Citation
Mechanics of the proximal pulmonary artery in pulmonary hypertension

Material Information

Title:
Mechanics of the proximal pulmonary artery in pulmonary hypertension
Creator:
Burgett, Shawna L. ( author )
Language:
English
Physical Description:
1 electronic resource (241 pages) : ;

Subjects

Subjects / Keywords:
Pulmonary hypertension ( lcsh )
Pulmonary hypertension -- artery -- Diseases ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Pulmonary hypertension (PH) is an incurable, progressive disease with poor survival. PH is defined as an increase in mean pulmonary arterial pressure above 25mmHg at rest. Despite the rapid advances in PH therapies, the progressive nature of the disease eventually results in right heart failure and death. PH is characterized by vasoconstriction, remodeling, and thickening of the small to medium arteries and arterioles, resulting in increased pulmonary vascular resistance. Pulmonary vascular resistance is often the single parameter used to assess disease severity and response to treatment. However, other markers of vascular function, for example pulmonary vascular stiffness, area strain and vascular capacitance have been shown as predictive of mortality in PH. ( ,, )
Review:
This thesis includes three studies that focus on the intersection of clinical and scientific investigations to evaluate mechanical changes in the proximal pulmonary artery during PH and their impact on patient outcomes. The first is a clinical study utilizing a range of clinical measures in combination to accurately predict future patient health status to facilitate proactive clinical therapies. The second and third studies involved animal models. The second study focused on the mechanical impacts of preconditioning tissue. The third study looked at the mechanical changes in the proximal pulmonary artery upon PH onset and recovery. All of these studies centered on defining the mechanical function of the proximal pulmonary arteries through time in order to assess health status. We aim to quantify changes in the proximal pulmonary arteries in the setting of PH to help guide therapeutic targets and clinical interventions.
Review:
The findings of these three studies support that the the proximal pulmonary artery mechanics are important for prediction of future patient health status. In animal studies of soft tissue using pressure-inflation testing, sample preparation may significantly affect the resulting mechanics measurements. Care must be taken in tissue preparation and testing in order to preserve the mechanical characteristics as closely as possible to in-vivo settings. Finally, proximal PA segments from rats that have chronic hypoxia-induced PH and are allowed to recover for six weeks in normoxia have impaired mechanics: decreased vessel diameters, decreased vessel compliance and decreased wall tension as compared with age matched controls which may cause longer term right heart dysfunction. We also found, with hypoxic exposure, the PA segments have decreased VSM contraction and that VSM contraction returns to normal levels upon recovery. We speculate that therapies aimed at reducing or preventing collagen deposition may improve proximal PA mechanics upon recovery from PH or upon normalization of the hemodynamics.
Thesis:
Thesis (Ph.D.) - University of Colorado Denver.
Bibliography:
Includes bibliographic references
System Details:
System requirements: Adobe Reader.
General Note:
Department of Bioengineering
Statement of Responsibility:
by Shawna L. Burgett.

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:
945986646 ( OCLC )
ocn945986646
Classification:
LD1193.E56 2015d B97 ( lcc )

Downloads

This item has the following downloads:


Full Text
MECHANICS OF THE PROXIMAL PULMONARY ARTERY IN PULMONARY
HYPERTENSION
by
SHAWNA L. BURGETT
M.S., University of Colorado Denver, 2009
B.S., Montana State University, Bozeman, 2001
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
Doctor of Philosophy
Bioengineering
2015


This thesis for the Doctor of Philosophy degree by
Shawna L. Burgett
has been approved for the
Department of Bioengineering
by
Kendall Hunter, Advisor
John S. Walker, Advisor
Robin Shandas, Chair
Dunbar Ivy
Reuben Blair Dodson
November 20, 2015
n


Burgett, Shawna L. (Ph.D., Bioengineering)
Mechanics of the Proximal Pulmonary Artery in Pulmonary Hypertension
Thesis directed by Assistant Professor Kendall Hunter and Assistant Professor John
S. Walker
ABSTRACT
Pulmonary hypertension (PH) is an incurable, progressive disease with poor
survival. PH is defined as an increase in mean pulmonary arterial pressure above
25mmHg at rest. Despite the rapid advances in PH therapies, the progressive nature
of the disease eventually results in right heart failure and death. PH is characterized
by vasoconstriction, remodeling, and thickening of the small to medium arteries and
arterioles, resulting in increased pulmonary vascular resistance. Pulmonary vascular
resistance is often the single parameter used to assess disease severity and response
to treatment. However, other markers of vascular function, for example pulmonary
vascular stiffness, area strain and vascular capacitance have been shown as predictive
of mortality in PH.
This thesis includes three studies that focus on the intersection of clinical and
scientific investigations to evaluate mechanical changes in the proximal pulmonary
artery during PH and their impact on patient outcomes. The first is a clinical study
utilizing a range of clinical measures in combination to accurately predict future pa-
tient health status to facilitate proactive clinical therapies. The second and third
studies involved animal models. The second study focused on the mechanical im-
pacts of preconditioning tissue. The third study looked at the mechanical changes in
the proximal pulmonary artery upon PH onset and recovery. All of these studies cen-
tered on defining the mechanical function of the proximal pulmonary arteries through
time in order to assess health status. We aim to quantify changes in the proximal


pulmonary arteries in the setting of PH to help guide therapeutic targets and clinical
interventions.
The findings of these three studies support that the the proximal pulmonary
artery mechanics are important for prediction of future patient health status. In
animal studies of soft tissue using pressure-inflation testing, sample preparation may
significantly affect the resulting mechanics measurements. Care must be taken in
tissue preparation and testing in order to preserve the mechanical characteristics as
closely as possible to in-vivo settings. Finally, proximal PA segments from rats that
have chronic hypoxia-induced PH and are allowed to recover for six weeks in normoxia
have impaired mechanics: decreased vessel diameters, decreased vessel compliance and
decreased wall tension as compared with age matched controls which may cause longer
term right heart dysfunction. We also found, with hypoxic exposure, the PA segments
have decreased VSM contraction and that VSM contraction returns to normal levels
upon recovery. We speculate that therapies aimed at reducing or preventing collagen
deposition may improve proximal PA mechanics upon recovery from PH or upon
normalization of the hemodynamics.
The form and content of this abstract are approved. I recommend its publication.
Approved: Kendall Hunter
Approved: John S. Walker
IV


DEDICATION
For my husband, Terry, my son, Derek Zane, and my daughter, Kendra Quin,
who inspire me every day to be myself and to be better than myself. You are my
heart. To my parents Joyce and Ken for inspiring me with the love of learning, the
love of experimenting and the confidence to do anything. To my sister, Jennifer, for
keeping me grounded by our shared roots, for the comfort of our similarities and
for inspiring me toward our differences which seem achievable and delightful by your
example. To my friends, Andrea, Nancy and Renee for your wonderful company, for
keeping me on your contact lists and giving my kids a second home. To Andrew for
your uncanny timing and insanely good edits. Again to Nancy-Ant, you are the best
softball player, lawyer, adviser, scheduler, editor, strategist, planner and day-saver
ever; there are no more Xs.
I am extremely lucky and eternally grateful to have you all in my life. Without
your love, friendship and support this endeavour would not have been possible. I
dedicate this thesis to you all.
v


ACKNOWLEDGMENT
Special thanks go to my advisers, Kendall Hunter and John S. Walker, to whom
I owe a great deal of gratitude for mentorship, support and guidance. Many thanks
to Robin Shandas for making all this possible through his discussions regarding this
work, his teaching and as the founder and visionary of UCD Bioengineering. Special
thanks to Blair Dodson for the access, set-up, mentoring, and support necessary to run
the mechanical testing apparatus and graciously sharing it for the following 3 years. I
would like to thank Dr. Dunbar Ivy, Dr. Steve Abman, and Dr. Lori Walker for many
discussions and suggestions on this work. I wish to thank Dr. Kurt Stenmark and
the Cardio-Vascular Pulmonary Research group for allowing access to their hypobaric
chambers. Special thanks to Dr. Michael Wunder for statistical methods mentoring.
Many thanks to Jules Harral for the hemodynamic testing and for always finding time
for these studies. Thanks also to Greg Seedorf for his generosity with equipment, set-
up and lab space. Thank you to Dr. Xiaotao Li and Yanmai Du for their help in the
lab, day in and day out. Thanks to Andrew Eitel, Nicholas Hobson and Melanie Dufva
for help with data collection. Finally, thank you to my colleagues in the first class of
UCD Bioengineering, especially, Bryan Yunker, Stephen Humphries and Derek Eilers
for their support and comradery as we navigated this new territory.
This thesis would not have been possible without the generous support from P.I.
Robin Shandas, NIH Training Grant 5T32HL072738-09, P.I. Kendall Hunter NIH
Mentored Quantitative Research Development Award NHLBI-K25 HL094749, lab
support from John S. Walker, lab equipment from R. Blair Dodson, and clinical data
and perspective from Dunbar D. Ivy.
vi


TABLE OF CONTENTS
Tables....................................................................... xii
Figures .................................................................... xiii
Chapter
1. Motivation and Specific Aims................................................ 1
2. Background.................................................................. 3
2.1 The Vascular Circuit Design.......................................... 3
2.1.1 Historical Perspective: de Motu Cordis......................... 6
2.2 Mean Blood Flow Depends on Pressure and Resistance .................. 7
2.2.1 Blood Flow..................................................... 7
2.2.1.1 Historical Perspective: Poiseuilles Law................. 9
2.2.2 Vascular Resistance .......................................... 10
2.2.3 Blood Pressure................................................ 11
2.2.3.1 Historical Perspective: Windkessel...................... 13
2.3 Artery Wall Anatomy................................................. 14
2.3.1 Intima........................................................ 15
2.3.2 Media......................................................... 15
2.3.3 Adventitia.................................................... 15
2.3.4 Artery Wall Composition....................................... 16
2.3.4.1 Endothelium............................................. 16
2.3.4.2 Smooth Muscle Cells..................................... 17
2.3.4.3 Collagen................................................ 18
2.3.4.4 Elastin................................................. 21
2.4 Regulation of Blood Vessels......................................... 22
2.5 Hypoxic Vasoconstriction............................................ 22
2.6 Pulmonary Hypertension.............................................. 23
2.7 Large Arteries in Pulmonary Hypertension............................ 25
vii


2.7.1 Extracellular Response........................................... 25
2.7.2 Cellular Response ............................................... 25
2.8 Evaluation of Arterial Mechanics..................................... 27
2.8.1 In-Vivo.......................................................... 28
2.8.1.1 Right Heart Catheterization............................... 28
2.8.2 In-Vitro......................................................... 29
2.8.2.1 Unixaial Test............................................. 29
2.8.2.2 Ring Test................................................. 30
2.8.2.3 Biaxial Test.............................................. 30
2.8.2.4 Pressure Inflation Test................................... 30
2.8.2.5 Pre-Conditioning.......................................... 31
2.9 Recovery Studies....................................................... 32
2.10 Studies in this Thesis................................................ 33
3. Study: Multivariate Prediction of Clinical Indicators ..................... 34
3.1 Abstract............................................................... 34
3.2 Introduction........................................................... 35
3.3 Methods................................................................ 36
3.3.1 Study Design..................................................... 36
3.3.2 Patient Population............................................... 36
3.3.3 Clinical Data Collection......................................... 37
3.3.4 Statistical Analysis............................................. 38
3.4 Results................................................................ 39
3.4.1 Single Variable Measures as Predictors of Outcome................ 40
3.4.2 Combining Multiple Measures as Predictors of Outcome ... 41
3.4.3 Survival Predictors.............................................. 42
3.5 Discussion............................................................. 43
3.5.1 Study Limitations ............................................... 45
viii


3.5.2 Conclusions................................................. 45
3.6 Figures & Tables.................................................. 46
4. Study: The Mechanics of Preconditioning Pulmonary Artery Segments . 50
4.1 Abstract.......................................................... 50
4.2 Introduction...................................................... 51
4.3 Methods........................................................... 53
4.3.1 Animals..................................................... 53
4.3.2 Isolated Vessel Chamber Testing System...................... 53
4.3.3 Artery Segment Preparation.................................. 54
4.3.4 Non-Preconditioned Mechanical Measurements ................. 54
4.3.5 Preconditioned Mechanical Measurements...................... 55
4.3.6 PA Compliance: ............................................. 56
4.3.7 PA Diameter & Vascular Smooth Muscle Activity............... 56
4.3.8 PA Wall Tension............................................. 56
4.3.9 Endothelium and VSM Response................................ 57
4.3.10 Statistics.................................................. 58
4.4 Results........................................................... 58
4.5 Discussion........................................................ 59
4.5.1 Limitations................................................. 60
4.5.2 Conclusions................................................. 60
4.6 Acknowledgments................................................... 61
4.7 Figures........................................................... 61
5. Study: Proximal Pulmonary Artery Mechanics in Pulmonary Hyptertension
and Recovery........................................................... 66
5.1 Abstract.......................................................... 66
5.2 Introduction...................................................... 67
5.3 Methods........................................................... 68
IX


5.3.1 Animals................................................... 68
5.3.2 Hypoxic Chambers.......................................... 68
5.3.3 Hemodynamic Measurements.................................. 68
5.3.4 Isolated Vessel Chamber Testing System.................... 69
5.3.5 Artery Segment Mechanical Measurements.................... 70
5.3.6 Derived Measures.......................................... 72
5.3.7 Histology................................................. 73
5.3.8 Experimental Design & Statistical Analysis................ 73
5.4 Results......................................................... 74
5.4.1 Animals................................................... 74
5.4.2 Hemodynamics.............................................. 74
5.4.3 Mechanical Measurements................................... 75
5.4.4 PA Compliance............................................. 76
5.4.5 Ability of Vascular Smooth Muscle to Alter Diameter....... 76
5.4.6 PA Wall Tension........................................... 77
5.4.7 Histology................................................. 78
5.5 Discussion...................................................... 79
5.5.1 Vessel Properties......................................... 79
5.5.2 Hemodynamics.............................................. 80
5.5.3 PH Development............................................ 81
5.5.4 Recovery from Hypoxic Exposure............................ 81
5.5.5 Limitations............................................... 82
5.5.6 Conclusions............................................... 83
5.6 Figures......................................................... 84
6. Conclusions......................................................... 89
6.1 Major Findings ................................................. 89
6.2 Clinical Relevance.............................................. 90
x


6.3 Scope within the Existing Literature.............................. 91
6.4 Future Studies.................................................... 92
6.5 Concluding Remarks................................................ 92
References.................................................................. 94
Appendix
A. Top Predictive Models................................................... 106
B. R-code Study 1, Multivariate Prediction .............................. 108
C. R-code Study 2, Preconditioning......................................... 164
D. R-code Study 3, Mechanics of the PPA.................................. 171
E. Summary Data of Study 3, PD Measurements and Calculations ............ 219
F. Animal Protocol Detail Report .......................................... 220
xi


TABLES
Table
3.1 Patient Demographics................................................. 46
3.2 Cox Proportional Hazards............................................. 48
4.1 Summary Data of Preconditioning Measurements and Calculations. ... 64
xii


FIGURES
Figure
2.1 The Cardio-Pulmonary Circuits............................................ 5
2.2 Blood Pressures through the Cardiovascular Circulation.................. 11
2.3 Windkessel Represents Arterial Compliance and Elastic Recoil............ 12
2.4 Vessel Wall Anatomy: Arteries and Veins ................................ 14
2.5 Smooth Muscle Cells..................................................... 16
2.6 Collagen Triple Helix Structure and Aggregation ........................ 18
2.7 Elastin................................................................. 20
2.8 In- Vitro Testing-System schematic diagrams............................. 27
2.9 Pressure Inflation Testing System Schematic............................. 31
3.1 Color M-Mode Tissue Doppler Image of the RPA, lumen inner walls and
echocardiograph......................................................... 47
3.2 Univariate analyses confirmed by three methods: Regression, Random
Forest, AIC............................................................. 48
3.3 Prediction Results by WHO-FC Category for PVRI and the Top Equation
Model................................................................... 49
3.4 Cox Proportional Hazards................................................ 49
4.1 Ten Successive Stress-Strain Curves..................................... 61
4.2 Schematic of the Pressure Inflation Testing Chamber Apparatus........... 62
4.3 Data Traces: diameter and pressure paired data points................... 62
4.4 Data from a Complete Test of One Control Artery......................... 63
4.5 Pressure-Diameter Measurements.......................................... 63
4.6 Difference in diameter between active and passive vessels, AD........... 64
4.7 Stress-Strain Curves.................................................... 65
5.1 Experimental Design..................................................... 84
5.2 Pressure-Diameter paired points and records............................. 84
xiii


5.3 PA Diameter and Wall Stress........................................... 85
5.4 Ability of Vascular Smooth Muscle to Change Diameter, AD, through PH
and Recovery........................................................... 86
5.5 mPAP, CO, PVR, wall stress, compliance, diameter..................... 87
5.6 Mean Diameter Changes with Pressure and Group........................ 87
5.7 Structural Changes to the PA Measured by Tissue Areas of Histological
Sections............................................................... 88
xiv


Chapter 1
1. Motivation and Specific Aims
Pulmonary hypertension (PH) is a disorder of the pulmonary vasculature charac-
terized by increases in arterial pressure. PH is a rare condition with an incidence rate
of 1.1 to 2.4 cases per million per year [37]. However, with a five-year mortality rate
of 40%, progressively declining quality of life for patients, right heart dysfunction,
and increases in hospitalization, PH poses significant medical challenges [53]. This
is a complex, multifaceted disorder. Reflecting this complexity, the World Health
Organization (WHO) clinically classified PH into categories of disorders that share
similar characteristics, pathogenesis or therapeutic management [17] [98].
PH is characterized by vasoconstriction, remodeling, and thickening of the small
to medium arteries and arterioles [81],[35]. PH affects the whole lung vasculature
by narrowing the distal pulmonary vascular bed, dramatically increasing pulmonary
vascular resistance (PVR). This increased PVR has a commensurate impact on the
demand placed on the right ventricle of the heart which must force blood through the
narrowed vascular tree. Because of the characteristic distal narrowing, studies of PH
have focused intently on the distal vasculature.
There have been fewer studies involving the proximal pulmonary arteries. The
studies that have focused on proximal pulmonary arteries (PPA) in PH commonly
report thickening of the media and adventitial layers. Area strain of the PPA and
vascular capacitance have been shown to be predictive of PH outcomes and PPA input
impedance improves prediction in children with PH. Fewer studies have examined the
mechanics of the PPA in order to quantify vascular changes with respect to disease
progression. Specifically, the impact of changes in active vessel mechanics upon PPA
remodeling and mechanical stiffening in PH are still not well understood.
We chose to focus on the intersection of clinical work and scientific studies to
evaluate mechanical changes in the PPA during PH and their impact on patient
outcomes. This thesis includes three studies. The first is a clinical study focused on
utilizing a range of clinical measures in concert to accurately predict future patient
health status to facilitate proactive clinical therapies. The second and third studies
involved animal models. The second study focused on the mechanical impacts of
preconditioning tissue. The third study looked at the mechanical changes in the PPA
upon PH onset and recovery. All of these studies focused on defining the mechanical
changes and measures involved in PH with a goal of identifying opportunities for
1


future development of therapeutic treatments that provide long-term resolution of
PH related health issues.
Specific Aims
1. Aim 1
Hypothesis: Right heart catheterization (RHC) and Color M-Mode
Tissue Doppler Images (CMM-TDI) of proximal pulmonary artery
function improve the accuracy of patient outcome predictions in con-
firmed cases of pediatric pulmonary hypertension.
Approach: Analysis of hemodynamic, clinical, ultrasound and RHC data for 78
pediatric patients with confirmed Pulmonary Arterial Hypertension to deter-
mine what combination of clinical measurements taken at the initial RHC most
accurately predict patient outcome as defined by WHO Functional Classification
at the time of follow-up.
2. Aim 2
Hypothesis: Preconditioning pulmonary artery segments compro-
mises passive and active vessel wall mechanics significantly altering
diameter, compliance, vasoactive contraction, stress, strain, and elas-
tic modulus compared to non-preconditioned segments.
Approach:Compare preconditioned and unconditioned pulmonary artery seg-
ments from rats using measures of pressure and diameter to assess the impact
of preconditioning upon mechanical artery properties: diameter, compliance,
wall-stress, and modulus. Present a method for achieving repeatable mechani-
cal testing results without preconditioning.
3. Aim 3
Hypothesis: Wall stress is maintained at a constant level despite
marked changes in pulmonary artery pressure and wall thickness dur-
ing development of and upon recovery from Pulmonary Hypertesion.
Approach: Examine proximal pulmonary artery mechanics through a physio-
logic range of pressures and time in control, hypoxic and recovery animal groups.
Use mechanical testing to quantify PPA mechanics through the development of
and recovery from PH to determine if the mechanical adaptations in the PPA
exist to maintain the wall stress at a constant level in the face of changing vessel
wall loads.
2


Chapter 2
2. Background
2.1 The Vascular Circuit Design
The human body has approximately 1013 total cells comprised of 200 different
types. All cells in the body need nutrients, the exchange of gases, waste removal,
hormones, temperature control, damage control and defense. Blood is the bodily fluid
that delivers the substances necessary for those vital supportive functions. Blood is
transported through a vast array of branching vessels, the vascular system, to the
tissues and cells of the body [1].
The heart pumps blood through the vascular circuit creating the cardio-vascular
system. The heart may be thought of as two pumps in series which together with the
vascular network form a closed system of two circuits: the pulmonary circuit and the
systemic circuit [60] [42] [97]. Figure 2.1 shows a schematic of the heart and the two
circuits.
The pulmonary circuit begins with the right ventricle (the first pump) located on
the right-side of the heart. The right ventricle sends blood to the lungs to exchange
C02 for 02. The blood then flows back to the heart via the pulmonary veins and into
the left atrium. The systemic circuits begins with the left ventricle, (the second pump)
which sends the oxygenated blood to all the tissues in the body. This systemic circuit
allows for the exchange of gases between the blood and cells of the body returning
blood with lower 02 concentration and higher C02 concentration back to the right
atrium to reenter the pulmonary circuit, completing the full cardio-vascular loop [42],
Blood moves through the pulmonary and systemic circulations through branching
vascular systems with similar gross anatomy. In both circuits, the ventricles of the
heart pump blood into the arteries which carry blood away from the heart. The
3


arteries branch forming arterial trees or networks of smaller and smaller branches
with arterioles forming the smallest branches of the artery system. Arterioles branch
into capillaries which are the smallest vessels and the site of gas exchange. Capillaries
connect the artery network to the venous network or blood-return system. On the
venous side, capillaries rejoin or merge to form venules which, like arterioles, are the
smallest vessels of the venous system. Venules merge to form veins which merge into
larger veins and return the blood to the atria chambers of the heart.
Capillary exchange of gasses and solutes between the blood and the tissue is
the central purpose of the blood circulation. Therefore, most of the activities of
the circulatory system are centered around providing the capillaries with blood [97].
However, underlying this simplified statement lies the need for an incredibly high
degree of precision due to the complexity of passing blood through a network of
approximately 60,000 miles of capillaries which are 5-10 microns in diameter and a
single cell in thickness. This includes the need for precise timing, flow rate, pressure,
concentration and quality of blood [60].
4


CaplllBry beds
ol lungs where
gas exchange
occurs
Venae
cavae
Pulmonary
veins
Aorta and
branches
Key;
1= Oxygen-rich,
C02-poor blood
1= Oxygen-poor,
COj-rlch blood
Capillary
beds ol all
body tissues
where gas
exchange
occurs
Copyright 2000 by W.B. Saunders Company,
Cardiopulmonary Anatomy and Physiology
Figure 2.1: The Cardio-Pulmonary Circuits
5


2.1.1 Historical Perspective: de Motu Cordis
In 1628 William Harvey, an English physician and Physician Extraordinary to
King James I and King Charles I, published Exercitatio Anatomica de Motu Cordis
et Sanguinis in Animalibus (de Motu Cordis). The title is Latin for An Anatomical
Exercise on the Motion of the Heart and Blood in Living Beings. Harveys work
was based on prior experiments and publications. However, de Motu Cordis created
the most complete set of detailed descriptions to date to introduce and support the
idea of blood moving through a closed circle or circuit in one direction according
to the one-way valves in the heart and veins. This compilation included extensive
coverage of methods, experiments, and reported results to introduce and support this
idea [101]. De Motu Cordis suggested that the circulation is comprised of a double
circuit, from the heart to the lungs (pulmonary circulation), back to the heart, and
out to the main circulation (systemic circulation). Harvey was not able to discern flow
through the capillaries, however, and his theory of blood movement refuted an earlier
theory, thus it was considered controversial. His hypothesis that blood flow existed in
a closed circuit was proven after his death in 1660 by Marcello Malpighi. Malpighi, an
Italian physician, discovered the connection between the artery and venous systems
when he identified capillary flow in a frog lung with the use of a microscope [101].
The impact of de Motu Cordis went far beyond basic physiology as it also intro-
duced the application of quantification and mechanistic organizational structure to
the heart and vascular system. The view that the heart could be analyzed mechani-
cally as a pump and that its output could be quantified as producing 72 heart beats
per minute and throwing 540 pounds of blood every hour was a major departure
from the mystical and more religious views of the time which regarded the heart as
the seat of the spirit. The introduction of this concept forms foundational aspects
of modern cardiology.
6


2.2 Mean Blood Flow Depends on Pressure and Resistance
Although the cardio-vascular system can be explained as a simple closed circuit,
the varying characteristics of the vascular network such as changes in vessel wall
composition, geometry and interactions with blood and supported tissues make its
functioning exceedingly elaborate. Central to the proper functioning of the cardio-
vascular system is the need for adequate and consistent perfusion. This is accom-
plished through the coordination of blood pressure and vascular resistance to moder-
ate blood how.
2.2.1 Blood Flow
Blood how through the cardio-vascular system is measured by the blood volume
that passes through a point in a given amount of time. Cardiac output (CO) is the
measure of this volume from the heart and is commonly measured in liters per minute.
For a typical adult at rest, CO is 5 1/min. CO is related to the stroke volume (SV) of
the heart and heart rate (HR). The relationship among these factors is represented
by the equation:
CO = SV x HR (2.1)
[l/min] = [l/beat] x [beats/min]
Since blood how is unidirectional through the heart and vascular system in the
normal anatomy and in the absence of disease, as seen in Figure 2.1, CO is neces-
sarily matched for both the systemic and pulmonary circuits. In order to send blood
through the vast systemic circuit and through the more proximal pulmonary circuit
with closely matched CO, the gross mechanics of the two circuits must diher to ac-
commodate the needs of the tissues supported by each circuit. Hence, regulation
is essential to ensure the blood pressure does not damage tissue, small vessels and
capillaries.
7


Ohms Law for electrical circuits may be applied to fluid flow for vascular circuits
to illustrate the concept of similar rates of blood flow through dissimilar circuits.
From Ohms Law we know that voltage potential is equal to the current and the
resistance to the flow of current:
Voltagepotentiai = Current x Resistance
V = I x R (2.2)
Converting Ohms Law for the flow of current in an electrical circuit into the flow of
fluid, and specifically to the flow of blood in an arterial circuit, results in an equation
relating pressure gradient to blood flow and resistance:
Pressureoutiet Pressure^st = BloodFlow x Resistance
A P = QxR (2.3)
This equation is valid given the following assumptions regarding the fluid and
the artery segment: the fluid must be Newtonian, flow must be laminar and steady,
there must be no-slip at the wall boundaries, the artery must be cylindrical and rigid.
As Milnor points out three of the assumptions are not valid: the flow is pulsitile,
not steady, and the arteries taper and may be elliptical and are distensible. Thus,
the equation is limited in its application and tends to overestimate blood flow [73].
However, the relationships are qualitatively correct and the equation illustrates how
flow may be matched in very different circuits, by commensurate changes in AP
and R. Consequently, to overcome a large total peripheral resistance like that of
the systemic circulation, there needs to be a large change in the pressure gradient.
Correspondingly, to match flow and overcome a smaller total peripheral resistance like
that of the pulmonary circulation, there needs to be a smaller change in the pressure
gradient.
8


The two pumps of the heart are adapted for these varied requirements. The left
ventricle of the heart produces high pressure to overcome the higher total resistance
in the systemic circulation. In the pulmonary circuit, the total resistance is much
lower so the right ventricle of the heart produces lower pressure. The proximal ar-
teries convert the higher pressure and highly pulsitile blood flow, that exists near the
heart, to lower pressures and a more steady flow by virtue of their wall structure
and differential composition along the tree. At the other end of the arterial tree, the
arterioles facilitate the regulation of blood flow to the billions of capillaries through
adjustments in lumen diameter altering branch resistance and thus flow.
2.2.1.1 Historical Perspective: Poiseuilles Law
In the 1830s and 1840s Jean Leonard Marie Poiseuille, a French physiologist
and physician, derived and published the relationship of the flow of distilled water
through narrow glass tubes and the influence changing individual variables. Poiseuille
iterated through his experiments altering temperature, tube diameters, bulb pressure,
and tube lengths making successive measurements with high precision [107]. He
experimentally arrived at the equation:
KPD4
Q L
where he noted that the constant, K, is a function of the temperature and the type
of liquid. He went on to report the results of fluid flow for different types of liquid
through the glass tubes such as aqueous salt solutions, aqueous acids and bases,
mineral water, teas, wines, spirits, and bovine serum. The rate of flow for the various
fluids was compared to distilled water indicating an interest in the microcirculation
of the body [107]. Poiseuille also recognized entrance and exit effects of the flow in
the tubes but did not mention viscosity directly.
The modern form of the equation, Poiseuilles law, replaces his constant, K, with
for diameter or ^ for radius where fi is the viscosity of the fluid. Interestingly,
9


the viscosity for distilled water derived from the constant K is accurate to 0.1%
[107].
Q
vrr4AP
8/iL
(2.4)
Poiseuilles law is one of few equations derived from applied mechanics that is
well known in the present medical community [107].
2.2.2 Vascular Resistance
The forces inhibiting the flow of blood through an artery segment is a measure
of resistance. Vascular resistance is impacted by three factors: blood viscosity, vessel
length and vessel radius. Since vessel length is generally consistent in adults as is
viscosity absent injury or disease, vessel radius necessarily has the greatest impact on
the level of resistance.
To estimate how changes in artery radius impact resistance, Poiseuilles Law for
flow in a cylindrical tube (eqn. 2.4) may be combined with the Ohms Law relationship
for flow in a circuit (eqn. 2.3). The resistance may be represented as
R
8 rjL
7rm
(2.5)
where rj is the fluid viscosity, L is the tube length and r is the radius of the tube. This
calculation requires certain assumptions being met including laminar flow, Newtonian
fluid, fully developed flow profiles, and tube geometry. While blood flow does not
meet all those requirements, this equation forms the starting point for evaluating
total peripheral resistance and is an adequate approximation [73].
Given an artery segment of constant length and blood at constant viscosity, the
relationship of vascular resistance to vessel radius becomes:


The inverse radius term raised to the 4th power emphasizes how small vessel geometry
changes may result in large changes in resistance thus permitting very precise local
control of flow through particular arterial branches.
Fundamentals of Anatomy and Physiology 4th ed. Martini
Figure 2.2: Blood Pressures through the Cardiovascular Circulation
2.2.3 Blood Pressure
Blood pressure is a measure of the circumferential force that circulating blood
exerts on the vessel walls. Blood flow occurs when a force causes a difference in
pressure over a vessel segment from inlet to outlet or a pressure gradient. Blood
flows across a pressure gradient from an area of higher pressure to lower pressure.
The pressure gradient is directly proportional to the rate of flow. Hence, the greater
the pressure gradient, the greater the rate of flow given the same vessel segment or
vessel geometry as seen in AP = QR. Circulation can only occur if the flow has
the necessary pressure gradient to overcome the total peripheral resistance within the
system.
A key element underlying blood pressure is the cardiac cycle which consists of
two distinct phases, systole and diastole. The cardiac cycle starts with the systolic
phase where ventricle contraction forces blood into the large proximal arteries (aorta
11


or pulmonary artery). The highest blood pressures are found in the aorta. During the
systolic phase, the proximal arteries must expand to accommodate the stroke volume
from the heart. This is followed by the diastolic phase where the ventricles of the
heart relax and refill. During this time, the proximal arteries contract and continue
to propel the blood through the vascular tree. When blood pressure and vascular
resistance are coupled with the elasticity of the proximal arteries, the blood flow is
converted from highly pulsatile to a more steady flow at the capillaries.
The systolic phase comprises 1/3 of the cycle time while the diastolic phase com-
prises 2/3. The process may be described through the concept of a two-stage pump,
where the ventricle contraction, systolic phase, is the first stage and the arterial re-
coil during the diastolic phase is the second stage. Mean pulmonary artery pressure
(MPAP) is the average pressure as measured through the cardiac cycle. Due to the
cardiac cycle timing, the resulting mean pulmonary artery pressure is less than the
arithmetic mean.
Figure 2.3: Windkessel Represents Arterial Compliance and Elastic Recoil
12


2.2.3.1 Historical Perspective: Windkessel
Stephen Hales was a clergyman and a Fellow of the Royal Society working in
England in the early 1700s. Hales wrote Vegetable Staticks in 1727 and Haemastat-
icks in 1733 describing plant and animal physiology respectively. Hales is attributed
with making the first measurement of the force of the blood or blood pressure. In
Haemastaticks he describes the blood force found when opening an artery of a horse,
inserted a brass pipe which he connected to a tall glass tube. He documented that
the blood in the tube rose 8 feet 3 inches above the horses heart [65]. Hales also
determined the main peripheral resistance was in the small arteries by making cuts
in arteries and veins and comparing the pulsation and how quickly blood flowed from
each [68].
Stephen Hales described the smoothing action of the great arteries on the pul-
satile blood flow from the heart as an elastic reservoir. When Haemastaticks was
translated, this was converted as Windkessel. Windkessel is a German term that
may be translated as air chamber, but with respect to the arterial system it refers
to the elastic reservoir of the proximal arteries. In the 1700s the German fire engines
were pulled by horses and the water was pumped by hand. Much like the pumping of
the heart, the hand water pump created a pulsatile flow. To diminish this pulsatility,
a Windkessel was attached inline with the pump and the hose outlet as seen in Figure
2.3.
The air chamber or Windkessel received the pumped water which would increase
the pressure inside the air chamber. The pressurized water could then be dispensed
as a more steady stream with less pulsatility [89]. Thus the reason that the damping
of the pulse pressure due to the elastic reservoir and recoil of the proximal arteries
are often referred to as the Windkessel effect throughout medical and physiology
literature.
13


Working in Germany in the early 1900s, Otto Frank went on to develop a math-
ematical model consisting of two rheological elements, a capacitor and resistor in
parallel. The capacitor represents the capacitance of the arteries and the resistor
represents the total peripheral resistance. In reference to Hales work, Frank coined
the term Windkessel effect while describing his mathematical model. This two-
element model helps illustrate the importance of the concept that the total load on
the heart includes both peripheral resistance and arterial compliance [119]. Models
with 3, 4 or more rheological elements examined in current literature are based on
Franks two-element Windkessel model.
Capillary
Copyright 2001 Benjamin Cummings, an imprint of Addison Wesley Longman. Inc.
Figure 2.4: Vessel Wall Anatomy: Arteries and Veins
2.3 Artery Wall Anatomy
Arteries are complex structures comprised of three distinct layers: tunica intima,
tunica media, and tunica externa or adventitia. Each layer has a specialized function
and thus a specialized composition of cellular and extracellular elements. The main
14


cellular elements include vascular smooth muscle (VSM) cells and endothelial cells.
The extracellular matrix (ECM) is fibrous tissue primarily composed of collagen and
elastin fibers.
2.3.1 Intima
The inner most layer of the artery is the tunica intima (intima), Latin for inner
coat. The intima is in direct contact with blood flowing through the vessel lumen.
The intima is composed of a single layer of endothelial cells. Surrounding and con-
necting the endothelial cells is the subendothelial tissue, a thin layer of connective
tissue. The subendothelial tissue is also in contact with the internal elastic lamina
which is a layer of elastic tissue thought to separate the intima from the media.
2.3.2 Media
The tunica media (media), Latin for middle coat, lies in the middle between
the intima and adventitia, or more specifically between the internal elastic lamina on
the intimal side and the external elastic lamina on the adventitial side of the vessle
walls. The media is a smooth muscle layer with interspersed elastin and collagen
fibers. The VSM is responsible for contraction and dilation of the artery, vasodilation
and vasoconstriction respectively. This muscular layer is controlled by both systemic
and local factors, including the central nervous system, hormonal control and the
endothelial regulation.
2.3.3 Adventitia
The tunica externa or tunica adventitia (adventitia), outer coat, is the outer-
most layer of the vessel. The adventitia provides the vessel with strength, structure,
and dispensability. The adventitia anchors the vessel and provides signaling pathways
15


Plaque
Intermediate filaments
of cytoskeleton
Actin filaments
Dense body
Myosin
Contracted smooth
muscle cells
Relaxed smooth
muscle cells
McGraw-Hill
Figure 2.5: Smooth Muscle Cells
to and from the surrounding tissues and may help prevent over distention [105]. The
adventitia is primarily collagen. Other tissues within the adventitia include extracel-
lular matrix, fibroblasts, interstitial cells, vaso vasorum, nerve cells and elastin fibers
[105].
2.3.4 Artery Wall Composition
2.3.4.1 Endothelium
In the intima, endothelium is a single layer of cells lining every blood vessel provid-
ing a barrier between the blood and vessel. Endothelial cells respond to environmental
changes such as chemical stimuli in the form of circulating hormones or other factors
in the blood and mechanical stimuli such as shear stress. In response, endothelial
cells produce vasoactive molecules, cell growth factors, adhesion and blood-clotting
16


molecules. In addition, endothelium promotes elastin and fibroblast synthesis, and
immune response. The vasoactive molecules direct the VSM to contract or relax,
thus providing local regulation of blood flow through particular artery segments via
vasoconstriction and vasodilation. Vasoactive molecules such as nitric oxide (NO)
and Endothelium-Derived Relaxation Factor (EDRF) are produced by the endothe-
lium and act on the smooth muscle through signaling pathways. While this is not a
complete list of all endothelium activities, it reinforces the key roles this single layer
of cells play in numerous artery functions, far beyond a mere barrier between the
blood and the artery as was once thought.
2.3.4.2 Smooth Muscle Cells
Smooth muscle cells (SMC) are long spindle-shaped cells approximately 20 nm
by 200 nm with a single nucleus. SMCs are arranged in helical sheets lining most
blood vessels and internal organs. They are regulated by the autonomic nervous and
endocrine systems (as well as various chemicals) and are activated by stretch [1],
SMC are so named because they are not striated like skeletal and cardiac muscle and,
therefore, have a smooth appearance. The lack of striation is due to the myofibril
arrangement. Instead of being arranged in regular parallel patterns the myofibrils
of myosin and actin crisscross the cell, connecting to dense bodies on the cell walls
causing the cell to bunch upon contraction.
Smooth muscle contraction is initiated by an increase of intracellular Ca+2. In-
tracellular Ca+2 binds with calmodulin, a calcium-binding protein present in all cells.
The Ca-calmodulin complex is able to attach to and activate myosin light chain kinase
(MLCK). The Ca-calmodulin-MLCK complex phosphorylates the myosin regulatory
light chain of the myosin filament. Phosphorylation activates the myosin head allow-
ing the head to attach to the actin filament. The phosphorylated myosin heads attach
and detach from the actin filament forming cycling cross bridges which generate force
and cause contraction. If the myosin light chain is dephosphorylated by myosin light
17


chain phosphotase (MLCP) while attached to an actin filament, the detachment from
the actin filament happens much more slowly than detachment from a phosporylated
crossbridge. This slow detachment of the unphosphorylated crossbridge is termed the
latch state. Latch is a hallmark of smooth muscle contraction and uniquely distin-
guishes it from skeletal or cardiac muscle [60]. The formation of latch bridges allows
the arteries to maintain force at a lower energy cost than striated muscle [60].
67 nm
Fibril
i--------------------------1
Alpha chain
Amino acid
sequence
Gly X Y Gly X Y Gly X Y -
Fratzl & Weinkamer j.pmatsci, 52(8): 1263-1334, 2007.
Figure 2.6: Collagen Triple Helix Structure and Aggregation
2.3.4.3 Collagen
Collagen is the most abundant protein in the human body and is the most com-
mon fibrous protein in the extracellular matrix. At 25-35% of all protein in the body,
18


collagen provides structure, strength and is able to resist tensile forces throughout the
tissues. Although mature collagens are insoluble, the collagen matrix is not static.
With a half-life of 0.5-3 months, collagen is able to be remodeled to meet the needs
of changing tissue environments [58].
Procollagens, collagen precursors, are synthesized intracellularly, primarily by
fibroblasts and myofibroblasts [29]. Procollagen molecules are secreted extracellularly.
The ends of the procollagen molecules are cleaved and assembled into fibrils composed
of many individual procollagen molecules with a regular end to end and side to side
pattern. The molecules within the fibrils are connected by and stabilized with covalent
crosslinks [57]. It should be noted that collagen contains only 1-4 crosslinks per unit
whereas elastin contains 15-20 [115].
The individual procollagen chains or polypeptides are modified to facilitate de-
velopment of the chains into the characteristic triple helix molecules. The triple
helix is composed of 3 polypeptide chains which have repeating amino acid segments.
The three chains typically have a glysine with two other amino-acid chains. These
polypeptide chains are therefore commonly represented as (Gly-X-Y). The X and Y
represent any amino acid including Glysine, and may be the same or different, adding
considerable variability to the superfamily [1],
When arteries are slack, the collagen fibers appear wavy. As intraluminal pres-
sure increases, expanding the artery diameter, collagen straightens becoming aligned
circumferentially causing the fibers to bear some of the load [94], At physiologic
pressures, less than 10% of the collagen fibers are stretched and engaged [38]. At
higher pressures, more collagen fibers are engaged and the vessel walls become less
distensible [22]. The mechanical properties of collagen consistently report high elastic
moduli on the order of 100-1000 MPa, suggesting very stiff material [22],
Collagen is the most abundant and main structural protein within the connective
tissue. This directly affects the mechanical function with regard to compliance and
19


modulus of the artery wall and vessel geometry including diameter and thickness of
the artery [59].
Although 28 collagen proteins are found in the human body, 90% of these are
type I [58]. Type I collagen is hbrallar (able to form collagen fibers) and found in the
skin, tendon, bones, lungs, vasculature and cornea. Collagen Types III and V are also
hbral-forming. Type V is closely associated with Type I, but they differ from Type
I in their triple helical chain structures which may allow for functionality related to
wound healing [29]. Type IV is a basement membrane collagen adding support and
may have functions to help filter fluids.
Collagen makes up approximately %15 of the dry weight of the human lung and
is the major protein group. Collagens Type I and III are the most abundant in the
lung at a ratio of approximately 2:1 and are found co-localized in the vessel walls
[62] Collagen Type I is the primary collagen found in the adventitia [29]. The media
contains collagen at a ratio of approximately 30% Type I and 70% Type III [45]. The
non-fibrallar collagen found in blood vessels is Type IV found in the subendothelial
tissue of the intima.
elastic fiber
STRETCH
RELAX
single elastin molecule
Molecular Biology of th Coll S/o
Figure 2.7: Elastin
20


2.3.4.4 Elastin
Elastin is a protein that forms elastic fibers found in the connective tissue of the
lungs, skin, bladder, elastic ligaments, elastic cartilage and blood vessels in the body.
Elastic fibers are made of a network of approximately 90% elastin and 10% fibrillin.
Elastin and fibrillin are produced by fibroblasts. Elastin is produced primarily in
the perinatal phase of development, synthesized by embryonic or juvenile fibroblasts.
After this stage, production of elastin dramatically decreases. The half life of elastin
is approximately 40 years or the life span of the organism making elastin very stable
with less than 1% turn-over per year in adult humans.
Elastins extended life span and insolubility are directly related to its extensive
(15-20) crosslinks per unit [115]. Within this highly crosslinked structure, relaxed
elastin fibers form compact and contracted arrangements, appearing somewhat over-
lapped, coiled and disordered. This relaxed formation is a low energy state. Elastin
fibers may stretch to 1.5 times their contracted length. When stretched, the fibers
elongate and form sheets of orderly molecules connected through a large network of
interspersed cross links. Stretched elastin is in a high energy state, and is able to
passively recoil back to the contracted low-energy state.
Elastin has a low elastic modulus of approximately 0.4 MPa. meaning that it
will resist very little of the stress and strain placed on the arterial wall. The quantity
of elastin in the arteries is greatest proximally to the heart and decreases along the
arterial tree to the arterioles where it is much less abundant. These diminishing levels
of elastin along the arterial tree underscores elastins primary role of providing the
artery with dispensability and elastic recoil to assist in reducing pulsatility as blood
flows toward the capillaries [22],
While the role of elastin is limited, it is essential. Specifically, elastin also partici-
pates in the regulation of smooth muscle cell production. When elastin is not present,
smooth muscle cells continue production until the artery is completed blocked. [115].
21


2.4 Regulation of Blood Vessels
Cardiovascular regulation ensures blood perfusion and flow to meet the needs of
proximal and distal tissues. Blood flow must be optimized for gas exchange, delivery
of nutrients, removal of waste and pH balance to support approximately 10 billion
capillaries in the adult human body [60]. Therefore, regulation of the cardiovascular
circulation must be exquisitely controlled in time and space. Changes in blood flow
must occur in time scales varying from fractions of seconds to days or years. In
addition, these changes to the flow of blood must be made both locally within specific
tissues as well as systemically.
The regulation of blood flow may be thought of as a feedback loop of controllers,
sensors and actors working in concert to keep blood pressure at a homeostatic set
point, which in human adults is a mean arterial pressure of 90 mmHg [42], Regulation
of the cardiovascular system involves changes in the CO, pressure and peripheral
resistance. Three mechanisms allow those changes to be made: autoregulation, neural
regulation and hormonal regulation.
These three mechanisms act on both a short and long term basis. Short term
regulation includes: 1. Intrinsic (Baroreceptor Reflex, Chemo Receptor Reflex, Car-
diopulmonary Reflex, Heart & Vessel Auto Regulation); 2. Extrinsic Reflexes (Pain,
Exercise, Anger/Fears/Anxiety, and cooling/thermal); and 3. Hormones. Long Term
Regulation includes: 1. Changes in blood volume; and 2. Changes in blood vessel
quantity and size [42] [67].
2.5 Hypoxic Vasoconstriction
Hypoxic vasoconstriction is one of the principle regulators of blood flow respond-
ing acutely and chronically (short and long term regulation). Hypoxia is a condition
of low oxygen levels in the body or in specific areas of the body. Hypoxia is the
most powerful stimulant of the peripheral chemoreceptors [42], Hypoxic pulmonary
22


vasoconstriction (HPV) is a physiologic process where arteries in the lung contract in
response to hypoxia in the alveoli.
As yet, there is no consensus on the precise oxygen sensor or sensors which trigger
HPV, though a number of signaling systems have been suggested including: heme
proteins, ion channel conductance, and a mitochondria initiated decrease in reactive
02 species [19]. Current studies suggest one of the key mechanisms lies within the
pulmonary vascular smooth muscle [75][108]. With low p02 sensed, potassium ion
channels become blocked or inhibited, causing the cell membrane to depolarize [117].
Upon membrane depolarization, the voltage-gated calcium channels are activated
allowing an influx of Ca++. At the same time calcium is also released intracellularly
from the sarcoplasmic reticulum.
The total increase in the concentration of intracellular Ca++ initiates contrac-
tion of the smooth muscle and vasoconstriction of the artery. Specifically, the small
resistance arteries and arterioles contract, decreasing diameter, increasing resistance
to blood flow, thus decreasing blood flow through the poorly oxygenated capillaries.
The increased resistance in these branches encourages blood flow through branches
with less resistance, thus diverting blood flow toward capillaries with better p02 per-
fusion. This process is also known as the Euler-Liljestrand mechanism named after
Ulf vonEuler and Goran Liljestrand who first described the process [27]. Once this
occurs, a state of localized hypertension exists.
2.6 Pulmonary Hypertension
Pulmonary Hypertension is defined as a mean pulmonary arterial pressure above
25 mmHg at rest [44], RHC is conventionally used to measure pressures in the PA
and to diagnose PH [35]. In some sense, PH can be thought of as similar to general-
ized hypoxic vasoconstriction affecting the whole pulmonary bed which dramatically
increases pulmonary vascular resistance (PVR). Increased PVR makes the right ven-
23


tricle of the heart work harder to force the blood through the narrowed vascular
tree. Thus, PVR is often the single parameter used to assess disease severity and
response to treatment. Nevertheless, other markers of vascular function, for example
pulmonary vascular stiffness (PVS), have been shown as predictive of mortality in
PH [51] [36] [74],
The World Health Organization clinically classifies PH into five categories of dis-
orders by cause that share similar characteristics, pathogenesis or therapeutic man-
agement [98]. The groups include: Pulmonary arterial hypertension (Group 1), Pul-
monary hypertension due to left heart disease(Group 2), Pulmonary hypertension
due to lung diseases and/or hypoxia (Group 3) Chronic thromboembolic pulmonary
hypertension (Group 4), and Pulmonary hypertension with unclear multifactorial
mechanisms (Group 5) [98].
The definition and diagnosis of PH continue to evolve. For example, it has been
suggested that PH be redefined to include a resting mPAP of 20 instead of 25 mmHg.
Another option would be to characterize mPAP from 21 to 24 mmHg as borderline
PH. Previously, the definition included exercise induced PH which could be reintro-
duced as part of the diagnostic tools. Given the common use of PVR, it is a possible
factor to be included in the definition as well. There is also the option to include
pulmonary artery wedge pressure (PAWP) of 15 mmHg in the standard protocol to
distinguish between pre-capillary and post-capillary PH [44],
Despite the rapid advances in PH therapies, the progressive nature of the disease
eventually results in right heart failure and death. Consequently, medical therapies
focus on relieving symptoms and helping the heart pump, providing oxygen and re-
ducing edema. Medications that modulate vascular tone are one of the main pharma-
cological agents with the goal of improved hemodynamic profiles [3] [6]. Vasodilators
slow the progression of the disease and improve the quality of life for patients with
PH [42], These medications act to relax VSM contraction of the blood vessels by
24


regulating various intracellular pathways: reducing intracellular calcium, decreasing
the phosphorylation of myosin light chain, or altering intercellular potassium or ATP
levels. The common theme among these medications is the regulation of VSM for the
normalization of the cardiovascular circulation to prevent right heart dysfunction [6]
[95].
2.7 Large Arteries in Pulmonary Hypertension
Chronic pulmonary hypertension (PH) increases pulmonary artery stiffness and
decreases dispensability throughout the arterial tree and inducing changes and re-
modeling within the vessel wall. Increases in vascular collagen deposition as well as
increased wall thickness are well established and thought to be the primary cause of
vessel stiffening. However the process of PH causes many significant changes within
the vasculature.
2.7.1 Extracellular Response
Extracellular Matrix
During the development of PH, extracellular matrix deposition increases. In
proximal arteries, SMC and fibroblasts increase collagen and elastin production in
the media and adventitia. In the distal arteries, the endothelial cells increase the
production of Type IV Collagen as well as elastin [105].
2.7.2 Cellular Response
Endothelial Cell
Within the intima, studies show increases in endothelial cells as an acute response
in the PPA and as a chronic response in the distal arteries. The increase in the number
of endothelia cells is not likely to increase the stiffness of the artery walls. Whether
25


this change is dysfunctional or adaptive, is still under debate. In either case, the
quantity of endothelia cells does not appear to directly change the mechanics but
may indirectly impact artery mechanics through increased signaling to other tissues
and cells within the arteries.
PH induces a number of changes in endothelium function. Some of these changes
include: the increase in inflammatory response, the production of adhesion molecules,
the expression of coagulation factors and increased coagulation, and an increase in
endothelial cell permeability. The decreased endothelial barrier allows factors in the
blood to be in direct contact with the subendothelial layer, possibly leading to in-
creases in cell proliferation of the media and adeventitia [105]. The endothelium also
produces more vasoconstricting factors and reduces the production of vasodiliating
factors.
Smooth Muscle Cells
In PH, SMCs increase in size (hypertrophy) and number (hyperplasia) while in-
creasing production of collagen. Hypertrophy is more prevalent in the proximal vessels
while hyperplasia is more prevalent in the distal vessels. In the microvasculature, fi-
broblasts are recruited from the avelolar walls to synthesize SMCs in areas of the
microvasculature where they do not normally exist. In this context, it is important to
remember that there are different sub-populations of SMCs that respond differently
to environmental stimuli [103]. Thus it is possible that not all SMCs will respond
equally to PH.
Fibroblasts
Much like the SMC, there are different groups of fibroblasts which respond to en-
vironmental stimuli differently. Throughout the arterial tree, PH induces an increase
in some fibroblast populations to initiate increased production of collagen. In addi-
tion one subgroup in the adventitia changes phenotypically to myofibroblasts which
produce cu-SM actin during PH. Of interest, myofibroblast activity is also known to
26


be present in response to injury, wound healing and scar contraction. In the distal
arteries, the increased fibroblasts lead to an increase of collagen Type I and elastin,
diminishing the artery inner diameter.
Inflammatory Cells
Inflammatory cells produce a complex web of factors, including growth factors,
cytokines and reactive oxygen species. These factors interact in a vast signaling
matrix to regulate cell growth, immune response and aptosis. Inflammatory cells
are correlated with the remodeling that is characteristic of PH. This is supported by
the increased numbers of lymphocytes and macrophages present in the arterial wall.
Inflammatory cells are also suggested as a potential impact on endothelial dysfunction
and increased sympathetic activity.
Adapted from Tian and Chester (2012)
Figure 2.8: In-Vitro Testing-System schematic diagrams
2.8 Evaluation of Arterial Mechanics
As we know, the most interesting and the most difficult mechanical properties
to measure are the active part of the vascular smooth muscles [48]. Given all of the
above factors and their interwoven impacts on arterial characteristics and function,
27


predicting potential changes in mechanics over time is much more complex than it
may initially appear.
2.8.1 In-Vivo
While in-vivo tests allow examination of the living vessels. They are extremely
invasive and are further limited by physiological control from the body including
hormones, the central nervous system, and auto-regulation because, influences from
the bodys regulatory systems are often not controled or measured by the testing
system. However, many in-vivo studies provide the basis for artery diameter, stress,
strain and modulus data and comparisons [46] [32] [68].
2.8.1.1 Right Heart Catheterization
Right heart catheterization (RHC) is performed to gather hemodynamic mea-
surements within the heart and a number of measurements from within the proximal
pulmonary artery to diagnose and assess PH. A catheter is inserted through the ve-
nous system into the right side of the heart taking measurements of pressure in the
right atrium and right ventricle as it is passed into the main pulmonary artery. Pres-
sure is also measured in the PA. Measurements of pressures for the left side of the
heart are made indirectly by inflating a balloon at the end of the catheter in the
pulmonary artery. The resulting pressure is pulmonary artery occlusion pressure or
pulmonary capillary wedge pressure (PCWP) and approximates left atrial pressure
of the heart. RHC measurements are utilized as indicators of heart function and
possible ventricular failure. [4] [69].
Non-invasive PPA measurements are also available. Ultrasound, echocardiology
or echo-doppler, C02 rebreathing, bioimpedance and magnetic resonance imaging are
being considered as options to estimate geometry, flow, pressure and CO. Studies are
currently examining the accuracy of these non-invasive or less-invasive measurements
and calculations compared to RHC data. Despite the appeal of non-invasive options,
28


RHC remains the standard to diagnose, confirm and assess PH as it consistently pro-
vides the most accurate measurements of pressure from within the heart and proximal
arteries [113].
2.8.2 In-Vitro
The most commonly performed in-vitro tests are stress and strain. These are the
most common in-vitro tests performed. In order to be successful, this testing must
be completed while the tissue is submereged in a physiologic solution to ensure that
the tissue remains viable. It should be noted that each of the in-vitro tests must rely
on certain assumptions regarding constant tissue volume.
2.8.2.1 Unixaial Test
Uniaxial testing is one of the most common in-vitro methods of testing soft tissue
to measure stress and strain. Uniaxial tests are performed on rectangular strips of tis-
sue that are stretched in one direction as they are held in place with clamps or hooks.
The tissue can be stretched to approximate either circumferential or longitudinal dis-
placement. During this process, a load cell applies a force and the displacement of the
tissue is measured through the movement of the hook or clamp. The resulting data
are pairs of force and displacement. The force and displacement pairs are used along
with measurements for cross sectional areas of unloaded tissue to calculate stress and
strain:
Stress
Force
(2.6)
Area
(2.7)
The results typically form a J shaped curve illustrating the initial activity of the
elastin, the base of the J showing a transition and subsequently showing the initiation
29


of the collagen up to tissue strain limits. The slope of this curve will depict the degree
of stiffness in each region, known as the incremental modulus. While this testing
provides valuable information regarding uni-directional stress and strain, it does not
account for the 3-dimensional nature of the tissue as it interacts circumferentially,
longitudinally and radially in-vivo with intraluminal pressures.
2.8.2.2 Ring Test
Ring testing is similar to uniaxial testing except that a small segment of artery
is cut circumferentially. This ring is mounted on hooks and pulled uni-directionally.
The ring test again produces paired data of force and displacement making the strain
calculation similar to that of the uniaxial. However, the stress calculation includes a
factor of \ due to the two sides of the ring:
1 Force
Stress = x ----
2 Area
(2.8)
The ring test limitations are similar to the uniaxial test in that it also fails to account
for the 3-dimensional nature of the artery in-vivo.
2.8.2.3 Biaxial Test
The biaxial test uses a square or rectangular tissue sample which is stretched in
two directions. Thus, allowing data pairs of force and displacement to be collected
in both the circumferential and the longitudinal directions. Stress and strain may
be calculated as before making sure that the areas reflect the cross section of the
direction of stretch as illustrated in 2.8C.
2.8.2.4 Pressure Inflation Test
The pressure inflation apparatus is used for testing whole longitudinal segments of
an artery in its original shape. An artery segment is placed between the two cannulas
and submerged in a physiologic solution. At each end of the artery is a pressure
transducer connected to a cannula. The perfusate pump sends the physiological
30


Adapted from Ooi et al. (2009)
Figure 2.9: Pressure Inflation Testing System Schematic
solution through the cannula into the artery segment. A microscope captures the
diameter of the artery as the intraluminal pressure increases. Simultaneously, the
pressure at each end of the cannula is measured with the pressure transducers to
estimate the intraluminal pressure of the tissue. With each change in pressure, a new
data pair is recorded. Intraluminal pressure is calculated by the formula:
P.
-L VI
PI P2
(2.9)
The advantage of htis test is its measure of both radial and circumferential stress.
While it does not account for shear forces from blood flow or longitudinal stretch, it
much more closely mimics in-vivo environments than other in-vitro tests.
2.8.2.5 Pre-Conditioning
Pre-conditioning biological tissue is the process of cyclically loading and unloading
the tissue samples prior to taking test measurements. Some studies consider pre-
31


conditioning necessary to obtain reproducible stress-strain curves [32] [46]. During
pre-conditioning, tissues may be mounted as described in any of the above methods:
uniaxial, ring, biaxial or pressure-inflation. The tissues are then stretched or inflated
according to Youngs protocol; approximately 10 times at 1 Hz prior to testing [].
2.9 Recovery Studies
Previous studies of rat hypoxic recovery models suggest a range of results from
complete recovery to persistent changes related to PH remodeling. Poianis study of
10 days hypoxic exposure at 10% 02 and pressures between 74-80 mmHg and up to
10 days recovery in room air found collagen and elastin contents returned to normal
[87]. Meyricks results of 380 torr hypoxic exposure for 10 days showed the adventia
doubled in thickness with increased fibroblasts and collagen fibers. Upon recovery
in room air for up to 70 days, the adventitia thickness returned to normal but the
fibroblasts were smaller while the collagen fiber concentration remained increased
[72], Hislop and Reid found with exposure of 2 weeks at 380 mmHg and recovery up
to 8 weeks in room air that right ventricular hypertrophy resolved, but the loss of
small arteries (arterial pruning) with outer diameter = 200 m did not recover from
the hypoxic exposure [43]. With 10% 02 hypoxic exposure up to 3 weeks and recovery
in room air up to 20 weeks, Herget found that muscularization and arterial pruning
persisted while mPAP returned to normal [41]. Finally, with 5 weeks of exposure to
380 mmHg and up to 5 weeks of recovery, Heath found normalization of both medial
thickness and RV hypertrophy [40]. This base of information opens an opportunity
for a single study, across time and pressures, to create a single set of data to align
the findings into a comprehensive base of information.
32


2.10 Studies in this Thesis
The work in this dissertation aims to better quantify the impact pulmonary hy-
pertension has on the mechanics of the proximal pulmonary arteries. Specifically, the
aim is to assess changes in the PPA mechanics in PH onset, recovery or treatment.
To our knowledge, this is the most inclusive assessment of PPA mechanics upon
development and recovery from PH in an animal model. Further, this is the first
comprehensive statistical evaluation of available clinical measures to predict future
WHO FC.
The thesis results are supported by a human study and two animal studies. The
clinical data from the human study originates from pediatric right heart catheteriza-
tion reports used to confirm the diagnosis of pulmonary arterial hypertension as well
as associated non-invasive measures. The aim of the clinical study was to provide the
foundation for a model to predict future patient status using clinical measures. The
animal studies focused on invasive mechanical PPA measures that are traditionally
unavailable in human studies. The aim of this study was to evaluate the changes in
PPA mechanics across a full range of physiological pressures in PH onset and recov-
ery to quantify the function of the passive elements of the arterial wall within the
boundaries of smooth muscle activity. The ultimate goal of this study was to provide
insights to support potential strategies for future medical treatments.
33


Chapter 3
3. Study: Multivariate Prediction of Clinical Indicators
3.1 Abstract
Background: Pulmonary hypertension (PH) is an incurable, progressive disorder
with poor survival in children. Conventionally, cardiac catheterization is used for
diagnosis and serial monitoring in children. Significant disparity remains regarding
which clinical factors are most predictive of outcome. The goal of this study was to
evaluate multiple clinical factors, in combination, to help guide clinical interventions
and predict outcomes.
Methods: This study includes 78 patients evaluated for PH in the cardiac
catheterization laboratory at The Childrens Hospital Colorado and associated hemo-
dynamic measurements and imaging of the proximal pulmonary artery. We used
a multi-covariate proportional odds logistic regression to explore prediction of pa-
tient outcome represented by the World Health Organization Functional Classifica-
tion (WHO-FC) at follow-up. We were interested in providing prediction of WHO-FC
in 1 year and to provide objective measures of disease progression.
Results: Our findings support the combination of mean pulmonary arterial pres-
sure, systolic blood pressure, systolic proximal artery diameter and the ratio of mean
pulmonary arterial pressure to mean aortic pressure as the model combination that
predicts future WHO-FC with the highest accuracy. Systolic blood pressure was iden-
tified as the most accurate single variable predictor. However, the single measures
are significantly less accurate than the multivariate equation, therefore limiting their
use in accurately quantifying future WHO-FC.
Conclusions: Single variable predictors when combined provide incremental im-
provement that greatly surpassed their individual ability to predict the future health
34


status of patients. Specifically, some of the newer non-invasive mechanical measures
when combined with classic RHC measurements significantly increased prediction
accuracy.
3.2 Introduction
Pulmonary hypertension (PH) is a disorder involving primarily the arterial com-
ponent of the pulmonary vascular tree [99]. PH is defined as an increase in mean
pulmonary arterial pressure (MPAP) above 25 mmHg at rest [7] [56] [92], Beneath
this definition is a complex array of symptoms, multiple pathogenic pathways, mul-
tiple etiologies and associated diseases [4] [35] [70]. PH is conventionally diagnosed by
right heart catheterization (RHC); an invasive procedure that carries a number of
risks [4] [35] [69]. Despite the rapid advances in PH therapies, the progressive nature
of the disease eventually results in right heart failure and death [3] [30] [34] [69] [70].
PH is characterized by vasoconstriction, remodeling, and thickening of the small
to medium arteries and arterioles, resulting in increased PVR [36] [103]. PVR is of-
ten the single parameter used to assess disease severity and response to treatment
[7] [24] [121]. Yet, other markers of vascular function, for example pulmonary vascular
stiffness (PVS) area strain and vascular capacitance have been shown as predictive of
mortality in PH [15] [36] [51] [74], Numerous prior studies have suggested utilization of
various combinations of individual clinical indicators as predictive of survival, mor-
tality and outcomes for PH [10] [11] [12] [23] [109]. However, there is no comprehensive
evaluation showing incremental increases in predictive value when a vast array of
individual indicators is combined.
Given the broad base of available clinical indicators and associated data, there
is a ready opportunity to evaluate the complimentary predictive value of pediatric
indicators in PH. This is especially beneficial because pediatric PH lacks its own
comprehensive evaluative factors, relying heavily on adult data for reference. Due
35


to the important differences in pediatric PH etiology, symptoms, and presentation,
pediatric disease is unique and requires customized predictive clinical indicators [8]
[9] [23] [120] [121]. With this aim in mind, we used 4 statistical tools to explore numer-
ous combinations of hemodynamic measurements as well as the geometry, flow, and
pressure changes in the proximal pulmonary artery to identify a predictive tool for
pediatric PH outcomes. For the purposes of this study, outcome is defined as WHO
Functional Classification (WHO-FC) at the time of post RHC follow-up. We hy-
pothesize that measures of proximal pulmonary artery function, including right heart
catheterization (RHC) and Color M-Mode Tissue Doppler Images (CMM-TDI), in-
crease the accuracy of patient outcome predictions when utilizing the combinations
of measures from confirmed cases of pediatric PH.
3.3 Methods
3.3.1 Study Design
The study was carried out with the approval of the Colorado Multiple Insti-
tutional Review Board (COMIRB). All patients or their guardians consented after
being informed of the risks and benefits of the procedures and prospectively granted
permission for the utilization of gathered data for study.
3.3.2 Patient Population
We reviewed the medical records for any patients who underwent RHC and acute
vascular reactivity testing at Childrens Hospital Colorado from August 2003 to July
2010. Consecutive patient data was included for those diagnosed with PH before the
age of 18 years old by right heart catheterization (RHC). WHO-FC score was obtained
from medical records. Patients were reevaluated and assigned a follow up WHO-FC
scores within 2 years of initial RHC (4-26 months). Hemodynamic and demographic
36


baseline data collected at the time of initial RHC are shown in Table 3.1. For inclusion
in the study patient records needed to include: PA pressures, tissue doppler images
suitable for PA diameter analysis, WHO-FC at diagnosis and follow-up.
All patients were treated with standard therapies.
3.3.3 Clinical Data Collection
All patient data included a RHC performed at 21% FiC>2 and under anesthesia
as is routine for the institution. Pressure measurements of mean pulmonary arte-
rial pressure (MPAP), mean right atrial pressure (mRAP), and pulmonary capillary
PCWP pressure (PCWP) were continuously recorded using a Swan-Ganz catheter
(Transpac IV, Abbott Critical Care Systems, Abbott Park, IL). A systemic arterial
monitoring line recorded the mean systemic blood pressure. Cardiac output (CO)
was calculated via the Fick equation. Indexed PVR (PVRI) was calculated as the
difference between MPAP and PCWP, divided by the cardiac index. The ratio of
pulmonary vascular resistance to systemic resistance |y-§, as well a calculation of
stroke volume to pulse pressure were used as indicators of compliance.
Color M-Mode Tissue Doppler Image (CMM-TDI) of the right pulmonary artery
(RPA) wall were also collected at the end of the RHC. Imaging of the RPA wall were
obtained from the suprasternal short-axis view, providing a long axis representation
of the RPA perpendicular to the ultrasound beam angle [25] [50]. Prior to measure-
ment, the ultrasound beam was swept through the long axis of the RPA to locate
maximal diameter. Acquisition of the diameter measurements occurred along the
beam line as shown in Figure 3.1a. A custom software package based on EchoMAT
(v.2.1, GE Medical Systems Inc.) detected the lumen wall edges, shown on Figure
3.1b, and continuously recorded the pressure and echocardiograph traces shown in
Figure 3.1c. Based upon the ultrasound results, the systolic diameter (sDiam) of the
RPA, the difference between the systolic diameter and the diastolic diameter of the
37


right pulmonary artery diameter (AD), and the maximum diameter indexed by BSA
were calculated. Pulse pressure (PP) was calculated as the difference between
systemic and diastolic pressure. Strain was calculated as the difference between the
systolic and diastolic RPA diameters and normalized by the diastolic RPA diameter
(sOtum-dDtcim) ^ j)ynamjc compliance (Cdyn) was calculated as the normalized volume
change given an applied pressure; that is stJ^n.
3.3.4 Statistical Analysis
Statistical analysis was performed using R Core Development (R version 2.14.1,
2011) [90]. Where appropriate, statistical significance was determined as p<0.05.
Appendix 2 comprises the code for all statistical analyses, calculations and derived
measurements.
Akaike Information Criterion (AIC) was used to ascertain the accuracy of vari-
ables in predicting future patient WHO-FC [112]. AIC is similar to a likelihood
ratio test in that it provides a measure of the goodness of fit of different calculation
methods using identical data. AIC includes a penalty term to verify a suitable fit
(Appendix 2 for code implementation) [16]. Given the small cohort of this study and
the relatively large amount of data collected for each patient, the Random Forest
method was also used to verify prediction accuracy of individual variables (Appendix
2: code implementation) [106].
Proportional odds logistic regression was as used with single and combined clinical
indicators to predict patient WHO-FC scores upon follow up. We used 3 categorical
levels for patient outcome: FC-I, FC-II, and a combined classification of FC III-IV.
FC-III and FC-IV were combined as there was only 1 patient assessed at FC-IV
and 10 at FC-III. To evaluate the predictive accuracy of each set of indicators (both
individually and in aggregate), we used a leave-one-out training-testing.
38


We performed a Cox proportional hazards survival analysis of the time until a
PH-related event over the course of this study, 40 months. In addition to WHO-FC,
we also defined five patient events: 1. Hospitalization due to PH; 2. Worsening
of WHO Function Class (with the exception of changing from class I to class II);
3. Start of IV therapy; 4. Lung transplant; 5. Death. The time to event was
assessed with both univariate and multivariate analyses [106]. We reduced the number
of variables under consideration using the Random Survival Forest methods. The
Random Survival Forest methods are patterned after Random Forest modeling and
have similar accuracy prediction for complex data with non-linear effects and high
ordered interactions [55]. These additional patient events were evaluated in order
to determine whether variables identified for WHO-FC could be utilized to predict
alternative outcomes.
3.4 Results
A total of 78 patients were included in this study from an initial base of more
than 200 patients. The mean age was 8.986.08 years of age at the time of diagnosis.
Fifty-one percent of the patients were female. The mean time from diagnosis to follow-
up was 11 5.4 months. The diagnoses noted for the patients were: 19 idiopathic
PH, 5 Familial PH, 3 associated with congenital heart failure, 17 associaated with
Atrial Septal Defect, 13 associated with Atrioventricular septal defect, 3 associated
with high altitude pulmonary edema, 10 associated with patent ductus arteriosus, 2
associated with lung disease, 1 associated with pulmonary capillaritis, 2 associated
with antiphospholipid antibody syndrome and 1 associated with Overlap syndrome.
There were 3 deaths and no lung transplantations during the time of the study. These
data are summarized in Table 3.1.
39


3.4.1 Single Variable Measures as Predictors of Outcome
The data contained a number of patient measurements that were strongly
aligned or correlated. There were positive correlations between the following mea-
sures: MPAP and PVRI (r=0.09, p<0.05); PVR/SVR and PVRI (r=0.11, p<0.05);
PVR/SVR and MPAP (r=0.12, p<0.05); MPAP and PP (r=0.16, p<0.05); AD and
sDiam (r=0.26, p<0.05); PP and PVRI (r=0.27, p<0.05); and PVR/SVR and PP
(r=0.29, p<0.05). Typically, we remove measurements with strong correlations from
further statistical analysis. However, this study calls for the examination of a broad
base of available clinical indicators and associated data. Therefore, we kept highly
correlated variables in the models to ensure a comprehensive evaluation of potential
complimentary predictive values when these measures were grouped with other pe-
diatric indicators. Likewise, we kept values of strain and compliance despite their
relation to diameter and pressure.
The univariate analysis of AIC weights shown in Figure 3.2a suggests that SBP,
MPAP/AoP, MPAP, and PVR/SVR are the best prognostic indicators for patient
outcome prediction. The Random Forest estimation of variable importance suggests
that the variables SBP, MPAP/AoP, MPAP, PVR/SVR, PP, Cdyn, and sDiam re-
spectively may individually describe the majority of the variance of patient outcome.
In Figure 3.2b all indicators are arranged in variable importance from most important
to least.
Finally, we used ordered logistic regression with each clinical indicator as a uni-
variate model to predict the patient outcomes. Again the best predictor was SBP (53
predictions or 68% MSE 0.397) followed by: PVR/SVR, PP, MPAP/AoP, MPAP,
PVRI, Cdyn, sDiam, SV/PP, sDiam/BSA, pulmonary capillary wedge pressure, CO,
AD, Strain and mRAP shown in Figure 3.2c.
40


3.4.2 Combining Multiple Measures as Predictors of Outcome
Again we used ordered logistic regression now with every combination of the clin-
ical indicators. All combinations were generated and used as the explanatory vari-
able in the regression to determine the most accurate prediction of patient outcome.
When sDiam, MPAP, MPAP/AoP, SBP were used in combination, these variables
accurately predicted 59 of the 78 patients (76% MSE .32) as illustrated in Table 3.1.
The equation for the ordered logistical regression fit of the measures to functional
class outcome is:
M PAP
(0.68 x sDiam) (0.18 x MPAP) + (5.56 x ) + (0.15 x SBP) (3.1)
A second equation with a different set of 4 variables: SBD, PVR/SVR, MPAP,
Cdyn predicted 57 patient WHO-FCs correctly (73% MSE 0.35) using the equation:
(0.13 x SBP) (0.11 x MPAP) + (2.22 x ) (37.37 x Cdyn) (3.2)
SV R
A third equation of four variables: MPAP, PVRSVR, SBP, Strain, was accurate
in predicting 56 of the 78 patients (72% MSE .35):
(0.16 x SBP) (0.14 x MPAP) + (2.36 x ) (1.63 x Strain) (3.3)
SV K
As a comparison, PVRI alone was accurate for 46 of the 78 patients (60% MSE
.53).
0.23 x PVRI (3.4)
When the variables were limited to the best single indicators, SBP, the prediction
accuracy fell to 53 of the 78 patients (68% MSE .40). The algorithm for this analysis
and the resulting equations is illustrated in Appendix 2.
41


0.08 x SBP
(3.5)
The next best single indicator, PVR/SVR accurately predicted 52 of the 78 pa-
tients (67% MSE .41).
4.11 x
PVR
SVR
(3.6)
We illustrate the number of correct and incorrect predictions for each FC category
for PVRI and our top model equation (Figure 3.3). PVRI maps to FC I almost well
as the top model, multivariate equation 3.1, with 35 correct predictions (83%) out of
42 possible compared to 37 (88%) correct predictions for the multivariate equation.
However, in FC II and FC III-IV PVRI predictive ability drops. PVRI makes 6 out
of 25 (25%) and 5 of 11 (45%) correct predictions, respectively. Whereas, equation
3.1 makes 15 (60%) and 7 (63%) correct predictions.
3.4.3 Survival Predictors
We employed a Random Survival Forest model to select the most important
clinical indicators for the ensemble estimation of the cumulative hazard function.
The top five selected variables, PP, Cdyn, SBP, Cl, and mRAP, shown in Table 3.2
were used in a Cox proportional hazard analysis. A univariate analysis of the clinical
indicators as individual covariates may be compared to the multivariate analysis of the
clinical indicators as a single model. A Kaplan-Meier survival plot shown in Figure
3.4 illustrates the probability of a PH-related event given the number of months from
RHC.
42


3.5 Discussion
This study found that WHO Functional Classification may be predicted a year
in advance with 76% accuracy by utilizing multivariate equation 3.1, including the
measures: sDiam, MPAP, MPAP/AoP, SBP. Interestingly, as the combinations of
indicators were analyzed, it became apparent that adding indicators did not increase
predication accuracy. In addition, we identified that the individual indicator of sys-
tolic pressure and the ratio PVR/SVR predicts WHO functional class a year in ad-
vance with 68% and 67% accuracy respectively. By comparison, PVRI predicts future
functional classification with 59% accuracy.
Given that this study looked at predictive tools, we initially thought measures
that proved predicatively accurate for WHO-FC would align with those that were
predictive for time to patient event. While WHO-FC and time-to-event were most
accurately represented with a mix of hemodynamic and PA measures, their equa-
tions differed. This illustrates the complexity of this disease and how changes in the
information sought will directly impact what measures are most predictive.
Although clinical measurements are used to diagnose and guide treatment in
adult PH, fewer studies have focused on the pediatric patient. Given the impor-
tant differences between adult and pediatric PH, this left a significant gap in the
accurate prediction of future patient health for those who are less than 18 years old
when diagnosed with PH. There were also gaps in the utilization of measures beyond
hemo dy nami cs.
It should be noted that the greatest accuracy was not obtained by utilizing ei-
ther the most predictive individual measures or the most widely used clinical indi-
cators. Instead, we found that measures which are not highly predictive in and of
themselves often provide incremental value that enhances the overall accuracy of the
model. If we consider sDiam individually, the univariate regression suggests signif-
icance (p=0.0006), however the predictive value of 56% accuracy is not within the
43


top half of the most accurate individual indicators and as such may be eliminated
from consideration in the final model. However, sDiam adds important information
to the multivariate equation. Hence, the value of any given clinical indicator or mea-
surement must be tested in a multivariate setting. These nuances may highlight
differences between adult and pediatric progression of PH as well as help to guide the
research toward differentiated treatment plans.
We contend that this studys comprehensive evaluation of clinical indicators pro-
duced the best tool available for predicting future health status in the unique pediatric
PH population. Since these available measures provide 76% accuracy, there remains
opportunity to examine new measures to continue to improve prediction accuracy
going forward. Underlying this opportunity is the need to understand the precise
mechanisms driving these predictions and outcomes.
This study also provides a meaningful basis for additional studies of this or sim-
ilar equations to solidify the accuracy of this model for future clinical reference. A
predictive model for clinical use will facilitate the proactive therapies and interven-
tions needed to improve pediatric outcomes as well as creating a mathematical model
for more consistent treatment plans across clinical settings. As such, this study may
eventually reduce the number of necessary patient data measures, thereby diminishing
the time and invasive-nature of the collections.
Our study findings were supported when reviewed with our own statistical tools
and across other studies. When RHC measurements and clinical indicators are an-
alyzed individually using powerful statistical tools, AIC, Random Forest and or-
dered logistic regression prediction, the three methods produced results that were
well aligned in predicting patient outcomes. Specifically, each method identified the
same top six individual measures as the most accurate single predictors of patient
outcomes. These results also align with previous studies which suggest MPAP/AoP,
MPAP, PYR/SYR and PVRI as single measures most predictive of patient outcome.
44


3.5.1 Study Limitations
This is a single center study with outcomes assigned by one observer. The study
center is in Aurora, Colorado at an elevation of 5285 feet (ps 630mmHg, pC>2
130mmHg). The patient population had a low number of events (50) for the survival
statistics and few events for the event types: death (3) and lung transplantation
(0). Each patient was treated with individualized therapies and outcomes may be
differentially impacted. This study did not account for differences in treatment.
3.5.2 Conclusions
The detection and analysis of meaningful patterns with regard to PH outcomes
from large amounts of data is an essential step toward defining critical measurements
and the impact of the interactions among those measurements. Analysis of the top
individual predictors, either alone or in combination, does not create the greatest pre-
dictive value. Instead, our results suggest that the equation with the most predictive
accuracy of future WHO-FC includes systolic diamter of the PA, MPAP, MPAP/AoP
and SBP within the PA. This suggests that measurements of stress and straing within
the PA increase accuracy prediction of future patient health status.
45


3.6 Figures & Tables
Table 3.1: Patient Demographics
Number 78
Female 40
Age at Catheterization (yrs) 8.98 SD 6.08
Body Surface Area (mA2) 1.01 SD 0.48
Idiopathic 37
Time to Follow-Up (mos) 11.07 SD 5.4
WHO Functional Class at RHC
I 36
II 27
III 12
IV 3
WHO Functional Class at Follow Up
I 42
II 25
III 10
IV 1
Cdyn (mmHg-1) 0.02 SD 0.02
CO mL/(min*m2) 4.58 SD 2.75
delD (cm) 0.44 SD 0.15
sDiam (cm) 1.64 SD 0.65
sDiam/BSA (cm/m2) 1.71 SD 0.6
mPAP (mmHg) 35.4 SD 18.4
mPAP/Aop 0.56 SD 0.3
mRAP (mmHg) 5 SD 2.26
sPress (mmHg) 51.4 SD 23.7
PCWP (mmHg) 7.4 SD 2.35
PP (mmHg) 28.7 SD 10.8
PVRI (mmHg*min*m2/L) 7.94 SD 8.08
Pvr/Svr 0.5 SD 0.4
Strain 0.43 SD 0.23
SV/PP (/mmHg) 0 SD 0
PCWP (mmHg) 7.4 SD 2.35
46


Figure 3.1: Color M-Mode Tissue Doppler Image of the RPA, lumen inner walls and
echocardiograph.
47


Clinical Indicators
Figure 3.2: Univariate analyses confirmed by three methods: Regression, Random
Forest, AIC
Table 3.2: Cox Proportional Flazards
Univariate Analysis
Clinical Indicators Coefficient (6,) Hazard Ratio Exp (ft,) 95% Cl p-value R2
PP 0.05 1.05 (1.03-1.08) 4.2E-05 * 0.19
Cdyn 0.01 1.00 (1.002-1.01) 0.00032 * 0.11
sPress 0.02 1.02 (1.01-1.03) 0.00053 * 0.14
Cl -0.22 0.80 (0.65-0.99) 0.042 * 0.06
mRAP 0.09 1.09 (0.96-1.24) 0.18 0.02
48


Model Prediction of Functional Classification
SBD + mPAP + mPAP/Aop + sDiam
Prediction Results
WHO Functional Classification at Follow-Up
FCI FCII FC lll/IV
Total (78)
I---i Correct Predictions
| | Incorrect Predictions
Prediction Results
WHO Functional Classification at Follow-Up
§ >
o
c
o
CL
T3
O
2
FC I FC II FC lll/IV
Total (78) I 1 Correct Predictions
I | Incorrect Predictions
Figure 3.3: Prediction Results by WHO-FC Category for PVRI and the Top Equation
Model
Figure 3.4: Cox Proportional Flazards
49


Chapter 4
4. Study: The Mechanics of Preconditioning Pulmonary Artery
Segments
4.1 Abstract
Preconditioning biological tissue is the process of cyclically loading and unloading
the tissue prior to taking test measurements of the sample. Preconditioning is thought
to be necessary to obtain reproducible stress-strain curves. We wanted to know
if preconditioning negatively affects the mechanics of the pulmonary artery. We
hypothesized that preconditioning proximal pulmonary artery segments compromises
passive and active vessel wall mechanics significantly changing diameter, compliance,
stress, strain, and elastic modulus.
We randomly assigned 12-week old adult male Sprague-Dawley rats into 2 co-
horts, preconditioned (pre) and control (con). A segment of pre-hilar left pulmonary
artery was dissected, denuded of endothelium, cannulated with metal pipettes, se-
cured with silk ligatures and perfused with calcium-free Krebs-Henseleit (37C) in a
perfused vessel chamber. Luminal pressure was adjusted using a peristaltic pump and
the corresponding vessel diameter was recorded with a video dimension analyzer. Pre-
conditioned arteries were cyclically loaded and unloaded 10 times; changing luminal
pressure from 5mmHg to 60mmHg at 1Hz. Control arteries were not preconditioned.
The outer diameter of each artery was recorded as luminal pressure was changed. Re-
sulting pressure-diameter (P-D) measurements were used to calculate vessel compli-
ance (C=pressure/diameter), smooth muscle (SM) contraction, circumferential wall
stress and strain and elastic modulus (E=stress/strain).
Our results show that strain was significantly increased with luminal pressure
loads from 20mmHg (con=0.28.036, pre=0.440.030, p<0.05) through 55mmHg
50


(con=0.59.l, pre=0.820.08, p<0.05). Elastic modulus was significantly decreased
(con=126.8617.00 kPa, pre=106.647.42 kPa, p<0.05) at 20mmHg and was de-
creased but not significantly (con=203.5816.93, pre=184.967.32, p=0.095) at
55mmHg The compliance of the vessel was significantly increased (con=0.020.003mmHg/mm,
pre=0.030.002mmHg/mm, p<0.05), mean arterial diameter was significantly in-
creased (con=1.990.08 mm, pre=2.110.07 mm, p<.05) and smooth muscle con-
traction had decreased, but not significantly.
Based on these results, we conclude that preconditioned arteries are more com-
pliant, larger in diameter and have larger values of strain and decreased elastic modli
than control arteries that have not been preconditioned.
4.2 Introduction
Tissue mechanics are important for our understanding of anatomy and physiology
[114]. The functioning of bodies under physiologic loads in health and the changes
through disease help to model, diagnose and predict health status. Quantification
of the mechanics and processes of the body aims to improve diagnosis, guide ther-
apeutic treatments, and improve tissue engineering, grafts and prostheses [20]. The
foundation of quantification is experimentation, modeling, computation and mechan-
ical measurement of tissues [20]. However, quantification may not be comparable
across studies if in-vitro factors impact the distensibility of the tissues. Subsequently
testing protocols become central to the precision of the measurements to allow cross-
study comparison.
Tissue mechanics may be broadly categorized as hard tissue mechanics like bone
and teeth and soft tissue mechanics like skin, tendons, organs and blood vessels.
One major mechanical difference between these two broad categories is the amount
of deformation each experiences under physiological loads. Hard tissues typically
51


show very small amounts of deformation while soft tissues typically show larger de-
formations [20]. However, soft tissues generally do not conform well to mechanical
definitions as they are neither purely elastic nor purely plastic but show characteristics
of both (viscoelasticity). Therefore, soft tissues typically shows hysteresis between
loading and unloading and are generally anisotropic due to the fibrous arrangement
of the extracellular matrix [33] [46] [73]. These characteristics confound the ability to
mathematically quantify soft tissue properties.
There are varying protocols in the preparation of biomaterials and particularly
implant tissues [114]. Preconditioning is one form of soft tissue preparation. Pre-
conditioning is the process of cyclically loading and unloading tissue samples prior to
taking measurements. Proponents of preconditioning protocols suggest it is necessary
to obtain reproducible mechanical relationships such as stress-strain curves [47] [48].
Fung warns that preconditioning necessarily changes the internal structure of the
tissue in order to reach a steady state for mechanical measurements [32], However,
he leaves the statement open, raising the questions regarding changes to the internal
structure and the links to mechanical properties.
Cyclically loading soft tissue above physiological pressures or frequencies is not a
process found in the body. Accordingly, preconditioning tissue in a manner that may
change internal structural could also potentially alter the underlying characteristics
of the tissue and the resulting measurements and calculations.
Given the disparity between reproducible results and in-tack in-vivo structural
mechanics, we studied the impact of preconditioning on the mechanics of rat pul-
monary artery segments. We hypothesize that preconditioning proximal pulmonary
artery segments compromises passive and active vessel wall mechanics significantly
changing diameter, compliance, stress, strain, and elastic modulus. In addition, we
demonstrate that repeatable stress-strain curves and consistent mechanical response
can be obtained with non-preconditioned tissue.
52


4.3 Methods
4.3.1 Animals
We used 13 normal adult male Sprague-Dawley rats (Harlan Laboratories, USA).
All animals were treated according the University of Colorado Animal Care Program
and the Institutional Animal Care and Use Committee (IACUC) following protocol
101913(04)IE (Appendix 5).
4.3.2 Isolated Vessel Chamber Testing System
The isolated vessel chamber testing system (LSI) consists of two metal cannulas
running across a superfusate bath with approximately 2 mm between the cannulas at
the center of the bath. An isolated vessel was secured to each pipette to complete a
perfusate circuit consisting of two pressure transducers on either side of the cannulas,
a peristaltic pump, and a perfusate reservoir. The pressure transducers were placed
inline with the cannulas, one before the cannulas, and one after the cannulas. A
peristaltic pump (LSI) placed inline prior to the first transducer pumped perfusate
from a water jacketed aerated 30 ml reservoir through the first transducer, first can-
nula, tissue segment, second cannula, second pressure transducer and back into the
perfusate reservoir. The superfusate circuit consisted of the aerated superfusate 1-
liter reservoir and a second pump to circulate the solution through the isolated vessel
chamber bath and back to the reservoir. A third pump and water heater circulated
warmed distilled water around jacketed condenser glassware at the superfusate inflow
to the vessel bath and through the perfusate jacketed reservoir. The perfusate and
superfusate were re-circulated; the thermometers measured the temperature in the
reservoirs to verify a consistent temperature of 37C.
53


4.3.3 Artery Segment Preparation
The animals were anaesthetized using pentobarbital sodium. An anterior thoraco-
tomy was performed after deep pain reflexes became absent. Both lungs were removed
and placed in a buffered Krebs-Henseleit solution (mM: 118.0 NaCl, 4.75 KC1, 1.18
KH2P04, 1.18 MgS047H20, 24.80 NaHC03, 2.52 CaCl2, and 10.0 D-glucose satu-
rated with 95% 02 and 5% C02 maintaining pH at 7.45). We excised a segment of
left pre-hilar pulmonary artery approximately 4 mm in length between the pulmonary
trunk and the first hilar branch from each animal.
We mounted the arterial segment under solution (calcium-free Krebs Hense-
leit (Ca++-free KHS) solution (mM: 118.0 NaCl, 4.75 KC1, 1.18 KH2P04, 1.18
MgS047H20, 24.80 NaHC03, and 10.0 D-glucose saturated with 95% 02 and 5%
C02 maintaining pH at 7.45) on the metal pipettes in the isolated vessel chamber
testing system. The artery segments were gently rotated on a metal cannula to re-
move the endothelium and were then secured with silk ligatures at both ends to
the bath cannulas with a distance between cannula tips of about 2.5 mm. We per-
fused the cannulated artery segment with the same Ca++-free KHS solution as the
superfusate. The segments equilibrated for 1 hour at 37C less than 5mmHg of intra-
luminal pressure at a flow rate of approximately 0.5ml/min, prior to any experimental
manipulation.
We made all measurements at ambient pressure (Aurora, CO : p^ 630mmHg,
p02 ~ 130mmHg) and pressure measurements are expressed as changes relative to
ambient.
4.3.4 Non-Preconditioned Mechanical Measurements
After equilibrating, the luminal pressure of the isolated vessels was increased
in 5mmHg steps to approximately 55mmHg (range 44.8 to 55.3mmHg) at 3-minute
intervals using a peristaltic pump. At the end of each time interval, we recorded the
54


outer diameter of the cannulated artery with a video dimension analyzer (LSI) and
the corresponding intra-luminal pressure. Upon reaching an intra-luminal pressure of
approximately 55mmHg, we reduced the pressure by the same 5mmHg steps and 3-
minute time intervals to baseline, again recording the outer diameters and pressures.
The time interval was set with each new vessel to ensure that the change in diameter
following a pressure step change was complete.
The vessel, bath chamber, perfusate and superfusate reservoirs were then flushed
with high-potassium (Hi-K) KHS (mM: 14.72 NaCl, 107.98 KCL, 1.18 MgSO4-7H20,
1.18 KH2P04, 24.80 NaHC03, 2.52 CaCl2-2H20, 10.0 D-glucose saturated with 95%
02 and 5% C02 maintaining pH at 7.45). During the equilibration period, the vessels
decreased in diameter against the perfusion pressure of 5mmHg by an overall average
of 0.21 0.7 mm. Time intervals for pressure steps were evaluated as described above.
Following the equilibration period we repeated the sequence of 5mm Hg pressure steps
up to 55mmHg and back in Hi-K KHS.
4.3.5 Preconditioned Mechanical Measurements
After equilibrating in calcium-free Krebs Henseleit solution, we cyclically loaded
and unloaded the isolated vessels 10 times, changing intra-luminal pressure from
5mmHg to 60mmHg at 1Hz. The luminal pressure of the isolated vessels was then
increased in 5mmHg steps to approximately 55mmHg (range 44.8 to 55.3mmHg) at
45-second intervals using a peristaltic pump. At the end of each time interval, we
recorded the outer diameter of the cannulated artery with a video dimension analyzer
(LSI) and the corresponding intra-luminal pressure. Upon reaching an intra-luminal
pressure of approximately 55mmHg, we reduced the pressure by the same 5mmHg
steps and 45-second time intervals to baseline, again recording the outer diameters and
pressures. The time interval was set with each new vessel to ensure that the change
in diameter following a pressure step change was complete. Finally the pressure was
55


set at 30mmHg and 15mmHg and respective diameter measurements were taken to
make sure there was no drift.
The vessel, bath chamber, perfusate and superfusate reservoirs were then flushed
with high-potassium (Hi-K) KHS (mM: 14.72 NaCl, 107.98 KCL, 1.18 MgSOzi-7H20,
1.18 KH2P04, 24.80 NaHC03, 2.52 CaCl2-2H20, 10.0 D-glucose saturated with 95%
02 and 5% C02 maintaining pH at 7.45). The pressure-step and respective diameter
measurements were repeated.
4.3.6 PA Compliance:
The slope of a pressure-diameter (P-D) curve is an indication of the compliance of
the vessel. We estimated the vessel compliance in mm/mmHg for each artery (active
and passive) as the inverse of the slope of a line drawn between the 15mmHg point
and the 40mmHg on the PD curve.
4.3.7 PA Diameter & Vascular Smooth Muscle Activity
We calculated the PA diameter for each artery, solution type, and pressure step.
We report the vessel diameters as mean diameter (mm)SD. The diameter range for
each artery and solution type were calculated and reported as mean diameter range
(mm)SD.
We calculated the smooth muscle contraction as it shortens the vessel (?D) at
each pressure step for each vessel by subtracting the active diameter from the passive
diameter. Our study reports the ?D as (mm)SD.
4.3.8 PA Wall Tension
We calculate the vessel circumferential wall stress using the circumferential stress
equation for cylinders:
S
Pi x Ri Pax R0
RiX R0x (P0 Pi)
Rmid X (R0 Ri)
(4.1)
R0 Ri
56


where S = wall stress, Pi=luminal pressure, Ri=luminal radius, R0=outer radius,
Rmw2=mid-wall radius and PG=outer pressure, assumed here to be OmmHg, but kept
in the equation for completeness. Using the assumption that artery segments are
iso-volumetric, we calculated the cross sectional area for each artery segment using
the thickness of the slack artery and outer diameter at 5mmHg as initial conditions.
For each subsequent outer diameter measurement we calculated the corresponding
inner diameter assuming constant artery volume for each artery. We report mean
wall stress, maximum wall stress, and stress at mPAP for normoxic and hypoxic
conditions as (kPa)SD. We also calculated the wall stress using the thin-walled
approximation: also reported as (kPA)SD.
4.3.9 Endothelium and VSM Response
In series of experiments, we verified effective disruption of endothelial activity.
This was accomplished by adding acetylcholine (ACh) (10-5 M) to the perfusate and
superfusate solution of an activated (HiK) vessel segment over a range of pressures: 5,
20, 25, 30mmHg. The diameters remained consistent whether in the HiK solution or
in HiK with ACh supporting the disruption of endothelial activity due to the absence
of a relaxation response.
The absence of basal smooth muscle tone was verified by adding sodium nitro-
prusside (SNP) (10-5 M) to the perfusate and superfusate solution (Ca++-free KHS)
of a purportedly inactivated vessel segment over a range of pressures: 5, 20, 25,
30mmHg. The diameters remained consistent whether in the Ca++-free KHS solu-
tion or in Ca++-free KHS with SNP suggesting no pre-existing VSM tone as there
was no increase in diameter in the presence of the SNP vasodilator.
57


4.3.10 Statistics
Statistical analysis was performed using the R statistical software (Version 2.5 R
Development Core Team 2014). Tukey Honest Significant Difference (HSD) was used
to determine statistical significance set at p<0.05. Appendix 3 comprises the code
for the statistical analysis, calculations and derived measures.
4.4 Results
As compared to non-preconditioned tissue, the preconditioned vessel segments
are larger in diameter and the vessel more compliant as well as showing increases
in circumferential stress and strain while elastic modulus decreased. Smooth muscle
ability to decrease artery diameter is also impaired suggesting that preconditioning
compromises vasoactivity.
We analyzed strain at mean pulmonary arterial pressure (mPAP), corresponding
to a luminal load of 20mmHg which is commonly thought of as the elastic region
of the artery. We also analyzed strain at the top of the tested pressures, 55mmHg,
which is well outside the physiologic range, but commonly thought of as the region
of collagen engagement of the artery. Strain was significantly increased with luminal
pressure loads from 20mmHg (con=0.280.036, pre=0.440.030, p<0.05) through
55mmHg (con=0.59.l, pre=0.820.08, p<0.05). Elastic modulus was significantly
decreased (con=126.8617.00 kPa, pre=106.647.42 kPa, p<0.05) at 20mmHg and
was decreased but not significantly (con=203.5816.93, pre=184.967.32, p=0.095)
at 55mmHg (Figure 1). The compliance of the vessel was significantly increased
(con=.020.003mmHg/mm, pre=0.030.002mmHg/mm, p<0.05), mean arterial di-
ameter was significantly increased (con=1.990.08 mm, pre=2.110.07 mm, p<0.05)
and smooth muscle function decreased, but not significantly.
58


4.5 Discussion
This study reports novel data supporting the hypothesis that the mechanics of
preconditioned issue are significantly altered. Specifically this study finds:
(1) Evidence of geometrical changes to the PA wall: increased diameter and compli-
ance with preconditioning.
(2) Evidence of vascular smooth muscle changes: significant decreases in D over the
range of pressures with preconditioning and decreased T)max, though not significantly.
(3) Evidence of mechanical changes in the PA segments: increased stress and strain,
decreased elastic modulus with preconditioning.
We further demonstrate that preconditioning is not necessary to create repro-
ducible results. We illustrate repeatable stress strain curves without preconditioning
the arteries. This is the case for both activated and inactivated tissue. The exact me-
chanics behind these results are not clear, but it is possible that preconditioning rat
pulmonary arteries may isolate the mechanical properties of the extracellular matrix
proteins by breaking matrix cross-links while data from the control arteries describes
the mechanical properties of the composite material with intact matrix cross-links.
Previous studies have not compared preconditioned artery tissues to artery tissues
that had not been preconditioned.
The ability to make measurements without first preconditioning tissues may en-
hance our ability to quantify the mechanics of artery tissue and make cross-study
comparisons. This may be supported because the limited changes to the internal
structure of these tissues and may represent the state of the tissue in its natural bio-
logical setting. This leaves the opportunity for future studies to evaluate the impact
of various preconditioning protocols on the resulting stress-strain curves.
59


4.5.1 Limitations
As previously mentioned we allowed the vessel diameter to fully contract or relax
over a relatively long time interval with each pressure step prior to taking measure-
ments. Artery compliance has a time component and depends on the rate of change in
diameter. The vessel appears to be less compliant with faster changes to vessel diam-
eter. Alternatively, the vessel appears more compliant with slower diameter changes.
This viscoelastic effect helps protect the artery, attenuating high-frequency pulses and
dissipating the pulsatile energy from the heart, also acting as an elastic reservoir and
second stage pump for the heart [96]. This study removes as much of the viscoelastic
effect as possible in order to isolate the steady state elastic and contractile conditions
of the artery wall, thus neglecting the dynamic and pulsatile functions of the PA.
Protein analysis would help determine if collagen and elastin cross-links are bro-
ken upon preconditioning. Additionally, the peristaltic pump allowed us to set the
flow rate or to set the intra-luminal pressure. With pressure set, we approximated
the flow rate using a separate reservoir over a given time period. Lastly, the study
lacks biochemical data to more precisely quantify the vessel proteins, relying instead
on PA morphology to visualize the changes in cellular structure.
4.5.2 Conclusions
Taken together these data suggest that the mechanics of the rat proximal pul-
monary artery are significantly altered with preconditioning including: diameter,
compliance, stress, strain, and elastic modulus of the PA wall. We also demon-
strate the feasibility of testing and collecting mechanical measurements resulting in
reproducible and stress-strain relationships of rat proximal PA segments without pre-
conditioning.
60


4.6 Acknowledgments
This study was generously supported by NIH program Project Grant # NHLBI-K25
HL094749 (P.I. Kendall Hunter).
4.7 Figures
Holzapfel et al. Am J Physiol-Heart C, 289(5):H2048-H2058, 2005.
Figure 4.1: Ten Successive Stress-Strain Curves.
61


Figure 4.2: Schematic of the Pressure Inflation Testing Chamber Apparatus.
Time (S)
Figure 4.3: Data Traces: diameter and pressure paired data points
62


Intra-luminal Pressure (mmHg)
Figure 4.4: Data from a Complete Test of One Control Artery.
Figure 4.5: Pressure-Diameter Measurements.
63


Figure 4.6: Difference in diameter between active and passive vessels, AD.
Table 4.1: Summary Data of Preconditioning Measurements and Calculations.
Passive Vessel Mechanics Control Pre-Condition p-val
Stress at 55mniHgP [kPa] 119.3 12.71 157.4 9.60 p = .002 *
Strain at 55mmHg [mm/mm] 0.59 0.10 0.82 0.08 p = 001 *
E at 20mmHg [kPa] 126.86 17.00 106.64 7.42 p = .005 *
E at 55minHg [kPa] 203.58 16.93 189.96 7.32 p = .095
Compliance [mmHg/mm] 0.02 0.003 0.03 0.002 p = .006 *
ADmax N 0.39 0.1 0.34 0.07 p = .26
AD F(l,9) = 33.49 p = .0003*
Mean Diameter [mm] 1.99 0.08 2.11 0.07 p = .023 *
64


0.0 0.2 0.4 0.6
Strain (inm/nun)
Figure 4.7: Stress-Strain Curves.
0.8
65


Chapter 5
5. Study: Proximal Pulmonary Artery Mechanics in Pulmonary
Hyptertension and Recovery
5.1 Abstract
Recent studies suggest that proximal pulmonary artery impedance, area strain
and vascular capacitance play a major role in the pathology of Pulmonary Hyper-
tension (PH), especially in the development of right heart failure. The proximal pul-
monary artery is known to remodel and undergo mechanical stiffening in PH. However,
the changes in active vessel mechanics during and in recovery from PH are not as well
understood. This study examines changes in proximal pulmonary artery mechanics
during the development of and recovery from PH, induced by chronic hypoxia in
Sprague-Dawley rats. We studied five conditions, distinguished by exposure to vary-
ing lengths of normoxia (elevation=5285 ft, PB=632mmHg, p02=132.5mmHg), hy-
poxia (simulated elevation=17,000 ft, PB=410mmHg; p02=85.9mmHg), or hypoxia
followed by normoxia (3 Week Normoxic, 3 Week Hypoxic, 6 Week Hypoxic, 3 Week
Hypoxic + 6 Week Normoxic, 9 Week Normoxic). We found the maximum contrac-
tion of the cannulated pulmonary artery (30.45.2mmHg) was severely diminished
following 3 weeks of hypoxia (p<0.001) and was fully restored (p<0.001) in the re-
covery group. The artery wall compliance (^p) decreased dramatically with hypoxic
exposure (p<0.001). Upon recovery, compliance was increased but did not fully re-
solve (p<0.001). At any given pressure, the proximal artery diameter decreased with
hypoxic exposure and was restored to normal levels in the recovery group. Wall stress
and modulus of elasticity were decreased as well.
66


5.2 Introduction
Pulmonary hypertension (PH) is a rare but serious disorder of the pulmonary vas-
culature characterized by pulmonary artery (PA) pressures above 25mmHg [93] [98].
It has an incidence rate of 1.1 to 2.4 cases per million per year, is associated with
a significant incidence of right heart failure and a five year mortality rate of 40%
[371(53].
Pathological changes associated with PH include distal pulmonary artery vaso-
constriction, accompanied in the medium to long term by thickening and stiffening
of the artery wall [104] [26]. The narrowing of the distal pulmonary vascular bed
dramatically increases pulmonary vascular resistance (PVR) and alters the mechanic
loads on the pulmonary vascular tree [61] [79]. Consequently, studies of and treatment
for PH have focused intently on the distal vasculature [3] [6].
More recently, human studies have found that proximal pulmonary vessel me-
chanics are highly predictive of PH mortality including input impedance, area strain
and vascular capacitance [36] [52] [72], Other clinical studies have shown that proximal
stiffness is also predictive of PH progression. Given the apparent role of proximal PA
mechanics in PH, we sought to quantify PA mechanics to determine if the mechani-
cal adaptations in the proximal PA are an attempt to maintain the wall stress at a
constant level in the face of changing vessel wall loads created by PH [115].
We placed Sprague-Dawley rats into hypobaric chambers to initiate hypoxia in-
duced PH for three to six weeks and then returned them to normoxia to allow the
PH to resolve. We hypothesized that wall stress would change little despite marked
changes in pulmonary artery pressure and wall thickness due to PH suggested by
recent studies.
67


5.3 Methods
5.3.1 Animals
We used adult male Sprague-Dawley rats (Harlan Laboratories, USA) to avoid
potential sexually dimorphic effects on vascular smooth muscle [91]. All animals were
treated according the University of Colorado Animal Care Program and the Institu-
tional Animal Care and Use Committee (IACUC) following protocol 101913(04)IE
(Appendix 5).
5.3.2 Hypoxic Chambers
Hypoxia-induced pulmonary hypertension was achieved by exposing animals to
a low barometric pressure of 410mmHg (p02=85.9) in hypobaric chambers. There
were brief weekly interruptions for animal care that lasted no more than 20 minutes.
Normoxic conditions consisted of housing the animals near the barometric chamber
and providing animal care via brief weekly interventions. The barometric pressure
for these animals was about 632mmHg (p02=132mmHg) and was typical for Aurora,
CO USA (elevation=5285ft). The animals received unlimited access to water and
were all on the same 12 hour light cycle.
All animals were sacrificed following their respective environmental exposures.
We recorded the body weights and randomly allocated the animals to right heart
hemodynamics and mechanical testing. Forty-six rats were used for mechanical test-
ing and 44 for hemodynamic measurements.
5.3.3 Hemodynamic Measurements
Hemodynamic testing was performed by the University of Colorado Denver Car-
diovascular Physiology Core. Each measurement session lasted about 45 minutes.
A 1.9 French Pressure-Volume catheter FTE-1912B-6018 (Transonic Systems Inc.,
68


Ithaca, NY) inserted into the heart via right thoracotomy, measured right ventricu-
lar (RV) and PA pressures. An additional pressure-only catheter (Transonic FTH-
1611B-0018, 1.6F) placed in the femoral artery measured systemic arterial pressure.
Pressures are reported as meanistandard deviation.
The animals were induced with 5% isoflurane in room air at 1 1/min for ap-
proximately 2 minutes. From a supine position, the animals were intubated using a
tracheal cannula and connected to a 16g tubing adapter. The animal was then con-
nected to an Anesthesia Workstation (Hallowell) and anesthesia was maintained at
1.5-2.5% isoflurane in 100% oxygen. We kept peak airway pressure at 16-18cmH20,
respiratory rate at 50-150 breaths per minute, and oxygen flow at 0.5-0.8 1/min.
Each animal underwent a right thoracotomy at the level of the 6th and 7th ribs
to expose the heart. The pericardium was resected and an incision was made at
the base of the right ventricle (RV) using a 26 gauge needle. A pressure-volume
catheter was inserted through the site of incision and advanced along the length of
the RV. To eliminate ventilator artifact from the pressure-volume recordings, steady
state hemodynamics were collected with short pauses in ventilation (~ lOsec) or high-
frequency oscillatory ventilation. We occluded the inferior vena cava by threading a
suture around the vessel and pulling it taut for 10 seconds. The catheter was then
removed from the RV. Another incision was made just below the PA, and the catheter
was inserted to collect PA pressures. Data was continuously recorded with LabScribe
2 (iWorx, Dover, N) which automatically calculates and records heart rate, cardiac
output, and stroke volume. There were no complications.
5.3.4 Isolated Vessel Chamber Testing System
The isolated vessel chamber testing system (LSI, CH2) consists of two metal
cannulas running across a superfusate bath with approximately 2 mm between the
cannulas at the center of the bath. An isolated vessel was secured to each pipette to
69


complete a perfusate circuit consisting of two pressure transducers on either side of
the cannulas, a peristaltic pump, and a perfusate reservoir. The pressure transducers
were placed inline with the cannulas, one before the cannulas, and one after the
cannulas. A peristaltic pump (LSI, PS-200) placed inline prior to the first transducer
pumped perfusate from a water jacketed aerated 30 ml reservoir through the first
transducer, first cannula, tissue segment, second cannula, second pressure transducer
and back into the perfusate reservoir. The superfusate circuit consisted of the aerated
superfusate 1-liter reservoir and a second pump to circulate the solution through the
isolated vessel chamber bath and back to the reservoir. A third pump and water
heater circulated warmed distilled water around jacketed condenser glassware at the
superfusate inflow to the vessel bath and through the perfusate jacketed reservoir.
The perfusate and superfusate were re-circulated; the thermometers measured the
temperature in the reservoirs to verify a consistent temperature of 37C.
5.3.5 Artery Segment Mechanical Measurements
The animals were anaesthetized using pentobarbital sodium. An anterior thoraco-
tomy was performed after deep pain reflexes became absent. Both lungs were removed
and placed in a buffered Krebs-Henseleit solution (mM: 118.0 NaCl, 4.75 KC1, 1.18
KH2PO4, 1.18 MgS047H20, 24.80 NaHCCC, 2.52 CaCh, and 10.0 D-glucose satu-
rated with 95% 02 and 5% C02 maintaining pH at 7.45). We excised a segment of
left pre-hilar pulmonary artery approximately 4mm in length between the pulmonary
trunk and the first hilar branch from each animal.
We mounted the arterial segment under solution (calcium-free Krebs Hense-
leit (Ca++-free KHS) solution (mM: 118.0 NaCl, 4.75 KC1, 1.18 KH2P04, 1.18
MgS047H20, 24.80 NaHC03, and 10.0 D-glucose saturated with 95% 02 and 5%
C02 maintaining pH at 7.45) on the metal pipettes in the isolated vessel chamber
testing system. The artery segments were gently rotated on a metal cannula to remove
70


the endothelium and were then secured with silk ligatures at both ends to the bath
cannulas with a distance between ligatures of about 2.5 mm. We perfused the cannu-
lated artery segment with the same Ca++-free KHS solution as the superfusate. The
segments equilibrated for 1 hour at 37C less than 5mmHg of intraluminal pressure
at a flow rate of approximately 0.5ml/min, prior to any experimental manipulation.
Preliminary experiments showed an absence of a relaxation response to Acetylcholine
(10-5 M) in activated vessels, confirming the effective disruption of endothelial sig-
naling, and an absence of a relaxation response to sodium nitroprusside by vessels
in Ca++-free solution, demonstrating the absence of basal smooth muscle tone (data
not shown).
We made all measurements at ambient pressure (Aurora, CO: pb 630mmHg,
p02 130mmHg) and pressure measurements are expressed relative to ambient. Af-
ter equilibrating, the luminal pressure of the isolated vessels was increased in 5mmHg
steps to approximately 55mmHg (range 44.8 to 55.3mmHg) at 3-minute intervals
using a peristaltic pump. At the end of each time interval, we recorded the outer di-
ameter of the cannulated artery with a video dimension analyzer (LSI, VDA-10) and
the corresponding intra-luminal pressure. Upon reaching an intra-luminal pressure of
approximately 55mmHg, we reduced the pressure by the same 5mmHg steps and 3-
minute time intervals to baseline, again recording the outer diameters and pressures.
The time interval was set with new vessels to ensure that the change in diameter
following a pressure step change was complete.
The vessel, bath chamber, perfusate and superfusate reservoirs were then flushed
with high-potassium (Hi-K) KHS (mM: 14.72 NaCl, 107.98 KCL, 1.18 MgSO4-7H20,
1.18 KH2PO4, 24.80 NaHCCC, 2.52 CaCl2-2H20, 10.0 D-glucose saturated with 95%
O2 and 5% C02 maintaining pH at 7.45). The arteries were allowed to come to
equilibrium at 5mmHg and 37 C for 45 minutes (equilibration period). During the
equilibration period, the vessels decreased in diameter against the perfusion pressure
71


of 5mmHg by an overall average of 0.210.7 mm. Following the equilibration period
we repeated the sequence of 5mmHg pressure steps up to 55mmHg and back in Hi-K
KHS. The time intervals for pressure steps were evaluated as described above.
5.3.6 Derived Measures
We estimated the vessel compliance in mm/mmHg for each artery (active and
passive) as the slope of a line drawn between the 15mmHg point and the 40mmHg
on the PD curve. To quantify the impact of smooth muscle activation, we measured
the ability of the smooth muscle to shorten at a given pressure. To accomplish this,
we calculated the smooth muscle contraction as it shortens the vessel (AD) at each
pressure step for each vessel by subtracting the active diameter from the passive
diameter. We report the AD as (mm)SD.
We calculate the vessel circumferential wall stress using the circumferential stress
equation for cylinders:
s PjX Rj- PpX Rc Rj x R0x (Pp Pi)
R0 Ri Rmid X (7?o Ri)
where S = wall stress, Pi=luminal pressure, Ri=luminal radius, R0=outer radius,
Rmw2=rmd-wall radius and PG=outer pressure, assumed here to be OmmHg, but kept
in the equation for completeness. Using the assumption that artery segments are
iso-volumetric, we calculated the cross sectional area for each artery segment using
the thickness of the slack artery and outer diameter at 5mmHg as initial conditions.
For each subsequent outer diameter measurement we calculated the corresponding
inner diameter assuming constant artery volume for each artery. We report mean
wall stress, maximum wall stress, and stress at mPAP for normoxic and hypoxic
conditions as (kPa)SD. We also calculated the wall stress using the thin-walled
approximation: pf*R,----- also reported as (kPA)SD.
11 wallthickness 1 k /
72


5.3.7 Histology
After mechanical testing, a portion of each artery was fixed with 10% neutral-
buffered formalin and paraffin-embedded. Three transverse sections were cut from
each artery block at a thickness of 5/mi and stained for histological examination.
Haematoxylin and eosin (H&E) stain was used to quantify cell number and to visu-
alize the gross muscle structure and cell morphology. Massons Trichrome (American
MasterTech) was used to quantify the muscle and collagen structure. Verhoeffs
Elastin Stain (American MasterTech) was used to determine elastic fiber content and
structure. A Nikon Eclipse Ti microscope captured all artery images at 0.68 /im/pixel.
We used laboratory imaging software (NIS-Elements AR 4.20v) to quantify the nuclei
of the H&E stained arteries utilizing a combination of color, size and shape thresh-
olding of the high-resolution images. This study reports the mean number of nuclei
per fj,m2SD. We randomly selected six f00 fim transmural sections from each Mas-
sons Trichrome stained artery and from each Verhoeff Elastin stained artery. NIH
ImageJ (I.47v) was used to process the high-resolution images and measure the area
fractions of muscle, collagen, elastic fiber, as well as the total area via color threshold-
ing. Summary statistics report the mean area as fim2SD. ImageJ was also used to
measure the thickness of the media and adventitia layers of the transmural sections
with summary statistics reported as mean thickness fimSD.
5.3.8 Experimental Design & Statistical Analysis
We used a fully randomized design where six-week-old Sprague-Dawley rats were
randomly assigned to one of 5 groups: (1) HPX3 was exposed to 3 weeks of hypoxia.
(2) CNTL was kept in normoxic conditions for 3 weeks. (3) HPX6 was exposed to
6 weeks of hypoxia, a doubling of the hypoxic exposure. (4) REC was exposed to 3
weeks of hypoxia followed by 6 weeks of normoxia. (5) RECNTL was exposed to 9
weeks of normoxia.
73


Statistical analysis was performed using the R statistical software (Version 2.5 R
Development Core Team 2014). Tukey Honest Significant Difference (HSD) was used
to determine statistical significance set at p<0.05. Appendix 4 comprises the code
for the statistical analyses, calculations and derived measures.
5.4 Results
5.4.1 Animals
At the time of study entry, animals were approximately 6-week old Sprague-
Dawley Rats weighing 250-275 g. Our HPX3 and CNTL animals were 9 weeks old and
weighed the same when removed from the chambers. (CNTL 306.1010.16 g, HPX3
299.896.35 g). The HPX6 6-week hypoxic group was 12 weeks old and weighed
346.503.2 g upon removal. The REC and RECNTL groups were both 15 weeks
old and weighed 438.7512.41 g and 385.003.20 g respectively when removed from
the chamber. Figure 5.1 illustrates the experimental design and timing of the study
groups.
5.4.2 Hemodynamics
Hemodynamic measurements (Figure 5.5) show that the mean pulmonary artery
pressure (mPAP) is significantly higher HPX3 than in the CNTL (CNTL=21.863.5mmHg,
HPX3=35.326.2mmHg, p<0.05). Doubling the length of hypoxic exposure does
not cause further increases in mPAP (HPX6=34.612.4mmHg). After 6 weeks re-
covery in normoxia the mPAP returns to normal levels and is not different from the
control (REC=21.22.6mmHg, RECNTL=16.561.4mmHg, p=0.91). Heart rate,
stroke volume and cardiac output show no significant change among the groups.
74


5.4.3 Mechanical Measurements
The results of the P-D experiments are shown in Figure 5.3. The upper curves
represent the passive response and the lower curves representing the active response.
A first approximation indicates these two curves are parallel. However, a closer
examination reveals a widening of the gap between the two curves in the mid-range.
Subtracting the passive from the active curve we get a measure of the contribution
of the vascular smooth muscle which, if expressed as absolute diameter difference vs.
pressure has a peaked appearance as can be seen in Figure F and is further explained
below.
Looking at differences between the groups, the mean active diameter of the HPX3
group is reduced but not significantly different from the control group. By doubling
the hypoxic exposure the HPX6 diameter reduction is significant (CNTL=1.810.22,
HPX3=1.590.10 mm p=0.55, HPX6=1.360.1 mm £><0.05). In the recovery group
vessel diameters are close to control levels (RECNTL=1.710.14, REC=1.550.09
mm p=0.22). The mean diameter of the passive vessels follows a similar pattern.
The passive vessel diameters are reduced albeit not significantly after 3-weeks hypoxic
exposure. In the 6-week hypoxic group the diameter reduction becomes significant
(CNTL=2.070.27, HPX3=1.710.15 mm p=0.09, HPX6=1.560.08 mm p<0.05).
In the recovery group, vessel diameters are at normal levels (RECNTL=1.970.08
mm, REC=1.890.12 mm £>=0.99).
The full diameter range of the passive vessels in Ca++-free KHS is signifi-
cantly reduced in the hypoxic groups (CNTL=1.220.19 mm, HPX3=0.490.12
mm £><0.05, HPX6=0.540.10 mm £><0.05). The full diameter range for the re-
covery group remains impaired, not significantly different from the hypoxic groups
but significantly different from the recovery control group (REC=0.680.08 mm,
RECNTL=0.900.11 mm £>=0.04, HPX3 £>=0.05). The hypoxic diameter range
of the active vessels in Hi-K KHS is significantly reduced (CNTL=1.370.22 mm,
75


HPX3=0.460.14 mm p<0.05, HPX6=0.550.09 mm p<0.05). The active vessel
diameters return to normal levels upon recovery in normoxia (REC=0.800.11 mm,
RECNTL=0.980.12 mm p<0.05).
Examining the diameter at the hemodynamically measured MPAP for each an-
imal, the mean diameter of the control group is 1.460.16 mm at 21.9mmHg. The
diameter of the 3-week hypoxic group is 1.620.12 mm at mPAP=34.6, p<0.05.
5.4.4 PA Compliance
Vessel compliance (Figure 5.3) recovers slightly but not completely in the re-
turn group as it is different from both the hypoxic and control groups. Within
each group, vessel compliance with VSM activation is the same as compliance with-
out VSM activation as indicated by the parallel P-D curves. However, the compli-
ance changes significantly between groups. The 3-week hypoxic group is less com-
pliant than the control group (CNTL=0.030.006, HPX3=0.0110.003mmHg/mm,
p<0.05). The 6-week hypoxic group is no less compliant than of the 3-week hy-
poxic group (HPX3=0.0110.003, HPX6=0.0120.002mmHg/mm, p=0.99), sug-
gesting the vessel walls do not stiffen further when the time of exposure to hy-
poxia is doubled. After 6-weeks recovery in normoxia the active vessels increase
in compliance and approach time matched controls (REC=0.0160.004mmHg/mm,
RECNTL=0.0210.002, p=0.46). In contrast, the passive vessels without VSM ac-
tivation were slightly more compliant than the hypoxic groups, however, are not sig-
nificantly different from the hypoxic groups (p=0.07). These results are graphically
represented in Figure 5.3.
5.4.5 Ability of Vascular Smooth Muscle to Alter Diameter
The ability of the PA to change diameter is an indication of how well the VSM is
able to contract and is illustrated in Figure ??. We find the control group is able
76


to change diameter by up to 0.48 mm. With hypoxic exposure, AD is significantly
reduced by more than 50% of normal levels (CNTL=0.480.18, HPX3=0.160.10
mm p<0.05, HPX6=0.240.08 mm p<0.05). Some function has adapted by 6-weeks
of hypoxic exposure, though still significantly decreased. AD returns to normal levels
in the recovered group (RECNTL=0.360.06, REC=0.450.09 mm p=0.73). The
plots in Figure 5.4 are suggestive of the sliding filament theory, showing the maximum
ability of the SM to contract at approximately 30.45.2mmHg for each group. At
pressures significantly higher and lower than 30.4mmHg, AD decreases, suggesting
reduced contractile ability on either side of the peak function.
5.4.6 PA Wall Tension
The mean wall tension for each study group is illustrated in Figure 5.3, bot-
tom panel. PA Wall Stress with Ca++-KHS activation is significantly reduced with
hypoxic exposure (CNTL=76.436.07 kPa, HPX3=30.504.72 kPa, p<0.05 and
HPX6=26.732.42 kPa, p<0.05) and remains significantly reduced upon recovery in
normoxia (RECNTL=47.712.56 kPa, REC=32.804.90 kPa, p=0.02). The maxi-
mum wall tension with Ca++-KHS activation follows the same pattern: reduced with
hypoxic exposure and in recovery (CNTL=205.7840.76 kPa, HPX3=65.7713.30
kPa, p<0.05, HPX6=59.676.32 kPa, p<0.05, REC=81.4011.74 kPa, p=0.02). In
addition, the wall tension of the recovery controls that were kept in normoxia for
9 weeks were significantly lower than the younger controls that had only spent 3
weeks in normoxia (elevation=5285ft, p02=132.5) (p<0.05). The approximation of
the thin-walled cylinder was significantly lower than the exact equation for circum-
ferential wall-stress in all groups (p<0.05) (Appendix E).
We calculated wall stresses at 21mmHg and 35mmHg representing mean operating
pressures of the normoxic and hypoxic groups respectively. Wall stress was signifi-
cantly lower in the hypoxic groups at a given pressure (p<0.05). However, we also
77


compared the control groups at their operating pressure, 21mmHg, to the hypoxic
groups at their mPAP, 35mmHg. At the normoxic operating pressure of 21mmHg
the wall stresses were 24.505.22kPa and 17.680.96kPA for the Control and Re-
covery Control groups respectively. At the hypoxic operating pressure of 35mmHg
the hypoxic arteries were 31.004.12kPa, 27.002.39kPa and 31.075.01kPa for
HPX3 and HPX6 groups respectively. There is then a mis-match with the recov-
ery group when the operating pressure returns to normoxic levels and the wall stress
at 21mmHg remains decreased and significantly different from the recovery control
group (REC=12.461.90kPa, RECNTL=17.680.96kpa, p<0.05).
5.4.7 Histology
Histological analysis and quantification of H&E stained artery tissue indicates
no change in the number of nuclei per unit area (no hyperplasia) among all groups.
Histological analysis of muscle area of the Massons Trichrome stained tissue shown in
Figure 5.7 also shows no hypertrophy among the five groups (CNTL=33.310.7 /mi2,
HPX3=36.711.7 11 m2 p=0.99, HPX6=38.07.5 /mi2, p=0.98, RECNTL=40.69.7
/mi2, REC=41.710.4 /mi2, p=0.99). The collagen tissue area doubles with hypoxic
exposure (CNTL=48.368.83 /mi2, HPX3=81.917.2 /mi2, p<0.05, HPX6=91.612.4
/mi2 p<0.05), remains doubled with 6-week hypoxic exposure, and remains doubled
in the recovery group (RECNTL=47.411.4 /mi2, REC=90.315.6 /mi2, p<0.05).
Quantification of elastic fiber area fraction of Verhoeffs Elastic stained tissue shows
no difference among the five groups (CNTL=25.16.3 /mi2, HPX3=19.08.5 /mi2
p=0.99, HPX6=18.94.9 /mi2, p=0.98, RECNTL=26.99.2 /mi2, REC=18.09.1
/mi2, p=0.99) (Figure 5.7). The tissue thickness was also measured for the Mas-
sons Trichrome stained sections. The media thickness did not change among
the five experimental groups (CNTL=46.27.7mm, HPX3=56.89.3mm p=0.25,
HPX6=59.57.7mm p=0.15, REC=60.05.3mm, RECNTL=58.97.0mm p=0.99).
78


The adventitia thickness increased with hypoxic exposure and remained increased in
recovery (CNTL=41.38.7mm, HPX3=77.26.5mm, p<0.05 HPX6=78.07.4mm,
p<0.05, REC=83.217.1mm, RECNTL=55.412.9mm p<0.05).
5.5 Discussion
With this study we are able to answer three main questions for each measured
property of PA mechanics and hemodynamics: How hypoxia affects each property
and is this within the range of prior studies, how much each property recovers from
hypoxia, and how recovery vessels compare to vessels that have not had hypoxic
exposure.
The important findings of this study are: (i) After 6 weeks of recovery following
3 weeks hypoxia, the compliance of the proximal PA remains impaired suggesting
a highly asymmetric process; (ii) The mean vessel diameter continues to decrease
through 6 weeks of hypoxic exposure and the maximum diameter remains reduced
after 6 weeks of recovery; (iii) Hi-K KHS induced VSM contraction is significantly re-
duced after 3 weeks of hypoxia. Contraction partially recovers after a further 3 weeks
of hypoxia, suggesting some form of hypoxic adaptation is occurring. Upon recovery
in normoxia, VSM contraction fully returns to normal levels; (iv) Maximum VSM
contraction with Hi-K KHS occurs at a mean pressure of 30.45.2mmHg regardless of
environmental exposure, (v) Wall stress in the proximal PA over the pressure range
of 5-55mmHg is significantly reduced with hypoxic exposure and remains reduced
upon recovery.
5.5.1 Vessel Properties
The nearly parallel shift in the P-D curves of vessels under all conditions (Figure
5.3) suggests that the action of smooth muscle is simply to change the diameter of the
vessel without significantly altering the compliance of the vessel wall. The similarity
79


of the active response to the passive response is consistent with passive elements dom-
inating the overall response. This would be aligned with the smooth muscle having a
much higher stiffness than the passive elements under both conditions. Active arterial
smooth muscle can generate peak stresses of approximately 2.5 x 105N/m2 which is
higher than the wall stress under our experimental conditions and would allow the
smooth muscle to act as a rigid connection [78]. Under purely passive conditions,
the smooth muscle could extend along its passive length tension curve until it also
became a nearly rigid connection between passive elements, which again dominate
the vessel response.
The mean diameter of the passive PA at various pressures is shown in Figure
5.6. As can be seen, the diameter decreases with hypoxia and recovers on return to
normoxia, however the time course is extremely slow. These are changes in passive
properties and suggest that the collagen length tension curve is shifting to shorter
lengths and smaller diameters. How this occurs remains a puzzle. One theory is that
hypoxic vasoconstriction in-situ results in smaller diameters allowing new collagen to
be laid down at shorter mean lengths. However, measurements of the PA in situ at
the pathological mPAP show that the PA increases in diameter rather than decreases
[111].
5.5.2 Hemodynamics
We demonstrate that while the hemodynamic profile returns to normal levels,
indicating that the heart is functioning normally, the mechanics of the proximal pul-
monary artery, including vessel diameter, compliance and wall stress remain impaired
upon recovery from PH (Figure 5.5, panels a,b,c or Appendix E). Further, full recov-
ery of the smooth muscle contraction suggests that a doubling of the artery thickness
and the amount of collagen does not in itself affect contractile ability.
80


5.5.3 PH Development
Our Cntl animals, despite being mildly hypoxic at Denver altitudes, had mPAPs
within the range 16-22mmHg, which were consistent with prior studies of hypoxia-
induced PH in rats. These animals have increased vessel stiffness, increased artery
thickness and increased collagen depositions [72] [87] [103]. Our results agree with
previous findings of normoxic mPAP values lying between 16 and 22mmHg. The
results in this study are consistent with published hemodynamic measurements of
mPAP in the hypoxic rat being about 35.52mmHg [18]. In the development of
PH, our findings support prior studies showing the proximal vessels not only stiffen
but decrease in diameter, decrease wall stress and increase in elastic modulus. We
also show that the proximal vessels lose some contractile function as evidenced by a
reduction in the ability of active contraction to shorten the vessel in hypoxia (Figure
5.4). A return to normoxia provides a full recovery in function. Hypoxia is known
to affect K-channel function and this may be responsible for the diminished function
[108] [117].
5.5.4 Recovery from Hypoxic Exposure
Previous studies of rat hypoxic recovery models have found a wide range of results
from complete recovery to persistent changes due to remodeling as a result of PH,
suggesting the factors that determine recovery are not fully understood. After 10 days
of hypoxic exposure at 10% 02 and pressures between 74-80mmHg and normoxia
up to 10 days found collagen and elastin contents at normal levels [37]. A fuller
description would indicate what happened to collagen and elastin durign hypoxia.
Hyoxic exposure at 380mmHg exposure for 10 days doubled the adventitia thick-
ness with increased fibroblasts and collagen fibers. During 70 days of recovery in room
air, the adventitia thickness returned to normal but the collagen fiber concentration
remained increased [72], This suggests that the breakdown of the additional collagen
81


is a slow process.
Hislop and Reid found that RV hypertrophy resolved, but the loss of small ar-
teries (arterial pruning) with outer diameters < 200/mi did not recover from hypoxic
exposure of 2 weeks at 380mmHg and recovery up to 8 weeks in room air [43]. With
10% 02 hypoxic exposure up to 3 weeks and recovery in room air up to 20 weeks, Her-
get found that muscularization and arterial pruning persisted while mPAP returned
to normal [41]. Finally, with 5 weeks of exposure to 380mmHg and up to 5 weeks of
recovery, Heath found normalization of both medial thickness and RV hypertrophy
[40].
At 3 weeks of hypoxic exposure and 6 weeks of recovery in normoxia, our study
falls within the time-frames of previous recovery studies. Thus, we expected collagen
area and adventitia thickness to decrease upon recovery in normoxia. Collagen may
have remained increased because normoxia for this study was at an elevation of 5285
ft. We found no significant changes in media thickness or area, nor did we find changes
in elastin area. Our hemodynamic measurements support mPAP and RV recovery.
Within the context of these structural and hemodynamic changes, we examine the
vessel diameter, AD, vessel compliance and wall tension to gain insight into the PA
mechanics upon recovery from PH.
5.5.5 Limitations
As previously mentioned we allowed the vessel diameter to fully contract or re-
lax over a long time interval with each pressure step prior to taking measurements.
Artery compliance has a time component and depends on the rate of change in diam-
eter. The vessel appears to be less compliant with faster changes to vessel diameter.
Alternatively, the vessel appears more compliant with slower diameter changes. This
viscoelastic effect helps protect the artery, attenuating high-frequency pulses and dis-
sipating the pulsatile energy from the heart, also acting as an elastic reservoir and
82


second stage pump for the heart [96]. This study removes as much of the viscoelastic
effect as possible in order to isolate the steady state elastic and contractile conditions
of the artery wall, thus neglecting the dynamic and pulsatile functions of the PA.
The vessels were pressurized up to approximately 55mmHg and not above
60mmHg. This was necessary to ensure full force development of the VSM throughout
the testing. Occasionally, if pressures exceeded 65mmHg, the VSM contraction was
decreased, often following the curves of the passive vessel curves, and the test would
not be used. In these cases, the VSM may have been stretched beyond the limits
for actin and myosin attachment. These data and tests while beyond the scope of
the current project, are potentially supported from unpublished data currently under
study.
Additionally, the peristaltic pump allowed us to set the flow rate or to set the
pressure. With pressure set, we approximated the flow rate using a separate reservoir
over a given time period. Lastly, the study lacks biochemical data to more precisely
quantify the vessel proteins, relying instead on PA morphology to visualize the changes
in cellular structure.
5.5.6 Conclusions
In summary, we report that proximal PA segments from rats that have chronic
hypoxia-induced PH and are allowed to recover for six weeks in normoxia have im-
paired mechanics: decreased vessel diameters, decreased vessel compliance and de-
creased wall tension as compared with age matched controls. We also found with
hypoxic exposure the PA segments have decreased VSM contraction in Hik KHS
solution and that VSM contraction returns to normal levels during recovery. These
findings suggest that proximal VSM contraction is not affected by collagen deposition
which persists in recovery, nor by the other vessel mechanics which remain decreased.
We speculate that therapies aimed at reducing or preventing collagen deposition will
83


improve proximal PA mechanics upon recovery from PH or upon normalization of the
hemo dy nami cs.
5.6 Figures
3 Week Normoxic
(Control)
3 Week Hypoxic
(PH)
6 Week Hypoxic
(PH)
3 Wk Hypoxic + 6 Wk Normoxic
(Recovery)
9 Week Normoxic
(Recovery Control)
Normoxic
Hypoxic
0 3 6 9
Time (weeks)
Figure 5.1: Experimental Design.
A.
100
Time (S)
B.
Figure 5.2: Pressure-Diameter paired points and records.
84


3-Wk Normoxic 3-Wk Hypoxic 6-Wk Hypoxic 3-Wk Hpx + 6-Wk Nx 9-Wk Normoxic
(Control) (Hpx) (Hpx) (Recovery) (Recovery Control)
n=8 n=8 n=6 n=7 n=8
3-Wk Normoxic 3-Wk Hypoxic 6-Wk Hypoxic 3-Wk Hpx + 6-Wk NX 9-Wk Normoxic
(Control) (Hpx) (Hpx) (Recovery) (Recovery Control)
n=8 n=8 n=6 n-7 n=8
Pressure (mmHg)
Figure 5.3: PA Diameter and Wall Stress.
85


10 20 30 40 50
Pressure (mmHg)
Figure 5.4: Ability of Vascular Smooth Muscle to Change Diameter, AD, through
PH and Recovery.
86


Full Text

PAGE 1

MECHANICSOFTHEPROXIMALPULMONARYARTERYINPULMONARY HYPERTENSION by SHAWNAL.BURGETT M.S.,UniversityofColoradoDenver,2009 B.S.,MontanaStateUniversity,Bozeman,2001 Athesissubmittedtothe FacultyoftheGraduateSchoolofthe UniversityofColoradoinpartialfulllment oftherequirementsforthedegreeof DoctorofPhilosophy Bioengineering 2015

PAGE 2

ThisthesisfortheDoctorofPhilosophydegreeby ShawnaL.Burgett hasbeenapprovedforthe DepartmentofBioengineering by KendallHunter,Advisor JohnS.Walker,Advisor RobinShandas,Chair DunbarIvy ReubenBlairDodson November20,2015 ii

PAGE 3

Burgett,ShawnaL.Ph.D.,Bioengineering MechanicsoftheProximalPulmonaryArteryinPulmonaryHypertension ThesisdirectedbyAssistantProfessorKendallHunterandAssistantProfessorJohn S.Walker ABSTRACT PulmonaryhypertensionPHisanincurable,progressivediseasewithpoor survival.PHisdenedasanincreaseinmeanpulmonaryarterialpressureabove 25mmHgatrest.DespitetherapidadvancesinPHtherapies,theprogressivenature ofthediseaseeventuallyresultsinrightheartfailureanddeath.PHischaracterized byvasoconstriction,remodeling,andthickeningofthesmalltomediumarteriesand arterioles,resultinginincreasedpulmonaryvascularresistance.Pulmonaryvascular resistanceisoftenthesingleparameterusedtoassessdiseaseseverityandresponse totreatment.However,othermarkersofvascularfunction,forexamplepulmonary vascularstiness,areastrainandvascularcapacitancehavebeenshownaspredictive ofmortalityinPH. Thisthesisincludesthreestudiesthatfocusontheintersectionofclinicaland scienticinvestigationstoevaluatemechanicalchangesintheproximalpulmonary arteryduringPHandtheirimpactonpatientoutcomes.Therstisaclinicalstudy utilizingarangeofclinicalmeasuresincombinationtoaccuratelypredictfuturepatienthealthstatustofacilitateproactiveclinicaltherapies.Thesecondandthird studiesinvolvedanimalmodels.Thesecondstudyfocusedonthemechanicalimpactsofpreconditioningtissue.Thethirdstudylookedatthemechanicalchangesin theproximalpulmonaryarteryuponPHonsetandrecovery.Allofthesestudiescenteredondeningthemechanicalfunctionoftheproximalpulmonaryarteriesthrough timeinordertoassesshealthstatus.Weaimtoquantifychangesintheproximal iii

PAGE 4

pulmonaryarteriesinthesettingofPHtohelpguidetherapeutictargetsandclinical interventions. Thendingsofthesethreestudiessupportthatthetheproximalpulmonary arterymechanicsareimportantforpredictionoffuturepatienthealthstatus.In animalstudiesofsofttissueusingpressure-inationtesting,samplepreparationmay signicantlyaecttheresultingmechanicsmeasurements.Caremustbetakenin tissuepreparationandtestinginordertopreservethemechanicalcharacteristicsas closelyaspossibletoin-vivosettings.Finally,proximalPAsegmentsfromratsthat havechronichypoxia-inducedPHandareallowedtorecoverforsixweeksinnormoxia haveimpairedmechanics:decreasedvesseldiameters,decreasedvesselcomplianceand decreasedwalltensionascomparedwithagematchedcontrolswhichmaycauselonger termrightheartdysfunction.Wealsofound,withhypoxicexposure,thePAsegments havedecreasedVSMcontractionandthatVSMcontractionreturnstonormallevels uponrecovery.Wespeculatethattherapiesaimedatreducingorpreventingcollagen depositionmayimproveproximalPAmechanicsuponrecoveryfromPHorupon normalizationofthehemodynamics. Theformandcontentofthisabstractareapproved.Irecommenditspublication. Approved:KendallHunter Approved:JohnS.Walker iv

PAGE 5

DEDICATION Formyhusband,Terry,myson,DerekZane,andmydaughter,KendraQuin, whoinspiremeeverydaytobemyselfandtobebetterthanmyself.Youaremy heart.TomyparentsJoyceandKenforinspiringmewiththeloveoflearning,the loveofexperimentingandthecondencetodoanything.Tomysister,Jennifer,for keepingmegroundedbyoursharedroots,forthecomfortofoursimilaritiesand forinspiringmetowardourdierenceswhichseemachievableanddelightfulbyyour example.Tomyfriends,Andrea,NancyandReneeforyourwonderfulcompany,for keepingmeonyourcontactlistsandgivingmykidsasecondhome.ToAndrewfor youruncannytimingandinsanelygoodedits.AgaintoNancy-Ant,youarethebest softballplayer,lawyer,adviser,scheduler,editor,strategist,plannerandday-saver ever;therearenomoreX's. Iamextremelyluckyandeternallygratefultohaveyouallinmylife.Without yourlove,friendshipandsupportthisendeavourwouldnothavebeenpossible.I dedicatethisthesistoyouall. v

PAGE 6

ACKNOWLEDGMENT Specialthanksgotomyadvisers,KendallHunterandJohnS.Walker,towhom Ioweagreatdealofgratitudeformentorship,supportandguidance.Manythanks toRobinShandasformakingallthispossiblethroughhisdiscussionsregardingthis work,histeachingandasthefounderandvisionaryofUCDBioengineering.Special thankstoBlairDodsonfortheaccess,set-up,mentoring,andsupportnecessarytorun themechanicaltestingapparatusandgraciouslysharingitforthefollowing3years.I wouldliketothankDr.DunbarIvy,Dr.SteveAbman,andDr.LoriWalkerformany discussionsandsuggestionsonthiswork.IwishtothankDr.KurtStenmarkand theCardio-VascularPulmonaryResearchgroupforallowingaccesstotheirhypobaric chambers.SpecialthankstoDr.MichaelWunderforstatisticalmethodsmentoring. ManythankstoJulesHarralforthehemodynamictestingandforalwaysndingtime forthesestudies.ThanksalsotoGregSeedorfforhisgenerositywithequipment,setupandlabspace.ThankyoutoDr.XiaotaoLiandYanmaiDufortheirhelpinthe lab,dayinanddayout.ThankstoAndrewEitel,NicholasHobsonandMelanieDufva forhelpwithdatacollection.Finally,thankyoutomycolleaguesintherstclassof UCDBioengineering,especially,BryanYunker,StephenHumphriesandDerekEilers fortheirsupportandcomraderyaswenavigatedthisnewterritory. ThisthesiswouldnothavebeenpossiblewithoutthegeneroussupportfromP.I. RobinShandas,NIHTrainingGrant5T32HL072738-09,P.I.KendallHunterNIH MentoredQuantitativeResearchDevelopmentAwardNHLBI-K25HL094749,lab supportfromJohnS.Walker,labequipmentfromR.BlairDodson,andclinicaldata andperspectivefromDunbarD.Ivy. vi

PAGE 7

TABLEOFCONTENTS Tables........................................xii Figures.......................................xiii Chapter 1.MotivationandSpecicAims.........................1 2.Background...................................3 2.1TheVascularCircuitDesign......................3 2.1.1HistoricalPerspective:deMotuCordis.............6 2.2MeanBloodFlowDependsonPressureandResistance.......7 2.2.1BloodFlow............................7 2.2.1.1HistoricalPerspective:Poiseuille'sLaw.........9 2.2.2VascularResistance.......................10 2.2.3BloodPressure..........................11 2.2.3.1HistoricalPerspective:Windkessel............13 2.3ArteryWallAnatomy..........................14 2.3.1Intima...............................15 2.3.2Media...............................15 2.3.3Adventitia.............................15 2.3.4ArteryWallComposition....................16 2.3.4.1Endothelium........................16 2.3.4.2SmoothMuscleCells...................17 2.3.4.3Collagen..........................18 2.3.4.4Elastin...........................21 2.4RegulationofBloodVessels.......................22 2.5HypoxicVasoconstriction........................22 2.6PulmonaryHypertension........................23 2.7LargeArteriesinPulmonaryHypertension..............25 vii

PAGE 8

2.7.1ExtracellularResponse......................25 2.7.2CellularResponse........................25 2.8EvaluationofArterialMechanics....................27 2.8.1 In-Vivo ..............................28 2.8.1.1RightHeartCatheterization...............28 2.8.2 In-Vitro ..............................29 2.8.2.1UnixaialTest.......................29 2.8.2.2RingTest.........................30 2.8.2.3BiaxialTest........................30 2.8.2.4PressureInationTest..................30 2.8.2.5Pre-Conditioning.....................31 2.9RecoveryStudies.............................32 2.10StudiesinthisThesis..........................33 3.Study:MultivariatePredictionofClinicalIndicators............34 3.1Abstract.................................34 3.2Introduction...............................35 3.3Methods.................................36 3.3.1StudyDesign...........................36 3.3.2PatientPopulation........................36 3.3.3ClinicalDataCollection.....................37 3.3.4StatisticalAnalysis........................38 3.4Results..................................39 3.4.1SingleVariableMeasuresasPredictorsofOutcome......40 3.4.2CombiningMultipleMeasuresasPredictorsofOutcome...41 3.4.3SurvivalPredictors........................42 3.5Discussion................................43 3.5.1StudyLimitations........................45 viii

PAGE 9

3.5.2Conclusions............................45 3.6Figures&Tables.............................46 4.Study:TheMechanicsofPreconditioningPulmonaryArterySegments..50 4.1Abstract.................................50 4.2Introduction...............................51 4.3Methods.................................53 4.3.1Animals..............................53 4.3.2IsolatedVesselChamberTestingSystem............53 4.3.3ArterySegmentPreparation...................54 4.3.4Non-PreconditionedMechanicalMeasurements........54 4.3.5PreconditionedMechanicalMeasurements...........55 4.3.6PACompliance:.........................56 4.3.7PADiameter&VascularSmoothMuscleActivity.......56 4.3.8PAWallTension.........................56 4.3.9EndotheliumandVSMResponse................57 4.3.10Statistics.............................58 4.4Results..................................58 4.5Discussion................................59 4.5.1Limitations............................60 4.5.2Conclusions............................60 4.6Acknowledgments............................61 4.7Figures..................................61 5.Study:ProximalPulmonaryArteryMechanicsinPulmonaryHyptertension andRecovery..................................66 5.1Abstract.................................66 5.2Introduction...............................67 5.3Methods.................................68 ix

PAGE 10

5.3.1Animals..............................68 5.3.2HypoxicChambers........................68 5.3.3HemodynamicMeasurements..................68 5.3.4IsolatedVesselChamberTestingSystem............69 5.3.5ArterySegmentMechanicalMeasurements...........70 5.3.6DerivedMeasures.........................72 5.3.7Histology.............................73 5.3.8ExperimentalDesign&StatisticalAnalysis..........73 5.4Results..................................74 5.4.1Animals..............................74 5.4.2Hemodynamics..........................74 5.4.3MechanicalMeasurements....................75 5.4.4PACompliance..........................76 5.4.5AbilityofVascularSmoothMuscletoAlterDiameter.....76 5.4.6PAWallTension.........................77 5.4.7Histology.............................78 5.5Discussion................................79 5.5.1VesselProperties.........................79 5.5.2Hemodynamics..........................80 5.5.3PHDevelopment.........................81 5.5.4RecoveryfromHypoxicExposure................81 5.5.5Limitations............................82 5.5.6Conclusions............................83 5.6Figures..................................84 6.Conclusions...................................89 6.1MajorFindings.............................89 6.2ClinicalRelevance............................90 x

PAGE 11

6.3ScopewithintheExistingLiterature..................91 6.4FutureStudies..............................92 6.5ConcludingRemarks...........................92 References......................................94 Appendix A.TopPredictiveModels.............................106 B.R-codeStudy1,MultivariatePrediction...................108 C.R-codeStudy2,Preconditioning........................164 D.R-codeStudy3,MechanicsofthePPA....................171 E.SummaryDataofStudy3,PDMeasurementsandCalculations......219 F.AnimalProtocolDetailReport........................220 xi

PAGE 12

TABLES Table 3.1PatientDemographics............................46 3.2CoxProportionalHazards..........................48 4.1SummaryDataofPreconditioningMeasurementsandCalculations....64 xii

PAGE 13

FIGURES Figure 2.1TheCardio-PulmonaryCircuits.......................5 2.2BloodPressuresthroughtheCardiovascularCirculation.........11 2.3WindkesselRepresentsArterialComplianceandElasticRecoil......12 2.4VesselWallAnatomy:ArteriesandVeins.................14 2.5SmoothMuscleCells.............................16 2.6CollagenTripleHelixStructureandAggregation.............18 2.7Elastin.....................................20 2.8 In-Vitro Testing-Systemschematicdiagrams................27 2.9PressureInationTestingSystemSchematic................31 3.1ColorM-ModeTissueDopplerImageoftheRPA,lumeninnerwallsand echocardiograph................................47 3.2Univariateanalysesconrmedbythreemethods:Regression,Random Forest,AIC..................................48 3.3PredictionResultsbyWHO-FCCategoryforPVRIandtheTopEquation Model.....................................49 3.4CoxProportionalHazards..........................49 4.1TenSuccessiveStress-StrainCurves.....................61 4.2SchematicofthePressureInationTestingChamberApparatus.....62 4.3DataTraces:diameterandpressurepaireddatapoints..........62 4.4DatafromaCompleteTestofOneControlArtery.............63 4.5Pressure-DiameterMeasurements.......................63 4.6Dierenceindiameterbetweenactiveandpassivevessels,D.......64 4.7Stress-StrainCurves..............................65 5.1ExperimentalDesign..............................84 5.2Pressure-Diameterpairedpointsandrecords................84 xiii

PAGE 14

5.3PADiameterandWallStress.........................85 5.4AbilityofVascularSmoothMuscletoChangeDiameter,D,throughPH andRecovery.................................86 5.5mPAP,CO,PVR,wallstress,compliance,diameter............87 5.6MeanDiameterChangeswithPressureandGroup.............87 5.7StructuralChangestothePAMeasuredbyTissueAreasofHistological Sections.....................................88 xiv

PAGE 15

Chapter1 1.MotivationandSpecicAims PulmonaryhypertensionPHisadisorderofthepulmonaryvasculaturecharacterizedbyincreasesinarterialpressure.PHisarareconditionwithanincidencerate of1.1to2.4casespermillionperyear[37].However,withave-yearmortalityrate of40%,progressivelydecliningqualityoflifeforpatients,rightheartdysfunction, andincreasesinhospitalization,PHposessignicantmedicalchallenges[53].This isacomplex,multifaceteddisorder.Reectingthiscomplexity,theWorldHealth OrganizationWHOclinicallyclassiedPHintocategoriesofdisordersthatshare similarcharacteristics,pathogenesisortherapeuticmanagement[17][98]. PHischaracterizedbyvasoconstriction,remodeling,andthickeningofthesmall tomediumarteriesandarterioles[81],[35].PHaectsthewholelungvasculature bynarrowingthedistalpulmonaryvascularbed,dramaticallyincreasingpulmonary vascularresistancePVR.ThisincreasedPVRhasacommensurateimpactonthe demandplacedontherightventricleoftheheartwhichmustforcebloodthroughthe narrowedvasculartree.Becauseofthecharacteristicdistalnarrowing,studiesofPH havefocusedintentlyonthedistalvasculature. Therehavebeenfewerstudiesinvolvingtheproximalpulmonaryarteries.The studiesthathavefocusedonproximalpulmonaryarteriesPPAinPHcommonly reportthickeningofthemediaandadventitiallayers.AreastrainofthePPAand vascularcapacitancehavebeenshowntobepredictiveofPHoutcomesandPPAinput impedanceimprovespredictioninchildrenwithPH.Fewerstudieshaveexaminedthe mechanicsofthePPAinordertoquantifyvascularchangeswithrespecttodisease progression.Specically,theimpactofchangesinactivevesselmechanicsuponPPA remodelingandmechanicalstieninginPHarestillnotwellunderstood. Wechosetofocusontheintersectionofclinicalworkandscienticstudiesto evaluatemechanicalchangesinthePPAduringPHandtheirimpactonpatient outcomes.Thisthesisincludesthreestudies.Therstisaclinicalstudyfocusedon utilizingarangeofclinicalmeasuresinconcerttoaccuratelypredictfuturepatient healthstatustofacilitateproactiveclinicaltherapies.Thesecondandthirdstudies involvedanimalmodels.Thesecondstudyfocusedonthemechanicalimpactsof preconditioningtissue.ThethirdstudylookedatthemechanicalchangesinthePPA uponPHonsetandrecovery.Allofthesestudiesfocusedondeningthemechanical changesandmeasuresinvolvedinPHwithagoalofidentifyingopportunitiesfor 1

PAGE 16

futuredevelopmentoftherapeutictreatmentsthatprovidelong-termresolutionof PHrelatedhealthissues. SpecicAims 1. Aim1 Hypothesis:RightheartcatheterizationRHCandColorM-Mode TissueDopplerImagesCMM-TDIofproximalpulmonaryartery functionimprovetheaccuracyofpatientoutcomepredictionsinconrmedcasesofpediatricpulmonaryhypertension. Approach:Analysisofhemodynamic,clinical,ultrasoundandRHCdatafor78 pediatricpatientswithconrmedPulmonaryArterialHypertensiontodeterminewhatcombinationofclinicalmeasurementstakenattheinitialRHCmost accuratelypredictpatientoutcomeasdenedbyWHOFunctionalClassication atthetimeoffollow-up. 2. Aim2 Hypothesis:Preconditioningpulmonaryarterysegmentscompromisespassiveandactivevesselwallmechanicssignicantlyaltering diameter,compliance,vasoactivecontraction,stress,strain,andelasticmoduluscomparedtonon-preconditionedsegments. Approach:Comparepreconditionedandunconditionedpulmonaryarterysegmentsfromratsusingmeasuresofpressureanddiametertoassesstheimpact ofpreconditioninguponmechanicalarteryproperties:diameter,compliance, wall-stress,andmodulus.Presentamethodforachievingrepeatablemechanicaltestingresultswithoutpreconditioning. 3. Aim3 Hypothesis:Wallstressismaintainedataconstantleveldespite markedchangesinpulmonaryarterypressureandwallthicknessduringdevelopmentofanduponrecoveryfromPulmonaryHypertesion. Approach:Examineproximalpulmonaryarterymechanicsthroughaphysiologicrangeofpressuresandtimeincontrol,hypoxicandrecoveryanimalgroups. UsemechanicaltestingtoquantifyPPAmechanicsthroughthedevelopmentof andrecoveryfromPHtodetermineifthemechanicaladaptationsinthePPA existtomaintainthewallstressataconstantlevelinthefaceofchangingvessel wallloads. 2

PAGE 17

Chapter2 2.Background 2.1TheVascularCircuitDesign Thehumanbodyhasapproximately10 13 totalcellscomprisedof200dierent types.Allcellsinthebodyneednutrients,theexchangeofgases,wasteremoval, hormones,temperaturecontrol,damagecontrolanddefense.Bloodisthebodilyuid thatdeliversthesubstancesnecessaryforthosevitalsupportivefunctions.Bloodis transportedthroughavastarrayofbranchingvessels,thevascularsystem,tothe tissuesandcellsofthebody[1]. Theheartpumpsbloodthroughthevascularcircuitcreatingthecardio-vascular system.Theheartmaybethoughtofastwopumpsinserieswhichtogetherwiththe vascularnetworkformaclosedsystemoftwocircuits:thepulmonarycircuitandthe systemiccircuit[60][42][97].Figure2.1showsaschematicoftheheartandthetwo circuits. Thepulmonarycircuitbeginswiththerightventricletherstpumplocatedon theright-sideoftheheart.Therightventriclesendsbloodtothelungstoexchange CO 2 forO 2 .Thebloodthenowsbacktotheheartviathepulmonaryveinsandinto theleftatrium.Thesystemiccircuitsbeginswiththeleftventricle,thesecondpump whichsendstheoxygenatedbloodtoallthetissuesinthebody.Thissystemiccircuit allowsfortheexchangeofgasesbetweenthebloodandcellsofthebodyreturning bloodwithlowerO 2 concentrationandhigherCO 2 concentrationbacktotheright atriumtoreenterthepulmonarycircuit,completingthefullcardio-vascularloop[42]. Bloodmovesthroughthepulmonaryandsystemiccirculationsthroughbranching vascularsystemswithsimilargrossanatomy.Inbothcircuits,theventriclesofthe heartpumpbloodintothearterieswhichcarrybloodawayfromtheheart.The 3

PAGE 18

arteriesbranchformingarterialtreesornetworksofsmallerandsmallerbranches witharteriolesformingthesmallestbranchesofthearterysystem.Arteriolesbranch intocapillarieswhicharethesmallestvesselsandthesiteofgasexchange.Capillaries connectthearterynetworktothevenousnetworkorblood-returnsystem.Onthe venousside,capillariesrejoinormergetoformvenuleswhich,likearterioles,arethe smallestvesselsofthevenoussystem.Venulesmergetoformveinswhichmergeinto largerveinsandreturnthebloodtotheatriachambersoftheheart. Capillaryexchangeofgassesandsolutesbetweenthebloodandthetissueis thecentralpurposeofthebloodcirculation.Therefore,mostoftheactivitiesof thecirculatorysystemarecenteredaroundprovidingthecapillarieswithblood[97]. However,underlyingthissimpliedstatementliestheneedforanincrediblyhigh degreeofprecisionduetothecomplexityofpassingbloodthroughanetworkof approximately60,000milesofcapillarieswhichare5-10micronsindiameteranda singlecellinthickness.Thisincludestheneedforprecisetiming,owrate,pressure, concentrationandqualityofblood[60]. 4

PAGE 19

Figure2.1:TheCardio-PulmonaryCircuits 5

PAGE 20

2.1.1HistoricalPerspective:deMotuCordis In1628WilliamHarvey,anEnglishphysicianand'PhysicianExtraordinary'to KingJamesIandKingCharlesI,publishedExercitatioAnatomicadeMotuCordis etSanguinisinAnimalibusdeMotuCordis.ThetitleisLatinfor"AnAnatomical ExerciseontheMotionoftheHeartandBloodinLivingBeings."Harvey'swork wasbasedonpriorexperimentsandpublications.However,deMotuCordiscreated themostcompletesetofdetaileddescriptionstodatetointroduceandsupportthe ideaofbloodmovingthroughaclosedcircleorcircuitinonedirectionaccording totheone-wayvalvesintheheartandveins.Thiscompilationincludedextensive coverageofmethods,experiments,andreportedresultstointroduceandsupportthis idea[101].DeMotuCordissuggestedthatthecirculationiscomprisedofa"double circuit",fromthehearttothelungspulmonarycirculation,backtotheheart,and outtothemaincirculationsystemiccirculation.Harveywasnotabletodiscernow throughthecapillaries,however,andhistheoryofbloodmovementrefutedanearlier theory,thusitwasconsideredcontroversial.Hishypothesisthatbloodowexistedin aclosedcircuitwasprovenafterhisdeathin1660byMarcelloMalpighi.Malpighi,an Italianphysician,discoveredtheconnectionbetweenthearteryandvenoussystems whenheidentiedcapillaryowinafroglungwiththeuseofamicroscope[101]. TheimpactofdeMotuCordiswentfarbeyondbasicphysiologyasitalsointroducedtheapplicationofquanticationandmechanisticorganizationalstructureto theheartandvascularsystem.Theviewthattheheartcouldbeanalyzedmechanicallyasapumpandthatitsoutputcouldbequantiedas"producing72heartbeats perminuteandthrowing540poundsofbloodeveryhour"wasamajordeparture fromthemysticalandmorereligiousviewsofthetimewhichregardedtheheartas the"seatofthespirit".Theintroductionofthisconceptformsfoundationalaspects ofmoderncardiology. 6

PAGE 21

2.2MeanBloodFlowDependsonPressureandResistance Althoughthecardio-vascularsystemcanbeexplainedasasimpleclosedcircuit, thevaryingcharacteristicsofthevascularnetworksuchaschangesinvesselwall composition,geometryandinteractionswithbloodandsupportedtissuesmakeits functioningexceedinglyelaborate.Centraltotheproperfunctioningofthecardiovascularsystemistheneedforadequateandconsistentperfusion.Thisisaccomplishedthroughthecoordinationofbloodpressureandvascularresistancetomoderatebloodow. 2.2.1BloodFlow Bloodowthroughthecardio-vascularsystemismeasuredbythebloodvolume thatpassesthroughapointinagivenamountoftime.CardiacoutputCOisthe measureofthisvolumefromtheheartandiscommonlymeasuredinlitersperminute. Foratypicaladultatrest,COis5l/min.COisrelatedtothestrokevolumeSVof theheartandheartrateHR.Therelationshipamongthesefactorsisrepresented bytheequation: CO = SV HR .1 [ l=min ]=[ l=beat ] [ beats=min ] Sincebloodowisunidirectionalthroughtheheartandvascularsysteminthe normalanatomyandintheabsenceofdisease,asseeninFigure2.1,COisnecessarilymatchedforboththesystemicandpulmonarycircuits.Inordertosendblood throughthevastsystemiccircuitandthroughthemoreproximalpulmonarycircuit withcloselymatchedCO,thegrossmechanicsofthetwocircuitsmustdiertoaccommodatetheneedsofthetissuessupportedbyeachcircuit.Hence,regulation isessentialtoensurethebloodpressuredoesnotdamagetissue,smallvesselsand capillaries. 7

PAGE 22

Ohm'sLawforelectricalcircuitsmaybeappliedtouidowforvascularcircuits toillustratetheconceptofsimilarratesofbloodowthroughdissimilarcircuits. FromOhm'sLawweknowthatvoltagepotentialisequaltothecurrentandthe resistancetotheowofcurrent: Voltage potential = Current Resistance V = I R .2 ConvertingOhm'sLawfortheowofcurrentinanelectricalcircuitintotheowof uid,andspecicallytotheowofbloodinanarterialcircuit,resultsinanequation relatingpressuregradienttobloodowandresistance: Pressure outlet )]TJ/F19 11.9552 Tf 11.955 0 Td [(Pressure inlet = BloodFlow Resistance P = Q R .3 Thisequationisvalidgiventhefollowingassumptionsregardingtheuidand thearterysegment:theuidmustbeNewtonian,owmustbelaminarandsteady, theremustbeno-slipatthewallboundaries,thearterymustbecylindricalandrigid. AsMilnorpointsoutthreeoftheassumptionsarenotvalid:theowispulsitile, notsteady,andthearteriestaperandmaybeellipticalandaredistensible.Thus, theequationislimitedinitsapplicationandtendstooverestimatebloodow[73]. However,therelationshipsarequalitativelycorrectandtheequationillustrateshow owmaybematchedinverydierentcircuits,bycommensuratechangesinP andR.Consequently,toovercomealargetotalperipheralresistancelikethatof thesystemiccirculation,thereneedstobealargechangeinthepressuregradient. Correspondingly,tomatchowandovercomeasmallertotalperipheralresistancelike thatofthepulmonarycirculation,thereneedstobeasmallerchangeinthepressure gradient. 8

PAGE 23

Thetwopumpsoftheheartareadaptedforthesevariedrequirements.Theleft ventricleoftheheartproduceshighpressuretoovercomethehighertotalresistance inthesystemiccirculation.Inthepulmonarycircuit,thetotalresistanceismuch lowersotherightventricleoftheheartproduceslowerpressure.Theproximalarteriesconvertthehigherpressureandhighlypulsitilebloodow,thatexistsnearthe heart,tolowerpressuresandamoresteadyowbyvirtueoftheirwallstructure anddierentialcompositionalongthetree.Attheotherendofthearterialtree,the arteriolesfacilitatetheregulationofbloodowtothebillionsofcapillariesthrough adjustmentsinlumendiameteralteringbranchresistanceandthusow. 2.2.1.1HistoricalPerspective:Poiseuille'sLaw Inthe1830'sand1840'sJeanLeonardMariePoiseuille,aFrenchphysiologist andphysician,derivedandpublishedtherelationshipoftheowofdistilledwater throughnarrowglasstubesandtheinuencechangingindividualvariables.Poiseuille iteratedthroughhisexperimentsalteringtemperature,tubediameters,bulbpressure, andtubelengthsmakingsuccessivemeasurementswithhighprecision[107].He experimentallyarrivedattheequation: Q = K PD 4 L wherehenotedthattheconstant,K",isafunctionofthetemperatureandthetype ofliquid.Hewentontoreporttheresultsofuidowfordierenttypesofliquid throughtheglasstubessuchasaqueoussaltsolutions,aqueousacidsandbases, mineralwater,teas,wines,spirits,andbovineserum.Therateofowforthevarious uidswascomparedtodistilledwaterindicatinganinterestinthemicrocirculation ofthebody[107].Poiseuillealsorecognizedentranceandexiteectsoftheowin thetubesbutdidnotmentionviscositydirectly. Themodernformoftheequation,Poiseuille'slaw,replaceshisconstant,K",with 128 fordiameteror 8 forradiuswhere istheviscosityoftheuid.Interestingly, 9

PAGE 24

theviscosityfordistilledwaterderivedfromtheconstantK"isaccurateto0 : 1% [107]. Q = r 4 P 8 L .4 Poiseuille'slawisoneoffewequationsderivedfromappliedmechanicsthatis wellknowninthepresentmedicalcommunity[107]. 2.2.2VascularResistance Theforcesinhibitingtheowofbloodthroughanarterysegmentisameasure ofresistance.Vascularresistanceisimpactedbythreefactors:bloodviscosity,vessel lengthandvesselradius.Sincevessellengthisgenerallyconsistentinadultsasis viscosityabsentinjuryordisease,vesselradiusnecessarilyhasthegreatestimpacton thelevelofresistance. Toestimatehowchangesinarteryradiusimpactresistance,Poiseuille'sLawfor owinacylindricaltubeeqn.2.4maybecombinedwiththeOhm'sLawrelationship forowinacircuiteqn.2.3.Theresistancemayberepresentedas R = 8 L r 4 .5 where istheuidviscosity,Listhetubelengthandristheradiusofthetube.This calculationrequirescertainassumptionsbeingmetincludinglaminarow,Newtonian uid,fullydevelopedowproles,andtubegeometry.Whilebloodowdoesnot meetallthoserequirements,thisequationformsthestartingpointforevaluating totalperipheralresistanceandisanadequateapproximation[73]. Givenanarterysegmentofconstantlengthandbloodatconstantviscosity,the relationshipofvascularresistancetovesselradiusbecomes: R / 1 r 4 10

PAGE 25

Theinverseradiustermraisedtothe4thpoweremphasizeshowsmallvesselgeometry changesmayresultinlargechangesinresistancethuspermittingverypreciselocal controlofowthroughparticulararterialbranches. Figure2.2:BloodPressuresthroughtheCardiovascularCirculation 2.2.3BloodPressure Bloodpressureisameasureofthecircumferentialforcethatcirculatingblood exertsonthevesselwalls.Bloodowoccurswhenaforcecausesadierencein pressureoveravesselsegmentfrominlettooutletorapressuregradient.Blood owsacrossapressuregradientfromanareaofhigherpressuretolowerpressure. Thepressuregradientisdirectlyproportionaltotherateofow.Hence,thegreater thepressuregradient,thegreatertherateofowgiventhesamevesselsegmentor vesselgeometryasseenin P = QR .Circulationcanonlyoccuriftheowhas thenecessarypressuregradienttoovercomethetotalperipheralresistancewithinthe system. Akeyelementunderlyingbloodpressureisthecardiaccyclewhichconsistsof twodistinctphases,systoleanddiastole.Thecardiaccyclestartswiththesystolic phasewhereventriclecontractionforcesbloodintothelargeproximalarteriesaorta 11

PAGE 26

orpulmonaryartery.Thehighestbloodpressuresarefoundintheaorta.Duringthe systolicphase,theproximalarteriesmustexpandtoaccommodatethestrokevolume fromtheheart.Thisisfollowedbythediastolicphasewheretheventriclesofthe heartrelaxandrell.Duringthistime,theproximalarteriescontractandcontinue topropelthebloodthroughthevasculartree.Whenbloodpressureandvascular resistancearecoupledwiththeelasticityoftheproximalarteries,thebloodowis convertedfromhighlypulsatiletoamoresteadyowatthecapillaries. Thesystolicphasecomprises1/3ofthecycletimewhilethediastolicphasecomprises2/3.Theprocessmaybedescribedthroughtheconceptofatwo-stagepump, wheretheventriclecontraction,systolicphase,istherststageandthearterialrecoilduringthediastolicphaseisthesecondstage.Meanpulmonaryarterypressure MPAPistheaveragepressureasmeasuredthroughthecardiaccycle.Duetothe cardiaccycletiming,theresultingmeanpulmonaryarterypressureislessthanthe arithmeticmean. Figure2.3:WindkesselRepresentsArterialComplianceandElasticRecoil 12

PAGE 27

2.2.3.1HistoricalPerspective:Windkessel StephenHaleswasaclergymanandaFellowoftheRoyalSocietyworkingin Englandintheearly1700's.HaleswroteVegetableStaticksin1727andHaemastaticksin1733describingplantandanimalphysiologyrespectively.Halesisattributed withmakingtherstmeasurementoftheforceoftheblood"orbloodpressure.In Haemastatickshedescribesthebloodforcefoundwhenopeninganarteryofahorse, insertedabrasspipewhichheconnectedtoatallglasstube.Hedocumentedthat thebloodinthetuberose8feet3inchesabovethehorse'sheart[65].Halesalso determinedthemainperipheralresistancewasinthesmallarteriesbymakingcuts inarteriesandveinsandcomparingthepulsationandhowquicklybloodowedfrom each[68]. StephenHalesdescribedthesmoothing"actionofthegreatarteriesonthepulsatilebloodowfromtheheartasanelasticreservoir.WhenHaemastatickswas translated,thiswasconvertedasWindkessel".WindkesselisaGermantermthat maybetranslatedasairchamber",butwithrespecttothearterialsystemitrefers totheelasticreservoiroftheproximalarteries.Inthe1700'stheGermanreengines werepulledbyhorsesandthewaterwaspumpedbyhand.Muchlikethepumpingof theheart,thehandwaterpumpcreatedapulsatileow.Todiminishthispulsatility, aWindkesselwasattachedinlinewiththepumpandthehoseoutletasseeninFigure 2.3. TheairchamberorWindkesselreceivedthepumpedwaterwhichwouldincrease thepressureinsidetheairchamber.Thepressurizedwatercouldthenbedispensed asamoresteadystreamwithlesspulsatility[89].Thusthereasonthatthedamping ofthepulsepressureduetotheelasticreservoirandrecoiloftheproximalarteries areoftenreferredtoastheWindkesseleectthroughoutmedicalandphysiology literature. 13

PAGE 28

WorkinginGermanyintheearly1900's,OttoFrankwentontodevelopamathematicalmodelconsistingoftworheologicalelements,acapacitorandresistorin parallel.Thecapacitorrepresentsthecapacitanceofthearteriesandtheresistor representsthetotalperipheralresistance.InreferencetoHales'work,Frankcoined theterm"Windkesseleect"whiledescribinghismathematicalmodel.Thistwoelementmodelhelpsillustratetheimportanceoftheconceptthatthetotalloadon theheartincludesbothperipheralresistanceandarterialcompliance[119].Models with3,4ormorerheologicalelementsexaminedincurrentliteraturearebasedon Frank'stwo-elementWindkesselmodel. Figure2.4:VesselWallAnatomy:ArteriesandVeins 2.3ArteryWallAnatomy Arteriesarecomplexstructurescomprisedofthreedistinctlayers:tunicaintima, tunicamedia,andtunicaexternaoradventitia.Eachlayerhasaspecializedfunction andthusaspecializedcompositionofcellularandextracellularelements.Themain 14

PAGE 29

cellularelementsincludevascularsmoothmuscleVSMcellsandendothelialcells. TheextracellularmatrixECMisbroustissueprimarilycomposedofcollagenand elastinbers. 2.3.1Intima Theinnermostlayerofthearteryisthetunicaintimaintima,Latinforinner coat".Theintimaisindirectcontactwithbloodowingthroughthevessellumen. Theintimaiscomposedofasinglelayerofendothelialcells.Surroundingandconnectingtheendothelialcellsisthesubendothelialtissue,athinlayerofconnective tissue.Thesubendothelialtissueisalsoincontactwiththeinternalelasticlamina whichisalayerofelastictissuethoughttoseparatetheintimafromthemedia. 2.3.2Media Thetunicamediamedia,Latinformiddlecoat",liesinthemiddlebetween theintimaandadventitia,ormorespecicallybetweentheinternalelasticlaminaon theintimalsideandtheexternalelasticlaminaontheadventitialsideofthevessle walls.Themediaisasmoothmusclelayerwithinterspersedelastinandcollagen bers.TheVSMisresponsibleforcontractionanddilationoftheartery,vasodilation andvasoconstrictionrespectively.Thismuscularlayeriscontrolledbybothsystemic andlocalfactors,includingthecentralnervoussystem,hormonalcontrolandthe endothelialregulation. 2.3.3Adventitia Thetunicaexternaortunicaadventitiaadventitia,outercoat",istheoutermostlayerofthevessel.Theadventitiaprovidesthevesselwithstrength,structure, anddispensability.Theadventitiaanchorsthevesselandprovidessignalingpathways 15

PAGE 30

Figure2.5:SmoothMuscleCells toandfromthesurroundingtissuesandmayhelppreventoverdistention[105].The adventitiaisprimarilycollagen.Othertissueswithintheadventitiaincludeextracellularmatrix,broblasts,interstitialcells,vasovasorum,nervecellsandelastinbers [105]. 2.3.4ArteryWallComposition 2.3.4.1Endothelium Intheintima,endotheliumisasinglelayerofcellsliningeverybloodvesselprovidingabarrierbetweenthebloodandvessel.Endothelialcellsrespondtoenvironmental changessuchaschemicalstimuliintheformofcirculatinghormonesorotherfactors inthebloodandmechanicalstimulisuchasshearstress.Inresponse,endothelial cellsproducevasoactivemolecules,cellgrowthfactors,adhesionandblood-clotting 16

PAGE 31

molecules.Inaddition,endotheliumpromoteselastinandbroblastsynthesis,and immuneresponse.ThevasoactivemoleculesdirecttheVSMtocontractorrelax, thusprovidinglocalregulationofbloodowthroughparticulararterysegmentsvia vasoconstrictionandvasodilation.VasoactivemoleculessuchasnitricoxideNO andEndothelium-DerivedRelaxationFactorEDRFareproducedbytheendotheliumandactonthesmoothmusclethroughsignalingpathways.Whilethisisnota completelistofallendotheliumactivities,itreinforcesthekeyrolesthissinglelayer ofcellsplayinnumerousarteryfunctions,farbeyondamerebarrierbetweenthe bloodandthearteryaswasoncethought. 2.3.4.2SmoothMuscleCells SmoothmusclecellsSMCarelongspindle-shapedcellsapproximately20 m by200 mwithasinglenucleus.SMCsarearrangedinhelicalsheetsliningmost bloodvesselsandinternalorgans.Theyareregulatedbytheautonomicnervousand endocrinesystemsaswellasvariouschemicalsandareactivatedbystretch[1]. SMCaresonamedbecausetheyarenotstriatedlikeskeletalandcardiacmuscleand, therefore,haveasmoothappearance.Thelackofstriationisduetothemyobril arrangement.Insteadofbeingarrangedinregularparallelpatternsthemyobrils ofmyosinandactincrisscrossthecell,connectingtodensebodiesonthecellwalls causingthecelltobunchuponcontraction. SmoothmusclecontractionisinitiatedbyanincreaseofintracellularCa +2 .IntracellularCa +2 bindswithcalmodulin,acalcium-bindingproteinpresentinallcells. TheCa-calmodulincomplexisabletoattachtoandactivatemyosinlightchainkinase MLCK.TheCa-calmodulin-MLCKcomplexphosphorylatesthemyosinregulatory lightchainofthemyosinlament.Phosphorylationactivatesthemyosinheadallowingtheheadtoattachtotheactinlament.Thephosphorylatedmyosinheadsattach anddetachfromtheactinlamentformingcyclingcrossbridgeswhichgenerateforce andcausecontraction.Ifthemyosinlightchainisdephosphorylatedbymyosinlight 17

PAGE 32

chainphosphotaseMLCPwhileattachedtoanactinlament,thedetachmentfrom theactinlamenthappensmuchmoreslowlythandetachmentfromaphosporylated crossbridge.Thisslowdetachmentoftheunphosphorylatedcrossbridgeistermedthe latchstate.Latchisahallmarkofsmoothmusclecontractionanduniquelydistinguishesitfromskeletalorcardiacmuscle[60].Theformationoflatchbridgesallows thearteriestomaintainforceatalowerenergycostthanstriatedmuscle[60]. Figure2.6:CollagenTripleHelixStructureandAggregation 2.3.4.3Collagen Collagenisthemostabundantproteininthehumanbodyandisthemostcommonbrousproteinintheextracellularmatrix.At25-35%ofallproteininthebody, 18

PAGE 33

collagenprovidesstructure,strengthandisabletoresisttensileforcesthroughoutthe tissues.Althoughmaturecollagensareinsoluble,thecollagenmatrixisnotstatic. Withahalf-lifeof0.5-3months,collagenisabletoberemodeledtomeettheneeds ofchangingtissueenvironments[58]. Procollagens,collagenprecursors,aresynthesizedintracellularly,primarilyby broblastsandmyobroblasts[29].Procollagenmoleculesaresecretedextracellularly. Theendsoftheprocollagenmoleculesarecleavedandassembledintobrilscomposed ofmanyindividualprocollagenmoleculeswitharegularendtoendandsidetoside pattern.Themoleculeswithinthebrilsareconnectedbyandstabilizedwithcovalent crosslinks[57].Itshouldbenotedthatcollagencontainsonly1-4crosslinksperunit whereaselastincontains15-20[115]. Theindividualprocollagenchainsorpolypeptidesaremodiedtofacilitatedevelopmentofthechainsintothecharacteristictriplehelixmolecules.Thetriple helixiscomposedof3polypeptidechainswhichhaverepeatingaminoacidsegments. Thethreechainstypicallyhaveaglysinewithtwootheramino-acidchains.These polypeptidechainsarethereforecommonlyrepresentedasGly-X-Y.TheXandY representanyaminoacidincludingGlysine,andmaybethesameordierent,adding considerablevariabilitytothesuperfamily[1]. Whenarteriesareslack,thecollagenbersappearwavy.Asintraluminalpressureincreases,expandingthearterydiameter,collagenstraightensbecomingaligned circumferentiallycausingtheberstobearsomeoftheload[94].Atphysiologic pressures,lessthan10%ofthecollagenbersarestretchedandengaged[38].At higherpressures,morecollagenbersareengagedandthevesselwallsbecomeless distensible[22].Themechanicalpropertiesofcollagenconsistentlyreporthighelastic moduliontheorderof100-1000MPa,suggestingverystimaterial[22]. Collagenisthemostabundantandmainstructuralproteinwithintheconnective tissue.Thisdirectlyaectsthemechanicalfunctionwithregardtocomplianceand 19

PAGE 34

modulusofthearterywallandvesselgeometryincludingdiameterandthicknessof theartery[59]. Although28collagenproteinsarefoundinthehumanbody,90%oftheseare typeI[58].TypeIcollagenisbrallarabletoformcollagenbersandfoundinthe skin,tendon,bones,lungs,vasculatureandcornea.CollagenTypesIIIandVarealso bral-forming.TypeViscloselyassociatedwithTypeI,buttheydierfromType Iintheirtriplehelicalchainstructureswhichmayallowforfunctionalityrelatedto woundhealing[29].TypeIVisabasementmembranecollagenaddingsupportand mayhavefunctionstohelplteruids. Collagenmakesupapproximately%15ofthedryweightofthehumanlungand isthemajorproteingroup.CollagensTypeIandIIIarethemostabundantinthe lungataratioofapproximately2:1andarefoundco-localizedinthevesselwalls [62]CollagenTypeIistheprimarycollagenfoundintheadventitia[29].Themedia containscollagenataratioofapproximately30%TypeIand70%TypeIII[45].The non-brallarcollagenfoundinbloodvesselsisTypeIVfoundinthesubendothelial tissueoftheintima. Figure2.7:Elastin 20

PAGE 35

2.3.4.4Elastin Elastinisaproteinthatformselasticbersfoundintheconnectivetissueofthe lungs,skin,bladder,elasticligaments,elasticcartilageandbloodvesselsinthebody. Elasticbersaremadeofanetworkofapproximately90%elastinand10%brillin. Elastinandbrillinareproducedbybroblasts.Elastinisproducedprimarilyin theperinatalphaseofdevelopment,synthesizedbyembryonicorjuvenilebroblasts. Afterthisstage,productionofelastindramaticallydecreases.Thehalflifeofelastin isapproximately40yearsorthelifespanoftheorganismmakingelastinverystable withlessthan1%turn-overperyearinadulthumans. Elastin'sextendedlifespanandinsolubilityaredirectlyrelatedtoitsextensive -20crosslinksperunit[115].Withinthishighlycrosslinkedstructure,relaxed elastinbersformcompactandcontractedarrangements,appearingsomewhatoverlapped,coiledanddisordered.Thisrelaxedformationisalowenergystate.Elastin bersmaystretchto1.5timestheircontractedlength.Whenstretched,thebers elongateandformsheetsoforderlymoleculesconnectedthroughalargenetworkof interspersedcrosslinks.Stretchedelastinisinahighenergystate,andisableto passivelyrecoilbacktothecontractedlow-energystate. Elastinhasalowelasticmodulusofapproximately0.4MPa.meaningthatit willresistverylittleofthestressandstrainplacedonthearterialwall.Thequantity ofelastininthearteriesisgreatestproximallytotheheartanddecreasesalongthe arterialtreetothearterioleswhereitismuchlessabundant.Thesediminishinglevels ofelastinalongthearterialtreeunderscoreselastin'sprimaryroleofprovidingthe arterywithdispensabilityandelasticrecoiltoassistinreducingpulsatilityasblood owstowardthecapillaries[22]. Whiletheroleofelastinislimited,itisessential.Specically,elastinalsoparticipatesintheregulationofsmoothmusclecellproduction.Whenelastinisnotpresent, smoothmusclecellscontinueproductionuntilthearteryiscompletedblocked.[115]. 21

PAGE 36

2.4RegulationofBloodVessels Cardiovascularregulationensuresbloodperfusionandowtomeettheneedsof proximalanddistaltissues.Bloodowmustbeoptimizedforgasexchange,delivery ofnutrients,removalofwasteandpHbalancetosupportapproximately10billion capillariesintheadulthumanbody[60].Therefore,regulationofthecardiovascular circulationmustbeexquisitelycontrolledintimeandspace.Changesinbloodow mustoccurintimescalesvaryingfromfractionsofsecondstodaysoryears.In addition,thesechangestotheowofbloodmustbemadebothlocallywithinspecic tissuesaswellassystemically. Theregulationofbloodowmaybethoughtofasafeedbackloopofcontrollers, sensorsandactorsworkinginconcerttokeepbloodpressureatahomeostaticset point,whichinhumanadultsisameanarterialpressureof90mmHg[42].Regulation ofthecardiovascularsysteminvolveschangesintheCO,pressureandperipheral resistance.Threemechanismsallowthosechangestobemade:autoregulation,neural regulationandhormonalregulation. Thesethreemechanismsactonbothashortandlongtermbasis.Shortterm regulationincludes:1.IntrinsicBaroreceptorReex,ChemoReceptorReex,CardiopulmonaryReex,Heart&VesselAutoRegulation;2.ExtrinsicReexesPain, Exercise,Anger/Fears/Anxiety,andcooling/thermal;and3.Hormones.LongTerm Regulationincludes:1.Changesinbloodvolume;and2.Changesinbloodvessel quantityandsize[42][67]. 2.5HypoxicVasoconstriction Hypoxicvasoconstrictionisoneoftheprincipleregulatorsofbloodowrespondingacutelyandchronicallyshortandlongtermregulation.Hypoxiaisacondition oflowoxygenlevelsinthebodyorinspecicareasofthebody.Hypoxiaisthe mostpowerfulstimulantoftheperipheralchemoreceptors[42].Hypoxicpulmonary 22

PAGE 37

vasoconstrictionHPVisaphysiologicprocesswherearteriesinthelungcontractin responsetohypoxiainthealveoli. Asyet,thereisnoconsensusonthepreciseoxygensensororsensorswhichtrigger HPV,thoughanumberofsignalingsystemshavebeensuggestedincluding:heme proteins,ionchannelconductance,andamitochondriainitiateddecreaseinreactive O 2 species[19].Currentstudiessuggestoneofthekeymechanismslieswithinthe pulmonaryvascularsmoothmuscle[75][108].WithlowpO 2 sensed",potassiumion channelsbecomeblockedorinhibited,causingthecellmembranetodepolarize[117]. Uponmembranedepolarization,thevoltage-gatedcalciumchannelsareactivated allowinganinuxofCa ++ .Atthesametimecalciumisalsoreleasedintracellularly fromthesarcoplasmicreticulum. ThetotalincreaseintheconcentrationofintracellularCa ++ initiatescontractionofthesmoothmuscleandvasoconstrictionoftheartery.Specically,thesmall resistancearteriesandarteriolescontract,decreasingdiameter,increasingresistance tobloodow,thusdecreasingbloodowthroughthepoorlyoxygenatedcapillaries. Theincreasedresistanceinthesebranchesencouragesbloodowthroughbranches withlessresistance,thusdivertingbloodowtowardcapillarieswithbetterpO 2 perfusion.ThisprocessisalsoknownastheEuler-Liljestrandmechanismnamedafter UlfvonEulerandGoranLiljestrandwhorstdescribedtheprocess[27].Oncethis occurs,astateoflocalizedhypertensionexists. 2.6PulmonaryHypertension PulmonaryHypertensionisdenedasameanpulmonaryarterialpressureabove 25mmHgatrest[44].RHCisconventionallyusedtomeasurepressuresinthePA andtodiagnosePH[35].Insomesense,PHcanbethoughtofassimilartogeneralizedhypoxicvasoconstrictionaectingthewholepulmonarybedwhichdramatically increasespulmonaryvascularresistancePVR.IncreasedPVRmakestherightven23

PAGE 38

tricleoftheheartworkhardertoforcethebloodthroughthenarrowedvascular tree.Thus,PVRisoftenthesingleparameterusedtoassessdiseaseseverityand responsetotreatment.Nevertheless,othermarkersofvascularfunction,forexample pulmonaryvascularstinessPVS,havebeenshownaspredictiveofmortalityin PH[51][36][74]. TheWorldHealthOrganizationclinicallyclassiesPHintovecategoriesofdisordersbycausethatsharesimilarcharacteristics,pathogenesisortherapeuticmanagement[98].Thegroupsinclude:PulmonaryarterialhypertensionGroup1,PulmonaryhypertensionduetoleftheartdiseaseGroup2,Pulmonaryhypertension duetolungdiseasesand/orhypoxiaGroup3,Chronicthromboembolicpulmonary hypertensionGroup4,andPulmonaryhypertensionwithunclearmultifactorial mechanismsGroup5[98]. ThedenitionanddiagnosisofPHcontinuetoevolve.Forexample,ithasbeen suggestedthatPHberedenedtoincludearestingmPAPof20insteadof25mmHg. AnotheroptionwouldbetocharacterizemPAPfrom21to24mmHgasborderline PH".Previously,thedenitionincludedexerciseinducedPHwhichcouldbereintroducedaspartofthediagnostictools.GiventhecommonuseofPVR,itisapossible factortobeincludedinthedenitionaswell.Thereisalsotheoptiontoinclude pulmonaryarterywedgepressurePAWPof15mmHginthestandardprotocolto distinguishbetweenpre-capillaryandpost-capillaryPH[44]. DespitetherapidadvancesinPHtherapies,theprogressivenatureofthedisease eventuallyresultsinrightheartfailureanddeath.Consequently,medicaltherapies focusonrelievingsymptomsandhelpingtheheartpump,providingoxygenandreducingedema.Medicationsthatmodulatevasculartoneareoneofthemainpharmacologicalagentswiththegoalofimprovedhemodynamicproles[3][6].Vasodilators slowtheprogressionofthediseaseandimprovethequalityoflifeforpatientswith PH[42].ThesemedicationsacttorelaxVSMcontractionofthebloodvesselsby 24

PAGE 39

regulatingvariousintracellularpathways:reducingintracellularcalcium,decreasing thephosphorylationofmyosinlightchain,oralteringintercellularpotassiumorATP levels.ThecommonthemeamongthesemedicationsistheregulationofVSMforthe normalizationofthecardiovascularcirculationtopreventrightheartdysfunction[6] [95]. 2.7LargeArteriesinPulmonaryHypertension ChronicpulmonaryhypertensionPHincreasespulmonaryarterystinessand decreasesdispensabilitythroughoutthearterialtreeandinducingchangesandremodelingwithinthevesselwall.Increasesinvascularcollagendepositionaswellas increasedwallthicknessarewellestablishedandthoughttobetheprimarycauseof vesselstiening.HowevertheprocessofPHcausesmanysignicantchangeswithin thevasculature. 2.7.1ExtracellularResponse ExtracellularMatrix DuringthedevelopmentofPH,extracellularmatrixdepositionincreases.In proximalarteries,SMCandbroblastsincreasecollagenandelastinproductionin themediaandadventitia.Inthedistalarteries,theendothelialcellsincreasethe productionofTypeIVCollagenaswellaselastin[105]. 2.7.2CellularResponse EndothelialCell Withintheintima,studiesshowincreasesinendothelialcellsasanacuteresponse inthePPAandasachronicresponseinthedistalarteries.Theincreaseinthenumber ofendotheliacellsisnotlikelytoincreasethestinessofthearterywalls.Whether 25

PAGE 40

thischangeisdysfunctionaloradaptive,isstillunderdebate.Ineithercase,the quantityofendotheliacellsdoesnotappeartodirectlychangethemechanicsbut mayindirectlyimpactarterymechanicsthroughincreasedsignalingtoothertissues andcellswithinthearteries. PHinducesanumberofchangesinendotheliumfunction.Someofthesechanges include:theincreaseininammatoryresponse,theproductionofadhesionmolecules, theexpressionofcoagulationfactorsandincreasedcoagulation,andanincreasein endothelialcellpermeability.Thedecreasedendothelialbarrierallowsfactorsinthe bloodtobeindirectcontactwiththesubendotheliallayer,possiblyleadingtoincreasesincellproliferationofthemediaandadeventitia[105].Theendotheliumalso producesmorevasoconstrictingfactorsandreducestheproductionofvasodiliating factors. SmoothMuscleCells InPH,SMCsincreaseinsizehypertrophyandnumberhyperplasiawhileincreasingproductionofcollagen.Hypertrophyismoreprevalentintheproximalvessels whilehyperplasiaismoreprevalentinthedistalvessels.Inthemicrovasculature,broblastsarerecruitedfromtheavelolarwallstosynthesizeSMCsinareasofthe microvasculaturewheretheydonotnormallyexist.Inthiscontext,itisimportantto rememberthattherearedierentsub-populationsofSMCsthatresponddierently toenvironmentalstimuli[103].ThusitispossiblethatnotallSMCswillrespond equallytoPH. Fibroblasts MuchliketheSMC,therearedierentgroupsofbroblastswhichrespondtoenvironmentalstimulidierently.Throughoutthearterialtree,PHinducesanincrease insomebroblastpopulationstoinitiateincreasedproductionofcollagen.Inadditiononesubgroupintheadventitiachangesphenotypicallytomyobroblastswhich produce -SMactinduringPH.Ofinterest,myobroblastactivityisalsoknownto 26

PAGE 41

bepresentinresponsetoinjury,woundhealingandscarcontraction.Inthedistal arteries,theincreasedbroblastsleadtoanincreaseofcollagenTypeIandelastin, diminishingthearteryinnerdiameter. InammatoryCells Inammatorycellsproduceacomplexweboffactors,includinggrowthfactors, cytokinesandreactiveoxygenspecies.Thesefactorsinteractinavastsignaling matrixtoregulatecellgrowth,immuneresponseandaptosis.Inammatorycells arecorrelatedwiththeremodelingthatischaracteristicofPH.Thisissupportedby theincreasednumbersoflymphocytesandmacrophagespresentinthearterialwall. Inammatorycellsarealsosuggestedasapotentialimpactonendothelialdysfunction andincreasedsympatheticactivity. Figure2.8: In-Vitro Testing-Systemschematicdiagrams 2.8EvaluationofArterialMechanics Asweknow,themostinterestingandthemostdicultmechanicalproperties tomeasurearetheactivepartofthevascularsmoothmuscles[48].Givenallofthe abovefactorsandtheirinterwovenimpactsonarterialcharacteristicsandfunction, 27

PAGE 42

predictingpotentialchangesinmechanicsovertimeismuchmorecomplexthanit mayinitiallyappear. 2.8.1 In-Vivo While in-vivo testsallowexaminationofthelivingvessels.Theyareextremely invasiveandarefurtherlimitedbyphysiologicalcontrolfromthebodyincluding hormones,thecentralnervoussystem,andauto-regulationbecause,inuencesfrom thebody'sregulatorysystemsareoftennotcontroledormeasuredbythetesting system.However,many in-vivo studiesprovidethebasisforarterydiameter,stress, strainandmodulusdataandcomparisons[46][32][68]. 2.8.1.1RightHeartCatheterization RightheartcatheterizationRHCisperformedtogatherhemodynamicmeasurementswithintheheartandanumberofmeasurementsfromwithintheproximal pulmonaryarterytodiagnoseandassessPH.Acatheterisinsertedthroughthevenoussystemintotherightsideofthehearttakingmeasurementsofpressureinthe rightatriumandrightventricleasitispassedintothemainpulmonaryartery.PressureisalsomeasuredinthePA.Measurementsofpressuresfortheleftsideofthe heartaremadeindirectlybyinatingaballoonattheendofthecatheterinthe pulmonaryartery.Theresultingpressureispulmonaryarteryocclusionpressureor pulmonarycapillarywedge"pressurePCWPandapproximatesleftatrialpressure oftheheart.RHCmeasurementsareutilizedasindicatorsofheartfunctionand possibleventricularfailure.[4][69]. Non-invasivePPAmeasurementsarealsoavailable.Ultrasound,echocardiology orecho-doppler,CO 2 rebreathing,bioimpedanceandmagneticresonanceimagingare beingconsideredasoptionstoestimategeometry,ow,pressureandCO.Studiesare currentlyexaminingtheaccuracyofthesenon-invasiveorless-invasivemeasurements andcalculationscomparedtoRHCdata.Despitetheappealofnon-invasiveoptions, 28

PAGE 43

RHCremainsthestandardtodiagnose,conrmandassessPHasitconsistentlyprovidesthemostaccuratemeasurementsofpressurefromwithintheheartandproximal arteries[113]. 2.8.2 In-Vitro Themostcommonlyperformed in-vitro testsarestressandstrain.Thesearethe mostcommon in-vitro testsperformed.Inordertobesuccessful,thistestingmust becompletedwhilethetissueissubmeregedinaphysiologicsolutiontoensurethat thetissueremainsviable.Itshouldbenotedthateachofthe in-vitro testsmustrely oncertainassumptionsregardingconstanttissuevolume. 2.8.2.1UnixaialTest Uniaxialtestingisoneofthemostcommon in-vitro methodsoftestingsofttissue tomeasurestressandstrain.Uniaxialtestsareperformedonrectangularstripsoftissuethatarestretchedinonedirectionastheyareheldinplacewithclampsorhooks. Thetissuecanbestretchedtoapproximateeithercircumferentialorlongitudinaldisplacement.Duringthisprocess,aloadcellappliesaforceandthedisplacementofthe tissueismeasuredthroughthemovementofthehookorclamp.Theresultingdata arepairsofforceanddisplacement.Theforceanddisplacementpairsareusedalong withmeasurementsforcrosssectionalareasofunloadedtissuetocalculatestressand strain: Stress = Force Area .6 Strain = L )]TJ/F19 11.9552 Tf 11.955 0 Td [(L o L o .7 TheresultstypicallyformaJshapedcurveillustratingtheinitialactivityofthe elastin,thebaseoftheJshowingatransitionandsubsequentlyshowingtheinitiation 29

PAGE 44

ofthecollagenuptotissuestrainlimits.Theslopeofthiscurvewilldepictthedegree ofstinessineachregion,knownastheincrementalmodulus.Whilethistesting providesvaluableinformationregardinguni-directionalstressandstrain,itdoesnot accountforthe3-dimensionalnatureofthetissueasitinteractscircumferentially, longitudinallyandradially in-vivo withintraluminalpressures. 2.8.2.2RingTest Ringtestingissimilartouniaxialtestingexceptthatasmallsegmentofartery iscutcircumferentially.Thisringismountedonhooksandpulleduni-directionally. Theringtestagainproducespaireddataofforceanddisplacementmakingthestrain calculationsimilartothatoftheuniaxial.However,thestresscalculationincludesa factorof 1 2 duetothetwosidesofthering: Stress = 1 2 Force Area .8 Theringtestlimitationsaresimilartotheuniaxialtestinthatitalsofailstoaccount forthe3-dimensionalnatureoftheartery in-vivo 2.8.2.3BiaxialTest Thebiaxialtestusesasquareorrectangulartissuesamplewhichisstretchedin twodirections.Thus,allowingdatapairsofforceanddisplacementtobecollected inboththecircumferentialandthelongitudinaldirections.Stressandstrainmay becalculatedasbeforemakingsurethattheareasreectthecrosssectionofthe directionofstretchasillustratedin2.8C. 2.8.2.4PressureInationTest Thepressureinationapparatusisusedfortestingwholelongitudinalsegmentsof anarteryinitsoriginalshape.Anarterysegmentisplacedbetweenthetwocannulas andsubmergedinaphysiologicsolution.Ateachendofthearteryisapressure transducerconnectedtoacannula.Theperfusatepumpsendsthephysiological 30

PAGE 45

Figure2.9:PressureInationTestingSystemSchematic solutionthroughthecannulaintothearterysegment.Amicroscopecapturesthe diameterofthearteryastheintraluminalpressureincreases.Simultaneously,the pressureateachendofthecannulaismeasuredwiththepressuretransducersto estimatetheintraluminalpressureofthetissue.Witheachchangeinpressure,anew datapairisrecorded.Intraluminalpressureiscalculatedbytheformula: P in = P 1 )]TJ/F19 11.9552 Tf 11.955 0 Td [(P 2 2 .9 Theadvantageofhtistestisitsmeasureofbothradialandcircumferentialstress. Whileitdoesnotaccountforshearforcesfrombloodoworlongitudinalstretch,it muchmorecloselymimics in-vivo environmentsthanother in-vitro tests. 2.8.2.5Pre-Conditioning Pre-conditioningbiologicaltissueistheprocessofcyclicallyloadingandunloading thetissuesamplespriortotakingtestmeasurements.Somestudiesconsiderpre31

PAGE 46

conditioningnecessarytoobtainreproduciblestress-straincurves[32][46].During pre-conditioning,tissuesmaybemountedasdescribedinanyoftheabovemethods: uniaxial,ring,biaxialorpressure-ination.Thetissuesarethenstretchedorinated accordingtoYoung'sprotocol;approximately10timesat1Hzpriortotesting[]. 2.9RecoveryStudies Previousstudiesofrathypoxicrecoverymodelssuggestarangeofresultsfrom completerecoverytopersistentchangesrelatedtoPHremodeling.Poiani'sstudyof 10dayshypoxicexposureat10%O 2 andpressuresbetween74-80mmHgandupto 10daysrecoveryinroomairfoundcollagenandelastincontentsreturnedtonormal [87].Meyrick'sresultsof380torrhypoxicexposurefor10daysshowedtheadventia doubledinthicknesswithincreasedbroblastsandcollagenbers.Uponrecovery inroomairforupto70days,theadventitiathicknessreturnedtonormalbutthe broblastsweresmallerwhilethecollagenberconcentrationremainedincreased [72].HislopandReidfoundwithexposureof2weeksat380mmHgandrecoveryup to8weeksinroomairthatrightventricularhypertrophyresolved,butthelossof smallarteriesarterialpruningwithouterdiameter=200mdidnotrecoverfrom thehypoxicexposure[43].With10%O 2 hypoxicexposureupto3weeksandrecovery inroomairupto20weeks,Hergetfoundthatmuscularizationandarterialpruning persistedwhilemPAPreturnedtonormal[41].Finally,with5weeksofexposureto 380mmHgandupto5weeksofrecovery,Heathfoundnormalizationofbothmedial thicknessandRVhypertrophy[40].Thisbaseofinformationopensanopportunity forasinglestudy,acrosstimeandpressures,tocreateasinglesetofdatatoalign thendingsintoacomprehensivebaseofinformation. 32

PAGE 47

2.10StudiesinthisThesis Theworkinthisdissertationaimstobetterquantifytheimpactpulmonaryhypertensionhasonthemechanicsoftheproximalpulmonaryarteries.Specically,the aimistoassesschangesinthePPAmechanicsinPHonset,recoveryortreatment. Toourknowledge,thisisthemostinclusiveassessmentofPPAmechanicsupon developmentandrecoveryfromPHinananimalmodel.Further,thisistherst comprehensivestatisticalevaluationofavailableclinicalmeasurestopredictfuture WHOFC. Thethesisresultsaresupportedbyahumanstudyandtwoanimalstudies.The clinicaldatafromthehumanstudyoriginatesfrompediatricrightheartcatheterizationreportsusedtoconrmthediagnosisofpulmonaryarterialhypertensionaswell asassociatednon-invasivemeasures.Theaimoftheclinicalstudywastoprovidethe foundationforamodeltopredictfuturepatientstatususingclinicalmeasures.The animalstudiesfocusedoninvasivemechanicalPPAmeasuresthataretraditionally unavailableinhumanstudies.Theaimofthisstudywastoevaluatethechangesin PPAmechanicsacrossafullrangeofphysiologicalpressuresinPHonsetandrecoverytoquantifythefunctionofthepassiveelementsofthearterialwallwithinthe boundariesofsmoothmuscleactivity.Theultimategoalofthisstudywastoprovide insightstosupportpotentialstrategiesforfuturemedicaltreatments. 33

PAGE 48

Chapter3 3.Study:MultivariatePredictionofClinicalIndicators 3.1Abstract Background:PulmonaryhypertensionPHisanincurable,progressivedisorder withpoorsurvivalinchildren.Conventionally,cardiaccatheterizationisusedfor diagnosisandserialmonitoringinchildren.Signicantdisparityremainsregarding whichclinicalfactorsaremostpredictiveofoutcome.Thegoalofthisstudywasto evaluatemultipleclinicalfactors,incombination,tohelpguideclinicalinterventions andpredictoutcomes. Methods:Thisstudyincludes78patientsevaluatedforPHinthecardiac catheterizationlaboratoryatTheChildrensHospitalColoradoandassociatedhemodynamicmeasurementsandimagingoftheproximalpulmonaryartery.Weused amulti-covariateproportionaloddslogisticregressiontoexplorepredictionofpatientoutcomerepresentedbytheWorldHealthOrganizationFunctionalClassicationWHO-FCatfollow-up.WewereinterestedinprovidingpredictionofWHO-FC in1yearandtoprovideobjectivemeasuresofdiseaseprogression. Results:Ourndingssupportthecombinationofmeanpulmonaryarterialpressure,systolicbloodpressure,systolicproximalarterydiameterandtheratioofmean pulmonaryarterialpressuretomeanaorticpressureasthemodelcombinationthat predictsfutureWHO-FCwiththehighestaccuracy.Systolicbloodpressurewasidentiedasthemostaccuratesinglevariablepredictor.However,thesinglemeasures aresignicantlylessaccuratethanthemultivariateequation,thereforelimitingtheir useinaccuratelyquantifyingfutureWHO-FC. Conclusions:Singlevariablepredictorswhencombinedprovideincrementalimprovementthatgreatlysurpassedtheirindividualabilitytopredictthefuturehealth 34

PAGE 49

statusofpatients.Specically,someofthenewernon-invasivemechanicalmeasures whencombinedwithclassicRHCmeasurementssignicantlyincreasedprediction accuracy. 3.2Introduction PulmonaryhypertensionPHisadisorderinvolvingprimarilythearterialcomponentofthepulmonaryvasculartree[99].PHisdenedasanincreaseinmean pulmonaryarterialpressureMPAPabove25mmHgatrest[7][56][92].Beneath thisdenitionisacomplexarrayofsymptoms,multiplepathogenicpathways,multipleetiologiesandassociateddiseases[4][35][70].PHisconventionallydiagnosedby rightheartcatheterizationRHC;aninvasiveprocedurethatcarriesanumberof risks[4][35][69].DespitetherapidadvancesinPHtherapies,theprogressivenature ofthediseaseeventuallyresultsinrightheartfailureanddeath[3][30][34][69][70]. PHischaracterizedbyvasoconstriction,remodeling,andthickeningofthesmall tomediumarteriesandarterioles,resultinginincreasedPVR[36][103].PVRisoftenthesingleparameterusedtoassessdiseaseseverityandresponsetotreatment [7][24][121].Yet,othermarkersofvascularfunction,forexamplepulmonaryvascular stinessPVSareastrainandvascularcapacitancehavebeenshownaspredictiveof mortalityinPH[15][36][51][74].Numerouspriorstudieshavesuggestedutilizationof variouscombinationsofindividualclinicalindicatorsaspredictiveofsurvival,mortalityandoutcomesforPH[10][11][12][23][109].However,thereisnocomprehensive evaluationshowingincrementalincreasesinpredictivevaluewhenavastarrayof individualindicatorsiscombined. Giventhebroadbaseofavailableclinicalindicatorsandassociateddata,there isareadyopportunitytoevaluatethecomplimentarypredictivevalueofpediatric indicatorsinPH.ThisisespeciallybenecialbecausepediatricPHlacksitsown comprehensiveevaluativefactors,relyingheavilyonadultdataforreference.Due 35

PAGE 50

totheimportantdierencesinpediatricPHetiology,symptoms,andpresentation, pediatricdiseaseisuniqueandrequirescustomizedpredictiveclinicalindicators[8] [9][23][120][121].Withthisaiminmind,weused4statisticaltoolstoexplorenumerouscombinationsofhemodynamicmeasurementsaswellasthegeometry,ow,and pressurechangesintheproximalpulmonaryarterytoidentifyapredictivetoolfor pediatricPHoutcomes.Forthepurposesofthisstudy,outcomeisdenedasWHO FunctionalClassicationWHO-FCatthetimeofpostRHCfollow-up.Wehypothesizethatmeasuresofproximalpulmonaryarteryfunction,includingrightheart catheterizationRHCandColorM-ModeTissueDopplerImagesCMM-TDI,increasetheaccuracyofpatientoutcomepredictionswhenutilizingthecombinations ofmeasuresfromconrmedcasesofpediatricPH. 3.3Methods 3.3.1StudyDesign ThestudywascarriedoutwiththeapprovaloftheColoradoMultipleInstitutionalReviewBoardCOMIRB.Allpatientsortheirguardiansconsentedafter beinginformedoftherisksandbenetsoftheproceduresandprospectivelygranted permissionfortheutilizationofgathereddataforstudy. 3.3.2PatientPopulation WereviewedthemedicalrecordsforanypatientswhounderwentRHCandacute vascularreactivitytestingatChildren'sHospitalColoradofromAugust2003toJuly 2010.ConsecutivepatientdatawasincludedforthosediagnosedwithPHbeforethe ageof18yearsoldbyrightheartcatheterizationRHC.WHO-FCscorewasobtained frommedicalrecords.PatientswerereevaluatedandassignedafollowupWHO-FC scoreswithin2yearsofinitialRHC-26months.Hemodynamicanddemographic 36

PAGE 51

baselinedatacollectedatthetimeofinitialRHCareshowninTable3.1.Forinclusion inthestudypatientrecordsneededtoinclude:PApressures,tissuedopplerimages suitableforPAdiameteranalysis,WHO-FCatdiagnosisandfollow-up. Allpatientsweretreatedwithstandardtherapies. 3.3.3ClinicalDataCollection AllpatientdataincludedaRHCperformedat21%FiO 2 andunderanesthesia asisroutinefortheinstitution.PressuremeasurementsofmeanpulmonaryarterialpressureMPAP,meanrightatrialpressuremRAP,andpulmonarycapillary PCWPpressurePCWPwerecontinuouslyrecordedusingaSwan-Ganzcatheter TranspacIV,AbbottCriticalCareSystems,AbbottPark,IL.Asystemicarterial monitoringlinerecordedthemeansystemicbloodpressure.CardiacoutputCO wascalculatedviatheFickequation.IndexedPVRPVRIwascalculatedasthe dierencebetweenMPAPandPCWP,dividedbythecardiacindex.Theratioof pulmonaryvascularresistancetosystemicresistance PVR SVR ,aswellacalculationof strokevolumetopulsepressure SV PP ,wereusedasindicatorsofcompliance. ColorM-ModeTissueDopplerImageCMM-TDIoftherightpulmonaryartery RPAwallwerealsocollectedattheendoftheRHC.ImagingoftheRPAwallwere obtainedfromthesuprasternalshort-axisview,providingalongaxisrepresentation oftheRPAperpendiculartotheultrasoundbeamangle[25][50].Priortomeasurement,theultrasoundbeamwassweptthroughthelongaxisoftheRPAtolocate maximaldiameter.Acquisitionofthediametermeasurementsoccurredalongthe beamlineasshowninFigure3.1a.AcustomsoftwarepackagebasedonEchoMAT v.2.1,GEMedicalSystemsInc.detectedthelumenwalledges,shownonFigure 3.1b,andcontinuouslyrecordedthepressureandechocardiographtracesshownin Figure3.1c.Basedupontheultrasoundresults,thesystolicdiametersDiamofthe RPA,thedierencebetweenthesystolicdiameterandthediastolicdiameterofthe 37

PAGE 52

rightpulmonaryarterydiameter D ,andthemaximumdiameterindexedbyBSA sDiam BSA werecalculated.PulsepressurePPwascalculatedasthedierencebetween systemicanddiastolicpressure.Strainwascalculatedasthedierencebetweenthe systolicanddiastolicRPAdiametersandnormalizedbythediastolicRPAdiameter sDiam )]TJ/F20 7.9701 Tf 6.586 0 Td [(dDiam dDiam .DynamiccomplianceCdynwascalculatedasthenormalizedvolume changegivenanappliedpressure;thatis Strain PP 3.3.4StatisticalAnalysis StatisticalanalysiswasperformedusingRCoreDevelopmentRversion2.14.1, 2011[90].Whereappropriate,statisticalsignicancewasdeterminedas p < 0.05. Appendix2comprisesthecodeforallstatisticalanalyses,calculationsandderived measurements. AkaikeInformationCriterionAICwasusedtoascertaintheaccuracyofvariablesinpredictingfuturepatientWHO-FC[112].AICissimilartoalikelihood ratiotestinthatitprovidesameasureofthe`goodnessoft'ofdierentcalculation methodsusingidenticaldata.AICincludesapenaltyterm"toverifyasuitablet Appendix2forcodeimplementation[16].Giventhesmallcohortofthisstudyand therelativelylargeamountofdatacollectedforeachpatient,theRandomForest methodwasalsousedtoverifypredictionaccuracyofindividualvariablesAppendix 2:codeimplementation[106]. Proportionaloddslogisticregressionwasasusedwithsingleandcombinedclinical indicatorstopredictpatientWHO-FCscoresuponfollowup.Weused3categorical levelsforpatientoutcome:FC-I,FC-II,andacombinedclassicationofFCIII-IV. FC-IIIandFC-IVwerecombinedastherewasonly1patientassessedatFC-IV and10atFC-III.Toevaluatethepredictiveaccuracyofeachsetofindicatorsboth individuallyandinaggregate,weusedaleave-one-outtraining-testing. 38

PAGE 53

WeperformedaCoxproportionalhazardssurvivalanalysisofthetimeuntila PH-relatedeventoverthecourseofthisstudy,40months.InadditiontoWHO-FC, wealsodenedvepatientevents:1.HospitalizationduetoPH;2.Worsening ofWHOFunctionClasswiththeexceptionofchangingfromclassItoclassII; 3.StartofIVtherapy;4.Lungtransplant;5.Death.Thetimetoeventwas assessedwithbothunivariateandmultivariateanalyses[106].Wereducedthenumber ofvariablesunderconsiderationusingtheRandomSurvivalForestmethods.The RandomSurvivalForestmethodsarepatternedafterRandomForestmodelingand havesimilaraccuracypredictionforcomplexdatawithnon-lineareectsandhigh orderedinteractions[55].Theseadditionalpatienteventswereevaluatedinorder todeterminewhethervariablesidentiedforWHO-FCcouldbeutilizedtopredict alternativeoutcomes. 3.4Results Atotalof78patientswereincludedinthisstudyfromaninitialbaseofmore than200patients.Themeanagewas8 : 98 6 : 08yearsofageatthetimeofdiagnosis. Fifty-onepercentofthepatientswerefemale.Themeantimefromdiagnosistofollowupwas11 5 : 4months.Thediagnosesnotedforthepatientswere:19idiopathic PH,5FamilialPH,3associatedwithcongenitalheartfailure,17associaatedwith AtrialSeptalDefect,13associatedwithAtrioventricularseptaldefect,3associated withhighaltitudepulmonaryedema,10associatedwithpatentductusarteriosus,2 associatedwithlungdisease,1associatedwithpulmonarycapillaritis,2associated withantiphospholipidantibodysyndromeand1associatedwithOverlapsyndrome. Therewere3deathsandnolungtransplantationsduringthetimeofthestudy.These dataaresummarizedinTable3.1. 39

PAGE 54

3.4.1SingleVariableMeasuresasPredictorsofOutcome Thedatacontainedanumberofpatientmeasurementsthatwerestrongly alignedorcorrelated.Therewerepositivecorrelationsbetweenthefollowingmeasures:MPAPandPVRI r= 0.09, p < 0.05;PVR/SVRandPVRI r= 0.11, p < 0.05; PVR/SVRandMPAP r =0.12, p < 0.05;MPAPandPP r= 0.16, p < 0.05;Dand sDiam r= 0.26, p < 0.05;PPandPVRI r= 0.27, p < 0.05;andPVR/SVRandPP r= 0.29, p < 0.05.Typically,weremovemeasurementswithstrongcorrelationsfrom furtherstatisticalanalysis.However,thisstudycallsfortheexaminationofabroad baseofavailableclinicalindicatorsandassociateddata.Therefore,wekepthighly correlatedvariablesinthemodelstoensureacomprehensiveevaluationofpotential complimentarypredictivevalueswhenthesemeasuresweregroupedwithotherpediatricindicators.Likewise,wekeptvaluesofstrainandcompliancedespitetheir relationtodiameterandpressure. TheunivariateanalysisofAICweightsshowninFigure3.2asuggeststhatSBP, MPAP/AoP,MPAP,andPVR/SVRarethebestprognosticindicatorsforpatient outcomeprediction.TheRandomForestestimationofvariableimportancesuggests thatthevariablesSBP,MPAP/AoP,MPAP,PVR/SVR,PP,Cdyn,andsDiamrespectivelymayindividuallydescribethemajorityofthevarianceofpatientoutcome. InFigure3.2ballindicatorsarearrangedinvariableimportancefrommostimportant toleast. Finally,weusedorderedlogisticregressionwitheachclinicalindicatorasaunivariatemodeltopredictthepatientoutcomes.AgainthebestpredictorwasSBP predictionsor68% MSE 0.397followedby:PVR/SVR,PP,MPAP/AoP,MPAP, PVRI,Cdyn,sDiam,SV/PP,sDiam/BSA,pulmonarycapillarywedgepressure,CO, D ,StrainandmRAPshowninFigure3.2c. 40

PAGE 55

3.4.2CombiningMultipleMeasuresasPredictorsofOutcome Againweusedorderedlogisticregressionnowwitheverycombinationoftheclinicalindicators.Allcombinationsweregeneratedandusedastheexplanatoryvariableintheregressiontodeterminethemostaccuratepredictionofpatientoutcome. WhensDiam,MPAP,MPAP/AoP,SBPwereusedincombination,thesevariables accuratelypredicted59ofthe78patients% MSE .32asillustratedinTable3.1. Theequationfortheorderedlogisticalregressiontofthemeasurestofunctional classoutcomeis: : 68 sDiam )]TJ/F15 11.9552 Tf 11.955 0 Td [( : 18 MPAP + : 56 MPAP AoP + : 15 SBP .1 Asecondequationwithadierentsetof4variables:SBD,PVR/SVR,MPAP, Cdynpredicted57patientWHO-FC'scorrectly3%MSE0.35usingtheequation: : 13 SBP )]TJ/F15 11.9552 Tf 11.955 0 Td [( : 11 MPAP + : 22 PVR SVR )]TJ/F15 11.9552 Tf 11.955 0 Td [( : 37 Cdyn .2 Athirdequationoffourvariables:MPAP,PVRSVR,SBP,Strain,wasaccurate inpredicting56ofthe78patients% MSE .35: : 16 SBP )]TJ/F15 11.9552 Tf 11.955 0 Td [( : 14 MPAP + : 36 PVR SVR )]TJ/F15 11.9552 Tf 11.955 0 Td [( : 63 Strain .3 Asacomparison,PVRIalonewasaccuratefor46ofthe78patients% MSE .53. 0 : 23 PVRI .4 Whenthevariableswerelimitedtothebestsingleindicators,SBP,theprediction accuracyfellto53ofthe78patients% MSE .40.Thealgorithmforthisanalysis andtheresultingequationsisillustratedinAppendix2. 41

PAGE 56

0 : 08 SBP .5 Thenextbestsingleindicator,PVR/SVRaccuratelypredicted52ofthe78patients%MSE.41. 4 : 11 PVR SVR .6 WeillustratethenumberofcorrectandincorrectpredictionsforeachFCcategory forPVRIandourtopmodelequationFigure3.3.PVRImapstoFCIalmostwell asthetopmodel,multivariateequation3.1,with35correctpredictions%outof 42possiblecomparedto37%correctpredictionsforthemultivariateequation. However,inFCIIandFCIII-IVPVRIpredictiveabilitydrops.PVRImakes6out of25%and5of11%correctpredictions,respectively.Whereas,equation 3.1makes15%and7%correctpredictions. 3.4.3SurvivalPredictors WeemployedaRandomSurvivalForestmodeltoselectthemostimportant clinicalindicatorsfortheensembleestimationofthecumulativehazardfunction. Thetopveselectedvariables,PP,Cdyn,SBP,CI,andmRAP,showninTable3.2 wereusedinaCoxproportionalhazardanalysis.Aunivariateanalysisoftheclinical indicatorsasindividualcovariatesmaybecomparedtothemultivariateanalysisofthe clinicalindicatorsasasinglemodel.AKaplan-MeiersurvivalplotshowninFigure 3.4illustratestheprobabilityofaPH-relatedeventgiventhenumberofmonthsfrom RHC. 42

PAGE 57

3.5Discussion ThisstudyfoundthatWHOFunctionalClassicationmaybepredictedayear inadvancewith76%accuracybyutilizingmultivariateequation3.1,includingthe measures:sDiam,MPAP,MPAP/AoP,SBP.Interestingly,asthecombinationsof indicatorswereanalyzed,itbecameapparentthataddingindicatorsdidnotincrease predicationaccuracy.Inaddition,weidentiedthattheindividualindicatorofsystolicpressureandtheratioPVR/SVRpredictsWHOfunctionalclassayearinadvancewith68%and67%accuracyrespectively.Bycomparison,PVRIpredictsfuture functionalclassicationwith59%accuracy. Giventhatthisstudylookedatpredictivetools,weinitiallythoughtmeasures thatprovedpredicativelyaccurateforWHO-FCwouldalignwiththosethatwere predictivefortimetopatientevent.WhileWHO-FCandtime-to-eventweremost accuratelyrepresentedwithamixofhemodynamicandPAmeasures,theirequationsdiered.Thisillustratesthecomplexityofthisdiseaseandhowchangesinthe informationsoughtwilldirectlyimpactwhatmeasuresaremostpredictive. Althoughclinicalmeasurementsareusedtodiagnoseandguidetreatmentin adultPH,fewerstudieshavefocusedonthepediatricpatient.GiventheimportantdierencesbetweenadultandpediatricPH,thisleftasignicantgapinthe accuratepredictionoffuturepatienthealthforthosewhoarelessthan18yearsold whendiagnosedwithPH.Therewerealsogapsintheutilizationofmeasuresbeyond hemodynamics. Itshouldbenotedthatthegreatestaccuracywasnotobtainedbyutilizingeitherthemostpredictiveindividualmeasuresorthemostwidelyusedclinicalindicators.Instead,wefoundthatmeasureswhicharenothighlypredictiveinandof themselvesoftenprovideincrementalvaluethatenhancestheoverallaccuracyofthe model.IfweconsidersDiamindividually,theunivariateregressionsuggestssignificance p= 0.0006,howeverthepredictivevalueof56%accuracyisnotwithinthe 43

PAGE 58

tophalfofthemostaccurateindividualindicatorsandassuchmaybeeliminated fromconsiderationinthenalmodel.However,sDiamaddsimportantinformation tothemultivariateequation.Hence,thevalueofanygivenclinicalindicatorormeasurementmustbetestedinamultivariatesetting.Thesenuancesmayhighlight dierencesbetweenadultandpediatricprogressionofPHaswellashelptoguidethe researchtowarddierentiatedtreatmentplans. Wecontendthatthisstudy'scomprehensiveevaluationofclinicalindicatorsproducedthebesttoolavailableforpredictingfuturehealthstatusintheuniquepediatric PHpopulation.Sincetheseavailablemeasuresprovide76%accuracy,thereremains opportunitytoexaminenewmeasurestocontinuetoimprovepredictionaccuracy goingforward.Underlyingthisopportunityistheneedtounderstandtheprecise mechanismsdrivingthesepredictionsandoutcomes. Thisstudyalsoprovidesameaningfulbasisforadditionalstudiesofthisorsimilarequationstosolidifytheaccuracyofthismodelforfutureclinicalreference.A predictivemodelforclinicalusewillfacilitatetheproactivetherapiesandinterventionsneededtoimprovepediatricoutcomesaswellascreatingamathematicalmodel formoreconsistenttreatmentplansacrossclinicalsettings.Assuch,thisstudymay eventuallyreducethenumberofnecessarypatientdatameasures,therebydiminishing thetimeandinvasive-natureofthecollections. Ourstudyndingsweresupportedwhenreviewedwithourownstatisticaltools andacrossotherstudies.WhenRHCmeasurementsandclinicalindicatorsareanalyzedindividuallyusingpowerfulstatisticaltools,AIC,RandomForestandorderedlogisticregressionprediction,thethreemethodsproducedresultsthatwere wellalignedinpredictingpatientoutcomes.Specically,eachmethodidentiedthe sametopsixindividualmeasuresasthemostaccuratesinglepredictorsofpatient outcomes.TheseresultsalsoalignwithpreviousstudieswhichsuggestMPAP/AoP, MPAP,PVR/SVRandPVRIassinglemeasuresmostpredictiveofpatientoutcome. 44

PAGE 59

3.5.1StudyLimitations Thisisasinglecenterstudywithoutcomesassignedbyoneobserver.Thestudy centerisinAurora,Coloradoatanelevationof5285feetp B 630mmHg,pO 2 130mmHg.Thepatientpopulationhadalownumberofeventsforthesurvival statisticsandfeweventsfortheeventtypes:deathandlungtransplantation .Eachpatientwastreatedwithindividualizedtherapiesandoutcomesmaybe dierentiallyimpacted.Thisstudydidnotaccountfordierencesintreatment. 3.5.2Conclusions ThedetectionandanalysisofmeaningfulpatternswithregardtoPHoutcomes fromlargeamountsofdataisanessentialsteptowarddeningcriticalmeasurements andtheimpactoftheinteractionsamongthosemeasurements.Analysisofthetop individualpredictors,eitheraloneorincombination,doesnotcreatethegreatestpredictivevalue.Instead,ourresultssuggestthattheequationwiththemostpredictive accuracyoffutureWHO-FCincludessystolicdiamterofthePA,MPAP,MPAP/AoP andSBPwithinthePA.Thissuggeststhatmeasurementsofstressandstraingwithin thePAincreaseaccuracypredictionoffuturepatienthealthstatus. 45

PAGE 60

3.6Figures&Tables Table3.1:PatientDemographics 46

PAGE 61

Figure3.1:ColorM-ModeTissueDopplerImageoftheRPA,lumeninnerwallsand echocardiograph. 47

PAGE 62

Figure3.2:Univariateanalysesconrmedbythreemethods:Regression,Random Forest,AIC Table3.2:CoxProportionalHazards 48

PAGE 63

Figure3.3:PredictionResultsbyWHO-FCCategoryforPVRIandtheTopEquation Model Figure3.4:CoxProportionalHazards 49

PAGE 64

Chapter4 4.Study:TheMechanicsofPreconditioningPulmonaryArtery Segments 4.1Abstract Preconditioningbiologicaltissueistheprocessofcyclicallyloadingandunloading thetissuepriortotakingtestmeasurementsofthesample.Preconditioningisthought tobenecessarytoobtainreproduciblestress-straincurves.Wewantedtoknow ifpreconditioningnegativelyaectsthemechanicsofthepulmonaryartery.We hypothesizedthatpreconditioningproximalpulmonaryarterysegmentscompromises passiveandactivevesselwallmechanicssignicantlychangingdiameter,compliance, stress,strain,andelasticmodulus. Werandomlyassigned12-weekoldadultmaleSprague-Dawleyratsinto2cohorts,preconditionedpreandcontrolcon.Asegmentofpre-hilarleftpulmonary arterywasdissected,denudedofendothelium,cannulatedwithmetalpipettes,securedwithsilkligaturesandperfusedwithcalcium-freeKrebs-HenseleitCina perfusedvesselchamber.Luminalpressurewasadjustedusingaperistalticpumpand thecorrespondingvesseldiameterwasrecordedwithavideodimensionanalyzer.Preconditionedarterieswerecyclicallyloadedandunloaded10times;changingluminal pressurefrom5mmHgto60mmHgat1Hz.Controlarterieswerenotpreconditioned. Theouterdiameterofeacharterywasrecordedasluminalpressurewaschanged.Resultingpressure-diameterP-DmeasurementswereusedtocalculatevesselcomplianceC=pressure/diameter,smoothmuscleSMcontraction,circumferentialwall stressandstrainandelasticmodulusE=stress/strain. Ourresultsshowthatstrainwassignicantlyincreasedwithluminalpressure loadsfrom20mmHgcon=0.28 .036,pre=0.44 0.030, p < 0.05through55mmHg 50

PAGE 65

con=0.59 .1,pre=0.82 0.08, p < 0.05.Elasticmoduluswassignicantlydecreased con=126.86 17.00kPa,pre=106.64 7.42kPa, p < 0.05at20mmHgandwasdecreasedbutnotsignicantlycon=203.58 16.93,pre=184.96 7.32, p =0.095at 55mmHgThecomplianceofthevesselwassignicantlyincreasedcon=0.02 0.003mmHg/mm, pre=0.03 0.002mmHg/mm, p < 0.05,meanarterialdiameterwassignicantlyincreasedcon=1.99 0.08mm,pre=2.11 0.07mm, p < .05andsmoothmusclecontractionhaddecreased,butnotsignicantly. Basedontheseresults,weconcludethatpreconditionedarteriesaremorecompliant,largerindiameterandhavelargervaluesofstrainanddecreasedelasticmodli thancontrolarteriesthathavenotbeenpreconditioned. 4.2Introduction Tissuemechanicsareimportantforourunderstandingofanatomyandphysiology [114].Thefunctioningofbodiesunderphysiologicloadsinhealthandthechanges throughdiseasehelptomodel,diagnoseandpredicthealthstatus.Quantication ofthemechanicsandprocessesofthebodyaimstoimprovediagnosis,guidetherapeutictreatments,andimprovetissueengineering,graftsandprostheses[20].The foundationofquanticationisexperimentation,modeling,computationandmechanicalmeasurementoftissues[20].However,quanticationmaynotbecomparable acrossstudiesifin-vitrofactorsimpactthedistensibilityofthetissues.Subsequently testingprotocolsbecomecentraltotheprecisionofthemeasurementstoallowcrossstudycomparison. Tissuemechanicsmaybebroadlycategorizedashardtissuemechanicslikebone andteethandsofttissuemechanicslikeskin,tendons,organsandbloodvessels. Onemajormechanicaldierencebetweenthesetwobroadcategoriesistheamount ofdeformationeachexperiencesunderphysiologicalloads.Hardtissuestypically 51

PAGE 66

showverysmallamountsofdeformationwhilesofttissuestypicallyshowlargerdeformations[20].However,softtissuesgenerallydonotconformwelltomechanical denitionsastheyareneitherpurelyelasticnorpurelyplasticbutshowcharacteristics ofbothviscoelasticity.Therefore,softtissuestypicallyshowshysteresisbetween loadingandunloadingandaregenerallyanisotropicduetothebrousarrangement oftheextracellularmatrix[33][46][73].Thesecharacteristicsconfoundtheabilityto mathematicallyquantifysofttissueproperties. Therearevaryingprotocolsinthepreparationofbiomaterialsandparticularly implanttissues[114].Preconditioningisoneformofsofttissuepreparation.Preconditioningistheprocessofcyclicallyloadingandunloadingtissuesamplespriorto takingmeasurements.Proponentsofpreconditioningprotocolssuggestitisnecessary toobtainreproduciblemechanicalrelationshipssuchasstress-straincurves[47][48]. Fungwarnsthatpreconditioningnecessarilychangestheinternalstructureofthe tissueinordertoreachasteadystateformechanicalmeasurements[32].However, heleavesthestatementopen,raisingthequestionsregardingchangestotheinternal structureandthelinkstomechanicalproperties. Cyclicallyloadingsofttissueabovephysiologicalpressuresorfrequenciesisnota processfoundinthebody.Accordingly,preconditioningtissueinamannerthatmay changeinternalstructuralcouldalsopotentiallyaltertheunderlyingcharacteristics ofthetissueandtheresultingmeasurementsandcalculations. Giventhedisparitybetweenreproducibleresultsandin-tackin-vivostructural mechanics,westudiedtheimpactofpreconditioningonthemechanicsofratpulmonaryarterysegments.Wehypothesizethatpreconditioningproximalpulmonary arterysegmentscompromisespassiveandactivevesselwallmechanicssignicantly changingdiameter,compliance,stress,strain,andelasticmodulus.Inaddition,we demonstratethatrepeatablestress-straincurvesandconsistentmechanicalresponse canbeobtainedwithnon-preconditionedtissue. 52

PAGE 67

4.3Methods 4.3.1Animals Weused13normaladultmaleSprague-DawleyratsHarlanLaboratories,USA. AllanimalsweretreatedaccordingtheUniversityofColoradoAnimalCareProgram andtheInstitutionalAnimalCareandUseCommitteeIACUCfollowingprotocol 101913IEAppendix5. 4.3.2IsolatedVesselChamberTestingSystem TheisolatedvesselchambertestingsystemLSIconsistsoftwometalcannulas runningacrossasuperfusatebathwithapproximately2mmbetweenthecannulasat thecenterofthebath.Anisolatedvesselwassecuredtoeachpipettetocompletea perfusatecircuitconsistingoftwopressuretransducersoneithersideofthecannulas, aperistalticpump,andaperfusatereservoir.Thepressuretransducerswereplaced inlinewiththecannulas,onebeforethecannulas,andoneafterthecannulas.A peristalticpumpLSIplacedinlinepriortothersttransducerpumpedperfusate fromawaterjacketedaerated30mlreservoirthroughthersttransducer,rstcannula,tissuesegment,secondcannula,secondpressuretransducerandbackintothe perfusatereservoir.Thesuperfusatecircuitconsistedoftheaeratedsuperfusate1literreservoirandasecondpumptocirculatethesolutionthroughtheisolatedvessel chamberbathandbacktothereservoir.Athirdpumpandwaterheatercirculated warmeddistilledwateraroundjacketedcondenserglasswareatthesuperfusateinow tothevesselbathandthroughtheperfusatejacketedreservoir.Theperfusateand superfusatewerere-circulated;thethermometersmeasuredthetemperatureinthe reservoirstoverifyaconsistenttemperatureof37 C. 53

PAGE 68

4.3.3ArterySegmentPreparation Theanimalswereanaesthetizedusingpentobarbitalsodium.Ananteriorthoracotomywasperformedafterdeeppainreexesbecameabsent.Bothlungswereremoved andplacedinabueredKrebs-HenseleitsolutionmM:118.0NaCl,4.75KCl,1.18 KH 2 PO 4 ,1.18MgSO 4 7H 2 O,24.80NaHCO 3 ,2.52CaCl 2 ,and10.0D-glucosesaturatedwith95%O 2 and5%CO 2 maintainingpHat7.45.Weexcisedasegmentof leftpre-hilarpulmonaryarteryapproximately4mminlengthbetweenthepulmonary trunkandthersthilarbranchfromeachanimal. Wemountedthearterialsegmentundersolutioncalcium-freeKrebsHenseleitCa ++ -freeKHSsolutionmM:118.0NaCl,4.75KCl,1.18KH 2 PO 4 ,1.18 MgSO 4 7H 2 O,24.80NaHCO 3 ,and10.0D-glucosesaturatedwith95%O 2 and5% CO 2 maintainingpHat7.45onthemetalpipettesintheisolatedvesselchamber testingsystem.Thearterysegmentsweregentlyrotatedonametalcannulatoremovetheendotheliumandwerethensecuredwithsilkligaturesatbothendsto thebathcannulaswithadistancebetweencannulatipsofabout2.5mm.WeperfusedthecannulatedarterysegmentwiththesameCa ++ -freeKHSsolutionasthe superfusate.Thesegmentsequilibratedfor1hourat37 Clessthan5mmHgofintraluminalpressureataowrateofapproximately0.5ml/min,priortoanyexperimental manipulation. WemadeallmeasurementsatambientpressureAurora,CO:p B 630mmHg, pO 2 130mmHgandpressuremeasurementsareexpressedaschangesrelativeto ambient. 4.3.4Non-PreconditionedMechanicalMeasurements Afterequilibrating,theluminalpressureoftheisolatedvesselswasincreased in5mmHgstepstoapproximately55mmHgrange44.8to55.3mmHgat3-minute intervalsusingaperistalticpump.Attheendofeachtimeinterval,werecordedthe 54

PAGE 69

outerdiameterofthecannulatedarterywithavideodimensionanalyzerLSIand thecorrespondingintra-luminalpressure.Uponreachinganintra-luminalpressureof approximately55mmHg,wereducedthepressurebythesame5mmHgstepsand3minutetimeintervalstobaseline,againrecordingtheouterdiametersandpressures. Thetimeintervalwassetwitheachnewvesseltoensurethatthechangeindiameter followingapressurestepchangewascomplete. Thevessel,bathchamber,perfusateandsuperfusatereservoirswerethenushed withhigh-potassiumHi-KKHSmM:14.72NaCl,107.98KCL,1.18MgSO 4 7H 2 0, 1.18KH 2 PO 4 ,24.80NaHCO 3 ,2.52CaCl 2 2H 2 O,10.0D-glucosesaturatedwith95% O2and5%CO2maintainingpHat7.45.Duringtheequilibrationperiod,thevessels decreasedindiameteragainsttheperfusionpressureof5mmHgbyanoverallaverage of0.21 0.7mm.Timeintervalsforpressurestepswereevaluatedasdescribedabove. Followingtheequilibrationperiodwerepeatedthesequenceof5mmHgpressuresteps upto55mmHgandbackinHi-KKHS. 4.3.5PreconditionedMechanicalMeasurements Afterequilibratingincalcium-freeKrebsHenseleitsolution,wecyclicallyloaded andunloadedtheisolatedvessels10times,changingintra-luminalpressurefrom 5mmHgto60mmHgat1Hz.Theluminalpressureoftheisolatedvesselswasthen increasedin5mmHgstepstoapproximately55mmHgrange44.8to55.3mmHgat 45-secondintervalsusingaperistalticpump.Attheendofeachtimeinterval,we recordedtheouterdiameterofthecannulatedarterywithavideodimensionanalyzer LSIandthecorrespondingintra-luminalpressure.Uponreachinganintra-luminal pressureofapproximately55mmHg,wereducedthepressurebythesame5mmHg stepsand45-secondtimeintervalstobaseline,againrecordingtheouterdiametersand pressures.Thetimeintervalwassetwitheachnewvesseltoensurethatthechange indiameterfollowingapressurestepchangewascomplete.Finallythepressurewas 55

PAGE 70

setat30mmHgand15mmHgandrespectivediametermeasurementsweretakento makesuretherewasnodrift. Thevessel,bathchamber,perfusateandsuperfusatereservoirswerethenushed withhigh-potassiumHi-KKHSmM:14.72NaCl,107.98KCL,1.18MgSO 4 7H 2 0, 1.18KH 2 PO 4 ,24.80NaHCO 3 ,2.52CaCl 2 2H 2 O,10.0D-glucosesaturatedwith95% O2and5%CO2maintainingpHat7.45.Thepressure-stepandrespectivediameter measurementswererepeated. 4.3.6PACompliance: Theslopeofapressure-diameterP-Dcurveisanindicationofthecomplianceof thevessel.Weestimatedthevesselcomplianceinmm/mmHgforeacharteryactive andpassiveastheinverseoftheslopeofalinedrawnbetweenthe15mmHgpoint andthe40mmHgonthePDcurve. 4.3.7PADiameter&VascularSmoothMuscleActivity WecalculatedthePAdiameterforeachartery,solutiontype,andpressurestep. Wereportthevesseldiametersasmeandiametermm SD.Thediameterrangefor eacharteryandsolutiontypewerecalculatedandreportedasmeandiameterrange mm SD. Wecalculatedthesmoothmusclecontractionasitshortensthevessel?Dat eachpressurestepforeachvesselbysubtractingtheactivediameterfromthepassive diameter.Ourstudyreportsthe?Dasmm SD. 4.3.8PAWallTension Wecalculatethevesselcircumferentialwallstressusingthecircumferentialstress equationforcylinders: S = P i R i )]TJ/F19 11.9552 Tf 11.956 0 Td [(P o R o R o )]TJ/F19 11.9552 Tf 11.955 0 Td [(R i )]TJ/F19 11.9552 Tf 13.151 8.088 Td [(R i R o P o )]TJ/F19 11.9552 Tf 11.955 0 Td [(P i R mid R o )]TJ/F19 11.9552 Tf 11.955 0 Td [(R i .1 56

PAGE 71

whereS=wallstress,P i =luminalpressure,R i =luminalradius,R o =outerradius, R mid =mid-wallradiusandP o =outerpressure,assumedheretobe0mmHg,butkept intheequationforcompleteness.Usingtheassumptionthatarterysegmentsare iso-volumetric,wecalculatedthecrosssectionalareaforeacharterysegmentusing thethicknessoftheslackarteryandouterdiameterat5mmHgasinitialconditions. Foreachsubsequentouterdiametermeasurementwecalculatedthecorresponding innerdiameterassumingconstantarteryvolumeforeachartery.Wereportmean wallstress,maximumwallstress,andstressatmPAPfornormoxicandhypoxic conditionsaskPa SD.Wealsocalculatedthewallstressusingthethin-walled approximation: P i R wallthickness alsoreportedaskPA SD. 4.3.9EndotheliumandVSMResponse Inseriesofexperiments,weveriedeectivedisruptionofendothelialactivity. ThiswasaccomplishedbyaddingacetylcholineACh )]TJ/F17 7.9701 Tf 6.586 0 Td [(5 Mtotheperfusateand superfusatesolutionofanactivatedHiKvesselsegmentoverarangeofpressures:5, 20,25,30mmHg.ThediametersremainedconsistentwhetherintheHiKsolutionor inHiKwithAChsupportingthedisruptionofendothelialactivityduetotheabsence ofarelaxationresponse. TheabsenceofbasalsmoothmuscletonewasveriedbyaddingsodiumnitroprussideSNP )]TJ/F17 7.9701 Tf 6.586 0 Td [(5 MtotheperfusateandsuperfusatesolutionCa ++ -freeKHS ofapurportedlyinactivatedvesselsegmentoverarangeofpressures:5,20,25, 30mmHg.ThediametersremainedconsistentwhetherintheCa ++ -freeKHSsolutionorinCa ++ -freeKHSwithSNPsuggestingnopre-existingVSMtoneasthere wasnoincreaseindiameterinthepresenceoftheSNPvasodilator. 57

PAGE 72

4.3.10Statistics StatisticalanalysiswasperformedusingtheRstatisticalsoftwareVersion2.5R DevelopmentCoreTeam2014.TukeyHonestSignicantDierenceHSDwasused todeterminestatisticalsignicancesetat p < 0.05.Appendix3comprisesthecode forthestatisticalanalysis,calculationsandderivedmeasures. 4.4Results Ascomparedtonon-preconditionedtissue,thepreconditionedvesselsegments arelargerindiameterandthevesselmorecompliantaswellasshowingincreases incircumferentialstressandstrainwhileelasticmodulusdecreased.Smoothmuscle abilitytodecreasearterydiameterisalsoimpairedsuggestingthatpreconditioning compromisesvasoactivity. WeanalyzedstrainatmeanpulmonaryarterialpressuremPAP,corresponding toaluminalloadof20mmHgwhichiscommonlythoughtofastheelasticregion oftheartery.Wealsoanalyzedstrainatthetopofthetestedpressures,55mmHg, whichiswelloutsidethephysiologicrange,butcommonlythoughtofastheregion ofcollagenengagementoftheartery.Strainwassignicantlyincreasedwithluminal pressureloadsfrom20mmHgcon=0.28 0.036,pre=0.44 0.030, p < 0.05through 55mmHgcon=0.59 .1,pre=0.82 0.08, p < 0.05.Elasticmoduluswassignicantly decreasedcon=126.86 17.00kPa,pre=106.64 7.42kPa, p < 0.05at20mmHgand wasdecreasedbutnotsignicantlycon=203.58 16.93,pre=184.96 7.32, p =0.095 at55mmHgFigure1.Thecomplianceofthevesselwassignicantlyincreased con=.02 0.003mmHg/mm,pre=0.03 0.002mmHg/mm, p < 0.05,meanarterialdiameterwassignicantlyincreasedcon=1.99 0.08mm,pre=2.11 0.07mm, p < 0.05 andsmoothmusclefunctiondecreased,butnotsignicantly. 58

PAGE 73

4.5Discussion Thisstudyreportsnoveldatasupportingthehypothesisthatthemechanicsof preconditionedissuearesignicantlyaltered.Specicallythisstudynds: EvidenceofgeometricalchangestothePAwall:increaseddiameterandcompliancewithpreconditioning. Evidenceofvascularsmoothmusclechanges:signicantdecreasesinDoverthe rangeofpressureswithpreconditioninganddecreasedD max ,thoughnotsignicantly. EvidenceofmechanicalchangesinthePAsegments:increasedstressandstrain, decreasedelasticmoduluswithpreconditioning. Wefurtherdemonstratethatpreconditioningisnotnecessarytocreatereproducibleresults.Weillustraterepeatablestressstraincurveswithoutpreconditioning thearteries.Thisisthecaseforbothactivatedandinactivatedtissue.Theexactmechanicsbehindtheseresultsarenotclear,butitispossiblethatpreconditioningrat pulmonaryarteriesmayisolatethemechanicalpropertiesoftheextracellularmatrix proteinsbybreakingmatrixcross-linkswhiledatafromthecontrolarteriesdescribes themechanicalpropertiesofthecompositematerialwithintactmatrixcross-links. Previousstudieshavenotcomparedpreconditionedarterytissuestoarterytissues thathadnotbeenpreconditioned. Theabilitytomakemeasurementswithoutrstpreconditioningtissuesmayenhanceourabilitytoquantifythemechanicsofarterytissueandmakecross-study comparisons.Thismaybesupportedbecausethelimitedchangestotheinternal structureofthesetissuesandmayrepresentthestateofthetissueinitsnaturalbiologicalsetting.Thisleavestheopportunityforfuturestudiestoevaluatetheimpact ofvariouspreconditioningprotocolsontheresultingstress-straincurves. 59

PAGE 74

4.5.1Limitations Aspreviouslymentionedweallowedthevesseldiametertofullycontractorrelax overarelativelylongtimeintervalwitheachpressuresteppriortotakingmeasurements.Arterycompliancehasatimecomponentanddependsontherateofchangein diameter.Thevesselappearstobelesscompliantwithfasterchangestovesseldiameter.Alternatively,thevesselappearsmorecompliantwithslowerdiameterchanges. Thisviscoelasticeecthelpsprotecttheartery,attenuatinghigh-frequencypulsesand dissipatingthepulsatileenergyfromtheheart,alsoactingasanelasticreservoirand secondstagepumpfortheheart[96].Thisstudyremovesasmuchoftheviscoelastic eectaspossibleinordertoisolatethesteadystateelasticandcontractileconditions ofthearterywall,thusneglectingthedynamicandpulsatilefunctionsofthePA. Proteinanalysiswouldhelpdetermineifcollagenandelastincross-linksarebrokenuponpreconditioning.Additionally,theperistalticpumpallowedustosetthe owrateortosettheintra-luminalpressure.Withpressureset,weapproximated theowrateusingaseparatereservoiroveragiventimeperiod.Lastly,thestudy lacksbiochemicaldatatomorepreciselyquantifythevesselproteins,relyinginstead onPAmorphologytovisualizethechangesincellularstructure. 4.5.2Conclusions Takentogetherthesedatasuggestthatthemechanicsoftheratproximalpulmonaryarteryaresignicantlyalteredwithpreconditioningincluding:diameter, compliance,stress,strain,andelasticmodulusofthePAwall.Wealsodemonstratethefeasibilityoftestingandcollectingmechanicalmeasurementsresultingin reproducibleandstress-strainrelationshipsofratproximalPAsegmentswithoutpreconditioning. 60

PAGE 75

4.6Acknowledgments ThisstudywasgenerouslysupportedbyNIHprogramProjectGrant#NHLBI-K25 HL094749P.I.KendallHunter. 4.7Figures Figure4.1:TenSuccessiveStress-StrainCurves. 61

PAGE 76

Figure4.2:SchematicofthePressureInationTestingChamberApparatus. Figure4.3:DataTraces:diameterandpressurepaireddatapoints 62

PAGE 77

Figure4.4:DatafromaCompleteTestofOneControlArtery. Figure4.5:Pressure-DiameterMeasurements. 63

PAGE 78

Figure4.6:Dierenceindiameterbetweenactiveandpassivevessels,D. Table4.1:SummaryDataofPreconditioningMeasurementsandCalculations. 64

PAGE 79

Figure4.7:Stress-StrainCurves. 65

PAGE 80

Chapter5 5.Study:ProximalPulmonaryArteryMechanicsinPulmonary HyptertensionandRecovery 5.1Abstract Recentstudiessuggestthatproximalpulmonaryarteryimpedance,areastrain andvascularcapacitanceplayamajorroleinthepathologyofPulmonaryHypertensionPH,especiallyinthedevelopmentofrightheartfailure.TheproximalpulmonaryarteryisknowntoremodelandundergomechanicalstieninginPH.However, thechangesinactivevesselmechanicsduringandinrecoveryfromPHarenotaswell understood.Thisstudyexamineschangesinproximalpulmonaryarterymechanics duringthedevelopmentofandrecoveryfromPH,inducedbychronichypoxiain Sprague-Dawleyrats.Westudiedveconditions,distinguishedbyexposuretovaryinglengthsofnormoxiaelevation=5285ft,PB=632mmHg,pO 2 =132.5mmHg,hypoxiasimulatedelevation=17,000ft,PB=410mmHg;pO 2 =85.9mmHg,orhypoxia followedbynormoxiaWeekNormoxic,3WeekHypoxic,6WeekHypoxic,3Week Hypoxic+6WeekNormoxic,9WeekNormoxic.Wefoundthemaximumcontractionofthecannulatedpulmonaryartery.4 5.2mmHgwasseverelydiminished following3weeksofhypoxia p < 0.001andwasfullyrestored p < 0.001intherecoverygroup.Thearterywallcompliance D P decreaseddramaticallywithhypoxic exposure p < 0.001.Uponrecovery,compliancewasincreasedbutdidnotfullyresolve p < 0.001.Atanygivenpressure,theproximalarterydiameterdecreasedwith hypoxicexposureandwasrestoredtonormallevelsintherecoverygroup.Wallstress andmodulusofelasticityweredecreasedaswell. 66

PAGE 81

5.2Introduction PulmonaryhypertensionPHisararebutseriousdisorderofthepulmonaryvasculaturecharacterizedbypulmonaryarteryPApressuresabove25mmHg[93][98]. Ithasanincidencerateof1.1to2.4casespermillionperyear,isassociatedwith asignicantincidenceofrightheartfailureandaveyearmortalityrateof40% [37][53]. PathologicalchangesassociatedwithPHincludedistalpulmonaryarteryvasoconstriction,accompaniedinthemediumtolongtermbythickeningandstiening ofthearterywall[104][26].Thenarrowingofthedistalpulmonaryvascularbed dramaticallyincreasespulmonaryvascularresistancePVRandaltersthemechanic loadsonthepulmonaryvasculartree[61][79].Consequently,studiesofandtreatment forPHhavefocusedintentlyonthedistalvasculature[3][6]. Morerecently,humanstudieshavefoundthatproximalpulmonaryvesselmechanicsarehighlypredictiveofPHmortalityincludinginputimpedance,areastrain andvascularcapacitance[36][52][72].Otherclinicalstudieshaveshownthatproximal stinessisalsopredictiveofPHprogression.GiventheapparentroleofproximalPA mechanicsinPH,wesoughttoquantifyPAmechanicstodetermineifthemechanicaladaptationsintheproximalPAareanattempttomaintainthewallstressata constantlevelinthefaceofchangingvesselwallloadscreatedbyPH[115]. WeplacedSprague-DawleyratsintohypobaricchamberstoinitiatehypoxiainducedPHforthreetosixweeksandthenreturnedthemtonormoxiatoallowthe PHtoresolve.Wehypothesizedthatwallstresswouldchangelittledespitemarked changesinpulmonaryarterypressureandwallthicknessduetoPHsuggestedby recentstudies. 67

PAGE 82

5.3Methods 5.3.1Animals WeusedadultmaleSprague-DawleyratsHarlanLaboratories,USAtoavoid potentialsexuallydimorphiceectsonvascularsmoothmuscle[91].Allanimalswere treatedaccordingtheUniversityofColoradoAnimalCareProgramandtheInstitutionalAnimalCareandUseCommitteeIACUCfollowingprotocol101913IE Appendix5. 5.3.2HypoxicChambers Hypoxia-inducedpulmonaryhypertensionwasachievedbyexposinganimalsto alowbarometricpressureof410mmHgpO 2 =85.9inhypobaricchambers.There werebriefweeklyinterruptionsforanimalcarethatlastednomorethan20minutes. Normoxicconditionsconsistedofhousingtheanimalsnearthebarometricchamber andprovidinganimalcareviabriefweeklyinterventions.Thebarometricpressure fortheseanimalswasabout632mmHgpO 2 =132mmHgandwastypicalforAurora, COUSAelevation=5285ft.Theanimalsreceivedunlimitedaccesstowaterand wereallonthesame12hourlightcycle. Allanimalsweresacricedfollowingtheirrespectiveenvironmentalexposures. Werecordedthebodyweightsandrandomlyallocatedtheanimalstorightheart hemodynamicsandmechanicaltesting.Forty-sixratswereusedformechanicaltestingand44forhemodynamicmeasurements. 5.3.3HemodynamicMeasurements HemodynamictestingwasperformedbytheUniversityofColoradoDenverCardiovascularPhysiologyCore.Eachmeasurementsessionlastedabout45minutes. A1.9FrenchPressure-VolumecatheterFTE-1912B-6018TransonicSystemsInc., 68

PAGE 83

Ithaca,NYinsertedintotheheartviarightthoracotomy,measuredrightventricularRVandPApressures.Anadditionalpressure-onlycatheterTransonicFTH1611B-0018,1.6Fplacedinthefemoralarterymeasuredsystemicarterialpressure. Pressuresarereportedasmean standarddeviation. Theanimalswereinducedwith5%isouraneinroomairat1l/minforapproximately2minutes.Fromasupineposition,theanimalswereintubatedusinga trachealcannulaandconnectedtoa16gtubingadapter.TheanimalwasthenconnectedtoanAnesthesiaWorkstationHallowellandanesthesiawasmaintainedat 1.5-2.5%isouranein100%oxygen.Wekeptpeakairwaypressureat16-18cmH 2 O, respiratoryrateat50-150breathsperminute,andoxygenowat0.5-0.8l/min. Eachanimalunderwentarightthoracotomyatthelevelofthe6thand7thribs toexposetheheart.Thepericardiumwasresectedandanincisionwasmadeat thebaseoftherightventricleRVusinga26gaugeneedle.Apressure-volume catheterwasinsertedthroughthesiteofincisionandadvancedalongthelengthof theRV.Toeliminateventilatorartifactfromthepressure-volumerecordings,steady statehemodynamicswerecollectedwithshortpausesinventilation 10secorhighfrequencyoscillatoryventilation.Weoccludedtheinferiorvenacavabythreadinga suturearoundthevesselandpullingittautfor10seconds.Thecatheterwasthen removedfromtheRV.AnotherincisionwasmadejustbelowthePA,andthecatheter wasinsertedtocollectPApressures.DatawascontinuouslyrecordedwithLabScribe 2iWorx,Dover,Nwhichautomaticallycalculatesandrecordsheartrate,cardiac output,andstrokevolume.Therewerenocomplications. 5.3.4IsolatedVesselChamberTestingSystem TheisolatedvesselchambertestingsystemLSI,CH2consistsoftwometal cannulasrunningacrossasuperfusatebathwithapproximately2mmbetweenthe cannulasatthecenterofthebath.Anisolatedvesselwassecuredtoeachpipetteto 69

PAGE 84

completeaperfusatecircuitconsistingoftwopressuretransducersoneithersideof thecannulas,aperistalticpump,andaperfusatereservoir.Thepressuretransducers wereplacedinlinewiththecannulas,onebeforethecannulas,andoneafterthe cannulas.AperistalticpumpLSI,PS-200placedinlinepriortothersttransducer pumpedperfusatefromawaterjacketedaerated30mlreservoirthroughtherst transducer,rstcannula,tissuesegment,secondcannula,secondpressuretransducer andbackintotheperfusatereservoir.Thesuperfusatecircuitconsistedoftheaerated superfusate1-literreservoirandasecondpumptocirculatethesolutionthroughthe isolatedvesselchamberbathandbacktothereservoir.Athirdpumpandwater heatercirculatedwarmeddistilledwateraroundjacketedcondenserglasswareatthe superfusateinowtothevesselbathandthroughtheperfusatejacketedreservoir. Theperfusateandsuperfusatewerere-circulated;thethermometersmeasuredthe temperatureinthereservoirstoverifyaconsistenttemperatureof37 C. 5.3.5ArterySegmentMechanicalMeasurements Theanimalswereanaesthetizedusingpentobarbitalsodium.Ananteriorthoracotomywasperformedafterdeeppainreexesbecameabsent.Bothlungswereremoved andplacedinabueredKrebs-HenseleitsolutionmM:118.0NaCl,4.75KCl,1.18 KH 2 PO 4 ,1.18MgSO 4 7H 2 O,24.80NaHCO 3 ,2.52CaCl 2 ,and10.0D-glucosesaturatedwith95%O 2 and5%CO 2 maintainingpHat7.45.Weexcisedasegmentof leftpre-hilarpulmonaryarteryapproximately4mminlengthbetweenthepulmonary trunkandthersthilarbranchfromeachanimal. Wemountedthearterialsegmentundersolutioncalcium-freeKrebsHenseleitCa ++ -freeKHSsolutionmM:118.0NaCl,4.75KCl,1.18KH 2 PO 4 ,1.18 MgSO 4 7H 2 O,24.80NaHCO 3 ,and10.0D-glucosesaturatedwith95%O 2 and5% CO 2 maintainingpHat7.45onthemetalpipettesintheisolatedvesselchamber testingsystem.Thearterysegmentsweregentlyrotatedonametalcannulatoremove 70

PAGE 85

theendotheliumandwerethensecuredwithsilkligaturesatbothendstothebath cannulaswithadistancebetweenligaturesofabout2.5mm.WeperfusedthecannulatedarterysegmentwiththesameCa ++ -freeKHSsolutionasthesuperfusate.The segmentsequilibratedfor1hourat37 Clessthan5mmHgofintraluminalpressure ataowrateofapproximately0.5ml/min,priortoanyexperimentalmanipulation. PreliminaryexperimentsshowedanabsenceofarelaxationresponsetoAcetylcholine )]TJ/F17 7.9701 Tf 6.587 0 Td [(5 Minactivatedvessels,conrmingtheeectivedisruptionofendothelialsignaling,andanabsenceofarelaxationresponsetosodiumnitroprussidebyvessels inCa ++ -freesolution,demonstratingtheabsenceofbasalsmoothmuscletonedata notshown. WemadeallmeasurementsatambientpressureAurora,CO:p B 630mmHg, pO 2 130mmHgandpressuremeasurementsareexpressedrelativetoambient.Afterequilibrating,theluminalpressureoftheisolatedvesselswasincreasedin5mmHg stepstoapproximately55mmHgrange44.8to55.3mmHgat3-minuteintervals usingaperistalticpump.Attheendofeachtimeinterval,werecordedtheouterdiameterofthecannulatedarterywithavideodimensionanalyzerLSI,VDA-10and thecorrespondingintra-luminalpressure.Uponreachinganintra-luminalpressureof approximately55mmHg,wereducedthepressurebythesame5mmHgstepsand3minutetimeintervalstobaseline,againrecordingtheouterdiametersandpressures. Thetimeintervalwassetwithnewvesselstoensurethatthechangeindiameter followingapressurestepchangewascomplete. Thevessel,bathchamber,perfusateandsuperfusatereservoirswerethenushed withhigh-potassiumHi-KKHSmM:14.72NaCl,107.98KCL,1.18MgSO 4 7H 2 0, 1.18KH 2 PO 4 ,24.80NaHCO 3 ,2.52CaCl 2 2H 2 O,10.0D-glucosesaturatedwith95% O 2 and5%CO 2 maintainingpHat7.45.Thearterieswereallowedtocometo equilibriumat5mmHgand37 Cfor45minutesequilibrationperiod.Duringthe equilibrationperiod,thevesselsdecreasedindiameteragainsttheperfusionpressure 71

PAGE 86

of5mmHgbyanoverallaverageof0.21 0.7mm.Followingtheequilibrationperiod werepeatedthesequenceof5mmHgpressurestepsupto55mmHgandbackinHi-K KHS.Thetimeintervalsforpressurestepswereevaluatedasdescribedabove. 5.3.6DerivedMeasures Weestimatedthevesselcomplianceinmm/mmHgforeacharteryactiveand passiveastheslopeofalinedrawnbetweenthe15mmHgpointandthe40mmHg onthePDcurve.Toquantifytheimpactofsmoothmuscleactivation,wemeasured theabilityofthesmoothmuscletoshortenatagivenpressure.Toaccomplishthis, wecalculatedthesmoothmusclecontractionasitshortensthevesselDateach pressurestepforeachvesselbysubtractingtheactivediameterfromthepassive diameter.WereporttheDasmm SD. Wecalculatethevesselcircumferentialwallstressusingthecircumferentialstress equationforcylinders: S = P i R i )]TJ/F19 11.9552 Tf 11.956 0 Td [(P o R o R o )]TJ/F19 11.9552 Tf 11.955 0 Td [(R i )]TJ/F19 11.9552 Tf 13.151 8.088 Td [(R i R o P o )]TJ/F19 11.9552 Tf 11.955 0 Td [(P i R mid R o )]TJ/F19 11.9552 Tf 11.955 0 Td [(R i .1 whereS=wallstress,P i =luminalpressure,R i =luminalradius,R o =outerradius, R mid =mid-wallradiusandP o =outerpressure,assumedheretobe0mmHg,butkept intheequationforcompleteness.Usingtheassumptionthatarterysegmentsare iso-volumetric,wecalculatedthecrosssectionalareaforeacharterysegmentusing thethicknessoftheslackarteryandouterdiameterat5mmHgasinitialconditions. Foreachsubsequentouterdiametermeasurementwecalculatedthecorresponding innerdiameterassumingconstantarteryvolumeforeachartery.Wereportmean wallstress,maximumwallstress,andstressatmPAPfornormoxicandhypoxic conditionsaskPa SD.Wealsocalculatedthewallstressusingthethin-walled approximation: P i R wall )]TJ/F20 7.9701 Tf 6.587 0 Td [(thickness alsoreportedaskPA SD. 72

PAGE 87

5.3.7Histology Aftermechanicaltesting,aportionofeacharterywasxedwith10%neutralbueredformalinandparan-embedded.Threetransversesectionswerecutfrom eacharteryblockatathicknessof5 mandstainedforhistologicalexamination. HaematoxylinandeosinH&Estainwasusedtoquantifycellnumberandtovisualizethegrossmusclestructureandcellmorphology.Masson'sTrichromeAmerican MasterTechwasusedtoquantifythemuscleandcollagenstructure.Verhoe's ElastinStainAmericanMasterTechwasusedtodetermineelasticbercontentand structure.ANikonEclipseTimicroscopecapturedallarteryimagesat0.68 m/pixel. WeusedlaboratoryimagingsoftwareNIS-ElementsAR4.20vtoquantifythenuclei oftheH&Estainedarteriesutilizingacombinationofcolor,sizeandshapethresholdingofthehigh-resolutionimages.Thisstudyreportsthemeannumberofnuclei per m 2 SD.Werandomlyselectedsix100 mtransmuralsectionsfromeachMasson'sTrichromestainedarteryandfromeachVerhoeElastinstainedartery.NIH ImageJ.47vwasusedtoprocessthehigh-resolutionimagesandmeasurethearea fractionsofmuscle,collagen,elasticber,aswellasthetotalareaviacolorthresholding.Summarystatisticsreportthemeanareaas m 2 SD.ImageJwasalsousedto measurethethicknessofthemediaandadventitialayersofthetransmuralsections withsummarystatisticsreportedasmeanthickness m SD. 5.3.8ExperimentalDesign&StatisticalAnalysis Weusedafullyrandomizeddesignwheresix-week-oldSprague-Dawleyratswere randomlyassignedtooneof5groups: HPX3 wasexposedto3weeksofhypoxia. CNTL waskeptinnormoxicconditionsfor3weeks. HPX6 wasexposedto 6weeksofhypoxia,adoublingofthehypoxicexposure. REC wasexposedto3 weeksofhypoxiafollowedby6weeksofnormoxia. RECNTL wasexposedto9 weeksofnormoxia. 73

PAGE 88

StatisticalanalysiswasperformedusingtheRstatisticalsoftwareVersion2.5R DevelopmentCoreTeam2014.TukeyHonestSignicantDierenceHSDwasused todeterminestatisticalsignicancesetat p < 0.05.Appendix4comprisesthecode forthestatisticalanalyses,calculationsandderivedmeasures. 5.4Results 5.4.1Animals Atthetimeofstudyentry,animalswereapproximately6-weekoldSpragueDawleyRatsweighing250-275g.OurHPX3andCNTLanimalswere9weeksoldand weighedthesamewhenremovedfromthechambers.CNTL306.10 10.16g,HPX3 299.89 6.35g.TheHPX66-weekhypoxicgroupwas12weeksoldandweighed 346.50 3.2guponremoval.TheRECandRECNTLgroupswereboth15weeks oldandweighed438.75 12.41gand385.00 3.20grespectivelywhenremovedfrom thechamber.Figure5.1illustratestheexperimentaldesignandtimingofthestudy groups. 5.4.2Hemodynamics HemodynamicmeasurementsFigure5.5showthatthemeanpulmonaryartery pressuremPAPissignicantlyhigherHPX3thanintheCNTLCNTL=21.86 3.5mmHg, HPX3=35.32 6.2mmHg, p < 0.05.Doublingthelengthofhypoxicexposuredoes notcausefurtherincreasesinmPAPHPX6=34.61 2.4mmHg.After6weeksrecoveryinnormoxiathemPAPreturnstonormallevelsandisnotdierentfromthe controlREC=21.2 2.6mmHg,RECNTL=16.56 1.4mmHg, p =0.91.Heartrate, strokevolumeandcardiacoutputshownosignicantchangeamongthegroups. 74

PAGE 89

5.4.3MechanicalMeasurements TheresultsoftheP-DexperimentsareshowninFigure5.3.Theuppercurves representthepassiveresponseandthelowercurvesrepresentingtheactiveresponse. Arstapproximationindicatesthesetwocurvesareparallel.However,acloser examinationrevealsawideningofthegapbetweenthetwocurvesinthemid-range. Subtractingthepassivefromtheactivecurvewegetameasureofthecontribution ofthevascularsmoothmusclewhich,ifexpressedasabsolutediameterdierencevs. pressurehasapeakedappearanceascanbeseeninFigureFandisfurtherexplained below. Lookingatdierencesbetweenthegroups,themeanactivediameteroftheHPX3 groupisreducedbutnotsignicantlydierentfromthecontrolgroup.Bydoubling thehypoxicexposuretheHPX6diameterreductionissignicantCNTL=1.81 0.22, HPX3=1.59 0.10mm p =0.55,HPX6=1.36 0.1mm p < 0.05.Intherecoverygroup vesseldiametersareclosetocontrollevelsRECNTL=1.71 0.14,REC=1.55 0.09 mm p =0.22.Themeandiameterofthepassivevesselsfollowsasimilarpattern. Thepassivevesseldiametersarereducedalbeitnotsignicantlyafter3-weekshypoxic exposure.Inthe6-weekhypoxicgroupthediameterreductionbecomessignicant CNTL=2.07 0.27,HPX3=1.71 0.15mm p =0.09,HPX6=1.56 0.08mm p < 0.05. Intherecoverygroup,vesseldiametersareatnormallevelsRECNTL=1.97 0.08 mm,REC=1.89 0.12mm p =0.99. ThefulldiameterrangeofthepassivevesselsinCa ++ -freeKHSissignicantlyreducedinthehypoxicgroupsCNTL=1.22 0.19mm,HPX3=0.49 0.12 mm p < 0.05,HPX6=0.54 0.10mm p < 0.05.Thefulldiameterrangefortherecoverygroupremainsimpaired,notsignicantlydierentfromthehypoxicgroups butsignicantlydierentfromtherecoverycontrolgroupREC=0.68 0.08mm, RECNTL=0.90 0.11mm p =0.04,HPX3 p =0.05.Thehypoxicdiameterrange oftheactivevesselsinHi-KKHSissignicantlyreducedCNTL=1.37 0.22mm, 75

PAGE 90

HPX3=0.46 0.14mm p < 0.05,HPX6=0.55 0.09mm p < 0.05.Theactivevessel diametersreturntonormallevelsuponrecoveryinnormoxiaREC=0.80 0.11mm, RECNTL=0.98 0.12mm p < 0.05. ExaminingthediameteratthehemodynamicallymeasuredMPAPforeachanimal,themeandiameterofthecontrolgroupis1.46 0.16mmat21.9mmHg.The diameterofthe3-weekhypoxicgroupis1.62 0.12mmatmPAP=34.6, p < 0.05. 5.4.4PACompliance VesselcomplianceFigure5.3recoversslightlybutnotcompletelyinthereturngroupasitisdierentfromboththehypoxicandcontrolgroups.Within eachgroup,vesselcompliancewithVSMactivationisthesameascompliancewithoutVSMactivationasindicatedbytheparallelP-Dcurves.However,thecompliancechangessignicantlybetweengroups.The3-weekhypoxicgroupislesscompliantthanthecontrolgroupCNTL=0.03 0.006,HPX3=0.011 0.003mmHg/mm, p < 0.05.The6-weekhypoxicgroupisnolesscompliantthanofthe3-weekhypoxicgroupHPX3=0.011 0.003,HPX6=0.012 0.002mmHg/mm, p =0.99,suggestingthevesselwallsdonotstienfurtherwhenthetimeofexposuretohypoxiaisdoubled.After6-weeksrecoveryinnormoxiatheactivevesselsincrease incomplianceandapproachtimematchedcontrolsREC=0.016 0.004mmHg/mm, RECNTL=0.021 0.002, p =0.46.Incontrast,thepassivevesselswithoutVSMactivationwereslightlymorecompliantthanthehypoxicgroups,however,arenotsignicantlydierentfromthehypoxicgroups p =0.07.Theseresultsaregraphically representedinFigure5.3. 5.4.5AbilityofVascularSmoothMuscletoAlterDiameter TheabilityofthePAtochangediameterisanindicationofhowwelltheVSMis abletocontractandisillustratedinFigure ?? .Wendthecontrolgroupisable 76

PAGE 91

tochangediameterbyupto0.48mm.Withhypoxicexposure,Dissignicantly reducedbymorethan50%ofnormallevelsCNTL=0.48 0.18,HPX3=0.16 0.10 mm p < 0.05,HPX6=0.24 0.08mm p < 0.05.Somefunctionhasadaptedby6-weeks ofhypoxicexposure,thoughstillsignicantlydecreased.Dreturnstonormallevels intherecoveredgroupRECNTL=0.36 0.06,REC=0.45 0.09mm p =0.73.The plotsinFigure5.4aresuggestiveoftheslidinglamenttheory,showingthemaximum abilityoftheSMtocontractatapproximately30.4 5.2mmHgforeachgroup.At pressuressignicantlyhigherandlowerthan30.4mmHg,Ddecreases,suggesting reducedcontractileabilityoneithersideofthepeakfunction. 5.4.6PAWallTension ThemeanwalltensionforeachstudygroupisillustratedinFigure5.3,bottompanel.PAWallStresswithCa ++ -KHSactivationissignicantlyreducedwith hypoxicexposureCNTL=76.43 6.07kPa,HPX3=30.50 4.72kPa, p < 0.05and HPX6=26.73 2.42kPa, p < 0.05andremainssignicantlyreduceduponrecoveryin normoxiaRECNTL=47.71 2.56kPa,REC=32.80 4.90kPa, p =0.02.ThemaximumwalltensionwithCa ++ -KHSactivationfollowsthesamepattern:reducedwith hypoxicexposureandinrecoveryCNTL=205.78 40.76kPa,HPX3=65.77 13.30 kPa, p < 0.05,HPX6=59.67 6.32kPa, p < 0.05,REC=81.40 11.74kPa, p =0.02.In addition,thewalltensionoftherecoverycontrolsthatwerekeptinnormoxiafor 9weeksweresignicantlylowerthantheyoungercontrolsthathadonlyspent3 weeksinnormoxiaelevation=5285ft,pO 2 =132.5 p < 0.05.Theapproximationof thethin-walledcylinderwassignicantlylowerthantheexactequationforcircumferentialwall-stressinallgroups p < 0.05AppendixE. Wecalculatedwallstressesat21mmHgand35mmHgrepresentingmeanoperating pressuresofthenormoxicandhypoxicgroupsrespectively.Wallstresswassignicantlylowerinthehypoxicgroupsatagivenpressure p < 0.05.However,wealso 77

PAGE 92

comparedthecontrolgroupsattheiroperatingpressure,21mmHg,tothehypoxic groupsattheirmPAP,35mmHg.Atthenormoxicoperatingpressureof21mmHg thewallstresseswere24.50 5.22kPaand17.68 0.96kPAfortheControlandRecoveryControlgroupsrespectively.Atthehypoxicoperatingpressureof35mmHg thehypoxicarterieswere31.00 4.12kPa,27.00 2.39kPaand31.07 5.01kPafor HPX3andHPX6groupsrespectively.Thereisthenamis-matchwiththerecoverygroupwhentheoperatingpressurereturnstonormoxiclevelsandthewallstress at21mmHgremainsdecreasedandsignicantlydierentfromtherecoverycontrol groupREC=12.46 1.90kPa,RECNTL=17.68 0.96kpa, p < 0.05. 5.4.7Histology HistologicalanalysisandquanticationofH&Estainedarterytissueindicates nochangeinthenumberofnucleiperunitareanohyperplasiaamongallgroups. HistologicalanalysisofmuscleareaoftheMasson'sTrichromestainedtissueshownin Figure5.7alsoshowsnohypertrophyamongthevegroupsCNTL=33.3 10.7 m 2 HPX3=36.7 11.7 m 2 p =0.99,HPX6=38.0 7.5 m 2 p =0.98,RECNTL=40.6 9.7 m 2 ,REC=41.7 10.4 m 2 p =0.99.Thecollagentissueareadoubleswithhypoxic exposureCNTL=48.36 8.83 m 2 ,HPX3=81.9 17.2 m 2 p < 0.05,HPX6=91.6 12.4 m 2 p < 0.05,remainsdoubledwith6-weekhypoxicexposure,andremainsdoubled intherecoverygroupRECNTL=47.4 11.4 m 2 ,REC=90.3 15.6 m 2 p < 0.05. QuanticationofelasticberareafractionofVerhoe'sElasticstainedtissueshows nodierenceamongthevegroupsCNTL=25.1 6.3 m 2 ,HPX3=19.0 8.5 m 2 p =0.99,HPX6=18.9 4.9 m 2 p =0.98,RECNTL=26.9 9.2 m 2 ,REC=18.0 9.1 m 2 p =0.99Figure5.7.ThetissuethicknesswasalsomeasuredfortheMasson'sTrichromestainedsections.Themediathicknessdidnotchangeamong theveexperimentalgroupsCNTL=46.2 7.7mm,HPX3=56.8 9.3mm p =0.25, HPX6=59.5 7.7mm p =0.15,REC=60.0 5.3mm,RECNTL=58.9 7.0mm p =0.99. 78

PAGE 93

Theadventitiathicknessincreasedwithhypoxicexposureandremainedincreasedin recoveryCNTL=41.3 8.7mm,HPX3=77.2 6.5mm, p < 0.05HPX6=78.0 7.4mm, p < 0.05,REC=83.2 17.1mm,RECNTL=55.4 12.9mm p < 0.05. 5.5Discussion Withthisstudyweareabletoanswerthreemainquestionsforeachmeasured propertyofPAmechanicsandhemodynamics:Howhypoxiaaectseachproperty andisthiswithintherangeofpriorstudies,howmucheachpropertyrecoversfrom hypoxia,andhowrecoveryvesselscomparetovesselsthathavenothadhypoxic exposure. Theimportantndingsofthisstudyare: i After6weeksofrecoveryfollowing 3weekshypoxia,thecomplianceoftheproximalPAremainsimpairedsuggesting ahighlyasymmetricprocess; ii Themeanvesseldiametercontinuestodecrease through6weeksofhypoxicexposureandthemaximumdiameterremainsreduced after6weeksofrecovery; iii Hi-KKHSinducedVSMcontractionissignicantlyreducedafter3weeksofhypoxia.Contractionpartiallyrecoversafterafurther3weeks ofhypoxia,suggestingsomeformofhypoxicadaptationisoccurring.Uponrecovery innormoxia,VSMcontractionfullyreturnstonormallevels; iv MaximumVSM contractionwithHi-KKHSoccursatameanpressureof30.4 5.2mmHgregardlessof environmentalexposure. v WallstressintheproximalPAoverthepressurerange of5-55mmHgissignicantlyreducedwithhypoxicexposureandremainsreduced uponrecovery. 5.5.1VesselProperties ThenearlyparallelshiftintheP-DcurvesofvesselsunderallconditionsFigure 5.3suggeststhattheactionofsmoothmuscleissimplytochangethediameterofthe vesselwithoutsignicantlyalteringthecomplianceofthevesselwall.Thesimilarity 79

PAGE 94

oftheactiveresponsetothepassiveresponseisconsistentwithpassiveelementsdominatingtheoverallresponse.Thiswouldbealignedwiththesmoothmusclehavinga muchhigherstinessthanthepassiveelementsunderbothconditions.Activearterial smoothmusclecangeneratepeakstressesofapproximately2.5x10 5 N/m 2 whichis higherthanthewallstressunderourexperimentalconditionsandwouldallowthe smoothmuscletoactasarigidconnection[78].Underpurelypassiveconditions, thesmoothmusclecouldextendalongit'spassivelengthtensioncurveuntilitalso becameanearlyrigidconnectionbetweenpassiveelements,whichagaindominate thevesselresponse. ThemeandiameterofthepassivePAatvariouspressuresisshowninFigure 5.6.Ascanbeseen,thediameterdecreaseswithhypoxiaandrecoversonreturnto normoxia,howeverthetimecourseisextremelyslow.Thesearechangesinpassive propertiesandsuggestthatthecollagenlengthtensioncurveisshiftingtoshorter lengthsandsmallerdiameters.Howthisoccursremainsapuzzle.Onetheoryisthat hypoxicvasoconstriction in-situ resultsinsmallerdiametersallowingnewcollagento belaiddownatshortermeanlengths.However,measurementsofthePAinsituat thepathologicalmPAPshowthatthePAincreasesindiameterratherthandecreases [111]. 5.5.2Hemodynamics Wedemonstratethatwhilethehemodynamicprolereturnstonormallevels, indicatingthattheheartisfunctioningnormally,themechanicsoftheproximalpulmonaryartery,includingvesseldiameter,complianceandwallstressremainimpaired uponrecoveryfromPHFigure5.5,panelsa,b,corAppendixE.Further,fullrecoveryofthesmoothmusclecontractionsuggeststhatadoublingofthearterythickness andtheamountofcollagendoesnotinitselfaectcontractileability. 80

PAGE 95

5.5.3PHDevelopment OurCntlanimals,despitebeingmildlyhypoxicatDenveraltitudes,hadmPAP's withintherange16-22mmHg,whichwereconsistentwithpriorstudiesofhypoxiainducedPHinrats.Theseanimalshaveincreasedvesselstiness,increasedartery thicknessandincreasedcollagendepositions[72][87][103].Ourresultsagreewith previousndingsofnormoxicmPAPvalueslyingbetween16and22mmHg.The resultsinthisstudyareconsistentwithpublishedhemodynamicmeasurementsof mPAPinthehypoxicratbeingabout35.5 2mmHg[18].Inthedevelopmentof PH,ourndingssupportpriorstudiesshowingtheproximalvesselsnotonlystien butdecreaseindiameter,decreasewallstressandincreaseinelasticmodulus.We alsoshowthattheproximalvesselslosesomecontractilefunctionasevidencedbya reductionintheabilityofactivecontractiontoshortenthevesselinhypoxiaFigure 5.4.Areturntonormoxiaprovidesafullrecoveryinfunction.Hypoxiaisknown toaectK-channelfunctionandthismayberesponsibleforthediminishedfunction [108][117]. 5.5.4RecoveryfromHypoxicExposure Previousstudiesofrathypoxicrecoverymodelshavefoundawiderangeofresults fromcompleterecoverytopersistentchangesduetoremodelingasaresultofPH, suggestingthefactorsthatdeterminerecoveryarenotfullyunderstood.After10days ofhypoxicexposureat10%O2andpressuresbetween74-80mmHgandnormoxia upto10daysfoundcollagenandelastincontentsatnormallevels[37].Afuller descriptionwouldindicatewhathappenedtocollagenandelastindurignhypoxia. Hyoxicexposureat380mmHgexposurefor10daysdoubledtheadventitiathicknesswithincreasedbroblastsandcollagenbers.During70daysofrecoveryinroom air,theadventitiathicknessreturnedtonormalbutthecollagenberconcentration remainedincreased[72].Thissuggeststhatthebreakdownoftheadditionalcollagen 81

PAGE 96

isaslowprocess. HislopandReidfoundthatRVhypertrophyresolved,butthelossofsmallarteriesarterialpruningwithouterdiameters 200 mdidnotrecoverfromhypoxic exposureof2weeksat380mmHgandrecoveryupto8weeksinroomair[43].With 10%O2hypoxicexposureupto3weeksandrecoveryinroomairupto20weeks,HergetfoundthatmuscularizationandarterialpruningpersistedwhilemPAPreturned tonormal[41].Finally,with5weeksofexposureto380mmHgandupto5weeksof recovery,HeathfoundnormalizationofbothmedialthicknessandRVhypertrophy [40]. At3weeksofhypoxicexposureand6weeksofrecoveryinnormoxia,ourstudy fallswithinthetime-framesofpreviousrecoverystudies.Thus,weexpectedcollagen areaandadventitiathicknesstodecreaseuponrecoveryinnormoxia.Collagenmay haveremainedincreasedbecausenormoxiaforthisstudywasatanelevationof5285 ft.Wefoundnosignicantchangesinmediathicknessorarea,nordidwendchanges inelastinarea.OurhemodynamicmeasurementssupportmPAPandRVrecovery. Withinthecontextofthesestructuralandhemodynamicchanges,weexaminethe vesseldiameter,D,vesselcomplianceandwalltensiontogaininsightintothePA mechanicsuponrecoveryfromPH. 5.5.5Limitations Aspreviouslymentionedweallowedthevesseldiametertofullycontractorrelaxoveralongtimeintervalwitheachpressuresteppriortotakingmeasurements. Arterycompliancehasatimecomponentanddependsontherateofchangeindiameter.Thevesselappearstobelesscompliantwithfasterchangestovesseldiameter. Alternatively,thevesselappearsmorecompliantwithslowerdiameterchanges.This viscoelasticeecthelpsprotecttheartery,attenuatinghigh-frequencypulsesanddissipatingthepulsatileenergyfromtheheart,alsoactingasanelasticreservoirand 82

PAGE 97

secondstagepumpfortheheart[96].Thisstudyremovesasmuchoftheviscoelastic eectaspossibleinordertoisolatethesteadystateelasticandcontractileconditions ofthearterywall,thusneglectingthedynamicandpulsatilefunctionsofthePA. Thevesselswerepressurizeduptoapproximately55mmHgandnotabove 60mmHg.ThiswasnecessarytoensurefullforcedevelopmentoftheVSMthroughout thetesting.Occasionally,ifpressuresexceeded65mmHg,theVSMcontractionwas decreased,oftenfollowingthecurvesofthepassivevesselcurves,andthetestwould notbeused.Inthesecases,theVSMmayhavebeenstretchedbeyondthelimits foractinandmyosinattachment.Thesedataandtestswhilebeyondthescopeof thecurrentproject,arepotentiallysupportedfromunpublisheddatacurrentlyunder study. Additionally,theperistalticpumpallowedustosettheowrateortosetthe pressure.Withpressureset,weapproximatedtheowrateusingaseparatereservoir overagiventimeperiod.Lastly,thestudylacksbiochemicaldatatomoreprecisely quantifythevesselproteins,relyinginsteadonPAmorphologytovisualizethechanges incellularstructure. 5.5.6Conclusions Insummary,wereportthatproximalPAsegmentsfromratsthathavechronic hypoxia-inducedPHandareallowedtorecoverforsixweeksinnormoxiahaveimpairedmechanics:decreasedvesseldiameters,decreasedvesselcomplianceanddecreasedwalltensionascomparedwithagematchedcontrols.Wealsofoundwith hypoxicexposurethePAsegmentshavedecreasedVSMcontractioninHikKHS solutionandthatVSMcontractionreturnstonormallevelsduringrecovery.These ndingssuggestthatproximalVSMcontractionisnotaectedbycollagendeposition whichpersistsinrecovery,norbytheothervesselmechanicswhichremaindecreased. Wespeculatethattherapiesaimedatreducingorpreventingcollagendepositionwill 83

PAGE 98

improveproximalPAmechanicsuponrecoveryfromPHoruponnormalizationofthe hemodynamics. 5.6Figures Figure5.1:ExperimentalDesign. Figure5.2:Pressure-Diameterpairedpointsandrecords. 84

PAGE 99

Figure5.3:PADiameterandWallStress. 85

PAGE 100

Figure5.4:AbilityofVascularSmoothMuscletoChangeDiameter,D,through PHandRecovery. 86

PAGE 101

Figure5.5:mPAP,CO,PVR,wallstress,compliance,diameter Figure5.6:MeanDiameterChangeswithPressureandGroup. 87

PAGE 102

Figure5.7:StructuralChangestothePAMeasuredbyTissueAreasofHistological Sections. 88

PAGE 103

Chapter6 6.Conclusions 6.1MajorFindings Thisdissertationcontainsaseriesofstudiesthatmeasuredandcharacterized mechanicsofthePPAthroughanin-vivoclincalsettingandin-vitroanimalstudies. Inbothsettings,changesinthePPAmechanicsasaresultofchronicPHmaterially alterthecompositionoftheartery.However,theexactmechanismsresultinginthis arestillnotcompletelyunderstood.Assuch,wehavecombinedthesetwosettings hopingthattheintersectionoftheseperspectiveswillshedlightonoptionsforfuture studiesandpotentialtherapeutictreatmentstosupplementcurrentstandards. Themajorndingsofthesestudiesinclude: WHOFunctionalClassicationmaybepredictedayearinadvancewith76% accuracybyutilizingmultivariateequationequation3.1whichincludesthe measures:sDiam,MPAP,MPAP/AoP,SBP,suggestingproximalanddistal mechanicalchangesareimportantindicatorsoffuturehealthstatus. Ofthetop20modelstopredictfutureWHOFC,17ofthosemodelsincluded thediameterofthePPAinsomeform:strain,dynamiccompliance,systolicor diastolicdiameter.ThisisanindicationofvesselstinessandPVRaspredictors ofhumanPHprogression. IndividualindicatorsofsystolicpressureinthePAandtheratioPVR/SVR predictWHOfunctionalclassayearinadvancewith68%and67%accuracy respectively.Bycomparison,PVRIpredictsfuturefunctionalclassicationwith 59%accuracy. 89

PAGE 104

Inanimalstudiesofsofttissueusingpressure-inationtesting,preconditioningsignicantlyaltersmechanicalmeasurementssuchasdiameter,compliance, stress,strainandelasticmodulus. InanimalstudiesofrecoveryfromhypoxiainducedPHthecomplianceofthe PPAremainsimpairedsuggestingahighlyasymmetricprocess. { Themeanvesseldiametercontinuestodecreasethrough6weeksofhypoxic exposure. { Themaximumdiameterremainsreducedafter6weeksofrecovery. { TheVSMabilitytodecreasevesseldiameterissignicantlyreducedafter 3weeksofhypoxia,partiallyrecoversafterafurther3weeksofhypoxia, suggestingsomeformofhypoxicadaptationisoccurring. { Uponrecoveryinnormoxia,VSMabilitytodecreasevesseldiameterfully returnstonormallevels. { Finally,wallstressintheproximalPAoverthepressurerangeof555mmHgissignicantlyreducedwithhypoxicexposureandremainsreduceduponrecovery. 6.2ClinicalRelevance TheworkpresentedinthisdissertationdemonstratestheimportanceofmechanicalpropertiesofthePPAtoguideclinicaltherapyinPH.Specically,a model/calculationthatcouldbeutilizedinaclinicalsettingtosupportproviders whendeningcurrentandfuturetherapeuticneeds.Inaddition,theanimalstudiesshedlightonthemechanicsofthePPA,especiallyinthecontexttheinterplay betweencollagenandsmoothmusclefunction.ThisopensavenuesforfurtherunderstandingthealtereddepositionofcollageninPH.Underlyingthisisthepotential 90

PAGE 105

forfuturetherapiestoregulatecollagendepositionandencourageareturntonormal levelsforpotentiallongertermtreatments. 6.3ScopewithintheExistingLiterature PreviousstudieshaveshownuniquedierencesinpediatricPHcomparedtoadult PHwithrespecttoetiology,symptoms,andpresentation.Pediatricdiseasethus requirescustomizedpredictiveclinicalindicators[23][8][9][120][121].PVRisoftenthe singleparameterusedtoassessdiseaseseverityandresponsetotreatment[7][121][24]. Yet,othermarkersofvascularfunction,forexamplepulmonaryvascularstiness, areastrainandvascularcapacitancehavebeenshownaspredictiveofmortalityin PH.[36][51][15][74].Numerouspriorstudieshavesuggestedutilizationofvarious combinationsofindividualclinicalindicatorsaspredictiveofsurvival,mortalityand outcomesforPH[12][10][109][11][23]. Thestudiesinthisthesisaddtothegrowingevidencesuggestingtheimportance ofPPAmechanics.MeasuresofPPAdiameter,strainanddynamiccomplianceincreasethepredictionofPHoutcomesthroughacomprehensiveevaluationshowing incrementalincreasesinpredictivevaluefromanarrayofindividualindicators.We thenexaminethemechanicsofthePPAwallwithin-vitroanimalstudiestoanswer questionsregardingthetimecourseofmechanicschangesandtheirpotentialresolution.Thisallowedafurtherunderstandingoftheunderlyingmechanicalchanges involvedwiththemeasuresidentiedintheclinicalstudy.Thereforethisworkadds tothebodyofliteratureregardingtheinuencesofPPAmechanicsinpredictingthe impactsofdiseaseprogressioninPH. 91

PAGE 106

6.4FutureStudies InadditiontoansweringquestionsregardingPPAmechanics,theworkpresented herealsoraisesnewquestionsandsuggestsnewavenuesforcontinuedstudy.We wishtomakeamorecomprehensivestudyofthenon-invasivemeasuresandalso toincorporatenewclinicalmarkersintotheassessmentoffuturepatientoutcomes. Clinicalmeasurementssuchasbrainnatriureticpeptideandtricuspidannularplane systolicexcursionasrecentlysuggestedbyPloegstraetal.areshowntobeappropriatetreatmentgoals[84][85][86]andmayfurtherenhanceandimprovetheprediction ofpediatricPH. Wehopetoexpandourexaminationofpreconditionedtissuetosystemicvessels whichhavedierentratiosofsmoothmuscleandcollagencomparedtothepulmonary arteries.Inadditionwewishtocompareamongtissueshavingundergonelevelsof maximumloadingwithinthephysiologicalrange,abovethephysiologicalrange,and withoutpreconditioning.Finally,thetissueswillbeanalyzedtodeterminechanges withintheinternalcellularandmolecularstructure. InthethirdstudywetestedattheboundariesofVSMactivation:completely activatedandcompletelyinactivated.WewishtocontinueanexaminationofVSM alterationswithinchronicPHinamidrangeofactivationusingthesamephysiologic solutionswithdierentconcentrationsofpotassiumtogetanintermediatelevelsof activationbythesmoothmuscle.Inadditionwehopetoexpandthestudytoinclude pulsatileow.Wealsoaimtowardananimalmodelofincreasedcollagenaseactivity withinarecoverystudy. 6.5ConcludingRemarks ThedetectionandanalysisofmeaningfulpatternsinPHoutcomesfromlarge amountsofdataisacriticalstep.Continuedanalysisofthetopindividualpredictors, 92

PAGE 107

eitheraloneorincombination,willensureourabilitytoreinforceourresultswhich suggestthattheequationwiththemostpredictiveaccuracyoffutureWHO-FCis: sDiam,MPAP,MPAP/AoP,SBP-PAaccordingtoequation3.1.Thissuggeststhat theincreasesinaccuracyareincrementalinnatureandthegeometryofthePPA providessynergisticpredictiveaccuracy.Astheclinicalmeasurescontinuetoevolve, sowilltheequationandourabilitytopredictoutcomes. Inanimalstudiesofsofttissueusingpressure-inationtesting,samplepreparationmaysignicantlyaecttheresultingmechanicsmeasurements.Forinstance, prolongedtissuedissectionandpreparationmayleadtoincreasesinstinessand overstretchingthetissueduringpreparationmaydecreasethestiness.Caremust betakenintissuepreparationandtestinginordertopreservethemechanicalcharacteristicsascloselyaspossibletoin-vivosettings. ProximalPAsegmentsfromratsthathavechronichypoxia-inducedPHandare allowedtorecoverforsixweeksinnormoxiahaveimpairedmechanics:decreasedvesseldiameters,decreasedvesselcomplianceanddecreasedwalltensionascompared withagematchedcontrols.Wealsofound,withhypoxicexposure,thePAsegments havedecreasedVSMcontractionandthatVSMcontractionreturnstonormallevels uponrecovery.ThesendingssuggestthattheabilityoftheproximalVSMtocontractreturnsdespitecollagendepositionandotheralteredmechanicsthatremain. Wespeculatethattherapiesaimedatreducingorpreventingcollagendepositionmay improveproximalPAmechanicsuponrecoveryfromPHoruponnormalizationofthe hemodynamics. 93

PAGE 108

REFERENCES [1]BruceAlberts,AlexanderJohnson,JulianLewis,MartinRa,KeithRoberts, andPeterWalter. MolecularBiologyoftheCell .GarlandScience,5edition, 2008. [2]SilviaM.Arribas,AleksanderHinek,andM.CarmenGonzlez.Elasticbresandvascularstructureinhypertension. Pharmacology&Therapeutics 111:771{791,September2006. [3]DavidBBadesch,StevenHAbman,GeraldSimonneau,LewisJRubin,and VallerieVMcLaughlin.Medicaltherapyforpulmonaryarterialhypertension: updatedACCPevidence-basedclinicalpracticeguidelines. Chest ,131:1917{ 1928,June2007. [4]DavidB.Badesch,HunterC.Champion,MiguelAngelGomezSanchez,MariusM.Hoeper,JamesE.Loyd,AlessandraManes,MichaelMcGoon,Robert Naeije,HorstOlschewski,RonaldJ.Oudiz,andAdamTorbicki.Diagnosis andAssessmentofPulmonaryArterialHypertension. JAmCollCardiol 54s1:S55{S66,June2009. [5]R.J.Barst,S.I.Ertel,MauriceBeghetti,andD.D.Ivy.Pulmonaryarterial hypertension:acomparisonbetweenchildrenandadults. EuropeanRespiratory Journal ,37:665{677,2011. [6]RobynJBarst,MichaelMcGoon,AdamTorbicki,OlivierSitbon,MichaelJ Krowka,HorstOlschewski,andSeanGaine.Diagnosisanddierentialassessmentofpulmonaryarterialhypertension. JAmCollCardiol ,43s1:S40{S47, June2004. [7]RobynJ.Barst,MichaelD.McGoon,C.GregoryElliott,AimeeJ.Foreman, DaveP.Miller,andD.DunbarIvy.SurvivalinChildhoodPulmonaryArterial HypertensionInsightsFromtheRegistrytoEvaluateEarlyandLong-TermPulmonaryArterialHypertensionDiseaseManagement. Circulation ,125:113{ 122,January2012. [8]RobynJBarst.Pulmonaryhypertension:Past,presentandfuture. Annalsof ThoracicMedicine ,3:1,2008. [9]RobynJBarst.Classicationofpediatricpulmonaryhypertensivevasculardisease:Doesitneedtobedierentfromtheadultclassication? Pulmonary Circulation ,1:134,2011. 94

PAGE 109

[10]RaymondL.Benza,MardiGomberg-Maitland,DaveP.Miller,AdaaniFrost, RobertP.Frantz,AimeeJ.Foreman,DavidB.Badesch,andMichaelD.McGoon.THerevealregistryriskscorecalculatorinpatientsnewlydiagnosed withpulmonaryarterialhypertension. Chest ,141:354{362,February2012. [11]RaymondL.Benza,DaveP.Miller,MardiGomberg-Maitland,RobertP. Frantz,AimeeJ.Foreman,ChristopherS.Coey,AdaaniFrost,RobynJ.Barst, DavidB.Badesch,C.GregoryElliott,TheodoreG.Liou,andMichaelD.McGoon.PredictingSurvivalinPulmonaryArterialHypertensionInsightsFrom theRegistrytoEvaluateEarlyandLong-TermPulmonaryArterialHypertensionDiseaseManagementREVEAL. Circulation ,122:164{172,July2010. [12]RolfMFBerger,MauriceBeghetti,TilmanHumpl,GaryERaskob,DDunbarIvy,Zhi-ChengJing,DamienBonnet,IngramSchulze-Neick,andRobynJ Barst.Clinicalfeaturesofpaediatricpulmonaryhypertension:aregistrystudy. Lancet ,379:537{546,February2012. [13]FredTBosmanandIvanStamenkovic.Functionalstructureandcomposition oftheextracellularmatrix. J.Pathol. ,200:423{428,July2003. [14]PierreBoutouyrie,CarolineBussy,PatrickLacolley,XavierGirerd,Brigitte Laloux,andStphaneLaurent.AssociationBetweenLocalPulsePressure,Mean BloodPressure,andLarge-ArteryRemodeling. Circulation ,100:1387{1393, September1999. [15]PierreBoutouyrie,AnneIsabelleTropeano,RolandAsmar,IsabelleGautier, AthanaseBenetos,PatrickLacolley,andStephaneLaurent.Aorticstinessis anindependentpredictorofprimarycoronaryeventsinhypertensivepatients alongitudinalstudy. Hypertension ,39:10{15,2002. [16]KennethP.BurnhamandDavidR.Anderson. ModelSelectionandMultimodelInferenceaPracticalInformation-TheoreticApproach .Springer-Verlag newYork,Inc.,seconded.2010edition,2010. [17]RCarbone,EBossone,GBottino,AMonselise,andMRubenre.Secondary pulmonaryhypertension{diagnosisandmanagement. Europeanreviewformedicalandpharmacologicalsciences ,9:331{342,2004. [18]MariaACavasin,KimDemos-Davies,ToddRHorn,LoriAWalker,DouglasD Lemon,NicholasBirdsey,MaryCMWeiser-Evans,JulieHarral,DavidCIrwin,AdilAnwar,etal.Selectiveclassihistonedeacetylaseinhibitionsuppresseshypoxia-inducedcardiopulmonaryremodelingthroughanantiproliferativemechanism. Circulationresearch ,110:739{748,2012. [19]NavdeepS.ChandelandPaulT.Schumacker.Cellularoxygensensingby mitochondria:oldquestions,newinsight. JournalofAppliedPhysiology 88:1880{1889,May2000. 95

PAGE 110

[20]StephenCorteenCowinandJayDowellHumphrey. Cardiovascularsofttissue mechanics .Springer,2001. [21]L.DebelleandA.M.Tamburro.Elastin:moleculardescriptionandfunction. TheInternationalJournalofBiochemistry&CellBiology ,31:261{272, February1999. [22]P.B.Dobrin.Mechanicalpropertiesofarteries. PhysiologicalReviews 58:397{460,1978. [23]JohannesM.Douwes,RosaL.E.vanLoon,ElkeS.Hoendermis,AntonVonkNoordegraaf,MarcusT.R.Roofthooft,MelleD.Talsma,HansL.Hillege,and RolfM.F.Berger.Acutepulmonaryvasodilatorresponseinpaediatricand adultpulmonaryarterialhypertension:occurrenceandprognosticvaluewhen comparingthreeresponsecriteria. EurHeartJ ,32:3137{3146,December 2011. [24]JohannesM.Douwes,MarcusT.R.Roofthooft,BeatrijsBartelds,MelleD. Talsma,HansL.Hillege,andRolfM.F.Berger.Pulsatilehaemodynamicparametersarepredictorsofsurvivalinpaediatricpulmonaryarterialhypertension. InternationalJournalofCardiology ,168:1370{1377,September2013. [25]KarrieDyer,CraigLanning,BibhutiDas,Po-FengLee,D.DunbarIvy,LilliamValdes-Cruz,andRobinShandas.NoninvasiveDopplerTissueMeasurementofPulmonaryArteryComplianceinChildrenwithPulmonaryHypertension. JournaloftheAmericanSocietyofEchocardiography ,19:403{412, April2006. [26]KarimaEt-Taouil,MichelSafar,andGrardEPlante.Mechanismsandconsequencesoflargearteryrigidity. Can.J.Physiol.Pharmacol. ,81:205{211, March2003. [27]U.S.v.EulerandG.Liljestrand.ObservationsonthePulmonaryArterial BloodPressureintheCat. ActaPhysiologicaScandinavica ,12:301{320, August1946. [28]GFaury.Functionstructurerelationshipofelasticarteriesinevolution:from microbrilstoelastinandelasticbres. PathologieBiologie ,49:310{325, 2001. [29]peterfratzl. Collagen:StructureandMechanics .SpringerUS. [30]AdaaniE.Frost.TheChangingPictureofPatientsWithPulmonaryArterial HypertensionintheUnitedStates:HowREVEALDiersFromHistoricand Non-USContemporaryRegistries. CHESTJournal ,139:128,January2011. [31]YCFung.Biomechanics:materialpropertiesoflivingtissues,1993. 96

PAGE 111

[32]Y.C.Fung. BIOMECHANICS:CIRCULATION,2NDEDITION. Springer US,2edition,1997. [33]YCFung,KFronek,andPPatitucci.Pseudoelasticityofarteriesandthe choiceofitsmathematicalexpression. AmericanJournalofPhysiology-Heart andCirculatoryPhysiology ,237:H620{H631,1979. [34]NazzarenoGali,HosseinA.Ghofrani,AdamTorbicki,RobynJ.Barst,LewisJ. Rubin,DavidBadesch,ThomasFleming,TamizaParpia,GaryBurgess,AngeloBranzi,FriedrichGrimminger,MarcinKurzyna,andGraldSimonneau. SildenalCitrateTherapyforPulmonaryArterialHypertension. NewEngland JournalofMedicine ,353:2148{2157,2005. [35]NazzarenoGali,MariusM.Hoeper,MarcHumbert,AdamTorbicki,Jean-Luc Vachiery,JoanAlbertBarbera,MauriceBeghetti,PaulCorris,SeanGaine, J.SimonGibbs,MiguelAngelGomez-Sanchez,GuillaumeJondeau,WalterKlepetko,ChristianOpitz,AndrewPeacock,LewisRubin,MichaelZellweger,GeraldSimonneau,AlecVahanian,AngeloAuricchio,JeroenBax,ClaudioCeconi,VeronicaDean,GerasimosFilippatos,ChristianFunck-Brentano, RichardHobbs,PeterKearney,TheresaMcDonagh,KeithMcGregor,BogdanA.Popescu,ZeljkoReiner,UdoSechtem,PerAntonSirnes,MichalTendera,PanosVardas,PetrWidimsky,UdoSechtem,NawwarAlAttar,FelicitaAndreotti,MichaelAschermann,RiccardoAsteggiano,RayBenza,Rolf Berger,DamienBonnet,MarionDelcroix,LukeHoward,AnastasiaN.Kitsiou, IreneLang,AldoMaggioni,JensErikNielsen-Kudsk,MyungPark,Pasquale Perrone-Filardi,SuzannaPrice,MariaTeresaSubiranaDomenech,AntonVonkNoordegraaf,andJoseLuisZamorano.GuidelinesforthediagnosisandtreatmentofpulmonaryhypertensionTheTaskForcefortheDiagnosisandTreatmentofPulmonaryHypertensionoftheEuropeanSocietyofCardiologyESC andtheEuropeanRespiratorySocietyERS,endorsedbytheInternational SocietyofHeartandLungTransplantationISHLT. EurHeartJ ,30:2493{ 2537,October2009. [36]C.Tji-JoongGan,Jan-WillemLankhaar,NicoWesterhof,J.TimMarcus,AnnemarieBecker,JosW.R.Twisk,AncoBoonstra,PieterE.Postmus,and AntonVonk-Noordegraaf.NOninvasivelyassessedpulmonaryarterystiness predictsmortalityinpulmonaryarterialhypertension*. Chest ,1326:1906{ 1912,December2007. [37]MaryGGeorge,LindaJSchieb,CarmaAyala,AnjaliTalwalkar,andShaleah Levant.Pulmonaryhypertensionsurveillance:Unitedstates,2001to2010. CHESTJournal ,146:476{495,2014. [38]S.E.Greenwald,J.E.Moore,A.Rachev,T.P.C.Kane,andJ.J.Meister. ExperimentalInvestigationoftheDistributionofResidualStrainsintheArtery Wall. JournalofBiomechanicalEngineering ,119:84{93,2010. 97

PAGE 112

[39]T.S.Hakim,A.Pedoto,D.Mangar,andE.M.Camporesi.Nitricoxideplays aminimalroleinhypoxicpulmonaryvasoconstrictioninisolatedratlungs. RespiratoryPhysiology&Neurobiology ,189:93{98,October2013. [40]D.Heath,C.Edwards,M.Winson,andP.Smith.Eectsontherightventricle,pulmonaryvasculature,andcarotidbodiesoftheratofexposureto,and recoveryfrom,simulatedhighaltitude. Thorax ,28:24{28,January1973. [41]J.Herget,A.J.Suggett,E.Leach,andG.R.Barer.Resolutionofpulmonary hypertensionandotherfeaturesinducedbychronichypoxiainratsduringcompleteandintermittentnormoxia. Thorax ,33:468{473,August1978. [42]GeorgeH.Hicks. CardiopulmonaryAnatomyandPhysiology .W.B.Saunders Company,2000. [43]A.HislopandL.Reid.Changesinthepulmonaryarteriesoftheratduringrecoveryfromhypoxia-inducedpulmonaryhypertension. BrJExpPathol 58:653{662,December1977. [44]MariusM.Hoeper,HarmJanBogaard,RobinCondlie,RobertFrantz,Dinesh Khanna,MarcinKurzyna,DavidLangleben,AlessandraManes,ToruSatoh, FernandoTorres,MartinR.Wilkins,andDavidB.Badesch.Denitionsand DiagnosisofPulmonaryHypertension. JAmCollCardiol ,62 S,December 2013. [45]GerhardA.Holzapfel.Collageninarterialwalls:biomechanicalaspects.In Collagen ,pages285{324.SpringerUS,2008. [46]GerhardA.Holzapfel,ThomasC.Gasser,andRayW.Ogden.ANewConstitutiveFrameworkforArterialWallMechanicsandaComparativeStudyof MaterialModels. JournalofElasticity ,61-3:1{48,July2000. [47]GerhardAHolzapfel,GerhardSommer,ChristianTGasser,andPeterRegitnig.Determinationoflayer-specicmechanicalpropertiesofhumancoronaryarterieswithnonatheroscleroticintimalthickeningandrelatedconstitutive modeling. AmericanJournalofPhysiology-HeartandCirculatoryPhysiology 289:H2048{H2058,2005. [48]WeiHuang,DavidDelgado-West,JonathanT.Wu,andYuanChengFung. TissueRemodelingofRatPulmonaryArteryinHypoxicBreathing.II.Course ofChangeofMechanicalProperteis. AnnalsofBiomedicalEngineering ,29:552{ 562,2001. [49]MarcHumbert,OlivierSitbon,andGeraldSimonneau.Treatmentofpulmonaryarterialhypertension. NewEnglandJournalofMedicine ,351:1425{ 1436,2004. 98

PAGE 113

[50]KendallS.Hunter,JosephA.Albietz,Po-FengLee,CraigJ.Lanning,StevenR. Lammers,StephenH.Hofmeister,PhilipH.Kao,H.JerryQi,KurtR.Stenmark,andRobinShandas.Invivomeasurementofproximalpulmonaryartery elasticmodulusintheneonatalcalfmodelofpulmonaryhypertension:developmentandexvivovalidation. JournalofAppliedPhysiology ,108:968{975, April2010. [51]KendallS.Hunter,Po-FengLee,CraigJ.Lanning,D.DunbarIvy,K.Scott Kirby,LoriR.Claussen,K.ChenChan,andRobinShandas.Pulmonaryvascularinputimpedanceisacombinedmeasureofpulmonaryvascularresistance andstinessandpredictsclinicaloutcomesbetterthanpulmonaryvascularresistancealoneinpediatricpatientswithpulmonaryhypertension. American HeartJournal ,155:166{174,January2008. [52]KendallSHunter,Po-FengLee,CraigJLanning,DDunbarIvy,KScottKirby, LoriRClaussen,KChenChan,andRobinShandas.Pulmonaryvascularinput impedanceisacombinedmeasureofpulmonaryvascularresistanceandstiness andpredictsclinicaloutcomesbetterthanpulmonaryvascularresistancealone inpediatricpatientswithpulmonaryhypertension. Americanheartjournal 155:166{174,2008. [53]AlexandraHyduk,JanetBCroft,CarmaAyala,KanZheng,Zhi-JieZheng, andGeorgeAMensah. Pulmonaryhypertensionsurveillance:Unitedstates, 1980-2002 .USDepartmentofHealthandHumanServices,2005. [54]HopeD.IntenganandErnestoL.Schirin.StructureandMechanicalPropertiesofResistanceArteriesinHypertensionRoleofAdhesionMoleculesand ExtracellularMatrixDeterminants. Hypertension ,36:312{318,September 2000. [55]HemantIshwaranandUdayaB.Kogalur.PackagerandomSurvivalForest. CRAN [56]D.DunbarIvy,StevenH.Abman,RobynJ.Barst,RolfM.F.Berger,Damien Bonnet,ThomasR.Fleming,SheilaG.Haworth,J.UshaRaj,ErikaB.Rosenzweig,IngramSchulzeNeick,RobinH.Steinhorn,andMauriceBeghetti.PediatricPulmonaryHypertension. JAmCollCardiol ,62 S,December2013. [57]JohannaMyllyharjuandKariI.Kivirikko.Collagensandcollagen-relateddiseases. AnnalsofMedicine ,33:7{21,2001. [58]KarlE.Kadler,ClairBaldock,JordiBella,andRaymondP.Boot-Handford. Collagensataglance. JCellSci ,120:1955{1958,June2007. [59]RyanW.Kobs,NidalE.Muvarak,JensC.Eickho,andNaomiC.Chesler. Linkedmechanicalandbiologicalaspectsofremodelinginmousepulmonary arterieswithhypoxia-inducedhypertension. AmericanJournalofPhysiologyHeartandCirculatoryPhysiology ,288:H1209{H1217,March2005. 99

PAGE 114

[60]BruceM.KoeppenandBruceA.Stanton. Berne&LevyPhysiology .Mosby Elsevier,6edition,2008. [61]Jan-WillemLankhaar,NicoWesterhof,TheoJCFaes,CTji-JoongGan, KoenMMarques,AncoBoonstra,FredGvandenBerg,PieterEPostmus, andAntonVonk-Noordegraaf.Pulmonaryvascularresistanceandcompliance stayinverselyrelatedduringtreatmentofpulmonaryhypertension. European heartjournal ,2008. [62]GJLaurent.Lungcollagen:morethanscaolding. Thorax ,41:418{428, 1986. [63]StphaneLaurent,PierreBoutouyrie,RolandAsmar,IsabelleGautier,Brigitte Laloux,LouisGuize,PierreDucimetiere,andAthanaseBenetos.AorticStinessIsanIndependentPredictorofAll-CauseandCardiovascularMortalityin HypertensivePatients. Hypertension ,37:1236{1241,May2001. [64]MatthewN.LevyandAchillesJ.Pappano. CardiovascularPhysiology .Mosby Elsevier,9edition,2007. [65]OliveLewis.Stephenhalesandthemeasurementofbloodpressure. Journalof humanhypertension ,8:865{871,1994. [66]LuisA.Martinez-Lemus,MichaelA.Hill,andGeraldA.Meininger.ThePlastic NatureoftheVascularWall:AContinuumofRemodelingEventsContributingtoControlofArteriolarDiameterandStructure. Physiology ,24:45{57, February2009. [67]FredericH.Martini. FundamentalsofAnatomyandPhysiology .PrenticeHall, 4edition,1998. [68]DonaldA.McDonald. Bloodowinarteries .CamelotPressLtd,Southampton, 2edition,1974. [69]MichaelMcGoon,DavidGutterman,VirginiaSteen,RobinBarst,DouglasC. McCrory,TerryA.Fortin,andJamesE.Loyd.Screening,earlydetection,and diagnosisofpulmonaryarterialhypertension*:Accpevidence-basedclinical practiceguidelines. Chest ,126 suppl:14S{34S,July2004. [70]VallerieV.McLaughlin,StephenL.Archer,DavidB.Badesch,RobynJ.Barst, HarrisonW.Farber,JonathanR.Lindner,MichaelA.Mathier,MichaelD.McGoon,MyungH.Park,RobertS.Rosenson,LewisJ.Rubin,VictorF.Tapson, andJohnVarga.ACCF/AHA2009ExpertConsensusDocumentonPulmonary HypertensionAReportoftheAmericanCollegeofCardiologyFoundationTask ForceonExpertConsensusDocumentsandtheAmericanHeartAssociation DevelopedinCollaborationWiththeAmericanCollegeofChestPhysicians; AmericanThoracicSociety,Inc.;andthePulmonaryHypertensionAssociation. JAmCollCardiol ,53:1573{1619,April2009. 100

PAGE 115

[71]VallerieVMcLaughlin,KennethWPresberg,RamonaLDoyle,StevenHAbman,DouglasCMcCrory,TerryFortin,andGregoryAhearn.Prognosisof pulmonaryarterialhypertension*:Accpevidence-basedclinicalpracticeguidelines. CHESTJournal ,126 suppl:78S{92S,2004. [72]B.MeyrickandL.Reid.Hypoxia-inducedstructuralchangesinthemediaand adventitiaoftherathilarpulmonaryarteryandtheirregression. AmJPathol 100:151{178,July1980. [73]WilliamR.Milnor.Hemodynamics.September1982. [74]GaryF.Mitchell,Shih-JenHwang,RamachandranS.Vasan,MartinG.Larson,MichaelJ.Pencina,NaomiM.Hamburg,JosephA.Vita,DanielLevy,and EmeliaJ.Benjamin.ArterialStinessandCardiovascularEventsTheFraminghamHeartStudy. Circulation ,121:505{511,February2010. [75]RohitMoudgil,EvangelosD.Michelakis,andStephenL.Archer.Hypoxicpulmonaryvasoconstriction. JournalofAppliedPhysiology ,98:390{403,January2005. [76]M.J.MulvanyandC.Aalkjaer.Structureandfunctionofsmallarteries. PhysiologicalReviews ,70:921{961,October1990. [77]R.A.Murphy.Whatisspecialaboutsmoothmuscle?Thesignicanceof covalentcrossbridgeregulation. FASEBJ ,8:311{318,March1994. [78]RAMurphy,JeremiahTHerlihy,andJ1Megerman.Force-generatingcapacityandcontractileproteincontentofarterialsmoothmuscle. TheJournalof generalphysiology ,64:691{705,1974. [79]RobertNaeije.Pulmonaryvascularresistance. Intensivecaremedicine 29:526{529,2003. [80]BobbyD.Nossaman,BrittniA.Scruggs,VaughnE.Nossaman,SubramanyamN.Murthy,andPhilipJ.Kadowitz.HistoryofRightHeartCatheterization:100YearsofExperimentationandMethodologyDevelopment. Cardiol Rev ,18:94{101,2010. [81]WorldHealthOrganization,InternationalSocietyofHypertensionWritingGroup,etal.2003worldhealthorganizationwho/internationalsociety ofhypertensionishstatementonmanagementofhypertension. Journalof hypertension ,21:1983{1992,2003. [82]WorldHealthOrganization,WorldHealthOrganization,etal. 1999World HealthOrganization-InternationalSocietyofHypertensionguidelinesforthe managementofhypertension .WHO,1999. 101

PAGE 116

[83]JPinheiro,DBates,SDebRoy,andDSarkar.Rdevelopmentcoreteam nlme:linearandnonlinearmixedeectsmodels.rpackageversion3.1-103. R FoundationforStatisticalComputing,Vienna ,2013. [84]Mark-JanPloegstra,SanneArjaans,WillemijnMHZijlstra,JohannesM Douwes,TheresiaRVissia-Kazemier,MarcusTRRoofthooft,HansLHillege, andRolfMFBerger.Clinicalworseningascompositestudyendpointinpediatricpulmonaryarterialhypertension. CHESTJournal ,2015. [85]Mark-JanPloegstra,JohannesMDouwes,MarcusTRRoofthooft, WillemijnMHZijlstra,HansLHillege,andRolfMFBerger.Identication oftreatmentgoalsinpaediatricpulmonaryarterialhypertension. European RespiratoryJournal ,pageserj00304{2014,2014. [86]Mark-JanPloegstra,WillemijnMHZijlstra,JohannesMDouwes,HansLHillege,andRolfMFBerger.Prognosticfactorsinpediatricpulmonaryarterial hypertension:Asystematicreviewandmeta-analysis. Internationaljournalof cardiology ,184:198{207,2015. [87]G.J.Poiani,C.A.Tozzi,S.E.Yohn,R.A.Pierce,S.A.Belsky,R.A.Berg, S.Y.Yu,S.B.Deak,andD.J.Riley.Collagenandelastinmetabolismin hypertensivepulmonaryarteriesofrats. CirculationResearch ,66:968{978, April1990. [88]RobinR.PrestonandThadE.Wilson. Physiology .LippincottWilliams& Wilkins,September2012. [89]R.R.PrestonandT.E.Wilson. Physiology .Lippincott'sIllustratedReviews Series.WoltersKluwerHealth,2012. [90]RCoreTeam. R:ALanguageandEnvironmentforStatisticalComputing .R FoundationforStatisticalComputing,Vienna,Austria,2013. [91]MARLENERabinovitch,WALTERJGamble,OLLISMiettinen,andLReid. Ageandsexinuenceonpulmonaryhypertensionofchronichypoxiaandon recovery. AmericanJournalofPhysiology-HeartandCirculatoryPhysiology 240:H62{H72,1981. [92]Rich,S,Rubin,LJ,Abenhail,L,andetal.ExecutiveSummaryfromthe WorldSymposiumonPrimaryPulmonaryHypertension1998Evian,France, September610,1998. [93]StephanRosenkranz.Pulmonaryhypertension2015:currentdenitions,terminology,andnoveltreatmentoptions. ClinicalResearchinCardiology 104:197{207,2014. [94]SylvainRoy,ChristopheBoss,RanaRezakhaniha,NikosStergiopulos,and S.Roy.Experimentalcharacterizationofthedistributionofcollagenberrecruitment. Journalofbiorheology ,24:84{93,2010. 102

PAGE 117

[95]LewisJRubin.Primarypulmonaryhypertension. NewEnglandJournalof Medicine ,336:111{117,1997. [96]DanielBiaSantana,JuanGabrielBarra,JuanCarlosGrignola,FernandoFlorencioGines,andRicardoLuisArmentano.Pulmonaryarterysmoothmuscle activationattenuatesarterialdysfunctionduringacutepulmonaryhypertension. JournalofAppliedPhysiology ,98:605{613,2005. [97]LauraleeSherwood. Humanphysiology:fromcellstosystems .Wadsworth PublishingCompany,1997. [98]GeraldSimonneau,MichaelA.Gatzoulis,IanAdatia,DavidCelermajer,Chris Denton,ArdeschirGhofrani,MiguelAngelGomezSanchez,R.KrishnaKumar, MichaelLandzberg,RobertoF.Machado,HorstOlschewski,IvanM.Robbins, andRogieroSouza.UpdatedClinicalClassicationofPulmonaryHypertension. JournaloftheAmericanCollegeofCardiology ,62,Supplement:D34{ D41,December2013. [99]GeraldSimonneau,MichaelAGatzoulis,IanAdatia,DavidCelermajer,Chris Denton,ArdeschirGhofrani,MiguelAngelGomezSanchez,RKrishnaKumar, MichaelLandzberg,RobertoFMachado,etal.Updatedclinicalclassication ofpulmonaryhypertension. JournaloftheAmericanCollegeofCardiology 62:D34{D41,2013. [100]GraldSimonneau,IvanM.Robbins,MauriceBeghetti,RichardN.Channick, MarionDelcroix,ChristopherP.Denton,C.GregoryElliott,SeanP.Gaine, MarkT.Gladwin,Zhi-ChengJing,MichaelJ.Krowka,DavidLangleben,NorifumiNakanishi,andRogrioSouza.UpdatedClinicalClassicationofPulmonaryHypertension. JAmCollCardiol ,54s1:S43{S54,June2009. [101]CharlesSingeretal.Ahistoryofbiology. Ahistoryofbiology. ,1950. [102]N.Sommer,A.Dietrich,R.T.Schermuly,H.A.Ghofrani,T.Gudermann, R.Schulz,W.Seeger,F.Grimminger,andN.Weissmann.Regulationofhypoxic pulmonaryvasoconstriction:basicmechanisms. Eur.Respir.J. ,32:1639{ 1651,December2008. [103]KurtR.Stenmark,KarenA.Fagan,andMariaG.Frid.Hypoxia-InducedPulmonaryVascularRemodelingCellularandMolecularMechanisms. Circulation Research ,99:675{691,September2006. [104]KurtRStenmark,BarbaraMeyrick,NazzarenoGalie,WolterJMooi,and IvanFMcMurtry.Animalmodelsofpulmonaryarterialhypertension:the hopeforetiologicaldiscoveryandpharmacologicalcure. AmericanJournalof Physiology-LungCellularandMolecularPhysiology ,297:L1013{L1032,2009. [105]KurtRStenmark,EvaNozik-Grayck,EvgeniaGerasimovskaya,AdilAnwar, MinLi,SuzetteRiddle,andMariaFrid.Theadventitia:essentialroleinpulmonaryvascularremodeling. ComprehensivePhysiology ,2011. 103

PAGE 118

[106]CarolinStrobl,JamesMalley,andGerhardTutz.AnIntroductiontoRecursive Partitioning:Rationale,ApplicationandCharacteristicsofClassicationand RegressionTrees,BaggingandRandomForests. PsycholMethods ,14:323{ 348,December2009. [107]SalvatorePSuteraandRichardSkalak.Thehistoryofpoiseuille'slaw. Annual ReviewofFluidMechanics ,25:1{20,1993. [108]J.T.Sylvester,LarissaA.Shimoda,PhilipI.Aaronson,andJeremyP.T.Ward. HypoxicPulmonaryVasoconstriction. PhysiologicalReviews ,92:367{520, January2012. [109]ThenappanThenappan.ValidationofthePulmonaryHypertensionConnection EquationforSurvivalPredictioninPulmonaryArterialHypertension. CHEST Journal ,141:642,March2012. [110]LianTianandNaomiC.Chesler.Invivoandinvitromeasurementsofpulmonaryarterialstiness:Abriefreview. PulmCirc ,2:505{517,2012. [111]UyenTruong,BrianFonseca,JamieDunning,ShawnaBurgett,CraigLanning, DDunbarIvy,RobinShandas,KendallHunter,andAlexJBarker.Wallshear stressmeasuredbyphasecontrast.2013. [112]WilliamNVenablesandDavidMSmith.therdevelopmentcoreteam. An IntroductiontoR.NotesonR:AProgrammingEnvironmentforDataAnalysis andGraphics ,2005. [113]Jen-Louisvincent,AndrewRhodes,AzrielPerel,GregS.Martin,Giorgio DellaRocca,andmervynsinger.Clinicalreview:Updateonhemodynamic monitorying-aconsensusof16. CriticalCare ,15,2011. [114]AndreasFvonRecum. Handbookofbiomaterialsevaluation:scientic,technicalandclinicaltestingofimplantmaterials .CRCPress,1998. [115]JessicaE.WagenseilandRobertMecham.VascularExtracellularMatrixand ArterialMechanics. PhysiologicalReviews ,89:957{989,2008. [116]J.S.Walker,C.J.Wingard,andR.A.Murphy.Energeticsofcrossbridge phosphorylationandcontractioninvascularsmoothmuscle. Hypertension ,23 Pt2:1106{1112,June1994. [117]E.K.WeirandS.L.Archer.Themechanismofacutehypoxicpulmonary vasoconstriction:thetaleoftwochannels. FASEBJ. ,9:183{189,February 1995. [118]NorbertWeissmann,NataschaSommer,RalphTheoSchermuly,HosseinArdeschirGhofrani,WernerSeeger,andFriedrichGrimminger.Oxygensensorsin hypoxicpulmonaryvasoconstriction. Cardiovasc.Res. ,71:620{629,September2006. 104

PAGE 119

[119]NicoWesterhof,Jan-WillemLankhaar,andBerendE.Westerhof.Thearterial Windkessel. MedBiolEngComput ,47:131{141,June2008. [120]Dan-ChenWu,Hong-DaZhang,andZhi-ChengJing.PediatricPulmonary ArterialHypertension. CurrHypertensRep ,15:606{613,December2013. [121]WillemijnMH.Zijlstra,JohannesM.Douwes,ErikaB.Rosenzweig,SandorSchokker,UshaKrishnan,MarcusTR.Roofthooft,KathleenMiller-Reed, HansL.Hillege,DDunbarIvy,andRolfMF.Berger.Survivaldierencesin pediatricpulmonaryarterialhypertension:cluestoabetterunderstandingof outcomeandoptimaltreatmentstrategies. JAmCollCardiol ,March2014. 105

PAGE 120

APPENDIXA.TopPredictiveModels ModelNamePredictionsAccuracy Pmax+Mpap+MpapAop+Dmax5975.64% Pmax+Mpap+MpapAop+Dmin5975.64% Pmax+PvrSvr+Mpap+Cdyn5773.08% Pmax+PvrSvr+Mpap+Strain5671.79% MpapAop+DmaxBSA5570.51% MpapAop+DmaxBSA+SvPP5570.51% Pmax+Mpap+Cdyn+Pmin5570.51% Pmax+Mpap+DmaxBSA+Pmin5570.51% Pmax+Mpap+MpapAop5570.51% Pmax+Mpap+MpapAop+Pmin5570.51% Pmax+Mpap+MpapAop+Strain5570.51% Pmax+Mpap+Strain+Pmin5570.51% Pmax+MpapAop+Cdyn+DmaxBSA5570.51% Pmax+PvrSvr+Mpap5570.51% Pmax+PvrSvr+Mpap+SvPP5570.51% PvrSvr+Cdyn+SvPP5570.51% MpapAop+DmaxBSA+CO5570.51% PvrSvr+Cdyn+Dmax+CO5570.51% PvrSvr+Cdyn+SvPP+CO5570.51% MpapAop+Cdyn+DmaxBSA+Pmin5469.23% MpapAop+Cdyn+DmaxBSA+Strain5469.23% MpapAop+DmaxBSA+CI5469.23% MpapAop+DmaxBSA+SvPP+CI5469.23% MpapAop+DmaxBSA+SvPP+Pmin5469.23% Pmax+Mpap+Cdyn+SvPP5469.23% 106

PAGE 121

Pmax+Mpap+Dmax5469.23% Pmax+Mpap+MpapAop+Cdyn5469.23% Pmax+Mpap+MpapAop+CI5469.23% Pmax+Mpap+MpapAop+SvPP5469.23% Pmax+Mpap+Strain+SvPP5469.23% Pmax+MpapAop+CI5469.23% Pmax+PVRI+Mpap+MpapAop5469.23% Pmax+PvrSvr+DmaxBSA5469.23% Pmax+PvrSvr+DmaxBSA+CI5469.23% Pmax+PvrSvr+Mpap+CI5469.23% Pmax+PvrSvr+Mpap+DmaxBSA5469.23% Pmax+PvrSvr+Mpap+MpapAop5469.23% Pmax+PvrSvr+MpapAop+DmaxBSA5469.23% Pmax+PvrSvr+PVRI+DmaxBSA5469.23% Pmax+PvrSvr+PVRI+Mpap5469.23% PVRI+MpapAop+DmaxBSA5469.23% PVRI+MpapAop+DmaxBSA+SvPP5469.23% PvrSvr+Cdyn+SvPP+CI5469.23% PvrSvr+MpapAop+Cdyn+DmaxBSA5469.23% PvrSvr+MpapAop+DmaxBSA5469.23% PvrSvr+MpapAop+DmaxBSA+CI5469.23% PvrSvr+PVRI+Cdyn+SvPP5469.23% MpapAop+Dmax+DmaxBSA+Mrap5469.23% MpapAop+DmaxBSA+CI+Mrap5469.23% 107

PAGE 122

APPENDIXB.R-codeStudy1,MultivariatePrediction ##------------------------------------------------------------------##SetupDataFramesfromthescrubbedcatheterizationandechodata ##Thesedatahavealreadybeenprocessedandarereadyforanalysis ##------------------------------------------------------------------allDataW1Ch1.df<-read.csv"C:/cathDB/Analyz/allDataW1Ch1.csv",header=TRUE allDataW2Ch1.df<-read.csv"C:/cathDB/Analyz/allDataW2Ch1.csv",header=TRUE allDataW1Ch3.df<-read.csv"C:/cathDB/Analyz/allDataW1Ch3.csv",header=TRUE allDataW2Ch3.df<-read.csv"C:/cathDB/Analyz/allDataW2Ch3.csv",header=TRUE allDataWdelCh1.df<-read.csv"C:/cathDB/Analyz/allDataWdelCh1.csv",header=TRUE allDataWdelCh3.df<-read.csv"C:/cathDB/Analyz/allDataWdelCh3.csv",header=TRUE ch1.diam.df<-read.csv"C:/cathDB/Analyz/ch1.diam.csv",header=TRUE ch3.diam.df<-read.csv"C:/cathDB/Analyz/ch3.diam.csv",header=TRUE timeToEvent<-read.csv"C:/cathDB/Analyz/adverseEvents_sm.csv",header=TRUE ##Getfunctions:allCombos,makeDF,allPredictPolrintoworkingmemory ##makeDFfunctioninrunEvals.R ##runWho1RoomAirCh1 #DF<-makeDFallDataW1Ch1.df,ch1.diam.df ##runWho2RoomAirCh1 DF<-makeDFallDataW2Ch1.df,ch1.diam.df 108

PAGE 123

##--------------------------------------------------------------------##Setuptheexplanatoryvariables.Processingtakesawhile,so ##trydifferentmixesofthevariables,replacingvarsthatdo ##notcontribute"significantly" ##------------------------------------------------------------------vars<-c"Pmax","PvrSvr","PVRI","Mpap","MpapAop","Cdyn","Dmax","DmaxBSA", "Strain","SvPP","CI","Dmin" ##PPandDmaxout-colinearwithothervars vars<-c"Pmax","PvrSvr","PVRI","Mpap","MpapAop","Cdyn","Dmax","DmaxBSA", "Strain","SvPP","CI","Wedge","Mrap" ##checkPmaxSBPandPminDBPtogether vars<-c"Pmax","PvrSvr","PVRI","Mpap","MpapAop","Cdyn","Dmax","DmaxBSA", "Strain","SvPP","CI","Pmin" ##runallvarswithsignifcantp-vals>0.05 vars<-c"Pmax","PvrSvr","MpapAop","Pmin","Mpap","PVRI","Cdyn","Dmax", "Dmin","Strain" ##allCombosfunctionin ac<-allCombosvars ##----------------------------------------------------------------------##Removeallrowswith5ormorevariablecombinations-thoseequations ##becomeunwieldyarelikelymorespecifictothispopulationrather 109

PAGE 124

##thanbeingagoodgeneralsolution ##---------------------------------------------------------------------ac<-ac[-562:-nrowac,]##cutdownto4variablesorfewer-allcombosofthose ac<-ac[-794:-nrowac,]##cutdownto4varsandfewer... ac<-ac[-1093:-nrowac,] ac<-ac[-638:-nrowac,] ##--------------------------------------------------------------------##Runthepredictions-orderedlogisticregressionpolr ## ##Thenwriteouttheresults ##Toomanyvariablestoconsiderallatonce.Breakthemapart ##indifferentconfigurations,alwayskeepingtopvarsandtrading ##lesspredictivevarsinandout.==>manyresultsfiles. ##Mergealltheresultsfilesbasedonuniquemodelnames. ##---------------------------------------------------------------------all_predictions<-allPredictPolrvars,DF,ac write.csvall_predictions,file="C:/cathDB/Results/.csv",na="" write.csvresults1,file="C:/cathDB/Results/results1.csv",na="" write.csvresults2,file="C:/cathDB/Results/results2.csv",na="" write.csvresults3,file="C:/cathDB/Results/results3.csv",na="" write.csvresults4,file="C:/cathDB/Results/results4.csv",na="" write.csvresults5,file="C:/cathDB/Results/results5.csv",na="" tmp<-mergeresults5,results4,by=c"ModelName","CorrectPreds","MSE", all.x=TRUE,all.y=TRUE res<-tmp 110

PAGE 125

tmp<-mergeres,results3,by=c"ModelName","CorrectPreds","MSE", all.x=TRUE,all.y=TRUE res<-tmp tmp<-mergeres,results2,by=c"ModelName","CorrectPreds","MSE", all.x=TRUE,all.y=TRUE res<-tmp tmp<-mergeres,results1,by=c"ModelName","CorrectPreds","MSE", all.x=TRUE,all.y=TRUE res<-tmp write.csvres,file="C:/cathDB/Results/All_Predictions.csv",na="" write.csvall_predictions,file="C:/cathDB/Results/pvalResults.csv",na="" ##--------------------------------------------------------##Runindividualmodels,lookatfitandsummarydataand ##calculatethep-valsfortheexplanatoryvariables ##Runbyhandorgeneratecombosandstorethoseexplanatory ##modelstringsintheconveniencevariable:modl ##--------------------------------------------------------#fit<-polrfactorWho2~Pmin,data=DF,Hess=TRUE #fit<-polrfactorWho2~sDiam,data=DF,Hess=TRUE #modl<-"Pmin+sDiam" fit<-polrfactorWho2~modl,data=DF,Hess=TRUE fit 111

PAGE 126

summaryfit ctable<-coefsummaryfit p<-pnormabsctable[,"tvalue"],lower.tail=FALSE*2 ctable<-cbindctable,"pvalue"=p ctable ##---------------------------------------------------------------------##PlottheIndividualresults-polr,AIC,RandForestasbarchartsfor ##comparison ##---------------------------------------------------------------------uniDF<-read.csv"C:/cathDB/Results/Univariate_all.csv",header=TRUE ##CorrectPredictions ##AIC ##RandomForest tmpDF<-uniDF tmpDF$clinicalInd<-factoruniDF$clinicalInd,levels=uniDF[order-uniDF$CorrectPredictions,"clinicalInd"] tmpDF<-renametmpDF,c"CorrectPredictions"="thisVar" p1<-clinicIndsBarPlottmpDF,"NumberofCorrectPredictionsnusingOrderedLogisticRegression" p1 tmpDF<-uniDF tmpDF$clinicalInd<-factoruniDF$clinicalInd,levels=uniDF[orderuniDF$AIC,"clinicalInd"] tmpDF<-renametmpDF,c"AIC"="thisVar" p2<-clinicIndsBarPlottmpDF,"RelativeMeasureofGoodness-of-FitnusingAICWeight" p2 112

PAGE 127

tmpDF<-uniDF tmpDF$clinicalInd<-factoruniDF$clinicalInd,levels=uniDF[order-uniDF$RandomForest,"clinicalInd"] tmpDF<-renametmpDF,c"RandomForest"="thisVar" p3<-clinicIndsBarPlottmpDF,"RelativeMeasureofVariableImportancenusingRandomForest" p3 png"C:/cathDB/Figures/singleVarResults.png",width=3000,height=2400,res=300 multiplotp1,p2,p3,cols=2 dev.off ggsave"C:/cathDB/Figures/singleVarResults.png",width=11,height=6,dpi=300 ##---------------------------------------------------------------------## ##---------------------------------------------------------------------clinicIndsBarPlot<-functionthisDF,thisYlab{ ifmissingthisYlab thisYlab="" p<-ggplotthisDF,aesx=clinicalInd,y=thisVar,width=.6+ geom_barposition=position_dodge.9,stat="identity",fill="#999999"+ xlab"nClinicalIndicators"+ ylabthisYlab p<-p+theme_bw p<-p+themeaxis.text.x=element_textsize=10,angle=45,hjust=1, 113

PAGE 128

axis.text.y=element_textsize=12, axis.title.x=element_textsize=12, axis.title.y=element_textsize=12, strip.text.x=element_textsize=12 #p<-p+themeaxis.text.x=element_textangle=45,hjust=1 p } ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ## ##RunthePolar,VGLMorGLMcallsoneverycombinationofvariables ##givenspecifiedWHOfunctionalclassificationsandVentillationChallenges ## ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##-----------------------##SetEnv_Cath-setuptheevironment ##-----------------------#rmlist=ls#cleanslate setwd"C:/Users/Owner/Desktop/cathDB" sink"C:/Users/Owner/Desktop/cathDB/console.output.txt",split=T libraryaod#AICfunctions libraryAICcmodavg#AICmodelaveraging libraryreshape#rename 114

PAGE 129

libraryVGAM allDataW1Ch1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW1Ch1.csv",header=TRUE allDataW2Ch1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW2Ch1.csv",header=TRUE allDataW1Ch3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW1Ch3.csv",header=TRUE allDataW2Ch3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW2Ch3.csv",header=TRUE allDataWdelCh1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataWdelCh1.csv",header=TRUE allDataWdelCh3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataWdelCh3.csv",header=TRUE ch1.diam.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/ch1.diam.csv",header=TRUE ch3.diam.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/ch3.diam.csv",header=TRUE #timeToEvent<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/adverseEvents_sm.csv",header=TRUE timeToEvent<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/AdverseEventsData_sm2.csv",header=TRUE numOfEvents<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/AdverseEvents_visits_sm.csv",header=TRUE ##GoldStandard GSvars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","CI" ##NewGuys PPvars<-c"delD","Cdyn","Strain","PP","Pmax","Dmax" ##mix #vars<-c"Dmax","Strain","PVRI","Wedge","Pmax","PvrSvr","PP","CI" 115

PAGE 130

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Functions ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##SetuptheDataFrameusingtheKinematicdataand ##thedeomographic,... ##~~~~~~~~~~~~~~~~~~~~~~~~~~~ makeDF<-functionkinematic.df,diameter.df{ df<-mergekinematic.df,diameter.df,by="Patient_num" df<-na.omitdf PvrSvr<-df$PVRI/df$SVRI PP<-df$Pmax-df$Pmin delD<-df$Dmax-df$Dmin Strain<-delD/df$Dmin Cdyn<-Strain/PP S1<-Strain/df$HR SvPP<-df$SV/PP MpapAop<-df$Mpap/df$Aop DmaxBSA<-df$Dmax/df$BSA df<-cbinddf,PvrSvr,PP,delD,Strain,Cdyn,S1,SvPP,MpapAop,DmaxBSA returndf }#endFunctionmakeDF 116

PAGE 131

##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##ThisistheactualVGLMorGLMcall.Itiswrappedinafunction ##sothatitcanberunwithinaTRYincasethereisanerror. ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ doVglm<-functionrespVar,anyModl,dataSet{ #returnevalparsetext=paste"vglmrespVar~",anyModl,",family=multinomial,data=",dataSet,"" #returnevalparsetext=paste"vglmrespVar~",anyModl,",family=cumulative,data=",dataSet,"" #returnevalparsetext=paste"vglmrespVar~",anyModl,",family=ordpoisson,data=",dataSet,"" returnevalparsetext=paste"vglmWho2~",anyModl,",family=cumulative,data=dataSet" #returnevalparsetext=paste"glmrespVar~",anyModl,",family=gaussian,data=",dataSet,"" }#endFunctiondoVglm ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Createalistofallpossiblecombinationsofthevariablesand ##putthemintheformatthattheVGLMorGLMcallcanuse,ie. ##variable+variable+variable+... ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ allCombos<-functionvars{ libraryreshape#forrename ##---Thelistisalittleunwieldy,butcombnisveryconvenient ##---ttransformsthematrix combos<-list foriin1:lengthvars##allcombinationsofallvariables #foriin1:5##5variablesorless 117

PAGE 132

{ combos[[i]]<-tcombnvars,i,simplify=TRUE } ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Usingthelistofallpossiblecombosofvariablescreateavector ##withallpossiblemodelnames:var+var+... ##Thevectoriseasiertousetoevaluatethemodels. ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ indx<-1 allModels<-NULL kvals<-NULL modelStr<-NULL foriin1:lengthvars{##Thecombinationlevels numrows<-lengthcombos[[i]][,1] forjin1:numrows{##Withinacomblevel,the modelStr<-pastecombos[[i]][j,1],"",sep=""##Putinfirstvariable k<-2##startcountingthesecondvariableforloop whilek<=i{##Foreachrow,pastetogetherallremainingvariablesfortotalofi #putstringtogether modelStr<-pastemodelStr,"+",sep="" modelStr<-pastemodelStr,combos[[i]][j,k],sep="" k<-k+1 } 118

PAGE 133

##Addthemodeltothearray allModels[indx]<-modelStr kvals[indx]<-k indx<-indx+1 } } allModComb<-data.frameallModels,kvals allModComb<-renameallModComb,callModels="Model",kvals="Kval" returnallModComb }#endFunctionallCombs ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Evaluatethemodels:createthelistofthemodels,runthe ##VGLMorGLMcommandoneachmodel,calculateAIC,AICc,getthe ##logLikelihood,putalltheseinanewlistandorderthembyAICc ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ evalVGLM<-functionexplanVars,respVar,DF{ modlNames<-allCombosexplanVars model.list<-list foriin1:lengthmodlNames$Model{ model.list[[i]]<-trydoVglmrespVar,modlNames$Model[i],"DF",TRUE } 119

PAGE 134

errlist<-sapplymodel.list,functionanyModl!inheritsanyModl,"try-error" new_model.list<-unlistmodel.list[sapplymodel.list,functionanyModl!inheritsanyModl,"try-error"] #printerrlist ##UsethisforVGLM kvals<-NULL aicVals<-NULL llVals<-NULL aiccVals<-NULL modelResults<-NULL j<-1 newNames<-NULL ##Ifthevglmcommandcamebackwithanerror,skiptheerrorsand ##usetherestofthelist foriin1:lengtherrlist{ iferrlist[i]=="TRUE"{ #kvals[j]<-modlNames$Kval[i] newNames[j]<-pastemodlNames$Model[i] j<-j+1 } } ##Calculationsforeachmodel foriin1:lengthnew_model.list{ llVals[i]<-logLiknew_model.list[[i]] #aicVals[i]<-2*kvals[i]-2*llVals[i] aicVals[i]<-AICnew_model.list[[i]] kval<-nprednew_model.list[[i]] numobs<-nobsnew_model.list[[i]] 120

PAGE 135

#aiccVals[i]<-aicVals[i]+*kvals[i]*kvals[i]+1/nrowDF-kvals[i]-1 aiccVals[i]<-aicVals[i]+*kval*kval+1/numobs-kval-1 } tmp<-data.framenewNames,aicVals,aiccVals,llVals tmp<-renametmp, cnewNames="ModelName",aicVals="AIC",aiccVals="AICc",llVals="LogLikelihood" modelResults<-tmp[ordertmp$AICc,] returnmodelResults }#endFunctionevalModels #evalGLM<-functionexplanVars,respVar,DF{ evalGLM<-functionexplanVars,DF{ dataSet<-"DF" modlNames<-allCombosexplanVars model.list<-list foriin1:lengthmodlNames$Model{ model.list[[i]]<-evalparsetext=paste"glmWho2~",modlNames$Model[i],",family=gaussian,data=",dataSet,"" } aicResults<-aictabcand.set=model.list,modnames=modlNames,sort=TRUE returnaicResults 121

PAGE 136

#printaictabcand.set=model.list,modnames=modlNames,sort=TRUE, #digits=2,LL=TRUE }#endFunctionevalGLM ##---------------------------------------##aicPolr-AICforeachmodel ##----------------------------------------aicPolr<-functionexplanVars,localDF{ libraryAICcmodavg modlNames<-allCombosexplanVars fit.list<-list foriin1:lengthmodlNames$Model{ #model.list[[i]]<-evalparsetext=paste"polrWho2~",modlNames$Model[i],",data=",dataSet,"" fit.list[[i]]<-evalparsetext=paste"polrfactorWho2~",modlNames$Model[i],",data=localDF,Hess=TRUE" } aicResults<-aictabcand.set=fit.list,modnames=modlNames,sort=TRUE returnaicResults }#endFunctionaicPolr 122

PAGE 137

##-----------------------------------------## ##-----------------------------------------allPredictMultinom<-functionexplanVars,localDF{ libraryreshape#forrename librarynnet#formultinom dataSet<-"localDF" modlNames<-allCombosexplanVars fitglm<-list fit<-NULL pred<-NULL tmp<-NULL mse1<-NULL mse2<-NULL correctPreds<-NULL foriin1:lengthmodlNames$Model{##Predictwitheachmodel ##debugtryfirst20 ##foriin1:20{ diffs<-NULL diffs2<-NULL gotit<-0 forjin1:nrowlocalDF{##ForeachmodeldoaLeave-One-OutPrediction trainset<-localDF[-j,] testset<-localDF[j,] #fitglm[[j]]<-evalparsetext=paste"glmrespVar~",modlNames$Model[i],",family=gaussian,data=",trainset,"" 123

PAGE 138

#fitglm[[j]]<-evalparsetext=paste"glmWho2~",modlNames$Model[i],",family=poisson,data=trainset" #fitglm[[j]]<-evalparsetext=paste"glmlWho2~",modlNames$Model[i],",family=poisson,data=trainset" #fitglm[[j]]<-evalparsetext=paste"glmWho2~",modlNames$Model[i],",family=gaussian,data=trainset" fit<-evalparsetext=paste"multinomfactorWho2~",modlNames$Model[i],",data=trainset,trace=FALSE" #fit<-fitmnom[[j]] pred[j]<-evalparsetext=paste"predictfit,newdata=testset" iftestset$Who2-pred[j]==0gotit<-gotit+1 diffs[j]<-testset$Who2-pred[j] } mse1[i]<-sumdiffs^2/nrowlocalDF correctPreds[i]<-gotit } tmp<-data.framemodlNames$Model,correctPreds,mse1 ##Debug-checkfirst20 #tmp<-data.framemodlNames$Model[1:20],correctPreds,mse1,mse2 tmp<-renametmp,cmodlNames.Model="ModelName",correctPreds="CorrectPreds",mse1="MSE" predResults<-tmp[order-tmp$CorrectPreds,tmp$MSE,] returnpredResults } 124

PAGE 139

##-----------------------------------------## ##-----------------------------------------allPredictPolr<-functionexplanVars,localDF,ac{ libraryreshape#forrename libraryMASS#forpolr #modlNames<-allCombosexplanVars modlNames<-ac fit<-NULL pred<-NULL tmp<-NULL mse1<-NULL correctPreds<-NULL foriin1:lengthmodlNames$Model{##Predictwitheachmodel diffs<-NULL diffs2<-NULL gotit<-0 forjin1:nrowlocalDF{##ForeachmodeldoaLeave-One-OutPrediction trainset<-localDF[-j,] testset<-localDF[j,] fit<-evalparsetext=paste"polrfactorWho2~",modlNames$Model[i],",data=trainset,Hess=TRUE" pred[j]<-evalparsetext=paste"predictfit,newdata=testset" 125

PAGE 140

iftestset$Who2-pred[j]==0gotit<-gotit+1 diffs[j]<-testset$Who2-pred[j] } mse1[i]<-sumdiffs^2/nrowlocalDF correctPreds[i]<-gotit } tmp<-data.framemodlNames$Model,correctPreds,mse1 tmp<-renametmp,cmodlNames.Model="ModelName",correctPreds="CorrectPreds",mse1="MSE" predResults<-tmp[order-tmp$CorrectPreds,tmp$MSE,] returnpredResults } ##-----------------------------------------## ##-----------------------------------------predictPolr<-functionmodl,localDF{ libraryreshape#forrename libraryMASS#forpolr fit<-NULL pred<-NULL tmp<-NULL 126

PAGE 141

mse1<-NULL correctPreds<-NULL diffs<-NULL gotit<-0 forjin1:nrowlocalDF{##ForeachmodeldoaLeave-One-OutPrediction trainset<-localDF[-j,] testset<-localDF[j,] fit<-evalparsetext=paste"polrfactorWho2~",modl,",data=trainset,Hess=TRUE" pred[j]<-evalparsetext=paste"predictfit,newdata=testset" iftestset$Who2-pred[j]==0gotit<-gotit+1 diffs[j]<-testset$Who2-pred[j] } mse1<-sumdiffs^2/nrowlocalDF correctPreds<-gotit tmp<-data.framemodl,correctPreds,mse1 tmp<-renametmp,cmodl="ModelName",correctPreds="CorrectPreds",mse1="MSE" predsTbl<-tablelocalDF$Who2,pred predsTbl mosaicplotpredsTbl,color=FALSE,shade=FALSE,legend=FALSE #mosaicplotpredsTbl,gp=shading_max,split_vertical=TRUE, #main="PredictionResults" 127

PAGE 142

returntmp } ##-----------------------------------------## ##-----------------------------------------allPredictGLM<-functionexplanVars,localDF{ libraryreshape#forrename dataSet<-"localDF" modlNames<-allCombosexplanVars fitglm<-list predglm<-list tmp<-NULL mse1<-NULL mse2<-NULL correctPreds<-NULL foriin1:lengthmodlNames$Model{##Predictwitheachmodel ##debugtryfirst20 ##foriin1:20{ diffs<-NULL diffs2<-NULL 128

PAGE 143

gotit<-0 forjin1:nrowlocalDF{##ForeachmodeldoaLeave-One-OutPrediction trainset<-localDF[-j,] testset<-localDF[j,] #fitglm[[j]]<-evalparsetext=paste"glmrespVar~",modlNames$Model[i],",family=gaussian,data=",trainset,"" #fitglm[[j]]<-evalparsetext=paste"glmWho2~",modlNames$Model[i],",family=poisson,data=trainset" #fitglm[[j]]<-evalparsetext=paste"glmlWho2~",modlNames$Model[i],",family=poisson,data=trainset" fitglm[[j]]<-evalparsetext=paste"glmWho2~",modlNames$Model[i],",family=gaussian,data=trainset" fit<-fitglm[[j]] predglm[[j]]<-evalparsetext=paste"predictfit,newdata=data.frametestset,type="response",se.fit=T" iftestset$Who2-roundpredglm[[j]]$fit==0gotit<-gotit+1 diffs[j]<-testset$Who2-roundpredglm[[j]]$fit diffs2[j]<-testset$Who2-predglm[[j]]$fit #predictions[i]<-roundpredglm$fit #who2[i]<-testset$Who2 } mse1[i]<-sumdiffs^2/nrowlocalDF mse2[i]<-sumdiffs2^2/nrowlocalDF correctPreds[i]<-gotit } tmp<-data.framemodlNames$Model,correctPreds,mse1,mse2 ##Debug-checkfirst20 129

PAGE 144

#tmp<-data.framemodlNames$Model[1:20],correctPreds,mse1,mse2 tmp<-renametmp,cmodlNames.Model="ModelName",correctPreds="CorrectPreds",mse1="MSE_rounded",mse2="MSE" predResults<-tmp[order-tmp$CorrectPreds,tmp$MSE,] returnpredResults } ##-----------------------------------------## ##-----------------------------------------allPredictVGLM<-functionexplanVars,localDF{ libraryVGAM libraryreshape#forrename dataSet<-"localDF" modlNames<-allCombosexplanVars fitglm<-list fitvglm<-list predglm<-list predvglm<-list tmp<-NULL mse1<-NULL mse2<-NULL correctPreds<-NULL 130

PAGE 145

newModelNames<-NULL idxI<-1 foriin1:lengthmodlNames$Model{##Predictwitheachmodel ##debugtryfirst20 ##foriin1:20{ diffs<-NULL diffs2<-NULL gotit<-0 idxJ<-1 forjin1:nrowlocalDF{##ForeachmodeldoaLeave-One-OutPrediction trainset<-localDF[-j,] testset<-localDF[j,] #fitglm[[j]]<-evalparsetext=paste"glmrespVar~",modlNames$Model[i],",family=gaussian,data=",trainset,"" #fitvglm[[j]]<-evalparsetext=paste"vglmWho2~",modlNames$Model[i],",family=multinomial,data=trainset" #fitvglm[[j]]<-evalparsetext=paste"vglmWho2~",modlNames$Model[i],",family=cumulative,data=trainset" #fitvglm[[j]]<-evalparsetext=paste"vglmlWho2~",modlNames$Model[i],",family=ordpoisson,data=trainset" fitvglm[[j]]<-trydoVglmtrainset$Who2,modlNames$Model[i],trainset,TRUE errlist<-sapplyfitvglm,functionanyModl!inheritsanyModl,"try-error" new_fit.list<-unlistfitvglm[sapplyfitvglm,functionanyModl!inheritsanyModl,"try-error"] ##Ifthevglmcommandcamebackwithanerror,skiptheerrorsand ##usetherestofthelist iferrlist[j]=="TRUE"{ fit<-fitvglm[[j]] predvglm[[idxJ]]<-evalparsetext=paste"predictfit,newdata=data.frametestset,type="res"" 131

PAGE 146

iftestset$Who2-which.maxpredvglm[[idxJ]]==0gotit<-gotit+1 diffs[idxJ]<-testset$Who2-which.maxpredvglm[[idxJ]] idxJ<-idxJ+1 } } iferrlist[1]=="TRUE"{ mse1[idxI]<-sumdiffs^2/nrowlocalDF correctPreds[idxI]<-gotit newModelNames[idxI]<-pastemodlNames$Model[i] idxI<-idxI+1 } printi } tmp<-data.framenewModelNames,correctPreds,mse1 ##Debug-checkfirst20 #tmp<-data.framemodlNames$Model[1:20],correctPreds,mse1 tmp<-renametmp,cnewModelNames="ModelName",correctPreds="CorrectPreds",mse1="MSE" predResults<-tmp[order-tmp$CorrectPreds,tmp$MSE,] returnpredResults } 132

PAGE 147

evalCoxph<-functionexplanVars,DF{ #dataSet<-"DF" modlNames<-allCombosexplanVars coxfit.list<-list aicVals<-NULL respVar<-SurvtteDF$Time,tteDF$Event logtest<-NULL ltr<-NULL foriin1:lengthmodlNames$Model{ coxfit.list[[i]]<-evalparsetext=paste"coxphrespVar~",modlNames$Model[i],",tteDF" } foriin1:lengthcoxfit.list{ thisFit<-coxfit.list[[i]] aicVals[i]<-extractAICcoxfit.list[[i]][2] #logtest[i]<--2*thisFit$loglik[1]-thisFit$loglik[2] #degFree<-nrowthisFit$var #ltr[i]<--pchisqlogtest,degFree } tmp<-data.framemodlNames$Model,aicVals tmp<-renametmp, cnewNames="ModelName",aicVals="AIC" #tmp<-data.framemodlNames$Model,aicVals,ltr #tmp<-renametmp, #cnewNames="ModelName",aicVals="AIC",ltr="LikelyTestRatioP" 133

PAGE 148

modelResults<-tmp[ordertmp$AIC,] returnmodelResults }#endFunctionevalCoxPH ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Summarystatsforthetopmodelsforbothconventionalandproximalsmodels ##runaovforthevariablesignificance ##runlmforthemodelR2 ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ sumStats<-functionmodelList,df,resp_var,VGLM{ topModel<-modelList[1,] dataset<-"df" #evalparsetext=paste"glmresp_var~",topModel,",family=gaussian,data=",dataset,"" modl<-topModel$ModelName ifVGLM==TRUE modl<-topModel$ModelName ifVGLM==FALSE##thenit'sGLM modl<-as.charactertopModel$Modnames.Model aovfit<-evalparsetext=paste"aovresp_var~",modl,",data=",dataset,"" sprintf"Anovaforvariablesignificance" printaovfitSum<-summaryaovfit 134

PAGE 149

lmfit<-evalparsetext=paste"lmresp_var~",modl,",data=",dataset,"" sprintf"LinearModelFitforR2" printlmfitSum<-summarylmfit #ifvglm==TRUE{ #null_dev<-evalparsetext=paste"vglmresp_var~",1,",family=multinomial,data=",dataset,"" #res_dev<-evalparsetext=paste"vglmresp_var~",topModel,",family=multinomial,data=",dataset,"" #printpseudo_R2<-1-devianceres_dev/deviancenull_dev #} #ifvglm!=TRUE{ #devs<-evalparsetext=paste"glmresp_var~",topModel,",family=gaussian,data=",dataset,"" #} }#endFunctionsumStats libraryrandomSurvivalForest datatteDF,package="randomSurvivalForest" cath.out<-rsfSurvTime,Event~.,data=tteDFsm printcath.out plotcath.out varSelcath.out #datapbc,package="randomSurvivalForest" datatteDFtiny,package="randomSurvivalForest" rsf.f<-as.formulaSurvTime,Event~.,tteDFtiny pbc3.out<-rsfrsf.f,tteDFtiny,nsplit=10,mtry=2 135

PAGE 150

B<-100 cox.err<-repNA,B #pbc.data<-pbc[applyis.natteDFtiny,1,sum==0,]##removeNA cat"Out-of-bagCoxAnalysis...","n" forbin1:B{ cat"Coxbootstrap:",b,"n" bag.sample<-sample:nrowtteDFtiny,nrowtteDFtiny,replace=TRUE oob.sample<-setdiff:nrowtteDFtiny,bag.sample train<-tteDFtiny[bag.sample,] test<-tteDFtiny[oob.sample,] cox.out<-tryCatch{coxphrsf.f,train},error=functionex{NULL} ifis.listcox.out{ cox.predict<-predictcox.out,test cox.err[b]<-rcorr.censcox.predict, Survpbc.data$days[oob.sample], pbc.data$status[oob.sample][1] } } cat"Errorrates:","n" cat"RandomSurvivalForests:",pbc3.out$err.rate[pbc3.out$ntree],"n" cat"CoxRegression:",meancox.err,na.rm=TRUE,"n" ##polr-OrderedLogisticRegression libraryMASS#forpolr fit<-polrfactorWho2~Mpap+PvrSvr+Pmax+Cdyn,data=DF,Hess=TRUE fit<-polrfactorWho2~PVRI+Wedge+Mpap+Mrap+PvrSvr+CI+SvPP+MpapAop,data=DF,Hess=TRUE 136

PAGE 151

fit<-polrfactorWho2~delD+Cdyn+Strain+PP+Pmax+DmaxBSA+Dmax,data=DF,Hess=TRUE fit<-polrfactorWho2~PVRI+Mpap+PvrSvr+CO+SvPP+MpapAop+Pmax+Dmax+Strain+PP+delD+DmaxBSA+Wedge+Cdyn,data=DF,Hess=TRUE Dmax+Mpap+Mpap/Aop+Pmax ##--------------------##Prediction,P-valsandAICforindividualmodels ##pasteintomodlandtheexplanatoryvars ##------------------modl<-"Pmax+Mpap+MpapAop+Dmax" predictPolrmodl,DF fit<-polrfactorWho2~Pmax+Mpap+MpapAop+Dmax,data=DF,Hess=TRUE summaryfit ##P-vals ctable<-coefsummaryfit p<-pnormabsctable[,"tvalue"],lower.tail=FALSE*2 ctable<-cbindctable,`pvalue`=p tablewho2s,preds #tabletestset$Who2,which.maxpredglm PredictionResults<-tablewho2s,preds #mosaicplotPredictionResults,shade=TRUE,legend=TRUE mosaicplotPredictionResults,gp=shading_max,split_vertical=TRUE, 137

PAGE 152

main="Mpap+PmaxPredictionResults" #cdplotas.factorWho2~Pmax,data=DF #withDF,rugjitterMpap,col="white",quiet=TRUE spineplotas.factorWho2~Pmax,data=DF,breaks=6 ##ConfidenceIntervals ci<-confintfit confint.defaultfit expcoeffit allVars<-c"PVRI","Mpap","PvrSvr","CO","SvPP","MpapAop","Pmax","Dmax", "Strain","delD","DmaxBSA","PP","Wedge","Cdyn","Mrap","CI" #allModls<-PVRI+Mpap+PvrSvr+SvPP+MpapAop+Pmax+Dmax+Strain+delD+DmaxBSA+PP+Wedge+Cdyn+Mrap+CI+CO ##Proximal prox<-c"delD","Cdyn","Strain","PP","Pmax","DmaxBSA","Dmax" modlNames<-modlNames ##Conventional conv<-c"PVRI","Mpap","PvrSvr","CO","SvPP","MpapAop","Wedge","Mrap" 138

PAGE 153

#DmaxBSA<-tteDF$Dmax/tteDF$BSA #1940rowsfor4vars modlNames<-modlNames[1:1940,] librarysurvival DFcox<-mergeallDataW1Ch1.df,ch1.diam.df,by="Patient_num" #timeToEvent<-read.csv"C:/cathDB/Analyz/AdverseEventsData_sm2.csv",header=TRUE timeToEvent<-read.csv"C:/cathDB/timeToEvent2.csv",header=TRUE ##makeDFmergesthetwodf'sandcbindssomecalculatedvariablesie.strain eventDF<-makeDFDF,timeToEvent #example:coxphSurvtime,status~x+stratasex,test1 coxfit<-coxphSurvTime,Event~PVRI,eventDF coxfit<-coxphSurvTime,Event~Dmax+Strain,eventDF coxfit<-coxphSurvTime,Event~Pmax,eventDF coxfit<-coxphSurvTime,Event~PP,eventDF ##tteDF-combinationofDFandtime-to-event tteDF<-read.csv"C:/cathDB/tteDF.csv",header=TRUE evalCoxphallVars,tteDF ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ## 139

PAGE 154

##Setupnewdatabasesfromtherawdata ##Readintherawdata,scrubthevariables,repackagetheminto ##groupsbyWHOandVentChallenge,savethosedataframesas ##*.csvfilessothisonlyneedstoberunwhentheRedHatdatabase ##isupdated. ## ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ #rmlist=ls#cleanslate libraryaod#AICfunctions libraryAICcmodavg#AICmodelaveraging libraryreshape#rename libraryVGAM ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Readintherawdata ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##C:/Users/Owner/Desktop/cathDB/Analyz2/ #demograph<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_Demographics_May13.csv",header=TRUE #ch1<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/",header=TRUE #ch3<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_Challenge3Sh_2013-05-21_1418.csv",header=TRUE #bloodwork<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_BloodSh_2013-05-21_1418.csv",header=TRUE #demograph<-read.csv"C:/Users/Owner/Downloads/Report_DemographicsSh_2013-06-01_0856.csv",header=TRUE #ch1<-read.csv"C:/Users/Owner/Downloads/Report_Challenge1Sh_2013-05-30_0922.csv",header=TRUE 140

PAGE 155

#ch3<-read.csv"C:/Users/Owner/Downloads/Report_Challenge3Sh_2013-05-30_0922.csv",header=TRUE #bloodwork<-read.csv"C:/Users/Owner/Downloads/Report_BloodSh_2013-05-30_0922.csv",header=TRUE demograph<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_DemographicsSh_2013-09-23_1612.csv",header=TRUE ch1<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_Challenge1Sh_2013-09-23_1610.csv",header=TRUE ch3<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_Challenge3Sh_2013-09-23_1611.csv",header=TRUE bloodwork<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz2/Report_BloodSh_2013-09-23_1612.csv",header=TRUE #timeToEvent<-read.csv"C:/Users/Owner/Desktop/cathDB/eventTime.csv",header=TRUE timeToEvent<-read.csv"C:/Users/Owner/Desktop/cathDB/AdverseEvents_sm.csv",header=TRUE #attachdemograph attachbloodwork ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Scrubthedata-renamingthevariables ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ demograph<-renamedemograph, cMRN..mrn.="MRN",Name..name.="Name",Patient....patient.="Patient_num", Birth.Date..birth_date.="Birth_date",Age.at.Cath..months...age_at_cath.="Age", Sex..sex.="Sex",Date.of.Death..death.="Death_date",Height..cm...height.="Height", Weight..kg...weight.="Weight",BSA..m.2...bsa.="BSA", Initial.WHO..int_who.="Who1",Initial.Cath.Date..initial_cath_date.="Cath_date1", Follow.Up.WHO..fu_who.="Who2",Follow.Up.Cath.Date..fu_cath.="Cath_date2", 141

PAGE 156

Outcome.Score..outcome_score.="Outcome",Ethnicity..ethnicity.="Ethnicity", Other.Ethnicity..other_ethnicity.="Ethnicity_other",Race..race.="Race", Other.Race..other_race.="Race_other",Daignosis.and.Procedures..d_p.="Diagnosis_proceedures", Code....code.="Diag" demograph$X<-NULL bloodwork<-renamebloodwork, cMRN..mrn.="MRN",Name..name.="Name",Patient....patient.="Patient_num", Name..name.="Name",Blood.Date..blood_date.="Blood_date",BNP..bnp.="BNP", BNP.List..bnp_list.="BNP_list",Nt.Pro.BNP..nt_pro_bnp.="NtProBnp", Nt.Pro.BNP.List..nt_pro_bnp_list.="NtProBnp_list",Na..na.="Sodium", K..k.="K",Cl..cl.="Cl",CO2..co2.="CO2",Glucose..glucose.="Glucose", BUN..bun.="BUN",Creatine..creatine.="Creatine", X6.Minute.Walk..minute_walk.="6Min_walk", Cycle.Exercise.Test..cycle_exercise_test.="Cycle_test" bloodwork$X<-NULL ch1<-renamech1, cMRN..mrn.="MRN",Name..name.="Name",Patient....patient.="Patient_num", Challenge..1..challenge_1.="Challenge1", MPA.Pressure..1..mmHg..max..mpa_max.="MPA_max",MPA.min..1..mpa_min.="MPA_min", MPA.Avg...1..mpa_avg.="MPA_avg",MPA.PP..1..mpa_pp.="MPA_pp", RPA.Pressure..1..mmHg..max..rpa_max.="RPA_max", RPA.min..1..rpa_min.="RPA_min",RPA.Avg...1...rpa_avg.="RPA_avg", RPA.PP..1..rpa_pp.="RPA_pp", LPA.Pressure..1..mmHg..max..lpa_max.="LPA_max", 142

PAGE 157

LPA.min..1..lpa_min.="LPA_min",LPA.Avg...lpa_avg.="LPA_avg", LPA.PP..1..lpa_pp.="LPA_pp",Qp..L.min...1..qp.="Qp", Qp.index..1..qp_index.="QpI",Qs..L.min...1..qs.="Qs", Qs.index..1..qs_index.="QsI",Qp.Qs..1..qp_qs.="QpDivQs", Rp.index..1..PVRI...rp.="PVRI",Rs.index..1..SVRI...rs.="SVRI", Rp.units..1..PVR...rp_units.="PVR",Rs.units..1..SVR...rs_units.="SVR", HR..bpm...1..heart_rate.="HR",CO.L.min..1..cardiac_output.="CO", CI.Cardiac.Index..ci_cardiac_index.="CI", PVS..mmHg.mL...1..pvs.="PVS", PVS..Qp..mmHg.mL...1..pvs_qp.="PVSDivQp", Pulse.Pressure..1..pulse_pressure.="PP", Stroke.Volume..Qp..1..sv_qp.="SV_Qp", Stroke.Volume..CO..1..sv_co.="SV_CO", RAP..rap.="RAP",AoP..aop.="AoP",PCWP..pcwp.="PCWP" ch1$X<-NULL ch3<-renamech3, cMRN..mrn.="MRN",Name..name.="Name",Patient....patient.="Patient_num", Challenge..3..challenge_3.="Challenge3", MPA.Pressure..3..mmHg..max..mpa_pressure_3_mmhg_max.="MPA_max",MPA.min..3..mpa_min_3.="MPA_min", MPA.Avg...3..mpa_avg_3.="MPA_avg",MPA.PP..3..mpa_pp_3.="MPA_pp", RPA.Pressure..3..mmHg..max..rpa_pressure_3_mmhg_max.="RPA_max", RPA.min..3..rpa_min_3.="RPA_min",RPA.Avg...3...rpa_avg_3.="RPA_avg", RPA.PP..3..rpa_pp_3.="RPA_pp", LPA.Pressure..3..mmHg..max..lpa_pressure_3_mmhg_max.="LPA_max", LPA.min..3..lpa_min_3.="LPA_min",LPA.Avg...3..lpa_avg_3.="LPA_avg", LPA.PP..3..lpa_pp_3.="LPA_pp",Qp..L.min...3..qp_l_min_3.="Qp", 143

PAGE 158

Qp.index..3..qp_index_3.="QpI",Qs..L.min...3..qs_l_min_3.="Qs", Qs.index..3..qs_index_3.="QsI",Qp.Qs..3..qp_qs_3.="QpDivQs", Rp.index..3..PVRI...rp_index_3.="PVRI",Rs.index..3..SVRI...rs_index_3.="SVRI", Rp.units..3..PVR...rp_units_3.="PVR",Rs.units..3..SVR...rs_units_3.="SVR", HR..bpm...3..hr_bpm_3.="HR",CO.L.min..3..co_l_min_3.="CO", CI.Cardiac.Index..3..ci_cardiac_index_3.="CI", PVS..mmHg.mL...3..pvs_mmhg_ml_3.="PVS", PVS.Qp..mmHg.mL...3..pvs_qp_mmhg_ml_3.="PVSDivQp", Pulse.Pressure..3..pulse_pressure_3.="PP", Stroke.Volume.Qp..3..stroke_volume_qp_3.="SV_Qp", Stoke.Volume.CO..3..stoke_volume_co_3.="SV_CO", RAP..3..rap_3.="RAP",AoP..3..aop_3.="AoP",PCWP..3..pcwp_3.="PCWP" ch3$X<-NULL #recodex,"c,2='A';else='B'" #recodetest.words,'"cat"="kissa";"dog"="koira";"coffee"="kahvi";"tea"="tee";"tee"="tii"' ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Scrubthedata-codeallwhoscoresas1,2,3,4 ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ librarycar who1<-demograph$Who1 who1<-recodewho1,'c"I-II","1-2","I-II","1,2","2,1"=2; c"II-III","2-3","II-III","2,3"=3; c"III-IV","III-IV","3-4","3,4"=4;"IV"=4; c"none","NONE"="";"?"="";"nonotesyet"="" 144

PAGE 159

' who2<-demograph$Who2 who2<-recodewho2,'c"I-II","1-2","I-II","1,2","2,1"=2; c"II-III","2-3","II-III","2,3"=3; c"III-IV","III-IV","3-4","3,4"=4;"IV"=4; c"none","NONE"="";"?"="";"nonotesyet"="" #Leftoversbyhand i<-grep"I-II",demograph$Who1 who1[i]<-2 i<-grep"II-III",demograph$Who2 who2[i]<-3 who1[who1==""]<-NA who2[who2==""]<-NA ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Scrubthedata-makepressuresallnumericvalsues ##andcreateapress_*setwiththemostcomplete ##setofpressuredata ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##SetupChallenge1dataframe: mpa_min<-recodech1$MPA_min,'"nodata"=""' lpa_max<-recodech1$LPA_max,'"nodata"=""' lpa_min<-recodech1$LPA_min,'"nodata"=""' lpa_avg<-recodech1$LPA_avg,'"nodata"=""' 145

PAGE 160

hr<-recodech1$HR,'"?"=""' co<-recodech1$CO,'c"na",""=""' bnp<-recodebloodwork$BNP,'"<15"="15"' ##OrsetupforChallenge3dataframe: mpa_min<-recodech3$MPA_min,'"nodata"=""' lpa_max<-recodech3$LPA_max,'"nodata"=""' lpa_min<-recodech3$LPA_min,'"nodata"=""' lpa_avg<-recodech3$LPA_avg,'"nodata"=""' hr<-recodech3$HR,'"?"=""' co<-recodech3$CO,'c"na",""=""' bnp<-recodebloodwork$BNP,'"<15"="15"' mpa_min[mpa_min==""]<-NA lpa_max[lpa_max==""]<-NA lpa_min[lpa_min==""]<-NA lpa_avg[lpa_avg==""]<-NA hr[hr==""]<-NA co[co==""]<-NA bnp[bnp==""]<-NA mpa_min<-as.numericas.charactermpa_min lpa_max<-as.numericas.characterlpa_max lpa_min<-as.numericas.characterlpa_min lpa_avg<-as.numericas.characterlpa_avg hr<-as.numericas.characterhr co<-as.numericas.characterco bnp<-as.numericas.characterbnp 146

PAGE 161

press_max<-bigSet3ch1$RPA_max,ch1$MPA_max,lpa_max press_min<-bigSet3ch1$RPA_min,mpa_min,lpa_min press_avg<-bigSet3ch1$RPA_avg,ch1$MPA_avg,lpa_avg sv<-bigSet2ch1$SV_Qp,ch1$SV_CO PH<-rep,lengthdemograph$Diag i<-grep"175",demograph$Diag PH[i]<-1 i<-grep"176",demograph$Diag PH[i]<-2 ##FinishChallenge1dataframe ch1<-cbindch1,press_max,press_min,press_avg,hr,co,sv demograph<-cbinddemograph,who1,who2 sv_calc<-ch1$co/ch1$hr svAll<-bigSet2ch1$sv,sv_calc co_calc<-ch1$sv*ch1$hr coAll<-bigSet2ch1$co,co_calc ch1<-cbindch1,coAll,svAll,bnp ##OrFinishChallenge3dataframe ch3<-cbindch3,press_max,press_min,press_avg,hr,co,sv 147

PAGE 162

demograph<-cbinddemograph,who1,who2 sv_calc<-ch3$co/ch1$hr svAll<-bigSet2ch1$sv,sv_calc co_calc<-ch3$sv*ch1$hr coAll<-bigSet2ch1$co,co_calc ch3<-cbindch3,coAll,svAll,bnp ##WriteCh1andCh3toafilesodon'tneedtodothiseverytimeonstart-up!! ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##PuttogetheradfwithMRN,Patient_num,Who1,Who2 ##omittingtheNA's-thisisthemaximumsetofobservations ##tobeusedbecausetheWHOvaluesarenecessary. ## ##Withthisnewsubset-createothersubsetspullingfromoriginal ##datausingtheseMRN'sandPatientNums. ## ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ #tmpdf<-data.frameMRN=demograph$MRN,Patient_num=demograph$Patient_num,Who1=who1,Who2=who2 #who.df<-na.omittmpdf##thisomitsforbothwho1andwho2-intersection ##goodfordel_who who1.tmp<-data.frameMRN=demograph$MRN,Patient_num=demograph$Patient_num, Who1=demograph$who1,BSA=demograph$BSA,PH=PH, 148

PAGE 163

Cath_date1=demograph$Cath_date1 #who1.tmp<-na.omitwho1.tmp who2.tmp<-data.frameMRN=demograph$MRN,Patient_num=demograph$Patient_num, Who2=demograph$who2,BSA=demograph$BSA,PH=PH, Cath_date2=demograph$Cath_date2 #who2.tmp<-na.omitwho2.tmp tmpdf<-NULL who1.tmp$Who1<-as.factoras.characterwho1.tmp$Who1 who2.tmp$Who2<-as.factoras.characterwho2.tmp$Who2 ##SetupTimetoWorseningbyWho delWho.df<-mergewho1.tmp,who2.tmp,by="Patient_num" delWho.df$Who1<-as.numericdelWho.df$Who1 delWho.df$Who2<-as.numericdelWho.df$Who2 delWho<-na.passdelWho.df$Who1-delWho.df$Who2 delWho.df<-cbinddelWho.df,delWho delWho.df$BSA.y<-NULL delWho.df$PH.y<-NULL delWho.df$MRN.y<-NULL delWho.df$MRN.x<-NULL whoWorse.df<-delWho.df[delWho.df$delWho<0,] whoWorse.df<-na.omitwhoWorse.df write.csvwhoWorse.df,"C:/Users/Owner/Desktop/cathDB/whoWorse.csv",na="" 149

PAGE 164

who1.tmp$Cath_date1<-NULL who2.tmp$Cath_date2<-NULL who1.tmp<-na.omitwho1.tmp who2.tmp<-na.omitwho2.tmp ##Verifiedthatnon-consentingpatientsarenotinthedf #PHT0004,PHT0010,PHT0012,PHT0017,PHT0022,PHT0030,PHT0032,PHT0033, #PHT0034,PHT0048,PHT0051,PHT0066,PHT0097,PHT0108,PHT0112,PHT0114, #PHT0115,PHT0118,PHT0138,PHT0147,PHT0150,PHT0156,PHT0158,PHT0160, #PHT0163,PHT0169,PHT0170,PHT0171,PHT0174 ##pulloutpatientswithmorethanonecath ##Patient_numisunique,MRNisspecifictoaperson tmp<-mergewho1.tmp,ch1,by="Patient_num" allDataW1Ch1.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who1=tmp$Who1,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min tmp<-mergewho2.tmp,ch1,by="Patient_num" 150

PAGE 165

allDataW2Ch1.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who2=tmp$Who2,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min tmp<-mergewho1.tmp,ch3,by="Patient_num" allDataW1Ch3.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who1=tmp$Who1,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,#Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min tmp<-mergewho2.tmp,ch3,by="Patient_num" allDataW2Ch3.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who2=tmp$Who2,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,#Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min 151

PAGE 166

tmp<-mergedelWho.df,ch1,by="Patient_num" allDataWdelCh1.df<-data.frameMRN=tmp$MRN,Patient_num=tmp$Patient_num, delWho=tmp$delWho,PH=tmp$PH.x, BSA=tmp$BSA.x,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA.x,CI=tmp$coAll/tmp$BSA.x, Pmax=tmp$press_max,Pmin=tmp$press_min tmp<-mergedelWho.df,ch3,by="Patient_num" allDataWdelCh3.df<-data.frameMRN=tmp$MRN,Patient_num=tmp$Patient_num, delWho=tmp$delWho,PH=tmp$PH.x, BSA=tmp$BSA.x,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,#Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA.x,CI=tmp$coAll/tmp$BSA.x, Pmax=tmp$press_max,Pmin=tmp$press_min tmp<-mergewho1.tmp,ch1,by="Patient_num" allDataW1Ch1BNP.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who1=tmp$Who1,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min,BNP=tmp$bnp 152

PAGE 167

tmp<-mergewho2.tmp,ch1,by="Patient_num" allDataW2Ch1BNP.df<-data.frameMRN=tmp$MRN.x,Patient_num=tmp$Patient_num, Who2=tmp$Who2,PH=tmp$PH, BSA=tmp$BSA,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA,CI=tmp$coAll/tmp$BSA, Pmax=tmp$press_max,Pmin=tmp$press_min,BNP=tmp$bnp tmp<-mergedelWho.df,ch1,by="Patient_num" allDataWdelCh1BNP.df<-data.frameMRN=tmp$MRN,Patient_num=tmp$Patient_num, delWho=tmp$delWho,PH=tmp$PH.x, BSA=tmp$BSA.x,PVRI=tmp$PVRI,SVRI=tmp$SVRI, Wedge=tmp$PCWP,PVR=tmp$PVR, Mpap=tmp$press_avg,Mrap=tmp$RAP,Aop=tmp$AoP, HR=tmp$hr,SV=tmp$svAll,CO=tmp$coAll, SVI=tmp$svAll/tmp$BSA.x,CI=tmp$coAll/tmp$BSA.x, Pmax=tmp$press_max,Pmin=tmp$press_min,BNP=tmp$bnp write.csvallDataW1Ch1.df,"C:/Users/Owner/Desktop/cathDB/allDataW1Ch1.csv",na="" write.csvallDataW2Ch1.df,"C:/Users/Owner/Desktop/cathDB/allDataW2Ch1.csv",na="" write.csvallDataW1Ch3.df,"C:/Users/Owner/Desktop/cathDB/allDataW1Ch3.csv",na="" write.csvallDataW2Ch3.df,"C:/Users/Owner/Desktop/cathDB/allDataW2Ch3.csv",na="" write.csvallDataWdelCh1.df,"C:/Users/Owner/Desktop/cathDB/allDataWdelCh1.csv",na="" write.csvallDataWdelCh3.df,"C:/Users/Owner/Desktop/cathDB/allDataWdelCh3.csv",na="" 153

PAGE 168

##~~~~RungetDiams~~~~~~ write.csvch1.diam.df,"C:/Users/Owner/Desktop/cathDB/ch1.diam.csv",na="" write.csvch3.diam.df,"C:/Users/Owner/Desktop/cathDB/ch3.diam.csv",na="" ##~~~~~~~~~~~~RuntheAICStuff~~~~~~~~~~~~~~~~~~~~ #rmlist=ls#cleanslate libraryaod#AICfunctions libraryAICcmodavg#AICmodelaveraging libraryreshape#rename libraryVGAM allDataW1Ch1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW1Ch1.csv",header=TRUE allDataW2Ch1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW2Ch1.csv",header=TRUE allDataW1Ch3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW1Ch3.csv",header=TRUE allDataW2Ch3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataW2Ch3.csv",header=TRUE allDataWdelCh1.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataWdelCh1.csv",header=TRUE allDataWdelCh3.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/allDataWdelCh3.csv",header=TRUE ch1.diam.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/ch1.diam.csv",header=TRUE ch3.diam.df<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/ch3.diam.csv",header=TRUE timeToEvent<-read.csv"C:/Users/Owner/Desktop/cathDB/Analyz/adverseEvents_sm.csv",header=TRUE 154

PAGE 169

##runWho1RoomAirCh1 DF<-mergeallDataW1Ch1.df,ch1.diam.df,by="Patient_num" DF<-na.omitDF resp_var<-DF$Who1 attachDF ##runWho2RoomAirCh1 DF<-mergeallDataW2Ch1.df,ch1.diam.df,by="Patient_num" DF<-na.omitDF resp_var<-DF$Who2 attachDF ##rundelWhoRoomAir DF<-mergeallDataWdelCh1.df,ch1.diam.df,by="Patient_num" DF<-na.omitDF resp_var<-DF$delWho attachDF ##runWho1NOch3 DF<-mergeallDataW1Ch3.df,ch3.diam.df,by="Patient_num" DF<-na.omitDF resp_var<-DF$Who1 attachDF ##runWho2NOch3 DF<-mergeallDataW2Ch3.df,ch3.diam.df,by="Patient_num" 155

PAGE 170

DF<-na.omitDF resp_var<-DF$Who2 attachDF ##rundelWhoNOch3 DF<-mergeallDataWdelCh3.df,ch3.diam.df,by="Patient_num" DF<-na.omitDF resp_var<-DF$delWho attachDF #DF<-mergeallDataW1Ch1BNP.df,ch1.diam.df,by="Patient_num" #DF<-na.omitDF #DF<-mergeallDataW2Ch1BNP.df,ch1.diam.df,by="Patient_num" #DF<-na.omitDF #DF<-mergeallDataWdelCh1BNP.df,ch1.diam.df,by="Patient_num" #DF<-na.omitDF #demo_sm<-data.framePatient_num=demograph$Patient_num, #who1=demograph$who1,Cath_date1=demograph$Cath_date1, #who2=demograph$who2,Cath_date2=demograph$Cath_date2 #eventTime<-mergeDF,demo_sm,by="Patient_num" #whoWorse_sm.cvscombinedwithAdverseEvents_sm.csv 156

PAGE 171

##Eventtime DF<-mergeDF,timeToEvent,by="Patient_num" PvrSvr<-DF$PVRI/DF$SVRI PP<-DF$Pmax-DF$Pmin delD<-DF$Dmax-DF$Dmin Strain<-delD/DF$Dmin Cdyn<-Strain/DF$PP S1<-Strain/DF$HR ##LogTransformthesedatatonormalize lnMpap<-logDF$Mpap lnPVRI<-logDF$PVRI lnPvrSvr<-logPvrSvr lnCI<-logDF$CI lnCdyn<-logCdyn lnSVRI<-logDF$SVRI lnSVI<-logDF$SVI lnPmax<-logDF$Pmax lnDmax<-logDF$Dmax dataSet="DF" 157

PAGE 172

#resp_var<-as.numericDF$Who1 resp_var<-DF$Who1 attachDF resp_var<-DF$Who2 attachDF resp_var<-DF$delWho attachDF resp_var<-DF$Time #vars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","HR","PP" #vars<-c"PVRI","Wedge","Mpap","PvrSvr","HR","Mrap","PP" #vars<-c"PVRI","Wedge","Strain","CI","mRap","diamMaxAve","delD", #"delP","mPap","RpPerRs","Cdyn","Ep" #vars<-c"PVRI","Mpap","PP","HR","Dmax","delD","Strain" vars<-c"lnPVRI","lnMpap","PP","HR","lnDmax","delD","Strain" vars<-c"lnPVRI","delD","lnMpap","Wedge","lnDmax","Strain" ##HR,SVI,CI ##GoldStandard vars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","CI" #vars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","HR" vars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","CI","HR"### ##NewGuys 158

PAGE 173

vars<-c"delD","Cdyn","Strain","PP","Pmax","Dmax" vars<-c"delD","Cdyn","Strain","PP","Pmax","Dmax","HR"### ##Non-Invasive vars<-c"delD","Strain","Dmax","HR","S1","Mpap" ##Mix vars<-c"Dmax","Strain","PVRI","Wedge","Mpap","PvrSvr" #vars<-c"lnDmax","Strain","PVRI","Wedge","Mpap","PvrSvr" vars<-c"Dmax","Strain","PVRI","Wedge","Mpap","PvrSvr","PP"##** vars<-c"delD","Cdyn","Dmax","PVRI","Wedge","Mpap","PP" vars<-c"Dmax","Strain","PVRI","Wedge","Pmax","PvrSvr","PP"##** vars<-c"delD","Cdyn","Dmax","PVRI","Wedge","Mpap","PP","Pmax" vars<-c"Dmax","Strain","PVRI","Wedge","Pmax","PvrSvr","PP","CI" vars<-c"Pmax","PvrSvr","Wedge","PP","Strain" vars<-c"Wedge","Pmax" vars<-c"Dmax","Strain","PVRI","Wedge","Mpap","PvrSvr" vars<-c"Dmax","Strain","PVRI","Wedge","Mpap" vars<-c"Dmax","Strain","PVRI","Wedge" ##WithBNP vars<-c"PVRI","Wedge","Mpap","Mrap","PvrSvr","CI","BNP" vars<-c"delD","Cdyn","Strain","PP","Pmax","Dmax","BNP" vars<-c"Dmax","Strain","PVRI","Wedge","Pmax","PvrSvr","PP","BNP" vars<-c"delD","Cdyn","Dmax","PVRI","Wedge","Mpap","BNP","Pmax" 159

PAGE 174

##createallmodelcombinationsofgivenvariables modlNames<-allCombosvars model.list<-list foriin1:lengthmodlNames$Model{ model.list[[i]]<-evalparsetext=paste"glmresp_var~",modlNames$Model[i],",family=gaussian,data=",dataSet,"" #model.list[[i]]<-evalparsetext=paste"vglmresp_var~",modlNames$Model[i],",family=multinomial,data=",dataSet,"" } #commandlinetestofvglmtocheckresults #vglmresp_var~Dmax+Strain+PVRI+Wedge,family=multinomial,data=DF ##UsethisforVGLM aicVals<-NULL llVals<-NULL aiccVals<-NULL modelResults<-NULL foriin1:lengthmodel.list{ kval<-modlNames$Kval[i] llVals[i]<-logLikmodel.list[[i]] aicVals[i]<-2*kval-2*llVals[i] aiccVals[i]<-aicVals[i]+*kval*kval+1/nrowDF-kval-1 } tmp<-data.framemodlNames$Model,aicVals,aiccVals,llVals tmp<-renametmp, 160

PAGE 175

cmodlNames="ModelName",aicVals="AIC",aiccVals="AICc",llVals="LogLikelihood" modelResults<-tmp[ordertmp$AICc,] modelResults ##UsethisforstraightGLM aicResults<-aictabcand.set=model.list,modnames=modlNames,sort=TRUE printaictabcand.set=model.list,modnames=modlNames,sort=TRUE, digits=2,LL=TRUE ##Geteachvariablesweightacrossthissetofmodels variableWts<-varWeightsaicResults,vars ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##MakeabarplotfortheVariableImportanceRanking ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ normWts<-variableWts normWts$Cumulative<-NULL barplottnormWts,main="VariableImportanceRankingbyAICWeight", ylab="VariableWeight/CumulativeWeight", xlab="PredictorVariables", 161

PAGE 176

col="lightgrey" #endbarplot ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ##Functionsusedinthiscodetomake1setascompleteaspossiblefrom ##2or3sets ##Could/shouldbe1function-set3=FALSE... ##~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ bigSet3<-functionset1,set2,set3{ bigset<-repNA,lengthset1 foriin1:lengthset1{ if!is.naset3[i] {bigset[i]<-set3[i]} if!is.naset2[i] {bigset[i]<-set2[i]} if!is.naset1[i] {bigset[i]<-set1[i]} } returnbigset } bigSet2<-functionset1,set2{ bigset<-repNA,lengthset1 foriin1:lengthset1{ 162

PAGE 177

if!is.naset2[i] {bigset[i]<-set2[i]} if!is.naset1[i] {bigset[i]<-set1[i]} } returnbigset } 163

PAGE 178

APPENDIXC.R-codeStudy2,Preconditioning ##-------------------------------------------------------------## ##Pre-ConditioningStudy ##Comparecontrolstopreconditionedarteries ##PrintthePDcurves ##CalculateandprintdelDiamterandStress-Strain ## ##-------------------------------------------------------------##RunsetUpS1S2 precondDF_orig<-s2##Keepacopyoforginalbase tmpDF<-s2##copytoscrubforthisstudyandplots tmpDF$DOB<-NULL tmpDF$Chamber<-NULL tmpDF$Hemo<-NULL tmpDF$Study<-NULL Thickness<-repNA,nrowtmpDF tmpDF<-cbindtmpDF,Thickness amcntls<-s1[s1$Treatment=="amcntl",] amcntls$Study<-NULL amcntls$Hemo<-NULL precDF<-rbindtmpDF,amcntls 164

PAGE 179

precDF$Treatment[precDF$Treatment=="amcntl"]<-"cntl" ##Takealookatthedata-notproductiongraphs: p<-ggplotaesP,D,data=precDF p<-p+geom_pointaescolor=Solution+scale_color_brewertype="qual",palette="Set1" p<-p+facet_grid.~Treatment p<-p+stat_smoothaesgroup=Solution,color=Solution p<-p+stat_summary"mean_cl_normal",geom="pointrange",mapping=aesgroup=1 ##Error:ggplot2doesntknowhowtodealwithdataofclasscharacter p<-p+xlab"PressuremmHg"+ylab"Pulmonaryarterydiametermm" p<-p+theme_bw p ##TakealookatVSMabilitytocontract-thediffbetweenpassiveandactive ##prelimplots-notproduction #Createnewpressurevectorfornewdataframe P<-seq,55,by=1 ##amcntl passive.loess<-loessD~P,span=0.8,data=tmpDF[tmpDF$Treatment=="amcntl"&tmpDF$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=tmpDF[tmpDF$Treatment=="amcntl"&tmpDF$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict 165

PAGE 180

#createatreatmentvectoramdmakeadataframetostoretheresults. Treatment<-rep"AMCNTL",lengthdeltaDCap DCap.d<-data.frameTreatment,P,deltaDCap #0.1HzPrecond active.loess<-loessD~P,span=0.8,data=tmpDF[tmpDF$Treatment=="precond"&tmpDF$Solution=="hik",] passive.loess<-loessD~P,span=0.8,data=tmpDF[tmpDF$Treatment=="precond"&tmpDF$Solution=="krebs",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict Treatment=rep"precond",lengthdeltaDCap tmp.d<-data.frameTreatment,P,deltaDCap DCap.d<-rbindDCap.d,tmp.d #rmP,Treatment,deltaDCap p<-ggplotaesx=P,y=deltaDCap,color=Treatment,data=DCap.d p<-p+geom_linesize=1+xlab"PressuremmHg"+ylab"ChangeinVesselDiametermm" p ##----------P-Dcurvesforhi-resprinting------------------##usetheloesslineforthefitforeacharteryandputthat ##datatogetherasthedatafortherestoftheanalysis ##configNewDFisinFig5_cohortWithSE.R 166

PAGE 181

krebs_data<-precDF[precDF$Solution=="krebs",] newKrebs_data<-configNewDFkrebs_data newKrebs_data<-na.omitnewKrebs_data hik_data<-precDF[precDF$Solution=="hik",] newHik_data<-configNewDFhik_data newHik_data<-na.omitnewHik_data df1<-summarySEnewKrebs_data,measurevar="newD",groupvars=c"Treatment","newP" df2<-summarySEnewHik_data,measurevar="newD",groupvars=c"Treatment","newP" Solution<-rep"krebs",nrowdf1 df1<-cbinddf1,Solution Solution<-rep"hik",nrowdf2 df2<-cbinddf2,Solution precSumDF<-rbinddf1,df2 vessels.d<-precSumDF vessels.d$Treatment<-factorvessels.d$Treatment,levels=c"cntl","precond" levelsvessels.d$Treatment<-c"Cntlnn=8","PreCondnn=5" #vessels.d$ArteryNum<-as.factorvessels.d$ArteryNum summaryvessels.d p<-ggplotvessels.d,aesx=newP,y=newD,group=Solution p<-p+geom_errorbaraesymin=newD-se,ymax=newD+se,colour="black" p<-p+geom_lineaeslinetype=Solution,size=.9 167

PAGE 182

p<-p+geom_pointaesshape=Solution,fill="white",size=3.2 p<-p+scale_linetype_manualvalues=c"solid","solid" p<-p+scale_shape_manualvalues=c,19 p<-p+facet_grid.~Treatment p<-p+xlab"nPressuremmHg"+ylab"PulmonaryArteryDiametermmn" p<-p+theme_bw p<-p+themelegend.position="none", #axis.text.y=element_blank, #axis.title.y=element_blank axis.text.x=element_textsize=15, axis.text.y=element_textsize=15, axis.title.x=element_textsize=20, axis.title.y=element_textsize=20, strip.text.x=element_textsize=14 p ggsave"C:/PreCond/Figures/PD_se.png",width=15,height=10,dpi=300 ##--------------DeltaDanindicatorofVSMcontractileability write.csvnewKrebs_data,file="C:/PreCond/newKrebs_data.csv",na="" write.csvnewHik_data,file="C:/PreCond/newHik_data.csv",na="" deltaDdf<-read.csv"C:/PreCond/precond_deltaD.csv",header=TRUE 168

PAGE 183

deltaDsum<-summarySEdeltaDdf,measurevar="deltaD",groupvars=c"Treatment","newP" deltaDsum$Treatment<-factordeltaDsum$Treatment,levels=c"cntl","precond" levelsdeltaDsum$Treatment<-c"Cntln=8","PreCondn=5" #write.csvdeltaDsum,file="C:/preCond/deltaDsum.csv",na="" #deltaDsum<-read.csv"C:/PreCond/deltaDsum.csv",na="" p<-ggplotdeltaDsum,aesx=newP,y=deltaD,group=Treatment p<-p+geom_errorbaraesymin=deltaD-se,ymax=deltaD+se,colour="black" p<-p+stat_smoothse=FALSE,colour="black",size=.9 p<-p+geom_pointaesshape=Treatment,fill="white",size=3.2 p<-p+scale_linetype_manualvalues=c"solid","solid" p<-p+scale_shape_manualvalues=c,15 p<-p+xlab"nPressuremmHg"+ylab"deltaDmmn" p<-p+theme_bw p<-p+themelegend.position=c,1, legend.justification=c,1, legend.title=element_blank, legend.key=element_blank,#nolegendborders #axis.text.y=element_blank, #axis.title.y=element_blank axis.text.x=element_textsize=15, axis.text.y=element_textsize=15, axis.title.x=element_textsize=20, axis.title.y=element_textsize=20, strip.text.x=element_textsize=14 p 169

PAGE 184

#ggsave"C:/PreCond/Figures/deltaD_se.png",width=10,height=10,dpi=300 ##-------------CalculateStress&Strain prec_krebs<-precDF[precDF$Solution=="krebs",] prec_hik<-precDF[precDF$Solution=="hik",] precSSdf_krebs<-calcStressStrainprec_krebs,1 precSSdf_hik<-calcStressStrainprec_hik,1 precSSdf_krebs_print<-calcStressStrainprec_krebs,5 precSSdf_hik_print<-calcStressStrainprec_hik,5 170

PAGE 185

APPENDIXD.R-codeStudy3,MechanicsofthePPA ##--------------------------------------------------------------------## ##Pressure-DiameterStudy ##code,data,meta-data,plots ## ##--------------------------------------------------------------------#rmlist=ls#cleanslate libraryggplot2#ggplot librarydoBy#forsummaryBy libraryplyr#forrename librarygridExtra #libraryCairo librarylattice libraryggplot2 libraryreshape2 libraryRColorBrewer librarysciplot librarynlme libraryxtable libraryMASS requiregrid libraryHmisc librarysplines 171

PAGE 186

##---Readthedatain #all_data<-read.csvfile.choose all_data<-read.csv"C:/PD/PD_artery_data.csv",header=TRUE ##--Scrubthedata-thesetissuesweredeadordidn'tworkforvariousreasons ##--Seenotebookforspecifics. all_data<-all_data[all_data$TestDate!="16-Jul",]# all_data<-all_data[all_data$TestDate!="13-Aug",]#problemswith2xmMEGTA all_data<-all_data[all_data$TestDate!="31-Jul",]#dead-ish-slippedoff #cannula,closedupforawhile,alsohikhadn'tbeenpre-bubbled all_data<-all_data[all_data$TestDate!="30-Jul",] all_data<-all_data[all_data$TestDate!="4-Oct",] all_data<-all_data[all_data$TestDate!="9-Oct",] all_data<-all_data[all_data$TestDate!="18-Nov",] ##-------------------------------------------------## ##Setuptreatmentsandlevelstoplotpressurevs.diam ## ##-------------------------------------------------vessels.d<-all_data vessels.d$Treatment<-factorvessels.d$Treatment,levels=c"cntl","3hpx","6hpx","6rtn","amcntl","nvcntl","12rtn" 172

PAGE 187

#levelsvessels.d$Treatment<-c"Control","3weekshypoxia","6weekshypoxia","Returned" levelsvessels.d$Treatment<-c"Control","3weekshypoxia","6weekshypoxia","6weeksreturned","Returncontrols","NaiveControls","12weeksreturned" vessels.d$Artery<-as.factorvessels.d$Artery summaryvessels.d ##---PlotasafunctionofPressure-prettiergraphs #p<-ggplotaesP,D,data=vessels.d p<-ggplotaesP,D,data=vessels.d p<-p+geom_pointaescolor=Solution+scale_color_brewertype="qual",palette="Set1" p<-p+facet_grid.~Treatment p<-p+stat_smoothaesgroup=Solution,color=Solution p<-p+stat_summary"mean_cl_normal",geom="pointrange",mapping=aesgroup=1 ##Error:ggplot2doesntknowhowtodealwithdataofclasscharacter p<-p+xlab"PressuremmHg"+ylab"Pulmonaryarterydiametermm" p<-p+theme_bw p ##geom_smooth ##--------------------------------------------------------------## ##Smoothmusclefunctionalability-CapacitytoShorten ## ##--------------------------------------------------------------#Createnewpressurevectorfornewdataframe 173

PAGE 188

P<-seq,60,by=1 ##ControlDcapcurve passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="Control"&vessels.d$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="Control"&vessels.d$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P #sincethesearediametersunderpressure,theactivepressureissmaller #thanthepassiveandsowesubtractactivefrompassive. deltaDCap<-passive.predict-active.predict #createatreatmentvectoramdmakeadataframetostoretheresults. Treatment<-rep"Controls:3weeksnormoxia",56 DCap.d<-data.frameTreatment,P,deltaDCap #hpx passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="3weekshypoxia"&vessels.d$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="3weekshypoxia"&vessels.d$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict Treatment=rep"3weekshypoxia",56 tmp.d<-data.frameTreatment,P,deltaDCap DCap.d<-rbindDCap.d,tmp.d #6hpx passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="6weekshypoxia"&vessels.d$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="6weekshypoxia"&vessels.d$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict 174

PAGE 189

Treatment=rep"6weekshypoxia",56 tmp.d<-data.frameTreatment,P,deltaDCap DCap.d<-rbindDCap.d,tmp.d #rtn=6rtn passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="Returncontrols"&vessels.d$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="Returncontrols"&vessels.d$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict Treatment=rep"3weekshypoxian+6weeksnormoxia",56 tmp.d<-data.frameTreatment,P,deltaDCap DCap.d<-rbindDCap.d,tmp.d #amcntl=9wkcntl passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="NaiveControls"&vessels.d$Solution=="krebs",] active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="NaiveControls"&vessels.d$Solution=="hik",] passive.predict<-predictpassive.loess,data.frameP=P active.predict<-predictactive.loess,data.frameP=P deltaDCap<-passive.predict-active.predict Treatment=rep"Controls:9weeksnormoxia",56 tmp.d<-data.frameTreatment,P,deltaDCap DCap.d<-rbindDCap.d,tmp.d #12weeksreturned=12rtn #passive.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="12weeksreturned"&vessels.d$Solution=="krebs",] #active.loess<-loessD~P,span=0.8,data=vessels.d[vessels.d$Treatment=="12weeksreturned"&vessels.d$Solution=="hik",] #passive.predict<-predictpassive.loess,data.frameP=P #active.predict<-predictactive.loess,data.frameP=P 175

PAGE 190

#deltaDCap<-passive.predict-active.predict #Treatment=rep"3weekshypoxian+12weeksnormoxia",56 #tmp.d<-data.frameTreatment,P,deltaDCap #DCap.d<-rbindDCap.d,tmp.d rmP,Treatment,deltaDCap p<-ggplotaesx=P,y=deltaDCap,color=Treatment,data=DCap.d p<-p+geom_linesize=1+xlab"PressuremmHg"+ylab"ChangeinVesselDiametermm" p ########################################################################### ## ##Calculatethechordslopeusingpointsat20mmHGand40mmHG ##calcastraightlinethroughthosepointsandreturntheslope ## ########################################################################### findSlope<-functionP,D{ fit<-loessD~P pNew<-seq,55,5#NewPressures predictD<-predictfit,pNew#NewlypredictedDiameters indx<-whichpNew%in%15 diam15<-predictD[indx] 176

PAGE 191

indx<-whichpNew%in%40 diam40<-predictD[indx] slope<-diam40-diam15/-15 returnslope #Checktomakesureitlooksgood #plotP,D,col="red" #linesloess.smoothP,D #pointspNew,predictD,col="blue",pch=2 } ##-----------------------------------------##Peakshorteningcapacity ##-----------------------------------------findPeak<-functionpHik,dHik,pKrebs,dKrebs{ fitHik<-loessdHik~pHik fitKrebs<-loessdKrebs~pKrebs pNew<-seq,55,5#NewPressures predHikD<-predictfitHik,pNew#NewlypredictedHikDiameters predKrebsD<-predictfitKrebs,pNew#NewlypredictedKrebsDiameters diff<-predKrebsD-predHikD diff<-na.omitdiff 177

PAGE 192

peak<-maxdiff returnpeak } ##---------------------------------##Meandiameterforeachcurve ##--------------------------------findMeanD<-functionP,D{ fit<-loessD~P,na.rm=TRUE pNew<-seq,55,1#NewPressures predictD<-predictfit,pNew#NewlypredictedDiameters ##findthediamataparticularpressure-sayworkingpressure #indx<-whichpNew%in%21.6 #diam20<-predictD[indx] meanD<-meanpredictD,na.rm=TRUE returnmeanD #returndiam20 } ##----------------------------------##DiameterataparticularpressureMpap 178

PAGE 193

##----------------------------------findDiam<-functiondf,atPress{ D<-df$D P<-df$P fit<-loessD~P,na.rm=TRUE pNew<-seq,55,1#NewPressures predictD<-predictfit,pNew#NewlypredictedDiameters ##findthediamataparticularpressure-sayworkingpressure indx<-whichpNew%in%roundatPress newDiam<-predictD[indx] returnnewDiam } ##---------------------------------------------------------------##ReturnthelmFitCoeficientsforachord-inparticular: ##FitthechordforPressuresbetweenabout20mmHgto40mmHg ##usethissubsetofD~Pvaluestocalculatethelmfit. ##Needtopredictthevaluesbetween20-40becausethedataiscollected ##byincreasinganddecreasingthepressure-sothereareapproximately ##twodatapointsperstretchstep.Combinethesedatapointsbyfitting ##abest-fitlinethroughthemusingloessandthenpredictnewdiameters ##usingthechordbetween20-40,thenrunlmonthenewly ##predicteddatasetandreturnthecoeficientsinterceptandslope. 179

PAGE 194

##---------------------------------------------------------------lmCoef<-functionthisArt{ fit<-loessthisArt$D~thisArt$P #pNew<-seq,40,5 pNew<-seq,40,5#NewPressures predictD<-predictfit,pNew#NewlypredictedDiameters lmCoef_return<-coeflmpredictD~pNew returnlmCoef_return } ################################################################################# ## ##CreateaDFofeveryartery.calledminiDF ##Onerowforeacharteryincluding:testdate,ratweight,treatment, ##krebschordslope,andhikchordslope,lm-Interceptandlm-Slope ## ################################################################################# #distillDF<-functionlocalDF{ libraryplyr#forrename ##-------------------------##Getthedataandsetupkrebsandhiksetsforanalysis 180

PAGE 195

##-------------------------localDF<-all_data #localDF<-tmpDF #iflocalDF==NULLlocalDF<-read.csv"C:/PD/PD_artery_data_new.csv",header=TRUE krebs_data<-localDF[localDF$Solution=="krebs",] hik_data<-localDF[localDF$Solution=="hik",] ##----------------------##Initializations ##----------------------allSlopesK<-NULL allSlopesHik<-NULL allTestDate<-NULL allRatWeight<-NULL allSolution<-NULL allOrigTreatment<-NULL allTreatment<-NULL allTimeline<-NULL allPeakCapacity<-NULL allDiamMeanK<-NULL allDiamMeanHik<-NULL allGroup<-NULL allAge<-NULL allHypoxicTime<-NULL allReturnTime<-NULL allInterceptK<-NULL allInterceptHik<-NULL 181

PAGE 196

allSlopeK<-NULL allSlopeHik<-NULL ##---------------------##Loopthroughallthearteriesmanyrowsofdatapersingleartery ##collectdatatousefromthefirstrow[1]ofeachartery ##---------------------#all_data<-na.omitlocalDF artNumMax<-maxlocalDF$ArteryNum #artNumMax<-35 foriin1:artNumMax{ slopeK<-NA slopeHik<-NA TestDate<-NA RatWeight<-NA Solution<-NA OrigTreatment<-NA Treatment<-NA Timeline<-NA dMeanK<-NA dMeanHik<-NA Age<-NA Group<-NA HypoxicTime<-NA ReturnTime<-NA lmCoefK<-NA lmCoefHik<-NA 182

PAGE 197

##------------------##Getasetofarterydata,distiguishedbysolution ##------------------tmp<-paste"",i,"",sep="" artDataK<-krebs_data[krebs_data$ArteryNum==tmp,] artDataHik<-hik_data[hik_data$ArteryNum==tmp,] ##---------------------##Justincasethereareunusedarteriesindatabase-skipthem ##Otherwise,collectspecificdatavaluesthatareconstantforeach ##artery,i.e.arterynumber,weight,date,treatment ##---------------------iflengthartDataK$P!=0&lengthartDataHik$P!=0{ slopeK<-findSlopeartDataK$P,artDataK$D slopeHik<-findSlopeartDataHik$P,artDataHik$D peakCapacity<-findPeakartDataHik$P,artDataHik$D,artDataK$P,artDataK$D TestDate<-artDataK$TestDate[1] RatWeight<-artDataK$RatWeight[1] Treatment<-artDataK$Treatment[1] #ifartDataK$OrigTreatment=="cntl"Timeline<-0 #ifartDataK$OrigTreatment=="3hpx"Timeline<-3 #ifartDataK$OrigTreatment=="6hpx"Timeline<-6 #ifartDataK$OrigTreatment=="6rtn"Timeline<-9 #ifartDataK$OrigTreatment=="12rtn"Timeline<-15 dMeanK<-findMeanDartDataK$P,artDataK$D dMeanHik<-findMeanDartDataHik$P,artDataHik$D #OrigTreatment<-artDataK$OrigTreatment[1] #Age<-artDataK$Age[1] #HypoxicTime<-artDataK$HypoxicTime[1] 183

PAGE 198

#ReturnTime<-artDataK$ReturnTime[1] lmCoefK<-lmCoefartDataK lmCoefHik<-lmCoefartDataHik allSlopesK[i]<-slopeK allSlopesHik[i]<-slopeHik allPeakCapacity[i]<-peakCapacity allTestDate[i]<-pasteTestDate allRatWeight[i]<-RatWeight #allOrigTreatment[i]<-pasteOrigTreatment allTreatment[i]<-pasteTreatment #allTimeline[i]<-Timeline allDiamMeanK[i]<-dMeanK allDiamMeanHik[i]<-dMeanHik #allAge[i]<-Age #allHypoxicTime[i]<-HypoxicTime #allReturnTime[i]<-ReturnTime allInterceptK[i]<-lmCoefK[1] allSlopeK[i]<-lmCoefK[2] allInterceptHik[i]<-lmCoefHik[1] allSlopeHik[i]<-lmCoefHik[2] } } ##--------------------------##Putthecondenseddataframetogetherandrenamethecolumns ##--------------------------miniDF<-data.frameallTestDate,allRatWeight, 184

PAGE 199

allTreatment, #allTimeline, allSlopesK,allSlopesHik,allPeakCapacity, allDiamMeanK,allDiamMeanHik, allInterceptK,allSlopeK, allInterceptHik,allSlopeHik #allAge,allHypoxicTime,allReturnTime miniDF<-renameminiDF,c"allTestDate"="TestDate","allRatWeight"="RatWeight", "allTreatment"="Treatment", #"allTimeline"="Time", "allSlopesK"="SlopeKrebs","allSlopesHik"="SlopeHik", "allPeakCapacity"="PeakCapacity", "allDiamMeanK"="meanDiamK", "allDiamMeanHik"="meanDiamHik", "allInterceptK"="lmInterceptK", "allSlopeK"="lmSlopeK", "allInterceptHik"="lmInterceptHik", "allSlopeHik"="lmSlopeHik" #TO_DO:Automate-ifnoratweightthensetto"" #miniDF$RatWeight[1]<-"" #miniDF$RatWeight[8]<-"" #miniDF<-na.omitminiDF ##HereisthenewlydistilleddataFrame-onerowperartery 185

PAGE 200

miniDF write.csvminiDF,file="C:/PD/fateDataMechPerArt.csv",na="" ##------------------------------------------------------------------------##Nowdistillthemechanicsdataonestepfurther-createsummarydataof ##theabovedistilleddataframeminiDFbyTreatment. ##Reducedownto1rowofdataperTreatment. ##------------------------------------------------------------------------fateDataMech<-summaryBySlopeKrebs+PeakCapacity+SlopeHik+meanDiamK+meanDiamHik+lmInterceptK+lmSlopeK+lmInterceptHik+lmSlopeHik~Treatment,data=miniDF, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} fateDataMech$Treatment<-gsub"amcntl","9wkcntl",fateData$Treatment,fixed=TRUE fateDataMech$Treatment<-gsub"nvcntl","0wkcntl",fateData$Treatment,fixed=TRUE fateDataMech #write.csvfateDataMech,file="C:/PD/fateDataMech.csv",na="" ##------------------------------------------------------------------------##Similarly,distillthehemodynamicdata-createsummarydataofthe ##ofthehemodynamictestsbyTreatment. ##Readinthedata,renamethevariablessotheyaremanagableinR,and ##performthesummarytoreducetheinfoto1rowofhemodataperTreatement. ##------------------------------------------------------------------------allHemo<-read.csv"C:/PD/allHemo.csv",header=TRUE allHemo<-renameallHemo, 186

PAGE 201

cSystemic.Systolic.BP..mmHg.="maxSAP", Systemic.Diastolic.BP..mmHg.="minSAP", Systemic.Mean.BP..mmHg.="mSAP", Heart.Rate..bpm.="HR", PA.Systolic.BP..mmHg.="maxPAP", PA.Diastolic.BP..mmHg.="minPAP", PA.Mean.BP..mmHg.="mPAP", Systemic.Pulse.Pressure..mmHg.="PPsys", PA.Pulse.Pressure..mmHg.="PPpa", Treatment="Treatment", Mechanics="Mechanics", study="Study", weight="Weight", ESP...mmHg.="ESP", EDP...mmHg.="EDP", PMax....mmHg.="Pmax", PMin...mmHg.="Pmin", dPMax="dPmax", dPMin="dPmin", VMax...ul.="Vmax", Vmin...ul.="Vmin", ESV...uL.="ESV", EDV...uL.="EDV", Stroke.Volume...ml.="SV", Cardiac.Output..ml.min.="CO", Ejection.Fraction.....="EjctFract", Stroke.Work...mmHg.ml.="StrokeWork", maxPwr="maxPwr", plPwr="plPwr", 187

PAGE 202

Ea...mmHg.ml.="Ea", PVA="PVA", PE="PE", Eff....="Eff", Tau..W.="TauW", Tau..G.="TauG", ESPVR...slope="ESPVRslope", V0="V0", EDPVR.stiffness="EDPVRstiff", PRSW..slope="PRSWslope", PRSW.volume.axis.intercept="PRSWvolAxis", Emax..peak="EmaxPeak", Emax....Time..ms.after.ED.="EmaxTimeAfterED" #Re-ordertheTreatments allHemo$Treatment<-factorallHemo$Treatment,levels=c"0wkcntl","cntl", "3hpx","6hpx","6rtn","9wkcntl","12rtn" fateDataHemo<-summaryBymaxSAP+minSAP+mSAP+HR+maxPAP+minPAP+mPAP+PPsys+PPpa+ Weight+ESP+EDP+Pmax+Pmin+dPmax+dPmin+Vmax+Vmin+ ESV+EDV+SV+CO+EjctFract+StrokeWork+maxPwr+plPwr+ Ea+PVA+PE+Eff+TauW+TauG+ESPVRslope+V0+EDPVRstiff+ PRSWslope+PRSWvolAxis+EmaxPeak+EmaxTimeAfterED ~Treatment,data=allHemo, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} fateDataHemo 188

PAGE 203

##-------------------------------------------------------------------##MakeoneverywidebutshallowdfcontainingeveryvariablebyTreatement ##DoublechecktheTreatmenttypestomakesurethereareusual7: ##0wkcntl,cntl,9wkcntl,3hpx,6hpx,6rtn,12rtn ## ##Datawillbeintheformof: ##.mforthemeanvalue ##.sforthestdev ##-------------------------------------------------------------------allFateData<-mergefateDataMech,fateDataHemo,"Treatment" allFateData #write.csvallFateData,file="C:/PD/allFateData.csv",na="" ##--likelysomevariablesneedtobealtered,i.e.multipliedbyafactor,etc. newD<-functionfunP,funD{ #fit<-loessfunD~funP,na.rm=TRUE #fit<-loessfunD~funP fit<-loessfunD~funP,control=loess.controlsurface="direct" pNew<-seq,55,5#NewPressures predictD<-predictfit,pNew#NewlypredictedDiameters 189

PAGE 204

returnpredictD } localDF<-all_data localDF<-s1 allDkrebs<-NULL allDhik<-NULL newP<-NULL allP<-NULL allTreat<-NULL allArt<-NULL artNumMax<-maxlocalDF$ArteryNum #artNumMax<-35 foriin1:artNumMax{ #i<-1 newDkrebs<-NULL newDhik<-NULL tmp<-paste"",i,"",sep="" #artDataK<-krebs_data[krebs_data$ArteryNum==tmp,] #artDataHik<-hik_data[hik_data$ArteryNum==tmp,] artDataK<-localDF[localDF$ArteryNum==tmp&localDF$Solution=="krebs",] artDataHik<-localDF[localDF$ArteryNum==tmp&localDF$Solution=="hik",] iflengthartDataK$P!=0{ newDkrebs<-newDartDataK$P,artDataK$D 190

PAGE 205

newDhik<-newDartDataHik$P,artDataHik$D } #appendx,values,after=lengthx allDkrebs<-appendallDkrebs,newDkrebs,after=lengthallDkrebs allDhik<-appendallDhik,newDhik,after=lengthallDhik newP<-seq,55,5 allP<-appendallP,newP,after=lengthallP tmpTreat<-pasteartDataK$Treatment tmpTreat<-tmpTreat[1:lengthnewP] allTreat<-appendallTreat,pastetmpTreat,after=lengthallTreat##--toomany allArt<-appendallArt,artDataK$ArteryNum[1:lengthnewP],after=lengthallArt } newDF<-data.frameallArt,allTreat,allP,allDkrebs,allDhik newDF<-newDF[newDF$allTreat!="12rtn",] newDF<-newDF[newDF$allTreat!="nvcntl",] df<-summarySEnewDF,measurevar="allDkrebs",groupvars=c"allTreat","allP" df<-df[-56,] df2<-summarySEnewDF,measurevar="allDhik",groupvars=c"allTreat","allP" df2<-df2[-56,] write.csvdf,file="C:/PD/pdWithSum.csv",na="" write.csvdf2,file="C:/PD/pdWithSum2.csv",na="" ##Byhand-appendedthetwofilesandadjustedthecolumnnamesandaddedSolution 191

PAGE 206

df<-read.csv"C:/PD/pdWithSum.csv",header=TRUE vessels.d<-df vessels.d$Treatment<-factorvessels.d$Treatment,levels=c"cntl","3hpx","6hpx","6rtn","amcntl","nvcntl","12rtn" #levelsvessels.d$Treatment<-c"3-WkNormoxicnControl","3-WkHypoxicnPH","6weekshypoxia","Returned" levelsvessels.d$Treatment<-c"3-WkNormoxicnControl","3-WkHypoxicnHpx","6-WkHypoxicnHpx","3-WkHpx+6-WkNXnRecovery","9-WkNormoxicnRecoveryControl","NaiveControls","12weeksreturned" vessels.d$Artery<-as.factorvessels.d$Artery summaryvessels.d #setwd"C:/PD/Figures_Final" #requiregraphics #setEPS #postscript"cohortsPD2.eps",height=10,width=20 #dev.off ##---PlotasafunctionofPressure-prettiergraphs p<-ggplotvessels.d,aesx=P,y=D,group=Solution p<-p+geom_errorbaraesymin=D-se,ymax=D+se,colour="black" p<-p+geom_lineaeslinetype=Solution,size=1.2 p<-p+geom_pointaesshape=Solution,fill="white",size=3.2 p<-p+scale_linetype_manualvalues=c"solid","dashed" p<-p+scale_shape_manualvalues=c,21 p<-p+facet_grid.~Treatment #p<-p+stat_summary"mean_cl_normal",geom="pointrange",mapping=aesgroup=1 ##Error:ggplot2doesntknowhowtodealwithdataofclasscharacter 192

PAGE 207

p<-p+xlab"PressuremmHg"+ylab"PulmonaryArteryDiametermm" p<-p+theme_bw p<-p+themelegend.position="none", #axis.text.y=element_blank, #axis.title.y=element_blank axis.text.x=element_textsize=15, axis.text.y=element_textsize=15, axis.title.x=element_textsize=20, axis.title.y=element_textsize=20, strip.text.x=element_textsize=14 p dev.off ggsave"C:/PD/Figures_Final/cohortsPD5.png",width=20,height=10,dpi=300 libraryplyr#rbind.fill #RunSetUpS1S2 #bothDF<-rbind.fills1,s2 s1<-s1[s1$Treatment!="12rtn",] s1<-s1[s1$Treatment!="nvcntl",] #s1<-s1[s1$Treatment!="6hpx",] #s1<-s1[s1$Treatment!="3hpx",] #s1<-s1[s1$Treatment!="6rtn",] 193

PAGE 208

#s1<-s1[s1$Treatment!="cntl",] #s1$Treatment<-factors1$Treatment,levels=c"nvcntl","amcntl" #levelss1$Treatment<-c"cntl","nvcntl","amcntl" #s1$Artery<-as.factors1$Artery s1$Treatment<-factors1$Treatment,levels=c"cntl","3hpx","6hpx","6rtn","amcntl" #s3<-read.csv"C:/PD/Data/PD_artery_data_precond2.csv",header=TRUE #krebsDF<-s3[s3$Solution=="krebs",] #hikDF<-s3[s3$Solution=="hik",] krebsDF<-s1[s1$Solution=="krebs",] hikDF<-s1[s1$Solution=="hik",] ##--------------------------------------------------## ##CalculateStressandStrainfromtheP-Ddata ##PuttogetheranewdataFrameforthisdataandreturnthenew ##dataFrame-alltheusualsuspectsplusStressandStrainandnewPressure ##df<-data.frameTestDate,ArteryNum,RatWeight,Solution,Treatment, ##Thickness,Stress,Strain,pNew,thinWall ## ##--------------------------------------------------calcStressStrain<-functionthisDF,Step{ #ssDF2<-NULL 194

PAGE 209

#ssDF1<-NULL ssDF<-NULL df<-NULL ##--Doforeachartery artNumMax<-maxthisDF$ArteryNum #foriin10:10{#Debugging,doonlyonce #foriin14:21{#preconditiondata #foriin14:15{ foriin1:artNumMax{ Strain<-NA Rin<-NA Rout<-NA Rmid<-NA Stress<-NA fit<-NA thisArt<-NA predictD<-NA thinWall<-NA A<-NA ##--setupasmalldataframeforjustthisarterytoworkwith tmp<-paste"",i,"",sep="" thisArt<-thisDF[thisDF$ArteryNum==tmp,] iflengththisArt$P!=0{ 195

PAGE 210

##Bestfitcurve D<-thisArt$D P<-thisArt$P fit<-loessD~P,na.rm=TRUE ##Useastep-sizeof1fordataanalysis,orifstep-sizeisspecified ##inthefunctioncall-usethat. ifmissingStep Step<-1 #pNew<-seq,55,1#NewPressures##Tryfrom5###### #pNew<-seq,55,1#Usefor1fordataanalysis,use5forPlots pNew<-seq,55,Step #Getthenewlypredicteddiametervectorbesedonthenewpressurevector #Pressurevectorisnewtomakeituniformandregular-asopposed #totherawdata predictD<-predictfit,pNew#NewlypredictedDiameters ##firstindex==NAisproblematic,pushthenextvalueintoit ##butnotspendingmoretime/effort/computationonafewpoints ##willna.omitpriortorunningsummarystatsandplotting-anylingering ##NAproblemswillbeweededoutwithna.omitoris.na ifis.napredictD[1] predictD[1]<-predictfit,pNew+.1 ##converttocircumference circf<-pi*predictD ##Use5mmHgDiamasarteryslacklengthforinitialconditions 196

PAGE 211

Do<-predictD[1] circ5mmHg<-circf[1] ##Strain=dL/Dvector #Strain<-predictD-Do/Do Strain<-circf-circ5mmHg/circ5mmHg ##Thicknessexample~=112micrometer ##convertmicrometertometers.Singlevalueofarterythickness ##atslackdiameter,heredefinedasdiameterat5mmHg #Thick<-114.25/1000000 Thick<-thisArt$Thickness[1]/1000000 ##ConvertDiam->Radius ##ConvertMillimeters->Meters ##VectorofouterRadii Rout<-predictD/2/1000 ##OuterRadius-Thickness=>InnerRadiusSingleValueofinner ##radiusatslacklengthof5mmHg R1<-Rout[1]-Thick ##Calculatetheareaofthevesselat5mmHg. ##AreaisconstantSingleValue A=pi*Rout[1]^2-R1^2 ##UseconstantAreaandallouterradiitocalculateinnerradiifor ##allpressurestepsVector Rin=Rout^2-A/pi^.5 197

PAGE 212

##Calculatethemidpointofthearterywall-halfwaybetweenouter ##andinnerradiiVector Rmid<-Rout-Rout-Rin/2 ##ConvertpressuretokPA:1mmHg=133.32pascals=.13332kPA Pin<-pNew*.13332 Pout<-0##Outerpressureassumedtobe0 ##CalculateStress!Calculatetheregularorthickwalledequation ##andthin-walledapproximation ##http://www.engineeringtoolbox.com/stress-thick-walled-tube-d_949.html ##[pi*ri2-poro2/ro2-ri2]-[ri2ro2po-pi/r2ro2-ri2] Stress<-Pin*Rin^2-Pout*Rout^2/Rout^2-Rin^2-Rin^2*Rout^2*Pout-Pin/Rmid^2*Rout^2-Rin^2 thinWall<-Pin*Rmid/Thick ##SetupthenewdataFramevaluesforeachartery newLen<-lengthStress TestDate<-repthisArt$TestDate[1],newLen RatWeight<-repthisArt$RatWeight[1],newLen Treatment<-repthisArt$Treatment[1],newLen ArteryNum<-repthisArt$ArteryNum[1],newLen Thickness<-repthisArt$Thickness[1],newLen Solution<-repthisArt$Solution[1],newLen df<-data.frameTestDate,ArteryNum,RatWeight,Solution,Treatment,Thickness,Stress,Strain,pNew,thinWall ##PutthenewdataFrametogetheraddingeachartery'svaluesaswego 198

PAGE 213

ssDF<-rbindssDF,df } } returnssDF } ssDFkrebs<-calcStressStrainkrebsDF ssDFhik<-calcStressStrainhikDF ##CallsforManuscriptPlots ssDFkrebs<-calcStressStrainkrebsDF,5 ssDFhik<-calcStressStrainhikDF,5 ##PlotwithplotStressPressure.R ##--------------------------------##BarPlotsofMechanicsusingmyfunction,myBarPlot-plotting ##thevariableofinterestwithrespecttoTreatmentcohort ##--------------------------------tmpDF<-ssDFhik tmpDF$RatWeight<-NULL#weighthasNA's-don'tneedforthis tmpDF<-na.omittmpDF sumDF<-summarySEtmpDF,measurevar="Stress",groupvars=c"Treatment" sumDF<-renamesumDF,c"Stress"="thisVar" 199

PAGE 214

p1<-myBarPlotsumDF,"StressmmHgn" p1 sumDF<-summarySEtmpDF,measurevar="Strain",groupvars=c"Treatment" sumDF<-renamesumDF,c"Strain"="thisVar" p3<-myBarPlotsumDF,"Strainmm/mmn" p3 tmpDF<-read.csv"C:/PD/Data/newMiniMechs.csv",header=TRUE tmpDF<-tmpDF[tmpDF$Treatment!="12rtn",] tmpDF<-tmpDF[tmpDF$Treatment!="nvcntl",] tmpDF$Treatment<-factortmpDF$Treatment,levels=c"cntl","3hpx","6hpx","6rtn","amcntl" tmpDF$RatWeight<-NULL#weighthasNA's-don'tneedthisthiscolumnhere tmpDF<-na.omittmpDF sumDF<-summarySEtmpDF,measurevar="SlopeHik",groupvars=c"Treatment" sumDF<-renamesumDF,c"SlopeHik"="thisVar" p5<-myBarPlotsumDF,"Compliancemm/mmHg" p5 sumDF<-summarySEtmpDF,measurevar="PeakCapacity",groupvars=c"Treatment" sumDF<-renamesumDF,c"PeakCapacity"="thisVar" p7<-myBarPlotsumDF,"VSMContractionmm" p7 sumDF<-summarySEtmpDF,measurevar="meanDiamK",groupvars=c"Treatment" sumDF<-renamesumDF,c"meanDiamK"="thisVar" p2<-myBarPlotsumDF,"MeanDiameterwithKrebsActivationmm" p2 200

PAGE 215

sumDF<-summarySEtmpDF,measurevar="diamsHik21",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik21"="thisVar" p4<-myBarPlotsumDF,"Diameterat21mmHgmm" p4 sumDF<-summarySEtmpDF,measurevar="diamsHik30",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik30"="thisVar" p6<-myBarPlotsumDF,"Diameterat30mmHgmm" p6 sumDF<-summarySEtmpDF,measurevar="diamsHik40",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik40"="thisVar" p8<-myBarPlotsumDF,"Diameterat40mmHgmm" p8 multiplotp4,p6,p8,cols=3 multiplotp1,p2,p3,p4,p5,p6,p7,p8,cols=4 ##mechPlots ##-Insteadmakeadot-plotwiththediamdata sumDF<-summarySEtmpDF,measurevar="diamsHik21",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik21"="thisVar" wrkPress<-rep"20",nrowsumDF sumDF1<-cbindsumDF,wrkPress p4<-myDotPlotsumDF1,"Diameterat21mmHgmm" p4 sumDF<-summarySEtmpDF,measurevar="diamsHik30",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik30"="thisVar" 201

PAGE 216

wrkPress<-rep"30",nrowsumDF sumDF2<-cbindsumDF,wrkPress p6<-myDotPlotsumDF2,"Diameterat30mmHgmm" p6 sumDF<-summarySEtmpDF,measurevar="diamsHik40",groupvars=c"Treatment" sumDF<-renamesumDF,c"diamsHik40"="thisVar" wrkPress<-rep"40",nrowsumDF sumDF3<-cbindsumDF,wrkPress p8<-myDotPlotsumDF3,"Diameterat40mmHgmm" p8 sumDF<-rbindsumDF1,sumDF2,sumDF3 p<-myDotPlotsumDF p ##ggsave"C:/PD/Figures/"pastethisYlab".png",width=11,height=6,dpi=300 ggsave"C:/PD/Figures/mechsBarPlots8.png",width=20,height=15,dpi=300 ##runSetupHemo hemoDF<-allHemo hemoDF<-hemoDF[allHemo$Treatment!="12rtn",] hemoDF<-hemoDF[hemoDF$Treatment!="nvcntl",] hemoDF<-hemoDF[hemoDF$Treatment!="0wkcntl",] hemoDF$Treatment<-factorhemoDF$Treatment,levels=c"cntl","3hpx","6hpx","6rtn","9wkcntl" #levelshemoDF$Treatment<-c"3-WkNormoxicnControlnn=10","3-WkHypoxicnHpxnn=9","6-WkHypoxicnHpxnn=6","3-WkHpx+6-WkNXnRecoverynn=7","9-WkNormoxicnRecoveryControlnn=8","NaiveControls","12weeksreturned" 202

PAGE 217

levelshemoDF$Treatment<-c"3-WkNormoxicnControl","3-WkHypoxicnHpx","6-WkHypoxicnHpx","3-WkHpx+6-WkNXnRecovery","9-WkNormoxicnRecoveryControl","NaiveControls","12weeksreturned" hemoDF$CO sumDF<-summarySEhemoDF,measurevar="mPAP",groupvars=c"Treatment" sumDF<-renamesumDF,c"mPAP"="thisVar" p1<-myBarPlotsumDF,"mPAPmmHg" p1 sumDF<-summarySEhemoDF,measurevar="mSAP",groupvars=c"Treatment" sumDF<-renamesumDF,c"mSAP"="thisVar" p2<-myBarPlotsumDF,"mSAPmmHg" p2 sumDF<-summarySEhemoDF,measurevar="HR",groupvars=c"Treatment" sumDF<-renamesumDF,c"HR"="thisVar" p3<-myBarPlotsumDF,"HeartRatebeats/min" p3 sumDF<-summarySEhemoDF,measurevar="CO",groupvars=c"Treatment" sumDF<-renamesumDF,c"CO"="thisVar" p5<-myBarPlotsumDF,"CardiacOutputl/min" p5 sumDF<-summarySEhemoDF,measurevar="SV",groupvars=c"Treatment" sumDF<-renamesumDF,c"SV"="thisVar" p4<-myBarPlotsumDF,"StrokeVolumel/beat" p4 sumDF<-summarySEhemoDF,measurevar="PVR",groupvars=c"Treatment" sumDF<-renamesumDF,c"PVR"="thisVar" p6<-myBarPlotsumDF,"PVRdynes*s*cm^-5" p6 203

PAGE 218

multiplotp1,p2,p3,p4,p5,p6,cols=3 ##------------------------------------## ##--------------------------------------localDF<-ssDF[!is.nassDF$Stress,] krebs_data<-localDF[localDF$Solution=="krebs",] hik_data<-localDF[localDF$Solution=="hik",] stressKall<-NULL stressHikall<-NULL treatK<-NULL treatHik<-NULL artNumMax<-maxlocalDF$ArteryNum foriin1:artNumMax{ #i<-2 stressK<-NA stressHik<-NA tmp<-paste"",i,"",sep="" artDataK<-krebs_data[krebs_data$ArteryNum==tmp,] artDataHik<-hik_data[hik_data$ArteryNum==tmp,] iflengthartDataK$pNew!=0{ stressK<-findStressartDataK,35 stressKall[i]<-stressK 204

PAGE 219

treatK[i]<-pasteartDataK$Treatment[1] } iflengthartDataHik$pNew!=0{ stressHik<-findStressartDataHik,35 stressHikall[i]<-stressHik treatHik[i]<-pasteartDataHik$Treatment[1] } } ##Analyzeataparticularpressure tmpK<-data.framestressKall,treatK tmpHik<-data.framestressHikall,treatHik distillStressK<-summaryBystressKall ~treatK, data=tmpK, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} distillStressK distillStressHik<-summaryBystressHikall ~treatHik, data=tmpHik, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} distillStressHik ##----------------------------------205

PAGE 220

##StressataparticularpressureMpap ##----------------------------------findStress<-functiondf,atPress{ S<-df$Stress P<-df$pNew #fit<-loessS~P,na.rm=TRUE #pNew<-seq,55,1#NewPressures #predictS<-predictfit,pNew#NewlypredictedDiameters ##findthediamataparticularpressure-sayworkingpressure indx<-whichP%in%roundatPress newStress<-S[indx] returnnewStress } librarymultcomp newD<-functionfunX,funY,mx,step{ #fit<-loessfunD~funP,na.rm=TRUE #fit<-loessfunD~funP fit<-loessfunY~funX,control=loess.controlsurface="direct" 206

PAGE 221

xNew<-seq,mx,step#NewPressures predY<-predictfit,xNew#NewlypredictedDiameters returnpredY } tinyDF<-NULL maxStress<-NULL meanStress<-NULL maxStrain<-NULL eMax<-NULL localDF<-ssDF localDF<-localDF[localDF$Treatment!="12rtn",] localDF<-localDF[localDF$Treatment!="nvcntl",] artNumMax<-maxlocalDF$ArteryNum foriin2:artNumMax{ tmp<-paste"",i,"",sep="" thisArt<-localDF[localDF$ArteryNum==tmp,] iflengththisArt$Stress!=0{ maxStress<-maxna.omitthisArt$Stress meanStress<-meanna.omitthisArt$Stress maxStrain<-maxna.omitthisArt$Strain meanThinWall<-meanna.omitthisArt$thinWall eMax<-maxStress/maxStrain 207

PAGE 222

Treatment<-thisArt$Treatment[1] newRow<-data.framemaxStress,meanStress,meanThinWall,maxStrain,eMax,Treatment tinyDF<-rbindtinyDF,newRow } } distill<-summaryBymaxStress+meanStress+meanThinWall+maxStrain+eMax ~Treatment,data=tinyDF, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} distill tinyDF<-tinyDF[-9,] fit<-aovmaxStress~Treatment,data=tinyDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" fit<-aovmeanStress~Treatment,data=tinyDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" 208

PAGE 223

fit<-aovmaxStrain~Treatment,data=tinyDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" fit<-aoveMax~Treatment,data=tinyDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" ##ssDFfromstressStrain.RforPRECONDITIONEDData localDF<-ssDF newSSdf<-NULL artNumMax<-maxlocalDF$ArteryNum #artNumMax<-35 foriin2:artNumMax{ #i<-1 newStrain<-NA newStress<-NA 209

PAGE 224

TestDate<-NA RatWeight<-NA Treatment<-NA ArteryNum<-NA df<-NA newLen<-NA mx<-NA step<-NA tmp<-paste"",i,"",sep="" thisArt<-localDF[localDF$ArteryNum==tmp,] iflengththisArt$Stress!=0{ mx<-distill$maxStress.m[1]##foramcntl step<-mx/10 ifthisArt$Treatment[1]=="precond"{ mx<-distill$maxStress.m[2]##forprecond step<-mx/10 } newStrain<-newDthisArt$Stress,thisArt$Strain,mx,step } newLen<-lengthnewStrain newStress<-seq,mx,step TestDate<-repthisArt$TestDate[1],newLen RatWeight<-repthisArt$RatWeight[1],newLen Treatment<-repthisArt$Treatment[1],newLen ArteryNum<-repthisArt$ArteryNum[1],newLen 210

PAGE 225

df<-data.frameTestDate,ArteryNum,RatWeight,Treatment,newStress,newStrain newSSdf<-rbindnewSSdf,df } krebsSSdf<-newSSdf sumKrebsSSdf<-summarySEkrebsSSdf,measurevar="newStrain",groupvars=c"Treatment","newStress" printDF<-sumKrebsSSdf printDF$Treatment<-factorprintDF$Treatment,levels=c"amcntl","precond" levelsprintDF$Treatment<-c"ArterieswithoutPre-Conditioning","Pre-ConditionedArteries" #printDF$Artery<-as.factorprintDF$Artery p<-ggplotprintDF,aesx=newStrain,y=newStress p<-p+geom_errorbarhaesxmin=newStrain-se,xmax=newStrain+se,colour="black",size=.7,height=1.3 p<-p+geom_lineaeslinetype=Treatment,size=1.0 p<-p+geom_pointaesshape=Treatment,fill="white",size=3.2 p<-p+scale_linetype_manualvalues=c"solid","solid" p<-p+scale_shape_manualvalues=c,19 #p<-p+facet_grid.~Treatment 211

PAGE 226

#p<-p+stat_summary"mean_cl_normal",geom="pointrange",mapping=aesgroup=1 ##Error:ggplot2doesntknowhowtodealwithdataofclasscharacter p<-p+xlab"Strainmm/mm"+ylab"ArterialWallStresskPa" p<-p+theme_bw p<-p+themelegend.position="none", #axis.text.y=element_blank, #axis.title.y=element_blank axis.text.x=element_textsize=15, axis.text.y=element_textsize=15, axis.title.x=element_textsize=20, axis.title.y=element_textsize=20, panel.border=element_blank, axis.line=element_line, strip.text.x=element_textsize=14 p localDF<-krebsSSdf modulus<-localDF$newStress/localDF$newStrain localDF<-cbindlocalDF,modulus df20mmhg<-NULL artNumMax<-maxlocalDF$ArteryNum #artNumMax<-35 foriin1:artNumMax{ 212

PAGE 227

tmp<-paste"",i,"",sep="" thisArt<-localDF[localDF$ArteryNum==tmp,] iflengththisArt$newStress!=0{ df20mmhg<-rbinddf20mmhg,thisArt[4,] } } distill2<-summaryBynewStress+newStrain+modulus ~Treatment,data=df20mmhg, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} distill2 fit<-aovnewStrain~Treatment,data=df20mmhg summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" fit<-aovmodulus~Treatment,data=df20mmhg summaryfit TukeyHSDfit parmar=c,4,6,2 213

PAGE 228

tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" distill3<-summaryBySlopeKrebs+SlopeHik+PeakCapacity+meanDiamK+meanDiamHik+lmSlopeK+lmSlopeHik ~Treatment,data=miniDF, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} distill3 fit<-aovSlopeKrebs~Treatment,data=miniDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" fit<-aovPeakCapacity~Treatment,data=miniDF summaryfit TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" fit<-aovmeanDiamK~Treatment,data=miniDF summaryfit 214

PAGE 229

TukeyHSDfit parmar=c,4,6,2 tuk<-glhtfit,linfct=mcpTreatment="Tukey" plotcldtuk,level=.05,col="lightgrey" allHemo<-read.csv"C:/PD/Data/allHemo.csv",header=TRUE allHemo<-renameallHemo, cSystemic.Systolic.BP..mmHg.="maxSAP", Systemic.Diastolic.BP..mmHg.="minSAP", Systemic.Mean.BP..mmHg.="mSAP", Heart.Rate..bpm.="HR", PA.Systolic.BP..mmHg.="maxPAP", PA.Diastolic.BP..mmHg.="minPAP", PA.Mean.BP..mmHg.="mPAP", Systemic.Pulse.Pressure..mmHg.="PPsys", PA.Pulse.Pressure..mmHg.="PPpa", Treatment="Treatment", Mechanics="Mechanics", study="Study", weight="Weight", ESP...mmHg.="ESP", EDP...mmHg.="EDP", PMax....mmHg.="Pmax", PMin...mmHg.="Pmin", dPMax="dPmax", dPMin="dPmin", 215

PAGE 230

VMax...ul.="Vmax", Vmin...ul.="Vmin", ESV...uL.="ESV", EDV...uL.="EDV", Stroke.Volume...ml.="SVml", Cardiac.Output..ml.min.="COmlmin", SV_liter="SV", CO_LperMin="CO", Ejection.Fraction.....="EjctFract", Stroke.Work...mmHg.ml.="StrokeWork", maxPwr="maxPwr", plPwr="plPwr", Ea...mmHg.ml.="Ea", PVA="PVA", PE="PE", Eff....="Eff", Tau..W.="TauW", Tau..G.="TauG", ESPVR...slope="ESPVRslope", V0="V0", EDPVR.stiffness="EDPVRstiff", PRSW..slope="PRSWslope", PRSW.volume.axis.intercept="PRSWvolAxis", Emax..peak="EmaxPeak", Emax....Time..ms.after.ED.="EmaxTimeAfterED" libraryplyr#forrename librarydoBy#forsummaryBy 216

PAGE 231

#allHemo<-read.csv"C:/PD/allHemo.csv",header=TRUE allHemo<-read.csv"C:/PD/DAtaBackupRevs/allHemoR_2.csv",header=TRUE allHemo<-renameallHemo, cSystemicSystolicBP="maxSAP", SystemicDiastolicBP="minSAP", SystemicMeanBP="mSAP", HeartRate="HR", PASystolicBP="maxPAP", PADiastolicBP="minPAP", PAMeanBP="mPAP", SystemicPP="PPsys", PA_PP="PPpa" #Re-ordertheTreatments allHemo$Treatment<-factorallHemo$Treatment,levels=c"nvcntl","cntl", "3hpx","6hpx","6rtn","amcntl","12rtn" fateDataHemo<-summaryBymaxSAP+minSAP+mSAP+HR+maxPAP+minPAP+mPAP+PPsys+PPpa+ weight+ESP+EDP+Pmax+Pmin+dPMax+dPMin+Vmax+Vmin+ ESV+EDV+StrokeVolume+CardiacOutput+EjectionFraction+StrokeWork+maxPwr+plPwr+ Ea+PVA+PE+Eff+TauW+TauG+ESPVRslope+V0+EDPVRstiff+ PRSWslope+PRSWvolAxis+EmaxPeak+EmaxTimeAfterED ~Treatment,data=allHemo, FUN=functionx{cm=meanx,na.rm=TRUE,s=sdx,na.rm=TRUE} 217

PAGE 232

fateDataHemo #write.csvfateDataHemo,file="C:/PD/fateDataHemo.csv",na="" 218

PAGE 233

APPENDIXE.SummaryDataofStudy3,PDMeasurementsand Calculations 219

PAGE 234

APPENDIXF.AnimalProtocolDetailReport 220

PAGE 235

221

PAGE 236

222

PAGE 237

223

PAGE 238

224

PAGE 239

225

PAGE 240

226

PAGE 241

227