Citation
Chemotherapy treatment, brain gray matter, and markers of inflammation in women with breast cancer

Material Information

Title:
Chemotherapy treatment, brain gray matter, and markers of inflammation in women with breast cancer a longitudinal cohort
Creator:
Kairys, Anson E. ( author )
Language:
English
Physical Description:
1 electronic file (58 pages). : ;

Subjects

Subjects / Keywords:
Brain -- Physiology ( lcsh )
Breast -- Cancer ( lcsh )
Inflammation ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
Breast cancer (BrCA) is associated with the highest mortality rate of any type of cancer besides that of the lung. Although the prevalence, mortality, and costs associated with breast cancer are quite high, recent advances in screening and treatment have increased long term survivorship. Thus, understanding the outcomes of treatment in patients with BrCA is of great importance to help better understand quality of life and biological markers of treatment-related outcomes in this population. To this end, the present study evaluated changes in brain gray matter volume, burden of inflammation, and levels of anxiety and depression associated with chemotherapy treatment in women with breast cancer. Additionally, this study tested whether there is an association between levels of inflammation, alterations in gray matter, and levels of depression and anxiety associated with chemotherapy. This was accomplished using previously collected prospective data involving women with BrCA who have undergone chemotherapy treatment, women with cancer who did not receive chemotherapy, and age-matched healthy controls. Data were collected at three time points: prior to chemotherapy, 3 months after chemotherapy, and finally 9 months after treatment completion. Results of the analyses showed a significant interaction between group and time on a priori gray matter regions of interest found to be altered in previous research within this population. However, the findings of the present study did not show consistent decreases in GM volumes associated with chemotherapy treatment as has been found previously. Correlation analyses showed significant negative relationships between changes in IL6 and gray matter volumes in the middle frontal gyrus, and the parahippocampal gyrus following chemotherapy treatment. Further, there was also a negative relationship between changes in gray matter volume in the cingulate gyrus and levels of depression in the chemotherapy-treated patients. These results are discussed in light of previous research, and future directions for research are provided.
Thesis:
Thesis (M.A.)--University of Colorado Denver.
Bibliography:
Includes bibliographic references
System Details:
System requirements: Adobe Reader.
General Note:
Department of Psychology
Statement of Responsibility:
by Anson E. Kairys.

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:
951985643 ( OCLC )
ocn951985643

Downloads

This item has the following downloads:


Full Text
CHEMOTHERAPY TREATMENT, BRAIN GRAY MATTER, AND MARKERS OF
INFLAMMATION IN WOMEN WITH BREAST CANCER: A LONGITUDINAL COHORT
STUDY
By
ANSON E. KAIRYS
B.A., University of Michigan
A thesis submitted to the
Faculty of the Graduate school of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Arts
Clinical Health Psychology Program
2015


2015
ANSON E. KAIRYS
ALL RIGHTS RESERVED
11


This thesis for the Master of Arts degree by
Anson E. Kairys
has been approved for the
Clinical Health Psychology Program
by
Jim Grigsby, Chair
Edward Dill
Kristin Kilbourn
Date: January 4, 2016


Kairys, Anson E. (M. A. Clinical Health Psychology)
Chemotherapy Treatment, Brain Gray Matter, and Markers of Inflammation in Women with
Breast Cancer: a Longitudinal Cohort Study
Thesis directed by Professor Jim Grigsby
ABSTRACT
Breast cancer (BrCA) is associated with the highest mortality rate of any type of cancer
besides that of the lung. Although the prevalence, mortality, and costs associated with breast
cancer are quite high, recent advances in screening and treatment have increased long term
survivorship. Thus, understanding the outcomes of treatment in patients with BrCA is of great
importance to help better understand quality of life and biological markers of treatment-related
outcomes in this population. To this end, the present study evaluated changes in brain gray
matter volume, burden of inflammation, and levels of anxiety and depression associated with
chemotherapy treatment in women with breast cancer. Additionally, this study tested whether
there is an association between levels of inflammation, alterations in gray matter, and levels of
depression and anxiety associated with chemotherapy. This was accomplished using previously
collected prospective data involving women with BrCA who have undergone chemotherapy
treatment, women with cancer who did not receive chemotherapy, and age-matched healthy
controls. Data were collected at three time points: prior to chemotherapy, 3 months after
chemotherapy, and finally 9 months after treatment completion. Results of the analyses showed a
significant interaction between group and time on a priori gray matter regions of interest found
to be altered in previous research within this population. However, the findings of the present
study did not show consistent decreases in GM volumes associated with chemotherapy treatment
as has been found previously. Correlation analyses showed significant negative relationships
between changes in IL6 and gray matter volumes in the middle frontal gyrus, and the
IV


parahippocampal gyrus following chemotherapy treatment. Further, there was also a negative
relationship between changes in gray matter volume in the cingulate gyrus and levels of
depression in the chemotherapy-treated patients. These results are discussed in light of previous
research, and future directions for research are provided.
The form and content of this abstract are approved. I recommend its publication
Approved: Jim Grigsby
v


TABLE OF CONTENTS
CHAPTER
I. BACKGROUND
Overview of Inflammation..............................................2
Chemotherapy-Associated Inflammation..................................3
How Pro-Inflammatory Cytokines may affect the Brain...................5
Brain Imaging Studies in Depression and Anxiety.......................6
Gray Matter Changes Associated with Chemotherapy......................7
Hypotheses and Specific Aims.........................................13
II. METHOD
Participants.........................................................16
Procedure............................................................17
Measures.............................................................19
Data Analysis........................................................23
III. RESULTS..............................................................25
Primary Analyses.....................................................25
Exploratory Analysis.................................................33
IV. DISCUSSION...........................................................34
Strengths and Limitations............................................39
Future Directions....................................................42
Conclusions..........................................................42
REFERENCES................................................................45
vi


LIST OF TABLES
TABLE
1. Inclusion and Exclusion Criteria...............................................17
2. Baseline Demographics and Clinical Characteristics.............................25
3. Results of MANCOVA Analysis....................................................26
4. Results of Correlation Analysis Among Change Scores (3 months - baseline)....28
5. Results of Correlation Analysis Among Change Scores (9 months - 3 months).....30
vii


LIST OF FIGURES
FIGURE
1. Correlation Between IL6 and Right Middle Frontal Gyrus GM Volume..........29
2. Correlation Between CES-D and Left Cingulate Gyrus GM Volume..............31
3. Correlation Between IL6 and Left Parahippocampal Gyrus GM Volume..........32
viii


CHAPTER I
BACKGROUND
Breast cancer (BrCA) is associated with the second highest mortality rate of any type of
cancer (American Cancer Society, 2013). BrCA occurs most commonly in women over the age
of 40, with an estimated global incidence of 1.6 million cases in 2010. An estimated 39,000
women in America died from BrCA alone in 2013 (American Cancer Society, 2013). BrCA is
responsible for the largest portion of spending on cancer-related care in the United States; an
estimated $16.5 billion of cancer-related spending went towards BrCA in 2010 (Mariotto, Robin
Yabroff, Shao, Feuer, & Brown, 2011). Although the prevalence, mortality, and costs associated
with BrCA are quite high, recent advances in screening and treatment have increased long term
survivorship of BrCA patients (McDonald, Conroy, Ahles, West, & Saykin, 2010). This increase
in long-term survivorship has in turn led to increased attention to survivors daily functioning
and quality of life (QOL). Therefore, understanding the outcomes of treatment in patients with
BrCA is of great importance to help better understand QOL and biological markers of treatment-
related outcomes in this population.
The most common form of treatment for breast cancer, after surgery, is adjuvant
chemotherapy (American Cancer Society, 2013). The term adjuvant refers to a treatment
applied secondary to an initial treatment, typically to suppress further tumor formation
(American Cancer Society, 2013). Although chemotherapy is a successful treatment in many
cases, there are many side effects associated with the treatment. Common side effects of
chemotherapy include: nausea and vomiting, hair loss, appetite loss, pain, fatigue, depression,
anxiety, problems with concentration and memory, and sleep disturbances, among others
(American Cancer Society, 2013). These side effects take a toll on patients, and contribute


significantly to decreased QOL, even after finishing treatment (McDonald et al., 2010). Systemic
inflammation, resulting from both the disease process and treatment with chemotherapy, has
been suggested as an underlying factor contributing to the observed side effects. Research into
these inflammatory processes appears to support this notion (Bower et al., 2011; Pomykala,
Ganz, et al., 2013).
Overview of Inflammation
When tissue injury occurs, a network of chemical signals activate a host response which is
designed to protect and heal the damaged tissues (i.e., inflammation) (Coussens & Werb, 2002).
This process involves the activation and migration of leukocytes (neutrophils, monocytes, and
eosinophils) to the site of damage as well as recruitment of tissue mast cells (Coussens & Werb,
2002). Additionally, this inflammatory response involves migration of macrophage progenitors,
which travel through the venous system to the site of injury. This migratory process is directed to
the site of tissue damage by a host of chemotactic factors, among these are cytokines including
interleukin-ip (IL-ip) and tumor-necrosis factor alpha (TNF-a) (Coussens & Werb, 2002). Once
present at the site of injury, these macrophage progenitors will differentiate into either mature
macrophages or immature dendritic cells.
After activation, macrophages become the main source of cytokines (TNF-a and IL-1),
which help modulate tissue repair and can have drastic effects on the local microenvironment
(Coussens & Werb, 2002). The major pro-inflammatory cytokines (PICs), which include
Interleukin-1 (ILl-a and IL1-|3), Interleukin-6 (IL-6), and Tumor Necrosis Factor-alpha (TNF-a)
promote inflammation. They are regulated and modulated by other molecules, including anti-
inflammatory cytokines such as IL4, IL10, and IL13. The coordinated dynamic activity of the
various classes of cytokines and receptors allows the appropriate development of inflammation,
2


and its down regulation when it is no longer necessary. Sabotage of cell death and repair
programs occurs in chronically inflamed tissues, thus resulting in DNA replication and
proliferation of cells that have lost normal growth control.
Normal inflammation is self-limiting, because the production of anti-inflammatory cytokines
follows the pro-inflammatory cytokines closely. However, chronic inflammation seems to be due
to persistence of the initiating factors or a failure of mechanisms required for resolving the
inflammatory response. For example, the PIC TNF-a mediates many factors of the inflammatory
process. During early development of tumors in breast cancers, TNF-a regulates a cascade of
cytokines and growth factors which actually may be one of the ways inflammation acts as a
tumor promoter when these cascades become unregulated (Coussens & Werb, 2002). Another
example of this unregulated inflammatory response involves the PIC IL-6. During acute
inflammation, IL-6 stimulates and regulates protein production, while concurrently controlling
the level of the inflammatory response by regulating anti-inflammatory cytokines. Therefore,
during acute inflammation, IL-6 possesses both pro-inflammatory and anti-inflammatory
properties. These properties change during chronic inflammation, a state in which IL-6 seems to
mediate only pro-inflammatory responses and also enhances white blood cell infiltration at the
sites of inflammation (Dethlefsen, Hojfeldt, & Hojman, 2013). This becomes important in BrCA,
a disease that is multifactorial, and is known to involve this chronic inflammatory response.
Chemotherapy-Associated Inflammation
Inflammation is associated with the side-effects of chemotherapy mentioned above, and
has been shown to affect cognition as well, which has become associated with sickness
behavior and may be related to chemobrain in this population (Dantzer & Kelley, 2007;
Wang et al., 2015). It has been proposed that inflammation, as measured by plasma levels of
3


PICs, is the underlying cause of many of the side-effects observed with chemotherapy treatment,
namely depression, fatigue, sleep disturbances, and problems with memory and cognition
(Bower et al., 2011; Mohamed, Al-Raawi, Sabet, & El-Shinawi, 2014). For example, a recent
study of 103 women with breast cancer who received primary treatment (i.e., surgery,
radiotherapy, or chemotherapy) found that women treated with chemotherapy reported higher
levels of depression, fatigue, and sleep disturbances than women treated with surgery or
radiotherapy. The women treated with chemotherapy also had higher plasma levels of tumor
necrosis factor receptor type II (sTNF-RII), a marker of inflammation, when compared to the
other treatment groups, and healthy controls. Finally, this group also showed a strong
relationship between reported levels of fatigue, and sTNF-RII levels (Bower et al., 2011). This
study indicates that measures of cytokine plasma levels can be used as biomarkers to track
inflammation in patients with breast cancer who have undergone chemotherapy, and these
biomarkers seem to correlate with behavioral and psychological symptoms observed in this
population. These biomarkers hold importance as they are related to quality of life and other
health-related outcomes. Additional biomarkers associated with the cognitive declines found in
this population have been thought to occur in the central nervous system (CNS). Therefore,
brain-imaging methods have been employed in order to discover potential biomarkers in the
brain (e.g., changes in white matter or gray matter integrity), which may have implications for
disease progression and behavioral symptom development. Importantly, few studies to date have
investigated the etiology of alterations in brain Gray matter (GM) following chemotherapy. This
study seeks to fill this gap in the literature by assessing burden of inflammation and changes in
brain GM morphology as well as outcomes on measures of anxiety and depression.
4


How Pro-Inflammatory Cytokines may effect the Brain
Chemotherapeutic agents are thought to have limited direct access to brain tissues due to
the blood-brain barrier (BBB). However, animal studies suggest that chemotherapeutic agents
which are not known to readily cross the BBB (e.g., doxorubicin) may be associated with
reduced neurogenesis (Janelsins et al., 2010). Further, even small amounts of chemotherapy may
cause long-term damage in the brain as both dividing and non-dividing glial cells have been
shown to be particularly vulnerable to chemotherapeutic agents (Dietrich, 2010).
The PICs (IL1, IL6, and TNF-a) typically have limited access to the brain, though they
may cross the BBB in small numbers via active transport (Goehler et al., 2000). These PICs may
also enter the cerebrospinal fluid through circumventricular organs (e.g., the choroid plexus) and
the vagus nerve, while also being synthesized endogenously in the brain (Goehler et al., 2000).
Receptors for these PICs have been found throughout the brain, which the PICs must bind to in
order to have an effect. Once these cytokines bind to receptors in the brain, they may exacerbate
damage to neurons and glial cells and lead to neuronal death as well as to certain types of
cognitive impairment (Dietrich, 2010). IL6 increases BBB permeability, while IL1 and TNF-a
induce free radicals that may then affect the BBB (Merrill & Benveniste, 1996). Once in the
brain, these cytokines lead to activation of microglia and astrocytes, and thus promote even
greater inflammation (Coussens & Werb, 2002).
Herein, it is proposed that chemotherapeutic agents may, due to a breakdown of the BBB,
induce acute effects as well as long-term alterations in brain GM. It is suggested that this
increased permeability of the capillary endothelial layer allows PICs access to the brain, inducing
a state of brain inflammation. Thus, chemotherapy may be associated with a mild increase in
5


permeability of the BBB, producing some degree of brain inflammation, with associated changes
in levels of PICs and observable changes in brain GM morphology.
Brain Imaging Studies in Depression and Anxiety
Structural imaging studies looking at changes in GM associated with depression have
shown mixed results (Lai, 2013). A recent meta-analysis, however, revealed that there does seem
to be consistent GM decreases in the brains of depressed patients occurring in the anterior
cingulate cortex, or ACC (Lai, 2013). This region is thought to be involved in attention,
cognition, affective regulation, and motivation (Lai, 2013). Decreased GM in the ACC is thought
to disrupt the visceromotor network which in turn may change ones ability to cope with stress.
The ACC is thus a potential target for GM changes in patients with BRCA as they experience
significant stress, and show signs of comorbid depression following chemotherapy (McDonald,
Conroy, Smith, West, & Saykin, 2013).
Neuroimaging studies looking at symptoms of anxiety are limited. Recently, Moon and
colleagues (2014) conducted a cross-sectional study using VBM to evaluate GM changes in
patients with generalized anxiety disorder (GAD). 22 patients with GAD (13 men and 9 women)
were compared with 22 age- and sex-matched healthy controls. When compared to healthy
controls, patients with GAD showed significantly decreased GM volume in the hippocampus,
thalamus, insula, and superior temporal gyrus (Moon, Kim, & Jeong, 2014). One interesting
aspect is the decreased GM volume found in the hippocampus of GAD patients. The
hippocampus has been found to play an important role in the hormone regulation involved in the
hypothalamic-pituitary-adrenal axis (Moon et al., 2014). The authors speculate that a decreased
GM volume in the hippocampus may be related to dysregulation in neuroendocrine systems, thus
6


leading to more stress for the individual. Again, this seems to fit well with how BRCA patients
are conceptualized and with the prevalence of anxiety in these patients.
Gray Matter Changes Associated with Chemotherapy
Previous research has focused on the central nervous system (CNS) in search of
biomarkers associated with the many side-effects observed following chemotherapy treatment
(Cimprich et al., 2010; Conroy et al., 2012; de Ruiter et al., 2012; Inagaki et al., 2007; McDonald
et al., 2013; McDonald et al., 2010). One method that can be used to assess structural changes in
the CNS at the level of the brain is Voxel-Based Morphometry (VBM; (Ashburner & Friston,
2000). VBM is a fully automated procedure for assessing gray matter (GM) and white matter
(WM) integrity in the brain using structural magnetic resonance images (MRI). VBM is a
method to quantitatively assess tissue changes in GM and WM on a voxel-by-voxel basis based
on an a priori statistical threshold. Thus, VBM is an unbiased, highly accurate measure, sensitive
to local changes in the brain (Ashburner & Friston, 2000).
Inagaki and colleagues (2007) conducted one of the first cross-sectional VBM analyses
on BrCA survivors treated with chemotherapy. Patients were split into two groups: those treated
with chemotherapy less than one year after surgery (c+, n=55), and those treated more than 3
years after surgery (c+, n=73). These two groups were then compared to similar groups of
patients who did not receive chemotherapy at 1 or 3 years (c-, n=55 and 59, respectively), and
healthy controls (HC; n=55 and n=37, respectively) (Inagaki et al., 2007). In the less than 1-year
study, when compared to the C- group, the C+ group showed decreased GM in the right middle
and superior frontal gyrus, and the right parahippocampal gyrus. Additionally, they found
evidence in the C+ group of less WM in the right cingulate gyrus, bilateral middle frontal gyrus,
left parahippocampal gyrus, and the left precuneus, when compared to the C- group. No
7


differences were observed in GM or WM between either C+ or C- groups when compared to HC.
In the 3-year study, no differences between or within C+ or C- were observed, suggesting the
effects of chemotherapy on GM and WM may recover over time (Inagaki et al., 2007).
McDonald and colleagues (2010) conducted the first longitudinal, prospective study of
structural brain changes in women with BrCA receiving chemotherapy, using VBM. They
included three groups: BrCA patients treated with chemotherapy (C+, n=17), BrCA patients who
had not received chemotherapy (C-, n=12), and a matched healthy control group (HC, n=18)
(McDonald et al., 2010). Each group was evaluated before chemotherapy, at one month, and
finally one year after completion of treatment. They reported no between-group differences in
brain GM at baseline. Comparisons between groups revealed decreased GM density in the
middle frontal gyrus (bilaterally) and cerebellum at 1 month in both cancer groups, when
compared to HC (McDonald et al., 2010). However, the declines were much more extensive and
persistent for the C+. Within the C+ group, over 1 month there were GM density decreases in
bilateral frontal, temporal, and cerebellar regions, right thalamus and bilateral temporal lobes
including medial temporal structures. Some frontal areas of grey matter reduction persisted to the
1 year time point, while some areas recovered (McDonald et al., 2010). An important thing to
note is that comparisons between C+ and C- groups indicated no differences in GM at 1 month
or at 1 year. Although this study failed to detect differences between C+ and C- patients, these
findings suggest an acute effect of chemotherapy in GM density with subsequent partial recovery
over time (at 1 year). The relatively small sample could also have contributed to the lack of
significant findings. This study laid the groundwork for further work in this area.
These same researchers sought to replicate and extend their findings of decreased GM
density in this population (McDonald et al., 2013). BrCA patients treated with and without
8


chemotherapy (n=27 and n = 28, respectively) and matched HC (n = 24) were scanned at
baseline (prior to treatment) and 1 month following chemotherapy (or matched intervals). GM
reductions after 1 month were found in left middle and superior frontal gyrus, when compared to
baseline. Additionally, these researchers found that chemotherapy treated BrCA patients reported
significantly more self-perceived symptoms in their ability to initiate a task and generate
problem-solving strategies, as measured by the BRIEF-A. Finally, they found a negative
correlation between difference scores on self-reported executive functioning and GM density in
the left middle frontal gyrus (MFG). The authors infer from this result that a reduction in GM in
the MFG may be associated with increased cognitive complaints as evidenced by the BRIEF-A.
The result also suggests a relationship between reductions in frontal GM volume and executive
functioning in BrCA patients who have undergone chemotherapy.
These two studies (McDonald et al., 2013; McDonald et al., 2010) suggest that GM
reductions are a result of chemotherapy treatment in BrCA patients, and these reductions seem to
be localized to prefrontal regions. One may then speculate about the relationship between
reductions in prefrontal cortex GM and the observed cognitive complaints in this population,
another notion supported by the self-reports of current executive functioning in the previous
study.
de Ruiter and colleagues (2012) investigated the long-term effects of chemotherapy
treatment in BrCA patients more than nine years after receiving treatment. This study included
17 BrCA patients receiving high dose chemotherapy (C+) with 15 patients who were untreated
(C-). This study utilized a multimodal imaging approach to test their hypotheses. Their VBM
analysis showed significant differences in GM volume between the two groups. Namely, they
found reductions in posterior regions of the brain such as the posterior parietal cortex, the
9


precuneus (bilaterally), the left occipital cortex, and the cerebellum, bilaterally in the C+ group
but not the C- group. These GM volume reductions were found to overlap with fMRI task-related
hypoactivations (de Ruiter et al., 2012). Further, these same regions showed evidence of changes
in WM microstructure, thought to represent WM integrity. These results are suggestive of long-
term neurotoxic side effects of high-dose adjuvant chemotherapy for BrCA. However, these
results also contradict some of the previous work mentioned above. This study would suggest
that there is a longitudinal effect of chemotherapy treatment, while the study by Inagaki and
colleagues (2007) suggests acute effects of chemotherapy on brain GM, with a restoration of
these effects 3 years later. There are several limitations to the study by Inagaki et al. (2007) that
should be considered when interpreting their results. One such limitation is the wide variety of
chemotherapeutic agents used within the sample, thereby making it impossible to determine the
effects of a specific regimen on GM values. A second significant limitation is that two, only
partially overlapping, samples were used when evaluating the effects of chemotherapy at the 3
years time point. This poses difficulties in interpreting their results, but also led researchers in
the right direction towards ways in which experimental control can be best established to answer
these questions.
Koppelman and colleagues (2013) conducted the first study to determine the long-term
effects of chemotherapy treatment on brain volume in a relatively homogenous sample
(Koppelmans, Breteler, Boogerd, Seynaeve, & Schagen, 2013). In this study, MRI scans from
184 BrCA patients who had undergone chemotherapy (on average 21 years prior) were
compared to those of 368 age-matched healthy controls (Koppelmans et al., 2013). These
researchers reported that BrCA patients who underwent chemotherapy had significantly less total
brain volume (sum of GM, WM, and Cerebrospinal fluid or CSF), and total GM volume than the
10


healthy controls. VBM analysis showed no local GM differences between the two groups. With
regard to the differences in total brain volume and total GM, they found these reductions were
long-term. In this sample, these researchers showed that every year of aging was associated with
a 0.75 ml reduction in total GM volume in the C+ group (Koppelmans et al., 2013). Based on
their observed differences in total GM volume between groups, these authors suggested that
exposure to chemotherapy is equivalent to 3.9 years of additional aging. Another possible
interpretation is that the reductions in total GM volume is related to the stress associated with the
treatment or dealing with the cancer itself.
In addition to these studies on GM changes associated with chemotherapy treatment,
there have been only two studies that have investigated the relationship between neuroimaging
outcomes and markers of inflammation in BrCA patients who have undergone chemotherapy
treatment (Kesler et al., 2013; Pomykala, Ganz, et al., 2013).
Pomykala and colleagues (2013) conducted a positron-emission topography (PET) study
to examine the relationships between PICs, cerebral metabolism, and cognitive complaints
following chemotherapy treatment in BrCA patients. In this study, 23 BrCA patients who
underwent chemotherapy and 10 BrCA patients who did not were analyzed at baseline
(following treatment) and then again one year later. Data collection involved brain imaging data
(cerebral metabolism via PET) and cytokine markers, including IL-1, sTNF-RII, CRP, and IL-6.
These researchers found that at baseline, there was a positive correlation between the levels of
cytokines and cerebral metabolism in the left medial prefrontal cortex and right inferior lateral
anterior temporal cortex in the chemotherapy group. This correlation was not found in the non-
chemotherapy group. At the 1-year time point, there was a positive correlation between cytokine
levels and metabolism in the medial prefrontal cortex, and the anterior temporal cortex. Again,
11


this correlation was not present in the non-chemotherapy group (Pomykala, Ganz, et al., 2013).
These results suggest increased cerebral metabolism may be associated with increased burden of
inflammation. The authors speculate that an initial inflammatory response (perhaps due to
exposure to chemotherapeutic agents) begins a cascade that ends up affecting brain metabolism.
It is possible that the observed alterations in metabolism could lead to the changes observed in
other studies looking at GM changes associated with chemotherapy as increased metabolism in
some areas could serve as a compensatory mechanism for loss of neuronal functioning or
integrity in other regions.
Kesler and colleagues (2013) conducted the only study to date directly testing the
relationship between cytokine markers of inflammation and brain GM volume. In this cross-
sectional study, MRI, cytokine data, and measures of memory performance were collected from
20 BrCA patients who had undergone chemotherapy (average time since chemotherapy ~
4.8years) and 23 healthy controls. These researchers focused specifically on the hippocampus in
search of GM alterations associated with markers of inflammation. This was accomplished by
manually tracing the hippocampus, bilaterally, in order to obtain hippocampal GM volumes.
Their results showed that left hippocampal volume and memory performance were reduced in
the BrCA group compared to the healthy controls. They also found increased levels of IL-6 and
TNF-a in the BrCA group. Correlation analyses showed that lower hippocampal volume was
associated with higher levels of TNF-a (negative correlation), but lower levels of IL-6 (positive
correlation). The authors suggest that these contradictory findings are due to both the anti- and
pro-inflammatory roles that IL-6 plays in the immune system. Finally, they found a significant
interaction between these cytokines, such that the authors propose that IL-6 may have a
moderating effect on TNF-a levels. This study provides insight into the possible effects
12


chemotherapy may have on burden of inflammation, and the association between brain GM
changes and cytokine markers of inflammation (Kesler et al., 2013). However, this study was
limited by the fact that they were not able to determine the acute effects of chemotherapy
treatment on these measures due to the long duration post-treatment. Additionally, this study did
not include BrCA patients who did not receive chemotherapy as a necessary control group.
As can be seen, research evaluating the association between PICs and GM volume in
BrCA patients undergoing chemotherapy treatment is limited. Further, despite the prevalence of
anxiety and depressive symptoms in this population, little is known about the relationship
between changes in GM volume and symptoms of depression and anxiety. Thus, this necessary
step can bridge the gap between observed differences in GM volumes in this condition, and serve
to tie the associated symptoms of chemotherapy treatment and markers of inflammation into a
cohesive whole so that underlying mechanisms can be elucidated and models of symptom
development can be established.
Hypotheses and Specific Aims
There have been many imaging studies conducted in this population; however, there has
been little consistency in the findings. Therefore, this study intended to replicate and extend on
previous work done with this population. Specifically, based on previous research, we expected
to see significant changes in the following brain regions following 3 months of chemotherapy
treatment: the parahippocampal gyrus (PHG), the middle frontal gyrus (MFG), the superior
frontal gyrus (SFG), the precentral gyrus (PCG), and the cingulate gyrus (CG). Further, we
expected these changes to persist at 9 months. These regions have been previously identified, and
highlighted in a recent review of imaging studies in this population (Pomykala, de Ruiter,
Deprez, McDonald, & Silverman, 2013). Based on the coordinates provided in these recent
13


imaging studies, we built regions of interest (ROIs) and tested these a priori regions specifically
for all hypotheses.
Aim 1: To test whether chemotherapy treatment is associated with alterations in brain
GM volume in regions previously found to change with chemotherapy treatment
(replication and extension).
Hypothesis 1.1: At baseline (prior to chemotherapy treatment), there will be no significant
differences in regional brain GM volume.
Hypothesis 1.2: At baseline (prior to chemotherapy treatment), there will be no significant
differences in regional brain GM volume in a priori ROIs or PIC levels between BrCA
patients and HC.
Hypothesis 1.3: Compared with BrCA patients who have not received chemotherapy and
HC, patients who have received chemotherapy will show decreased brain GM volume
following chemotherapy treatment in the a priori ROIs mentioned above.
Aim 2: To explore the relationship between brain GM volume changes, burden of
inflammation, and levels of depression and anxiety in chemotherapy treated breast cancer
patients.
Hypothesis 2.1: Between baseline and 3 months following completion of chemotherapy
treatment, there will be a negative relationship between changes in regional brain GM
volume in a priori ROIs and PIC levels in BrCA patients who have undergone
chemotherapy (BrCA+).
Hypothesis 2.2: Between 3 and 9 months following completion of chemotherapy, there will
be a negative relationship between changes in brain GM volume in pre-defined ROIs and
markers of inflammation in the BrCA+ group.
14


Specifically, it has been shown previously that receptors for the PICs collected in the present
study have been found throughout the brain, to which the PICs must bind in order to have an
effect. Once these cytokines bind to receptors in the brain, they may exacerbate damage to
neurons and glial cells and lead to neuronal death as well as to certain types of cognitive
impairment (Dietrich, 2010). This suggests that inflammation may be associated with decreases
in brain GM, as PICs can both enter the brain, and may play a role in neuronal cell death, as
evidenced by decreased GM.
Hypothesis 2.3: Between baseline and 3 months following completion of chemotherapy
treatment, there will be a negative relationship between changes in brain GM volume in a
priori ROIs and changes on two self-report measures of depression and anxiety in the
BrCA+ group.
Hypothesis 2.4: Between 3 and 9 months following completion of chemotherapy treatment,
there will be a negative relationship between changes in brain GM volume in a priori ROIs,
and changes on two self-report measures of depression and anxiety in the BrCA+ group.
Hypothesis 2.5: Between baseline and 3 months, and between 3 months and 9 months
following chemotherapy treatment, there will be positive relationships between burden of
inflammation (change in PIC levels) and changes on self-report measures of depression and
anxiety in BrCA patients who have undergone chemotherapy.
Hypothesis 2.6: In this exploratory analysis, we hypothesize that brain GM volumes in a
priori ROIs will be negatively correlated with two-self report items related to patients
memory and stress associated with chemotherapy treatment, in BrCA patients who have
undergone chemotherapy.
15


CHAPTER II
METHOD
Participants
An existing dataset was used for the present study. Following a protocol approved by the
Colorado Multiple Institutional Review Board (COMIRB) and consistent with HIPAA
regulations, written informed consent was obtained from all participants. Two hundred and nine
females with BrCA (stage 0-3) were recruited to take part in a larger study of the cognitive
trajectory and quality of life of women with BrCA undergoing chemotherapy. A subset of
eligible participants was randomly selected to undergo additional MRI testing.
Thirty-one females with BrCA who underwent chemotherapy treatment (BrCA+; mean
age in years SD; 57.96 6.64), 18 BrCA patients who did not undergo chemotherapy
treatment (BrCA-; mean age in years SD; 67.6 5.5), and 20 healthy controls (mean age in
years SD; 63.2 7.84) were randomly selected for MRI testing. In this total subsample,
approximately 95% (55/59 patients) identified themselves as White/Caucasian. Three patients
identified themselves as being black, and another identified herself as being Asian. Although we
tried to maximize recruitment of racial and ethnic minorities, the demographics of the sampling
frame made this difficult. The data in the present analysis is based on this subset of BrCA
patients and HC that received MRI testing (n=59) taken from the larger sample of 209 BrCA
patients. A trained MRI technician collected all MRI data. See Table 1 for specific inclusion and
exclusion criteria.
16


Table 1. Inclusion and Exclusion Criteria
Inclusion Criteria Exclusion Criteria
English-speaking females Stage 4 disease
Confirmed diagnosis of Stage 0-3 BrCA Planned high-dose chemotherapy regimen
Post-operative, lumpectomy or mastectomy Sensory deficits that would limit interview data
Scheduled for adjuvant or neoadjuvant standard dose chemotherapy (chemotherapy group) Hx of delirium, dementia, mental retardation, or psychosis
Not scheduled for chemotherapy (control group) Hx of significant alcohol/drug use past 18 mos.
Age > 45 Significant CNS comorbidity (e.g. CVA, MS)
Acute medical condition, exacerbation of a chronic condition, or medical-surgical treatment affecting cognitive emotional functioning in the prior two weeks
Moderate-severe congestive heart failure, significant hepatic or renal disease, COPD requiring oxygen
Acute medical condition other than cancer, acute exacerbation of a chronic condition, or medical-surgical treatment affecting ability to tolerate assessment or that compromised ability to participate
BrCA=Breast Cancer; CNS=Central Nervous System; COPD=Chronic Obstructive Pulmonary
Disease; CVA=Cerebrovascular Accident (stroke); MS=Multiple Sclerosis; Hx=History.
Procedure
Data were collected at three time points. For chemotherapy patients, these were prior to
beginning the first chemotherapy cycle (baseline), three months after completion of
chemotherapy (first follow-up), and nine months after completion of chemotherapy (second
follow-up). For the non-chemotherapy participants, the post-operative baseline evaluation
occurred at approximately the same time following surgery as for chemotherapy patients, while
the two follow-up examinations were scheduled six months and fifteen months later. Healthy
controls were examined at the time of enrollment, and again after intervals of six and fifteen
months. BrCA+ patients received a combination of anthracycline-based chemotherapies, taxanes
17


or some combination of the two, which are considered standard chemotherapy treatments. The
protocol involved collecting medical history as well as demographic, QOL, neuropsychological,
imaging, and molecular (cytokine) data. Here we report on the demographic, imaging (MRI),
self-report, and pro-inflammatory cytokine data.
Subjects with BrCA were recruited from one of three organizations for this study. These
included: 1) Rocky Mountain Cancer Centers (RMCC), which has 13 oncology clinics within 1.5
hours of the Denver metro area along the Front Range of Colorado; 2) the University of
Colorado Hospital Breast Center (UCHBC); and 3) Kaiser Permanente of Denver. Some patients
volunteered in response to brochures placed in clinic office and exam rooms, but initially most
were approached about participation in the study by physicians and nursing staff. Patients who
expressed interest in participation were contacted by phone by study nurses and screened for
eligibility. If a patient met eligibility criteria, the study was explained to her, she was told she
would be paid for her participation, and verbal consent was obtained. It was then explained that a
random subset of eligible participants would be chosen to undergo a MRI scan at each of the
time points of assessment. MRIs were conducted at the University of Colorado Denver Brian
Imaging Center. Participants were paid $50 for each testing session (for a total of $150 if they
completed the study). Participants randomly chosen to receive MRIs were given an additional
$50 for each MRI visit. Patients were then scheduled for a baseline appointment, during which
we obtained written consent and conducted study interviews.
The interviewers were registered nurses with experience in health services research and
home health care. They administered all neuropsychological, functional, QOL, and mood
measures, and drew blood at the beginning of each visit. Interviews took place in participants
homes unless they preferred to come to the oncology clinic. Over 95% of all clinical data
18


collection was done in the home. Following clinical interview, the MRI participants were
scheduled to receive an MRI scan as close to the interview as possible. MRI scans were
conducted at the University of Colorado Health Sciences Brain Imaging Center.
Magnetic resonance imaging (MRI) was performed on a 3.0 Tesla GE Signa scanner
(HDx release). For each subject, a T-l weighted spoiled gradient recalled (SPGR) data set (TR
10ms, TE 2.1ms, flip angle 10, FOV 256x256, yielding 124 sagittal slices with a defined voxel
size of .86x.86x.86mm, resampled to lxlximm) was acquired. An Eclipse 3.0 T 94 quadrature
head coil was used. Inspection of individual T1 MR-images revealed no gross morphological
abnormality for any participant.
Measures
Gray Matter Volume: GM volume was obtained using a standardized procedure
described previously (Ashburner, 2007). Baseline Tl-weighted structural images were first co-
registered with the follow-up scans. Co-registration involves aligning images, based on their
edges, to allow for more accurate segmentation. Next, the images were segmented into white
matter (WM), gray matter (GM) and cerebrospinal fluid (CSF) using the new segment function
in Statistical Parametric Mapping, version 8 (SPM8; Wellcome Department of Imaging
Neuroscience, London, United Kingdom). SPM8 is free, open-source software that is available
for download at: http://www.fil.ion.ucl.ac.uk/spm/software/spm8/. The resulting GM segments
were then processed using the VBM8 toolbox within SPM8 (http://dbm.neuro.uni-iena.de/vbm).
Using the diffeomorphic registration algorithm (DARTEL) described by Ashburner
(Ashburner, 2007), the individual GM and WM segments in native-space were then nonlinearly
normalized to the DARTEL-Template supplied with the VBM8 toolbox (see
http://dbm.neuro.uni-jena.de). The voxel values of the normalized tissue segments were then
19


multiplied (modulated) with the nonlinear component of the Jacobian determinant, which was
derived from the aforementioned normalization step. The resulting GM segments thus preserve
the local GM volume as identified in native space corrected for total brain size (Buckner et al.,
2004). A quality check was performed using tools from the SPM Toolbox and individual visual
assessment, which yielded no artifacts or failed segmentation and normalization of the data. Prior
to statistical analysis the segments were smoothed with an 8 X 8 X 8mm FWHM kernel.
Normalized, smoothed data were then entered into statistical analyses within SPM8. Significant
regions of interest (ROIs) were extracted using the open-source software platform MARSBAR
v0.44 (http://marsbar.sourceforge.net/). Extracted GM volume values were then analyzed further
in SPSS version 22. A priori ROIs were also created using MARSBAR v0.44. These ROIs were
created as 10mm spheres, and were centered based on Montreal Neurological Institute (MNI)
coordinates found previously in the literature. Specifically, we looked at GM volumes found to
be altered over the course of chemotherapy treatment by in two separate studies (McDonald et
al., 2013; McDonald et al., 2010).
Cytokine Assays: Blood was drawn at each timepoint, prior to the examination;
ethylenediamine tetraacetic acid (EDTA) was used to prevent clotting, and the blood samples
were then spun down, and plasma was collected and frozen at -20 C. Quantitative enzyme-
linked immunosorbent assays (ELISA) were done to calculate peripheral IL-1, IL-6, TNF-a, and
C-reactive protein levels. The first 3 are PICs, while C-reactive protein is a general indicator of
inflammation. Because of diurnal variability in cytokine levels, blood was drawn for these assays
at about 9 AM and about 2 PM on the day of a cognitive exam. For older persons, research
suggests that the circadian pattern is relatively flat for both IL6 and TNF-a (Vgontzas et al.,
2003), so there is not much variation between the 2 levels.
20


Center for Epidemiologic Studies Depression Scale (CES-D): Depression was evaluated
using the Center for Epidemiologic Studies Depression Scale (CES-D); (Radioff, 1977). Because
the CES-D reflects cognitive and affective symptoms rather than somatic symptoms of
depression, it is highly recommended for use with patients experiencing medical problems (Liu
et al., 2012).
The CES-D is a 20-item self-report measure. Self-reported responses were scored on a 4-
point Likert scale for how often participants endorsed the items over the past week, rated by
participants from Rarely or None of the Time (<1 day) to Most or All of the Time (5-7 days).
Items in the measures include 16 questions that are scored summed such as (e.g. I was bothered
by things that dont usually bother me), and 4 items that are reverse-scored (e.g. I felt that I
was just as good as other people). Research demonstrates that the CES-D is both a valid and
reliable instrument that can be used in research to screen for common symptoms of major
depression. Internal consistency using coefficient alpha is estimated to be .85 for the general
population and .90 in patient samples (Radloff, 1977). Estimates of test-retest reliability ranging
from two weeks to twelve months fall in the moderate range of .45 to .70 (Radloff, 1977).
Concurrent validity of the CES-D was evaluated by determining the degree to which
CES-D scores were in agreement with other measures of depression. The CES-D was found to
have correlations ranging from .50 to .80 with the Hamilton Rating Scale, .30 to .80 with the
Raskin rating scale, .40 to .50 with the Lubin Depression Adjective Checklist, .60 with the
Bradburn Affect Balance Scales Negative Affect and Positive Affect Scales, and .43 with the
Cantril life satisfaction ladder (Radloff, 1977). Thus, this measure shows good convergent
validity with other measures of depression and negative affect. The discriminant validity of the
21


CES-D, however, was found to be less successful in distinguishing between depression and other
types of emotional responses, such as anger, fear, and boredom (Radioff, 1977).
State-Trait Anxiety Inventory (STAI): Current anxiety (state anxiety) and trait anxiety was
assessed using the STAI (Barnes, Harp, & Jung, 2002). This scale consists of 20 statements
about how the person feels at the moment (state), and 20 items for assessing trait anxiety.
State-anxiety items include, I am tense, and I feel nervous. Trait-anxiety items include, I
worry too much over something that doesnt matter. Test-retest reliability coefficients on initial
development (Barnes et al., 2002) ranged from 0.31 to 0.86, with intervals ranging from 1 hour
to 104 days. Because the state anxiety scale tends to detect brief states, test-retest coefficients
were lower for the state anxiety as compared to the trait anxiety (Barnes et al., 2002).
Throughout the development of the overall STAI (state and trait subscales), more than 10,000
adults and adolescents were tested. To assess content validity, items were selected from other
anxiety measures on the basis of strong associations with the Taylor Manifest Anxiety Scale
(Taylor, 1953). The overall correlation between the STAI and this measure was 0.73,
demonstrating concurrent validity. Validity of the State subscale alone was originally established
from testing in situations characterized by state high stress, such as classroom examinations and
military training programs. This subscale was shown to be a valid measure of state-anxiety, and
distinguished from trait-anxiety (Julian, 2011).
Self-report Items Related to Stress and Perceived Memory Problems: Participants were
given two self-report items related to patients perceived effects of stress on their health
associated with their illness and their perception of cognitive problems, which are most
important for the present study and in line with our research hypotheses. One of the self-report
items comes from the Functional Assessment of Cancer Therapy-Breast (FACT-B). This self-
22


report question states: During the past 7 days, I worry about the effects of stress on my illness,
and responses were scored on a 5-point Likert scale ranging from 0 not at all, to 4 very
much. A second item was administered as part of a self-report questionnaire used to assess
aspects of memory and concentration. This question states: During the past two weeks I have
noticed that my concentration and memory are not what they used to be. Responses were also
scored on 5-point Likert scale ranging from 0 not at all, to 4 very much. This question is
not part of a validated measure; therefore, interpretations of this self-report item should be
interpreted with caution. Analyses including these two items were intended to be exploratory in
order to provide potential implications for future research.
Data Analysis
Data were analyzed using IBM SPSS Statistics version 22 (SPSS Inc., 2013) and
Statistical Parametric Mapping version 8 (SPM8; Wellcome Department of Imaging
Neuroscience, London, United Kingdom).
Preliminary Analyses
Prior to our main analyses, descriptive statistics (i.e. means, standard deviations,
frequency distributions) were calculated to describe the sample, and an analysis of variance
(ANOVA) was conducted to compare the three groups on baseline demographics. Continuous
variables (e.g. GM volumes and cytokine values) were plotted as histograms in order to check
for normal distributions; appropriate transformations were applied as necessary to correct for
skewed distributions.
Hypothesis Testing
AIM 1: Hypothesis 1.1 was tested using an independent samples t-test within SPM8 to compare
whole-brain, smoothed GM images between groups (HC vs. BrCA+ and BrCA-) at baseline. To
23


test hypothesis 1.2, baseline GM volumes in a priori ROIs were extracted from SPM8 and
entered into SPSS. GM volumes in a priori ROIs and PIC levels were then compared across the
three groups using a one-way analysis of variance (ANOVA). Hypothesis 1.3 was tested using a
repeated measures multivariate analysis of covariance (MANCOVA) within SPSS. GM volumes
in a priori ROIs were used to determine changes in regional brain GM between the BrCA+, the
BrCA-, and the HC groups over the course of treatment. This analysis included one within-
subjects factor (time) and one between subjects factor (group). Age was entered as a covariate in
order to control for GM decreases associated with the aging process. A priori ROIs were applied
as dependent variables and compared between the three groups, across time.
AIM 2: To test hypotheses 2.1 through 2.5, change scores were calculated for all variables of
interest (GM volume in a priori regions, PICs, CES-D scores, and STAI scores) between the first
two time points (baseline and after 3 months of chemotherapy) and between the second and third
time points (3 months and 9 months after chemotherapy). To calculate the change scores, the
earlier time points were subtracted from the latter time points (e.g., 3 month value subtracted
from baseline value, or post minus pre). These change scores were then entered into
bivariate correlation analyses within SPSS and included only the BrCA+ group. Finally, to test
the exploratory analysis (hypothesis 2.6), change scores on the two self report items were entered
into bivariate correlations with the change scores of GM volume for a priori ROIs.
24


CHAPTER III
RESULTS
Primary Analyses
Analyses were conducted to address the two Specific Aims and test the above
hypotheses. Table 2 presents demographics, baseline clinical characteristics, and results of the
preliminary ANOVA, including baseline depression scores on the CES-D, and baseline state and
trait anxiety scores on the STAI.
Table 2. Baseline Demographics and Clinical Characteristics
Variable HC BrCA+ BrCA- p-value
N=59 18 24 17
Age (years) 63.33 (8.28) 58.1 (5.6) 68 (5.4) /K0.001*
Education (years) 16.9 (2.6) 15.8 (2.3) 15.3 (2.3) p= 0.149
Ethnicity
White/Caucasian 18 (100%) 23 (95%) 14 (82%)
Black 0 1 (5%) 2 (13%)
Asian 0 0 1 (5%)
CES-D 27.6 (6.6) 31.3 (7.9) 29.3 (9) p=Q.29\
STAI-S 36.7 (6.9) 29.6 (8.7) 28.5 (9.5) p=0.551
STAI-T 29.1 (7) 31.4 (7.7) 30.6(10) p= 0.674
Values presented as: mean (standard deviation) or frequencies (percentage); HC = Healthy Controls, BrCA+ =
Breast cancer patients in the chemotherapy group, BrCA- = Breast cancer patients in the non-chemotherapy group,
CES-D = Center for the Epidemiological Study of Depression scale, STAI-S = State-Trait Anxiety Inventory (State
scale), STAI-T = State-Trait Anxiety Inventory (Trait scale).
* = significantly different based on ANOVA of baseline values.
25


Evaluation of histograms showed a reasonably normal distribution for all continuous
variables included in the study, and thus no transformations were necessary prior to further
analysis. Results of the preliminary ANOVA showed that the ages among the three groups were
significantly different, F (2,56) = 11.95,/) <0.001. Tukeyspost-hoc test showed that the HC and
BrCA- groups were significantly older than the BrCA+ group (mean difference = 5.26,p = 0.031
and mean difference = 9.95,p< 0.001, respectively), but were not significantly different from
each other in age (p = 0.091). No other demographic variables were significantly different among
the groups (all p > 0.05). As hypothesized, there were no significant regional GM volume
differences at baseline between groups, when comparing whole-brain GM volumes in SPM8.
Results of the ANOVA also revealed no significant group differences at baseline for GM
volumes in a priori ROIs or PICs (all p > 0.05). Results of the MANCOVA are presented in
Table 3.
Table 3. Results of MANCOVA Analysis
Effect Wilks' Lambda F Error df Sig. Partial Eta Squared Observed Power
Between Subjects Age 0.635 2.938 46 0.008 0.365 0.934
Group 0.437 2.619 92 0.001 0.339 0.993
Within Subjects Time 0.564 1.590 37 0.114 0.436 0.807
Time Age 0.533 1.802 37 0.064 0.467 0.865
Time Group 0.269 1.909 74 0.01 0.481 0.993
Dependent Variables = a priori Gray Mater Regions of Interest
There was a significant overall between-subjects effect of Age on GM volumes in the a
priori ROIs, F(9,46) = 2.94,p = 0.008; Wilks A = 0.635, partial r|2 = 0.365, observed power =
0.93. There was also a significant overall between-subjects effect of Group on GM volumes in
theapriori ROIs, F( 18,92) = 2.62, p = 0.001; Wilks A = 0.437, partial r|2 = 0.339, observed
26


power = 0.993. The within-subjects interaction between Age and Time was insignificant (p =
0.064). There was a significant within-subjects interaction between Group and Time on GM
volumes in the a priori ROIs when controlling for age, A(36,74) = 1.91,/? = 0.01; Wilks A =
0.269, partial r|2 = 0.481, observed power = 0.993. Post-hoc pairwise comparisons were used to
further evaluate the observed differences between groups over time and specify the specific
regions adjusted by these factors. In order to correct for multiple comparisons, a Bonferroni
correction was applied. Significant pairwise mean differences were found in the right precentral
gyrus (PCG), and in the left superior frontal gyrus (SFG). Specifically, the BrCA+ group showed
significantly less mean GM volume in the right PCG compared to the BrCA- group (mean
difference = -0.04,/? = 0.031). Additionally, patients in the BrCA- group showed significantly
more mean GM volume in the left SFG compared to the HC group (mean difference = -0.02,/? =
0.047).
Results of the bivariate correlations among change scores between baseline and 3 months
post-chemotherapy treatment are presented in Table 4. The bivariate correlations of these change
scores showed that IL6 levels were negatively correlated with GM volume in the right middle
frontal gyrus (MFG; r = -0.43,/? = 0.045). A scatter plot graph of this result is presented in
Figure 1.
27


00
(N
Table 4. Results of Correlation Analysis Among Change Scores (3 months post-treatment baseline)
l|phg R MFG R SFG R Precentral gyrus L MFG L SFG L Cingulate CESD STAI State STAI Trait CRP TNF-A IL6 IL1B
L PHG 1.00
R MFG -.04 1.00
R SFG -.04 .42* 1.00
R Precentral gyrus .172 .62* .33 1.00
L MFG .186 -.05 -.06 .34 1.00
L SFG .03 .26 .68* .41* .25 1.00
L Cingulate .32 -.2 -.32 -.06 -.07 .13 1.00
CESD .05 -.15 .12 -.37 -.05 .13 .16 1.00
STAI State -.1 .12 -.15 -.03 .07 .03 .06 .08 1.00
STAI Trait .05 -.03 -.36 -.24 -.22 -.21 .15 .31 .80* 1.00
CRP -.22 -.43* -.34 -.27 -.02 -.25 .11 -.13 -.37 -.28 1.00
TNF-A .144 -.15 .21 .07 .25 .13 -.02 .23 .38 .19 .00 1.00
IL6 .33 -.43* -.17 .07 .20 -.12 .20 -.03 -.03 .02 -.06 .02 1.00
IL1B -.06 .03 .19 -.31 .03 .13 .10 .12 .15 -.05 -.24 .14 -.03 1.00
Values presented are Pearsons R-values.
* = Significant at p<0.05
N =24
PHG = parahippocampal gyrus, MFG = middle frontal gyrus, SFG = superior frontal gyrus, CESD = Center for the Epidemiological Study of Depression scale,
STAI = State-Trait Anxiety Inventory, CRP = C-reactive protein, TNF-a = tumor necrosis factor alpha, IL6 = interleukin-6, IL1B = interleukin 1 beta.


.05-
IL6_A (post-pre)
Figure 1. Correlation Between IL6 and Right Middle Frontal Gyrus GM Volume
Blue dot represents region of interest placed in the middle frontal gyrus (image is in radiological format).
Post-pre = change scores computed between baseline (pre) and 3 months (post); R MFG = Right Middle Frontal
Gyrus Gray Matter volume, IL6_A = Interleukin 6 value.
Results of the bivariate correlations among change scores between 3 months and 9
months post-chemotherapy treatment are presented in Table 5. Results of this analysis showed
that CES-D scores were negatively associated with GM volume in the Left Cingulate cortex (r =
-.408,/) = .048; Figure 2), and positively associated with changes in STAI scores on both
subscales (state: r = 0.636, p = .001; trait: r = 0.678,/) <0.001). Changes on the State-anxiety
portion of the STAI measure showed a positive correlation with changes in IL6 (r =
0.528,/) = 0.02). Finally, changes in IL6 were also negatively correlated with GM volume in the
left parahippocampal gyrus (r = -0.632,p = 0.004; Figure 3).
29


o
Table 5. Results of Correlation Analysis Among Change Scores (9 months 3 months post-treatment)
L PHG RMFG R SFG R Precentral gyrus L MFG L SFG L Cingulate CESD STAI State STAI Trait CRP TNF-a IL6 IL1B
L PHG 1.00
RMFG .288 1.00
R SFG -.26 .35 1.00
R Precentral gyrus .22 .45* .3 1.00
L MFG .02 .39 .41* .34 1.00
L SFG .04 .14 .49* .39 .25 1.00
L Cingulate -.12 .29 .1 -.13 .07 .10 1.00
CESD .01 -.19 -.19 -.37 -.04 .13 .16 1.00
STAI State -.12 -.13 -.29 .01 .03 .06 .13 .08 1.00
STAI Trait -.03 -.14 -.27 -.24 -.24 -.21 .16 .31 .80 1.00
CRP -.4 -.43 -.39 -.27 -.05 -.25 .13 -.12 -.37 -.28 1.00
TNF-A .17 -.19 .05 .07 .21 .13 .02 .23 .38 .19 .00 1.00
IL6 -.63* -.41 -.22 .07 .20 -.12 .20 -.03 -.03 .02 -.06 .02 1.00
IL1B .03 .01 -.18 -.31 .02 .13 .10 .12 .15 -.05 -.24 .14 -.03 1.00
Values presented are Pearsons R-values. PHG = parahippocampal gyrus
* = Significant at p<0.05
N = 24
PHG = parahippocampal gyrus, MFG = middle frontal gyrus, SFG = superior frontal gyrus, CESD = Center for the Epidemiological Study of Depression scale,
STAI = State-Trait Anxiety Inventory, CRP = C-reactive protein, TNF-a = tumor necrosis factor alpha, IL6 = interleukin-6, IL1B = interleukin 1 beta.


.075
O
.050-
0
w
Q.
----1----------1----------1-----------1----------1------------1----------1---
20.00 -10.00 .00 10.00 20.00 30.00 40.00
CES-D (post-pre)
Figure 2. Correlation Between CES-D and Left Cingulate GM Volume
Blue dot represents region of interest placed in the Left Cingulate Gyrus (image is in radiological format).
Post-pre = change scores computed between 3 months (pre) and 9 months (post); L Cingulate= gray matter volume
in this region; CES-D = Center for the Epidemiologic Studies Depression Scale.
31


IL6_A (post-pre)
Figure 3. Correlation Between IL6 and Left Parahippocampal Gyrus GM Volume
Blue dot represents region of interest placed in the Left Parahippocampal Gyrus (image is in radiological format).
Post-pre = change scores computed between 3 months (pre) and 9 months (post); L parahippocampal gyrus = gray
matter volume in this region; IL6 = interleukin-6.
32


Exploratory Analysis
Results of the bivariate correlations among change scores between baseline and 3 months
showed that there was a significant negative correlation between CRP and perceived effects of
stress on health associated with treatment as measured by the single item from the FACT-B (r = -
.475,/) = 0.026) as well as CRP and self-reported problems with memory (r = -.659,/> = .001). It
was also found that GM volume in the right middle frontal gyrus (MFG) was positively
correlated with the measure of stress associated with treatment (r = 0.434,/) = .034).
Results of the bivariate correlations among change scores between 3 and 9 months post-
chemotherapy showed that changes in CES-D scores were positively associated with changes in
perceived effects of stress associated with treatment (r = 0.518,/) = 0.01). Further, changes in
both state- and trait-anxiety, as measured by the STAI, were positively correlated with memory
complaints in these patients (r = 0.419,/) = 0.041 and r = 0.429,/) = 0.036, respectively).
33


CHAPTER IV
DISCUSSION
The current study tested the notion that BrCA patients may experience reductions in brain
GM volume following chemotherapy treatment. To this end, relationships between
chemotherapy treatment, brain GM volume, markers of inflammation, and measures of
depression and anxiety were evaluated in a sample of BrCA patients undergoing chemotherapy
and compared with BrCA patients who did not undergo chemotherapy, and HC.
The preliminary analysis showed that the groups differed significantly by age (Table 2).
Although the groups were significantly different from each other on age, the BrCA+ patients had
a mean age of 58 years, which is comparable to the ages of participants of previous studies
(Pomykala, de Ruiter, et al., 2013). Due to the significant difference in ages, age was entered as a
covariate in the MANCOVA analysis to help control for the effects of age on GM volume.
The results support the first hypothesis, that there would be no significant differences
between the groups at baseline (prior to treatment) on measures of regional GM volume and in a
priori GM ROIs. Further, there were no significant differences on measures of PICs at baseline.
This has been observed in previous studies (McDonald et al., 2013; Pomykala, de Ruiter, et al.,
2013) and thus serves to replicate what has been found previously.
Results of the MANCOVA showed that there were overall between-subjects effects of
both Group and Age on the GM volume of ROIs included in the analysis. Further, there was also
a significant within-subjects interaction effect between Group and Time. These results are not
wholly surprising; we expected to find differences between groups over time in the a priori
ROIs. Also, following the preliminary ANOVA, and based on previous work (McDonald et al.,
2013), age was expected to have an effect on GM volume over time and thus was controlled for
34


in subsequent analyses comparing groups. What was surprising about these results, and
contradictory to findings from other studies, is that GM volumes in the a priori ROIs showed
that there were no significant differences between the BrCA+ and HC. The only ROI wherein
BrCA+ patients showed significantly less GM volume than the BrCA- group was in the
precentral gyrus. In this sample, on average, HC and BrCA- patients had less GM volume than
the BrCA+ group in the a priori ROIs included in the analyses. These results are inconsistent
with previous reports of decreased GM volumes in patients with BrCA who have undergone
chemotherapy in these specific regions that were tested in the present study (McDonald et al.,
2013; McDonald et al., 2010). It is possible that differences in sample size, VBM processing
techniques, and measurement intervals could account for these differences. However, these
findings challenge the notion that chemotherapy causes significant differences in brain GM by
testing the specific regions where such differences were found previously. Further analyses of
these specific regions could help clarify this discrepancy. The precentral gyrus did show
significant GM reductions in the BrCA+ group over the course of treatment when compared with
the BrCA- group. This finding is partially in line with our hypothesis, but still failed to show
significant differences between the HC group and the BrCA+ patients. Moreover, this result in
the precentral gyrus is difficult to interpret since basic motor functioning, as mediated by the
precentral gyrus, does not have a logical relationship to chemotherapy treatment other than the
fact that this region was found to be decreased following chemotherapy treatment in previous
research (McDonald et al., 2010). It may be possible that measures of motor functioning (e.g.,
the Purdue Peg Test) could show some relationship with GM volume changes in the precentral
gyrus. This should be evaluated in a future analysis or research study. This same research by
McDonald and Colleagues (2010) also showed a rebound effect of GM values over time when
35


observed at 1 year post-treatment. It could be that if we had an additional measurement point at 1
year or longer we could have observed the same sort of effect in this sample.
When evaluating the second aim, that there would be relationships between changes in
GM in the a priori ROIs and both measures of inflammation and scores on depression and
anxiety scales, we found several relationships that supported the hypotheses both between
baseline and 3 months and between 3 months and 9 months post-treatment. With regard to
changes between baseline and 3 months, there was a significant negative correlation between
GM volume in the right MFG and levels of IL-6 (Figure 1). These findings are in line with our
hypothesis, but are somewhat inconsistent with previous work evaluating relationships between
GM changes and markers of inflammation (Kesler et al., 2013). Kesler and colleagues (2013)
found a positive relationship between hippocampal GM volume and IL6 levels following
chemotherapy treatment, and a negative relationship between hippocampal GM volume and
TNF-a levels. We did not find any significant relationships with changes in TNF-a levels over
the course of treatment in the BrCA+ group. It could be that these opposing findings are due to
both the anti- and pro-inflammatory roles that IL-6 plays in the immune system. For example,
levels of IL-6 could be increased initially due to their pro-inflammatory response, but then levels
could decrease or increase because of the various roles this cytokine plays in the inflammation
process. The study by Kesler and colleagues was limited by the fact that they were not able to
determine the acute effects of chemotherapy treatment on these measures due to the long
duration post-treatment. This is an advantage of the present study over previous research. We did
not find significant relationships between GM volume changes and changes in depression or
anxiety between baseline and 3 months, or between changes in PIC levels and depression and
anxiety during this interval.
36


When evaluating relationships between changes on levels of GM volume, PICs, and
levels of anxiety and depression between 3 months and 9 months post-treatment, we observed
several correlations that supported our hypotheses, as well as one finding that was in
contradiction to our hypotheses. Results of these change score correlations showed that changes
in the left Cingulate cortex were negatively associated with changes in CES-D scores, such that
less GM volume in the cingulate was associated with increased depressive symptoms. Previous
VBM research has found significant reductions in the cingulate gyrus in patients experiencing
depression, when compared to HC (Rodriguez-Cano et al., 2014). This is in line with our finding
that reductions in GM volume in the cingulate may be associated with increased depressive
symptoms. However, another recent study found increased GM volume in this region in a sample
of un-medicated depressed patients (Yang et al., 2015). Thus, the relationship between GM
volume in the left cingulate gyrus and levels of depression as measured by the CES-D
contributes to this knowledge base in support of a negative relationship between these two
variables. Future research should test this region and its association with depressive symptoms,
specifically in BrCA patients undergoing various treatments that could affect quality of life.
Contradictory to our hypothesis, there was a significant positive correlation between left
Cingulate GM volumes and both state and trait anxiety as measured by the STAI. Previous
research has shown reductions in GM volumes of patients with anxiety when compared to
healthy controls (Moon et al., 2014). However, longitudinal effects on GM volume associated
with symptoms of both state and trait anxiety are less well understood. One possible explanation
for the findings in the present study is a nonlinearity, such that levels of anxiety did not reach
clinically significant levels (as defined by DSM-5 criteria), and thus do not show the effects
previously demonstrated in the review by Moon and Colleagues (2014). This finding also
37


suggests that there may be inverse effects of depressive and anxious symptoms in the left
Cingulate cortex, such that depression is associated with decreased GM volume, while symptoms
of anxiety are associated with increased GM volume. Future longitudinal studies should address
this question in order to further evaluate the effects of psychological characteristics on GM
volumes in the cingulate cortex.
Further, we observed a negative relationship between GM volume in the
parahippocampal gyrus and levels of IL6. This finding further supports our hypotheses
predicting a negative relationship between levels of inflammation and GM volumes. In the study
done by Kesler and Colleagues (2013), lower hippocampal volumes were associated with lower
levels of IL6. This is contrary to the findings in the present study. Again, this may be due to the
dual role (both pro- and anti-inflammatory) IL6 plays in the inflammatory process. The
parahippocampal gyrus is a region that encompasses the hippocampus and is involved in
memory encoding and retrieval. Thus, it is possible that increased levels of IL6 could be related
to measures of memory storage and retrieval. Future research should directly test this hypothesis.
One can speculate that levels of IL6, which also showed contradictory findings, could moderate
the decrease in GM volume in the cingulate gyrus in relation to reductions in STAI scores.
Finally, our exploratory analyses showed a negative relationship between CRP levels and
perceived stress in the BrCA+ group between baseline and 3 months post-treatment. Further, a
positive relationship was observed between changes in right MFG GM volume and perceived
stress as measured by the single self-report item. These results did not support our exploratory
hypotheses, and seem to contradict previous research as well. Between 3 months and 9 months
post-treatment, it was found that CES-D scores were positively associated with changes in
perceived stress associated with treatment. Further, changes in both state- and trait-anxiety were
38


positively correlated with memory complaints in these patients. These results are in support of
our proposed exploratory hypotheses and seem to be in line with previous research evaluating
cognitive complaints in this population (Simo, Rifa-Ros, Rodriguez-Fomells, & Bruna, 2013). It
seems that when measuring psychological self-report variables (e.g., perceived effects of stress
on health, depression, anxiety), differences occur later in treatment, and the acute effects are not
as prominent in this sample.
Strengths and Limitations
There are several limitations that must be considered when drawing inferences from the
results of the present study. One limitation is how to interpret GM changes. Although the exact
biological mechanism remains unclear, GM variability can be due to a number of different
factors such as differences in cell volume, synaptic densities, or blood flow and interstitial fluid
(Gage, 2002). Because of this variability in potential mechanisms leading to changes in GM, it is
difficult to draw valid conclusions in regard to the outcomes that correlate with such a measure.
Although there is no way to best interpret GM changes, the present analyses included the
cytokine data in order to allow for more direct associations between chemotherapy,
inflammation, and GM changes. Cytokines are a choice candidate as a related biomarker, as
some cytokines have been shown to pass through the blood-brain barrier and could have
potential consequences on brain neurogenesis and function (McDonald et al., 2013).
Further, in order for VBM to be valid, several assumptions must be met (Ashbumer &
Friston, 2000). First, the initial segmentation must correctly identify GM and WM. This is
determined by several factors such as image quality and level of contrast between GM and WM
tissue (Ashbumer & Friston, 2000). In order to verify that segmentation occurred correctly, all
GM segmentations were visually inspected and plotted using a box plot to determine if there
39


were any outliers in the data set. A second assumption is that all confounding effects will be
eliminated or accounted for in the statistical model (Ashbumer & Friston, 2000). We attempted
to account for this assumption by co-registering the data within-subjects, and then between-
subjects in order to reduce the noise produced over time, and across subjects data. Co-
registration is a method used to correct for motion between images, thus reducing the potential
noise in the data between time points. Additionally, thresholds were applied to exclude any GM
values that are <0.1, which indicates it may be an incorrectly labeled WM voxel. A third issue
related to the validity of VBM is the assumptions held by the statistical tests. It is important to
understand how the distribution of the data will affect the statistical tests. Parametric tests
assume the data are normally distributed. If data are not normally distributed, then a non-
parametric test is indicated in order to maintain the validity of the VBM analysis (Ashbumer &
Friston, 2000).
Another limitation of this longitudinal design is the history of the participants. During the
intervals between assessment visits, participants may have experienced a wide array of different
personal and emotional factors that could lead to changes in the measures under study. However,
in the present study, the measures selected represent constructs that are not thought to be as
variable as other potential psychological outcomes. Therefore, experiences outside the range of
testing likely had minimal influence on the final results.
A third limitation is that peripheral cytokines are not directly related to GM changes in
the CNS. However, there is evidence that inflammation in the periphery can have effects on the
CNS via a weakened BBB as mentioned above (Merrill & Benveniste, 1996). Additionally, the
brain produces cytokines in response to central inflammation (Merrill & Benveniste, 1996). It is
not feasible to determine definitively whether PICs in brain tissue directly cause GM MRI
40


changes, especially among women in this population. The data permitted, however, assessment
of the magnitude of the association between the effects of chemotherapy, GM changes, and
burden of inflammation, and yield additional relevant data that adds to what is currently known
about these relationships.
A final limitation is that the intended sample size to be included in this study was
significantly reduced in association with a 19% cut in the budgets of new research grants
awarded by the National Institute on Aging (NIA) in 2003. Originally, it was proposed that we
would collect neuroimaging data from all three time points for every participant, but the budget
shortfall made this impossible, leaving the study slightly under-powered for these analyses.
Despite the limitations mentioned above, there were several strengths of this study. First,
despite the need to reduce the use of imaging in response to the NIH budget cuts, we were able to
obtain a relatively large sample for an imaging study and should be able to draw valid inferences
from these data. Second, the longitudinal design allows us to evaluate changes over time, and
this is one of the first studies to look at these changes in relation to PICs, considered to be
markers of inflammation. Additionally, the prospective design furthered our understanding of the
effects of chemotherapy on the brain in women with BrCA.
A final strength of this study is that the data collection involved multiple components
(e.g., psychosocial data, brain imaging, markers of inflammation). Each of these constructs
contributes differently to the overall outcomes in this population. Therefore, the data collection
methods employed in the present study attempted to answer questions from a multi-dimensional
approach, which would not be possible without the multiple components involved in the study.
41


Future Directions
Given the results of the present study, future research and analyses of these data can seek
to replicate and extend on these findings. For example, evaluation of specific cognitive measures
across time in this population could shed further light on the relationship between changes in
GM volume in these specific regions and self-reported memory complaints in patients
undergoing chemotherapy treatment. A future study could also look at a longer duration
following chemotherapy treatment in order to evaluate the longer-term effects of chemotherapy
treatment specially looking at regions seen in previous work. Another interesting study could
be to examine scores on task of motor processing (e.g. Trails test or Purdue Pegboard test) in
order to further evaluate the GM changes in the precentral gyrus. It may be the case that
alterations of GM volume in the precentral gyrus are associated with impaired processing speed
and/or motor coordination. Future research should also continue to look at the relationship
between the cingulate gyrus and measures of depression. Decreased GM volume in this region
has consistently been shown in the past in patients with BrCA and with mood disorders such as
depression, and this fits with the integrative and introceptive functioning of the cingulate gyrus
in the context of depressive symptoms. Finally, more studies should try to replicate GM changes
in specific regions found previously.
Conclusions
The relationship between GM volume changes and chemotherapy treatment has been a
relatively consistent finding in the literature. The most consistent findings have been in areas in
the prefrontal cortex with other areas found less consistently (e.g. hippocampus, cingulate) to
change with chemotherapy treatment. What the research has shown thus far is that the
relationship between GM volume changes and chemotherapy is very complex, and involves
42


numerous factors at the biological level of the individual cells and at the psychological level of
the patient as a whole. This study attempted to add to the literature and help further our
understanding of the relationships between chemotherapy and GM volume changes by testing
specific regions that have been found in previous work. One conclusion that can be made is that
GM changes are difficult to replicate. Sampling error becomes an issue when different MRI
scanners are used and various processing techniques are applied to accomplish the task of
quantifying GM volume. However, VBM studies in this population have not sought to replicate
findings in any specific regions. More research is needed that seeks to replicate previous GM
changes in specific regions that have been found to change with chemotherapy treatment. It
seems that different regions show different responses to the treatment process some may show
increases while others show decreases. VBM as a method is limited in that there is no clear
biological explanation for these changes and discrepancies in results.
The more prominent conclusion from the present analyses is that there are relatively
strong negative associations between changes in GM volume and measures of cytokines, namely
IL6. IL6 has an inhibitory effect on TNF-a, and in the present study there were no significant
relationships between changes in the GM ROIs and TNF-a. One possible explanation for this is
that BrCA+ patients with higher levels of IL6 show lower levels of TNF-a, which may fall below
a threshold that is detectable by the correlations conducted in the present study.
In conclusion, more research is needed to help bolster our understanding of the process of
chemotherapy in BrCA patients. Once these relationships are better understood, we may then
turn to clinical implications for these patients that could help promote decreased inflammation
and perhaps in turn help mitigate some of the cognitive complaints these patients experience.
However, we remain quite far from these treatment implications, and studies such as this, which
43


seek to replicate previous work, are needed in order to develop a clearer picture of the processes
of disease progression and changes on biological and psychological measures with chemotherapy
treatment.
44


REFERENCES
American Cancer Society. Breast Cancer Facts & Figures 2013-2014. Atlanta: American
Cancer Society, Inc. 2013.
Ashbumer, J. (2007). A fast diffeomorphic image registration algorithm. Neuroimage, 35(1), 95-
113. doi: 10.1016/j.neuroimage.2007.07.007
Ashbumer, J., & Friston, Karl J. (2000). Voxel-Based MorphometryThe Methods.
Neuroimage, 11(6), 805-821. doi: 10.1006/nimg.2000.0582
Bames, L. L. B., Harp, D., & Jung, W. S. (2002). Reliability Generalization of Scores on the
Spielberger State-Trait Anxiety Inventory. Educational and Psychological Measurement,
62(4), 603-618. doi: 10.1177/0013164402062004005
Bower, J. E., Ganz, P. A., Irwin, M. R., Kwan, L., Breen, E. C., & Cole, S. W. (2011).
Inflammation and behavioral symptoms after breast cancer treatment: do fatigue,
depression, and sleep disturbance share a common underlying mechanism? J Clin Oncol,
29(26), 3517-3522. doi: 10.1200/JC0.2011.36.1154
Buckner, R. L., Head, D., Parker, J., Fotenos, A. F., Marcus, D., Morris, J. C., & Snyder, A. Z.
(2004). A unified approach for morphometric and functional data analysis in young, old,
and demented adults using automated atlas-based head size normalization: reliability and
validation against manual measurement of total intracranial volume. Neuroimage, 23(2),
724-738. doi: 10.1016/j.neuroimage.2004.06.018
Cimprich, Bernadine, Reuter-Lorenz, Patricia, Nelson, James, Clark, Patricia M, Therrien,
Barbara, Normolle, Daniel, Welsh, Robert C. (2010). Prechemotherapy alterations in
brain function in women with breast cancer. Journal of Clinical and Experimental
Neuropsychology, 32(3), 324-331. doi: 10.1080/13803390903032537
45


Conroy, Susan K, McDonald, Brenna C, Smith, Dori J, Moser, Lyndsi R, West, John D,
Kamendulis, Lisa M, Saykin, Andrew J. (2012). Alterations in brain structure and
function in breast cancer survivors: effect of post-chemotherapy interval and relation to
oxidative DNA damage. Breast Cancer Research and Treatment, 137(2), 493-502. doi:
10.1007/sl0549-012-2385-x
Coussens, L. M., & Werb, Z. (2002). Inflammation and cancer. Nature, 420(6911), 860-867. doi:
10.103 8/nature01322
Dantzer, R., & Kelley, K. W. (2007). Twenty years of research on cytokine-induced sickness
behavior. BrainBehavImmun, 21(2), 153-160. doi: 10.1016/j.bbi.2006.09.006
de Ruiter, Michiel B, Reneman, Liesbeth, Boogerd, Willem, Veltman, Dick J, Caan, Matthan,
Douaud, Gwenaelle, Schagen, Sanne B. (2012). Late effects of high-dose adjuvant
chemotherapy on white and gray matter in breast cancer survivors: Converging results
from multimodal magnetic resonance imaging. Human Brain Mapping, 33(12), 2971-
2983. doi: 10.1002/hbm.21422
Dethlefsen, C., Hojfeldt, G., & Hojman, P. (2013). The role of intratumoral and systemic IL-6 in
breast cancer. Breast Cancer Res Treat, 138(3), 657-664. doi: 10.1007/sl0549-013-2488-
z
Dietrich, J. (2010). Chemotherapy associated central nervous system damage. Adv Exp Med Biol,
678, 77-85.
Gage, F. H. (2002). Neurogenesis in the adult brain. JNeurosci, 22(3), 612-613.
Goehler, L. E., Gaykema, R. P., Hansen, M. K., Anderson, K., Maier, S. F., & Watkins, L. R.
(2000). Vagal immune-to-brain communication: a visceral chemosensory pathway. Auton
Neurosci, 55(1-3), 49-59. doi: 10.1016/S 1566-0702(00)00219-8
46


Inagaki, Masatoshi, Yoshikawa, Eisho, Matsuoka, Yutaka, Sugawara, Yuriko, Nakano,
Tomohito, Akechi, Tatsuo, Group, and The Breast Cancer Survivors' Brain
MRI Database. (2007). Smaller regional volumes of brain gray and white matter
demonstrated in breast cancer survivors exposed to adjuvant chemotherapy. Cancer,
109(1), 146-156. doi: 10.1002/cncr.22368
Janelsins, M. C., Roscoe, J. A., Berg, M. J., Thompson, B. D., Gallagher, M. J., Morrow, G. R., .
. Gross, R. A. (2010). IGF-1 partially restores chemotherapy-induced reductions in
neural cell proliferation in adult C57BL/6 mice. Cancer Invest, 28(5), 544-553. doi:
10.3109/07357900903405942
Julian, Laura J. (2011). Measures of anxiety: State-Trait Anxiety Inventory (STAI), Beck
Anxiety Inventory (BAI), and Hospital Anxiety and Depression Scale-Anxiety (HADS-
A). Arthritis Care & Research, 63(S11), S467-S472. doi: 10.1002/acr.20561
Kesler, S., Janelsins, M., Koovakkattu, D., Palesh, O., Mustian, K., Morrow, G., & Dhabhar, F.
S. (2013). Reduced hippocampal volume and verbal memory performance associated
with interleukin-6 and tumor necrosis factor-alpha levels in chemotherapy-treated breast
cancer survivors. Brain Behav Immun, 30 Suppl, S109-116. doi:
10.1016/j.bbi.2012.05.017
Koppelmans, V., Breteler, M. M., Boogerd, W., Seynaeve, C., & Schagen, S. B. (2013). Late
effects of adjuvant chemotherapy for adult onset non-CNS cancer; cognitive impairment,
brain structure and risk of dementia. CritRev OncolHematol, 55(1), 87-101. doi:
10.1016/j.critrevonc.2013.04.002
47


Lai, Chien-Han. (2013). Gray matter volume in major depressive disorder A meta-analysis of
voxel-based morphometry studies. Psychiatry Research: Neuroimaging, 211(1), 37-46.
doi: 10.1016/j.pscychresns.2012.06.006
Liu, L., Mills, P. J., Rissling, M., Fiorentino, L., Natarajan, L., Dimsdale, J. E., Ancoli-Israel,
S. (2012). Fatigue and sleep quality are associated with changes in inflammatory markers
in breast cancer patients undergoing chemotherapy. Brain Behav Immun, 26(5), 706-713.
doi: 10.1016/j .bbi.2012.02.001
Mariotto, A B, Robin Yabroff, K, Shao, Y, Feuer, E J, & Brown, M L. (2011). Projections of the
Cost of Cancer Care in the United States: 2010-2020. JNCI Journal of the National
Cancer Institute, 103(2), 117-128. doi: 10.1093/jnci/djq495
McDonald, B. C., Conroy, S. K., Smith, D. J., West, J. D., & Saykin, A. J. (2013). Frontal gray
matter reduction after breast cancer chemotherapy and association with executive
symptoms: a replication and extension study. Brain Behav Immun, 30 Suppl, SI 17-125.
doi: 10.1016/j .bbi.2012.05.007
McDonald, B. C., Conroy, Susan K, Ahles, Tim A, West, John D, & Saykin, Andrew J. (2010).
Gray matter reduction associated with systemic chemotherapy for breast cancer: a
prospective MRI study. Breast Cancer Research and Treatment, 123(3), 819-828. doi:
10.1007/s 10549-010-1088-4
Merrill, J.E., & Benveniste, E.N. (1996). Cytokines in inflammatory brain lesions- helpful and
harmful. Trends Neuroscience, 19, 331-338.
Mohamed, M. M., Al-Raawi, D., Sabet, S. F., & El-Shinawi, M. (2014). Inflammatory breast
cancer: New factors contribute to disease etiology: A review. JAdv Res, 5(5), 525-536.
doi: 10.1016/j .jare.2013.06.004
48


Moon, Chung-Man, Kim, Gwang-Won, & Jeong, Gwang-Woo. (2014). Whole-brain gray matter
volume abnormalities in patients with generalized anxiety disorder. NeuroReport, 25(3),
184-189. doi: 10.1097/WNR.0000000000000100
Pomykala, K. L., de Ruiter, M. B., Deprez, S., McDonald, B. C., & Silverman, D. H. (2013).
Integrating imaging findings in evaluating the post-chemotherapy brain. Brain Imaging
Behav, 7(4), 436-452. doi: 10.1007/sll682-013-9239-y
Pomykala, K. L., Ganz, P. A., Bower, J. E., Kwan, L., Castellon, S. A., Mallam, S., .
Silverman, D. H. (2013). The association between pro-inflammatory cytokines, regional
cerebral metabolism, and cognitive complaints following adjuvant chemotherapy for
breast cancer. Brain Imaging Behav, 7(4), 511-523. doi: 10.1007/sl 1682-013-9243-2
Radloff, L S. (1977). The CES-D Scale: A Self-Report Depression Scale for Research in the
General Population. Applied Psychological Measurement, 7(3), 385-401. doi:
10.1177/0146621677001003 06
Rodriguez-Cano, E., Sarro, S., Monte, G. C., Maristany, T., Salvador, R., McKenna, P. J., &
Pomarol-Clotet, E. (2014). Evidence for structural and functional abnormality in the
subgenual anterior cingulate cortex in major depressive disorder. PsycholMed, 44(15),
3263-3273. doi: 10.1017/S0033291714000841
Simo, Marta, Rifa-Ros, Xavier, Rodriguez-Fornells, Antoni, & Bruna, Jordi. (2013).
Chemobrain: A systematic review of structural and functional neuroimaging studies.
Neuroscience andBiobehavioralReviews, 37(8), 1311-1321. doi:
10.1016/j.neubiorev.2013.04.015
Taylor, J. A. (1953). A personality scale of manifest anxiety. JAbnorm Psychol, 45(2), 285-290.
49


Vgontzas, A. N., Zoumakis, M., Bixler, E. O., Lin, H. M., Prolo, P., Vela-Bueno, A., .
Chrousos, G. P. (2003). Impaired nighttime sleep in healthy old versus young adults is
associated with elevated plasma interleukin-6 and cortisol levels: physiologic and
therapeutic implications. J Clin EndocrinolMetab, 55(5), 2087-2095. doi:
10.1210/jc.2002-021176
Wang, X. M., Walitt, B., Saligan, L., Tiwari, A. F., Cheung, C. W., & Zhang, Z. J. (2015).
Chemobrain: A critical review and causal hypothesis of link between cytokines and
epigenetic reprogramming associated with chemotherapy. Cytokine, 72(1), 86-96. doi:
10.1016/j.cyto.2014.12.006
Yang, X., Ma, X., Li, M., Liu, Y., Zhang, J., Huang, B., Ma, X. (2015). Anatomical and
functional brain abnormalities in unmedicated major depressive disorder. Neuropsychiatr
Dis Treat, 11, 2415-2423. doi: 10.2147/NDT.S93055
50


Full Text

PAGE 1

CHEMOTHERAPY TREATMENT, BRAIN GRAY MATTER AND MARKERS OF INFLAMMATION IN WOMEN WITH BREAST CANCER: A LONGITUDINAL COHORT STUDY By ANSON E. KAIRYS B.A., University of Michigan A t hesis submitted to the Faculty of the Graduate school of the University of Colorado in partial fulfillment o f the requirements for the degree of Master of Arts Clinical Health Psychology Program 201 5

PAGE 2

ii 2015 ANSON E. KAIRYS ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Arts degree by Anson E. Kairys h as been approved for the Clinical Health Psychology Program by Jim Grigsby, C hair Edward Dill Kristin Kilbourn Date: January 4 2016

PAGE 4

iv Kairys, Anson E. (M.A. Clinical Health Psychology) Chemotherapy Treatment, Brain G ray Matter, and Markers of Inflammation in Women with Breast Cancer: a Longitudinal Cohort S tudy Thesis directed by Professor Jim Grigsby ABSTRACT Breast cancer (BrCA) is associated with the highest mortality rate of any type of cancer besides that of the lung. Although the prevalence, mortality, and costs associated with breast cancer are quite high, recent advances in screening and treatment have increased long term s urvivorship. Thus, understanding the outcomes of treatment in patients with BrCA is of great importance to help be tter understand quality of life and biological markers of treatment related outcomes in this population. To this end, the present study evalua te d changes in brain gray matter volume burden of inflammation and levels of anxiety and depression associated with chemotherapy treatment in women with breast cance r. Additionally, this study test ed whether there is an association between levels of infl ammati on alterations in gray matter and lev e ls of depression and anxiety associated with chemotherapy This was accomplished using previously collected prospective data involving women with BrCA who have undergone chemotherapy treatment, wo men with cance r who did not receive chemotherapy and age matched healthy controls. Data were collected at three time points: prior to chemotherapy, 3 months after chemotherapy, and finally 9 months after treatment comple tion. Results of the analyses showed a significan t interaction between group and time on a priori gray matter regions of interest found to be altered in previous research within this population. However, the findings of the present study did not show consistent decreases in GM volumes associated with chemotherapy treatment as has been found previously Correlation analyses showed significant negative relationships between changes in IL6 and gray matter volumes in the middle frontal gyrus, and the

PAGE 5

v parahippocampal gyrus following chemotherapy treatment. Further, there was also a negative relationship bet ween changes in gray matter volume in the cingulate gyrus and levels of depression in the chemotherapy treated patients These results are discussed in light of previous research, and future directions for research are provided. The form and content of t his abstract are approved. I recommend its publication Approved: Jim Grigsby

PAGE 6

vi TABLE OF CONTENTS CHAPTER I. BACKGROUND Overview of Inflammation ................................ ................................ ............................. 2 Chemotherapy Associated Inflammation ................................ ................................ ...... 3 How Pro Inflammatory Cytokines may affect the Brain ................................ ............... 5 Brain Imaging Studies in Depression and Anxiety ................................ ........................ 6 Gray Matter Changes Associated with Chemotherapy ................................ .................. 7 Hypotheses and Specific Aims ................................ ................................ .................... 13 II. METHOD Participants ................................ ................................ ................................ ................... 16 Procedure ................................ ................................ ................................ ..................... 17 Measures ................................ ................................ ................................ ...................... 19 Data Analysis ................................ ................................ ................................ ............... 23 III. RESULTS ................................ ................................ ................................ .................... 25 Primary Analyses ................................ ................................ ................................ ......... 25 Exploratory Analysis ................................ ................................ ................................ ... 33 IV. DISCUSSION ................................ ................................ ................................ .............. 34 Strengths and Limitations ................................ ................................ ............................ 39 Future Directions ................................ ................................ ................................ ......... 42 Conclusions ................................ ................................ ................................ .................. 42 REFERENCES ................................ ................................ ................................ ........................ 45

PAGE 7

vii LIST OF TABLES TABLE 1. Inclusion and Exclusion Criteria ................................ ................................ .................. 17 2. Baseline Demographics and Clinical Characteristics ................................ .................. 25 3. Results of MANCOVA Analysis ................................ ................................ ................. 26 4. Results of Correlation Analysis Among Change Scores (3 months baseline) .......... 28 5. Results of Correlation Analysis Among Chang e Scores (9 months 3 months) ........ 30

PAGE 8

viii LIST OF FIGURES FIGURE 1. Correlation Between IL6 and Right Middle Frontal Gyrus GM Volume .................... 29 2. Correlation Between CES D and Left Cingulate Gyrus GM Volume ......................... 31 3. Correlation Between IL6 and Left Parahippocampal Gyrus GM Volume .................. 32

PAGE 9

CHAPTER I B ACKGROUND Breast cancer ( BrCA ) is associated with the second highest mortality rate of any type of cancer (American Canc er Society, 2013). BrCA occurs most commonly i n women over the age of 40, with an estimated global incidence of 1.6 million cases in 2010 An estimated 39,000 women in America died from BrCA alone in 2013 (American Cancer Society, 2013). BrCA is responsible for the largest portion of spending on cancer related care in the United States; an estimated $16.5 billion of cancer related spending went towards BrCA in 2010 ( Mariotto, Robin Yabroff, Shao, Feuer, & Brown, 2011 ) Although the prevalence, mortality, and costs associated with BrCA are quite high, recent advances in screen ing and treatment have increased long term survivorship of BrCA patients ( McDonald, Conroy, Ahles, West, & Saykin, 2010 ) This increase in long term survivorship has in turn led to increased attention to survivors' daily functioning and quality of life (QOL) Therefore, understanding the outcomes of treatment in patients with BrCA is of great importance to help better understand QOL and biological markers of treatment related outcomes in this population The most common form of treatment for breast canc er after surgery is adjuvant chemotherapy (American Cancer Society, 2013). The term "adjuvant" refer s to a treatment applied secondary to an initial treatment, typically to suppress further tumor formation (American Cancer Society, 2013). Although chemot herapy is a successful treatment in many cases, there are many side effects associated with the treatment. Common side effects of chemotherapy include: nausea and vomiting, hair loss, appetite loss, pain, fatigue, depression, anxiety, problems with concent ration and memory, and sleep disturbances, among others (American Cancer Society, 2013) These side effects take a toll on patients, and contribute

PAGE 10

2 significantly to decreased QOL even after finishing treatment ( McDonald et al., 2010 ) Systemic i nflammation resulting from bot h the disease process and treatment with chemotherapy, has been suggested as an underlying factor contributing to the observed side effects. Research into these inflammatory processes appears to support this notion ( Bower et al., 2011 ; Pomykala, Ganz, et al., 2013 ) Overview of Inflammation When tissue injury occurs, a network of chemical signals activate a host response which is designed to protect and heal the damaged tissues (i.e. inflammation) ( Coussens & Werb, 2002 ) This process involves the activation and migration of leukocytes (neutrophils, monocytes, and eosinophils ) to the site of damage as well as recruitment of tissue mast cells ( Coussens & Werb, 2002 ) Additionally, this inflammatory response involves migration of macrophage progenitors, which travel through the venous system to the site of injury. This migratory process is directed to the site of tissue damage by a host of chemot actic factors, a mong th e se are cytokines including interleukin 1 ( IL 1 ) and tumor necrosis factor alpha ( TNF ") ( Coussens & Werb, 2002 ) Once present at the site of injury, these macrophage progenitors will differentiate into either mature macrophages or immature dendritic cells. After activation, macrophages become the main source of cytokines ( TNF and IL 1 ), which help modulate tissue repair and can have drastic effects on the local microenvironment ( Coussens & Werb, 2002 ) The major pro inflamm atory cytokines (PICs), which include Interleukin 1 (IL1 and IL1 ), Interleukin 6 (IL 6), and Tumor Necrosis Factor alpha (TNF ) promote inflammation. They are regulated and modulated by other molecules, including anti inflammatory cytokines such as IL 4, IL10, and IL13. The coordinated dynamic activity of the various classes of cytokines and receptors allows the appropriate development of inflammation,

PAGE 11

3 and its down regulation when it is no longer necessary. Sabotage of cell death and repair programs occ urs in chronically inflamed tissues, thus resulting in DNA replication and proliferation of cells that h ave lost normal growth cont rol. Normal inflammation is self limiting, because the production of anti inflammatory cytokines follows the pro inflammatory cytokines closely. However, chronic inflammation seems to be due to persistence of the initiating factors or a failure of mechanisms required for resolving the inflammatory response. For example, the PIC TNF mediates many factors of the inflammatory pro cess. During early development of tumors in breast cancers, TNF re gulates a cascade of cytokines and growth factors which actually may be one of the ways inflammation acts as a tumor promoter when these cascades become unregulated ( Coussens & Werb, 2002 ) Another example of this unregulated inflammatory response involves the PIC IL 6 During acute inflammation, IL 6 stimulates and regulates protein production, while concurrently controlling th e level of the inflammatory response by regulating anti inflammatory cytokines. Therefore, during acute inflammation, IL 6 possesses both pro inflammatory and anti inflammatory properties. These properties change during chronic inflammation, a state in whi ch IL 6 seems to mediate only pro inflammatory responses and also enhances white blood cell infiltration at the sites of inflammation ( Dethlefsen, Hojfeldt, & Hojman, 2013 ) This becomes important in BrCA a disease tha t is multifactorial, and is known to involve this chronic inflammatory response. Chemotherapy A ssociated Inflammation Inflammation is associated with the side e ffects of chemotherapy mentioned above, and has been shown to affect cognition as well which has become associated with "sickness behavior and may be related to "chemobrain in this population ( Dantzer & Kelley, 2007 ; Wang et al., 2015 ) It has been proposed that inflamma tion, as measured by plasma levels of

PAGE 12

4 PICs is the underlying cause of many of the side effects observed with chemotherapy treatment, namely depression, fatigue, sleep disturbances and problems with memory and cognition ( Bower et al., 2011 ; Mohamed, Al Raawi, Sabet, & El Shinawi, 2014 ) For example, a recent study of 103 women with breast cancer who received primary treatment ( i.e. surgery, radiotherapy or chemotherapy) found that women treated with chemotherapy reported higher levels of depression, fatigue, and sleep disturbances than wome n treated with surgery or radiotherapy. The women treated with chemotherapy also had higher plasma levels of tumor necrosis factor receptor type II (sTNF RII), a marker of inflammation, when compared to the other treatment groups, and healthy controls. Fin ally, this group also showed a strong relationship between reported levels of fatigue, and sTNF RII levels ( Bower et al., 2011 ) This study indicates that measure s of cytokine plasma levels can be used as biomarker s to track inflammation in patients wi th breast cancer who have undergone chemotherapy, and these biomarkers seem to correlate with behavioral and psychological symptoms observed in this population. These biomarkers hold importance as they are related to quality of life and other health rel ate d outcomes Additional biomarkers associated with the cognitive declines found in this population have been thought to occur in the central nervous system (CNS). Therefore, brain imaging methods have been employed in order to discover potential biomarkers in the brain (e.g. changes in white m atter or gray matter integrity), which may have implications for disease progression and behavioral symptom development. Importantly, few studies to date have investigated the etiology of alterations in brain Gray matt er ( GM ) following chemotherapy. This study seeks to fill this gap in the literature by assessing burden of inflammation and changes in brain GM morphology as well as outcomes on measures of anxiety and depression

PAGE 13

5 How Pro Inflammatory Cytokines may e ffec t the Brain Chemotherapeutic agents are thought to have limited direct access to brain tissues due to the blood brain barrier (BBB). However, animal studies suggest that chemotherapeutic agents which are not known to readily cross the BBB (e.g. doxorubicin) may be associated with reduced neurogenesis ( Janelsins et al., 2010 ) Further, even small amounts of chemotherapy may cause long term damage in the brain as both dividing and non dividing glial cells have be en shown to be particularly vulnerable to chemotherapeutic agents ( Dietrich, 2010 ) The PICs (IL1, IL6, and TNF ) typically have limited access to the brain, though they may cross the BBB in smal l numbers via active transport ( Goehler et al., 2000 ) These PICs may also enter the cerebrospinal fluid through circumventricular organs (e.g. the choroid plexus) and the vagus nerve, while also be ing synthesized endoge nously in the brain ( Goehler et al., 2000 ) Receptors for these PICs have been found throughout the brain, which the P ICs must bind to in order to have an effect. Once these cytokines bind to receptors in the brain, they may exacerbate damage to neurons and glial cells and lead to neuronal death as well as to certain types of cognitive impairment ( Dietrich, 2010 ) IL6 increases BBB permeability, while IL1 and TNF induce free radical s that may then affect the BBB ( Merrill & Benveniste, 1996 ) Once in the brain, these cytokines lead to activation of microglia and astro cytes, and thus promote even greater inflamm ation ( Coussens & Werb, 2002 ) Herein, it is proposed that chemotherapeutic agents may, due to a breakdown of the BBB, induce acute effects as well as long term alterations in brain GM It is s uggest ed that this increased permeability of the capillary endothelial layer allows PICs access to the brain inducing a state of brain inflammation. Thus chemotherapy may be associated with a mild increase in

PAGE 14

6 permeability of the BBB, producing some degre e of brain inflammation, with associated changes in levels of PICs and observable changes in brain GM morphology Brain Imaging Studies in Depression and Anxiety Structural imaging studies looking at changes in GM associated with depression have shown mixed results ( Lai, 2013 ) A recent meta analysis, however, revealed that there does seem to be consistent GM decreases in the brains of depressed patients occurring in the an terior cingulate cortex or ACC ( Lai, 2013 ) This region is thought to be involved in attention, cognition, affective regulation, and motivation ( Lai, 2013 ) Decreased GM in the ACC is thought to disrupt the "visceromotor network" which in turn may change one's ability to cope with stress. The ACC is thus a potential target for GM changes in patien ts with BRCA as they experience significant stress, and show signs of comorbid depression following chemotherapy ( McDonald, Conroy, Smith, West, & Saykin, 2013 ) Neuroimaging studies looking at symptoms of anxiety are limited. Recently, Moon and colleagues (2014) conducted a cross sectional study using VBM to evaluate GM changes in patients with generalized anxiety disorder (GAD). 22 patients with GAD (13 men and 9 women) were compared with 22 age and sex matched healthy controls. When compared t o healthy controls, patients with GAD showed significantly decreased GM volume in the hippocampus, thalamus, insula, and superior temporal gyrus ( Moon, Kim, & Jeong, 2014 ) One interesting aspect is the decreased GM volume found in the hippocampus of GAD patients. The hippocampus has been found to play an important role in the hormone regulation involved in the hypothalamic pituitary adrenal axis ( Moon et al., 2014 ) The authors speculate that a decreased GM volume in the hippocampus may be related to dys regulation in neuroendocrine systems, thus

PAGE 15

7 leading to more stress for the individual. Again, this seems to fit well with how BRCA patients are conceptualized and with the prevalence of anxiety in these patients. Gray Matter Changes Associated with Chemothe rapy Previous research has focused on the central nervous system (CNS) in search of biomarkers associated with the many side effects observed following chemotherapy treatment ( Cimprich et al., 2010 ; Conroy et al., 2012 ; de Ruiter et al., 2012 ; Inagak i et al., 2007 ; McDonald et al., 2013 ; McDonald et al., 2010 ) One method that can be used to assess structural changes in the CNS at the level of the brain is Voxel Based Morphometry (VBM; ( Ashburner & Friston, 2000 ) VBM is a fully automated procedure for assessing gray matter (GM) and white matter (WM) integrity in the brain using struct ural magnetic resonance images (MRI). VBM is a method to quantitatively assess tissue changes in GM and WM on a voxel by voxel basis based on an a priori statistical threshold. Thus, VBM is an unbiased, highly accurate measure, sensitive to local changes i n the brain (Ashburner & Friston, 2000). Inagaki an d colleagues (2007) conducted one of the first cross sectional VBM analyse s on BrCA survivors treated with chemotherapy. Patients were split into two groups: those treated with chemotherapy less than one year after surgery (c+, n=55), and those treated more than 3 years after surgery (c+, n=73). These two groups were then compared to similar groups of patients who did not receive chemotherapy at 1 or 3 years (c n=55 and 59, respectively), and healthy con trols ( HC ; n= 55 and n=37, respectively) ( Inagaki et al., 2007 ) In the less than 1 year study, when compared to the C group, the C+ group showed decreased GM in the right middle and superior frontal gyrus, and the right parahippocampal gyrus. Additionally, they found evidence in the C+ group of less WM in the right cingulate gyrus, bilateral middle frontal gyrus left parahippocampal gyrus, and the left precuneus, when compared to the C group. No

PAGE 16

8 differences were observed in GM or WM between either C+ or C groups when compared to HC In the 3 year study, no differences between or within C+ or C were observed suggesting the effects of chemotherapy on GM and WM may recover over time ( Inagaki et al., 2007 ) McDonald an d c olleagues (2010) conducted the first longitudinal prospective study of structural brain changes in women with BrCA receiving chemotherapy, using VBM. They included three groups: BrCA patients treated with chemotherapy (C+, n=17), BrCA patients who had not received chemotherapy (C n= 12), and a match ed healthy control group ( HC n=18) ( McDonald et al., 2010 ) Each group was evaluated before chemotherapy, at one month, and finally one year after completion of treatment. They reported no between group diffe rences i n brain GM at baseline. Comparisons between groups revealed decreased GM density in the middle frontal g yrus (bilaterally) and cerebellum at 1 month in both cancer groups, when compared to HC ( McDonald et al., 2010 ) However, the declines were much more extensive and persistent for t he C+. Within the C+ group, over 1 month there were GM density decreases in bilateral frontal, temporal, and cerebellar regions, right thalamus and bilateral temporal lobes including medial temporal structures. Some frontal areas of grey matter reduction p ersisted to the 1 year time point, while some areas recovered ( McDonald et al., 2010 ) A n important thing to no te is that comparisons between C + and C groups indicated no differences in GM at 1 month or at 1 year. Although this study failed to detect differences between C+ and C patients, t hese findings suggest an acute effect of chemotherapy in GM density with subsequent partial recovery over time (at 1 year). The relatively small sample could also have contributed to the lack of significant findings. This study laid the groundwo rk for further work in this area. These same researchers sought to replicate and extend their findings of decreased GM density in this population ( McDonald et al., 2013 ) BrCA patients treated with and without

PAGE 17

9 chemotherapy ( n=27 and n = 28 respectively ) and matched HC (n = 24) were scanned at baseline (prior to treatment) and 1 month following che motherapy (or matched intervals). GM reductions after 1 month were found in left middle and superior frontal gyrus when compared to baseline Additionally, these researchers foun d that chemotherapy treated BrCA patients reported significantly more self perceived symptoms in their ability to initiate a task and generate problem solving strategies, as measured by the BRIEF A. Finally, they found a negative correlation between differ ence scores on self reported executive functioning and GM density in the left middle frontal gyrus (MFG) The authors infer from this result that a reduction in GM in the MFG may be associated with increased cognitive complaints as evide nced by the BRIEF A. The result also suggest s a relationship between reductions in frontal GM volume and executive functioning in BrCA patients who have undergone chemotherapy. These two studies ( McDonald et al., 2013 ; McDonald et al., 2010 ) suggest that GM reductions are a result of chemotherapy treatment in BrCA patients, and these reductions seem to be localized to pref rontal regions. One may then speculate about the relationship between reductions in prefrontal cortex GM and the observed cognitive complaints in this population, another notion supported by the self reports of current executive functioning in the previous study. de Ruiter and colleagues (2012 ) investigated the long term effects of chemotherapy treatment in BrCA patients more than nine years after receiving treatment This study included 17 BrCA patients receiving high dose chemotherapy (C+) with 15 patients who were untreated (C ) This study utilized a multimodal imaging approach to test their hypotheses. Their VBM analysis showed significant differences in GM volume between the two groups. Namely, they found reductions in posterior regions of t he brain such as the posterior parietal cortex the

PAGE 18

10 precuneus (bilaterally), the left occipital cortex, and the cerebellum, bilaterally in the C+ group but not the C group. These GM volume reductions were found to overlap with fMRI task related hypoactiva tions ( de Ruiter et al., 2012 ) Further, the se same regions showed evidence of changes in WM microstructure, thought to represent WM integrity. These results are suggestive of long term neurotoxic side effects of high dose adjuvan t chemotherapy for BrCA However, these results also contradict some of the p revious work mentioned above. This study would suggest that there is a longitudinal effect of chemotherapy treatment, while the study by In a gaki and colleagues (2007) suggests acute effects of chemotherapy on brain GM, with a restoration of these effects 3 years later. There are several limitations to the study by Inagaki et al. (2007) that should be considered when interpreting their results. One such limitation is the wide variety of chemotherapeutic agents used within the sample, thereby making it imposs ible to determine the effects of a specific regimen on GM values. A second significant limitation is that two, only partially overlapping, samples were used when evaluating the effects of chemotherapy at the 3 years time point. This poses difficulties in i nterpreting their results, but also led researchers in the right direction towards ways in which experimental control can be best established to answer these questions. Koppelman and colleagues (2013) conducted the first study to determine the long term e ffects of chemotherapy treatment on brain volume in a relatively homogenous sample ( Koppelmans, Breteler, Boogerd, Seynaeve, & Schagen, 2013 ) In this study, MRI scans from 1 84 BrCA patients who had undergone chemotherapy (on average 21 years prior) were compared to those of 368 age matched healthy controls ( Koppelmans et al., 2013 ) These researchers reported that BrCA patients who underwent chemotherapy had significantly less total brain volume (sum o f GM, WM, and Cerebrospinal fluid or CSF ), and total GM volume than the

PAGE 19

11 healthy controls. VBM analysis showed no local GM differ ences between the two groups. With regard to the differences in total brain volume and total GM, they found these reductions were long term. In this sample, these researchers showed that every year of aging was associated with a 0.75 ml reduction in total GM volume in the C+ group ( Koppelmans et al., 2013 ) Based on their observed differences in total GM volume between groups, these authors suggested that exposure to chemotherapy is equivalent to 3.9 years of additional aging Another possible in terpretation is that the reductions in total GM volume is related to the stress associated with the treatment or dealing with the cancer itself In addition to these studies on GM changes associated with c hemotherapy treatment, there have been only two st udies that have investigated the relationship between neuroimaging outcomes and markers of inflammation in BrCA patients who have undergone chemotherapy treatment ( Kesler et al., 2013 ; Pomykala, Ganz, et al., 2013 ) Pomykala and colleagues (2013 ) conducted a positron emission topography (PET) study to examine the relationships between PICs, cerebral metabolism, and cognitive complaints following chemotherapy treatment in BrCA patients. In this study, 23 BrCA patients who underwent chemotherapy an d 10 BrCA patients who did not were analyzed at baseline (following treatment) and then again one year later. Data collection involved brain imaging data (cerebral metabolism via PET) and cytokine markers, including IL 1, sTNF RII, CRP, and IL 6. These researchers found that at baseline, there was a positive correlation between the levels of cytokines and cerebral metabolism in the left medial prefrontal cortex and right inferior lateral anterior temporal cortex in the chemotherapy group. This corr elation was not found in the non chemotherapy group. At the 1 year time point, there was a positive correlation between cytokine levels and metabolism in the medial prefrontal cortex, and the anterior temporal cortex. Again,

PAGE 20

12 this correlation was not presen t in the non chemotherapy group ( Pomykala, Ganz, et al., 2013 ) These results suggest increased cerebral metabol ism may be associated with increased burden of inflammation. The authors speculate that an initial inflammatory response (perhaps due to exposure to chemotherapeutic agents) begins a cascade that ends up affecting brain metabolism. It is possible that the observed alterations in metabolism could lead to the changes observed in other studies looking at GM chang es associated with chemotherapy as i ncreased metabolism in some areas could serve as a compensatory mechanism for loss of neuronal functioning or inte grity in other regions Kesler and colleagues (2013) conducted the only study to date directly testing the relationship between cytokine markers of inflammation and brain GM volume. In this cross sectional study, MRI, cytokine data, and measures of memory performance were collected from 20 BrCA patients who had undergone chemotherapy (average time since chemotherapy ~ 4.8years) and 23 healthy controls. These researchers focused specifically on the hippocampus in search of GM alterations associated with mar kers of inflammation. This was accomplished by manually tracing the hippocampus, bilaterally, in order to obtain hippocampal GM volumes. Their results showed that left hippocampal volume and memory performance were reduced in the BrCA group compared to the healthy controls. They also found increased levels of IL 6 and TNF in the BrCA group. Correlation analyses showed that lower hippocampal volume was associated with higher levels of TNF (negative correlation), but lower levels of IL 6 (positive correla tion). The authors suggest that these contradictory findings are due to both the anti and pro inflammatory roles that IL 6 plays in the immune system. Finally, they found a significant interaction between these cytokines, such that the authors propose tha t IL 6 may have a m oderating effect on TNF levels. This study provides insight into the possible effects

PAGE 21

13 chemotherapy may have on burden of inflammation, and the association between brain GM changes and cytokine markers of inflammation ( Kesler et al., 2013 ) However, this study was li mited by the fact that they were not able to determine the acute effects of chemotherapy treatment on these measures due to the long duration post treatment. Additionally, this study did not includ e BrCA patients who did not receive chemotherapy as a necessary control group. As can be seen, research evaluating the association between PICs and GM volume in BrCA patients undergoing chemotherapy treatment is limited. Further, despite the prevalence of anxiety and depressive symptoms in this population, little is known about the relationship between changes in GM volume and symptoms of depression and anxiety. Thus, this necessary step can bridge the gap between observed differences in GM volumes in this condition, and serve to tie the associated symptoms of chemotherapy treatment and markers of inflammation into a cohesive whole so that underlying mechanisms can be elucidated and models of symptom development can be esta blished. Hypotheses and Specific Aims There have been many imaging studi es conducted in this population; however, there has been little consistency in the findings. There fore, this study intended to replicate and extend on previous work done with this popu lation. Specifically, based on previous research, we expect ed to see significant changes in the following brain regions follow ing 3 months of chemotherapy treatment: the parahippocampal gyrus (PHG) the middle frontal gyrus (MFG), the superior frontal gyru s (SFG), the precentral gyrus (PCG), and the cingulate gyrus (CG) Further, we expected these changes to persist at 9 months. These regions have been previously identified, and highlighted in a recent review of imaging studies in this population ( Pomykala, de Ruiter, Deprez, McDonald, & Silverman, 2013 ) Based on the coordinates provided in these recent

PAGE 22

14 imaging studies, we built regions of interest ( ROIs ) and test ed these a priori regions specifically for all hypotheses Aim 1 : To test whether chemotherapy treatment is associated with alterations in brain GM volume in regions previously found to change with chemotherapy treatment (replication and extension) Hypothesis 1 .1 : A t baseline (prior to chemotherapy treatment), there will be no significant differences in regional brain GM volume Hypothesis 1 .2 : A t baseline (prior to chemotherapy treatment), there will be no significant differences in regional brain GM volume in a priori ROIs or PIC levels between BrCA patients and HC Hypothesis 1. 3 : C ompared with BrCA patients who have not received chemotherapy and HC patients who have received chemotherapy will show decreased brain GM volume fo llowing chemotherapy treatment in the a priori ROIs mentioned above Aim 2 : To explore the relationship between brain GM volume changes burden of inflammation and level s of depression and anxiety in chemotherapy treated breast cancer patients. Hypothesis 2.1 : B etween baseline and 3 months f ollowing completion of chemotherapy treatment, there will be a negative relationship between changes in regional brain GM volume in a priori ROIs and PIC levels in BrCA patients who have undergone chemotherapy (BrCA+) Hypothesis 2.2 : B etween 3 and 9 months following completion of chemotherapy, there will be a negative relationship between changes in brain GM volume in pre defined ROIs and markers of inflammation in the BrCA+ group

PAGE 23

15 Specifically, it has been shown previously that receptors for the PICs collected in the present study have been found throughout the brain, to which the PICs must bind in order to have an effect. Once these cytokines bind to receptors in the brain, they may exacerbate damage to neurons and glial cells and lead to neuronal death as well as to certain types of cognitive impairment ( Dietrich, 2010 ) This suggests that inflammation may be associated with decreases in brain GM, as PICs can both enter the brain, and may play a role in neuronal cell death, as evidenced by decreased GM. Hypothesis 2.3 : B etween baseline and 3 months following completion o f chemotherapy treatment, there will be a negative relationship between changes in brain GM volume in a priori ROIs and changes on two self report measures of depression and anxiety in the BrCA + group Hypothesis 2.4 : B etween 3 and 9 months following completion of chemotherapy treatment, there will b e a negative relationship between changes in brain GM volume in a priori ROIs, and changes on two self report measures of depression and anxiety in the BrCA+ group Hypothesis 2.5 : B etween baseline and 3 months and between 3 months and 9 months following chemotherapy treatment, there will be positive relationship s between burden of inflammation ( change in PIC levels) and changes on self report measures of depression and anxiety in Br CA patients who have undergone chemotherapy Hypothesis 2.6 : In this exploratory analysis, we hypothesize that brain GM volumes in a priori ROIs will be negatively correlated with two self report items related to patient's memory and stress associated with chemotherapy treatment in BrCA patients who have undergone chemotherapy.

PAGE 24

16 CHAPTER II M ETHOD Participants An existing dataset was used for the present study. Following a protocol approved by the Colorado Multiple Institutional Review Board (COMIRB) and consistent with HIPAA regulations, written informed consent was obtained from all participants. Two hundred and nine females with BrCA (stage 0 3) were recruited to take part in a larger study of the cognitive trajectory and quality of life of women with B rCA undergoing chemotherapy. A subset of eligible participants was randomly selected to undergo additional MRI testing. Thirty one females with BrCA who underwent chemotherapy treatment ( BrCA+; mean age in years SD; 57.96 6.64), 18 BrCA patients who did not undergo chemotherapy treatment ( BrCA ; mean age in years SD; 67.6 5.5), and 20 healthy controls (mean age in years SD; 63.2 7.84) were randomly selected f or MRI testing In this total subsample, approximately 95 % (55/5 9 patients) identifie d th emselves as W hite/Caucasian. Three patient s identified themselves as being black, and another identified herself as being Asian. Although we tried to maximize recruitment of racial and ethnic minorities, the demographics of the sampling frame made this difficult. The data in the present analysis is based on this sub set of BrCA patients and HC that received MRI testing (n=5 9 ) taken from the larger sample of 209 BrCA patients A trained MRI technician collected all MRI data. See Table 1 for specific in clusion and exclusion criteria.

PAGE 25

17 Table 1 Inclusion and Exclusion Criteria BrCA=Breast Cancer; CNS=Central Nervous System; COPD=Chronic Obstructive Pulmonary Disease; CVA=Cerebrovascular Accident (stroke); MS=Multiple Sclerosis; Hx=History. Procedure Data were collected at three time points For chemotherapy patients, these were prior to beginning the first chemotherapy cycle (baseline), three months after completion of chemotherapy (first follow up), and nine months after completion of chemotherapy (secon d follow up). For the non chemotherapy participants, the post operative baseline evaluation occurred at approximately the same time following surgery as for chemotherapy patients, while the two follow up examinations were scheduled six months and fifteen m onths later. Healthy controls were examined at the time of enrollment, and aga in after intervals of six and fifteen months. BrCA + patients received a combination of anthracycline based chemotherapies, taxanes Inclusion Criteria Exclusion Criteria English speaking females Stage 4 disease Confirmed diagnosis of Stage 0 3 BrCA Planned high dose chemotherapy regimen Post operative, lumpectomy or mastectomy Sensory deficits that would limit interview data Scheduled for adjuvant or neoadjuvant standard dose chemotherapy (chemotherapy group) Hx of delirium, dementia, mental retardation, or psychosis Not scheduled for chemotherapy (control group) Hx of significant alcohol/drug use past 18 mos. Age # 45 Significant CNS comorbidity (e.g. CVA, MS) Acute medical condition, exacerbation of a chronic condition, or medical surgical treatment affecting cognitive emotional functioning in the prior two weeks Moderate severe congestive heart failure, significant hepatic or renal disease, COPD requiring oxygen Acute medical condition other than cancer, acute exacerbation of a chronic condition, or medical surgical treatment affecting ability to tolerate assessment or that com promised ability to participate

PAGE 26

18 or some combination of the two which are consi dered "standard" chemotherapy treatments. The protocol involved collecting medical history as well as demographic, QOL neuropsychological, imaging, and molecular (cytokine) data Here we report on the demographic, imaging (MRI), self report, and pro inflammatory cytokine data Subjects with BrCA were recruited from one of three organizations for this study. These included: 1) Rocky Mountain Cancer Centers (RMCC), which has 13 oncology clinics within 1.5 hours of the Denver metro area along the F ront Range of Colorado; 2) the University of Colorado Hospital Breast Center (UCHBC); and 3) Kaiser Permanente of Denver. Some patients volunteered in response to brochures placed in clinic office and exam rooms, but initially most were approached about pa rticipation in the study by physicians and nursing staff. Patients who expressed interest in participation were contacted by phone by study nurses and screened for eligibility. If a patient met eligibility criteria, the study was explained to her, she was told she would be paid for her participation, and verbal consent was obtained. It was then explained that a random subset of eligible participants would be chosen to undergo a MRI scan at each of the time points of assessment. MRIs were conducted at the Un iversity of Colorado Denver Brian Imaging Center. Participants were paid $50 for each testing session (for a total of $150 if they completed the study). Participants randomly chosen to receive MRIs were given an addi tional $50 for each MRI visit. Patients were then scheduled for a baseline appointment, during which we obtained written consent and conducted study interviews. The interviewers were registered nurses with experience in health services research and home health care. They administered all neurops ychological, f unctional, QOL and mood measures, and drew blood at the beginning of each visit. Interviews took place in participants' homes unless they preferred to come to the oncology clinic. Over 95% of all clinical data

PAGE 27

19 collection was done in the home Following clinical interview, the MRI participants were scheduled to receive an MRI scan as close to the interview as possible. MRI scans were conducted at the University of Colorado Health Sciences Brain Imaging Center. Magnetic resonance imaging (MRI) was pe rformed on a 3.0 Tesla GE Signa scanner (HDx re lease ). For each subject, a T 1 weighted spoiled gradient recalled (SPGR) data set (TR 10ms, TE 2.1ms, flip angle 10 ¡, FOV 256$ 256, yielding 124 sagittal slice s with a defined voxel size of .86$ .86 $ .86 mm resampled to 1$ 1 $ 1mm ) was acquired. An Eclipse 3.0 T 94 quadrature head coil was used. Inspection of individual T1 MR images revealed no gross morphological abnormality for any participant. Measures Gray Matter Volume : GM volume was obtained using a standardized procedure described previously ( Ashburner, 2007 ) Baseline T1 weighted structural images were first co registered with the follow up scans. Co registration involves aligning images, based on their edges, to allow for more accurate segmentation. Next, the images were segmented into white matter (WM), gray matter (GM) and cerebrospinal fluid (CSF) using the "new seg ment" function in Statistical Parametric Mapping, version 8 (SPM8 ; Wellcome Department of Imaging Neuroscience, London, United Kingdom ). SPM8 is free, open source software that is available for download at: http://www.fil.ion.ucl.ac.uk/spm/software/spm8/ The resulting GM segments were th en processed using the VBM8 toolbox within SPM8 ( http://dbm.neuro.uni jena.de/vbm ) Using the diffeomorphic registration algorithm (DARTEL) described by Ashburner ( Ashburner, 2007 ) t he individual GM and WM segments in native space were then nonlinearly normalized to the DARTEL Template supplied with the VBM8 toolbox (see http://dbm.neuro.uni jena.de). The voxel values of the normalized tissue segments were then

PAGE 28

20 multiplied (modulated '') with the nonlinear component of the Jacobian determinant, which was derived from the aforementioned normalization step. The resulting GM segments thus preserve the local GM volume as identified in native space corrected for total brain size ( Buckner et al., 2004 ) A quality check was performed using tools from the SP M Toolbox and individual visual assessment, which yielded no artifacts or failed segmentation and normalization of the data. Prior to statistical analysis the segments were smoothed with an 8 X 8 X 8mm FWHM kernel. Normalized, smoothed data were then enter ed into statistical analyses within SPM8. Significan t regions of interest (ROI s ) were extracted using the open source software platform MARSBAR v0.44 ( http://marsbar.sourceforge.net/ ). Extracted GM volume val ues were then analyzed further in SPSS version 22. A priori ROIs were also created using MARSBAR v0.44. These ROIs were created as 10mm spheres, and were centered based on Montreal Neurological Institute (MNI) coordinates found previously in the literature. Specifically, we looked at GM volumes found to be altered over the course of chemotherapy treatment by in two separate studies ( McDonald et al., 2013 ; McDonald et al., 2010 ) Cytokine Assays : B lood was drawn at each timepoint, prior to the examination ; ethylenediamine tetraacetic acid (EDTA) was used to prevent clotting and t he blood samples were then spun down, and plasma was collected and frozen at 20 ¡ C Quantita tive enzyme linked im munosorbent assays (ELISA) were done to calculate peripheral IL 1, IL 6, TNF and C reactive protein levels. The first 3 are PICs, while C reactive protein is a general indicator of inflammation. Because of diurnal variability in cyt okine levels, blood was drawn for these assays at about 9 AM a nd about 2 PM on the day of a cognitive exam. For older persons, research suggests that the circadian pattern is relatively flat for both IL6 and TNF ( Vgontzas et al., 2003 ) so there is not much variation between the 2 levels.

PAGE 29

21 Center for Epidemiologic Studies Depression Scale (CES D): Depre ssion was evaluated using the Center for Epidemiologic Studies Depress ion Scale (CES D ) ; ( Radloff, 1977 ) Because the CES D reflects cognitive and affective symptoms rather than somatic symptoms of depression, it is highly recomm ended for use with patients experiencing medical problems ( Liu et al., 2012 ) The CES D is a 20 item self report measure. Self repor ted responses were scored on a 4 point Likert scale for how often participants endorsed the items over the past week rated by participants from Rarely or None of the Time (<1 day) to Most or All of the Time (5 7 days). Items in the measures include 16 questions that are scored summed such as (e.g. "I was bothered by things that don't usually bother me"), and 4 items that are reverse scored (e.g. "I felt that I was just as good as other peo ple"). Research demonstrates that the CES D is both a valid and reliable instrument that can be used in research to screen for common symptoms of major depression. Internal consistency using coefficient alpha is estimated to be .85 for t he general populati on and .90 in patient samples ( Radloff, 1977 ) Estimates of test retest reliability ranging from two weeks to twelve months fall in the moderate range of .45 to .70 ( Radloff, 1977 ) Concurrent validity of the CES D was evaluated by determining the degree to which CES D scores were in agreement with other measures of depression. The CES D was found to have correlations ranging fro m .50 to .80 with the Hamilton Rating S cale, .30 to .80 with the Raskin rating scale, .40 to .50 with the Lubin Depression Adjective Checklist, .60 with the Bradburn Affect Balance Scale's Negative Affect and Positive Affect Scales, and .43 with the Cantril life sat isfaction ladder ( Radloff, 1977 ) Thus, this measure shows good convergent validity with other measures of depression and negative affect. The discriminant validity of the

PAGE 30

22 CES D, however, was found to be less successful in distinguishing between depression and other types of emotional re sponses, such as anger, f ear, and boredom ( Radloff, 1977 ) State Trait Anxiety Inventory (STAI): C urrent anxiety (state anxiety) and trait anxiety was assessed using the ST AI ( Barnes, Harp, & Jung, 2002 ) This scale consists of 20 statements about how the person fee ls "at the moment" (state), and 20 items for assessing trait anxiety State anxiety items include "I am tense," and "I feel nervous." Trait anxiety items i nclude, "I worry too much over something that doesn't matter." Test retest reliability coefficients on initial development ( Barnes et al., 2002 ) ranged from 0.31 to 0.86, with intervals ranging from 1 hour to 104 days. Because the state anxiety scale tends to d etect brief states, test retest coefficients were lower for the state a nxiety as compared to the trait a nxiety ( Barnes et al., 2002 ) Throughout the development of the overall STAI (state and trait subscales), more than 10,000 adults and adolescents were tested To assess content validity, items were selected from other anxiety measures on the basis of strong associations with the Taylor Manifest Anxiety Scale ( Taylor, 1953 ) The overall correlation between the STAI and this measure was 0.73 demonstrating co ncurrent validity. Validity of the State subscale alone was originally established from testing in situations characterized by state high stress, such as classroom examinations and military training programs. This subscale was shown to be a valid measure o f state anxiety, and distinguished from trait anxiety ( Julian, 2011 ) Self report Items Related to Stress and Perceived Memory Problems : Participants w ere given t wo self report items related to patient's perceived effects of stress on their health associated with their illness and their perception of cognitive problems which are most important for the present study and in line with our research hypothes es One of the self report items comes from the Functional Assessment of Cancer Therapy Breast (FACT B). This self

PAGE 31

23 report question states: "During the past 7 days, I worry about the effects of stress on my illness," and responses wer e scored on a 5 point Likert scale ranging from 0 "not at all," to 4 "very much." A second item was administered as part of a self report questionnaire used to assess aspects of memory and concentration. This question states: "During the past two weeks I have noticed that m y concentration and memory are not what they used to be." Responses were also scored on 5 point Likert scale ranging from 0 "not at all," to 4 "very much." This question is not part of a validated measure; therefore interpretations of this self report item should be interpreted with caution. A nalyses including these two items were intended to be exploratory in order to provide potential implications for future research. Data Analysi s Data were analyzed using IBM SPSS Statistics version 22 (SPSS Inc., 2013) and Statistical Parametric Mapping version 8 (SPM8 ; Wellcome Department of Imaging Neuroscience, London, United Kingdom ) Preliminary Analyses Prior to our main analyses, d escriptive statistics (i.e mean s standard deviations, frequency distributi ons) were calculated to describe the sample and an analysis of variance (ANOVA) was conducted to compare the three groups on baseline demographics Continuous variables (e.g. GM volumes and cytokine values) were plotted as histograms in order to check for normal distributions; appr opriate transformations were applied as necessary to correct for skewed distributions. Hypothesis Testing AIM 1 : Hypothesis 1.1 was tested u sing an independ ent samples t test within SPM8 to compare whole brain, smoothed GM images between groups ( HC vs. BrCA+ and BrCA ) at baseline To

PAGE 32

24 test h ypothesis 1.2 b aseline GM volumes in a priori ROIs were extracted from SPM8 and entered into SPSS. GM volumes in a prior i ROIs and PIC levels were then compared across the three groups u sing a one way analysis of variance (ANOVA) Hypothesis 1.3 was tested using a repeated measures multivariate analysis of covariance (MANCOVA) within SPSS. GM volumes in a priori ROIs were used to determine changes in regional brain GM b etween the BrCA+, the BrCA and the HC groups over the course of treatment. This analysis included one within subjects factor (time) and one between subjects factor (group) Age was entered as a covariate in order to control for GM decreases associated wi th the aging process. A priori ROIs were applied as dependent variables and compared between the three groups, across time. AIM 2: To test hypotheses 2.1 through 2.5, change scores were calculated for all variables of interest (GM volume in a prior i regio ns, PICs, CES D scores, and STAI scores) between the firs t two time points (baseline and after 3 months of chemotherapy) and between the second and third time points (3 months and 9 months after chemotherapy). To calculate the change scores, the earlier ti me points were subtracted from the latter time points (e.g. 3 month value subtracted from baseline value, or "post" minus "pre"). These change scores were then entered into bivariate correlation analyse s within SPSS and included only the BrCA+ group. Finally, to test the exploratory analysis (hypothesis 2.6) change scores on the two self report items were entered into bivariate correlations with the change scores of GM volume for a priori ROIs.

PAGE 33

25 CHAPT ER III RESULTS Primary Analyses A nalyses were conducted to addre s s the two Specific Aims and test the above hypotheses. Table 2 presents demographics, baseline clinical characteristics, and results of the preliminary ANOVA, including baseline depression sc ores on the CES D and baseline state and trait anxiety scores on the STAI Table 2. Baseline Demographics and Clinical Characteristics Variable HC BrCA+ BrCA p value N=59 18 24 17 Age (years) 63.33 (8.28) 58.1 (5.6) 68 (5.4) p< 0.001* Education (years) 16.9 (2.6) 15.8 (2.3) 15.3 (2.3) p =0.149 Ethnicity White/Caucasian 18 (100%) 23 (95%) 14 (82%) Black 0 1 (5%) 2 (13%) Asian 0 0 1 (5%) CES D 27.6 (6.6) 31.3 (7.9) 29.3 (9) p =0.291 STAI S 36.7 (6.9) 29.6 (8.7) 28.5 (9.5) p =0.557 STAI T 29.1 (7) 31.4 (7.7) 30.6 (10) p =0.674 Values presented as: mean (standard deviation) or frequencies (percentage); HC = Healthy Controls, BrCA+ = Breast cancer patients in the chemotherapy group, BrCA = Breast cancer patients in the non chemotherapy group, CES D = Center for the Epidemiological Study of Depression scale, STAI S = State Trait Anxiety Inventory (State scale), STAI T = State Trait Anxiety Inventory (Trait scale). = significantly different based on ANOVA of baseline values.

PAGE 34

26 Evaluation of h istograms showed a reasonably normal distribution for all continuous variables included in the study, and thus no transformations were necessary prior to further analysis. Results of the preliminary ANOVA showed that the ages among the three groups were significantly different, F ( 2,56) = 11.95, p <0.001. Tukey's post hoc tes t showed that the HC and BrCA groups were significantly older than the BrCA+ group (mean difference = 5.26, p = 0.031 and mean difference = 9.95, p < 0.001, respectively), but were not significantly different from each other in age ( p = 0.091 ). No other demographic variables were significantly different among the groups (all p > 0.05). As hypothesized, there were no significant regional GM volume differences at baseline between groups, when comparing whole brain GM volumes in SPM8. Results of the ANOVA also revealed no significant group differences at baseline for GM volumes in a priori ROIs or PICs (all p > 0.05). Results of the MANCOVA are presented in Table 3 Table 3. Results of MANCOVA Analysis Dependent Variables = a priori Gray Mater Regions of Interest There was a significant overall between subjects effect of Age on GM volumes in the a priori ROIs, F (9,46) = 2.94 p = 0.008 ; Wilk s % = 0.635 partial &2 = 0 .365, observed power = 0.9 3. There was also a significant overall between subjects effect of Group on GM volumes in the a priori ROIs, F (18,92) = 2.62, p = 0.001; Wilks' % = 0.437 partial &2 = 0 .33 9, observed Effect Wilks' Lambda F Error df Sig. Partial Eta Squared Observed Power Between Subjects Age 0.635 2.938 46 0.008 0.365 0.934 Group 0.437 2.619 92 0.001 0.339 0.993 Within Subjects Time 0.564 1.590 37 0.114 0.436 0.807 Time Age 0.533 1.802 37 0.064 0.467 0.865 Time Group 0.269 1.909 74 0.01 0.481 0.993

PAGE 35

27 power = 0.993. The within subjects interaction between Age an d Time was insignificant ( p = 0.064). There was a significant within subjects interaction between Group and Time on GM volumes in the a priori ROIs when controlling for age F (36,74) = 1.91 p = 0.01; Wilk s % = 0.269 partial &2 = 0.481 observed power = 0.993 Post hoc pairwise comparisons were used to further evaluate the observed differences between groups over time and specify the specific regions adjusted by these factors. In order to correct for multiple comparisons, a Bonferroni correction was applied. Significant pairwise mean differences were found in the right precentral gyrus (PCG), and in the left superior frontal gyrus (SFG). Specifically, the BrCA+ group showed significantly less mean GM volume in the right PCG compared to the BrCA group ( mean difference = 0.04, p = 0.031). Additionally, patients in the BrCA group showed significantly more mean GM volume in the left SFG compared to the HC group (mean difference = 0.02, p = 0.047). Results of the bivariate correlations among change scores bet ween baseline and 3 months post chemotherapy treatment are presented in Table 4 The bivariate correlations of these change scores showed that IL6 levels were negatively correlated with GM volume in the right middle frontal gyrus (MF G; r = 0.43, p = 0.045). A scatter plot graph of this result is presented in Figure 1.

PAGE 36

28

PAGE 37

29 Figure 1. Correlation Between IL6 and Right Middle Frontal Gyrus GM Volume Blue dot represents region of interest placed in the middle frontal gyrus (image is in radiological format). Post pre = change scores computed between baseline (pre) and 3 months (post); R MFG = Right Middle Frontal Gyrus Gray Matter volume, IL6_A = Interleukin 6 value. Results of the bivaria te correlations among change scores bet ween 3 months and 9 months post chemotherapy tr eatment are presented in Table 5 Results of this analysis showed that CES D scores were negatively associated with GM volume in the Left Cingulate cortex (r = .408, p = .048 ; Figure 2 ), and positively associated with changes in STAI scores on both subscales (state: r = 0.636, p = .001; trait: r = 0.678, p <0.001). Changes on the State anxiety portion of the STAI measure showed a positive correlation with changes in IL6 ( r = 0.528, p = 0.02). Finally, changes in IL6 were also negatively correlated with GM volume in the left parahippocampal gyrus (r = 0.632, p = 0.004; Figure 3).

PAGE 38

30

PAGE 39

31 Figure 2. Correlation Between CES D and Left Cingulate GM Volume Blue dot represents region of interest placed in the Left Cingulate Gyrus (image is in radiological format). Post pre = change scores computed between 3 months (pre) and 9 months (post); L Cingulate= gray matter volume in this regi on; CES D = Center for the Epidemiologic Studies Depression Scale.

PAGE 40

32 Figure 3. Correlation Between IL6 and Left Parahippocampal Gyrus GM Volume Blue dot represents region of interest placed in the Left Parahippocampal Gyrus (image is in radiological format). Post pre = change scores computed between 3 months (pre) and 9 months (post); L parahippocampal gyrus = gray matter volume in this region; IL6 = interleukin 6.

PAGE 41

33 Exploratory Analysi s Results of the bivariate correlations among change scores between baseline and 3 months showed that t here was a significant negative correlation be tween CRP and perceived effects of stress on health associated with treatment as measured by the single item from the FACT B (r = .475, p = 0.026) as well as CRP and self reported problems with memory (r = .659, p = .001). It was also found that GM volume in the right middle frontal gyrus (MFG) was positively correlated with the measure of stress associated with treatment (r = 0.434, p = .034). Results of the bivariate correlations among change sco res between 3 and 9 months post chemotherapy showed that changes in CES D scores were positively associated with changes in perceived effects of stress associated with treatment (r = 0.518, p = 0.01). Further, changes in both state and trait anxiety as measured by the STAI were positively correlated with memory complaints in these patients ( r = 0.419, p = 0.041 and r = 0.429, p = 0.036, respectively).

PAGE 42

34 CHAPTER IV DISCUSSION The current study tested the notion that BrCA patients may experience reductions in brain GM volume following chemotherapy treatment. To this end, relationships bet ween chemotherapy treatment, brain GM volume markers of inflammation, and measu res of depression and anxiety were evaluated in a sample of BrCA patients undergoing chemotherapy and compared with BrCA patients who did not undergo chemotherapy, and HC The preliminary analysis showed that the groups differed significantly by age (Tab le 2). Although the groups were significantly different from each other on age, the BrCA+ patients had a mean age of 58 years, which is comparable to the ages of participants of previous studies ( Pomykala, de Ruiter, et al., 2013 ) Due to the significant difference in ages, age was entered as a covariate in the MANCOVA analysis to help control for the effects of age on GM volume The results support the first hypothesis, that there would be no significant differences between the groups at baseline (prior to treatment) on measures of regional GM volume and i n a priori GM ROIs. Further, there were no significant differences on measures of PICs at baseline. This has been observed in previous studies ( McDonald et al., 2013 ; Pomykala, de Ruiter, et al., 2013 ) and thus serves to replicate what has been found previous ly. Results of the MANCOVA showed that there were overall between subjects effect s of both Group and Age on the GM volume of ROIs included in the analysis. Further, there was also a significant within subjects interaction effect between Group and Time. Th ese results are not whol ly surprising; we expected to find differences between groups over time in the a priori ROIs. Also, following the preliminary ANOVA, and based on previous work ( M cDonald et al., 2013 ) age was expected to have an ef fect on GM volume over time and thus was controlled for

PAGE 43

35 in subsequent analyses comparing groups. What was surprising about these results, and contradictory to findings from other studies, is that GM volumes in the a priori ROIs showed that there were no si gnificant differences between the BrCA+ and HC The only ROI wherein BrCA+ patients showed significantly less GM volume than the BrCA group was in the precentral gyrus In this sample, on average, HC and BrCA patients had less GM volume than the BrCA+ gr oup in the a priori ROIs included in the analyses These results are inconsistent with previous reports of decreased GM volumes in pat i ents with BrCA who have undergone chemotherapy in these specific regions that were tested in the present study ( McDonald et al., 2013 ; McDonald et al., 2010 ) It is possible that differences in sample size, VB M processing techniques, and measurement intervals could account for these differences. However, these findings challenge the notion that chemotherapy causes significant differences in brain GM by testing the specific regions where such differences were fo und previously Further analyses of these speci fic regions could help clarify this discrepancy. The precentral gyrus did show significant GM reductions in the BrCA+ group over the course of treatment when compared with the BrCA group This finding is part ially in line with our hypothesis, but still failed to show significant differences between the HC group and the BrCA+ patients. Moreover, t his result in the precent r al gyrus is difficult to interpret since basic moto r functioning as mediated by the prece ntral gyrus, does not have a logical relationship to chemotherapy treatment other than the fact that this region was found to be decreased following chemotherapy treatment in previous research ( McDonald et al., 2010 ) It may be possible that measures of motor functioning (e.g. the P urdue Peg Test) could show some relationship with GM volume changes in the precentral gyrus. This should be evaluated in a fut ure analysis or research study. This same research by McDonald and Colleagues (2010) also showed a "rebound" effect of GM values o ver time when

PAGE 44

36 observed at 1 year post treatment. It could be that if we had an additional measurement point at 1 year or longer we could have observed the same sort of effect in this sample. When evaluating the second aim that there would be relationships between changes in GM in the a priori ROIs and both measures of inflammation and scores on depression and anxiety scales, we found several relationships that supported the hypotheses both between baseline and 3 months and between 3 months and 9 months post treatment. With regard to changes between baseline and 3 months, there was a significant negative correlation between GM volume in the right MFG and levels of IL 6 (Figure 1). These findings are in line with our hypot hesis, but are somewhat inconsistent with previous work evaluating relationships between GM changes and markers of inflammation ( Kesler et al., 2013 ) Kesler and colleagues (2013) found a positive relationship between hippocampa l GM volume and IL6 levels fo llowing chemotherapy treatment, and a negative relationship between hippocampal GM volume and TNF levels. We did not find any significant relationships with changes in TNF levels over the course of treatment in the BrCA+ gr oup. It could be that these opposing findings are due to both the anti and pro inflammatory roles that IL 6 plays in the immune system. For example, levels of IL 6 could be increased initially due to their pro inflammatory response, but then levels could decrease or increase because of the various roles this cytokine plays in the inflammation process. The study by Kesler and colleagues was limited by the fact that they were not able to determine the acute effects of chemotherapy treatment on these measures due to the long duration post treatment. This is an advantage of the present study over previous research. We did not find significant relationships between GM volume changes and changes in depression or anxiety between baseline and 3 months, or between c hanges in PIC levels and depression and anxiety during this interval.

PAGE 45

37 When evaluating relationships between changes on levels of GM volume, PICs, and le vels of anxiety and depression between 3 months and 9 months post treatment, we observed several correl ations that supported our hypotheses, as well as one finding that was in contradiction to our hypotheses Results of these change score correlations showed that changes in the left Cingulate cortex were negatively associated with changes in CES D scores s uch that less GM volume in the cingulate was associated with increased depressive symptoms. Previous VBM research has found significant reductions in the cingulate gyrus in patients experiencing depression, when compared to HC ( Rodriguez Cano et al., 2014 ) This is in line with our finding that reductions in GM volume in the cingulate may be associated with increased depressive symptoms. However, another recent study found increased GM volume in this region in a sample of un medicated depressed patients ( Yang et al., 2015 ) Thus, the relationship between GM volume in the left cingulate gyrus and levels of depression as measured by the CES D contributes to this knowledge base in support of a negative r elationship between these two variables. Future research should test this region and its association with depressive symptoms, specifically in BrCA patients undergoing various treatments that could affect quality of life. Co ntradict ory to our hypothesis there was a significant positive correlation between left Cingulate GM volumes and both st ate and trait anxiety as measured by the STAI. Previous research has shown reductions in GM volumes of patients with anxiety when compa red to healthy controls ( Moon et al., 2014 ) However, longitudinal effects on GM vo lume associated with symptoms of both state and trait anxiety are less well understood. One possible explanation for the findings in the present study is a nonlinearity, such that levels of anxiety did not reach clinically significant levels (as defined by DSM 5 criteria) and thus do not show the effects previously demonstrated in the review by Moon and Colleagues (2014) This finding also

PAGE 46

38 suggests that there m a y be inverse effects of depressive and anxious symptoms in the left Cingulate cortex such that depression is associated with decreased GM volume, while symptoms of anxiety are associated with increased GM volume. Future longitudinal studies should address this question in order to further evaluate the effects of psychological characteristi cs on GM volumes in the cingulate cortex. Further, we observed a negative relationship between GM volume in the parahippocampal gyrus and levels of IL6 This finding further supports our hypotheses predicting a negative relationship between levels of in fl ammation and GM volumes. In the study done by Kesler and Colleagues (2013), lower hippocampal volumes were associated with lower levels of IL6. This is contrary to the findings in the present study. Again, this may be due to the dual role (both pro and an ti inflammatory) IL6 plays in the inflammatory process. The parahippocampal gyrus is a region that encompasses the hippocampus and is involved in memory encoding and retrieval. Thus, it is possible that increased levels of IL6 could be related to measures of memory storage and retrieval. Future research should directly test this hypothesis. One can speculate that levels of IL6, which also showed contradictory findings, could moderate the decrease in GM volume in the cingulate gyrus in relation to reductions in STAI scores Finally, our exploratory analyses showed a negative relationship between CRP levels and perceived stress in the BrCA+ group between baseline and 3 months post treatment. Further, a positive relationship was observed between changes in rig ht MFG GM volume and perceived stress as measured by the single self report item. These results did not support our exploratory hypotheses, and seem to contradict previous research as well. Between 3 months and 9 months post treatment, it was found that CE S D scores were positively associated with changes in perceived stress associated with treatment. Further, changes in both state and trait anxiety were

PAGE 47

39 positively correlated with memory complaints in these patients. These results are in support of our pro posed exploratory hypotheses and seem to be in line with previous research evaluating cognitive complaints in this population ( Sim—, Rifˆ Ros, Rodriguez Fornells, & Bruna, 2013 ) It seems that when measuring psychological self report variables (e.g. perceived effe cts of stress on health depression, anxiety), diffe rences occur later in treatment, and the acute effects are not as prominent in this sample. Strengths and Limitations There are several limitations that must be conside red when drawing inferences from the results of the present study. One limitation is how to interpret GM changes. Although the exact biological mechanism remains unclear, GM variability can be due to a number of different factors such as differences in cell volume, synaptic densities, or blo od flow and interstitial fluid ( Gage, 2002 ) Because of this variability in potential mechanisms leading to changes in GM, it is difficult to draw valid conclusions in regard to the outcomes that correlate with such a measure. Although there is no way to "best" interpret GM changes, the present analyses include d the cytokine data in order to allow for more direct associations between chemothe rapy inflammation, and GM changes. Cytokines are a choice candidate as a related biomarker as some cytokines have been shown to pass through the blood brain barrier and could have potential consequences on brain neurogenesis and function ( McDonald et al., 2013 ) Further, in order for VBM to be valid, several assumptions must be met ( Ashburner & Friston, 2000 ) First, the initial segmentation must correctly identify GM and WM. This is determined by several factors such as image qual ity and level of contrast between GM and WM tissue ( Ashburner & Friston, 2000 ) In order to verify that segmentation occurred corre ctly, all GM segmentations were visually inspected and plotted using a box plot to determine if there

PAGE 48

40 we re any outliers in the data set. A second assumption is that all confounding effects will be eliminated or accounted for in the statistical model ( Ashburner & Friston, 2000 ) We atte mpt ed to account for this assumption by co registering the data within subjects, and then between subjects in order to reduce the noise produced over time, and across subject's data. Co registration is a method used to correct for motion between images, th us reducing the potential noise in the data between time point s. Additionally, thresholds were applied to exclude any GM values that are <0.1, which indicates it may be an incorrectly labeled WM voxel. A third issue related to the validity of VBM is the as sumptions held by the statistical tests. It is important to understand how the distribution of the data will affect the statistical tests. Parametric tests assume the data are normally distributed. If data are not normally distributed, then a non parametri c test is indicated in order to maintain the validity of the VBM analysis ( Ashburner & Friston, 2000 ) Another limitation of this longitudinal design is the history of the participants. During the intervals betw een assessment visits, participants may have experienced a wide array of different personal and emotional factors that could lead to changes in the measures under study However, in the present study, the measures selected represent constructs that are not thought to be as variable as other potential psychological outcomes Therefore, experiences o utside the range of testing likely had minimal influence on the final results. A third limitation is that peripheral cytokines are not directly related t o GM ch anges in the CNS However, there is evidence that inflammation in the periphery can have effects on the CNS via a "weakened" BBB as mentioned above ( Merrill & Benveniste, 1996 ) Additionally, the brain produces cytokines in response to central inflammation ( Merrill & Benveniste, 1996 ) It is not feasible to determine definitively whether PICs in brain tissue directly cause GM MRI

PAGE 49

41 changes, especially among women in this population The data permit ted however, assessment of the magnitude of the asso ciation between the effects of chemotherapy GM changes, and burden of inflammation, and yield additional relevant data that adds to what is curr ently known about these relationships A final limitation is that the intended sample size to be included in this study was significantly reduced in association with a 19% cut in the budgets of new research grants awarded by the National Institute on Agi ng (NIA) in 2003 Originally, it was proposed that we would collect neuroimaging data from all three time points for every participant, but the budget shortfall made this impossible, leaving the study slightly under powered for these analyses. Despite the limita tions mentioned above, there were sever al strengths of this study. First, despite the need to reduce the use of imaging in response to the NIH budget cuts, we were able to obtain a relatively large sample for an imaging study and should be able to d raw valid inferences from these data. Second, the longitudinal design allows us to evaluate changes over time, and this is one of the first studies to look at these changes in relation to PICs, considered to be markers of inflammation. Additionally, the pr ospective design further ed our understanding of the effects of chemotherapy on the brain in women with BrCA A final strength of this study is that the data collection involved multiple components (e.g. psychosocial data, brain imaging, markers of inflammation ). Each of these constructs contribute s differently to the overall outcomes in this population. Therefore, the data collection methods employed in the present study attempted to answer questions from a multi dimensional approach, w hich would no t be possible with out the multiple components involved in the study.

PAGE 50

42 Future Directions Given the results of the present study, future research and analyses of these data can seek to replicate and extend on these findings. For example, evaluation of spe cific cognitive measures across time in this population could shed further light on the relationship between changes in GM volume in these specific regions and self reported memory complaints in patients undergoing chemotherapy treatment. A future study co uld also look at a longer duration following chemotherapy treatment in order to evaluate the longer term effects of chemotherapy treatment specially looking at regions seen in previous work. Another interesting study could be to examine scores on task of motor processing (e.g. Trails test or Purdue Pegboard test) in order to further evaluate the GM changes in the precentral gyrus. It may be the case that alterations of GM volume in the precentral gyrus are associated with impaired processing speed and/or motor coordination. Future research should also continue to look at the relationship between the cingulate gyrus and measures of depression. Decreased GM volume in this region has cons istently been shown in the past in patients with BrCA and with mood diso rders such as depression, and this fits with the integrative and introceptive functioning of the cingulate gyrus in the context of depressive symptoms. Finally, more studies should try to replicate GM changes in specific regions found previously. Conclusio ns The relationship between GM volume changes and chemotherapy treatment has been a relatively consistent finding in the literature. The most consistent findings have been in areas in the prefrontal cortex with other areas found less consistently (e.g. hi ppocampus, cingulate) to change with chemotherapy treatment. What the research has shown thus far is that the relationship between GM volume changes and chemotherapy is very complex, and involves

PAGE 51

43 numerous factors at the biological level of the individual c ells and at the psychological level of the patient as a whole. This study attempted to add to the literature and help further our understanding of the relationships between chemotherapy and GM volume changes by testing specific regions that have been found in previous work. One conclusion that can be made is that GM changes are difficult to replicate. Sampling error becomes an issue when different MRI scanners are used and various processing techniques are applied to accomplish the task of quantifying GM vo lume. However, VBM studies in this population have not sought to replicate findings in any specific regions. More research is needed that seeks to replicate previous GM changes in specific regions that have been found to change with chemotherapy treatment. It seems that different regions show different responses to the treatment process some may show increases while others show decreases. VBM as a method is limited in that there is no clear biological explanation for these changes and discrepancies in res ults. The more prominent conclusion from the present analyses is that there are relatively strong negative associations between changes in GM volume and measures of cytokines, namely IL6. IL6 has an inhibitory effect on TNF ", and in the present study there were no significant relationships between changes in the GM ROIs and TNF ". One possible explanation for this is that BrCA+ pat ients with higher levels of IL6 show lower levels of TNF ", which may fall below a threshold that is detectable by the corr elations conducted in the present study. In conclusion, more research is needed to help bolster our understanding of the process of chemotherapy in BrCA patients. Once these relationships are better understood, we may then turn to clinical implications f or these patients that could help promote decreased inflammation and perhaps in turn help mitigate some of the cognitive complaints these patients experience. However, we remain quite far from these treatment implications and studies such as this, which

PAGE 52

44 s eek to replicate previous work, are needed in order to develop a clearer picture of the processes of disease progression and changes on biological and psychological measure s with chemotherapy treatment.

PAGE 53

45 R EFERENCES American Cancer Society. Breast Cancer Facts & Figures 2013 2014 Atlanta: American Cancer Soc iet y, Inc. 2013 Ashburner, J. (2007). A fast diffeomorphic image registration algorithm. Neuroimage, 38 (1), 95 113. doi: 10.1016/j.neuroimage.2007.07.007 Ashburner, J., & Friston, Karl J. (2000). Voxel Based Morphometry The Methods. Neuroimage, 11 (6), 805 821. doi: 10.1006/nimg.2000.0582 Barnes, L. L. B., Harp, D., & Jung, W. S. (2002). Reliability Generalization of Scores on the Spielberger State Trait Anx iety Inventory. Educational and Psychological Measurement, 62 (4), 603 618. doi: 10.1177/0013164402062004005 Bower, J. E., Ganz, P. A., Irwin, M. R., Kwan, L., Breen, E. C., & Cole, S. W. (2011). Inflammation and behavioral symptoms after breast cancer trea tment: do fatigue, depression, and sleep disturbance share a common underlying mechanism? J Clin Oncol, 29 (26), 3517 3522. doi: 10.1200/JCO.2011.36.1154 Buckner, R. L., Head, D., Parker, J., Fotenos, A. F., Marcus, D., Morris, J. C., & Snyder, A. Z. (2004) A unified approach for morphometric and functional data analysis in young, old, and demented adults using automated atlas based head size normalization: reliability and validation against manual measurement of total intracranial volume. Neuroimage, 23 (2) 724 738. doi: 10.1016/j.neuroimage.2004.06.018 Cimprich, Bernadine, Reuter Lorenz, Patricia, Nelson, James, Clark, Patricia M, Therrien, Barbara, Normolle, Daniel, Welsh, Robert C. (2010). Prechemotherapy alterations in brain function in women with breast cancer. Journal of Clinical and Experimental Neuropsychology, 32 (3), 324 331. doi: 10.1080/13803390903032537

PAGE 54

46 Conroy, Susan K, McDonald, Brenna C, Smith, Dori J, Moser, Lyndsi R, West, John D, Kamendulis, Lisa M, Saykin, Andrew J. (2012). Alte rations in brain structure and function in breast cancer survivors: effect of post chemotherapy interval and relation to oxidative DNA damage. Breast Cancer Research and Treatment, 137 (2), 493 502. doi: 10.1007/s10549 012 2385 x Coussens, L. M., & Werb, Z. (2002). Inflammation and cancer. Nature, 420 (6917), 860 867. doi: 10.1038/nature01322 Dantzer, R., & Kelley, K. W. (2007). Twenty years of research on cytokine induced sickness behavior. Brain Behav Immun, 21 (2), 153 160. doi: 10.1016/j.bbi.2006.09.006 de Ruiter, Michiel B, Reneman, Liesbeth, Boogerd, Willem, Veltman, Dick J, Caan, Matthan, Douaud, Gwena‘lle, Schagen, Sanne B. (2012). Late effects of high dose adjuvant chemotherapy on white and gray matter in breast cancer survivors: Converging resul ts from multimodal magnetic resonance imaging. Human Brain Mapping, 33 (12), 2971 2983. doi: 10.1002/hbm.21422 Dethlefsen, C., Hojfeldt, G., & Hojman, P. (2013). The role of intratumoral and systemic IL 6 in breast cancer. Breast Cancer Res Treat, 138 (3), 6 57 664. doi: 10.1007/s10549 013 2488 z Dietrich, J. (2010). Chemotherapy associated central nervous system damage. Adv Exp Med Biol, 678 77 85. Gage, F. H. (2002). Neurogenesis in the adult brain. J Neurosci, 22 (3), 612 613. Goehler, L. E., Gaykema, R. P., Hansen, M. K., Anderson, K., Maier, S. F., & Watkins, L. R. (2000). Vagal immune to brain communication: a visceral chemosensory pathway. Auton Neurosci, 85 (1 3), 49 59. doi: 10.1016/S1566 0702(00)00219 8

PAGE 55

47 Inagaki, Masatoshi, Yoshikawa, Eisho, Matsuoka, Yutaka, Sugawara, Yuriko, Nakano, Tomohito, Akechi, Tatsuo, Group, and The Breast Cancer Survivors' Brain MRI Database. (2007). Smaller regional volumes of brain gray and white matter demonstrated in breast cancer survivors exposed to adjuvant chemotherapy. Cancer, 109 (1), 146 156. doi: 10.1002/cncr.22368 Janelsins, M. C., Roscoe, J. A., Berg, M. J., Thompson, B. D., Gallagher, M. J., Morrow, G. R., Gross, R. A. (2010). IGF 1 partially restores chemotherapy induced reductions in neural cel l proliferation in adult C57BL/6 mice. Cancer Invest, 28 (5), 544 553. doi: 10.3109/07357900903405942 Julian, Laura J. (2011). Measures of anxiety: State Trait Anxiety Inventory (STAI), Beck Anxiety Inventory (BAI), and Hospital Anxiety and Depression Scale Anxiety (HADS A). Arthritis Care & Research, 63 (S11), S467 S472. doi: 10.1002/acr.20561 Kesler, S., Janelsins, M., Koovakkattu, D., Palesh, O., Mustian, K., Morrow, G., & Dhabhar, F. S. (2013). Reduced hippocampal volume and verbal memory performance associated with interleukin 6 and tumor necrosis factor alpha levels in chemotherapy treated breast c ancer survivors. Brain Behav Immun, 30 Suppl S109 116. doi: 10.1016/j.bbi.2012.05.017 Koppelmans, V., Breteler, M. M., Boogerd, W., Seynaeve, C., & Schagen, S. B. (2013). Late effects of adjuvant chemotherapy for adult onset non CNS cancer; cognitive impa irment, brain structure and risk of dementia. Crit Rev Oncol Hematol, 88 (1), 87 101. doi: 10.1016/j.critrevonc.2013.04.002

PAGE 56

48 Lai, Chien Han. (2013). Gray matter volume in major depressive disorder A meta analysis of voxel based morphometry studies. Psychiatr y Research: Neuroimaging, 211 (1), 37 46. doi: 10.1016/j.pscychresns.2012.06.006 Liu, L., Mills, P. J., Rissling, M., Fiorentino, L., Natarajan, L., Dimsdale, J. E., Ancoli Israel, S. (2012). Fatigue and sleep quality are associated with changes in in flammatory markers in breast cancer patients undergoing chemotherapy. Brain Behav Immun, 26 (5), 706 713. doi: 10.1016/j.bbi.2012.02.001 Mariotto, A B, Robin Yabroff, K, Shao, Y, Feuer, E J, & Brown, M L. (2011). Projections of the Cost of Cancer Care in th e United States: 2010 2020. JNCI Journal of the National Cancer Institute, 103 (2), 117 128. doi: 10.1093/jnci/djq495 McDonald, B. C., Conroy, S. K., Smith, D. J., West, J. D., & Saykin, A. J. (2013). Frontal gray matter reduction after breast cancer chemot herapy and association with executive symptoms: a replication and extension study. Brain Behav Immun, 30 Suppl S117 125. doi: 10.1016/j.bbi.2012.05.007 McDonald, B. C., Conroy, Susan K, Ahles, Tim A, West, John D, & Saykin, Andrew J. (2010). Gray matter r eduction associated with systemic chemotherapy for breast cancer: a prospective MRI study. Breast Cancer Research and Treatment, 123 (3), 819 828. doi: 10.1007/s10549 010 1088 4 Merrill, J.E., & Benveniste, E.N. (1996). Cytokines in inflammatory brain lesio ns helpful and harmful. Trends Neuroscience, 19 331 338. Mohamed, M. M., Al Raawi, D., Sabet, S. F., & El Shinawi, M. (2014). Inflammatory breast cancer: New factors contribute to disease etiology: A review. J Adv Res, 5 (5), 525 536. doi: 10.1016/j.jare .2013.06.004

PAGE 57

49 Moon, Chung Man, Kim, Gwang Won, & Jeong, Gwang Woo. (2014). Whole brain gray matter volume abnormalities in patients with generalized anxiety disorder. NeuroReport, 25 (3), 184 189. doi: 10.1097/WNR.0000000000000100 Pomykala, K. L., de Ruiter, M. B., Deprez, S., McDonald, B. C., & Silverman, D. H. (2013). Integrating imaging findings in evaluating the post chemotherapy brain. Brain Imaging Behav, 7 (4), 436 452. doi: 10.1007/s11682 013 9239 y Pomykala, K. L., Ganz, P. A., Bower, J. E., Kwan, L., Castellon, S. A., Mallam, S., Silverman, D. H. (2013). The association between pro inflammatory cytokines, regional cerebral metabolism, and cognitive complaints following adjuvant chemotherapy for breast cancer. Brain Ima ging Behav, 7 (4), 511 523. doi: 10.1007/s11682 013 9243 2 Radloff, L S. (1977). The CES D Scale: A Self Report Depression Scale for Research in the General Population. Applied Psychological Measurement, 1 (3), 385 401. doi: 10.1177/014662167700100306 Rodrig uez Cano, E., Sarro, S., Monte, G. C., Maristany, T., Salvador, R., McKenna, P. J., & Pomarol Clotet, E. (2014). Evidence for structural and functional abnormality in the subgenual anterior cingulate cortex in major depressive disorder. Psychol Med, 44 (15) 3263 3273. doi: 10.1017/S0033291714000841 Sim—, Marta, Rifˆ Ros, Xavier, Rodriguez Fornells, Antoni, & Bruna, Jordi. (2013). Chemobrain: A systematic review of structural and functional neuroimaging studies. Neuroscience and Biobehavioral Reviews, 37 (8), 1311 1321. doi: 10.1016/j.neubiorev.2013.04.015 Taylor, J. A. (1953). A personality scale of manifest anxiety. J Abnorm Psychol, 48 (2), 285 290.

PAGE 58

50 Vgontzas, A. N., Zoumakis, M., Bixler, E. O., Lin, H. M., Prolo, P., Vela Bueno, A., Chrousos, G. P. (2 003). Impaired nighttime sleep in healthy old versus young adults is associated with elevated plasma interleukin 6 and cortisol levels: physiologic and therapeutic implications. J Clin Endocrinol Metab, 88 (5), 2087 2095. doi: 10.1210/jc.2002 021176 Wang, X M., Walitt, B., Saligan, L., Tiwari, A. F., Cheung, C. W., & Zhang, Z. J. (2015). Chemobrain: A critical review and causal hypothesis of link between cytokines and epigenetic reprogramming associated with chemotherapy. Cytokine, 72 (1), 86 96. doi: 10.101 6/j.cyto.2014.12.006 Yang, X., Ma, X., Li, M., Liu, Y., Zhang, J., Huang, B., Ma, X. (2015). Anatomical and functional brain abnormalities in unmedicated major depressive disorder. Neuropsychiatr Dis Treat, 11 2415 2423. doi: 10.2147/NDT.S93055