disorders: Journal of the International Association for the Study of Obesity, 21 (1), 14 21. Krieg, E. F., Chrislip, D. W., Letz, R. E., Otto, D. A., Crespo, C. J., BRIGHTWELL, W. S., & Ehrenberg, R. L. (2001). Neurobehavioral test performance in the third National Health and Nutrition Examination Survey. Neurotoxicology and Teratology, 23 (6), 5 69 589. Krieger, D. R., & Landsberg, L. (1988). Mechanisms in obesity related hypertension: role of insulin and catecholamines. American Journal of Hypertension, 1 (1), 84 90. Krikorian, R., Shidler, M. D., Dangelo, K., Couch, S. C., Benoit, S. C., & Cleg g, D. J. (2012). Dietary ketosis enhances memory in mild cognitive impairment. Neurobiology of Aging, 33 (2), e19 e27. doi: 10.1016/j.neurobiolaging.2010.10.006 Kuh, D., & Ben Shlomo, Y. (2004). A Life Course Approach to Chronic Diseases Epidemiology (2nd e d.): Oxford University Press. Kukull, W. A., Higdon, R., Bowen, J. D., McCormick, W. C., Teri, L., Schellenberg, G. D., . Larson, E. B. (2002). Dementia and Alzheimer disease incidence: a prospective cohort study. Archives of Neurology, 59 (11), 1737. Kwiterovich, P. O., Vining, E. P. G., Pyzik, P., Skolasky, R., & Freeman, J. M. (2003). Effect of a high fat ketogenic diet on plasma levels of lipids, lipoproteins, and apolipoproteins in children. JAMA, 290 (7), 912 920. Kyrou, I., & Tsigos, C. (2007). S tress mechanisms and metabolic complications. Hormone and Metabolic Research, 39 (6), 430 438. doi: 10.1055/s 2007 981462 Laitala, V., Kaprio, J., Koskenvuo, M., Raiha, I., Rinne, J., & Silventoinen, K. (2011). Association and causal relationship of midlife obesity and related metabolic disorders with old age cognition. Current Alzheimer Research, 8 (6), 699 706. doi: 10.2174/156720511796717186 Lam, J., Sharma, S., & Lam, B. (2010). Obstructive sleep apnoea: definitions, epidemiology & natural history. Indian Journal of Medical Research, 131 165 170. Landfield, P. W., Blalock, E. M., Chen, K. C., & Porter, N. M. (2007). A new glucocorticoid Current Alzheimer Research, 4 (2), 205.

PAGE 265

250 Landsberg, L. (2001). Insulin mediated sympathetic stimulation: role in the pathogenesis of obesity related hypertension (or, how insulin affects blood pressure, and why). Journal of Hypertension, 19 (3), 523 528. Laron, Z. (2001). Insulin like growth factor 1 (IGF 1): a growth hormone. Molecular Pathology, 54 (5), 311 316. Laron, Z. (2002). Effects of growth hormone and insulin like growth factor 1 deficiency on ageing and longevity. Novartis Foundation Symposium, 242 125 137; discussion 137 142. Laske, C., Stransky, E., Leyhe, T., Eschweiler, G. W., Maetzler, W., Wittorf, A., . Schott, K. (2007). BDNF serum and CSF concentrations in Alzheimer's disease, normal pressure hydrocephalus and healthy controls. Journal of Psychiatric Research, 41 (5), 387 394. doi: 10.101 6/j.jpsychires.2006.01.014 Laske, C., Stransky, E., Leyhe, T., Eschweiler, G. W., Wittorf, A., Richartz, E., . Schott, K. (2006). Stage dependent BDNF serum concentrations in Alzheimer's disease. Journal of Neural Transmission, 113 (9), 1217 1224. doi: 10.1007/s00702 005 0397 y Launer, L. J., Ross, G. W., Petrovitch, H., Masaki, K., Foley, D., White, L. R., & Havlik, R. J. (2000). Midlife blood pressure and dementia: the Honolulu Asia aging study< sup> . Neurobiology of Aging, 21 (1), 49 55. Laurin D., Verreault, R., Lindsay, J., MacPherson, K., & Rockwood, K. (2001). Physical activity and risk of cognitive impairment and dementia in elderly persons. Archives of Neurology, 58 (3), 498 504. Lautenschlager, N. T., & Almeida, O. P. (2006). Physical ac tivity and cognition in old age. Curr Opin Psychiatry, 19 (2), 190 193. doi: 10.1097/01.yco.0000214347.38787.37 Lauterio, T. J. (1992). The effects of IGF I and IGF II on cell growth and differentiation in the central nervous system. Advances in Experimenta l Medicine and Biology, 321 31 36. Lee, C. H., Olson, P., Hevener, A., Mehl, I., Chong, L. W., Olefsky, J. M., . Peters, J. M. (2006a). Proceedings of the National Academy of Sciences of the United States of America, 103 (9), 3444 3449. Lee, G. D., Wilson, M. A., Zhu, M., Wolkow, C. A., De Cabo, R., Ingram, D. K., & Zou, S. (2006b). Dietary deprivation extends lifespan in Caenorhabditis elegans. Aging cell, 5 (6), 515 524. Lee, J., Duan, W., Long, J. M., Ingram, D. K., & Mattson, M. P. (2000). Dietary restriction increases the number of newly generated neural cells, and induces BDNF expression, in the dentate gyrus of rats. Journal of Molecular Neuroscience, 15 (2), 99 108. Lee, J., Duan, W., required for basal neurogenesis and mediates, in part, the enhancement of neurogenesis by dietary restriction in the hippocampus of adult mice. Journal of Neurochemistry, 82 (6), 13 67 1375.

PAGE 266

251 Lee, L., Kang, S. A., Lee, H. O., Lee, B. H., Park, J. S., Kim, J. H., . Lee, J. E. (2001). Relationships between dietary intake and cognitive function level in Korean elderly people. Public Health, 115 (2), 133 138. Lemura, L. M., von Duvill ard, S. P., & Mookerjee, S. (2000). The effects of physical training of functional capacity in adults. Ages 46 to 90: a meta analysis. Journal of Sports Medicine and Physical Finess, 40 (1), 1 10. Letz, R. (1989). Quantitative neurobehavioral testing in hu mans for assessing potential effects of occupational exposure. International Journal of Toxicology, 8 (2), 303 309. Letz, R. (1990). Neurobehavioral evaluation system: an international effort. Liebermeister, H., & Schroter, K. (1983). Absence of detriment al changes of cognitive parameters during fasting. International Journal of Obesity, 7 (1), 45 51. Lindwall, M., & Hassmen, P. (2006). [Exercise and self confidence -keys to better mental health of the elderly]. Lakartidningen, 103 (47), 3710 3713. Liu Amb rose, T., Nagamatsu, L. S., Graf, P., Beattie, B. L., Ashe, M. C., & Handy, T. C. (2010). Resistance training and executive functions: a 12 month randomized controlled trial. Archives of Internal Medicine, 170 (2), 170. Lo, A. H., Pachana, N. A., Byrne, G. J., Sachdev, P. S., & Woodman, R. J. (2012). Relationship between changes in body weight and cognitive function in middle aged and older women. International Journal of Geriatric Psychiatry, 27 (8), 863 872. doi: 10.1002/gps.2797 Lopez, P. P., Stefan, B., Schulman, C. I., & Byers, P. M. (2008). Prevalence of sleep apnea in morbidly obese patients who presented for weight loss surgery evaluation: more evidence for routine screening for obstructive sleep apnea before weight loss surgery. The American surgeon, 74 (9), 834 838. Lotfi, S., Madani, M., Tazi, A., Boumahmaza, M., & Talbi, M. (2010). [Variation of cognitive functions and glycemia during physical exercise in Ramadan fasting]. Revue Neurologique, 166 (8 9), 721 726. doi: 10.1016/j.neurol.2010.01.016 Lot tenberg, S., Giannella Neto, D., Derendorf, H., Rocha, M., Bosco, A., Carvalho, S., . Wajchenberg, B. (1998). Effect of fat distribution on the pharmacokinetics of cortisol in obesity. International Journal of Clinical Pharmacology and Therapeutics, 36 (9), 501. Lu, B. (2003). BDNF and activity dependent synaptic modulation. Learning and Memory, 10 (2), 86 98. Lu, Y., Christian, K., & Lu, B. (2008). BDNF: a key regulator for protein synthesis dependent LTP and long term memory? Neurobiology of Learning and Memory, 89 (3), 312 323. doi: 10.1016/j.nlm.2007.08.018

PAGE 267

252 Luchsinger, J. A., Cheng, D., Tang, M. X., Schupf, N., & Mayeux, R. (2012). Central obesity in the elderly is related to late onset Alzheimer disease. Alzheimer Disease and Associated Disorders, 26 (2), 101 105. doi: 10.1097/WAD.0b013e318222f0d4 Luchsinger, J. A., & Gustafson, D. R. (2009). Adiposity and Alzheimer's disease. Current Opinion in Clinical Nutrition and Metabolic Care, 12 (1), 15 21. doi: 10.1097/MCO.0b013e32831c8c71 Luchsinger, J. A., Pa tel, B., Tang, M. X., Schupf, N., & Mayeux, R. (2007). Measures of adiposity and dementia risk in elderly persons. Archives of Neurology, 64 (3), 392 398. Luchsinger, J. A., Patel, B., Tang, M. X., Schupf, N., & Mayeux, R. (2008). Body mass index, dementia and mortality in the elderly. Journal of Nutrition, Health & Aging, 12 (2), 127 131. Lukaski, H. C., Johnson, P. E., Bolonchuk, W. W., & Lykken, G. I. (1985). Assessment of fat free mass using bioelectrical impedance measurements of the human body. Ameri can Journal of Clinical Nutrition, 41 (4), 810 817. Lupien, S. J., Buss, C., Schramek, T. E., Maheu, F., & Pruessner, J. (2005a). Hormetic influence of glucocorticoids on human memory. Nonlinearity in Biology, Toxicology, Medicine, 3 (1), 23 56. doi: 10.220 1/nonlin.003.01.003 Lupien, S. J., Fiocco, A., Wan, N., Maheu, F., Lord, C., Schramek, T., & Tu, M. T. (2005b). Stress hormones and human memory function across the lifespan. Psychoneuroendocrinology, 30 (3), 225 242. doi: 10.1016/j.psyneuen.2004.08.003 Lup ien, S. J., King, S., Meaney, M. J., & McEwen, B. S. (2000). Child's stress hormone levels correlate with mother's socioeconomic status and depressive state. Biological Psychiatry, 48 (10), 976 980. Lupien, S. J., Nair, N. P., Briere, S., Maheu, F., Tu, M. T., Lemay, M., . Meaney, M. J. (1999). Increased cortisol levels and impaired cognition in human aging: implication for depression and dementia in later life. Reviews in the Neurosciences, 10 (2), 117 139. Luppino, F. S., de Wit, L. M., Bouvy, P. F., Stijnen, T., Cuijpers, P., Penninx, B. W., & Zitman, F. G. (2010). Overweight, obesity, and depression: a systematic review and meta analysis of longitudinal studies. Archives of General Psychiatry, 67 (3), 220. Lurie, A. (2011). Obstructive sleep apnea in adults: epidemiology, clinical presentation, and treatment options. Advances in Cardiology, 46 1. Maalouf, M., Rho, J. M., & Mattson, M. P. (2009). The neuroprotective properties of calorie restriction, the ketogenic diet, and ketone bodies. Brain Res R ev, 59 (2), 293 315. doi: 10.1016/j.brainresrev.2008.09.002 Mager, D. E., Wan, R., Brown, M., Cheng, A., Wareski, P., Abernethy, D. R., & Mattson, M. P. (2006). Caloric restriction and intermittent fasting alter spectral measures of heart rate

PAGE 268

253 and blood pre ssure variability in rats. FASEB Journal, 20 (6), 631 637. doi: 10.1096/fj.05 5263com Makar, T. K., Trisler, D., Sura, K. T., Sultana, S., Patel, N., & Bever, C. T. (2008). Brain derived neurotrophic factor treatment reduces inflammation and apoptosis in ex perimental allergic encephalomyelitis. Journal of the Neurological Sciences, 270 (1 2), 70 76. doi: 10.1016/j.jns.2008.02.011 Manschot, S. M., Biessels, G., de Valk, H., Algra, A., Rutten, G., van der Grond, J., & Kappelle, L. J. (2007). Metabolic and vascu lar determinants of impaired cognitive performance and abnormalities on brain magnetic resonance imaging in patients with type 2 diabetes. Diabetologia, 50 (11), 2388 2397. Marambaud, P., Zhao, H., & Davies, P. (2005). Resveratrol promotes clearance of Alz heimer's disease amyloid Journal of Biological Chemistry, 280 (45), 37377 37382. Martin, B., Mattson, M. P., & Maudsley, S. (2006). Caloric restriction and intermittent fasting: two potential diets for successful brain aging. Ageing Res Rev, 5 ( 3), 332 353. doi: 10.1016/j.arr.2006.04.002 Martin, B., Pearson, M., Kebejian, L., Golden, E., Keselman, A., Bender, M., . Mattson, M. P. (2007a). Sex dependent metabolic, neuroendocrine, and cognitive responses to dietary energy restriction and excess Endocrinology, 148 (9), 4318 4333. doi: 10.1210/en.2007 0161 Martin, C. K., Anton, S. D., Han, H., York Crowe, E., Redman, L. M., Ravussin, E., & Williamson, D. A. (2007b). Examination of cognitive function during six months of calorie restriction: result s of a randomized controlled trial. Rejuvenation Research, 10 (2), 179 190. Masoro, E. J. (2000). Caloric restriction and aging: an update. Experimental Gerontology, 35 (3), 299 305. Maswood, N., Young, J., Tilmont, E., Zhang, Z., Gash, D. M., Gerhardt, G. A., . Ingram, D. K. (2004). Caloric restriction increases neurotrophic factor levels and attenuates neurochemical and behavioral deficits in a primate model of Parkinson's disease. Proceedings of the National Academy of Sciences of the United States o f America, 101 (52), 18171 18176. doi: 10.1073/pnas.0405831102 Mattson, M. P. (2000). Apoptosis in neurodegenerative disorders. Nature Reviews Molecular Cell Biology, 1 (2), 120 130. Mattson, M. P. (2005). Energy intake, meal frequency, and health: a neurob iological perspective. Annual Review of Nutrition, 25 237 260. doi: 10.1146/annurev.nutr.25.050304.092526 Mattson, M. P. (2008). Hormesis defined. Ageing Res Rev, 7 (1), 1 7. doi: 10.1016/j.arr.2007.08.007

PAGE 269

254 Mattson, M. P., & Calabrese, E. J. (Eds.). (2010). Hormesis: A Revolution in Biology, Toxicology and Medicine New York: Springer. Mattson, M. P., & Cheng, A. (2006). Neurohormetic phytochemicals: Low dose toxins that induce adaptive neuronal stress responses. Trends in Neurosciences, 29 (11), 632 639. Ma ttson, M. P., Duan, W., & Guo, Z. (2003). Meal size and frequency affect neuronal plasticity and vulnerability to disease: cellular and molecular mechanisms. Journal of Neurochemistry, 84 (3), 417 431. Mattson, M. P., Maudsley, S., & Martin, B. (2004a). BD NF and 5 HT: a dynamic duo in age related neuronal plasticity and neurodegenerative disorders. Trends in Neurosciences, 27 (10), 589 594. Mattson, M. P., Maudsley, S., & Martin, B. (2004b). A neural signaling triumvirate that influences ageing and age rela ted disease: insulin/IGF 1, BDNF and serotonin. Ageing Res Rev, 3 (4), 445 464. doi: 10.1016/j.arr.2004.08.001 Mattson, M. P., & Wan, R. (2005). Beneficial effects of intermittent fasting and caloric restriction on the cardiovascular and cerebrovascular sys tems. J Nutr Biochem, 16 (3), 129 137. doi: 10.1016/j.jnutbio.2004.12.007 McAuley, E., Kramer, A. F., & Colcombe, S. J. (2004). Cardiovascular fitness and neurocognitive function in older adults: a brief review. Brain, Behavior, and Immunity, 18 (3), 214 220 doi: 10.1016/j.bbi.2003.12.007 McAuley, P. A., Artero, E. G., Sui, X., Lee, D. c., Church, T. S., Lavie, C. J., . Blair, S. N. (2012). The obesity paradox, cardiorespiratory fitness, and coronary heart disease. Paper presented at the Mayo Clinic Proc eedings. McAuley, P. A., Sui, X., Church, T. S., Hardin, J. W., Myers, J. N., & Blair, S. N. (2009). The joint effects of cardiorespiratory fitness and adiposity on mortality risk in men with hypertension. American Journal of Hypertension, 22 (10), 1062 106 9. McDowell, I., Xi, G., Lindsay, J., & Tierney, M. (2007). Mapping the connections between education and dementia. Journal of Clinical and Experimental Neuropsychology, 29 (2), 127 141. McEwen, B. S. (1997). Possible mechanisms for atrophy of the human h ippocampus. Molecular Psychiatry, 2 (3), 255 262. McEwen, B. S. (1998). Protective and damaging effects of stress mediators. New England Journal of Medicine, 338 (3), 171 179. doi: 10.1056/nejm199801153380307 McEwen, B. S. (2000). The neurobiology of stress : from serendipity to clinical relevance. Brain Research, 886 (1 2), 172 189.

PAGE 270

255 McEwen, B. S. (2004). Protection and damage from acute and chronic stress: allostasis and allostatic overload and relevance to the pathophysiology of psychiatric disorders. Annal s of the New York Academy of Sciences, 1032 1 7. doi: 10.1196/annals.1314.001 McEwen, B. S. (2008). Central effects of stress hormones in health and disease: Understanding the protective and damaging effects of stress and stress mediators. European Journa l of Pharmacology, 583 (2 3), 174 185. doi: 10.1016/j.ejphar.2007.11.071 McEwen, B. S., Magarinos, A. M., & Reagan, L. P. (2002). Studies of hormone action in the hippocampal formation: possible relevance to depression and diabetes. Journal of Psychosomatic Research, 53 (4), 883 890. McGaugh, J. L., & Roozendaal, B. (2002). Role of adrenal stress hormones in forming lasting memories in the brain. Current Opinion in Neurobiology, 12 (2), 205 210. McGeer, E. G., Klegeris, A., & McGeer, P. L. (2005). Inflammati on, the complement system and the diseases of aging. Neurobiology of Aging, 26 Suppl 1 94 97. doi: 10.1016/j.neurobiolaging.2005.08.008 the treatment of dem entia. Cochrane Database Syst Rev, 8 (8). McKhann, G. M., Knopman, D. S., Chertkow, H., Hyman, B. T., Jack, C. R., Jr., Kawas, C. H., . Phelps, C. H. (2011). The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Instit ute on Aging Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7 (3), 263 269. doi: 10.1016/j.jalz.2011.03.005 McNay, E. C., Fries, T. M., & Gold, P. E. (2000). Decreases in rat extracellular hippocampal glucose concentration associated with cognitive demand during a spatial task. Proceedings of the National Academy of Sciences of the United States of America, 97 (6), 2881 2885. doi: 10.1073/pnas.050583697 Means, L. W., Higgins, J., & Fernandez, T. (1993). Mid life onset of dietary restriction extends life and prolongs cognitive functioning. Physiology and Behavior, 54 (3), 503 508. hypothesis: a systematic review with meta analyses and qualitative analyses. PloS One, 7 (6), e38268. Messier, C. (2005). Impact of impaired glucose tolerance and type 2 diabetes on cognitive aging. Neurobiology of Aging, 26 Suppl 1 26 30. doi: 10.1016/j.neurobiolaging.2005.09.014 Middl eton, L. E., & Yaffe, K. (2009). Promising strategies for the prevention of dementia. Archives of Neurology, 66 (10), 1210 1215. doi: 10.1001/archneurol.2009.201

PAGE 271

256 Mielke, M. M., Zandi, P., Sjgren, M., Gustafson, D., stling, S., Steen, B., & Skoog, I. (2005 ). High total cholesterol levels in late life associated with a reduced risk of dementia. Neurology, 64 (10), 1689 1695. Miller, H. W. (1973). Plan and operation of the health and nutrition examination survey. United states -1971 1973. Vital and Health Sta tistics. Series 1: Programs and Collection Procedures (10a), 1 46. Miller, J. L., Couch, J., Schwenk, K., Long, M., Towler, S., Theriaque, D. W., . Leonard, C. M. (2009). Early childhood obesity is associated with compromised cerebellar development. De velopmental Neuropsychology, 34 (3), 272 283. doi: 10.1080/87565640802530961 Miller, S., & Wolfe, R. (2008). The danger of weight loss in the elderly. The Journal of Nutrition Health and Aging, 12 (7), 487 491. Miranda, M. J., Turner, Z., & Magrath, G. (201 2). Alternative diets to the classical ketogenic diet can we be more liberal? Epilepsy Research, 100 (3), 278 285. doi: 10.1016/j.eplepsyres.2012.06.007 Molteni, R., Wu, A., Vaynman, S., Ying, Z., Barnard, R. J., & Gomez Pinilla, F. (2004). Exercise revers es the harmful effects of consumption of a high fat diet on synaptic and behavioral plasticity associated to the action of brain derived neurotrophic factor. Neuroscience, 123 (2), 429 440. Morris, J. C. (2005). Dementia update 2005. Alzheimer Disease and Associated Disorders, 19 (2), 100 117. Morris, M. C., Evans, D. A., Bienias, J. L., Tangney, C. C., Bennett, D. A., Aggarwal, N., . Scherr, P. A. (2002). Dietary intake of antioxidant nutrients and the risk of incident Alzheimer disease in a biracial c ommunity study. JAMA: the journal of the American Medical Association, 287 (24), 3230 3237. Morris, M. C., Evans, D. A., Bienias, J. L., Tangney, C. C., Bennett, D. A., Wilson, R. S., . Schneider, J. (2003). Consumption of fish and n 3 fatty acids and risk of incident Alzheimer disease. Archives of Neurology, 60 (7), 940 946. Morris, M. C., Scherr, P. A., Hebert, L. E., Bennett, D. A., Wilson, R. S., Glynn, R. J., & Evans, D. A. (2000). The cross sectional association between blood pressure and Alzheime r's disease in a biracial community population of older persons. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 55 (3), M130 M136. Morton, N. M., Ramage, L., & Seckl, J. R. (2004). Down regulation of adipose 11beta hydroxys teroid dehydrogenase type 1 by high fat feeding in mice: a potential adaptive mechanism counteracting metabolic disease. Endocrinology, 145 (6), 2707 2712. doi: 10.1210/en.2003 1674 Mosek, A., Natour, H., Neufeld, M. Y., Shiff, Y., & Vaisman, N. (2009). Ket ogenic diet treatment in adults with refractory epilepsy: a prospective pilot study. Seizure, 18 (1), 30 33.

PAGE 272

257 Mu, J. S., Li, W. P., Yao, Z. B., & Zhou, X. F. (1999). Deprivation of endogenous brain derived neurotrophic factor results in impairment of spatia l learning and memory in adult rats. Brain Research, 835 (2), 259 265. Muller, E. E., Cella, S. G., De Gennaro Colonna, V., Parenti, M., Cocchi, D., & Locatelli, V. (1993). Aspects of the neuroendocrine control of growth hormone secretion in ageing mammals Journal of Reproduction and Fertility. Supplement, 46 99 114. Munck, A., & Naray Fejes Toth, A. (1994). Glucocorticoids and stress: permissive and suppressive actions. Annals of the New York Academy of Sciences, 746 115 130; discussion 131 113. Nader ali, E. K., Ratcliffe, S. H., & Dale, M. C. (2009). Obesity and Alzheimer's disease: a link between body weight and cognitive function in old age. American Journal of Alzheimer's Disease and Other Dementias, 24 (6), 445 449. doi: 10.1177/1533317509348208 Na m, S. Y., Lee, E. J., Kim, K. R., Cha, B. S., Song, Y. D., Lim, S. K., . Huh, K. B. (1997). Effect of obesity on total and free insulin like growth factor (IGF) 1, and their relationship to IGF binding protein (BP) 1, IGFBP 2, IGFBP 3, insulin, and gro wth hormone. International Journal of Obesity and Related Metabolic Disorders, 21 (5), 355 359. Naylor, A. S., Bull, C., Nilsson, M. K. L., Zhu, C., Bjork Eriksson, T., Eriksson, P. S., . Kuhn, H. G. (2008). Voluntary running rescues adult hippocampal neurogenesis after irradiation of the young mouse brain. Proceedings of the National Academy of Sciences of the United States of America, 105 (38), 14632 14637. Newman, A. B., Arnold, A. M., Sachs, M. C., Ives, D. G., Cushman, M., Strotmeyer, E. S., . Robbins, J. (2009). Long term function in an older cohort -the cardiovascular health study all stars study. Journal of the American Geriatrics Society, 57 (3), 432 440. doi: 10.1111/j.1532 5415.2008.02152.x Newman, W. P., & Brodows, R. G. (1983). Insulin ac tion during acute starvation: evidence for selective insulin resistance in normal man. Metabolism: Clinical and Experimental, 32 (6), 590 596. Newson, R. S., & Kemps, E. B. (2006). Cardiorespiratory fitness as a predictor of successful cognitive ageing. Jo urnal of Clinical and Experimental Neuropsychology, 28 (6), 949 967. doi: 10.1080/13803390591004356 Nguyen, N. T., Magno, C. P., Lane, K. T., Hinojosa, M. W., & Lane, J. S. (2008). Association of hypertension, diabetes, dyslipidemia, and metabolic syndrome with obesity: findings from the National Health and Nutrition Examination Survey, 1999 to 2004. Journal of the American College of Surgeons, 207 (6), 928 934. Nourhashemi, F., Deschamps, V., Larrieu, S., Letenneur, L., Dartigues, J., & Barberger Gateau, P. (2003). Body mass index and incidence of dementia: The PAQUID study. Neurology, 60 (1), 117 119.

PAGE 273

258 Nunomura, A., Moreira, P. I., Castellani, R. J., Lee, H. G., Zhu, X., Smith, M. A., & Perry, G. (2012). Oxidative damage to RNA in aging and neurodegenerative disorders. Neurotoxicity Research, 22 (3), 231 248. doi: 10.1007/s12640 012 9331 x O'Callaghan, R. M., Ohle, R., & Kelly, A. M. (2007). The effects of forced exercise on hippocampal plasticity in the rat: A comparison of LTP, spatial and non spatial learn ing. Behavioural Brain Research, 176 (2), 362 366. Ogunniyi, A., Gao, S., Unverzagt, F. W., Baiyewu, O., Gureje, O., Nguyen, J. T., . Hendrie, H. C. (2011). Weight loss and incident dementia in elderly Yoruba Nigerians: A 10 year follow up study. Inter national Psychogeriatrics, 23 (3), 387 394. doi: 10.1017/S1041610210001390 Onen, F., & Onen, H. (2010). Obstructive sleep apnea and cognitive impairment in the elderly]. Psychologie & Neuropsychiatrie du Vieillissement, 8 (3), 163. Ostrosky Solis, F., Mendo za, V. U., & Ardila, A. (2001). Neuropsychological profile of patients with primary systemic hypertension. International Journal of Neuroscience, 11 159 172. Owen, O. E., Morgan, A. P., Kemp, H. G., Sullivan, J. M., Herrera, M. G., & Cahill, G. F. (1967) Brain metabolism during fasting. Journal of Clinical Investigation, 46 (10), 1589 1595. Pacini, G., & Bergman, R. N. (1986). MINMOD: a computer program to calculate insulin sensitivity and pancreatic responsivity from the frequently sampled intravenous g lucose tolerance test. Computer Methods and Programs in Biomedicine, 23 (2), 113 122. Pan, W., Banks, W. A., Fasold, M. B., Bluth, J., & Kastin, A. J. (1998). Transport of brain derived neurotrophic factor across the blood brain barrier. Neuropharmacology, 37 (12), 1553 1561. Parekh, N., Roberts, C. B., Vadiveloo, M., Puvananayagam, T., Albu, J. B., & Lu Yao, G. L. (2010). Lifestyle, anthropometric, and obesity related physiologic determinants of insulin like growth factor 1 in the Third National Health and Nutrition Examination Survey (1988 1994). Annals of Epidemiology, 20 (3), 182 193. doi: 10.1016/j.annepidem.2009.11.008 Pasquali, R., Cantobelli, S., Casimirri, F., Capelli, M., Bortoluzzi, L., Flamia, R., . Barbara, L. (1993). The hypothalamic pituita ry adrenal axis in obese women with different patterns of body fat distribution. Journal of Clinical Endocrinology and Metabolism, 77 (2), 341 346. Pavelic, J., Matijevic, T., & Knezevic, J. (2007). Biological & physiological aspects of action of insulin l ike growth factor peptide family. Indian Journal of Medical Research, 125 (4), 511 522. Pavlik, V. N., Hyman, D. J., & Doody, R. (2004). Cardiovascular risk factors and cognitive function in adults 30 59 years of age (NHANES III). Neuroepidemiology, 24 (1 2 ), 42 50.

PAGE 274

259 Pawlikowska, L., Hu, D., Huntsman, S., Sung, A., Chu, C., Chen, J., . Ziv, E. (2009). Association of common genetic variation in the insulin/IGF1 signaling pathway with human longevity. Aging Cell, 8 (4), 460 472. doi: 10.1111/j.1474 9726.200 9.00493.x Pedersen, W. A., Culmsee, C., Ziegler, D., Herman, J. P., & Mattson, M. P. (1999). Aberrant stress response associated with severe hypoglycemia in a transgenic mouse model of Journal of Molecular Neuroscience, 13 (1 2), 159 16 5. Pelleymounter, M. A., Cullen, M. J., & Wellman, C. L. (1995). Characteristics of BDNF induced weight loss. Experimental Neurology, 131 (2), 229 238. Peskind, E., Wilkinson, C., Petrie, E., Schellenberg, G., & Raskind, M. (2001). Increased CSF cortisol in AD is a function of APOE genotype. Neurology, 56 (8), 1094 1098. Petrovitch, H., White, L., Izmirilian, G., Ross, G., Havlik, R., Markesbery, W., . Foley, D. (2000). Midlife blood pressure and neuritic plaques, neurofibrillary tangles, and brain wei ght at death: the HAAS< sup> . Neurobiology of Aging, 21 (1), 57 62. Pillar, G., & Shehadeh, N. (2008). Abdominal Fat and Sleep Apnea The chicken or the egg? Diabetes Care, 31 (Supplement 2), S303 S309. Plassman, B. L., Williams, J. W., Jr., Burke, J R., Holsinger, T., & Benjamin, S. (2010). Systematic review: factors associated with risk for and possible prevention of cognitive decline in later life. Annals of Internal Medicine, 153 (3), 182 193. doi: 10.1059/0003 4819 153 3 201008030 00258 Ploughman M. (2008). Exercise is brain food: the effects of physical activity on cognitive function. Developmental Neurorehabilitation, 11 (3), 236 240. Poduslo, J. F., & Curran, G. L. (1996). Permeability at the blood brain and blood nerve barriers of the neurotr ophic factors: NGF, CNTF, NT 3, BDNF. Brain Research: Molecular Brain Research, 36 (2), 280 286. Pollitt, E., Cueto, S., & Jacoby, E. R. (1998). Fasting and cognition in well and undernourished schoolchildren: a review of three experimental studies. Ameri can Journal of Clinical Nutrition, 67 (4), 779S 784S. Pollitt, E., Lewis, N. L., Garza, C., & Shulman, R. J. (1982). Fasting and cognitive function. Journal of Psychiatric Research, 17 (2), 169 174. Popovic, V., & Duntas, L. H. (2005). Brain somatic cross talk: ghrelin, leptin and ultimate challengers of obesity. Nutritional Neuroscience, 8 (1), 1 5. Power, B. D., Alfonso, H., Flicker, L., Hankey, G. J., Yeap, B. B., & Almeida, O. P. (2011). Body adiposity in later life and the incidence of dementia: The he alth in men study. PloS One, 6 (3). doi: 10.1371/journal.pone.0017902

PAGE 275

260 Power, C., & Hertzman, C. (1997). Social and biological pathways linking early life and adult disease. British Medical Bulletin, 53 (1), 210 221. Qiu, C., Winblad, B., & Fratiglioni, L. (2009). Low diastolic pressure and risk of dementia in very old people: a longitudinal study. Dementia and Geriatric Cognitive Disorders, 28 (3), 213 219. Radak, Z., Kaneko, T., Tahara, S., Nakamoto, H., Pucsok, J., Sasvari, M., . Goto, S. (2001). Reg ular exercise improves cognitive function and decreases oxidative damage in rat brain. Neurochemistry International, 38 (1), 17 23. Radloff, L. S. (1977). The CES D Scale: A Self Report Depression Scale for Research in the General Population. Applied Psych ological Measurement, 1 (3), 385 401. doi: 10.1177/014662167700100306 Raffaitin, C., Feart, C., Le Goff, M., Amieva, H., Helmer, C., Akbaraly, T., . Barberger Gateau, P. (2011). Metabolic syndrome and cognitive decline in French elders: The Three City S tudy. Neurology, 76 (6), 518 525. doi: 10.1212/WNL.0b013e31820b7656 Raffaitin, C., Gin, H., Empa na, J. P., Helmer, C., Berr, C., Tzourio, C., . Barberger Gateau, P. (2009). Metabolic syndrome and risk for incident Alzheimer's disease or vascular dementia: the Three City Study. Diabetes Care, 32 (1), 169 174. doi: 10.2337/dc08 0272 Rasmuson, S., Na sman, B., Carlstrom, K., & Olsson, T. (2002). Increased levels of adrenocortical and gonadal hormones in mild to moderate Alzheimer's disease. Dementia and Geriatric Cognitive Disorders, 13 (2), 74 79. Rattan, S. I. (2008). Hormesis in aging. Ageing Resear ch Reviews, 7 (1), 63 78. doi: 10.1016/j.arr.2007.03.002 Razay, G., Williams, J., King, E., Smith, A. D., & Wilcock, G. (2009). Blood pressure, dementia and Dementia and Geriatric Cognitive Disorders, 28 (1 ), 70 74. Rea, T. D., Breitner, J. C., Psaty, B. M., Fitzpatrick, A. L., Lopez, O. L., Newman, A. B., . Lyketsos, C. G. (2005). Statin use and the risk of incident dementia: the Cardiovascular Health Study. Archives of Neurology, 62 (7), 1047. Regal, P., & Heatherington, E. (2012). Body mass index and cognition. Journal of the American Geriatrics Society, 60 (7), 1386 1387. doi: 10.1111/j.1532 5415.2012.04013.x Reitz, C., Luchsinger, J., Tang, M. X., Manly, J., & Mayeux, R. (2005). Impact of plasma lipi ds and time on memory performance in healthy elderly without dementia. Neurology, 64 (8), 1378 1383. doi: 10.1212/01.wnl.0000158274.31318.3c Reitz, C., Tang, M. X., Manly, J., Schupf, N., Mayeux, R., & Luchsinger, J. A. (2008). Plasma lipid levels in the el derly are not associated with the risk of mild cognitive impairment. Dementia and Geriatric Cognitive Disorders, 25 (3), 232 237.

PAGE 276

261 Renvall, M. J., Spindler, A. A., Nichols, J. F., & Ramsdell, J. W. (1993). Body composition of patients with Alzheimer's disea se. Journal of the American Dietetic Association, 93 (1), 47 52. Requejo, A., Ortega, R., Robles, F., Navia, B., Faci, M., & Aparicio, A. (2003). Influence of nutrition on cognitive function in a group of elderly, independently living people. European Jour nal of Clinical Nutrition, 57 S54 S57. Resende, R., Ferreiro, E., Pereira, C., & Oliveira, C. R. (2008). ER stress is involved in Abeta induced GSK 3beta activation and tau phosphorylation. Journal of Neuroscience Research, 86 (9), 2091 2099. doi: 10.1002 /jnr.21648 Richards, M., Hardy, R., & Wadsworth, M. E. J. (2003). Does active leisure protect cognition? Evidence from a national birth cohort. Social Science and Medicine, 56 (4), 785 792. Riley, K. P., Snowdon, D. A., Desrosiers, M. F., & Markesbery, W. R. (2005). Early life linguistic ability, late life cognitive function, and neuropathology: findings from the Nun Study. Neurobiology of Aging, 26 (3), 341 347. Rivenes, A. C., Harvey, S. B., & Mykletun, A. (2009). The relationship between abdominal fat, o besity, and common mental disorders: results from the HUNT study. Journal of Psychosomatic Research, 66 (4), 269. Rivera, E. J., Goldin, A., Fulmer, N., Tavares, R., Wands, J. R., & de la Monte, S. M. (2005). Insulin and insulin like growth factor expressi on and function deteriorate with progression of Alzheimer's disease: link to brain reductions in acetylcholine. Journal of Alzheimer's Disease, 8 (3), 247 268. Roberge, C., Carpentier, A. C., Langlois, M. F., Baillargeon, J. P., Ardilouze, J. L., Maheux, P ., & Gallo Payet, N. (2007). Adrenocortical dysregulation as a major player in insulin resistance and onset of obesity. American Journal of Physiology Endocrinology And Metabolism, 293 (6), E1465 E1478. Roe, C. M., Xiong, C., Miller, J. P., & Morris, J. C. (2007). Education and Alzheimer disease without dementia support for the cognitive reserve hypothesis. Neurology, 68 (3), 223 228. Rojas Vega, S., Knicker, A., Hollmann, W., Bloch, W., & Struder, H. K. (2010). Effect of resistance exercise on serum levels of growth factors in humans. Hormone and Metabolic Research, 42 (13), 982 986. doi: 10.1055/s 0030 1267950 Roky, R., Chapotot, F., Benchekroun, M. T., Benaji, B., Hakkou, F., Elkhalifi, H., & Buguet, A. (2003). Daytime sleepiness during Ramadan intermitten t fasting: polysomnographic and quantitative waking EEG study. Journal of Sleep Research, 12 (2), 95 101. Roky, R., Houti, I., Moussamih, S., Qotbi, S., & Aadil, N. (2004). Physiological and chronobiological changes during Ramadan intermittent fasting. Ann als of Nutrition and Metabolism, 48 (4), 296 303. doi: 10.1159/000081076

PAGE 277

262 Roky, R., Iraki, L., HajKhlifa, R., Lakhdar Ghazal, N., & Hakkou, F. (2000). Daytime alertness, mood, psychomotor performances, and oral temperature during Ramadan intermittent fasting Annals of Nutrition and Metabolism, 44 (3), 101 107. doi: 12830 Rollero, A., Murialdo, G., Fonzi, S., Garrone, S., Gianelli, M. V., Gazzerro, E., . Polleri, A. (1998). Relationship between cognitive function, growth hormone and insulin like growth fac tor I plasma levels in aged subjects. Neuropsychobiology, 38 (2), 73 79. Rosenberg, R., & Doghramji, P. (2009). Optimal treatment of obstructive sleep apnea and excessive sleepiness. Advances in Therapy, 26 (3), 295 312. Rosengren, A., Skoog, I., Gustafson D., & Wilhelmsen, L. (2005). Body mass index, other cardiovascular risk factors, and hospitalization for dementia. Archives of Internal Medicine, 165 (3), 321 326. Roth, G. S., Ingram, D. K., & Lane, M. A. (1999). Calorie restriction in primates: will it work and how will we know? Journal of the American Geriatrics Society, 47 (7), 896. Roth, G. S., Ingram, D. K., & Lane, M. A. (2001). Caloric restriction in primates and relevance to humans. Annals of the New York Academy of Sciences, 928 (1), 305 315. RO TH, G. S., LANE, M. A., & INGRAM, D. K. (2005). Caloric restriction mimetics: the next phase. Annals of the New York Academy of Sciences, 1057 (1), 365 371. Roth, H. L. (2012). Dementia and sleep. Neurologic Clinics, 30 (4), 1213 1248. Runchey, S. S., Poll ak, M. N., Valsta, L. M., Coronado, G. D., Schwarz, Y., Breymeyer, K. L., . Neuhouser, M. L. (2012). Glycemic load effect on fasting and post prandial serum glucose, insulin, IGF 1 and IGFBP 3 in a randomized, controlled feeding study. European Journal of Clinical Nutrition, 66 (10), 1146 1152. doi: 10.1038/ejcn.2012.107 Runcie, J., & Thomson, T. (1970). Prolonged starvation a dangerous procedure. British Medical Journal, 3 (5720), 432. Ruscheweyh, R., Willemer, C., Kruger, K., Duning, T., Warnecke, T., Sommer, J., . Floel, A. (2011). Physical activity and memory functions: an interventional study. Neurobiology of Aging, 32 (7), 1304 1319. doi: 10.1016/j.neurobiolaging.2009.08.001 Ryan, C. (2005). Diabetes, aging, and cognitive decline. Neurobiology of Aging, 26 (S1), 21 25. Sabia, S., Kivimaki, M., Shipley, M. J., Marmot, M. G., & Singh Manoux, A. (2009). Body mass index over the adult life course and cognition in late midlife: the Whitehall II Cohort Study. American Journal of Clinical Nutrition, 89 (2), 601 607. doi: 10.3945/ajcn.2008.26482 Sachs Ericsson, N. J., Sawyer, K. A., Corsentino, E. A., Collins, N. A., & Blazer, D. G. (2010). APOE 4 allele carriers: Biological, psychological, and social variables associated with cognitive impairment. Aging & Mental Health, 14 (6), 679 691. doi: 10.1080/13607860903292594

PAGE 278

263 Sahu, A. (2003). Leptin signaling in the hypothalamus: emphasis on energy homeostasis and leptin resistance. Frontiers in Neuroendocrinology, 24 (4), 225 253. Salehi, M., Ferenczi, A., & Zumof f, B. (2005). Obesity and cortisol status. Hormone and Metabolic Research, 37 (4), 193 197. doi: 10.1055/s 2005 861374 Sano, M., Bell, K., Galasko, D., Galvin, J., Thomas, R., van Dyck, C., & Aisen, P. (2011). A randomized, double blind, placebo controlled trial of simvastatin to treat Alzheimer disease. Neurology, 77 (6), 556 563. Sarzani, R., Salvi, F., Dess Fulgheri, P., & Rappelli, A. (2008). Renin angiotensin system, natriuretic peptides, obesity, metabolic syndrome, and hypertension: an integrated vie w in humans. Journal of Hypertension, 26 (5), 831 843. Scarmeas, N., Levy, G., Tang, M. X., Manly, J., & Stern, Y. (2001). Influence of leisure activity on the incidence of Alzheimer's disease. Neurology, 57 (12), 2236 2242. Scarmeas, N., Stern, Y., Mayeux R., & Luchsinger, J. A. (2006). Mediterranean diet, Alzheimer disease, and vascular mediation. Archives of Neurology, 63 (12), 1709. Scarpace, P. J., & Zhang, Y. (2009). Leptin resistance: a prediposing factor for diet induced obesity. American Journal o f Physiology: Regulatory, Integrative and Comparative Physiology, 296 (3), R493 500. doi: 10.1152/ajpregu.90669.2008 Seeman, T. E., McEwen, B. S., Singer, B. H., Albert, M. S., & Rowe, J. W. (1997). Increase in urinary cortisol excretion and memory declines : MacArthur studies of successful aging. Journal of Clinical Endocrinology and Metabolism, 82 (8), 2458 2465. Sevigny, J. J., Ryan, J. M., van Dyck, C. H., Peng, Y., Lines, C. R., & Nessly, M. L. (2008). Growth hormone secretagogue MK 677: no clinical effe ct on AD progression in a randomized trial. Neurology, 71 (21), 1702 1708. doi: 10.1212/01.wnl.0000335163.88054.e7 Shapiro, A., Mu, W., Roncal, C., Cheng, K. Y., Johnson, R. J., & Scarpace, P. J. (2008). Fructose induced leptin resistance exacerbates weight gain in response to subsequent high fat feeding. American Journal of Physiology Regulatory, Integrative and Comparative Physiology, 295 (5), R1370 R1375. Shapiro, A., Tmer, N., Gao, Y., Cheng, K. Y., & Scarpace, P. J. (2011). Prevention and reversal of d iet induced leptin resistance with a sugar free diet despite high fat content. British Journal of Nutrition, 106 (3), 390 397. Shirayama, Y., Chen, A. C. H., Nakagawa, S., Russell, D. S., & Duman, R. S. (2002). Brain derived neurotrophic factor produces an tidepressant effects in behavioral models of depression. Journal of Neuroscience, 22 (8), 3251 3261. Siervo, M., Nasti, G., Stephan, B. C., Papa, A., Muscariello, E., Wells, J. C., . Colantuoni, A. (2012). Effects of intentional weight loss on physical and cognitive function in middle

PAGE 279

264 aged and older obese participants: a pilot study. Journal of the American College of Nutrition, 31 (2), 79 86. Singh Manoux, A., Czernichow, S., Elbaz, A., Dugravot, A., Sabia, S., Hagger Johnson, G., . Kivimaki, M. (2 012). Obesity phenotypes in midlife and cognition in early old age: The Whitehall II cohort study. Neurology, 79 (8), 755 762. doi: 10.1212/WNL.0b013e3182661f63 Sinz, H., Zamarian, L., Benke, T., Wenning, G., & Delazer, M. (2008). Impact of ambiguity and ri sk on decision making in mild Alzheimer's disease. Neuropsychologia, 46 (7), 2043 2055. Smith, P. J., Blumenthal, J. A., Babyak, M. A., Craighead, L., Welsh Bohmer, K. A., Browndyke, J. N., . Sherwood, A. (2010a). Effects of the dietary approaches to s top hypertension diet, exercise, and caloric restriction on neurocognition in overweight adults with high blood pressure. Hypertension, 55 (6), 1331 1338. doi: 10.1161/HYPERTENSIONAHA.109.146795 Smith, P. J., Blumenthal, J. A., Hoffman, B. M., Cooper, H., S trauman, T. A., Welsh Bohmer, K., . Sherwood, A. (2010b). Aerobic exercise and neurocognitive performance: a meta analytic review of randomized controlled trials. Psychosomatic Medicine, 72 (3), 239 252. doi: 10.1097/PSY.0b013e3181d14633 Snowden, M., St einman, L., Mochan, K., Grodstein, F., Prohaska, T. R., Thurman, D. J., . Anderson, L. A. (2011). Effect of exercise on cognitive performance in community dwelling older adults: review of intervention trials and recommendations for public health practi ce and research. Journal of the American Geriatrics Society, 59 (4), 704 716. doi: 10.1111/j.1532 5415.2011.03323.x Soerensen, M., Dato, S., Christensen, K., McGue, M., Stevnsner, T., Bohr, V. A., & Christiansen, L. (2010). Replication of an association of variation in the FOXO3A gene with human longevity using both case control and longitudinal data. Aging Cell, 9 (6), 1010 1017. doi: 10.1111/j.1474 9726.2010.00627.x Sogawa, H., & Kubo, C. (2000). Influence of short term repeated fasting on the longevity of female (NZB NZW) F1 mice. Mechanisms of Ageing and Development, 115 (1), 61 71. Sohal, R. S., & Weindruch, R. (1996). Oxidative stress, caloric restriction, and aging. Science, 273 (5271), 59 63. Solfrizzi, V., D'Introno, A., Colacicco, A. M., Capurso, C. Del Parigi, A., Capurso, S., . Panza, F. (2005). Dietary fatty acids intake: possible role in cognitive decline and dementia. Experimental Gerontology, 40 (4), 257 270. doi: 10.1016/j.exger.2005.01.001 Solfrizzi, V., Panza, F., & Capurso, A. (2003). T he role of diet in cognitive decline. Journal of Neural Transmission, 110 (1), 95 110. Solomon, A., Kareholt, I., Ngandu, T., Winblad, B., Nissinen, A., Tuomilehto, J., . Kivipelto, M. (2007a). Serum cholesterol changes after midlife and late life cogn ition: twenty one

PAGE 280

265 year follow up study. Neurology, 68 (10), 751 756. doi: 10.1212/01.wnl.0000256368.57375.b7 Solomon, A., Kreholt, I., Ngandu, T., Winblad, B., Nissinen, A., Tuomilehto, J., . Kivipelto, M. (2007b). Serum cholesterol changes after midli fe and late life cognition Twenty one year follow up study. Neurology, 68 (10), 751 756. Sonntag, W. E., Ramsey, M., & Carter, C. S. (2005). Growth hormone and insulin like growth factor 1 (IGF 1) and their influence on cognitive aging. Ageing Res Rev, 4 (2 ), 195 212. doi: 10.1016/j.arr.2005.02.001 Sperling, R. A., Aisen, P. S., Beckett, L. A., Bennett, D. A., Craft, S., Fagan, A. M., . Phelps, C. H. (2011). Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement, 7 (3), 280 292. doi: 10.1016/j.jalz.2011.03.003 Spitzer, R., Yanovski, S., & Marcus, M. (1993). The questionnaire on eating and weight patterns revised (QEWP R). New York: New York State Psychiatric Institute Squier, T. C. (2001). Oxidative stress and protein aggregation during biological aging. Experimental Gerontology, 36 (9), 1539 1550. Steen, E., Terry, B. M., Rivera, E. J., Cannon J. L., Neely, T. R., Tavares, R., . de la Monte, S. M. (2005). Impaired insulin and insulin like growth factor expression and signaling mechanisms in Alzheimer's disease -is this type 3 diabetes? Journal of Alzheimer's Disease, 7 (1), 63 80. Stein, C ., & Moritz, I. (1999). A life course perspective of maintaining independence in older age. World Health Organization Stern, Y. (2002). What is Cognitive Reserve? Theory and research application of the reserve concept,. Journal of the International Neuro psychological Society, 8 448 460. Stern, Y. (2009). Cognitive reserve. Neuropsychologia, 47 (10), 2015 2028. doi: 10.1016/j.neuropsychologia.2009.03.004 Stewart, J., Mitchell, J., & Kalant, N. (1989). The effects of life long food restriction on spatial m emory in young and aged Fischer 344 rats measured in the eight arm radial and the Morris water mazes. Neurobiology of Aging, 10 (6), 669 675. Stewart, R., Masaki, K., Xue, Q. L., Peila, R., Petrovitch, H., White, L. R., & Launer, L. J. (2005). A 32 Year Pr ospective Study of Change in Body Weight and Incident Dementia. Archives of Neurology, 62 (1), 55 60. doi: 10.1001/archneur.62.1.55 Stote, K. S., Baer, D. J., Spears, K., Paul, D. R., Harris, G. K., Rumpler, W. V., . Mattson, M. P. (2007). A controlled trial of reduced meal frequency without caloric restriction in

PAGE 281

266 healthy, normal weight, middle aged adults. American Journal of Clinical Nutrition, 85 (4), 981 988. Strachan, M. W., Deary, I. J., Ewing, F. M., & Frier, B. M. (1997). Is type II diabetes asso ciated with an increased risk of cognitive dysfunction? A critical review of published studies. Diabetes Care, 20 (3), 438 445. Stranahan, A. M., Lee, K., Martin, B., Maudsley, S., Golden, E., Cutler, R. G., & Mattson, M. P. (2009a). Voluntary exercise and caloric restriction enhance hippocampal dendritic spine density and BDNF levels in diabetic mice. Hippocampus, 19 (10), 951 961. Stranahan, A. M., Zhou, Y., Martin, B., & Maudsley, S. (2009b). Pharmacomimetics of exercise: novel approaches for hippocampal ly targeted neuroprotective agents. Current Medicinal Chemistry, 16 (35), 4668 4678. Strand, B. H., Langballe, E. M., Hjellvik, V., Handal, M., Naess, O., Knudsen, G. P., . Bjertness, E. (2013). Midlife vascular risk factors and their association with dementia deaths: Results from a Norwegian prospective study followed up for 35 years. Journal of the Neurological Sciences, 324 (1 2), 124 130. doi: 10.1016/j.jns.2012.10.018 Strawbridge, W. J., Deleger, S., Roberts, R. E., & Kaplan, G. A. (2002). Physical activity reduces the risk of subsequent depression for older adults. American Journal of Epidemiology, 156 (4), 328 334. Sturman, M. T., de Leon, C. F., Bienias, J. L., Morris, M. C., Wilson, R. S., & Evans, D. A. (2008). Body mass index and cognitive decl ine in a biracial community population. Neurology, 70 (5), 360 367. Suhr, J. A., Patterson, S. M., Austin, A. W., & Heffner, K. L. (2010). The relation of hydration status to declarative memory and working memory in older adults. Journal of Nutrition, Heal th & Aging, 14 (10), 840 843. Suhr, J. A., Stewart, J. C., & France, C. R. (2004). The relationship between blood pressure and cognitive performance in the Third National Health and Nutrition Examination Survey (NHANES III). Psychosomatic Medicine, 66 (3), 291 297. Sultana, R., & Butterfield, D. A. (2008). Redox proteomics studies of in vivo amyloid beta peptide animal models of Alzheimer's disease: Insight into the role of oxidative stress. Proteomics: Clinical Applications, 2 (5), 685 696. doi: 10.1002/prc a.200780024 Suwa, M., Kishimoto, H., Nofuji, Y., Nakano, H., Sasaki, H., Radak, Z., & Kumagai, S. (2006). Serum brain derived neurotrophic factor level is increased and associated with obesity in newly diagnosed female patients with type 2 diabetes mellitu s. Metabolism: Clinical and Experimental, 55 (7), 852 857. doi: 10.1016/j.metabol.2006.02.012 Talbot, K., Wang, H. Y., Kazi, H., Han, L. Y., Bakshi, K. P., Stucky, A., . Arnold, S. E. (2012). Demonstrated brain insulin resistance in Alzheimer's disease patients is associated with

PAGE 282

267 IGF 1 resistance, IRS 1 dysregulation, and cognitive decline. Journal of Clinical Investigation, 122 (4), 1316 1338. doi: 10.1172/jci59903 Taliaz, D., Loya, A., Gersner, R., Haramati, S., Chen, A., & Zangen, A. (2011). Resilience to chronic stress is mediated by hippocampal brain derived neurotrophic factor. Journal of Neuroscience, 31 (12), 4475 4483. doi: 10.1523/JNEUROSCI.5725 10.2011 Tapia Arancibia, L., Rage, F., Givalois, L., & Arancibia, S. (2004). Physiology of BDNF: focus on hypothalamic function. Frontiers in Neuroendocrinology, 25 (2), 77 107. doi: 10.1016/j.yfrne.2004.04.001 Tarkowski, E., Andreasen, N., Tarkowski, A., & Blennow, K. (2003). Intrathecal inflammation precedes development of Alzheimer's disease. Journal of N eurology, Neurosurgery and Psychiatry, 74 (9), 1200 1205. Taylor Tavares, J. V., Clark, L., Cannon, D. M., Erickson, K., Drevets, W. C., & Sahakian, B. J. (2007). Distinct profiles of neurocognitive function in unmedicated unipolar depression and bipolar I I depression. Biological Psychiatry, 62 (8), 917 924. Tazearslan, C., Huang, J., Barzilai, N., & Suh, Y. (2011). Impaired IGF1R signaling in cells expressing longevity associated human IGF1R alleles. Aging Cell, 10 (3), 551 554. doi: 10.1111/j.1474 9726.201 1.00697.x Texel, S. J., & Mattson, M. P. (2011). Impaired adaptive cellular responses to oxidative stress and the pathogenesis of Alzheimer's disease. Antioxid Redox Signal, 14 (8), 1519 1534. doi: 10.1089/ars.2010.3569 Thilers, P. P., Macdonald, S. W., Nil sson, L. G., & Herlitz, A. (2010). Accelerated postmenopausal cognitive decline is restricted to women with normal BMI: longitudinal evidence from the Betula project. Psychoneuroendocrinology, 35 (4), 516 524. Tian, H. H., Aziz, A. R., Png, W., Wahid, M. F ., Yeo, D., & Constance Png, A. L. (2011). Effects of fasting during ramadan month on cognitive function in muslim athletes. Asian Journal of Sports Medicine, 2 (3), 145 153. Tillerson, J. L., Caudle, W. M., Reveron, M. E., & Miller, G. W. (2003). Exercise induces behavioral recovery and attenuates neurochemical deficits in rodent models of Parkinson's disease. Neuroscience, 119 (3), 899 911. Toogood, A. A., O'Neill, P. A., & Shalet, S. M. (1996). Beyond the somatopause: growth hormone deficiency in adults over the age of 60 years. Journal of Clinical Endocrinology and Metabolism, 81 (2), 460 465. Trivedi, M. H., Greer, T. L., Grannemann, B. D., Chambliss, H. O., & Jordan, A. N. (2006). Exercise as an augmentation strategy for treatment of major depression. Journal of Psychiatric Practice, 12 (4), 205 213.

PAGE 283

268 Tudor Locke, C., Brashear, M. M., Johnson, W. D., & Katzmarzyk, P. T. (2010). Accelerometer profiles of physical activity and inactivity in normal weight, overweight, and obese US men and women. Int J Behav Nutr Phys Act, 7 (1), 60. Vaishnavi, S. N., Vlassenko, A. G., Rundle, M. M., Snyder, A. Z., Mintun, M. A., & Raichle, M. E. (2010). Regional aerobic glycolysis in the human brain. Proceedings of the National Academy of Sciences, 107 (41), 17757 17762. van den Berg, E., Dekker, J. M., Nijpels, G., Kessels, R. P., Kappelle, L. J., de Haan, E. H., . Biessels, G. J. (2008). Cognitive functioning in elderly persons with type 2 diabetes and metabolic syndrome: the Hoorn study. Dementia and Geriatric Cognitiv e Disorders, 26 (3), 261 269. doi: 10.1159/000160959 van den Berg, E., Kloppenborg, R. P., Kessels, R. P., Kappelle, L. J., & Biessels, G. J. (2009). Type 2 diabetes mellitus, hypertension, dyslipidemia and obesity: A systematic comparison of their impact o n cognition. Biochimica et Biophysica Acta, 1792 (5), 470 481. doi: 10.1016/j.bbadis.2008.09.004 van Praag, H., Christie, B. R., Sejnowski, T. J., & Gage, F. H. (1999). Running enhances neurogenesis, learning, and long term potentiation in mice. Proceedings of the National Academy of Sciences of the United States of America, 96 (23), 13427 13431. van Praag, H., Kempermann, G., & Gage, F. H. (1999). Running increases cell proliferation and neurogenesis in the adult mouse dentate gyrus. Nature Neuroscience, 2 ( 3), 266 270. doi: 10.1038/6368 van Praag, H., Shubert, T., Zhao, C., & Gage, F. H. (2005). Exercise enhances learning and hippocampal neurogenesis in aged mice. Journal of Neuroscience, 25 (38), 8680 8685. doi: 10.1523/jneurosci.1731 05.2005 Vanhanen, M., K oivisto, K., Karjalainen, L., Helkala, E. L., Laakso, M., Soininen, H., & Riekkinen, P., Sr. (1997). Risk for non insulin dependent diabetes in the normoglycaemic elderly is associated with impaired cognitive function. Neuroreport, 8 (6), 1527 1530. Vanhan en, M., Koivisto, K., Moilanen, L., Helkala, E. L., Hanninen, T., Soininen, H., . Kuusisto, J. (2006). Association of metabolic syndrome with Alzheimer disease: a population based study. Neurology, 67 (5), 843 847. Varady, K. A., & Hellerstein, M. K. ( 2007). Alternate day fasting and chronic disease prevention: a review of human and animal trials. American Journal of Clinical Nutrition, 86 (1), 7 13. Vasselli, J. R., Scarpace, P. J., Harris, R. B., & Banks, W. A. (2013). Dietary Components in the Develo pment of Leptin Resistance. Advances in Nutrition: An International Review Journal, 4 (2), 164 175. Vaynman, S., & Gomez Pinilla, F. (2006). Revenge of the "sit": how lifestyle impacts neuronal and cognitive health through molecular systems that interface energy metabolism with

PAGE 284

269 neuronal plasticity. Journal of Neuroscience Research, 84 (4), 699 715. doi: 10.1002/jnr.20979 Vaynman, S. S., Ying, Z., Yin, D., & Gomez Pinilla, F. (2006). Exercise differentially regulates synaptic proteins associated to the functi on of BDNF. Brain Research, 1070 (1), 124 130. Venkateshappa, C., Harish, G., Mahadevan, A., Srinivas Bharath, M. M., & Shankar, S. K. (2012). Elevated oxidative stress and decreased antioxidant function in the human hippocampus and frontal cortex with inc reasing age: implications for neurodegeneration in Alzheimer's disease. Neurochemical Research, 37 (8), 1601 1614. doi: 10.1007/s11064 012 0755 8 Voss, M. W., Erickson, K. I., Prakash, R. S., Chaddock, L., Kim, J. S., Alves, H., . Kramer, A. F. (2013). Neurobiological markers of exercise related brain plasticity in older adults. Brain, Behavior, and Immunity, 28 90 99. Wadsworth, M. (1997). Health inequalities in the life course perspective. Social Science and Medicine, 44 (6), 859 869. Wan, R., Camand ola, S., & Mattson, M. P. (2003). Intermittent fasting and dietary supplementation with 2 deoxy D glucose improve functional and metabolic cardiovascular risk factors in rats. FASEB Journal, 17 (9), 1133 1134. Wannamethee, S. G., Shaper, A. G., & Alberti, K. G. (2000). Physical activity, metabolic factors, and the incidence of coronary heart disease and type 2 diabetes. Archives of Internal Medicine, 160 (14), 2108 2116. Wechsler, D. (1981). WAIS R Manual Cleveland: The Psychologic Corporation. Weindruch, R., & Walford, R. L. (1988). The retardation of aging and disease by dietary restriction : CC Thomas Springfield, IL. Weisskopf, M. G., Wright, R. O., Schwartz, J., Spiro, A., 3rd, Sparrow, D., Aron, A., & Hu, H. (2004). Cumulative lead exposure and prospec tive change in cognition among elderly men: The VA Normative Aging Study. American Journal of Epidemiology, 160 (12), 1184 1193. Wellen, K. E., & Hotamisligil, G. S. (2003). Obesity induced inflammatory changes in adipose tissue. Journal of Clinical Inve stigation, 112 (12), 1785 1788. doi: 10.1172/jci20514 Wellen, K. E., & Hotamisligil, G. S. (2005). Inflammation, stress, and diabetes. Journal of Clinical Investigation, 115 (5), 1111 1119. doi: 10.1172/jci25102 West, N. A., & Haan, M. N. (2009). Body adipos ity in late life and risk of dementia or cognitive impairment in a longitudinal community based study. The Journals of Gerontology: Series A: Biological Sciences and Medical Sciences, 64A (1), 103 109. doi: 10.1093/gerona/gln006

PAGE 285

270 Whitmer, R. A. (2007). The e pidemiology of adiposity and dementia. Curr Alzheimer Res, 4 (2), 117 122. Whitmer, R. A., Gunderson, E. P., Barrett Connor, E., Quesenberry, C. P., Jr., & Yaffe, K. (2005). Obesity in middle age and future risk of dementia: A 27 year longitudinal populati on based study. BMJ: British Medical Journal, 330 (7504), 1360. doi: 10.1136/bmj.38446.466238.E0 Whitmer, R. A., Gunderson, E. P., Quesenberry, C. P., Jr., Zhou, J., & Yaffe, K. (2007). Body mass index in midlife and risk of Alzheimer disease and vascular d ementia. Current Alzheimer Research, 4 (2), 103 109. Whitmer, R. A., Gustafson, D. R., Barrett Connor, E., Haan, M. N., Gunderson, E. P., & Yaffe, K. (2008). Central obesity and increased risk of dementia more than three decades later. Neurology, 71 (14), 1 057 1064. doi: 10.1212/01.wnl.0000306313.89165.ef Whitmer, R. A., Karter, A. J., Yaffe, K., Quesenberry, C. P., Jr., & Selby, J. V. (2009). Hypoglycemic episodes and risk of dementia in older patients with type 2 diabetes mellitus. JAMA, 301 (15), 1565 1572 doi: 10.1001/jama.2009.460 Wing, R. R., Vazquez, J. A., & Ryan, C. M. (1995). Cognitive effects of ketogenic weight reducing diets. International Journal of Obesity & Related Metabolic Disorders: Journal of the International Association for the Study of Obesity, 19 (11), 811 816. Wirth, R., Bauer, J. M., & Sieber, C. C. (2007). Cognitive function, body weight and body composition in geriatric patients. Zeitschrift fr Gerontologie und Geriatrie, 40 (1), 13 20. Witte, A., Fobker, M., Gellner, R., Knecht, S ., & Floel, A. (2009). Caloric restriction improves memory in elderly humans. PNAS Proceedings of the National Academy of Sciences of the United States of America, 106 (4), 1255 1260. doi: 10.1073/pnas.0808587106 Wolf, P. A., Beiser, A., Elias, M. F., Au, R ., Vasan, R. S., & Seshadri, S. (2007). Relation of obesity to cognitive function: importance of central obesity and synergistic influence of concomitant hypertension. The Framingham Heart Study. Current Alzheimer Research, 4 (2), 111 116. Wolozin, B., Kel lman, W., Ruosseau, P., Celesia, G. G., & Siegel, G. (2000). Decreased prevalence of Alzheimer disease associated with 3 hydroxy 3 methyglutaryl coenzyme A reductase inhibitors. Archives of Neurology, 57 (10), 1439. Wolozin, B., Wang, S. W., Li, N. C., Lee A., Lee, T. A., & Kazis, L. E. (2007). Simvastatin is associated with a reduced incidence of dementia and Parkinson's disease. BMC Medicine, 5 (1), 20. World Health Organization. (2012). Dementia: a public health priority United Kingdom: World Health O rganization.

PAGE 286

271 Wu, S. Y., Wang, T. F., Yu, L., Jen, C. J., Chuang, J. I., Wu, F. S., . Kuo, Y. M. (2011). Running exercise protects the substantia nigra dopaminergic neurons against inflammation induced degeneration via the activation of BDNF signaling p athway. Brain, Behavior, and Immunity, 25 (1), 135 146. doi: 10.1016/j.bbi.2010.09.006 Xiong, G. L., Plassman, B. L., Helms, M. J., & Steffens, D. C. (2006). Vascular risk factors and cognitive decline among elderly male twins. Neurology, 67 (9), 1586 1591. doi: 10.1212/01.wnl.0000242730.44003.1d Xu, K., Sun, X., Eroku, B. O., Tsipis, C. P., Puchowicz, M. A., & LaManna, J. C. (2010). Diet induced ketosis improves cognitive performance in aged rats. Advances in Experimental Medicine and Biology, 662 71 75. do i: 10.1007/978 1 4419 1241 1_9 Xu, W. L., Atti, A. R., Gatz, M., Pedersen, N. L., Johansson, B., & Fratiglioni, L. (2011). Midlife overweight and obesity increase late life dementia risk: a population based twin study. Neurology, 76 (18), 1568 1574. doi: 10 .1212/WNL.0b013e3182190d09 Yadav, A., Kataria, M. A., Saini, V., & Yadav, A. (2012). Role of leptin and adiponectin in insulin resistance. Clinica Chimica Acta, 417C 80 84. doi: 10.1016/j.cca.2012.12.007 Yaffe, K. (2007). Metabolic syndrome and cognitive decline. Curr Alzheimer Res, 4 (2), 123 126. Yaffe, K., Barnes, D., Nevitt, M., Lui, L. Y., & Covinsky, K. (2001). A prospective study of physical activity and cognitive decline in elderly women: women who walk. Archives of Internal Medicine, 161 (14), 1703 1708. Yaffe, K., Blackwell, T., Kanaya, A. M., Davidowitz, N., Barrett Connor, E., & Krueger, K. (2004a). Diabetes, impaired fasting glucose, and development of cognitive impairment in older women. Neurology, 63 (4), 658 663. Yaffe, K., Blackwell, T., Ka naya, A. M., Davidowitz, N., Barrett Connor, E., & Krueger, K. (2004b). Diabetes, impaired fasting glucose, and development of cognitive impairment in older women.[Summary for patients in Neurology. 2004 Aug 24;63(4):E9 10; PMID: 15326277]. Neurology, 63 (4 ), 658 663. Yaffe, K., Kanaya, A., Lindquist, K., Simonsick, E. M., Harris, T., Shorr, R. I., . Newman, A. B. (2004c). The metabolic syndrome, inflammation, and risk of cognitive decline. JAMA, 292 (18), 2237 2242. doi: 10.1001/jama.292.18.2237 Yasutak e, C., Kuroda, K., Yanagawa, T., Okamura, T., & Yoneda, H. (2006a). Serum BDNF, TNF alpha and IL 1beta levels in dementia patients: comparison between Alzheimer's disease and vascular dementia. European Archives of Psychiatry and Clinical Neuroscience, 256 (7), 402 406. doi: 10.1007/s00406 006 0652 8 Yasutake, C., Kuroda, K., Yanagawa, T., Okamura, T., & Yoneda, H. (2006b). Serum BDNF, TNF and IL European Archives of Psychiatry and Clinical Neuroscience, 256 (7), 402 406.

PAGE 287

272 Ye n, P. K. (2005). Relationship of dementia and body weight. Geriatric Nursing, 26 (2), 79 80. Yu, B. P., & Chung, H. Y. (2001). Stress resistance by caloric restriction for longevity. Annals of the New York Academy of Sciences, 928 39 47. Zamarian, L., Si nz, H., Bonatti, E., Gamboz, N., & Delazer, M. (2008). Normal aging affects decisions under ambiguity, but not decisions under risk. Neuropsychology, 22 (5), 645. Zandi, P. P., Sparks, D. L., Khachaturian, A. S., Tschanz, J., Norton, M., Steinberg, M., . Breitner, J. (2005). Do statins reduce risk of incident dementia and Alzheimer disease?: The Cache County Study. Archives of General Psychiatry, 62 (2), 217. Zeki Al Hazzouri, A., Haan, M. N., Whitmer, R. A., Yaffe, K., & Neuhaus, J. (2012). Central obe sity, leptin and cognitive decline: the Sacramento Area Latino Study on Aging. Dementia and Geriatric Cognitive Disorders, 33 (6), 400 409. doi: 10.1159/000339957 Zeki Al Hazzouri, A., Stone, K. L., Haan, M. N., & Yaffe, K. (2013). Leptin, mild cognitive im pairment, and dementia among elderly women. Journals of Gerontology. Series A: Biological Sciences and Medical Sciences, 68 (2), 175 180. doi: 10.1093/gerona/gls155 Zhang, Y., & Pardridge, W. M. (2006). Blood brain barrier targeting of BDNF improves motor f unction in rats with middle cerebral artery occlusion. Brain Research, 1111 (1), 227 229. doi: 10.1016/j.brainres.2006.07.005 Zhao, G., Ford, E., Li, C., Tsai, J., Dhingra, S., & Balluz, L. (2011). Waist circumference, abdominal obesity, and depression amon g overweight and obese US adults: national health and nutrition examination survey 2005 2006. BMC Psychiatry, 11 (1), 130. Zhao, H., Tuominen, E. K., & Kinnunen, P. K. (2004). Formation of amyloid fibers triggered by phosphatidylserine containing membranes Biochemistry, 43 (32), 10302 10307. Zhao, Q., Stafstrom, C. E., Fu, D. D., Hu, Y., & Holmes, G. L. (2004). Detrimental effects of the ketogenic diet on cognitive function in rats. Pediatric Research, 55 (3), 498 506. doi: 10.1203/01.pdr.0000112032.47575.d 1 Zhao, Z. Y., Lu, F. H., Xie, Y., Fu, Y. R., Bogdan, A., & Touitou, Y. (2003). Cortisol secretion in the elderly. Influence of age, sex and cardiovascular disease in a Chinese population. Steroids, 68 (6), 551 555. doi: 10.1016/s0039 128x(03)00083 7 Zhu, H ., Guo, Q., & Mattson, M. P. (1999). Dietary restriction protects hippocampal neurons against the death promoting action of a presenilin 1 mutation. Brain Research, 842 (1), 224 229. Ziegenhorn, A. A., Schulte Herbrggen, O., Danker Hopfe, H., Malbranc, M. Hartung, H. D., Anders, D., . Hellweg, R. (2007). Serum neurotrophins a study on the time course and influencing factors in a large old age sample. Neurobiology of Aging, 28 (9), 1436 1445.

PAGE 288

273 Zimmet, P., Boyko, E., Collier, G., & Courten, M. d. (1999). Etiology of the metabolic syndrome: potential role of insulin resistance, leptin resistance, and other players. Annals of the New York Academy of Sciences, 892 (1), 25 44. Zupec Kania, B. A., & Spellman, E. (2008). An overview of the ketogenic diet for pe diatric epilepsy. Nutrition in Clinical Practice, 23 (6), 589 596. doi: 10.1177/0884533608326138