Transfusional iron, anthracyclines and cardiac outcomes in childhood cancer survivors

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content

TRANSFUSIONAL IRON, ANTHRACYCLINES AND CARDIAC OUTCOMES IN
CHILDHOOD CANCER SURVIVORS

By

Kathleen Ruccione

______________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PREVENTIVE MEDICINE)

August 2013

Copyright 2013 Kathleen Ruccione
ii

DEDICATION

For all childhood cancer survivors.

To my father, Daniel Anthony Derse, and my brother, David Daniel Derse, who continue
to inspire me and who live on in my memory.

iii

ACKNOWLEDGEMENTS

I am deeply grateful to my superb dissertation committee members: Maryalice
Jordan-Marsh, Mark Krailo, Jean Richardson (esteemed chair), Luanne Rohrbach,
Stuart E. Siegel, and John Wood for providing a rich and ideal mix of guidance, research
acumen, and professional expertise. They have been exceedingly generous with their
time, support, and wisdom.
Many colleagues at Children’s Hospital Los Angeles and elsewhere helped make
this research a reality. I am particularly appreciative of David Freyer’s unwavering
confidence in the project’s worth and his willingness to make himself available for many
“curbside consults” as my cancer survivorship mentor. Heartfelt gratitude goes to John
Wood for his stellar conceptual thinking and for his expertise in interpreting the MRI
reports on all study participants. I am very grateful to Richard Sposto for his equally
stellar conceptual thinking and for his and Jemily Malvar’s data analytic skills. Deep
appreciation goes to our outstanding “front-line” study team (Octavio Zavala, Ada Santa
Cruz, Libertad Garcia, and Marielena Sandoval) who consistently went above and
beyond to help get the study evaluations done while showing compassion and sensitivity
to the needs of our patients and their families. I am grateful to our Clinical Trials Office
(Susan Caso, Karla Castillo, Corazon Cruz, Lakshmi Damerla, and Leo Mascarenhas),
Blood Bank (Joy Fridey and Suzy Ghazarossian), Department of Radiology (Marvin
Nelson, and Mike Cizmar, Mohammed Dar, Jennifer Dill, Art Kohatsu, Reina Morales,
Sarahi Rodriguez , Sandra Vidal), the CHLA CTSI (Edward Gomperts, and Kinnari
Jhaveri, Gannat Kilani, Suji Moon, Kosar Soufi, Rachel Tabada), and our
finance/administrative and sponsored projects staff (Christine Boyes, Dierra Morrow,
Riza Natividad-Awai, and George Weiss) – all of whom helped implement the study
where the “rubber met the road.” I am indebted to the outstanding medical students who
assisted with various vital aspects of the study: Cheng Chen, Crystal Fecht, Robert
iv

Martinez, and Kiran Mudambi). I am very thankful to Pam Hinds and Kathy Meeske for
their outstanding support and expert consultation in the realm of patient-reported
outcomes and health-related quality of life. Enduring gratitude goes to Stuart E. Siegel
and Marcio Malogolowkin for their enthusiastic support of my studies while I was a full
time faculty member in the Children’s Center for Cancer and Blood Diseases. The CHLA
“Heme Team,” particularly Tom Coates and the hematology nurses, is appreciated for
sharing their clinical experiences and insights about transfusional iron overload which
were, in fact, the spark that ignited this study.
This work would not have been possible without the support of several funding
agencies, including: the Children’s Hospital Los Angeles Clinical and Translational
Science Institute, the Concern Foundation, DAISY Foundation, ONS Foundation
(through an unrestricted grant from the Oncology Nursing Society and Sigma Theta Tau
International Foundation for Nursing), St. Baldrick’s Foundation, and the ThinkCure
Foundation. I am deeply appreciative of the vital investment these agencies made in the
project. A summary of funding secured for the project is included in the Appendix.
Grateful thanks goes to Marny Barovich for her unending practical assistance
during my time in the doctoral program, and to the outstanding faculty who have
enriched my thinking every step of the way. My friends and family are appreciated
beyond measure for being so supportive all along this journey, and more tolerant than I
had a right to expect about the impact of academic commitments on our time together.
Daniel’s sense of humor especially helped keep everything in perspective.
Finally, very special appreciation goes to each patient who participated in the
study, and to the family and friends who support them. Their participation was a gift paid
forward to help other young people with cancer in the future.
v

TABLE OF CONTENTS

Page

Dedication ii
Acknowledgements iii
List of Tables vi
List of Figures ix
List of Abbreviations xi
Abstract xiii
Chapter 1: Introduction 1
Chapter 2: Study 1
Prevalence and Distribution of Transfusion- Derived Iron Deposition
in a Sample of Childhood Cancer Survivors 45
Chapter 3: Study 2
Cardiac Outcomes in a Sample of Childhood Cancer Survivors in the
Early Post-Treatment Phase of Survivorship 66
Chapter 4: Study 3
Health Related Quality of Life among Childhood Cancer Survivors at
Risk for Cardiac Dysfunction 91
Chapter 5: Conclusion 118
References 131
Appendix 144

vi

LIST OF TABLES

Page

Table 1.1 Clinical Consequences of Iron Overload 10
Table 1.2 Potential Risks to Study Subjects 36
Table 2.1 Study Sample Clinical/Demographic Characteristics 49
Table 2.2 Hemochromatosis Mutation Analysis 51
Table 2.3 Iron Biomarkers 52
Table 2.4 Transfusion-Related Factors 52
Table 2.5 MRI Iron Quantitation in Heart, Liver, Pancreas 54
Table 2.6 Iron Concentration in Heart, Liver, Pancreas 54
Table 2.7 Spearman Rank Correlation of Iron Concentration
with Other Iron Parameters 55
Table 2.8 Univariate Analysis of Liver Iron Concentration and
Key Clinical and Demographic Variables 57
Table 2.9 Multiple Regression Analysis of Variables Predicting
Liver Iron Concentration 58
Table 3.1 Study Sample Clinical/Demographic Characteristics 70
Table 3.2 Cardiac Biomarkers in Study Cohort 73
Table 3.3 High Sensitivity C-reactive Protein by AHA/CDC
Guidelines in the Study Sample 73
Table 3.4 MRI Assessment of Cardiac Structure and Function 74
Table 3.5 Mean, SD (Range) LVEF in Study Sample Compared
to Published Reference Values 76
Table 3.6 Univariate Analysis of LVEF and Key Clinical and
Demographic Variables 77

vii

Table 3.7a Multiple Regression Analysis of Variables Predicting
LVEF: Model A 78
Table 3.7b Multiple Regression Analysis of Variables Predicting
LVEF: Model B 79
Table 3.8 Type of Malignancy, Age Peaks and Anthracycline Use 83
Table 3.9 Cardiac MRI Parameters at A Glance 86
Table 4.1 Study Sample Clinical/Demographic Characteristics 95
Table 4.2 Versions of PedsQL
TM
Generic Core Scale and Fatigue
Scale Completed by Participants and Parents/Guardians 98
Table 4.3 HRQOL by Patient and Parent Proxy Report: PedsQL
TM
Total Score, Physical Health, and Psychosocial Health 99
Table 4.4 PedsQL
TM
Generic Core Scales: At Risk Status for
Impaired HRQOL 100
Table 4.5 PedsQL
TM
Self-Reported Mean Total Scores, Physical
Health, and Psychosocial Health Summary Scores in
Various Samples 100
Table 4.6 Fatigue by Patient and Parent Proxy Report: PedsQL
TM
Multidimensional Fatigue Scale 101
Table 4.7 Self-Reported Mean PedsQL
TM
Multidimensional Fatigue
Scale Scores in Various Samples 102
Table 4.8 HRQOL by Patient and Parent Proxy Report: PedsQL
TM
Total Score, Physical Health, and Psychosocial Health 104
Table 4.9 Patient-Reported Global Health Score 106
Table 4.10 Sample Characteristics and Their Association with
Patient-Reported HRQOL 107
viii

Table 4.11 Multiple Regression Analysis of Variables Predicting
Psychosocial HRQOL 108

ix

LIST OF FIGURES
Page
Figure 1.1 Iron, Anthracyclines and Cardiac Outcomes among Childhood
Cancer Survivors: Overall Study Model 4
Figure 1.2 Iron Overload Leading to Formation of Non-Transferrin-Bound
Iron (NTBI) 7
Figure 1.3 Pathophysiology of Iron Overload 8
Figure 1.4 Haber-Weiss and Fenton Reactions 8
Figure 1.5 Proposed Pathway of Anthracycline-Induced Cardiac
Dysfunction/Death 15
Figure 1.6 Gradient Echo Images of Liver Collected at Four Different
Echo Times 18
Figure 1.7 Distribution of CHLA Childhood Cancer Survivor Cohort by
Cumulative Anthracycline Dose and PRBC Volume 28
Figure 1.8 Distribution of Sample by Cumulative Anthracycline Dose
and PRBC Volume 28
Figure 1.9 Dissertation Study Schema 36
Figure 1.10 Study Recruitment 39
Figure 2.1 Distribution of Pancreatic R2* Results in Males and Females 54
Figure 2.2 Correlation between LIC Projected from Transfusion Data and
LIC Calculated from Liver R2* 55
Figure 3.1 Distribution of Subjects Relative to Cumulative Anthracycline
Dose and PRBC Transfusion Volume 72
Figure 3.2 Cross-Section of the Human Heart 73
Figure 3.3 LVEF in Relation to Cumulative Anthracycline Dose among
Subjects who Received Anthracycline Chemotherapy 80
x

Figure 3.4 Cumulative Anthracycline Dose among Study Sample 81
Figure 4.1 HRQOL by Self-Report and Proxy Report 103
Figure 4.2 Fatigue by Self-Report and Proxy Report 103
Figure 4.3 Fitted Lines for ICC between Self-Report and Parent Proxy
Report: Psychosocial Health 104
Figure 4.4 Fitted Lines for ICC between Self-Report and Parent Proxy
Report: Physical Health 105
Figure 4.5 Fitted Lines for ICC between Self-Report and Parent Proxy
Report: Total Fatigue Score 105

xi

LIST OF ABBREVIATIONS
ALL acute lymphoblastic leukemia
AML acute myelogenous leukemia
ASTM American Society for Testing and Materials
BSA body surface area
BMI body mass index
BNP B type natriuretic peptide
C282Y HFE gene mutation
cc cubic centimeter
CCCBD Children’s Center for Cancer and Blood Diseases (CHLA)
CCSS childhood cancer survivor study
CHLA Children’s Hospital Los Angeles
CIC cardiac iron concentration
CCI Committee on Clinical Investigations (CHLA)
CPDA citrate-phosphate-dextrose-adnenine
cPRBCv cumulative packed red blood cell volume
CRP C-reactive protein
CTCAE Common Terminology Criteria for Adverse Events
CTSI Clinical and Translational Science Institute (CHLA)
DNA de-oxy ribonucleic acid
g gram
H63D HFE gene mutation
Hb, Hgb hemoglobin
HFE hemochromatosis gene
HIPAA Health Insurance Portability and Accountability Act of 1996
xii

HRQOL health-related quality of life
ICC intraclass correlation
IRB Institutional Review Board
IRP iron regulatory protein
LIC liver iron concentration
LPI labile plasma iron
LRT likelihood ratio test
LV left ventricle or ventricular
LVEF left ventricular ejection fraction
µM micromolar
mg milligram
mL milliliter
MRI magnetic resonance imaging
MUGA Multi Gated Acquisition Scan
MV mitral valve
NTBI non-transferrin bound iron
PI Principal Investigator
PRBC packed red blood cells (transfusion)
PRO patient-reported outcome
RBC red blood cell
ROS reactive oxygen species
SEER Surveillance, Epidemiology and End Results
SQUID superconducting quantum interference device
USC University of Southern California
USC CIC USC Clinical Investigations Committee
xiii

ABSTRACT

Background: Childhood cancer survivors receiving multiple packed red blood cell
(PRBC) transfusions may be at increased risk for iron deposition in vital organs such as
the heart, thereby increasing risk of serious late effects due to organ dysfunction with
associated symptoms, which may diminish their health-related quality of life (HRQOL).
Purpose: To determine the prevalence, distribution and severity of transfusional iron
deposition in heart, liver and pancreas among a cohort of childhood cancer survivors,
and to characterize the prevalence, distribution and severity of cardiac dysfunction and
pathological remodeling among a cohort of childhood cancer survivors; to assess
HRQOL using validated patient-reported outcome (PRO) measures in a cohort of
childhood cancer survivors.
Methods: Design: single institution cross-sectional study of childhood cancer survivor
cohort. Organ tissue iron was quantified by magnetic resonance imaging (MRI). Iron
status by serum markers and hemachromatosis gene mutation status were assessed, as
were cardiac biomarkers. Study subjects participated in a structured interview during
which they (and their parent/guardian, if available), completed the validated PRO
measures of HRQOL and fatigue in English or Spanish; a brief Cardiac Health Form was
also completed. In addition to data acquired from laboratory and MRI evaluations
obtained for this study, disease/treatment and demographic information was abstracted
from medical records. Data analysis included descriptive approaches for data
summarization and regression modeling for estimation and testing.
Results: Seventy-five patients were enrolled (median age 14 years, range 8-25.6 years
at evaluation); slightly more than half were male (40/75) and most were Hispanic/Latino
(45/75). Median follow up time was 4.2 years. Cancer diagnoses included ALL/AML
(n=33) and solid tumors (n=42). Sixty-seven patients had received PRBC transfusions
xiv

(median=11, range=1-47), with median cumulative PRBC transfusion volume (cPRBCv)
of 2727 mL (range, 300-14224 mL), and median cPRBCv adjusted for body weight of
90.6 mL/kg (range, 3.6-391.9). Forty-eight (64%) study participants had received
anthracycline chemotherapy as part of their treatment plan, with a median dose of
293mg/m
2
(range, 50-739mg/m
2
). Study 1: Based on liver R2* (n=73) and cardiac T2*
(n=74), tissue iron concentration (mean, median, range in mg/g) was calculated for liver
(2.6, 1.2, 0.9-23.7) and heart (0.7, 0.6, 0.5-0.9). No iron calibration exists for pancreas;
pancreas R2* results (mean, median, range in Hz) were 30.2, 26, 18-128 (n=72). Liver
iron concentration (LIC) was elevated in 36 (58.9%) patients and pancreas R2* was
abnormal in 34 (47.2%) patients, respectively. No abnormal cardiac iron concentrations
were observed. Statistically significant correlations between liver R2* and pancreatic R2*
were found. Both hepatic and pancreatic R2* were significantly positively correlated with
serum iron, iron binding capacity, serum ferritin, and cRBCv. HFE C282Y and H63D
mutation status was normal in 72/75 and 63/75, respectively. Predictors of LIC found to
be significant in univariate analyses included diagnosis of AML or solid tumor, older age
at diagnosis/age at MRI, greater treatment intensity, higher cumulative anthracycline
dose, and higher adjusted cPRBCv. In the final reduced multivariate model there were
statistically significant positive associations between elevated LIC and higher cPRBCv
(p<0.0001) and older age at diagnosis (p<0.0001). Study 2: Seventy-four study
participants completed cardiac MRI examination. There were two males in the study
sample whose left ventricular ejection fraction (LVEF) was ≤50%, a level which would
trigger clinical cardiology assessment and consideration of medical intervention.
Predictors of lower LVEF found to be significant in univariate analyses included
race/ethnicity (being Caucasian/non-Hispanic or other), diagnosis of AML or solid tumor,
older age at diagnosis or at MRI evaluation, greater treatment intensity, higher
cumulative anthracycline dose, use of dexrazoxane, larger body surface area, taller
xv

height at MRI evaluation, higher serum ferritin, and higher serum iron. Two reduced
multivariate models had nearly identical model statistics but slightly different
implications: in one model there were statistically significant associations between lower
LVEF and higher cumulative anthracycline dose (p=0.003) and older age at MRI
(p=0.009); in the other model there were statistically significant associations between
lower LVEF and the variables of higher cumulative anthracycline dose (p=0.005) and
taller height at MRI (p=0.006). Study 3: Seventy study participants and 63 parents
completed PedsQL
TM
Generic Core Scales; 71 study subjects and 63 parents completed
PedsQL
TM
Fatigue Scales. On average, childhood cancer survivors in this study had
HRQOL total scores, physical health summary scores and psychosocial health summary
scores that were approximately midway between those of pediatric patients with cancer
in active treatment and those of healthy children. In general, participants rated their
overall HRQOL as good, but there were subsets (13-17%) whose scores were ≥1 SD
below the mean, indicating at-risk status for diminished HRQOL. Study participants
endorsed more fatigue symptoms on every scale than did healthy children. On
subscales within the fatigue measure, subjects reported the fewest symptoms of general
fatigue, followed by sleep-rest fatigue, and they reported the most symptoms of cognitive
fatigue. In univariate analyses, factors found to be statistically significantly associated
with higher self-rated psychosocial HRQOL included endorsement of fewer symptoms of
fatigue and higher self-rated physical health. Risk factors for lower self-rated physical
health were female sex and lower LVEF. In multivariate analysis older age at evaluation
(p=0.0003), higher self-rated physical health (p< 0.0001) and endorsement of fewer
symptoms of fatigue (p< 0.0001) were predictive of higher self-rated psychosocial
HRQOL. A statistically significant positive linear relationship was observed between
patient self-report and parent proxy report on all aspects of HRQOL and fatigue in this
study, although there was cross-informant variance in ratings of individual items on study
xvi

measures. Participants self-rated their HRQOL slightly higher, but endorsed more
fatigue symptoms than did their parents’ proxy-reports.
Conclusion: These studies are the first to examine the prevalence of transfusional iron
deposition and cardiac dysfunction by MRI, and to evaluate associated HRQOL among a
relatively large sample of childhood cancer survivors. Study 1 demonstrated the
feasibility of using non-invasive MRI assessment of iron deposition in several organ
systems in one MRI session. Nearly 60% had elevated liver iron concentration and
nearly half had abnormal pancreatic R2* results. Findings point to the need for
development of tailored monitoring/management guidelines for childhood cancer
patients/survivors at risk of liver/pancreatic damage due to iron overload (e.g., tracking,
staging algorithms), provide evidence for adding iron overload to late effects surveillance
guidelines, and indicate the need for exploration of the additive risk of liver/pancreatic
damage from specific chemotherapeutic exposures, as well as for the design of patient
education interventions to minimize further liver/pancreatic damage. Study 2
demonstrated the feasibility of assessing cardiac structure and function by MRI relatively
early in the survivorship trajectory. Higher cumulative dose of anthracycline
chemotherapy was associated with lower LVEF, which was confirmatory of previous
studies. In contrast to most other published studies, older age and taller height at MRI
evaluation were identified as risk factors. A new finding was that higher serum ferritin
and serum iron were associated with lower LVEF. Study findings support the need for
ongoing clinical surveillance of childhood cancer survivors at risk for adverse cardiac
outcomes, and health education regarding survivors’ individualized survivorship care
plans and modifiable risk factors for heart health. Study 3: Most study participants rated
their HRQOL as good, but a significant subset may have increased risk for impaired
HRQOL. Protective factors for psychosocial HRQOL included older age, endorsement of
fewer symptoms of fatigue, and higher self-rated physical health. Risk factors for poorer
xvii

physical health were female sex and lower LVEF. Study findings point to the importance
of systematic assessment of HRQOL, fatigue and other symptoms using validated PRO
measures and obtaining information from both patient and parent whenever possible
during survivorship care, when there is a window of opportunity for targeted interventions
to mitigate adverse effects of the cancer experience. The study provides a baseline for
future comparisons that may help to better understand the dynamic nature of the
psychosocial impact of cancer during the first years of post-treatment survivorship.

Taken together, the three studies indicate a need for further research to advance
understanding of the biologic mechanisms of iron overload, including
genetic/genomic/epigenetic parameters that may mediate organ iron uptake.
Prospective, longitudinal studies are needed that: (1) link assessment of organ iron
content to organ function and to development of associated morbidity/mortality; (2) begin
at the inception of cancer treatment, and include large sample sizes as well as additional
relevant biological measures such as pubertal status, traditional cardiovascular risk
factor assessment, and cardiotoxicity-related genetic analyses; (3) explore HRQOL risk
and protective factors, and identify/evaluate targeted interventions aimed at minimizing
adverse effects of cancer treatment such as fatigue to maximize healthy survivorship.
1

CHAPTER 1: INTRODUCTION

With advances in anti-cancer therapy and supportive care, the current combined
5-year survival rate for all types of cancer occurring in young people under 20 years of
age in the United States is 81%, according to National Cancer Institute Surveillance,
Epidemiology and End Results (SEER) data (Ries., 2008). This dramatic improvement in
survival, occurring over the past four decades, has resulted in a large and growing
population of childhood cancer survivors, currently estimated to number >363,000 in the
United States (Howlader, 2012). Life-long health surveillance is recommended for these
childhood cancer survivors because cancer treatment given at an early age can
contribute to treatment-induced/accelerated chronic health conditions, which may occur
months to years after treatment completion. A landmark report from the Childhood
Cancer Survivor Study, a longitudinal study of more than 14,000 five-year survivors
diagnosed between 1970 and 1986 in the United States, showed that two-thirds of
survivors developed a chronic health condition, and more than one-third developed a
severe or life-threatening condition (Oeffinger et al., 2006). When compared with age-
and sex-matched siblings, survivors had an eight-fold increased risk of developing a
severe or life-threatening condition, with the risk of morbidity and premature mortality
continuing to increase over time (Oeffinger et al., 2006). Iron overload resulting from
packed red blood cell (PRBC) transfusions administered during childhood cancer
treatment is a recently recognized problem that may represent another clinically
significant late effect potentially contributing to excess morbidity/mortality following
childhood cancer treatment.
During cancer treatment, chemotherapy drugs and radiation therapy, as well as
the cancer disease process, can decrease bone marrow activity resulting in the
production of fewer red blood cells. Because administration of PRBC transfusions helps
overcome this problem, thereby making it possible to deliver needed cancer treatment,
2

PRBC transfusions are a mainstay of supportive care for childhood cancer patients, and
they have contributed to improved survival (Buchanon, 2005). However, relatively little
is known about the long-term consequences of multiple PRBC transfusions in this
population. One concern is that PRBC transfusions can disrupt iron homeostasis and
result in the accumulation of iron (iron overload) in many organ tissues (Kohgo, Ikuta,
Ohtake, Torimoto, & Kato, 2008). Among the major risks of iron overload are damage to
the liver, heart, and endocrine systems. In addition, it has been postulated that iron
overload can perturb the body’s immuno-regulatory balance, possibly increasing the
growth rate of cancer cells and infectious organisms, leading to (additional) cancer
development (Huang, 2003).
A substantial proportion of childhood cancer patients receive cumulative PRBC
volumes (cPRBCv) in the range known to be associated with iron overload. Transfusion-
related iron overload in pediatric oncology is incompletely characterized and only
partially understood. On the one hand, childhood cancer survivors usually terminate
their transfusions upon completion of cancer treatment, and might substantially mobilize
their iron load over time through continued growth and development. Similarly, the iron
burden of cancer survivors treated and transfused as smaller infants might be expected
to diminish over time to a greater extent than for those treated as larger, older
adolescents (Halonen et al., 2003). Alternatively, since humans have no mechanism for
passive excretion of excess iron (Andrews & Schmidt, 2007), children exposed to
multiple transfusions during cancer therapy may have lifelong exposure to toxic iron,
possibly resulting in organ dysfunction many years after cure of their malignancy.
Organ toxicity from iron overload may be potentiated by cancer chemotherapy.
Anthracycline chemotherapeutic agents used to treat many childhood cancers are
known to be associated with the development of cardiomyopathy, a serious long-term
heart condition, in dose-dependent fashion (Lipshultz et al., 1991; Steinherz, Steinherz,
3

Tan, Heller, & Murphy, 1991) . There is experimental evidence that anthracylines
increase cardiac iron deposition and that tissue-laden iron may moderate damage to
myocardial cells caused by anthracyclines (B. Anderson, 2005; Kwok & Richardson,
2003, 2004; Link, Tirosh, Pinson, & Hershko, 1996; Minotti, Cairo, & Monti, 1999;
Minotti, Ronchi, Salvatorelli, Menna, & Cairo, 2001; Panjrath et al., 2007; Simunek et al.,
2008; Xu, Persson, & Richardson, 2005). Therefore, patients who receive treatment
with anthracylines and multiple PRBC transfusions may be at increased risk for
developing cardiomyopathy.
Utilizing a large, established, single institution (Children’s Hospital Los Angeles,
CHLA) cohort of childhood cancer survivors treated between June 1, 2004 and
December 31, 2009, and representing a variety of diagnoses and ranges of treatment
intensity, this dissertation study sought to determine the prevalence of cardiac iron
overload, measured by MRI, and to analyze its influence on the risk for cardiomyopathy,
alone and in the context of other host- and treatment-related factors. It was hypothesized
that there would be evidence of increased myocardial iron deposition among survivors
who received large cPRBCv. The study assessed for indications of cardiac dysfunction,
hypothesizing that there would be evidence of impaired left ventricular (LV) function due
to myocardial iron deposition in survivors who received large cPRBCv, even after
controlling for anthracycline exposure and other treatment and host factors. The study
also evaluated selected patient-reported outcomes (PROs) under the rubric of health-
related quality of life (HRQOL). HRQOL constructs included physical and psychosocial
health (emotional functioning, social functioning, school functioning), and symptoms of
general fatigue, sleep/rest fatigue, and cognitive fatigue. Patient-reported fatigue and
HRQOL were evaluated for possible associations with transfusion history, organ iron
content, anthracycline exposure and impaired LV function. It was hypothesized that self-
reported indicators of HRQOL would be lower in survivors exhibiting cardiac iron
4

deposition and impaired LV function. In addition, the study sought to acquire data
regarding the prevalence of liver and pancreatic iron deposition among childhood cancer
survivors as an exploratory aim. The conceptual model for the study is depicted in
Figure 1.1
Figure 1.1

1.1 Iron Biology and Iron Overload
1.1.1 Hematological supportive care for anemia
Red blood cells (RBCs, also called erythrocytes), mainly produced in the bone
marrow, contain >270 million molecules of hemoglobin, an iron- and oxygen-binding
protein, in their cytoplasm. RBCs continuously circulate throughout the body so that
hemoglobin can take up oxygen in the lungs, deliver it in exchange for carbon dioxide in
muscles and organ tissues, and then transport the carbon dioxide back to the lungs
5

where oxygen is inhaled and carbon dioxide is exhaled. Anemia
1
(Barnard, 2004) is
defined as a deficiency of healthy RBCs or hemoglobin, leading to a reduction in the
oxygen-carrying capacity of blood (Agrawal, 2011). Anemia is common during treatment
for most forms of childhood cancer due to blood loss and impaired production. Signs and
symptoms of severe anemia (including, for example, loss of appetite, headache,
dizziness, fatigue, vertigo, tinnitus, dyspnea, irritability, faintness, pallor, inactivity,
changes in heart function, loss of concentration, changes in behavior and poor school
performance) compound patients’ suffering and diminish their quality of life during
cancer treatment. In addition, anemia may complicate effective cancer therapy and
adversely affect survival (Agrawal, 2011). PRBC transfusions constitute standard
therapy for treatment-induced anemia because they are simple, fast, effective, available,
relatively safe, and provide measurable direct benefit (Agrawal, 2011).
1.1.2 Iron biology and normal iron homeostasis
Iron is a trace element with an essential role in mammalian biological processes.
All cells require iron for normal functioning (MacKenzie, Iwasaki, & Tsuji, 2008). Iron is a
component of molecules that sense, transport and store oxygen and of enzymes
involved in energy production, intermediate metabolism, and host defense (Ganz &
Nemeth, 2011b). However, free soluble (“labile”) iron is highly toxic because it can
catalyze the production of free radicals from reactive oxygen species (ROS), which
damage cellular structures. To protect from such damage, iron atoms are bound to
proteins in the body. Iron-binding proteins include hemoglobin and myoglobin, as well as
the iron storage proteins, ferritin and hemosiderin, and the transporter protein,
transferrin. Hemoglobin (Hb, Hgb), as noted previously, carries and releases oxygen

1
Guidelines for assessment of anemia from the Children’s Oncology Group (COG) state that red
blood cell mass should be adequate to maintain oxygen carrying capacity and tissue oxygen
delivery, for which hemoglobin >7g/dl is usually adequate, thus defining anemia as hemoglobin
<7g/dl.
6

and carbon dioxide via the blood; myoglobin is involved in oxygen storage and transport
in skeletal muscles, including the heart (Meyer, 2004; Schechter, 2008). Dietary iron
circulates in plasma bound to transferrin, the principal iron-transporter protein, and is
stored in cells in the form of ferritin or hemosiderin (Andrews, 1999). Hemosiderin is
thought to be derived from ferritin and to represent a more stable and less available form
of stored iron than ferritin (Andrews, 2009).
Given its essential nature, iron is carefully conserved in the body, in part by
having no active physiologic mechanism for its excretion. Therefore iron homeostasis
requires tight regulation of iron uptake to counter its toxicity while assuring an adequate
supply for its functions (Ganz & Nemeth, 2011a; Siah, Ombiga, Adams, Trinder, &
Olynyk, 2006; Zhang & Enns, 2009). The concentration of iron in plasma and
extracellular fluid is maintained within a relatively narrow range of 10-30µM in healthy
individuals (Ganz & Nemeth, 2011a). Daily losses of about 1-2mg of iron occur through
sloughing of epithelial cells in the skin and intestine, and through sweating and minor
bleeding. These losses are replaced through intestinal absorption of 1-2mg of dietary
iron daily. The total amount of iron stored in the body is approximately 3-4g, distributed
in the body with roughly 1g in the liver and 2g in the blood (Agrawal, 2011). RBCs have
a lifespan of about 120 days so some RBCs become senescent every day and are
degraded by macrophages, which recycle the RBC iron to plasma (Ganz & Nemeth,
2011b).
In the current, evolving understanding of iron regulation, dietary absorption,
storage, and tissue distribution of iron are controlled systemically by the hepcidin-
ferroportin axis,and intracellularly by key iron regulatory proteins (IRPs) interacting with
iron-responsive elements (IREs) (Andrews, 2008; Ganz & Nemeth, 2011a; Rouault,
2006). Hepatocytes (liver cells) store iron, while also serving as endocrine cells that
produce the protein hepcidin (Ganz & Nemeth, 2011b). Hepcidin is regulated by
7

Figure 1.2.
Iron Overload Leading to Formation
of Non-Transferrin-Bound Iron (NTBI)
Source: www.ironhealthalliance.com
feedback from iron concentrations in plasma and the liver, as well as by the demand for
iron needed for the production of RBCs (erythropoiesis). This regulation is coordinated
by multiple proteins acting through various signaling pathways (Rouault, 2006; Zhang &
Enns, 2009). Ferroportin is a protein that serves as the conduit for transferring cellular
iron to plasma, and to date it is the only known molecular target for hepcidin (Ganz &
Nemeth, 2011a). When hepcidin binds to ferroportin on iron-exporting cells, it negatively
regulates the delivery of iron to plasma from sources such as the duodenal absorption of
dietary iron, release of recycled iron by macrophages, and release of stored iron from
hepatocytes. Thus, when iron is deficient, hepcidin production allows more iron to enter
plasma, but when iron is abundant, more hepcidin is produced to limit further iron
absorption and release from body stores (Ganz & Nemeth, 2011b).
1.1.3 Pathophysiology of iron overload
The homeostatic balance of iron storage and distribution can be disrupted when
exogenous iron is loaded by genetic factors, disease, or through repeated transfusions
(Kohgo et al., 2008). Circulating
iron is bound and transported by
transferrin, which is about 30%
saturated under normal balanced
conditions (Wood, 2008). In iron
overload (Figure 1.2), plasma
transferrin becomes fully
saturated and non-transferrin
bound iron (NTBI) can circulate. Labile plasma iron (LPI), the most weakly bound portion
of NTBI, accumulates in the cytoplasm of macrophages and then permeates cells of the
liver, heart, pancreas and other organs. As shown in Figures 1.3 and 1.4, high levels of
labile cellular iron generate oxygen radicals (reactive oxygen species, ROS) through
8

Figure 1.3.
Pathophysiology of Iron Overload
Source: www.ironhealthalliance.com
Haber-Weiss and Fenton reactions
(Gattermann & Rachmilewitz, 2011;
Kohgo et al., 2008). Eventually,
ROS production exceeds the cells’
antioxidant capacity, causing lipid
peroxidation, oxidation of amino
acids, protein fragmentation, and
DNA damage resulting in tissue
damage and organ dysfunction
(Fleming & Ponka, 2012; Gattermann & Rachmilewitz, 2011). Undetected, untreated
iron toxicity causes progressive organ damage that can ultimately be fatal.

Figure 1.4.
Haber-Weiss and Fenton Reactions
Source: Kell, D.B. (2010). Arch Toxicol, 84(11), 825-889 (p. 826)

9

1.1.4 Clinical consequences of transfusion-related iron overload
Since each unit of blood (approximately 500 cc whole blood or 250 cc PRBCs)
contains 200-250mg of iron, multiple PRBC transfusions can result in the administration
of large amounts of transfused iron over time. It is estimated that as few as 10-20
lifetime PRBC transfusions can result in transfusional iron toxicity (Porter, 2001). Iron-
overloaded patients may be asymptomatic until damage impairs function sufficiently to
cause symptoms related to the affected organ. NTBI preferentially enters certain cells,
including hepatocytes, cardiomyocytes, anterior pituitary cells, and pancreatic beta cells
(Brittenham, 2011). Thus, anatomical sites affected by iron overload include the liver,
heart, and endocrine systems, as listed in Table 1.1 and described below.
Liver. Chronic excessive iron accumulation in hepatocytes causes hepatocellular
injury, which leads to the development of fibrosis and cirrhosis (Bonkovsky, 1991). Iron
overload has been associated with hepatocellular carcinoma, especially in the presence
of hepatitis B or C infection or cirrhosis in studies of patients with hemochromatosis
(Huang, 2003; Lauret et al., 2002). In addition, hepatic disturbance of glucose utilization
accelerates pancreatic beta cell depletion due to hyperinsulinemia, leading to glucose
intolerance (Kohgo et al., 2008). Given that the liver is the main iron storage site in the
body, secondary iron overload appears first in the liver, followed by other extrahepatic
organ involvement (Kohgo et al., 2008).
Heart. In the heart, iron toxicity is thought to occur from the direct effect of NTBI
on cardiac muscle cells (myocytes). Iron deposition begins by entry of labile plasma iron
through L-type calcium channels; once inside the myocytes, iron is bound to ferritin and
transported to lysosomes for degradation and long-term storage (Gujja, Rosing, Tripodi,
& Shizukuda, 2010; Wood, 2008). When the capacity of cellular antioxidant systems is
exceeded, ROS-triggered free radical damage interferes with electrical and mechanical
processes and results in myocyte apoptosis (Oudit, Trivieri, Khaper, Liu, & Backx, 2006;
10

Wood, 2008). Clinical manifestations of cardiac iron overload include cardiomyopathy,
arrhythmias, systolic and diastolic dysfunction, and h
eart failure (Wood, Enriquez, Ghugre, Otto-Duessel, et al., 2005). In chronically
transfused patients with thalassemia, Noetzli and colleagues showed that pancreatic iron
loading preceded and predicted
cardiac iron loading (Noetzli,
Papudesi, Coates, & Wood, 2009)
Pancreas. Pancreatic beta
cells can be damaged by iron
overload, leading to glucose
intolerance and diabetes mellitus
(Kohgo et al., 2008). Pancreatic
iron toxicity may be moderated by other factors; for example, a study of thalassemia
patients found that pancreatic iron was the strongest predictor of beta cell toxicity, but
total body iron burden, age, and body habitus also influenced glucose regulation
(Noetzli, Mittelman, Watanabe, Coates, & Wood, 2011).
Pituitary and other endocrine organs. The anterior pituitary gland has been
shown to be sensitive to early toxic effects of transfusional iron deposits in children with
thalassemia (Kletzky et al., 1979). Pituitary iron deposition can cause endocrine
disturbances leading to delayed sexual maturation, growth failure, problems with bone
density, and infertility (Chatterjee & Katz, 2000; Costin, Kogut, Hyman, & Ortega, 1979;
Schafer et al., 1981). Other endocrine glands that can be affected by iron overload
include the thyroid and parathyroid gland. Functional hypothyroidism and
hypoparathyroidism have been observed in transfused thalassemia patients (Borgna-
Pignatti et al., 2004; Vogiatzi et al., 2009).
Table 1.1.
Clinical Consequences of Iron Overload
Organ or system Examples
Liver

• Portal fibrosis, cirrhosis
• Hepatocellular carcinoma
Heart

• Arrhythmias
• Cardiomyopathy
• Heart failure
Endocrine

• Diabetes mellitus
• Growth hormone deficiency
• Hypoparathyroidism
• Thyroid dysfunction
• Adrenal insufficiency
• Delayed puberty
• Gonadal failure/infertility
11

In addition to organ dysfunction from long-term iron exposure, there is an
association with development of malignancy and vascular disease. Hepatocellular
carcinoma has become a significant cause of death in patients with thalassemia
(Benetatos, Alymara, Vassou, & Bourantas, 2008) and reduction in iron by phlebotomy
has resulted in a reduction in new cancer risk and all-cause mortality in patients
diagnosed with cancer (Zacharski, 2010; Zacharski et al., 2008).
1.1.5. Treatment of transfusion-related iron overload
Importantly, when iron overload is detected treatment can be initiated to reduce
iron levels and the potential organ damage caused by high iron levels. Treatment
modalities include iron chelating medications and therapeutic phlebotomy. By binding
iron in complexes and promoting their excretion, chelating agents clear NTBI, rid cells of
excess iron, and restore safe levels of body iron (Brittenham, 2011). In North America,
two iron chelating medications are approved for use, one given by subcutaneous or
intravenous infusion over 8-12 hours 5-7 days/week (deferoxamine mesylate, Desferal)
and a newer orally absorbed iron chelator given on a once daily schedule (deferasirox,
Exjade). Therapeutic phlebotomy is the removal of blood to reduce iron through a
procedure similar to that used for volunteer blood donation (Cook, 2010).
1.1.6 Transfusion-related iron overload in childhood cancer survivors
Experience with transfusional iron overload comes primarily from patients with
hematologic disorders such as thalassemia, sickle cell disease, Diamond-Blackfan
syndrome, and conditions of bone marrow failure such as aplastic anemia and
myelodysplastic syndrome requiring multiple transfusions. Preliminary evidence for
transfusional iron overload in pediatric oncology patients has been reported in 13
publications, 5 of which are abstracts (Blair, 2009; Eng & Fish, 2011; Gurram, Newman,
& Kobrinsky, 2012; Halonen et al., 2003; Jaffer, 2012; Jastaniah, 2008; Lutz, von
Komorowski, Durken, Engelhardt, & Dinter, 2008; Nelson, 2007; Nottage et al., 2012;
12

Reddy, 2011, 2012; Ruccione et al., 2012; Vag, 2011). Most of these represent relatively
small single-institution studies and case reports utilizing various methods of direct tissue
iron measurement in patients diagnosed with acute leukemia. Total body iron was
estimated in two of the recent studies on the basis of PRBC transfusions received. In a
study of 107 children with acute lymphoblastic leukemia (ALL), higher cumulative
transfusion volume was found to be associated with high risk ALL (Eng & Fish, 2011).
In a study of 214 children with various malignancies (described in section 1.5,
“Preliminary Work”), higher cumulative transfusion volume – and hence higher projected
liver iron burden – was associated with greater treatment intensity (Ruccione et al.,
2012). A summary of reports of transfusion-related iron overload in childhood cancer
patients is included in the Appendix.
1.2 Cardiac Outcomes in Childhood Cancer Survivors
1.2.1 Anthracycline-induced cardiotoxicity
Results from the Childhood Cancer Survivor Study (CCSS) have shown that
within 30 years after diagnosis, childhood cancer survivors have significantly increased
risk of chronic health conditions, with cardiac complications being the leading non-
cancerous cause of death and disability (Oeffinger et al., 2006). Compared with their
siblings, survivors in the large CCSS cohort had a seven-fold increased risk of death
from cardiovascular events (Mertens et al., 2008; Mertens et al., 2001), and they were
five to seven times more likely to report cardiac complications (Mulrooney et al., 2009).
Risk factors for late cardiotoxicity in general include female sex, mediastinal irradiation,
younger age at start of treatment, and higher cumulative dose of anthracycline
chemotherapy (Tukenova et al., 2010).
More than 50% of children with cancer receive treatment that includes
anthracycline agents, most notably daunorubicin and doxorubicin (Kremer & Caron,
2004; Lipshultz & Adams, 2010). Anthracyclines (also including idarubicin/epirubicin and
13

a similar drug, mitoxantrone) are highly effective, contributing to dramatic increases in
survival since their introduction (Adamson, 2011; VanderHeide, 2007). However, this
success has been tempered by the recognition that anthracycline drugs carry a well-
recognized dose-dependent risk of life-threatening cardiotoxicity. Cardiac abnormalities
develop in up to 60% of patients who receive higher cumulative doses of anthracyclines
(Bryant et al., 2007; Kremer, van der Pal, Offringa, van Dalen, & Voute, 2002;
Mulrooney et al., 2009; Wouters, Kremer, Miller, Herman, & Lipshultz, 2005). Recently, it
has also been reported that low to moderate cumulative doses can produce
cardiotoxicity in some patients (Blanco et al., 2011; Mulrooney et al., 2009).
Established risk factors for anthracycline-specific cardiotoxicity include young
age at time of treatment, female sex, concomitant chest irradiation or other
antineoplastic drugs, length of post-therapy interval, and cumulative dose (Lipshultz,
Alvarez, & Scully, 2008; K. Sorensen, Levitt, Bull, Dorup, & Sullivan, 2003; Trachtenberg
et al., 2011). Single nucleotide polymorphisms in carbonyl reductase genes and several
other genes have been associated with increased anthracycline cardiomyopathy risk in
two studies (Blanco et al., 2011; Visscher et al., 2011). The relative risks of various
comorbidities (e.g., diabetes, hypertension, endocrinopathies) and lifestyle factors (e.g.,
diet, alcohol, smoking, physical activity, substance use) in anthracycline cardiotoxicity,
although clinically relevant, have been less well studied among childhood cancer
survivors (Trachtenberg et al., 2011).
Anthracycline cardiotoxicity may be acute or chronic. Acute cardiotoxicity, usually
manifested as conduction disturbances and arrhythmias, is generally reversible when
anthracyclines are discontinued (Berry & Jorden, 2005; Lipshultz et al., 2008). Chronic
cardiotoxicity, which can have an early onset (<1 year) or begin later (>1 year) after
treatment completion, is characterized by left ventricular (LV) dysfunction with clinical
features of restrictive and/or dilated cardiomyopathy and arrhythmia (Barry, Alvarez,
14

Scully, Miller, & Lipshultz, 2007; Lipshultz et al., 2008). Importantly, routine surveillance
echocardiography has been found to yield subclinical abnormalities as early as one year
post-treatment even in survivors considered to be at low risk of anthracycline
cardiotoxicity (Abosoudah, Greenberg, Ness, Benson, & Nathan, 2011), suggesting the
progressive nature of anthracycline cardiotoxicity. Chronic cardiotoxicity can lead to
cardiac failure requiring medical management, possible cardiac transplant, and
ultimately may result in premature cardiac death (Lipshultz & Colan, 2004; Ward et al.,
2004).
Anthracycline cardiomyopathy is thought to result from cardiac myocyte damage
and apoptosis, thereby adversely affecting myocyte structure/function or decreasing the
absolute number of myocardial cells. After cardiac injury, a process termed pathologic
remodeling occurs. As defined by Cohn and colleagues, cardiac remodeling includes
genome expression, molecular, cellular and interstitial changes that are manifested
clinically as changes in structure or geometry (dimensions, mass, shape) and function of
the left ventricle (Cohn, Ferrari, & Sharpe, 2000). With remodeling, the left ventricular
wall of the heart thins and dilates, and the increased afterload further increases stress
on the heart (Sawyer, Peng, Chen, Pentassuglia, & Lim, 2010; Wouters et al., 2005).
Progression from cardiomyocyte injury to decreased LV wall thickness, increase in LV
dimension, and increase in LV end-systolic wall stress leads to heart failure. Pathologic
remodeling affects the heart’s ability to meet the increased demand that occurs with
normal growth, pregnancy or other cardiac stresses (Lipshultz et al., 2008).
The precise molecular and cellular mechanisms underlying anthracycline
cardiotoxicity are incompletely understood (Gianni et al., 2008; Vergely, 2007).
Production of ROS resulting in free-radical mediated cardiomyocyte damage/death by
oxidative stress is the most frequently advanced explanatory model, based on evidence
from animal and cell culture studies (M. J. Adams & Lipshultz, 2005; Barry et al., 2007;
15

Figure 1.5.
Proposed Pathway of Anthracycline-Induced
Cardiac Dysfunction/Death
(ODFR, oxygen-derived free radicals)

Source: VanderHeide, R. S., L'Ecuyer, T.J. (2007). Heart
Metabolism, 35, 1-4, p.2

Lipshultz et al., 2008; Vergely, 2007). Oxidative stress damages DNA, proteins and
lipids in myocytes. Two possible converging pathways for oxidative stress-induced
anthracycline cardiotoxicity, shown in Figure 1.5, include a non-enzymatic pathway
utilizing iron and an enzymatic
mechanism using the
mitochondrial respiratory chain
(VanderHeide, 2007; Vergely,
2007). In the oxidative stress
model, it is suggested that
anthracyclines may selectively
damage the heart because (1)
its myocytes have abundant
mitochondria with high
concentrations of cardiolipin
essential for mitochondrial energy metabolism; (2) cardiolipin’s high affinity for
anthracyclines allows increased concentrations to enter the mitochondria; and (3) the
myocardium has fewer natural antioxidants than other organs and glutathione, one of its
antioxidants, is depleted in the presence of anthracycline (Trachtenberg et al., 2011).
Another line of research, using mouse cell cultures and genetically engineered mice,
suggests that the cardiomyocyte damage from anthracyclines may begin independent of
ROS, resulting instead from anthracycline interactions with cellular topoisomerase II,
which damage DNA and disrupt mitochondrial function (Sawyer, 2013).
Additional alternative hypotheses not dependent on an oxidative stress pathway
are being investigated. Examples of several of these possible mechanisms,
summarized by Wouters and colleagues (Wouters et al., 2005) include transcriptional
changes in intracellular ATP production in cardiac myocytes; reduced expression of
16

mRNA encoding for an enzyme leading to decreased cardiac contractility; and
prolonged drug-related depression in cardiac glutathione peroxidase activity. A recent
report suggests a role for increased carbonyl reductase gene expression with
consequent increase in the synthesis of cardiotoxic anthracycline alcohol metabolites in
the genesis of anthracycline cardiotoxicity (Blanco et al., 2011).
1.2.2 Role of iron in anthracycline-induced cardiotoxicity
Iron has been implicated in anthracycline cardiotoxicity since 1982 on the basis
of in vivo and in vitro experimental data indicating that free radicals, generated by
anthracyclines or produced from the formation of anthracycline-iron complexes, damage
heart cells (Myers, Gianni, Simone, Klecker, & Greene, 1982; Simunek et al., 2008).
Also, mouse and human data show increased anthracycline toxicity in individuals with
iron overload conditions (Cascales et al., 2010; Hershko, 1996; Link et al., 1996;
Miranda et al., 2003; Panjrath et al., 2007) . Further support for the role of iron in
cardiotoxicity is derived from the observation that dexrazoxane, an iron chelator, is an
effective anthracycline cardioprotectant (B. Anderson, 2005; Lipshultz et al., 2010),
although cardioprotection by iron chelation may occur through mechanisms other than
by inhibition of ROS-induced injury (Brittenham, 2011; Simunek et al., 2008).
Other possible iron-mediated effects that do not depend on ROS are being
investigated, including dysregulation of iron homeostasis through interaction with iron
regulatory proteins, decreasing binding to the iron responsive elements of mRNA and
modifying expression of the proteins needed for maintaining optimal intracellular iron
levels (Kaiserova et al., 2007; Minotti et al., 2001; Simunek et al., 2008). Another non-
ROS dependent mechanism being studied is the inhibition of iron mobilization from
ferritin (Kaiserova et al., 2007; Kwok & Richardson, 2003, 2004; Simunek et al., 2008;
Xu et al., 2005).
17

1.3 Measurement of organ iron deposition, cardiac structure and function
1.3.1 Indirect measurement
The most convenient test available to estimate body iron stores is serum ferritin.
Although the precise source of serum ferritin is not known, it appears to correlate with
the larger amount of cellular ferritin, reflecting total body iron stores in normal individuals
(Brittenham, Sheth, Allen, & Farrell, 2001; Fischer & Harmatz, 2009; Kohgo et al., 2008).
Trends in total body iron stores, reflecting iron deposition in all organs and systems, can
be monitored by periodic measurements of serum ferritin. However, because ferritin
levels are known to vary with clinical conditions other than iron overload (such as chronic
inflammation, chronic liver damage, and malignancies), as well as with intensity of
transfusion therapy, its value as an independent marker of iron balance is limited (Kohgo
et al., 2008; Wood, 2011).
1.3.2 Direct measurement
Since the liver is the major organ for iron storage and liver iron concentration
(LIC) correlates closely with total body iron stores, direct measurement of LIC by liver
biopsy has been the gold standard for accurately assessing body iron stores in the past;
however, liver biopsy is invasive, costly, and subject to sampling error (Fischer &
Harmatz, 2009; Wood, 2007). Direct measurement of cardiac iron concentration (CIC)
by cardiac biopsy is rarely performed because of the invasiveness and high technical
risk of the procedure and because of the possibility of sampling error due to
inhomogeneous distribution of iron in myocardial tissue (Fischer & Harmatz, 2009;
Kohgo et al., 2008). Newer non-invasive iron quantitative procedures include
biomagnetic liver susceptometry measured using a superconducting quantum
interference device (SQUID), and iron quantitation using magnetic resonance imaging
(MRI) (Kohgo et al., 2008). SQUID equipment has very limited availability. MRI has
become the dominant technique because of its availability, as well as its sensitivity,
18

The top four images were collected from a patient having a liver iron of 6
mg/g. The bottom four images were collected from a normal volunteer. All
images darken as the echo time (TE) lengthens, but the iron-heavy tissue
darkens faster. The half life of this process is called T2* and the rate is
called R2* (R2*=1000/T2*). Source: Wood, J.C. (2007). Curr Opin
Hematol, 14(3): 183-190, p. 184.

Figure 1.6.
Gradient Echo Images of Liver Collected at Four Different Echo Times
reproducibility, and the advantage of being able to image multiple organs in a single
imaging session (Wood, 2007).
The principles of MRI-imaging for iron measurement, as described by Wood
(Wood, 2007), are that the MRI transmits a microwave signal into the body, which
excites water protons into a higher magnetic state. As the protons relax back to an
unexcited state, the signal returning from the body is interpreted by the scanner and
generates an image. If the tissue is iron-loaded, the iron deposits act like tiny magnets
placed in a strong magnetic field. They disrupt the protons and darken the image more
quickly than in non-iron overloaded tissues (Figure 1.6). Results are reported as T2*
values, representing the time needed for a tissue to become twice as dark, or as R2*, its
rate of darkening (Wood, 2011). R2* is directly proportional to iron concentration and its
values are reciprocal to T2* (R2*= 1000/T2*). T2 and R2 images can be collected, using
radio waves rather than magnetic gradients. The darkening process seen on MRI can be
considered a “half-life,” such that the greater the tissue iron, the shorter the signal half-
life, and the smaller the T2 and T2* values (Wood, 2011).

19

Through pioneering work done by Dr. John Wood at CHLA/USC, MRI has been
established as the new “gold standard” for safe, non-invasive measurement of iron
deposition in several organ systems, including the heart (Vermylen, 2008; Wood, 2007).
According to Wood, MRI assessment of both LIC and CIC has achieved the critical
levels of availability, utility, and validity to serve as primary endpoints for clinical trials
(Wood, 2007). Cardiac MRI can also be used to assess heart structure and function,
and it may be superior to conventional echocardiography for that purpose (Armstrong et
al., 2012; Plein et al., 2004; Plein, Ridgway, Jones, Bloomer, & Sivananthan, 2002) .
Clinically, cardiac MRI has been applied almost exclusively to patients with primary
hematological disorders requiring frequent PRBC transfusions, where it has been found
to be predictive of developing heart failure among heavily transfused thalassemia
patients who demonstrated evidence of myocardial iron loading on MRI (Kirk et al.,
2009; Wood, 2009). Pancreas R2* can be obtained when the liver and heart are
scanned, offering complementary information. At CHLA, pancreatic MRI iron
measurements are being used along with liver and cardiac MRI for clinical
monitoring/modification of therapy in patients being treated with chelating agents (Wood,
2011). MRI measurement of pituitary iron remains investigational with further clinical
validation and technical standardization necessary. CHLA investigators have reported
preliminary results for studies to define age-specific norms for pituitary iron and volume
imaging (Noetzli et al., 2012; Wood et al., 2010). In the future, it is envisioned that
pituitary iron measurement by MRI will be used to detect/prevent pan-endocrine
dysfunction associated with iron overload.
For the purposes of the dissertation study, given the study aims as well as
limitations of time and funding, the heart was selected as the primary organ to be
assessed by MRI. Studying the heart in this fashion offered an additional advantage in
that a single study could assess both tissue iron (myocardial iron deposition) and organ
20

function (LVEF). In all other body systems, a combined approach of measuring the iron
with MRI and then separately assessing organ function by various other means would be
required. At the time of cardiac MRI assessment, liver and pancreas iron measurements
were acquired because they could be easily obtained at the same time as the cardiac
MRI for no additional cost. The study did not include MRI measurement of pituitary iron.
1.4 Assessment of health-related quality of life
1.4.1 Background
Over the past 20 years, there have been notable advances in defining,
conceptualizing and measuring HRQOL in children with cancer and childhood cancer
survivors, and in the past decade there has been a significant increase in the
development and utilization of pediatric HRQOL measures (Hinds, Burghen, Haase, &
Phillips, 2006; Varni, Burwinkle, & Lane, 2005). The growing attention to HRQOL
parallels the striking increases in childhood cancer survival. Improvements in survival
rates have been achieved through intensive multimodal therapy with combinations of
surgery, chemotherapy, radiation therapy, and HSCT, which – along with hospital stays,
intrusive procedures, late effects, and other aspects of the cancer experience – pose
risks to survivors’ HRQOL. Increased attention to HRQOL also has been driven by
efforts to harmonize health outcomes review criteria across regulatory agencies through
the use of patient-reported outcomes (PRO) (Lipscomb, Gotay, & Snyder, 2007).
Knowing what comprises HRQOL, how it can best be measured, and how to translate
those findings into improved clinical care has important ramifications for health care and
clinical research, as well as health services research and policy. Advantages of
accurate, sensitive and well-timed measures of HRQOL include providing alternative
endpoints when treatment arms in a clinical trial are being compared that may have
equivalent outcomes but differing demands on daily life, being able to more fully assess
immediate and delayed effects of therapeutic interventions, and obtaining a longitudinal
21

view of points in the cancer treatment/survivorship trajectory that are most challenging
and which may be targets for supportive interventions to minimize demands on the
patient and family system (Hinds et al., 2006).
1.4.2 HRQOL measurement
Measuring HRQOL in children and adolescents with cancer or in post-treatment
survivorship requires definitional clarity, which has been a longstanding challenge.
Advances in defining HRQOL occurred in the early 1990’s with the work of an
international group of investigators under the auspices of the World Health Organization,
which posited a multidimensional concept incorporating the individual’s perception of six
broad domains of quality of life: physical health, psychologic state, levels of
independence, social relationships, environmental features, and spiritual concerns
(World Health Organization, 1993). In the same era, participants in a U.S. National
Institutes of Health workshop on quality of life assessment adopted a working definition
of HRQOL as: “…the value assigned to duration of life as modified by the impairments,
functional states, perceptions and social opportunities influenced by disease, injury,
treatment or policy” (Patrick, 1993). Both definitions remain relevant, one emphasizing
the subjective nature and contextual importance of HRQOL, and the other reflecting the
need for scientific specificity and objectivity (Padilla, 2004). In the early 2000’s, the U.S.
National Cancer Institute’s Outcomes Measurement Working Group, after evaluating
hundreds of HRQOL measures used in oncology, endorsed the view that HRQOL
measures are distinguished by being patient-reported and involving the patient’s
subjective assessment or evaluation of important aspects of his/her well-being. More
recently, studies have been conducted to help define HRQOL from the perspective of
pediatric cancer patients (Hinds et al., 2004; Woodgate, 2005; Woodgate & Degner,
2003, 2004). Hinds has advanced a pediatric oncology-oriented definition of HRQOL,
based on interviews with children and adolescents receiving treatment for cancer, as “an
22

overall sense of well-being based on being able to participate in usual activities, to
interact with others and feel cared about; to cope with uncomfortable physical, emotional
and cognitive reactions; and to find meaning in the illness experience”(Hinds et al.,
2004).
Definitional differences translate into differences in measures of HRQOL,
resulting in measures of different aspects of HRQOL. A review of 11 of the more
commonly used pediatric HRQOL instruments found that the number of conceptual
domains and the number of items varies widely, with the former ranging from 4-14 and
the latter ranging from 1-87 (Hinds, 2010). In pediatric oncology, a theory-driven
approach is emerging that may modify definitions and decisions about what data to
collect and how to interpret it (Hinds et al., 2006). Examples of conceptual frameworks
under study include the Resilience Model of Illness (formerly called the Adolescent
Resilience Model), the Self-Sustaining Model for Adolescents, Post-Traumatic
Stress/Post-Traumatic Growth models, the Interaction Model of Client Health Behavior,
the Pediatric QOL at End-of-Life Model, and Relational Decision-Making at End of Life in
Pediatric Oncology (Cox, 2003; Haase, Heiney, Ruccione, & Stutzer, 1999; Hinds &
Martin, 1988; Hinds, Oakes, Hicks, & Anghelescu, 2005; Kazak et al., 2004; Nuss,
Hinds, & LaFond, 2005; Santacroce, 2003).
1.4.3 The Pediatric Quality of Life Inventory
TM
(PedsQL
TM
) Measurement Model
One of the more well-established pediatric HRQOL measures is the Pediatric
Quality of Life Inventory
TM
(PedsQL
TM
) Measurement Model, which is based on a
conceptualization of HRQOL as the patient’s perceptions of the impact of disease and
treatment functioning in a variety of dimensions, including physical and psychosocial
domains, with the latter including emotional, social, and role functioning as well as
related constructs (Varni, Burwinkle, Katz, Meeske, & Dickinson, 2002). The PedsQL
TM

Measurement Model is a modular approach to measuring pediatric HRQOL, developed
23

to (1) integrate the relative advantages of generic approaches (i.e., allowing
comparisons across pediatric populations and benchmarking with healthy populations)
and disease-specific approaches (i.e., enhancing measurement sensitivity for health
domains relevant to that condition) to measuring pediatric HRQOL, and (2) be used for
the broadest age group empirically feasible, specifically including child self-report for the
youngest children possible (Varni, Burwinkle, & Seid, 2005; Varni, Seid, & Rode, 1999).
In pediatric HRQOL measurement, reliable and valid child self-report across a broad age
range is desirable even with the availability of parent-proxy report.
The PedsQL
TM
4.0 Measurement Model includes a Cancer Module (Varni et al.,
2002) and Generic Core Scales. The Cancer Module is comprised of eight scales
measuring pain, nausea, procedural anxiety, treatment anxiety, worry, cognitive
problems, perceived physical appearance, and communication. For the dissertation
study, the Generic Core Scales, which were designed for application in both healthy and
patient populations, was selected because eligible study participants have completed
cancer treatment, making the Cancer Module scales less applicable. The PedsQL
TM
4.0
Generic Core Scales were developed over a 20 year period through an iterative process
that involved qualitative and quantitative methods (i.e., patient and parent focus groups,
individual focus interviews, item generation, cognitive interviewing, pretesting, field
testing following standardized protocols, and administration to a large population of
children and adolescents). The Core Scales include Physical Health (8 items) and a
composite of Emotional (5 items), Social (5 items), and School (5 items) Functioning
scales that together comprise Psychosocial Health. The scales are designed to be
developmentally appropriate with child self-report forms available for children in different
age groups and corresponding parent-proxy report forms.
Whereas the lower age limit for most pediatric HRQOL self-report instruments is
age eight years, the PedsQL
TM
4.0 Generic Core Scales include child self-report for ages
24

five to seven years. Feasibility, reliability and validity of the Core Scales have been
reported in various studies which included pediatric populations across age ranges.
Specifically relevant to the issue of validation of the measure in younger children, an
analysis of child self-report data on 8,591 children across age subgroups demonstrated
construct validity for the PedsQL
TM
Core Scales through the known groups approach, in
which scores of healthy children and children with chronic health conditions were
compared, including 358 children with cancer. Across age groups, including children as
young as five years, there was a statistically significant difference in HRQOL, with
healthy children having better HRQOL than children with known chronic health
conditions; most effect sizes were in the medium to large effect sizes range. This
difference held true for each of the PedsQL
TM
4.0 psychosocial scales (Varni, Limbers, &
Burwinkle, 2007a).
A number of studies have investigated the discriminant/convergent and criterion
validity of the PedsQL
TM
4.0 Generic Core Scales, beginning with the development of the
first generation PedsQL
TM
instrument. The original PedsQL
TM
was derived from data
collected from 291 pediatric cancer patients and their parents at various stages of
treatment; pediatric cancer served as the model using a noncategorical approach to
pediatric conditions that is based on the consequences of the condition rather than a
diagnostic label (Varni et al., 1999). Discriminant, or clinical validity, was demonstrated
for on-treatment patients versus those off treatment. Further construct validity was
shown by multimethod-multitrait matrix using standardized self-report measures of
emotional distress (Children’s Depression Inventory, State-Trait Anxiety Inventory for
Children), perceived competence (Self-Perception Profile for Children and Adolescents),
and social support/functioning (Social Support Scale for Children and Adolescents) and
parent proxy report measures (Child Behavior Checklist) (Varni et al., 1999). Since the
25

publication of that first report, validity of the PedsQL
TM
Generic Core Scales has been
described in more than 75 published reports (Varni, Burwinkle, & Seid, 2005) including
comparisons of healthy children and children with cancer (Varni et al., 2002),
comparisons of healthy children and children with acute/chronic conditions (Varni, Seid,
& Kurtin, 2001), and a comparative analysis of 2500 pediatric patients in 10 disease
clusters and 33 disease categories/severities with 9500 healthy children (Varni, Limbers,
& Burwinkle, 2007b).
1.5 Preliminary work
In preparation for the dissertation study, a cohort of childhood cancer survivors
was established with the approval of the CHLA Institutional Review Board (IRB), offering
the opportunity for longitudinal prospective survivorship research. Preliminary work
related to the establishment of the cohort and a retrospective analysis that is
summarized in the synopsis that follows. This work was published in 2012 (Ruccione et
al., 2012); a copy of the paper is included in the Appendix.
1.5.1 Synopsis
Background. Packed red blood cell (PRBC) transfusion is a mainstay in
childhood cancer treatment, but has potential for inducing iron overload. The purpose of
this study was to determine whether treatment intensity is predictive of projected iron
burden resulting from PRBC transfusions among survivors of several forms of childhood
cancer.
Procedure. This retrospective cohort study involved patients treated at CHLA
between June 1, 2004 and December 31, 2009. Clinical/demographic data were
abstracted from medical records. Treatment Intensity Level was determined for each
patient using a published scale. Adjusted cumulative PRBC transfusion volume for each
patient (ml/kg) was used to compute the adjusted total iron burden (mg/kg) based upon
the average hematocrit of the product.
26

Results. Median age of the cohort (n=214) was 7.9 years (range 0.2-20.2). 114
(53.3%) subjects were male and 129 (60.3%) were Hispanic/Latino. Diagnoses included
acute leukemia and six solid tumors, management of which represents a range of cancer
treatment intensities. The number of transfusions, transfusion volumes and projected
iron burden were significantly increased and exceeded upper limits of normal among
patients with higher treatment intensity. Multivariate analysis found younger age and
lower hemoglobin at diagnosis to be associated with greater iron burden after adjusting
for treatment intensity.
Conclusion. This study was the first to estimate transfusional iron burden
systematically among survivors of a variety of childhood cancers treated over a wide
range of chemotherapeutic intensity. Analysis of cohort data indicated that subsets of
patients who received moderate to large cPRBCv had projected total body and liver iron
burdens that exceed normal values, and presumably place them at elevated risk for all of
the complications known to be associated with iron overload. Stated another way,
greater treatment intensity was associated with the need for more PRBC transfusions,
and thus increased risk of iron overload among childhood cancer survivors. The results
of the study clearly pointed to the need for additional research, as well as a heightened
clinical awareness of the potential risk patients may incur as a result of PRBC
transfusions.
1.6 Design and methods of the dissertation study
1.6.1 Overall design
The dissertation study utilized a cross-sectional design to evaluate a purposeful
sample of 100 survivors selected from the established CHLA cancer survivor cohort,
described above, at a single call-back visit to CHLA where they underwent: (1) a focused
cardiac MRI study also encompassing the liver and pancreas; (2) a blood draw for a
panel of iron and cardiac biomarkers, and hemochromatosis gene mutation analysis;
27

and (3) assessment of PROs related to physical health, emotional functioning, social
functioning, school functioning, and symptoms of general fatigue, sleep/rest fatigue, and
cognitive fatigue. The study was supported by grants from several organizations;a list of
funding sources is provided in the Appendix.
1.6.2 Sample and Setting
Under the direction of Dr. Richard Sposto (Professor of Research, Preventive
Medicine and Pediatrics, CHLA/USC), a sample of potential subjects was selected
randomly in a way designed to maximize the predicted power of the multivariate
regression of anthracycline dose and iron deposition on LV function. In brief, a large
number of random samples of size 100 from the CHLA survivor cohort of 214 patients
was generated. These were ranked according to the corresponding predicted power of
the multivariate regression analysis, and one of the top 10 ranking samples was selected
randomly as the sample for this study by Dr. Sposto. Using this method, an initial list of
potential subjects was selected, which was subsequently supplemented by three
replacement lists generated in an effort to meet accrual goals. Figures 1.7 and 1.8
depict the total sample (n=214) from which potential subjects were selected and the
selected sample (n=167), respectively.
Potential subjects were offered enrollment according to the following criteria.
Inclusion: (1) diagnosed with cancer before age 21 between June 1, 2004 and
December 31, 2009; (2) treatment and transfusions given at CHLA; (3) no radiation to
the heart or surrounding tissues; (4) able to read and write in English or Spanish in order
to complete patient/parent-reported outcome measures; (5) able to tolerate an MRI
procedure. Exclusion: (1) subjects less than seven years of age to avoid the need for
prolonged pediatric sedation and potential interference with image acquisition; (2)
females who were pregnant.

28

29

1.6.3 Instruments/Measures
Magnetic resonance imaging. MRI scans for quantitated organ iron content
(heart, liver, pancreas) and cardiac function/pathologic remodeling changes were
performed under the guidance of Dr. John C. Wood (Associate Professor, Pediatric
Cardiology and Radiology, CHLA/USC). All examinations were performed on a 1.5
Tesla magnet with 80 mT/m gradients and a four element torso array coil.
Liver/pancreas R2*and heart T2* were assessed using multi-echo gradient echo
technique with echo times from 1 to 16 milliseconds. Liver R2, a second metric of liver
iron, was assessed using single spin echo imaging with echo times from 3 to 30
milliseconds. Custom Matlab routines (The MathWorks, Natick, MA) were used to
calculate R2 and R2* coefficients and convert them into iron values (Wood, Enriquez,
Ghugre, Tyzka, et al., 2005). Left ventricular mass and function were assessed using 15
short axis steady state free precession images spanning from cardiac base to apex,
post-processed using MASS 3.0 software (Medis, The Netherlands). In addition to the
study radiologist’s interpretation of MRIs relative to study aims, a CHLA radiologist
reviewed all MRI images for evidence of other malignant or non-malignant disease
processes, to be communicated to the participant’s attending MD if noted.
Laboratory tests. A limited number of biomarkers related to iron and cardiac
status was selected based on their potential association with outcomes of interest. For
iron status, ferritin and iron, iron-binding capacity, and percent transferrin saturation
levels were drawn by venipuncture on the same day as the MRI evaluation was
performed; each was measured in 1 mL of serum. Ferritin, the major soluble iron
storage protein, was included because serum ferritin levels are proportional to – and
thus reflective of – body iron stores. Levels of serum iron, iron-binding capacity (the
maximum amount of iron that the blood can carry), and percent transferrin saturation
were measured because they are indicators of the amount of iron bound to transferrin,
30

with high percentages of transferrin saturation reflecting iron storage disorders. In
addition, patients were tested for hereditary hemochromatosis, a condition involving the
HFE gene and occurring most often in individuals of northern European origin, which is
characterized by a predisposition to increased, inappropriate absorption of dietary iron
that can lead to complications of iron overload (Bacon, Adams, Kowdley, Powell, &
Tavill, 2011). Hemochromatosis represented a potential confounder in evaluating
transfusional iron overload in this study, and patients were tested for two related
mutations of the HFE gene: C282Y and H63D. Individuals who are homozygous for HFE
C282Y make up 80-85% of patients with hereditary hemochromatosis; the HFE H63D
mutation can be associated with iron loading in compound C282Y/H63D
heterozygotes.(Bacon et al., 2011).
For cardiac status, markers of myocyte stress (B-type natriuretic peptide),
myocyte injury (high sensitivity cardiac specific troponin I) and inflammation (high
sensitivity C-reactive protein) were measured (Pincott, 2011). Cardiac specific troponin I
and B-type natriuretic peptide (BNP) have been used to evaluate cancer treatment-
related cardiotoxicity (Lipshultz et al., 2012). Increases in troponin I and BNP are
indications of damage/death in myocardial cells and increased LV wall stress,
respectively. Another marker, C-reactive protein (CRP), is a more general indicator of
inflammation and may indicate cardiac stress associated with inflammation. In this
study, BNP levels were measured in 1 mL frozen plasma samples collected in a plastic
EDTA tube; levels of high sensitivity cardiac specific troponin I and high sensitivity CRP
were each measured in 1 mL of serum.
All laboratory tests were performed by the CHLA CLIA
2
-certified outpatient
clinical laboratory or by Quest Diagnostics, San Juan Capistrano, CA. Point-of-care

2
Clinical Laboratory Improvement Amendments; a program to ensure quality laboratory testing;
clinical laboratories must be properly certified to receive Medicare or Medicaid payments.
31

urine pregnancy tests were performed on the day of the MRI evaluation to avoid
exposure of pregnant females to MRI.
Patient-reported outcomes. Three survey measures were selected for
administration by a trained bilingual (English/Spanish) Study Coordinator or the Study
Principal Investigator (Principa Investigator Ruccione) on the day of the MRI evaluation
in order to capture PROs at the same time that cardiac/liver/pancreatic iron was
measured and laboratory tests were performed. Descriptive data primarily related to
symptoms of early cardiotoxicity were sought to see whether such data would/would not
correlate with data from the MRI and blood tests. It was understood that one-time
assessment would not capture change over time, but that it could serve as a baseline for
future studies in the same established cohort of childhood cancer survivors.
Targeted cardiac health history. A brief 6-point Likert-scale self-report form
(Cardiac Health Form) was used, which was adapted from one used in Stanford’s
Pediatric Cardiopulmonary Exercise Laboratory and also used by Cox et al. in a study of
subclinical cardiac toxicity in childhood cancer survivors (Cox, Rai, Rosenthal, Phipps, &
Hudson, 2008). The measure asks survivors or their parents to rate fatigue, palpitations,
and chest pain in the context of running, walking, climbing stairs, and school/work.
Health-related quality of life. The study used PedsQL™4.0 Generic Core Scales
Self-Report (23 items) to measure dimensions of HRQOL (Varni et al., 2002).
Developmentally appropriate patient self-report and parent proxy-report versions for
ages 5-7, 8-12, 13-18, and 18-25 years in English or Spanish were administered. For
each item, respondents indicated the extent to which it had been a problem (0=never to
4=almost always) during the preceding month. Items were reverse scored and linearly
transformed to a 0-100 scale with higher scores indicating better quality of life. The
PedsQL™ 4.0 Generic Core Scales have been tested with pediatric oncology patients
and demonstrate strong internal consistency and test-retest reliability; they also
32

accurately differentiate between children with cancer and healthy controls (Varni et al.,
2002).
Fatigue symptoms. The PedsQL™ Multidimensional Fatigue Scale, General
Fatigue subscale (six items) was used. It has developmentally appropriate patient self-
report and parent-proxy report versions for ages 5-7, 8-12, 13-18, and 18-25 years in
English and Spanish. The PedsQL™ Fatigue Scale has the same format, instructions,
Likert response scale and scoring method as the PedsQL™ 4.0 Generic Core Scales
(Varni et al., 2002). Items are reverse scored and linearly transformed to a 0-100 scale
with higher scores indicating fewer symptoms of fatigue. The scale has been tested with
pediatric oncology patients and demonstrated strong internal consistency and test-retest
reliability; it accurately differentiates between children with cancer on treatment versus
those off treatment, and between children with cancer on treatment and healthy controls
(Varni et al., 2002).
1.6.4 Study materials acquisition
With grant funding secured for the study, the PedsQL 4.0 Generic Core Scales
and Multidimensional Fatigue Scales were procured in English and Spanish for all
eligible age groups. The cardiac symptom checklist was formatted/printed using the
previously described cardiac function classification and self-report Likert scale. These
materials were organized and stored for retrieval in paper files and in an electronic
shared drive accessible in password-protected format to the study team.
1.6.5 Scientific and regulatory review/approval and document translation
The study clinical protocol and consent/assent forms were prepared and
submitted for review by the Clinical Investigations Committee of the University of
Southern California Norris Comprehensive Cancer Center (USC CIC) on September 7,
2011. The reviewers assigned the study a priority score of 1.2 (outstanding) and it was
approved pending response to several questions. After these questions were addressed,
33

the USC CIC officially approved the study on October 10, 2011. In a parallel process, the
study protocol, English consent/assent forms, patient recruitment letter, survey forms,
and telephone script for recruitment follow-up calls were submitted to the CHLA
Institutional Review Board (IRB) (called the Committee on Clinical Investigations, CCI, at
CHLA) on August 25, 2011 for expedited review as a minimal risk study. The CHLA CCI
process included review by representatives of involved Divisions and the Department of
Pediatrics, followed by review/requested clarification by IRB staff and response by the
study Principal Investigator (PI). The study was approved October 7, 2011 with
contingencies, and the response to contingencies was submitted the same day. The
study was approved for patient entry on October 11, 2011. Once approved by the CCI,
the English study consent/assent documents and recruitment letters were sent for
translation into Spanish to an agency approved by CHLA. The translated documents
were completed November 28, 2011 and approved by the CCI on November 30, 2011.
1.6.6 Study team orientation and training
Each member of the study team completed required training modules on human
subject protection and good clinical practice, which were documented in the on-line IRB
study file and regulatory binder. Team members participated in study start-up training
given by the CHLA IRB Quality Assurance Specialist, which included conducting the
informed consent conference, recording and reporting adverse events, clinical trial
record-keeping, and the IRB quality assurance program auditing/monitoring program.
The Study PI (Ruccione) or Study Co-Investigator (Freyer) conducted all initial consent
conferences. Team members observed several consent conferences, and then were
supervised/approved by the PI in order to be able to conduct consent conferences
independently with IRB approval. Bilingual (Spanish-English) team members were
recruited to avoid the need for obtaining consent/assent through interpreters. These
34

individuals took and passed the CHLA oral bilingual assessment and completed the 24-
hour mandatory instructor-led training for bilingual staff.
1.6.7 Coordination with relevant CHLA services
Meetings, telephone conferences, and email communications were conducted
regularly with the CHLA Clinical and Translational Science Institute, Children’s Center
for Cancer and Blood Diseases (CCCBD) Clinical Trials Office, Department of
Radiology, and CCCBD finance office. Collaboration with these entities facilitated study
start-up, preparation of standard operating procedures for study logistics (such as entry
of MRI and lab orders into the electronic medical record, patient scheduling, specimen
handling and processing, and correct charge entry/reconciliation), and troubleshooting of
problems in real time. Weekly study team meetings were also conducted to assure that
the study proceeded according to plan.
1.6.8 Participant recruitment and scheduling
For all potential study participants, confirmation was solicited from the treatment
team (attending MD, nurse care manager, social worker) that there were no clinical,
psychosocial, or developmental issues that would make study participation inappropriate
for an individual patient. Letters were sent to all vetted eligible patients/families, written
in plain language English on one side and Spanish on the other, inviting them to
participate in the study. Letters were addressed to survivors if they were age 18 years or
older since they could legally consent for themselves. Follow-up telephone calls were
made in English or Spanish, as appropriate, to schedule study appointments. These
telephone conversations provided an opportunity to assess understanding of the study,
as well as any developmental or logistical considerations that would affect study
participation. For example, congruent with the theory of emerging adulthood, the study
team found that many potential participants ≥18 years of age were residing with their
parents and were dependent upon the family for transportation. After experiencing a
35

missed appointment for one study participant, study procedures were amended so that,
when indicated by assessment and with an 18+ year old’s agreement, a member of the
study team would speak with a parent to ensure adherence to the scheduled
appointment day/time.
A study coordinator scheduled the patient’s MRI examination and laboratory tests
in coordination with the Department of Radiology and the CHLA CTSI, and made at least
one confirmation telephone call to the participant before the study visit. Confirmation
calls were found to be essential to prevent “no shows” on appointment days. On the
scheduled day, a Study Coordinator and the PI met with the patient/parent in the Clinical
Trials Unit to obtain assent/consent, administer the study measures, obtain blood
samples and perform a urine pregnancy test if indicated, escort the participant to/from
the MRI (and place ECG leads on female patients), and de-brief with the patient/family
afterward. The duration of each participant appointment was approximately three hours.
MRI scans were performed in the CHLA Department of Radiology using the established
MRI iron quantitation and cardiac protocol; on average the MRI procedure took 45 - 60
minutes. All MRI and laboratory findings were provided to the patient’s attending
physician who was responsible for return of results to study participants and clinical
management as indicated.
Partial compensation for study participation was offered to participants and their
parents via a $25 gift card. When gasoline prices rose and potential participants began
to state that they could not participate due to higher transportation costs, the PI
sought/received approval from grant funding sources and the IRB to offer a $20 gas card
as well. These gift card amounts were judged to be appropriate for inconveniences
without being coercive. The Study Coordinator/PI presented the gift cards to participants
upon completion of all required study measures. The study schema is shown in Figure
1.9.
36

Figure 1.9. Dissertation Study Schema

1.6.9 Protection of human subjects used for research
Potential risks associated with the study. Although the potential risks to
subjects in this study were considered no more than minimal risks according to the
Common Rule, the study team sought to prevent these risks. Table 1.2 lists the study’s
potential risks and the preventive interventions employed by the study team.
Table 1.2. Potential Risks to Study Subjects
Potential Risk Preventive Intervention
Reactions to having blood drawn (e.g., pain,
swelling, bruising, and occasional fainting reactions;
rarely, infection or small blood clots in blood
vessels)
• Blood was drawn only by a qualified and trained
person, well experienced in pediatric
venipuncture.
• Parents accompanied their child, depending on
age and child preference.
Discomfort from having to lie still during the MRI
scan and periodically holding their breath or feeling
cold
• Anticipatory guidance was provided regarding
the MRI procedure, with emphasis on sensory
experiences. Parents were allowed in the MRI
scanning room after clearance by the study
team/Radiology staff.
• Patients were informed that they could bring
DVD’s to watch or they could choose from
among those available to provide distraction
during the MRI scan.
• Warm blankets were placed under and on top of
patients if needed in the MRI scanning room.
A breach of MRI safety precautions • Standard safety precautions for MRI were
observed to prevent injury due to implants,
37

Table 1.2. Potential Risks to Study Subjects
Potential Risk Preventive Intervention
magnetic projectiles, and acoustic trauma.
• Patients (or their parents) were asked to provide
information about all implants before entering
the room for the MRI.
• Utilizing the ASTM International classification
system for implants and ancillary clinical
devices, we excluded patients with most forms
of biostimulation implants (e.g., pacemakers,
nerve stimulators, implantable cardioverter-
defibrillators, insulin pumps, cochlear implants).
• To reduce the risks of projectile accidents,
ferromagnetic objects and devices were
prohibited in proximity to the MRI scanner and
patients undergoing MRI examinations were
required to remove all metallic objects.
• Appropriate ear protection was provided for
anyone inside the MRI scanner room during the
examination.
Being inadequately prepared for the demands of the
study and thus being surprised by some aspect
• The study demands were explained by letter
and telephone prior to the study evaluation day.
• The consent conference preceding MRI, lab
tests, and survey completion provided
opportunity to re-explain and clarify all aspects
of the study.
A breach of confidentiality • Potential privacy risks and measures to
minimize them were explained during the
consent conference, and the HIPAA form was
explained prior to patient/parent signing it.

Informed consent, permission, and assent. Several consensus documents
disseminated by pediatric and pediatric oncology organizations regarding informed
consent and decision-making involving children formed the framework for the consent
process in the study (Braunwald, 2008; Horacek et al., 2008; Zethelius et al., 2008).
Two key concepts articulated in these statements are that: (1) only individuals with
appropriate decisional capacity and legal empowerment can give consent, meaning that
parents (or other surrogates) provide informed permission with the assent of the child
when appropriate; and (2) assent includes helping the patient achieve a developmentally
appropriate understanding, telling the patient what to expect, assessing the patient’s
understanding and the factors influencing how he/she is responding, and soliciting an
expression of the patient’s willingness to participate.
38

The consent process was initiated by asking the participants/parents for their
permission to share information with them about the study, after determining whether
communication would be in English or Spanish, or both (i.e., it was not unusual for the
parent to prefer communicating in Spanish, while the child/teen participant preferred
English). While explaining the study, the study team offered a clearly worded verbal
statement that participation was their choice, adding a particular emphasis on what was
required of participants. The team routinely stated at least two additional times during
this discussion that participation was voluntary and that declining was their right and
would be respected by the study team. The Experimental Subject’s Bill of Rights was
used as the outline for discussion. After the study was explained and
participants/parents had an opportunity to read the consent and assent documents,
permission was sought from the parents of eligible survivors who were age <18 years.
Conforming to the requirements of the CHLA IRB, assent was requested from children
age 7-13 years. Older children, ages 14-18 years, signed the parent’s consent
document; in doing so they affirmed that they had read and understood the consent
form, had the chance to ask questions and have them answered, and they agreed to
participate. Participants/parents were given copies of their signed consent/assent forms,
and the PI’s business card with contact information.
Confidentiality of study data. Study data forms (consent documents, the
Cardiac Health Form, PedsQL™4.0 Generic Core Scales, and PedsQL™
Multidimensional Fatigue Scales, MRI reports, and laboratory reports) were stored in a
locked file cabinet accessible to study personnel only. Study participants were identified
using a specific subject code. The code linking the participant’s name was on a
screening log that was kept confidential in a locked office. Any data collection on data
forms was coded using the participant’s specific study code. Research data were stored
in a secured database in the Clinical Trials Office under lock and key.
39

1.6.10 Study enrollment
The first study participant was enrolled in December 2011. Study enrollment was
closed in December 2012. Using the potential subject selection method described in
Section 1.6.2, the study sample drawn from the original cohort of 214 under the direction
of Dr. Richard Sposto. Three additional replacement samples were drawn by Dr. Sposto
as non-participant patients were identified. The original sample plus the 3 replacement
samples yielded the 167 potential participants. The ideal projected enrollment was 100.
After the exclusion of 92 potential participants for the reasons shown in Figure 1.10, 75
individuals were enrolled.

Of the 92 potential participants who were excluded, there were 13 patients who
had become deceased between the time when the CHLA cancer survivorship cohort
was established and the opening of this study. Of the 48 patients who declined to
participate: 15 were “passive” refusals (e.g., no response to three attempted contacts to
a known working phone number or two or more “no shows” for scheduled appointments),
Study Recruitment Flow Diagram
CHLA Cancer Survivorship Cohort
(n=214)
Sample randomly selected by
study statistician
(n=167)
Enrolled
(n=75)
Excluded (n=92)
•Deceased (n=13)
•Declined (n=48)
•Ineligible (n=12)
•LTFU (n=19)
Figure 1.10.
Study Recruitment

40

9 said they were not interested, 16 did not state a reason for declining, and 8 gave the
following reasons: unable to tolerate MRI procedure (n=1); parent did not want child
“poked/prodded” any more (n=1); patient/parent did not want to come to the hospital if
not medically necessary (n=1); no transportation (n=1); dealing with relapse (n=2);
patient has Crohn’s disease with “too many appointments” (n=1); and “too many health
issues and lives too far” (n=1). Thus, where a reason other than not being interested was
explicitly stated, there were 4 with health issues (Crohn’s disease, relapse, and
unspecified health issues). The 12 patients who were declared ineligible were found to
have received chest irradiation when the study team reviewed their medical record prior
to contact. Chest irradiation was an exclusion criterion for study enrollment because of
the potential effect of irradiation on the heart. The category of Lost to Follow Up (LTFU)
is comprised of individuals for whom there was no current contact information (e.g.,
moved out of the state/country and/or for whom no phone numbers on record were
active). The study was not funded to contract with a tracing service to locate these
patients.
1.6.11 Data management and statistical analysis plan
Study data were managed using RedCap (Research Electronic Data Capture), a
secure online application that allows for building and managing databases. The
database in this study contains patients’ demographic, laboratory, MRI report, and
questionnaire data, all of which were entered manually into the online database from the
original medical and research study records. Data management and quality assurance
procedures, including an audit in which 10% of RedCap study data files were verified
against original data sources, were the responsibility of the designated study coordinator
and Clinical Trials Office personnel under the close supervision of the Principal
Investigator and Co-Investigators.
41

Data analyses were performed under the direction of Dr. Richard Sposto. To
demonstrate that the patients enrolled were representative of the CHLA childhood
cancer survivor cohort, the distribution of patients enrolled in the study was compared to
the CHLA childhood cancer survivor cohort using the χ
2
test for each of the clinical and
demographic characteristics listed on Table 1 of Chapters 2, 3, and 4. Spearman rank
correlation was utilized to assess the relationship between liver iron concentration with
the other organ-specific iron measures and iron parameters. Linear regression analyses
were conducted to examine the effects of patient demographics and clinical traits on the
following three endpoints: liver iron concentration, left ventricular ejection fraction, and
patient-reported PedsQL
TM
psychosocial HRQOL score. Additionally, one-way
ANOVA was used to investigate the effects of sex or age group on the various
MRI parameters. Univariate and multivariate analyses were performed for each of the
three endpoints noted above. The unique multivariate models were built starting with
predictors with an F-test p-value less than 0.15 on the univariate analysis and following
a step-wise iterative process until the multivariate likelihood ratio test (LRT) p-value fell
below 0.05 for each of the remaining predictors. This approach was validated by
assessing changes in significance through reintroduction of eliminated predictors. Intra-
class correlation and Spearman rank correlation were utilized to examine patient-parent
dyad agreement and relationship, respectively, in the PedsQL
TM
HRQOL and fatigue
measures. All analyses were performed as two-sided tests with a significance level of
0.05. Statistical computations were performed using the statistical software package
Stata (version 11).
1.6.12 Adverse events/unanticipated problems and their effect on the project
There were no serious adverse events. One child (female, age nine years) who
attempted the MRI was unable to complete it during her initial study visit, which included
42

the consent conference, blood draw, and completion of surveys. She returned on a
subsequent morning after a good night’s rest and completed the MRI without difficulty.
Another child (male, age 11 years) became tearful during the consent conference and
expressed reluctance to proceed. He was given the opportunity to think about
participating and talk it over with his parents at home. He opted to return on another day,
at which time he was able to complete the cardiac MRI, although he was unable to lie
still long enough for liver and pancreas MRI image acquisition. One child (male, age 8
years) who was developmentally delayed tried to complete the MRI procedure, but was
unable to lie still for the duration of the MRI procedure and therefore no images could be
obtained.
The most challenging issue for the successful completion of the study was limited
access to the MRI for study evaluations, which affected the rate of participant accrual.
This problem was unanticipated and beyond the study team’s control. High demand for
clinical needs meant that the CHLA Department of Radiology was able to accommodate
only a very limited number of research study subjects who had to be scheduled in the
evening or on Saturdays when there were no patients scheduled for clinical MRIs. In
April 2012 the study PI was informed that one of the two MRI machines would be taken
out of service to be upgraded, further reducing MRI access for study participants
because this meant that study patients could only be scheduled on Sundays. This
barrier was partially overcome as the study team pressed to get as many study patients
as possible to come in on Sundays, and to advocate for recovery of additional
appointment slots once the new MRI was fully installed/approved for use. Project grant
funds were allocated to pay study coordinators and radiology technicians to work extra
hours on Sundays and this PI-staff team worked most Sundays to continue study accrual
despite the limited MRI availability. In July 2012, MRI availability was further reduced to
every other Sunday.
43

1.7 Significance
Although there is compelling evidence to suggest that iron overload, particularly
in the presence of toxic chemotherapeutic exposure, may adversely affect organ function
in childhood cancer survivors, there have been no systematic studies of a large cohort
with a variety of malignant diagnoses in whom iron burden is quantitated using MRI
technology. With cardiac complications identified in the Childhood Cancer Survivor
Study as the leading cause of death and disability among the approximately two-thirds of
survivors reporting chronic health conditions (Oeffinger et al., 2006), there has been a
critical need for research to establish the prevalence of cardiac iron overload and identify
contributing risk factors to guide clinical interventions. Currently pediatric cancer
treatment centers do not routinely track transfusion volumes for individual patients as
they do for anthracycline cumulative doses, nor are transfusion decisions typically made
in consideration of anthracycline exposure. Risk related to iron overload is not
addressed in the Children’s Oncology Group (COG) Long-term Follow-up (LTFU)
Guidelines v3.0, the principal resource for childhood cancer survivorship care developed
by the COG Nursing Discipline and Late Effects Committees (Children's Oncology
Group, 2008). In addition to determining the prevalence of cardiac iron overload, there
has been a need to link pertinent biological and patient-reported outcomes. PROs are
being increasingly recognized, and required, as a key aspect of clinical research. Only
by seeking the childhood cancer survivor’s voice in terms of symptom and HRQOL
experiences, measured by validated instruments, can threats to their functional status
and HRQOL be adequately addressed.
The dissertation study investigated these areas of concern and provided insights
about the prevalence, distribution, and severity of iron overload in the heart, liver, and
pancreas; characterized the prevalence, distribution, and severity of cardiac dysfunction
by MRI; and assessed childhood cancer survivors’ HRQOL using patient-reported
44

outcome measures. Study findings can be expected to contribute to late effects
management in childhood cancer survivors, an area where pediatric oncology specialists
play a prominent role in research and practice. Ultimately it is hoped that the study’s
findings will contribute to the reduction of morbidity and mortality among childhood
cancer survivors.
1.8 Primary Goals
The three research studies in this dissertation study examined the prevalence of
transfusional iron deposition, cardiac dysfunction, and associated health-related quality
of life among childhood cancer survivors. The following primary goals were fulfilled:
• To determine the prevalence, distribution and severity of transfusional iron deposition
in heart, liver and pancreas among a cohort of childhood cancer survivors.
• To characterize the prevalence, distribution and severity of cardiac dysfunction and
pathological remodeling among a cohort of childhood cancer survivors.
• To assess HRQOL using validated patient-reported outcome measures in a cohort of
childhood cancer survivors.
45

CHAPTER 2: PREVALENCE AND DISTRIBUTION OF TRANSFUSION-DERIVED
IRON DEPOSITION IN A SAMPLE OF CHILDHOOD CANCER SURVIVORS

2.1 Abstract
2.1.1. Background
Childhood cancer survivors receiving multiple PRBC transfusions may be at
increased risk for iron deposition in vital organs, thereby increasing risk of serious late
effects.
2.1.2 Purpose
To characterize the prevalence, distribution and severity of iron deposition in
heart, liver and pancreas among a cohort of childhood cancer survivors.
2.1.3. Methods
Design: single institution cross-sectional study of childhood cancer survivor
cohort. Organ tissue iron was quantified by magnetic resonance imaging (MRI). Iron
status by serum markers and hemachromatosis gene mutation status were assessed. In
addition to data acquired from laboratory and MRI evaluations obtained for this study,
disease/treatment and demographic information was abstracted from medical records.
Data analysis included descriptive approaches for data summarization and regression
modeling for estimation and testing.
2.1.4. Results
Seventy-five patients were enrolled (median age 14 years, range 8-25.6 years at
evaluation); slightly more than half were male (40/75) and most were Hispanic/Latino
(45/75). Median follow up time was 4.2 years. Cancer diagnoses included ALL/AML
(n=33) and solid tumors (n=42). Sixty-seven patients had received PRBC transfusions
(median=11, range=1-47), with median cumulative PRBC transfusion volume (cPRBCv)
of 2727 mL (range, 300-14224 mL), and median cPRBCv adjusted for body weight of
90.6 mL/kg (range, 3.6-391.9). Based on liver R2* (n=73) and cardiac T2* (n=74), tissue
46

iron concentration (mean, median, range in mg/g) was calculated for liver (2.6, 1.2, 0.9-
23.7) and heart (0.7, 0.6, 0.5-0.9). No iron calibration exists for pancreas; pancreas R2*
results (mean, median, range in Hz) were 30.2, 26, 18-128 (n=72). Liver and pancreas
R2* were abnormal in 36 (49.3%) and 34 (47.2%) patients, respectively. Liver iron
concentration was elevated in 36 (49.3%) patients. No abnormal cardiac iron
concentrations were observed. Statistically significant correlations between liver and
pancreatic R2* were found. Both hepatic and pancreatic R2* were significantly positively
correlated with serum iron, iron binding capacity, serum ferritin, and cRBCv. HFE C282Y
and H63D mutation status was normal in 72/75 and 63/75, respectively. Predictors of
liver iron concentration found to be significant in univariate analyses included diagnosis
of AML or solid tumor, older age at diagnosis/age at MRI, greater treatment intensity,
higher cumulative anthracycline dose, and higher adjusted cPRBCv. In the final reduced
multivariate model there were statistically significant positive associations between
higher cPRBCv (p<0.0001) and older age at diagnosis (p<0.0001).
2.1.5. Conclusion
While increased cardiac iron detectable by MRI examination was absent in this
cohort, excess iron deposition was demonstrated by LIC (49.3% of the sample) and
pancreas R2* (47.2% of the sample) results. Risk factors for elevated liver iron
concentration identified in multivariate regression analysis were higher cPRBCv and
older age at diagnosis. Study findings have implications for the development of
monitoring/management guidelines for childhood cancer patients and survivors at risk of
liver/pancreatic damage due to iron overload, exploration of the additive risk of
liver/pancreatic damage from specific chemotherapeutic/other exposures, as well as for
health education to minimize further liver/pancreatic damage. Results of this study
provide the groundwork for further research aimed at better understanding of the biologic
47

mechanisms of iron overload in pediatric oncology patients/survivors and translation of
that knowledge into clinical practice.
2.2 Aims and hypotheses
This study was aimed at characterizing the pattern and severity of myocardial
iron deposition among a cohort of childhood cancer survivors who received large
cumulative packed red blood cell transfusion volumes compared with those who did not.
The hypothesis to be tested was that there would be evidence of increased myocardial
iron deposition among survivors who received large cPRBCv. An exploratory aim was to
determine the pattern and severity of liver and pancreatic iron deposition, testing the
hypothesis that there would be evidence of increased liver and pancreatic iron
deposition among survivors who received large cPRBCv.
2.3 Analyses
The study’s primary outcome of interest was organ iron deposition (indicator: iron
measurement by MRI). In addition to data acquired from laboratory and MRI evaluations
obtained for this study, disease/treatment and demographic information was abstracted
from medical records. Univariate and multivariate analyses were performed for each
endpoint. Unique multivariate models were built starting with predictors with p<0.15 via
likelihood ratio test (LRT) on univariate analysis and following a step-wise iterative
process until the multivariate LRT p-value was <0.05 for remaining covariates. This
approach was validated by assessing changes in significance through stepwise removal
of predictors, while retaining adjusted cPRBCv in the model. Data analyses were
conducted using Stata 11 (StataCorp. 2009. Stata Statistical Software: Release 11.
College Station, TX: StataCorp LP).
2.4 Results
Seventy-five childhood cancer survivors were enrolled between December 2011
and December 2012. Relevant sample characteristics were compared with the original
48

cohort of 214 CHLA childhood cancer survivors from which the sample was drawn to
determine whether there were any statistically significant (p< 0.05) differences; Chi
square tests were used for categorical variables and linear regression analysis was used
for continuous variables. No statistically significant differences were found, and the
sample was therefore considered to be representative of the cohort population. These
results and the study sample clinical and demographic characteristics are shown in
Table 2.1.
There were more males (n=40, 53%) than females. Most members of the cohort
were identified as Hispanic/Latino (n=45, 60%), reflecting the racial/ethnic composition of
the CHLA patient population. The median age at the time of study evaluation was 14
years (range, 8-25.6). Median follow up time, defined as the length of time between
treatment completion and study evaluation, was 4.2 years (range, 0.2-7.6 years). Cancer
diagnoses included acute lymphoblastic leukemia or acute myeloid leukemia (n=33) and
six types of solid tumors (n=42). Of the cohort members with solid tumors (n=42), 33
(44%) had undergone tumor resection. Median hemoglobin level at diagnosis for the
study cohort was 10.6 g/dL (range=5.5-18). There were 67 (89%) patients who received
at least one PRBC transfusion while undergoing cancer treatment, with a median of 11
(range=1-47) PRBC transfusions received per patient. The median dose of
anthracycline chemotherapy in the full study sample was 75 mg/m
2
(range, 0-739
mg/m
2
); among the 48 (64%) study participants who received anthracyclines as part of
their treatment plan, the median dose was 293 mg/m
2
(range, 50-739 mg/m
2
).
3
Nine
(12%) patients received dexrazoxane as a cardioprotectant in association with
anthracycline chemotherapy. There were 47 (62. 7%) participants who received both

3
Anthracycline was the only agent for which calculated cumulative dose was calculated because
of interest in whether it would affect cardiac function, not because it was thought that it would be
a predictor of liver/pancreatic/cardiac iron loading per se.

49

anthracycline chemotherapy and PRBC transfusions. Treatment history included
hematopoietic stem cell transplant (HSCT) in 4 (5.3%) patients. Treatment intensity was
graded by the dissertation study author and an experienced pediatric oncologist, using
the validated Intensity of Treatment Rating Scale (ITR 2.0) (Werba et al., 2007), as very
or most intensive in nearly half the participants (n=36, 48%).
Table 2.1.
Study Sample Clinical and Demographic Characteristics
Characteristic Original Cohort
(n = 214)
Study Cohort
(n=75)
Χ
2
test or
F-test p-
value∞
Sex (%)
Male 114 (53.3) 40 (53.3) 0.99
Female 100 (46.7) 35 (46.7)
Race/ethnicity (%)
Hispanic/Latino 129 (60.3) 45 (60) 0.90
Caucasian/Non-Hispanic 51 (23.8) 19 (25.3)
Asian 20 (9.3) 6 (8)
Black/African American 9 (4.2) 4 (5.3)
Other 5 (2.3) 1 (1.3)
Age at diagnosis (years)
Median 7.8 7.7 0.59
Range 0.2-20.2 1.8-20.2
Age group at diagnosis (years)
0-4 70 (32.7) 21 (28) 0.43
5-7 38 (17.86) 18 (24)
8-12 41 (19.2) 15 (20)
13-18 63 (29.4) 20 (26.7)
19-25 2 (0.9) 1 (1.3)
Diagnosis (%)
Acute lymphoblastic leukemia 57 (26.6) 23 (30.7) 0.85
Germ cell tumor 40 (18.7) 14 (18. 7)
Acute myeloid leukemia 32 (15) 10 (13.3)
Osteosarcoma 26 (12.2) 9 (12)
Ewing sarcoma 22 (10.3) 7 (9.3)
Wilms tumor 15 (7) 4 (5.3)
Nasopharyngeal carcinoma 1 (0.5) 1 (1.3)
Rhabdomyosarcoma 21 (9.8) 7 (9.8)
Initial Hemoglobin (g/dL)
Median (Range) 10.6 (5.5-18) 10.6 (5.5-18) 0.72
Tumor resection (%)
N/A 89 (41.6) 33 (44)
No 25 (11.7) 7 (9.3) 0.70
Yes 100 (46.7) 35 (46.7)
HSC Transplant (%)
No 202 (94.4) 71 (94.7) 0.90
Yes 12 (5.6) 4 (5.3)
Treatment intensity (levels)
Least intensive - Moderately intensive 108 (50.5) 39 (52) 0.93
Very intensive 66 (30.8) 22 (29.3)
Most intensive 40 (18.7) 14 (18.7)
Anthracycline chemotherapy
No 81 (37.9) 27 (36) 0.61
Yes 133 (62.2) 48 (64)
50

Table 2.1.
Study Sample Clinical and Demographic Characteristics
• ≤150 mg/m
2
54 (25.2) 22 (29.3)
• 150-350mg/m
2
33 (15.4) 9 (12)
• >350 mg/m
2
46 (21.5) 17 (22.7)
Cumulative anthracycline dose (mg/m
2
) ^
Median (Range) 75 (0-739) 75^ (0-739) 0.80
Number of patients receiving dexrazoxane as
cardioprotectant
9 (12)
PRBC transfusions (%)
No 33 (15.4) 8 (10.7) 0.36
Yes 181 (84.6) 67 (89.3)
• <10 PRBC transfusions 85 (39.7) 31 (41.3)
• ≥10 PRBC transfusions 96 (44.9) 36 (48)
Number of PRBC transfusions
Median (Range) 8 (0-89) 9 (0-47) 0.61
Number of patients receiving anthracycline
chemotherapy plus PRBC transfusions
130 (60.7) 47 (62.7) 0.67

At time of Study Evaluation

Age (years)
Median (Range) 14 (8-25.6)
Age group (years)
5-7 2 (2.7)
8-12 33 (44)
13-18 25 (33.3)
19-25 15 (20)
Follow up time (years)
Median (Range) 4.2 (0.2-7.6)
Duration of follow up (years)
<1 1 (1.4)
1-3 34 (46)
4-5 23 (31.1)
6-7 16 (21.6)
Height (cm) §
Median (Range) 154.3 (113-190)
Weight (kg) §
Median (Range) 55.6 (19-103.6)
BMI (kg/m
2
) §
Median (Range) 23.6 (12.8-40.9)
BSA (m
2
) §
Median (Range) 1.5 (0.8-2.3)
Liver iron content >1.2 mg/g ¥
Normal 30 (41.1)
Abnormal 36 (49.3)
Pancreas R2* «
Normal 38 (52.8)
Abnormal 34 (47.2)
LVEF <50 §
No 72 (97.3)
Yes 2 (2.7)
§ n=74; ¥ n=73; «n=72
Abbreviations: HSCT, hematopoietic stem cell transplant; BMI=body mass index; BSA, body surface area
∞Chi square test used for categorical variables, and linear regression for continuous variables
^48 subjects received anthracycline chemotherapy with a median dose=293 mg/m
2
(range, 50-739 mg/m
2
)

51

Blood samples obtained from study participants on the day of their MRI
examination were tested for two mutations of the hereditary hemochromatosis gene
(HFE, located on chromosome 6), which have been associated with high iron absorption
and progressive body iron overload (Bacon et al., 2011). Homozygosity, particularly for
the HFE C282Y gene mutation, would represent a potential confounder in the analyses.
Individuals who are homozygous for C282Y make up 80-85% of patients with hereditary
hemochromatosis; the H63D mutation
can be associated with iron loading in
compound C282Y/H63D heterozygotes
(Bacon et al., 2011). As shown in
Table 2.2. HFE C282Y and H63D
mutation status was normal in 71/75 and 63/75 participants, respectively. None of the
HFE C282Y mutations were homozygous. Of the 12 HFE H63D mutations, 2 were
homozygous and 10 were heterozygous. Thus, data from the 15 patients with C282Y or
H63D mutations were retained in analyses.
Blood samples obtained from study participants on the day of their MRI
examination were tested for select iron-related biomarkers, which included serum ferritin,
iron, iron binding capacity, and percent transferrin saturation. Median values for these
biomarkers fell within the normal range in the study subjects, but the ranges were wide
and exceeded the normal reference ranges’ upper or lower bounds in 36 subjects (Table
2.3). These included values above the normal range for ferritin in 21 (28.4%), serum iron
in 2 (2.8%), iron binding capacity in 8 (11.3%), and percent transferrin saturation in 5
(7%). One serum ferritin extreme outlier value (5444 mcg/dL) was noted and removed
from further analyses.

52

Table 2.3. Iron Biomarkers
Iron Biomarker Study Cohort
(n=75)
Normal Value
Reference Range
No. (%) Outside Normal
Range
Below Above
Serum ferritin (ng/mL)^
Median (range)

122.5 (13-1833)

10-291 ng/mL

--

21 (28.4)
Serum iron (mcg/dL)*
Median (range)

76 (24-188)

27-164 mcg/dL

--

2 (2.8)
Iron binding capacity (mcg/dL)*
Median (range)

347 (200-535)

271-448 mcg/dL

5 (7)

3 (4)
Percent transferrin saturation*
Median (range)

21 (8-81)

8-45%)

--

5 (7)
^n=74; *n=71
Iron binding capacity=transferrin

Additional detail regarding transfusion-related factors is given in Table 2.4. For
the full study sample, the raw median cPRBCv was 2084 mL (range, 0-14224 mL), and
median cPRBCv adjusted for body weight was 71.7 mL/kg (range, 0-391.9). For the 67
individuals in the study sample who received PRBC transfusions, the raw median
cPRBCv was 2727 mL (range, 300-14224 mL), and median cPRBCv adjusted for body
weight was 90.6 mL/kg (range, 3.6-391.9). Among the full study sample, based on
transfusion volume received, the median total body iron content was projected to be
1437 mg (range, 0-9383), and the projected median liver iron content (LIC) was 1.2 mg/g
(range, 0.9-23.7).
Table 2.4 Transfusion-Related Factors
Study Cohort (n=75)
Cumulative PRBC volume (mL)
Median 2084 (2727*)
Range 0-14224 (300-14224*)
Adjusted cumulative PRBC volume (mL/kg)
Median 71.7 (90.6*)
Range 0-391.9 (3.6-391.9*)
Projected total iron burden (mg)
Median 1437
Range 0-9383
Projected liver iron content (mg/g liver, dry weight)
Median 1.2
Range 0.9-23.7
*n=67 cohort members who received PRBC transfusions

53

Results of MRI iron quantitation and calculated tissue iron concentration for
heart, liver, and pancreas are shown in Tables 2.5 and 2.6. No abnormal cardiac T2*
results were observed, based on a cut-point of >20 milliseconds (ms) for normal cardiac
T2*. The median T2* value in the study sample (n=74) was 32.8 ms (range, 24.3-40.5
ms). Using the cardiac T2* values, cardiac iron concentration (CIC) was calculated for
the study population using the equation [CIC=45x(T2*)^-1.22] (Wood, 2011), yielding
mean, median, and range values of 0.7, 0.6, 0.5-0.9, respectively. Normal cardiac iron
concentration is 1.2 mg/g. Thus, no abnormal cardiac iron concentrations were
observed.
Liver R2* results were abnormal in 36 (49.3%) patients, with normal liver R2*
being <40 Hertz (Hz). Using liver R2* results (n=73), LIC was calculated using the
equation LIC=R2*x 0.0254+0.2 units of mg/g dry weight (Wood, Enriquez, Ghugre,
Tyzka, et al., 2005). Normal LIC is 1.0±0.2 mg/g. The study sample’s LIC mean,
median, and range were 2.6, 1.2, 0.9-23.7 mg/g. Liver iron concentration was elevated in
36 (49.3%) patients, when analyzed using a conservative cut point of 1.2 mg/g. [If 1.0
mg/g had been used as the cut-off, there would be 67 (91.8%) with abnormal values; at
1.1 mg/g, there would be 54 (74%) with abnormal values.]
Because no pancreatic iron calibration exists to convert R2* values into
pancreatic iron concentration (G. J. Anderson, McLaren, Skoien, & Powell, 2011),
pancreatic iron loading was estimated by R2* results. Normal R2* is <30 Hz. Pancreas
R2* results (mean, median, range in Hz) for the study sample were 30.2, 26, 18-128
(n=72). Although 34/72 (47.2%) study subjects had abnormal pancreatic R2* results, the
degree of pancreatic loading was small in the majority of patients (Figure 2.1). One
outlier had a pancreatic R2* value of 128; when this value was verified in the medical
record, it was noted that this result was interpreted by the radiologist as moderate
pancreatic siderosis (values of 30-100Hz constitute mild pancreatic siderosis, 100-
54

400Hz moderate siderosis, and >400Hz severe siderosis) (Noetzli et al., 2009). There
was no statistically significant difference between males and females in R2* results.
Table 2.5.
MRI Iron Quantitation in Heart, Liver and Pancreas
Study cohort Normal Values
a
No. (%) in abnormal range
Heart (ms)^ T2* >20 None
Mean 32.6

Median (Range) 32.8 (24.3 - 40.5)
Liver (Hz)§ R2* <40 36 (49.3)
Mean 94.6
Median (Range) 40.8 (27.6-925.7)
Pancreas (Hz)* R2* <30 34 (47.2)
Mean 30.2
Median (Range) 26 (18-128)
^n=74; §n=73; * n=72
a
(Wood, 2011)
Abbreviations: ms=milliseconds; Hz=Hertz; MRI imaging values for iron measurement: T2* = time needed
for a tissue to become twice as dark, R2* = rate of darkening

Table 2.6.
Iron Concentration in Heart, Liver and Pancreas
Study Cohort
Median (Range)
Normal Values No. (%) in Abnormal Range
Heart (n=74)
a
0.6 mg/g (0.5-0.9) CIC <1.2 mg/g None
Liver (n=73)
a
1.2 mg/g (0.9-23.7) LIC <1.2 mg/g 36 (58.9)
Pancreas (n=72)
b
26 (18-128) R2* <30 34 (47.2)
a
Tissue iron concentrations calculated based on published normal values (cardiac T2* >20, liver
R2* <40) and equations for conversion: CIC = 45 x [(T2*)^-1.22] and LIC = (R2* x 0.0254) + 0.2
(Wood, 2011; Enriquez, Ghugre, Tyzka, et al., 2005).
b
Pancreas R2* is presented here because there is no conversion for pancreatic iron concentration.

0
10
20
30
40
50
60
70
80
90
100
110
120
130
Pancreatic R2* (Hz)
Female Male
Gender
Figure 2.1.
Distribution of Pancreatic R2* Results in
Males and Females

55

Statistically significant correlations between liver and pancreatic R2* were found
(Table 2.7). Both hepatic and pancreatic R2* were significantly positively correlated with
serum iron, iron binding capacity, serum ferritin, and weight adjusted cRBCv.
Given that preliminary work prior to the opening of the dissertation study included
projection of iron burden values in the CHLA childhood cancer survivor cohort based on
medical record review of cPRBCv, the agreement between projected transfusional liver
iron content (Ruccione et al., 2012) and actual liver iron burden measured by MRI was
evaluated. In the preliminary work, cumulative transfusion volumes were determined for
each patient by totaling the volume of individual PRBC transfusions administered.
Projected total iron burden was then computed for each patient using the average
hematocrit value from the previous
year’s quality control data for each
product type (i.e., 70% for CPDA
[citrate-phosphate-dextrose-adenine]
units and washed PRBCs, and 55%
for units with additive solutions [AS]).
Because physiological iron excretion
is negligible in comparison with
56

transfusional iron loading, for purposes of our estimates it was assumed that 100% of
transfused iron was incorporated as total body iron burden (Kohgo et al., 2008). For
example, the estimated iron burden resulting from administration of 265 mL of CPDA-
anticoagulated PRBCs is 186 mg (265 mL x 0.7). Projected transfusion-derived LIC was
calculated according to the following formula: hepatic iron burden (mg/g, dry weight) =
total body iron burden (mg/kg) ÷ 10.6 (Angelucci et al., 2000). This projected LIC was
compared to the LIC calculated from MRI liver R2* results. As shown in Figure 2.2, there
was a statistically significant correlation between LIC projected from transfusion data
and LIC calculated from measured liver R2* (Spearman rank correlation r = 0.472, p<
0.0001).
Because cardiac iron levels were within the normal range and pancreatic iron
levels were considered to be only mildly abnormal in this study sample (normative data
for R2* are sparse, and good toxicity data for R2* results that are only mildly abnormal
are lacking) (G. J. Anderson et al., 2011), liver iron concentration (defined as the LIC
variable in Table 2.6) was the outcome variable for univariate and multivariate analyses.
As seen in Table 2.8, variables significantly associated with increased LIC in univariate
analyses included being older at the time of diagnosis (p=0.01) and at MRI evaluation
(p=0.05), higher treatment intensity rating (p<0.01), and receiving higher cumulative
PRBC volume (p< 0.0001). Having undergone HSCT was associated with LIC (p<
0.0001), but because there were only 4 patients in this group, this variable was not used
in the multivariate analyses. In addition, as noted previously, one serum ferritin extreme
outlier value (5444 mcg/dL) was noted. The univariate analysis was performed with the
outlier included and again without it. Because the estimates changed greatly without the
outlier, it was removed from analyses. In addition, log transformed ferritin values were
used because they were more normalized than the raw ferritin values.

57

Table 2.8 Univariate Analysis of Liver Iron Concentration and Key Clinical/Demographic
Variables (n=73)
Variables Coef (SE)
F-test p-
value
Reference
(SE)
MV
candidate
Sex (Ref: Female) -0.70 (0.90) 0.44 3.0 (0.65)
Race/ethnicity (Ref: Hispanic/Latino) 1.95 (0.57) 0.19 --
Caucasian/Non-Hispanic 1.54 (1.06) --
Other 1.85 (1.28) --
Diagnosis (Ref: ALL) 1.32 (0.79) 0.07 --

AML 3.86 (1.42) --
Bone tumor 1.25 (1.25) --
Soft tissue tumor 1.40 (1.08) --
Age at diagnosis, in years 0.25 (0.09) 0.01 0.38 (0.91) *
Age at MRI, in years 0.19 (0.09) 0.05 -0.08 (1.43) *
Follow-up time (end treatment to
MRI)
-0.20 (0.27) 0.47 3.46 (1.23)
Treatment Intensity Rating,
continuous
2.02 (0.53) <0.01 -0.76 (0.98) Highly
correlated
with
cPRBCv,
not used
Tumor resection (Ref: No surgery) 2.96 (1.47) 0.97 --
Yes -0.35 (1.62) --
N/A (ALL/AML) -0.43 (1.62) --
Transplant (Ref: No) 9.91 (1.60) < 0.0001 2.06 (0.38) Sparse,
not used
Cumulative anthracycline dose,
mg/m
2

0.004 (0.002) 0.08 1.94 (0.58)

Adjusted cumulative PRBC volume,
mL/kg
0.03 (0.004) < 0.0001 -0.07 (0.54)
*
Initial hemoglobin -0.08 (0.17) 0.63 3.45 (1.81)
BSA, m
2
-0.08 (1.21) 0.95 2.73 (1.91)
BMI, kg/m
2
-0.07 (0.08) 0.37 4.21 (1.85)
Height at evaluation, cm 0.02 (0.03) 0.49 -0.14 (3.95)
Weight at evaluation, g -0.01 (0.02) 0.74 3.01 (1.27)
Abbreviations: Coef = coefficient, SE = standard error, MV = multivariable analysis
Note: Each predictor was entered separately into a linear regression model to model LIC.

For the multivariate analysis, cPRBCv was retained while other candidate
predictor variables with p<0.15 via likelihood ratio test on univariate analysis were
removed in stepwise model building until the multivariate LRT p-value was <0.05 for
remaining covariates. Treatment intensity ratings was not included in multivariate
regression modeling because it was highly correlated with cPRBCv. This process
yielded a final reduced model comprised of higher cPRBCv (p<0.0001) and older age at
diagnosis (p<0.0001). cPRBCv was associated with a 0.03 increase in LIC for each
mL/kg transfused. Age at diagnosis was associated with a 0.29 increase in LIC for each
one year increase in age. The final model explained 52% of the variance in LIC among
58

the sample of 73 childhood cancer survivors who underwent liver MRI examination.
Length of follow-up was added to the model to examine for change in model statistics.
This resulted in a 0.02 increase in R
2
and no change in p-value; therefore length of
follow-up was not retained in the final model (Table 2.9)
Table 2.9. Multiple Regression Analysis of Variables Predicting Liver Iron Concentration
Predictor Variables Coef (SE) LRT p-value Reference (SE) N

-2.78 (0.78) 73 Adjusted cumulative PRBC volume (mL/kg) 0.03 (0.004) < 0.0001
Age at diagnosis, in years 0.29 (0.07) < 0.0001
Model statistics
F= 37.5 (2, 70)
R
2
=0.52
P < 0.0001

2.5 Discussion
A substantial proportion of childhood cancer patients receive cumulative PRBC
transfusion volumes (cPRBCv) in the range known to be associated with iron overload
among non-oncology pediatric patients requiring transfusion support. Although
preliminary evidence for transfusional iron overload in pediatric oncology patients has
been documented in a limited number of published reports and meeting abstracts (Blair,
2009; Eng & Fish, 2011; Gurram et al., 2012; Halonen et al., 2003; Jaffer, 2012;
Jastaniah, 2008; Lutz et al., 2008; Nelson, 2007; Nottage et al., 2012; Reddy, 2011,
2012; Ruccione et al., 2012; Vag, 2011), the prevalence, organ distribution, and severity
of transfusion-related iron overload have not been characterized across a spectrum of
leukemias, solid tumors and treatment intensities in childhood cancer survivors.
In work preparatory for the dissertation study, a retrospective cohort study of
>200 CHLA childhood cancer survivors was conducted which found that greater
treatment intensity was associated with the need for more PRBC transfusions and thus
with increased risk of iron overload (Ruccione et al., 2012). While this finding
constituted an important step in elucidating factors influencing development of iron
overload during childhood cancer treatment and carried potential significance for a
59

sizeable proportion of childhood cancer survivors, the results pointed to the need for
further research in order to provide more direct evidence regarding the true prevalence
and anatomic distribution of iron overload. In the dissertation study, a sample of 75
childhood cancer survivors representing several forms of childhood cancer was
evaluated with a single assessment of organ tissue iron quantified by magnetic
resonance imaging (MRI), serum iron biomarkers, and hemachromatosis gene mutation
status.
Seventy-four study participants underwent cardiac MRI examinations. In this
study, we found no MRI evidence of iron loading in the heart, which had been foreseen
as the principal organ of interest because: (1) cardiac complications have been shown to
be the leading non-cancerous cause of morbidity and mortality among long-term
childhood cancer survivors, (2) cardiac iron overload may damage or kill cardiomyocytes
exposed to NTBI leading to cardiomyopathy and heart failure, and (3) many childhood
cancer patients receive anthracycline chemotherapy, which may potentiate cardiac
damage caused by iron overload. The absence of measurable cardiac iron is a pertinent
negative finding that should be reassuring to patients/survivors, their families, and
pediatric oncology health care professionals. Yet relief must be tempered by the
realization that much remains unknown about the fate of transfusional iron in children
treated for cancer. Further research is needed to elucidate how normal iron homeostasis
might be modified in children with cancer through suppression of eythropoiesis by
chemotherapy, the inflammatory effects of infections and tissue breakdown, altered
nutrition, and circulating factors emanating from the neoplasms themselves. There
remains a theoretical possibility that childhood cancer survivors who receive high cRBCv
transfusion volumes may be exposed to cardiac myocyte-damaging circulating NTBI
during treatment, resulting in organ damage that was not apparent among this sample of
off-treatment patients whose median time since treatment completion was 4.2 years. In
60

addition, cardiac iron loading is a problem that is known to occur in childhood cancer
survivors, as documented in several case reports, and it is possible that cardiac iron
might have been found if the sample size had been larger.
Elevated liver iron concentration was found in 36 (58.9%) of the 73 study
participants who underwent liver MRI examination. This finding has clinical significance
due to the risks posed by excessive iron accumulation in hepatocytes, including
hepatocellular injury that may eventually lead to the development of fibrosis and cirrhosis
(Bonkovsky, 1991), and hepatocellular carcinoma (Huang, 2003; Lauret et al., 2002).
Abnormal pancreatic R2* measurements were found in 34 (47.2%) of the 72 study
participants who had evaluable pancreatic MRI examinations. While the degree of
pancreatic iron loading was relatively small in the majority of study subjects, this finding
has potential clinical implications in that pancreatic beta cell damage due to iron
overload may lead to glucose intolerance and diabetes mellitus, and hepatic disturbance
of glucose utilization has been shown to accelerate pancreatic beta cell depletion due to
hyperinsulinemia (Kohgo et al., 2008). In this study, liver R2* and pancreatic R2* results
were highly correlated (p=0.0008). This finding is in contrast to observations in patients
with thalassemia where pancreatic iron burden was more closely correlated with cardiac
iron than with liver iron (Noetzli et al., 2009). The liver and pancreas have different
mechanisms and kinetics of iron uptake and clearance with liver iron uptake primarily
mediated by transferrin, while endocrine tissue and the heart take up circulating non-
transferrin bound iron (NTBI) (Wood, 2007). Although the liver is the major iron storage
depot and high LIC conveys a risk of liver damage, most iron toxicities occur in
extrahepatic iron-sensitive target organs such as the pancreas and heart as a result of
NTBI exposure. Further research is needed to better understand the prospective risk of
subsequent glucose dysregulation in individuals with increased pancreatic R2* values,
what threshold levels of pancreatic iron loading might be critical in patients who also
61

have excessive LIC, as well as the additive risk of liver/pancreatic damage that may
arise from specific chemotherapeutic or other toxic exposures. In addition, prospective,
longitudinal studies are needed to characterize the relationship between pancreatic R2*
measurements and pancreatic function, and the potential role of pancreatic R2*
evaluation in the early recognition of physiologic conditions suitable for NTBI-mediated
organ damage.
Cumulative PRBC transfusion volume was positively associated with liver iron
concentration in both univariate and multivariate analyses. Greater cPRBCv predicted
higher liver iron concentration as measured by MRI examination. A statistically
significant correlation between the LIC projected from cPRBCv and the LIC
measured/calculated from the MRI R2* results also was demonstrated. Statistically
significant correlations were found between liver R2*, pancreatic R2* and several serum
iron biomarkers, including ferritin. The correlation between liver iron (as reflected in liver
R2*) and serum ferritin, in particular, is confirmatory of other studies as well as the
longstanding clinical practice of utilizing trends in serum ferritin to indirectly monitor iron
stores in transfused patients (Fleming & Ponka, 2012). Ferritin assays are easily
available, relatively inexpensive, and well standardized (Fischer & Harmatz, 2009),
although ferritin’s relationship to body iron stores can be altered by inflammation,
ascorbate status, oxidative stress, and increased erythropoietin (Brittenham et al., 2001;
Wood, 2011), and ferritin may not have predictive value with respect to pancreatic iron
loading. Taken together, study findings suggest that systematic monitoring and tracking
of cumulative PRBC transfusion volumes would be useful in clinical practice to facilitate
the most judicious use of PRBC transfusions. This simple practice change could flag
patients whose management requires higher PRBC volumes and who therefore should
be monitored through serum iron biomarkers such as ferritin and MRI evaluation for
early detection of subclinical iron loading. Staging algorithms for MRI examinations
62

could be developed that classify pediatric oncology patients into low and high risk
categories, based on cPRBCv, ferritin levels, and clinical characteristics as has been
done for patients with thalassemia, with the advantage that such triaging could be cost-
effective and conserve magnet time (Noetzli et al., 2009).
In both univariate and multivariate analyses, older age was associated with
increased liver iron concentration. It has been speculated that because childhood
cancer survivors usually terminate their transfusions upon completion of cancer
treatment, they might substantially mobilize their iron load over time through continued
growth and development, and that the iron burden of cancer survivors treated and
transfused as younger and smaller children might be expected to diminish over time to a
greater extent than for those treated as larger, older adolescents (Halonen et al., 2003).
Results of this study suggest that given there is no mechanism for passive excretion of
excess iron, children exposed to multiple transfusions during cancer therapy who are
older at the time of cancer treatment may sustain lifelong exposure to toxic iron, possibly
resulting in organ dysfunction many years after cure of their malignancy. However, age
is a complex construct in pediatric populations because the timing and velocity of growth
and development vary among individuals and by sex, and are influenced by nutrition,
pubertal status, and other factors. There are variations in the types of malignancies that
occur at different ages within the pediatric age range as well, and these malignancies
vary in the intensity of treatment, and hence the supportive care, required for their
management. Further study is needed to better understand the interplay among multiple
variables to explain why older age was a risk factor for increased liver iron concentration
in this study.
Limitations of this study include its cross-sectional design, which constrains
inferences about causal relationships. Also by the cross-sectional nature of the study,
we do not have pre-MRI values for iron biomarkers, nor do we have longitudinal
63

information to portray trends in organ iron uptake and disposition. Nine (12%) study
subjects received treatment with dexrazoxane, a chelating agent used as a
cardioprotectant during the administration of cardiotoxic chemotherapy, and it is
acknowledged that the extent to which dexrazoxane administration may have affected
organ iron concentration is unknown. However, this agent is not used as a chelator in
the management of iron overload and it is administered only during anthracycline
chemotherapy; thus, it would not be expected to have had much effect on organ iron
concentration (Coates, 2013). A volunteer bias is recognized as a threat to the
generalizability of our findings because there may have been undetected differences
between participants and nonparticipants. Also, individuals who could not tolerate the
MRI examination were excluded, which may have inadvertently introduced bias if
patients who were more or less likely to have organ iron loading were more apt to
participate. Other recognized threats to generalizability include survivor bias (i.e.,
individuals had to have survived to be included in the study), and the exclusion of
children less than seven years old and those individuals who received chest irradiation.
In addition, it is not known to what extent the study sample is representative of all
childhood cancer survivors with the same diagnoses and a similar length of follow up
since treatment completion, which limits the generalizability of study findings. It is
acknowledged that there may have been individual observations that exerted particular
influence on the coefficients in the data analysis; while influence diagnostics were not
performed due to time constraints, the one outlier ferritin value was removed from the
analysis and this was believed to have taken care of the most extreme influence
(Sposto, 2013). For the final manuscript to be submitted for publication, all analyses will
be re-evaluated, including looking at the scale used (log versus natural) for both the
independent and dependent variables, and reviewing some influence statistics.
64

This study is the first to establish cardiac iron content as a pertinent negative
finding among a cohort of pediatric cancer survivors with various malignancies
representing a spectrum of clinical, demographic, and treatment characteristics. The
finding that 58.9% of the study subjects had liver iron loading, which was associated with
higher cumulative PRBC transfusion volume during cancer treatment and older age at
diagnosis, points to the need for increased awareness of transfusion-related risks of iron
overload among childhood cancer patients and survivors and for collaboration between
hematology and oncology subspecialties to develop appropriate tailored monitoring and
management plans for childhood cancer patients and survivors at risk for premature
morbidity and mortality related to organ iron overload. The study findings provide
evidence for adding the clinical problem of iron overload to existing late effects
surveillance guidelines for childhood cancer survivors. In addition, study findings
regarding liver and pancreatic iron-associated health risks can be used to guide health
education interventions aimed at modifiable risk factors to minimize additional insults to
the liver and pancreas because of the association of increased liver and pancreatic iron
content with serious and potentially life-threatening morbidities. Important protective
behaviors include drinking alcohol in moderation if at all; avoiding/minimizing risks
associated with piercings, unprotected sex, and needle sharing; being immunized
against hepatitis B; exercising caution and using protective gear around insecticides and
other toxic chemicals; choosing a healthy diet that limits high-fat foods; and maintaining
a healthy weight to avoid nonalcoholic fatty liver disease (Mayo Clinic, 2011b).
Although there is compelling evidence to suggest that iron overload may
adversely affect organ function in childhood cancer survivors, there have been no prior
systematic studies of a large cohort with a variety of malignant diagnoses in whom iron
burden was quantitated using MRI technology. Among the strengths of this study is its
demonstration of the feasibility of using non-invasive MRI measurement of iron
65

deposition in several organ systems given that MRI assessment of both LIC and CIC has
achieved the critical levels of availability, utility, and validity (Wood, 2007). MRI
characterization of tissue iron deposited in the pancreas, as well as in the pituitary gland,
bone marrow, and kidney is less well validated, but may become increasingly utilized in
assessment of transfusional iron overload (Wood, 2007). At that time, study results
regarding pancreatic iron concentration should provide useful reference for comparison.
2.6 Conclusion
While increased cardiac iron detectable by MRI examination was absent in this
cohort, excess iron deposition was demonstrated in the liver (58.9% of the sample) and
the pancreas (47.2% of the sample). These findings have implications for the
development of monitoring/management guidelines for childhood cancer patients and
survivors at risk of liver/pancreatic damage due to iron overload, exploration of the
additive risk of liver/pancreatic damage from specific chemotherapeutic/other exposures,
as well as for the design of patient education interventions aimed at minimizing further
liver/pancreatic damage. Results of this study provide the groundwork for further
research aimed at better understanding the biologic mechanisms of iron overload in
pediatric oncology patients/survivors and translation of that knowledge into clinical
practice. Importantly, further research – particularly of a prospective, longitudinal nature
– that links assessment of organ iron content to organ function as well as to
genetic/genomic/epigenetic parameters that may mediate iron organ iron uptake is
needed to improve understanding of the eventual fate of iron-loaded organs in childhood
cancer survivors.
66

CHAPTER 3: CARDIAC OUTCOMES IN A SAMPLE OF CHILDHOOD CANCER
SURVIVORS IN THE EARLY POST-TREATMENT PHASE OF SURVIVORSHIP

3.1 Abstract
3.1.1Background
Childhood cancer survivors receiving multiple PRBC transfusions may be at
increased risk for exposure to non-transferrin bound iron (NTBI) in the heart, thereby
increasing the risk of serious late effects such as cardiomyopathy, arrhythmias, systolic
and diastolic dysfunction, and heart failure. Iron also has been implicated in
anthracycline cardiotoxicity.
3.1.2 Purpose
To characterize the prevalence, distribution and severity of cardiac dysfunction
and pathological remodeling among a cohort of childhood cancer survivors.
3.1.3 Methods
Design: single institution cross-sectional study of childhood cancer survivor cohort.
Organ tissue iron and cardiac structure/function parameters were assessed by magnetic
resonance imaging (MRI). Cardiac (B-type natriuretic peptide, high sensitivity cardiac
specific troponin I, high sensitivity C-reactive protein) and iron biomarkers (serum ferritin,
iron, iron binding capacity, percent transferrin saturation) were assessed.
Hemachromatosis gene mutation analysis was performed. In addition to data acquired
from laboratory and MRI evaluations obtained for this study, disease/treatment and
demographic information was abstracted from medical records. Data analysis included
descriptive approaches for data summarization and regression modeling for estimation
and testing.
3.1.4 Results
Seventy-five patients were enrolled (median age 14 years, range 8-25.6 years at
evaluation); slightly more than half were male (40/75) and most were Hispanic/Latino
67

(45/75). Median follow up time was 4.2 years. Cancer diagnoses included ALL/AML
(n=33) and solid tumors (n=42). Treatment history included HSCT in 4/75. Of the cohort
members with solid tumors (n=42), 33 (44%) had undergone tumor resection. Sixty-
seven patients had received PRBC transfusions (median=11, range=1-47), with median
cumulative PRBC transfusion volume (cPRBCv) of 2727 mL (range, 300-14224 mL), and
median cPRBCv adjusted for body weight of 90.6 mL/kg (range, 3.6-391.9).
Forty-eight (64%) study participants had received anthracycline chemotherapy as
part of their treatment plan, with a median dose of 293mg/m
2
(range, 50-739mg/m
2
).
Seventy-four study participants completed cardiac MRI examination. There were two
males in the study sample whose left ventricular ejection fraction (LVEF) was ≤50%, a
level which would trigger clinical cardiology assessment and consideration of medical
intervention. Predictors of lower LVEF found to be significant in univariate analyses
included race/ethnicity (being Caucasian/non-Hispanic or other), diagnosis of AML or
solid tumor, older age at diagnosis or at MRI evaluation, greater treatment intensity,
higher cumulative anthracycline dose, use of dexrazoxane, larger body surface area,
taller height at MRI evaluation, higher serum ferritin, and higher serum iron. Two
reduced multivariate models had nearly identical model statistics: in one model there
were statistically significant associations between decreased LVEF and the variables of
higher cumulative anthracycline dose (p=0.003) and older age at MRI (p=0.009); in the
other model there were statistically significant associations between decreased LVEF
and the variables of higher cumulative anthracycline dose (p=0.005) and taller height at
MRI (p=0.006).
3.1.5 Conclusion
In this study, at a relatively early point in post-treatment follow-up, evidence of
clinically significant impaired left ventricular function was not found in the majority of
study subjects during MRI evaluation. As in numerous previous studies, higher
68

cumulative dose of anthracycline chemotherapy was associated with lower LVEF, but in
contrast to most other published studies, older age at diagnosis and taller height at the
time of MRI evaluation were identified as risk factors for LVEF decline. A new finding
was that higher serum ferritin and serum iron were associated with declines in LVEF.
This study also demonstrated the feasibility of assessing cardiac structure and function
by MRI. Study findings support the need for ongoing clinical surveillance of childhood
cancer survivors at risk for adverse cardiac outcomes, and health education regarding
survivors’ individualized survivorship care plans and modifiable risk factors for heart
health. Building on this study’s findings, future studies should employ study designs that
are prospective and longitudinal in nature, that begin at the inception of cancer
treatment, and that include larger sample sizes as well as additional relevant biological
measures such as pubertal status, traditional cardiovascular risk factor assessment, and
cardiotoxicity-related genetic analyses.
3.2 Aims and hypotheses
The aim of this study was to assess for evidence of cardiac dysfunction and
pathologic remodeling in a sample of childhood cancer survivors who received large
cPRBCv compared with those who did not. The hypothesis tested was that there would
be evidence of impaired left ventricular (LV) function and/or pathologic remodeling due
to myocardial iron deposition in survivors who received large cPRBCv, after controlling
for anthracycline exposure and other treatment and host factors. For purposes of this
study, pathologic remodeling was defined as changes in size, shape, and function of the
heart after cardiac injury (Lipshultz et al., 2012).
3.3 Analyses
The study’s primary outcome variable was LVEF as determined by cardiac MRI
evaluation. In addition to data acquired from laboratory and MRI evaluations obtained
for this study, disease/treatment and demographic information was abstracted from
69

medical records. Data analysis included descriptive approaches for data summarization
and regression modeling for estimation and testing. Univariate and multivariate analyses
were performed for each endpoint. Unique multivariate models were built starting with
predictors with p<0.15 via likelihood ratio test (LRT) on univariate analysis and following
a stepwise iterative process until the multivariate LRT p-value was <0.05 for remaining
covariates. This approach was validated by assessing changes in significance through
stepwise removal of predictors, while retaining cumulative anthracycline dose in the
model. Data analyses were conducted using Stata 11 (StataCorp. 2009. Stata Statistical
Software: Release 11. College Station, TX: StataCorp LP).
3.4 Results
Seventy-five patients were enrolled in the study between December 2011 and
December 2012. Relevant sample characteristics were compared with the original cohort
of 214 CHLA childhood cancer survivors from which the sample was drawn to determine
whether there were any statistically significant (p<0.05) differences; Chi square tests
were used for categorical variables and linear regression analysis was used for
continuous variables. No statistically significant differences were found, and the sample
was therefore considered to be representative of the CHLA survivor cohort population.
These results and the clinical and demographic characteristics of the study sample are
shown in Table 3.1.
There were more males (n=40, 53%) than females. Most members of the cohort
were identified as Hispanic/Latino (n=45, 60%), reflecting the racial/ethnic composition of
the CHLA patient population. The median age at the time of study evaluation was 14
years (range, 8-25.6). Median follow up time, defined as the length of time between
treatment completion and study evaluation, was 4.2 years (range, 0.2-7.6 years). Cancer
diagnoses included acute lymphoblastic leukemia or acute myeloid leukemia (n=33) and
six types of solid tumors (n=42). Of the cohort members with solid tumors (n=42), 33
70

(44%) had undergone tumor resection. Median hemoglobin level at diagnosis for the
study cohort was 10.6 g/dL (range=5.5-18). There were 67 (89%) patients who received
at least one PRBC transfusion while undergoing cancer treatment, with a median of 11
(range=1-47) PRBC transfusions received per patient. The median dose of
anthracycline chemotherapy in the full study sample was 75 mg/m
2
(range, 0-739
mg/m
2
); among the 48 (64%) study participants who received anthracyclines as part of
their treatment plan, the median dose was 293 mg/m
2
(range, 50-739 mg/m
2
). Nine
(12%) patients received dexrazoxane as a cardioprotectant in association with
anthracycline chemotherapy. There were 47 (62.7%) participants who received both
anthracycline chemotherapy and PRBC transfusions. Treatment history included
hematopoietic stem cell transplant (HSCT) in 4 (5.3%) patients. Treatment intensity was
graded by the dissertation study author and an experienced pediatric oncologist, using
the validated Intensity of Treatment Rating Scale (ITR 2.0) (Werba et al., 2007), as very
or most intensive in nearly half the participants (n=36, 48%).
Table 3.1. Study Sample Clinical and Demographic Characteristics
Characteristic Original Cohort
(n = 214)
Study Cohort
(n=75)
Χ
2
test or
F-test p-
value∞
Sex (%)
Male 114 (53.3) 40 (53.3) 0.99
Female 100 (46.7) 35 (46.7)
Race/ethnicity (%)
Hispanic/Latino 129 (60.3) 45 (60) 0.90
Caucasian/Non-Hispanic 51 (23.8) 19 (25.3)
Asian 20 (9.3) 6 (8)
Black/African American 9 (4.2) 4 (5.3)
Other 5 (2.3) 1 (1.3)
Age at diagnosis (years)
Median 7.8 7.7 0.59
Range 0.2-20.2 1.8-20.2
Age group at diagnosis (years)
0-4 70 (32.7) 21 (28) 0.43
5-7 38 (17.86) 18 (24)
8-12 41 (19.2) 15 (20)
13-18 63 (29.4) 20 (26.7)
19-25 2 (0.9) 1 (1.3)
Diagnosis (%)
Acute lymphoblastic leukemia 57 (26.6) 23 (30.7) 0.85
Germ cell tumor 40 (18.7) 14 (18. 7)
Acute myeloid leukemia 32 (15) 10 (13.3)
Osteosarcoma 26 (12.2) 9 (12)
71

Table 3.1. Study Sample Clinical and Demographic Characteristics
Ewing sarcoma 22 (10.3) 7 (9.3)
Wilms tumor 15 (7) 4 (5.3)
Nasopharyngeal carcinoma 1 (0.5) 1 (1.3)
Rhabdomyosarcoma 21 (9.8) 7 (9.8)
Initial Hemoglobin (g/dL)
Median (Range) 10.6 (5.5-18) 10.6 (5.5-18) 0.72
Tumor resection (%)
N/A 89 (41.6) 33 (44)
No 25 (11.7) 7 (9.3) 0.70
Yes 100 (46.7) 35 (46.7)
HSC Transplant (%)
No 202 (94.4) 71 (94.7) 0.90
Yes 12 (5.6) 4 (5.3)
Treatment intensity (levels)
Least intensive - Moderately intensive 108 (50.5) 39 (52) 0.93
Very intensive 66 (30.8) 22 (29.3)
Most intensive 40 (18.7) 14 (18.7)
Anthracycline chemotherapy
No 81 (37.9) 27 (36) 0.61
Yes 133 (62.2) 48 (64)
• ≤150 mg/m
2
54 (25.2) 22 (29.3)
• 150-350mg/m
2
33 (15.4) 9 (12)
• >350 mg/m
2
46 (21.5) 17 (22.7)
Cumulative anthracycline dose (mg/m
2
) ^
Median (Range) 75 (0-739) 75^ (0-739) 0.80
Number of patients receiving dexrazoxane as
cardioprotectant
9 (12)
PRBC transfusions (%)
No 33 (15.4) 8 (10.7) 0.36
Yes 181 (84.6) 67 (89.3)
• <10 PRBC transfusions 85 (39.7) 31 (41.3)
• ≥10 PRBC transfusions 96 (44.9) 36 (48)
Number of PRBC transfusions
Median (Range) 8 (0-89) 9 (0-47) 0.61
Number of patients receiving anthracycline
chemotherapy plus PRBC transfusions
130 (60.7) 47 (62.7) 0.67

At time of Study Evaluation

Age (years)
Median (Range) 14 (8-25.6)
Age group (years)
5-7 2 (2.7)
8-12 33 (44)
13-18 25 (33.3)
19-25 15 (20)
Follow up time (years)
Median (Range) 4.2 (0.2-7.6)
Duration of follow up (years)
<1 1 (1.4)
1-3 34 (46)
4-5 23 (31.1)
6-7 16 (21.6)
Height (cm) §
Median (Range) 154.3 (113-190)
Weight (kg) §
Median (Range) 55.6 (19-103.6)
BMI (kg/m
2
) §
72

Table 3.1. Study Sample Clinical and Demographic Characteristics
Median (Range) 23.6 (12.8-40.9)
BSA (m
2
) §
Median (Range) 1.5 (0.8-2.3)
Liver iron content >1.2 mg/g ¥
Normal 30 (41.1)
Abnormal 36 (49.3)
Pancreas R2* «
Normal 38 (52.8)
Abnormal 34 (47.2)
LVEF <50 §
No 72 (97.3)
Yes 2 (2.7)
§ n=74; ¥ n=73; «n=72
Abbreviations: HSCT, hematopoietic stem cell transplant; BMI=body mass index; BSA, body surface area
∞Chi square test used for categorical variables, and linear regression for continuous variables
^48 subjects received anthracycline chemotherapy with a median dose=293 mg/m
2
(range, 50-739 mg/m
2
)

Figure 3.1 shows the distribution of the study sample with respect to cumulative
anthracycline dose and cumulative transfusion volume. The sample was constructed to
include patients who had high
transfusion-high anthracycline, high
transfusion-low/no anthracycline, and
low/no transfusion-low/no anthracycline
exposure histories (see also Figure 1.8).
Blood samples obtained from
study participants on the day of their MRI
examination were tested for selected
cardiac biomarkers. As shown in Table
3.2, in the majority of the cohort B type natriuretic peptide (BNP), high sensitivity
troponin I, and high sensitivity C-reactive protein (CRP) levels were within normal
ranges. However, if American Heart Association/Centers for Disease Control and
Prevention recommendations for subdividing CRP reference ranges into the four
cardiovascular risk groups validated for adults were applied (Table 3.3), 20/72 (28%)
73

participants would be in the combined higher risk/highest risk groups on the basis of
their CRP levels.
Table 3.2.
Cardiac Biomarkers in Study Cohort (n=75)
Biomarker assay Indicator of Number (%) Normal values
(reference range)
B type natriuretic peptide (pg/mL)^ LV wall stress <100 pg/mL
Normal 74 (100)
Abnormal 0 (0)
High sensitivity cardiac specific
troponin I (1ng/mL)
myocyte injury
and necrosis
<0.05 ng/mL
Normal 74 (98.6)
Abnormal 1 (1.4)
High sensitivity C-reactive protein
(mg/L)§
inflammation ≤10 mg/L
Normal 69 (95.8)
Abnormal 3 (4.2)
^n=74; §=72

Table 3.3. High Sensitivity C-reactive Protein by AHA/CDC Guidelines in the Study Sample
Relative cardiovascular risk Ranges (mg/L) Number (%)
n=72
Lower risk <1 33 (46)
Average risk 1-3 19 (26)
Higher risk 3.1-10 17 (24)
Highest risk >10 3 (4)
(Pearson et al., 2003)

MRI examination was performed to assess cardiac size, shape, and function
during a single imaging session. [Labeled structures in a cross section of the human
heart are shown in Figure 3.2 for
reference (Encyclopaedia Britannica,
2013).] All measures were obtained
either directly from MRI scanning or
derived from those measures.
Volumes of the left and right ventricles
were measured at end-systole and
end-diastole, and stroke volume was
74

calculated as the difference between end-systolic and end-diastolic volumes. Ejection
fraction (LVEF) was calculated as the ratio of stroke volume to end-diastolic volumes
and expressed as a percentage. Aortic flow was measured, with cardiac index calculated
as the ratio of aortic flow to BSA. Early (E) and late (atrial - A) ventricular filling velocity
was measured, from which the E/A ratio was calculated. The muscle mass of the left
ventricle was calculated. Blood pressures were obtained while subjects were supine in
the MRI scanner and were used to enable estimation of total vascular resistance, the
ratio between mean blood pressure and cardiac index. LV mass, end-diastolic volume,
end-systolic volume, and stroke volume were adjusted for BSA. Seventy-four study
participants completed cardiac MRI examination, and cardiac MRI results (mean,
standard deviation, median, range) for the study cohort, and separately for
males/females in the cohort, are shown in Table 3.4
Table 3.4.
MRI Assessment of Cardiac Structure and Function
Parameter Cohort sample
completing cMRI (n=74)
Females (n=35) Males (n=40)
Functional Measures
Left ventricular ejection fraction (LVEF)
% *

Mean (SD),
Median (Range)
61.5 (5.3)
62.5 (42-71.2)
62.2 (4.1)
63.1 (51.6-70)
61 (6.2)
62 (42-71.2)
Aortic flow (L/min) ^
Mean (SD)
Median (Range)
4.7 (1.2)
4.8 (2-7.7)
4.4 (0.9)
4.6 (2-6.6)
5 (1.3)
5.1 (2.5-7.7)
Cardiac index (aortic flow/BSA)^
Mean (SD)
Median (Range)
3.1 (0.6)
3.1 (1.8-4.6)
3.1 (0.5)
3.1 (2.1-4.0)
3.2 (0.6)
3.1 (1.8-4.6)
Mitral valve (MV) flow velocity E (cm/s) ¥
Mean (SD)
Median (Range)
84.2 (11.1)
86 (55-115)
86.3 (11.1)
89 (60-115)
82.4 (11)
85 (55-105)
Mitral valve (MV) flow velocity A (cm/s) ¥
Mean (SD)
Median (Range)
48.4 (12.8)
47.5 (23-80)
46.9 (13.9)
45 (26-80)
49.8 (11.7)
50 (23-72)
E/A ratio ¥
Mean (SD)
Median (Range)
1.9 (0.6)
1.7 (1-3.9)
2 (0.6)
2 (1-3.3)
1.8 (0.5)
1.6 (1.1-3.9)
Anatomic Measures
LV mass (g)^
Mean (SD)
Median (Range)
53.9 (11)
52 (34.3-83.2)
48.2 (7)
48.1 (34.3-66.3)
59.1 (10.9)
60.3 (36.1-83.2)
LV remodeling index
(LV mass/LV end- diastolic volume)*

Mean (SD)
Median (Range)
0.7 (0.1)
0.7 (0.4-1)
0.7 (0.1)
0.7 (0.5-0.8)
0.8 (0.1)
0.8 (0.4-1)
LV end-diastolic volume (mL)*
Mean (SD)
Median (Range)
75.7 (12.3)
73.3 (48.5-105.7)
73.9 (9.1)
71.5 (61.3-93)
77.2 (14.5)
76.8 (48.5-105.7)
LV end-systolic volume (mL)*
75

Table 3.4.
MRI Assessment of Cardiac Structure and Function
Mean (SD)
Median (Range)
29.1 (6.4)
28.3 (18.8-51.8)
28.1 (4.9)
28.3 (21-42.9)
30.1 (7.4)
28.7 (18.8-51.8)
Stroke volume
Mean (SD)
Median (Range)
46.5 (8.3)
45.7 (24.8-70.5)
45.9 (6.1)
44.2 (35.8-59.6)
47.1 (10)
47.6 (24.8-70.5)
Systemic Measures
Systolic blood pressure (mm Hg) §
Mean (SD)
Median (Range)
107.4 (13.7)
106 (82-154)
104 (9.6)
104 (88-128)
110.5 (15.8),
101 (82-154)
Diastolic blood pressure (mm Hg) §
Mean (SD)
Median (Range)
62.4 (12.4)
60 (37-120)
62 (10.8)
62 (41-91)
62.7 (13.8)
60 (37-120)
Mean blood pressure (mm Hg) §
Mean (SD)
Median (Range)
80.1 (13)
79 (54-139)
78.9 (10.3)
77 (58-102)
81.1 (15)
80.5 (54-139)
Total vascular resistance (mean
BP/cardiac index)

Mean (SD)
Median (Range)
26.7(6.6)
26.1 (13.9-46)
26.6 (5.1)
27 (17.2-35)
26.8 (7.8)
25.4 (13.9-46)
*n=74; ^n=73; § n=71; ¥ n=70
Abbreviations: BSA, body surface area (m
2
); cMRI, cardiac MRI
LV remodeling index is also known as arterial thickness

Analysis of the full set of 15 cardiac MRI parameters listed in Table 3.4 in
comparison to age- and sex-specific reference values culled from the literature is
planned for the final manuscript that will be submitted for publication, but was
determined to be beyond the scope of the dissertation study report. For this report a
single well-established cardiac MRI parameter, LVEF, was selected as the outcome
variable. LVEF is defined as the volume of blood pumped out of the left ventricle when it
contracts (systole). LVEF is derived from measurements of ventricular size and volume,
and lower LVEF indicates that the heart muscle has been damaged and is less efficient
in pumping blood. Because anthracyclines cause systolic dysfunction (M. J. Adams,
Duffy, S.A., Constine, L.S., & Lipshultz, S.E., 2005; Kremer & Caron, 2004), LVEF
determination is a key indicator of cardiac function included in the recommended post-
anthracycline surveillance of childhood cancer survivors, according to the Children’s
Oncology Group Long-Term Follow-up Guidelines (Children's Oncology Group, 2008).
Published reference values for mean LVEF in samples of healthy children and
young adults were used for comparison to LVEF results in this study (Table 3.5).
76

Reference values for mean LVEF on cardiac MRI, as reported by Cain and colleagues,
are 67±15% for males ages 11-20 years and 71±15.5% for females in the same age
group. Reference values from Sarikouch et al. are 64.4± 4.9% for males 8-20 years of
age and 64.4±6.1% for females in the same age group (Cain et al., 2009; Sarikouch,
2010). The mean LVEF in study sample males and females was slightly lower than
these published reference values for healthy children/young adults but within the
respective normal reference ranges (Table 3.5). There were two individuals in the study
sample whose mean LVEF was ≤50%, which is considered a clinically meaningful cut-
point for referral for further cardiology assessment (Armstrong et al., 2012). Further
analytic comparisons of study sample and reference values are planned for the
manuscript to be submitted.
Table 3.5.
Mean, SD (Range) LVEF in Study Sample Compared to Published Reference Values
Study Sample (n=74)
Mean, SD (range)
Reference Values for
Individuals Ages 11-20
Years (n=20)
Mean, SD (range)
(Cain et al., 2009)
Reference Values for
Individuals Ages 8-20 Years
(n=114) Indexed for BSA
Mean, SD (range)
(Sarikouch, 2010)
Males 60.98±6.18% (54.8-67.2) 67±15% (52-82) 64.4±4.9% (59.5-69.3)
Females 62.2±4.1% (58.1-66.3) 71±15.5% (55-86) 63.4±6.1% (57.3-69.2)
LVEF ≤50% triggers clinical cardiology assessment and consideration of medical intervention (Armstrong et
al., 2012)

Univariate and multivariate analyses were performed with LVEF as the outcome
variable. The predictor variable of having undergone HSCT was not included in analyses
as there were only four patients in this group. The cardiac biomarker troponin I was
considered censored as a predictor variable in univariate analyses. For patients with
values in the normal range, the laboratory report provided results such as “<4.” For
troponin I, there were 74 observations within the normal range, but reported as being
below the minimum of the range, leaving only one observation with an actual numeric
value. In regression analysis, there is an assumption that there is a linear relationship
77

between the endpoint and predictors, and this relationship cannot be established when it
is unknown whether the value is 0, 1, 2 or 3 or any number between 0-4. There are
methods that can handle censored values as endpoints, but not as predictors; since
troponin I was being used as a predictor, it was necessary to censor it in the analyses.
As seen in Table 3.6, variables significantly associated with lower LVEF values in
univariate analysis included race/ethnicity (being Caucasian/non-Hispanic or other)
(p=0.05), diagnosis of AML or solid tumor (p=0.0002), older age at diagnosis (p=0.02) or
at MRI evaluation (p=0.02), greater treatment intensity (p=0.03), higher cumulative
anthracycline dose (p=0.003), use of dexrazoxane (p=0.08), larger body surface area
(p=0.05), taller height at MRI evaluation (p=0.0009), higher serum ferritin (p=0.03), and
higher serum iron (p=05). The hypothesized statistically significant association between
LVEF and cPRBCv was not found (p=0.20).
Table 3.6.
Univariate Analysis of LVEF and Key Clinical/Demographic Variables
Predictor variables
Coef (SE)
F-test p-
value
Reference
(SE)
N
MV
candidate
Sex (Ref: Female) -1.18 (1.23) 0.34 62.17 (0.90) 74
Race/ethinicity (Ref:
Hispanic/Latino)
62.55 (0.78) 0.05 --
74
*
Caucasian/Non-Hispanic -1.40 (1.41) -- 74
Other -4.31 (1.73) -- 74
Diagnosis (Ref: ALL) 63.82 (1.00) 0.0002 -- 74 Highly
correlated
with
cumulative
anthracycline
dose, not
used
AML -3.44 (1.80) -- 74
Bone tumor -6.80 (1.54) -- 74
Soft tissue -0.10 (1.36) --
74
Age at diagnosis -0.30 (0.12) 0.02 64.24 (1.27) 74 *
Age at MRI -0.31 (0.13) 0.02 66.00 (1.94) 74 *
Length follow up -0.11 (0.36) 0.76 61.88(1.65) 73
Treatment intensity,
continuous
-1.74 (0.77) 0.03 64.46 (1.43)
74 *
Tumor resection (Ref: No) 62.04 (1.20) 0.17 --
74

Yes -1.70 (2.17) --
N/A (ALL/AML) 0.70 (2.18) --
HSC Transplant -3.51 (2.71) 0.20 61.73 (0.63)
74
Sparse, not
used
Cumulative anthracycline dose
(mg/m
2
)
-0.01 (0.003) 0.003 63.07 (0.77)
74 *
Use of dexrazoxane -3.33 (1.85) 0.08 61.95( 0.65) 74 *
Adjusted cumulative PRBC
volume
-0.01 (0.01) 0.20 62.45 (0.93)
74
Initial hemoglobin -0.19 (0.23) 0.40 63.59 (2.48) 74
78

Table 3.6.
Univariate Analysis of LVEF and Key Clinical/Demographic Variables
Predictor variables
Coef (SE)
F-test p-
value
Reference
(SE)
N
MV
candidate
BSA -3.18 (1.62) 0.05 66.42 (2.56) 74 *
BMI 0.07 (0.10) 0.53 60.01 (2.53) 74
Height at evaluation -0.11 (0.03) 0.0009 78.47 (4.92)
74 *
Weight at evaluation -0.04 (0.03) 0.17 63.77 (1.73) 74
LIC (from Table 2.5) -0.15 (0.15) 0.30 62.21 (0.68) 74
BNP 0.10 (0.07) 0.36 60.28 (1.16) 74
CRP -0.21 (0.15) 0.35 62.24 (0.86) 71
Serum ferritin, log transformed -2.03 (0.90) 0.03 66.12 (1.99) 74 *
Serum iron -0.04 (0.02) 0.05 64.22 (1.53) 71 *
Serum iron binding 0 (0.01) 0.10 61.68 (3.47) 70
Percent transferrin saturation -0.03 (0.05) 0.55 62.34 (1.23) 70
Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; BSA, body surface area;
BMI, body mass index; LIC, liver iron concentration; BNP, B-type natriuretic peptide; CRP, C-reactive protein

For the multivariate analysis (Tables 3.7a and 3.7b), cumulative anthracycline
dose was retained while other candidate predictor variables with p<0.15 via likelihood
ratio test on univariate analysis were removed in stepwise model building until the
multivariate LRT value was <0.05 for remaining covariates. Diagnosis was not included
in multivariate regression modeling because it was highly correlated with cumulative
anthracycline dose. This process yielded two final reduced models with very similar
model statistics, one comprised only of cumulative anthracycline dose and age at MRI
evaluation as predictors of LVEF (Model A, Table 3.7a) and the other comprised only of
cumulative anthracycline dose and height at MRI evaluation as predictors of LVEF
(Model B, Table 3.7b). Both are presented because they have nearly identical model
statistics, yet suggest somewhat different implications.
Table 3.7a Multiple Regression Analysis of Variables Predicting LVEF: Model A
Predictor variables Coef (SE) LRT p-value Reference (SE) N
Cumulative anthracycline dose (mg/m
2
) -0.01 (0.003) <0.01
67.36 (1.83) 70
Age at MRI, in years -0.30 (0.114) 0.01
Model Statistics
F(2, 67) = 7.38
R
2
= 0.18
P = 0.0013

79

In Model A, cumulative anthracycline dose was associated with a 0.01% decrease in
LVEF for each mg/m
2
increase in dose, and with a 0.30% decrease in LVEF for each
year increase in age. The final model explained 18% of the variance in LVEF among the
sample of childhood cancer survivors who underwent cardiac MRI examination.
Table 3.7b Multiple Regression Analysis of Variables Predicting LVEF: Model B
Predictor variables Coef (SE) LRT p-value Reference (SE) N
Cumulative anthracycline dose (mg/m
2
) -0.01 (0.003) 0.01
75.59 (4.65) 70
Height at MRI, in cm -0.08 (0.03) 0.01
Model Statistics
F(2, 67) = 7.78
R
2
= 0.19
P = 0.0009

In Model B, cumulative anthracycline dose was associated with a 0.01% decrease in
LVEF for each mg/m
2
increase in dose, and with a 0.08% decrease in LVEF for each cm
increase in height. The final model explained 19% of the variance in LVEF among the
sample of childhood cancer survivors who underwent cardiac MRI examination.
3.5 Discussion
Because they are highly effective in curative therapy, anthracycline
chemotherapeutic agents are employed as part of the treatment regimen for more than
half of all children with cancer (Kremer & Caron, 2004; Lipshultz & Adams, 2010).
However, treatment with anthracycline chemotherapy has long been recognized as a
risk factor for cardiomyopathy (Kremer et al., 2002; Lipshultz et al., 2005; Steinherz et
al., 1991). For example, in a study of more than 14,000 childhood cancer survivors
diagnosed between 1970 and 1986 who were enrolled in the Childhood Cancer Survivor
Study (CCSS) cohort, exposure to 250 mg/m
2
or more of anthracyclines increased the
risk of heart failure, pericardial disease, and valvular disease by two to five times
compared to survivors who had not received anthracyclines (Mulrooney et al., 2009).
Serial echocardiographic or MUGA scan monitoring for changes in LVEF (or for changes
80

in another cardiac parameter known as left ventricular shortening fraction, LVSF) is the
established recommended follow-up for individuals at risk for cardiotoxicity in order to
detect subclinical abnormalities and initiate interventions to minimize progression of
cardiovascular disease (Cain et al., 2009). In the dissertation study, cardiac anatomic
and functional assessments were performed during MRI examination.
There was a statistically significant inverse association between cumulative
anthracycline dose and LVEF in both univariate (p=0.003) and multivariate analyses
(Model A, p=0.003; Model B, p=0.005) such that higher cumulative anthracycline dose
predicted lower LVEF. This finding is consistent with numerous previous reports showing
that anthracycline-induced cardiomyopathy can occur after any dose, but that the risk
increases with time and higher cumulative doses (Kremer et al., 2002). In the CCSS
cohort, Mulrooney and colleagues found
that the relative hazard of congestive
heart failure associated with
anthracycline treatment was 2.4-fold
higher at doses of less than 250 mg/m
2
and 5.2-fold higher at doses of 250
mg/m
2
or more among survivors as
compared to their non-affected siblings
(Mulrooney et al., 2009). Figure 3.3 shows the distribution of the 48 individuals in the
dissertation study who received anthracycline chemotherapy with respect to their
cumulative anthracycline dose and LVEF. There were 23/48 study subjects (31%) who
received cumulative anthracycline doses ≤250mg/m
2
, while 25/48 (69%) had doses
>250mg/m
2
.
The distribution of the 74 cardiac MRI patients by sex and cumulative
anthracycline dose is shown in Figure 3.4. Two individuals had abnormal LVEF values
81

(i.e., below a clinically meaningful cut-point of ≤50%). Both were male adolescents
diagnosed with bone tumors: one individual with a diagnosis of Ewing sarcoma received
a cumulative anthracycline dose of 420 mg/m
2
with dexrazoxane and had an LVEF of
49%; the other individual, with a diagnosis of osteogenic sarcoma, received a cumulative
anthracycline dose of 398 mg/m
2
without
dexrazoxane and had an LVEF of 42%.
The low yield of abnormal LVEF
values may partly reflect the use of the
cardioprotectant, dexrazoxane in 12% of
the anthracycline-treated study subjects,
although the extent to which cardiotoxicity
may have been reduced or averted by
this agent is unknown. Taken at face value, the low number of abnormal LVEF values at
the study’s median follow up time of 4.2 years appears somewhat reassuring. In other
studies, subclinical abnormalities have been found on routine surveillance
echocardiography as early as one year after treatment completion among survivors
treated with anthracyclines for AML, as well as among survivors considered at low risk of
anthracycline cardiotoxicity (Abosoudah et al., 2011; Orgel et al., 2013). However,
anthracycline cardiotoxicity is characterized as being asymptomatic and progressive so it
is possible that for some individuals, the study’s evaluation point was too early in the
survivorship trajectory for cardiac abnormalities to be manifested as LVEF changes. In a
review of a large pediatric cancer treatment center’s eight-year experience with
echocardiographic surveillance for asymptomatic late-onset cardiomyopathy, Abosoudah
and colleagues found echocardiogram abnormalities in 26% of individuals screened at
seven years off therapy, and in 35% of individuals screened at eight years post-
treatment (Abosoudah et al., 2011). The incidence curve of anthracycline-induced
82

cardiomyopathy does not seem to plateau, and may accelerate as survivors enter
adulthood (Abosoudah et al., 2011; Mulrooney et al., 2009). Therefore, all individuals in
the study cohort who received anthracycline chemotherapy will require continued clinical
monitoring for cardiovascular complications to enable timely intervention, including
medication as appropriate and effective health education regarding modifiable risk
factors for heart health (e.g., obesity, physical inactivity, tobacco use, diabetes mellitus).
Diabetes mellitus is of particular concern because a significant proportion of the study
sample was found to have increased liver/pancreatic iron concentrations in a related
study (see Chapter 2), and there are potential diabetes risks associated with organ iron
overload. Study findings add to the literature concerning childhood cancer survivors’
reported higher likelihood of developing cardiovascular risk factors such as hypertension
and diabetes than age- and sex-matched controls (Lipshultz et al., 2005).
Both older age at diagnosis (p=0.02) and older age at the time of MRI evaluation
(p=0.02) were statistically significant risk factors for lower LVEF in univariate analyses,
and age at MRI evaluation remained significant in one final reduced multivariate model
(Model A, p=0.01). This finding is at odds with the preponderance of prior reports
indicating that the relevant age-related risk factor for anthracycline-induced cardiotoxicity
is being younger at the time of diagnosis/treatment, especially being less than four years
of age (Lipshultz et al., 1991; Trachtenberg et al., 2011) . An exception is a report that
noted a borderline statistically significant association of older age at treatment with
reduced LVSF in a population of ALL survivors who received anthracycline doses up to
270 mg/m
2
. The study’s authors speculated that when somewhat lower anthracycline
doses are used, the critical demand on left ventricular mechanics may be associated
with accelerated somatic growth at puberty, whereas among patients who receive higher
doses, more severe myocardial damage may be independent of growth patterns and
therefore demonstrated at a younger age (K. Sorensen et al., 2003). Although pubertal
83

status was not assessed in the dissertation study, the sample included 33 (46.7%)
patients who were in the 8-18 year age group and thus most likely to be peri-pubertal,
and potentially subject to increased left ventricular demands associated with puberty-
related growth. Another possible explanation for older age as a predictor of reduced
LVEF is related to the incidence of the cancer diagnoses represented in the study (Table
3.8), which peak at varying ages within the pediatric age group such that some types of
cancer are more likely to occur in older children/teens; at the same time, there are
variations in which malignancies are treated with anthracycline chemotherapy. Thus,
older age as a predictor of reduced LVEF may actually reflect underlying relationships
among age, cancer type, anthracycline exposure, and the anthracycline-related use of
the cardioprotectant dexrazoxane. Further research employing study designs that are
prospective and longitudinal in nature and that include large sample sizes as well as
measures of pubertal status may help to clarify this issue.
Table 3.8.
Type of Malignancy, Age Peaks and Anthracycline Use
Malignancy Treatment May
Include
Anthracyclines
Age Peak (Years) Study Sample by
Diagnostic Category
No. (%)
Acute lymphoblastic
leukemia
Yes 2-4 23 (30.7)
Germ cell tumor No 15-19 14 (18.7)
Acute myeloid leukemia Yes 15-19
Osteogenic sarcoma/Ewing
sarcoma
Yes 13-18 16 (21.3)
Rhabdomyosarcoma Yes* <5-9 7 (9.8)
Wilms tumor Yes* <5 4 (5.3)
Nasopharyngeal carcinoma No 15-19 1 (1.3)
*patients in the study sample with these diagnoses did not receive anthracyclines as part of their treatment

Being taller and having a greater BSA at the time of MRI evaluation were
statistically significant risk factors for lower LVEF in univariate analysis (p=0.0009 and
p=0.05, respectively); greater height remained significant in the second multivariate
model (Model B, p=0. 01). Height and BSA are related since BSA is calculated using a
formula that includes height; thus, we would not expect both to be retained in the same
84

multivariate model. Body habitus has not been recognized as a risk factor for either late
cardiotoxicity in general or for anthracycline-specific cardiotoxicity in childhood cancer
survivors. Height has been identified as a risk factor for coronary heart disease in the
general population, but increased risk is associated with shorter rather than taller
stature, according to a recent meta-analysis of 52 studies which found that short stature
in adults was associated with increased cardiovascular morbidity and mortality in both
sexes (Paajanen, 2010). In the study sample, being taller could be a reflection of
maturation that relates to the age (and pubertal status) relationships with diagnosis and
treatment described previously. In addition, many factors affect height, with genetics
(i.e., mid-parental height) being the most important predictor of a child’s adult height. It
is theoretically possible, though perhaps unlikely, that genetic factors favoring taller
height also increased vulnerability to anthracycline cardiotoxicity in the study sample.
Genetic factors are being studied in relation to the high inter-patient variability that has
been observed in the development and progression of anthracycline-induced
cardiotoxicity, and polymorphisms have been identified that are associated with
increased cardiomyopathy risk (Blanco et al., 2011; Visscher et al., 2011). As with the
study’s findings regarding older age, prospective, longitudinal studies with larger sample
sizes and inclusion of measures associated with traditional cardiovascular risk factors
(e.g., waist circumference, waist-to-height ratio, lipid profiles, vitamin D status) as well as
analysis of single nucleotide polymorphisms associated with anthracycline
cardiomyopathy risk may advance understanding of the study findings associating height
and LVEF.
Race/ethnicity was marginally significant (p=0.05) in univariate analysis and
indicated that being in the Caucasian/non-Hispanic or “other” group (Asian, African-
American, and unspecified) conferred higher risk of abnormal LVEF. Greater risk for
decline in LVEF among non-Hispanic Caucasians contrasts with results of a recently
85

reported study in a cohort of pediatric AML survivors, which found that decline in LVSF
was significantly worse among Latino/Hispanic subjects compared with non-Hispanic
Caucasian subjects (Orgel et al., 2013). Although African-American ancestry was
associated with higher risk of early cardiotoxicity in a review of children treated on
Pediatric Oncology Group
4
studies between 1974-1990 (Krischer, 1997), there were too
few African American subjects in the dissertation study (n=4) to confirm such an
association.
Increased serum ferritin, reflecting the amount of iron stored in the body, and
increased serum iron, indicating the amount of iron circulating in the blood, were
associated with reduced LVEF in univariate analyses (p=0.03 and p=0.05, respectively).
This is a new observation in childhood cancer survivors. A statistically significant
association between cPRBCv and decreased LVEF was not found in this dissertation
study (the hypothesis tested), nor was cardiac iron deposition observed at the
transfusion burden seen in the related study of the same subjects (described in Chapter
2), however a significant subset demonstrated excess iron deposition in the liver and/or
pancreas. It is possible that small amounts of cardiac iron, below the limit of detection or
circulating during critical windows of exposure during cancer treatment, nonetheless
catalyze ROS damage to the heart, which may be associated with subsequent declines
in LVEF. As noted in Chapter 2, further research is needed to improve understanding of
organ iron transport and elimination in the context of active cancer treatment, including
how normal iron homeostasis might be modified in children with cancer through
suppression of erythropoiesis by chemotherapy, the inflammatory effects of infections
and tissue breakdown, altered nutrition, and circulating factors emanating from
malignant cells.

4
Formerly a pediatric cancer clinical trials cooperative group of treatment institutions.
86

It is recognized that LVEF is inadequate as a sole measure of subtle changes in
myocardial function and that by the time LVEF changes are detected, cardiac
dysfunction may proceed rapidly and may be irreversible. In addition to LVEF values,
data pertaining to a panel of cardiac parameters reflecting different aspects of cardiac
structure and function were collected in this study (Table 3.9). Planned analyses of
these cardiac MRI parameters in relation to reference ranges (indexed to sex and age)
in close collaboration with Dr. Wood is expected to illuminate relationships among the
parameters to help advance the exploration of the usefulness – and potential superiority
– of cardiac MRI over echocardiography in identifying previously undiagnosed rates of
subclinical cardiac dysfunction among long-term survivors, as has been suggested in a
recent report (G. T. Armstrong et al., 2012).
Table 3.9.
Cardiac MRI Parameters at A Glance
Cardiac MRI Parameter What It Measures or Indicates
LVEF The fraction of the end-diastolic volume that is ejected with each
beat (SV/EDV)
Aortic flow The rate of blood flow to the aorta
Cardiac index (aortic flow/BSA) Aortic flow, adjusted for body surface area
Mitral valve flow velocity E Pattern of blood flow across the mitral valve is an indicator of
diastolic ventricular function. E (early) filling velocity is the
accumulation of blood during the previous systole; A (atrial) is
accumulation of blood created by atrial contraction.
Mitral valve flow velocity A
E/A ratio E component is normally >A; reversal is a clinical marker of
diastolic heart failure (i.e., greater proportion of end-diastolic
volume results from late filling rather than early filling)
LV mass/BSA The mass of the left ventricle, adjusted for body surface area
LV remodeling index (LV mass/LV
end-diastolic volume)
Ratio of LV mass to the volume of blood within a ventricle
immediately before a contraction
LV end-systolic volume (ESV) Volume of blood left in a ventricle at the end of contraction
LV end-diastolic volume (EDV) Volume of blood within a ventricle immediately before a contraction
Stroke volume Difference between LV end-systolic and LV end-diastolic volumes
– the volume of blood ejected with each heart beat
Systolic blood pressure Pressure exerted by circulating blood on arterial blood vessel walls
during pumping of the heart (systole)
Diastolic blood pressure Pressure exerted by circulating blood on blood vessel walls
between contractions (diastole)
Mean blood pressure A calculation based on the values of systolic and diastolic blood
pressure (but not the average of the two)
Total vascular resistance (mean
BP/cardiac index)
The resistance of all vasculature in the systemic circulation.
Decrease in resistance results in increased flow to tissues and an
increased venous flow back to the heart; increased resistance
(e.g., vessel narrowing) decreases flow to tissues and decreases
venous flow back to the heart.

87

Limitations of this study include its cross-sectional design, which constrains
inferences about causal relationships. Also by the cross-sectional nature of the study,
we do not have pre-MRI values for cardiac biomarkers, nor do we have longitudinal
information to portray trends in the natural history of these biomarkers among the study
cohort, which were largely within normal limits at the time of MRI evaluation, but may
have fluctuated during and after cancer treatment. Nine (12%) study subjects received
treatment with dexrazoxane, a chelating agent used as a cardioprotectant during the
administration of cardiotoxic chemotherapy, and it is acknowledged that the extent to
which dexrazoxane administration may have affected LVEF results is unknown. There
are other unstudied and therefore undetected factors that may have affected LVEF
results, such as genetic predisposition and lipid profiles. It is also recognized that LVEF
in normal subjects is probably age-dependent in the pediatric age group, such that some
of the association of age and height might not reflect vulnerability to cardiotoxicity, but
simply normal physiologic adaptation not adequately captured by published norms. A
volunteer bias is recognized as a threat to the generalizability of our findings because
there may have been undetected differences between participants and nonparticipants.
Also, individuals who could not tolerate the MRI examination were excluded, which may
have inadvertently introduced bias if patients who were more or less likely to have organ
iron loading were more apt to participate. Other recognized threats to generalizability
include survivor bias (i.e., individuals had to have survived to be included in the study),
and the exclusion of children less than seven years old and those individuals who
received chest irradiation. It is not known to what extent the study sample is
representative of all childhood cancer survivors with the same diagnoses and a similar
length of follow up since treatment completion, which limits the generalizability of study
findings.
88

This study is the first to confirm the association between higher cumulative
anthracycline dose and decline in LVEF through the use of non-invasive cardiac MRI
examination in a large cohort of childhood cancer survivors with a variety of malignant
diagnoses in early survivorship phase. Among the strengths of this study is its
demonstration of the feasibility of acquiring MRI images to assess cardiac structure and
function in an imaging session that also included iron quantitation in the heart, liver, and
pancreas. In contrast to most previous reports of anthracycline cardiotoxicity, risk
factors associated with declines in LVEF in this study included older age at diagnosis,
and taller height, factors that suggest interplay among pubertal status, diagnosis,
treatment, traditional cardiovascular risk factors, and genetic polymorphisms that may be
fruitful avenues for further research. Study findings did not support the hypothesis that
higher cPRBCv was associated with declines in LVEF. However, the finding that
increased serum ferritin and serum iron were associated with declines in LVEF suggests
the possibility that study subjects, a significant subset of whom were shown in a related
study to have liver and pancreatic iron loading, may have sustained NTBI-induced
damage to cardiac myocytes during critical windows of exposure while receiving cancer
treatment and PRBC transfusions. Although there was a low yield of clinically significant
abnormal LVEF values in the study sample, the progressive and asymptomatic nature of
anthracycline-induced cardiomyopathy underscores the need for ongoing clinical
monitoring. Further study of data from a panel of cardiac parameters may contribute to
continuing evaluation of the most useful and cost-effective tools for such monitoring
(e.g., MRI, newer types of echocardiography, various biomarkers and genetic/genomic
tests). Study findings regarding heart-related health risks, as well as the diabetes risk
associated with the increased incidence of liver/pancreatic iron concentrations among
the study sample, can be used to guide health education interventions aimed at
modifiable risk factors. Important protective behaviors related to anthracycline-induced
89

and traditional cardiovascular risks include: (1) avoiding smoking or tobacco use; (2)
implementing an exercise program in consultation with a cardiologist and exercise
physiologist; (3) eating a heart healthy diet; (4) maintaining a healthy weight; (5) avoiding
illicit drug use and alcohol consumption; (6) participating in regular screening for blood
pressure, cholesterol, and diabetes (M. J. Adams, Duffy, S.A., Constine, L.S., &
Lipshultz, S.E., 2005; Mayo Clinic, 2011a). In addition, the health education curriculum
should help ensure that all childhood cancer survivors have a survivorship care plan and
are familiar with the evidence- and consensus-based guidelines for lifetime monitoring
for risks associated with their treatment history (Children's Oncology Group, 2008).
3.6 Conclusion
With the advent of population-based long-term follow up survivorship studies,
cardiovascular complications have emerged as the leading cause of non-cancer
morbidity and premature mortality in childhood cancer survivors (Oeffinger et al., 2006),
The problem of cancer treatment-related cardiac morbidity and mortality takes on even
broader significance given the increasing number of cancer survivors, the comorbidities
of an aging cancer survivor population, and the increasing or continued use of classic
and targeted chemotherapeutic agents that adversely affect the cardiovascular system.
This study demonstrated the feasibility of assessing cardiac structure and function by
MRI relatively early in the survivorship trajectory. Although we did not find evidence of
clinically significant impaired LVEF in the majority of study subjects, higher cumulative
dose of anthracycline chemotherapy was associated with declines in LVEF, which was
confirmatory of numerous previous studies. In contrast to most other published studies,
however, older age at MRI evaluation and taller height at the time of MRI evaluation
were identified as risk factors for LVEF decline. A new finding was that higher serum
ferritin and serum iron were associated with declines in LVEF. Study findings support
the need for ongoing clinical surveillance of childhood cancer survivors at risk for
90

adverse cardiac outcomes, health education regarding survivors’ individualized
survivorship care plans and modifiable risk factors for heart health. Building on this
study’s findings, future studies should employ study designs that are prospective and
longitudinal in nature, that begin at the inception of cancer treatment, and that include
larger sample sizes as well as additional relevant biological measures such as pubertal
status, traditional cardiovascular risk factor assessment, and cardiotoxicity-related
genetic analyses.
91

CHAPTER 4: HEALTH-RELATED QUALITY OF LIFE AMONG SURVIVORS OF
CHILDHOOD CANCER AT RISK FOR CARDIAC DYSFUNCTION

4.1 Abstract

4.1.1. Background
Childhood cancer survivors receiving multiple PRBC transfusions may be at
increased risk for iron deposition in vital organs, thereby increasing the risk of organ
dysfunction with associated symptoms such as fatigue that may diminish their health-
related quality of life (HRQOL).
4.1.2. Purpose
To assess HRQOL using patient-reported outcome measures (PROs) in a cohort
of childhood cancer survivors.
4.1.3 Methods
Design: single institution cross-sectional study of a childhood cancer survivor
cohort. Childhood cancer survivors were recruited to participate in a structured interview
coincident with the study-required MRI evaluation and laboratory tests. During the
interview, participants (and their parent/guardian, if available), completed the PedsQL
TM

4.0 Generic Core Scale and PedsQL™ Multidimensional Fatigue Scale (General Fatigue
subscale) in English or Spanish. A brief Cardiac Health Form was also completed. In
addition to data acquired from laboratory and MRI evaluations obtained for this study,
disease/treatment and demographic information was abstracted from medical records.
Data analysis included descriptive approaches for data summarization and regression
modeling for estimation and testing.
4.1.4. Results
Seventy study participants and 63 parents completed the PedsQL
TM
Generic
Core Scales; 71 study subjects and 63 parents completed the PedsQL
TM
Fatigue Scales.
On average, childhood cancer survivors in this study had HRQOL total scores, physical
92

health summary scores and psychosocial health summary scores that were
approximately midway between those of pediatric patients with cancer in active
treatment and those of healthy children. In general, participants rated their overall
HRQOL as good, but there were subsets ranging from 13% to 17% of the sample whose
scores were ≥1 SD below the mean, indicating at-risk status for diminished HRQOL. On
subscales within the fatigue measure, subjects reported the fewest symptoms of general
fatigue, followed by sleep-rest fatigue, and they reported the most symptoms of cognitive
fatigue. Study participants endorsed more fatigue symptoms on every scale than did
healthy children. In univariate analyses, factors found to be statistically significantly
associated with better self-rated psychosocial HRQOL included endorsement of fewer
symptoms of fatigue and better perceived physical health. Risk factors for poorer
physical health were female sex and lower LVEF. In multivariate analysis older age at
evaluation (p=0.0003), better perceived physical health (p< 0.0001) and fewer symptoms
of fatigue (p< 0.0001) were predictive of better self-rated psychosocial HRQOL. A
statistically significant positive linear relationship was observed between patient self-
report and parent proxy report on all aspects of HRQOL and fatigue in this study,
although there was cross-informant variance in ratings of individual items on study
measures. Participants’ self-rated their HRQOL slightly better, but endorsed more
fatigue symptoms than did their parents’ proxy-reports.
4.1.5. Conclusion

Although most study participants rated their psychosocial HRQOL as good, a
subset may have increased risk for diminished HRQOL. Protective factors for
psychosocial HRQOL included older age when evaluated, endorsement of fewer
symptoms of fatigue, and better perceived physical health. Risk factors for poorer
physical health were female sex and lower LVEF. Taken together, study findings point
to the importance of systematic assessment of HRQOL, fatigue and other symptoms
93

using validated measures. Study findings also point to the value of obtaining information
from both patient and parent whenever possible during survivorship care visits when
there is a window of opportunity to to identify symptoms and risk factors amenable to
intervention and to implement interventions that can mitigate adverse psychosocial
effects of the cancer experience and improve long-term HRQOL. As the initial
assessment point in longitudinal follow-up of the study sample, study data provide a
baseline for future comparisons that may help to better understand the dynamic nature
of the psychosocial impact of cancer during the first years of post-treatment survivorship.
Longitudinal studies are needed to further explore risk and protective factors, as well as
to identify and evaluate targeted interventions aimed at minimizing adverse effects of
cancer treatment such as fatigue, and maximizing healthy survivorship. Ultimately, what
is learned from research focused on psychosocial outcomes for survivors of childhood
cancer will help drive evidence-based interventions that support successful transitions
into adulthood and long-term survivorship for this large and growing population.
4.2 Aims and hypotheses
Using validated PRO measures, this study was aimed at evaluating HRQOL
constructs including psychosocial health (emotional functioning, social functioning,
school functioning), physical health, and symptoms of general fatigue, sleep/rest fatigue,
and cognitive fatigue in a cohort of childhood cancer survivors. Patient-reported HRQOL
and fatigue were evaluated for possible associations with transfusion history, organ iron
content, anthracycline exposure and impaired left ventricular ejection fraction (LVEF). It
was hypothesized that self-reported indicators of HRQOL would be lower in survivors
exhibiting organ iron deposition and/or impaired LVEF. An exploratory aim was to collect
parent/guardian proxy-reported data to assess congruence with patient self-report
regarding HRQOL and symptoms of fatigue.

94

4.3 Analyses
The study’s primary outcome variable was patient-reported psychosocial HRQOL
(indicator: PedsQL
TM
Psychosocial summary score). In addition to data acquired from
laboratory and MRI evaluations obtained for this study, disease/treatment and
demographic information was abstracted from medical records. Data analysis included
descriptive approaches for data summarization and regression modeling for estimation
and testing. Univariate and multivariate analyses were performed for each endpoint.
Unique multivariate models were built starting with predictors with p<0.15 via likelihood
ratio test (LRT) on univariate analysis and following a stepwise iterative process until the
multivariate LRT p-value was <0.05 for remaining covariates. This approach was
validated by assessing changes in significance through stepwise removal of predictors,
while retaining age at study evaluation and duration of follow-up in the model. Data
analyses were conducted using Stata 11 (StataCorp. 2009. Stata Statistical Software:
Release 11. College Station, TX: StataCorp LP).
4.4 Results
Seventy-five patients were enrolled in the study between December 2011 and
December 2012. Relevant sample characteristics were compared with the original cohort
of 214 CHLA childhood cancer survivors from which the sample was drawn to determine
whether there were any statistically significant (p<0 .05) differences; Chi square tests
were used for categorical variables and linear regression analysis was used for
continuous variables. No statistically significant differences were found, and the sample
was therefore considered to be representative of the CHLA survivor cohort population.
These results and the clinical and demographic characteristics of the study sample are
shown in Table 4.1.
There were more males (n=40, 53%) than females. Most members of the cohort
were identified as Hispanic/Latino (n=45, 60%), reflecting the racial/ethnic composition of
95

the CHLA patient population. The median age at the time of study evaluation was 14
years (range, 8-25.6). Median follow up time, defined as the length of time between
treatment completion and study evaluation, was 4.2 years (range, 0.2-7.6 years). Cancer
diagnoses included acute lymphoblastic leukemia or acute myeloid leukemia (n=33) and
six types of solid tumors (n=42). Of the cohort members with solid tumors (n=42), 33
(44%) had undergone tumor resection. Median hemoglobin level at diagnosis for the
study cohort was 10.6 g/dL (range=5.5-18). There were 67 (89%) patients who received
at least one PRBC transfusion while undergoing cancer treatment, with a median of 11
(range=1-47) PRBC transfusions received per patient. The median dose of
anthracycline chemotherapy in the full study sample was 75 mg/m
2
(range, 0-739
mg/m
2
); among the 48 (64%) study participants who received anthracyclines as part of
their treatment plan, the median dose was 293 mg/m
2
(range, 50-739 mg/m
2
). Nine
(12%) patients received dexrazoxane as a cardioprotectant in association with
anthracycline chemotherapy. There were 47 (62.7%) participants who received both
anthracycline chemotherapy and PRBC transfusions. Treatment history included
hematopoietic stem cell transplant (HSCT) in 4 (5.3%) patients. Treatment intensity was
graded by the dissertation study author and an experienced pediatric oncologist, using
the validated Intensity of Treatment Rating Scale (ITR 2.0) (Werba et al., 2007), as very
or most intensive in nearly half the participants (n=36, 48%).
Table 4.1.
Study Sample Clinical and Demographic Characteristics
Characteristic Original Cohort
(n = 214)
Study Cohort
(n=75)
Χ
2
test or
F-test p-
value∞
Sex (%)
Male 114 (53.3) 40 (53.3) 0.99
Female 100 (46.7) 35 (46.7)
Race/ethnicity (%)
Hispanic/Latino 129 (60.3) 45 (60) 0.90
Caucasian/Non-Hispanic 51 (23.8) 19 (25.3)
Asian 20 (9.3) 6 (8)
Black/African American 9 (4.2) 4 (5.3)
Other 5 (2.3) 1 (1.3)
Age at diagnosis (years)
96

Table 4.1.
Study Sample Clinical and Demographic Characteristics
Median 7.8 7.7 0.59
Range 0.2-20.2 1.8-20.2
Age group at diagnosis (years)
0-4 70 (32.7) 21 (28) 0.43
5-7 38 (17.86) 18 (24)
8-12 41 (19.2) 15 (20)
13-18 63 (29.4) 20 (26.7)
19-25 2 (0.9) 1 (1.3)
Diagnosis (%)
Acute lymphoblastic leukemia 57 (26.6) 23 (30.7) 0.85
Germ cell tumor 40 (18.7) 14 (18. 7)
Acute myeloid leukemia 32 (15) 10 (13.3)
Osteosarcoma 26 (12.2) 9 (12)
Ewing sarcoma 22 (10.3) 7 (9.3)
Wilms tumor 15 (7) 4 (5.3)
Nasopharyngeal carcinoma 1 (0.5) 1 (1.3)
Rhabdomyosarcoma 21 (9.8) 7 (9.8)
Initial Hemoglobin (g/dL)
Median (Range) 10.6 (5.5-18) 10.6 (5.5-18) 0.72
Tumor resection (%)
N/A 89 (41.6) 33 (44)
No 25 (11.7) 7 (9.3) 0.70
Yes 100 (46.7) 35 (46.7)
HSC Transplant (%)
No 202 (94.4) 71 (94.7) 0.90
Yes 12 (5.6) 4 (5.3)
Treatment intensity (levels)
Least intensive - Moderately intensive 108 (50.5) 39 (52) 0.93
Very intensive 66 (30.8) 22 (29.3)
Most intensive 40 (18.7) 14 (18.7)
Anthracycline chemotherapy
No 81 (37.9) 27 (36) 0.61
Yes 133 (62.2) 48 (64)
• ≤150 mg/m
2
54 (25.2) 22 (29.3)
• 150-350mg/m
2
33 (15.4) 9 (12)
• >350 mg/m
2
46 (21.5) 17 (22.7)
Cumulative anthracycline dose (mg/m
2
) ^
Median (Range) 75 (0-739) 75^ (0-739) 0.80
Number of patients receiving dexrazoxane as
cardioprotectant
9 (12)
PRBC transfusions (%)
No 33 (15.4) 8 (10.7) 0.36
Yes 181 (84.6) 67 (89.3)
• <10 PRBC transfusions 85 (39.7) 31 (41.3)
• ≥10 PRBC transfusions 96 (44.9) 36 (48)
Number of PRBC transfusions
Median (Range) 8 (0-89) 9 (0-47) 0.61
Number of patients receiving anthracycline
chemotherapy plus PRBC transfusions
130 (60.7) 47 (62.7) 0.67

At time of Study Evaluation

Age (years)
Median (Range) 14 (8-25.6)
Age group (years)
5-7 2 (2.7)
8-12 33 (44)
97

Table 4.1.
Study Sample Clinical and Demographic Characteristics
13-18 25 (33.3)
19-25 15 (20)
Follow up time (years)
Median (Range) 4.2 (0.2-7.6)
Duration of follow up (years)
<1 1 (1.4)
1-3 34 (46)
4-5 23 (31.1)
6-7 16 (21.6)
Height (cm) §
Median (Range) 154.3 (113-190)
Weight (kg) §
Median (Range) 55.6 (19-103.6)
BMI (kg/m
2
) §
Median (Range) 23.6 (12.8-40.9)
BSA (m
2
) §
Median (Range) 1.5 (0.8-2.3)
Liver iron content >1.2 mg/g ¥
Normal 30 (41.1)
Abnormal 36 (49.3)
Pancreas R2* «
Normal 38 (52.8)
Abnormal 34 (47.2)
LVEF <50 §
No 72 (97.3)
Yes 2 (2.7)
§ n=74; ¥ n=73; «n=72
Abbreviations: HSCT, hematopoietic stem cell transplant; BMI=body mass index; BSA, body surface area
∞Chi square test used for categorical variables, and linear regression for continuous variables
^48 subjects received anthracycline chemotherapy with a median dose=293 mg/m
2
(range, 50-739 mg/m
2
)

HRQOL and fatigue were assessed by self-report in English or Spanish using
developmentally appropriate versions of the PedsQL
TM
Generic Core Scale and
PedsQL
TM
Multidimensional Fatigue Scale and corresponding parent-proxy report forms.
Seventy study participants and 63 parents completed the PedsQL
TM
Generic Core
Scales; 71 study subjects and 63 parents completed the PedsQL
TM
Fatigue Scales
(Table 4.2). Although obtaining parent proxy-report of HRQOL and fatigue was not a
primary aim of the study, parental presence during the study evaluation provided an
opportunity to gather data regarding parents’ perceptions about their child’s HRQOL and
fatigue and to assess how similar or different patient and parent perceptions might be.
[Data pertaining to parent proxy-reports are included in Tables 4.2, 4.3, 4.4, and 4.6. for
reference.] The majority of study subjects completed the scales in English, while one
98

third of the parents completed the scales in Spanish, reflecting the linguistic distribution
of the CHLA patient population from whom the study sample was drawn, which is largely
Hispanic/Latino (59% in this study). Among this population, most of the children are
bilingual in English/Spanish or monolingual in English while their parents are bilingual in
English/Spanish or monolingual in Spanish.
Table 4.2.
Versions of PedsQL
TM
Generic Core Scale and
TM
Fatigue Scale
Completed by Participants and Parents/Guardians
PedsQL
TM
Generic Core Scale PedsQL
TM
Fatigue
Version Self-Report
Number (%)
Parent Proxy-Report
Number (%)
Self-Report
Number (%)
Parent Proxy-Report
Number (%)*
English Spanish English Spanish English Spanish English Spanish
5-7 years 2 (2.9) 0 2 (4.7) 0 2 (2.9) 0 2 (4.7) 0
8-12 years 30 (44.1) 1 (50) 22 (52.4) 11 (52.4) 30 (43.5) 1 (50) 22 (52.4) 11 (52.4)
13-18 years 21 (30.9) 1 (50) 11 (26.2) 10 (47.6) 22 (31.9) 1 (50) 11 (26.2) 10 (47.6)
18-25 years 15 (22.1) 0 7 (16.7) 0 15 (21.7) 0 7 (16.7) 0
Totals 68 2 42 21 69 2 42 21
70 63 71 63

The PedsQL
TM
Core Scale total score includes a physical health (8 items) score
and a composite of emotional (5 items), social (5 items), and school (5 items) functioning
scores that together comprise the psychosocial health score. When completing these
instruments, respondents indicate the extent to which each item had been a problem
(0=never to 4=almost always) during the preceding month. Items are reverse scored and
linearly transformed to a 0-100 scale with higher scores indicating better quality of life.
As shown in Table 4.3, with the exception of emotional functioning, mean scores for
survivors’ self-ratings were slightly higher than mean scores for parents’ proxy-ratings on
all measures of HRQOL. On subscales within the Psychosocial Health measure, both
subjects and parents rated social functioning highest, followed by emotional functioning,
with school functioning rated lowest.
99

Table 4.3.
HRQOL by Participant and Parent Proxy-Report:
PedsQL
TM
Total Score, Physical Health, and Psychosocial Health
Patient Self-Report (n=70) Parent Proxy-Report (n=63)
Measure Score Score
PedsQL
TM
Generic Core Scale Total Score
Mean (SD) 76.42 (16.54) 74.76 (16.99)
Median (range) 79.89 (17.39-100) 76.09 (32.61-100)
Physical Health
Mean (SD) 76.97 (19.71) 75.10 (21.58)
Median (range) 81.25 (12.5-100) 81.25 (12.5-100)
Psychosocial Health
Mean (SD) 76.14 (16.72) 74.59 (16.94)
Median (range) 78.33 (15-100) 73.33 (30-100)
• Emotional Functioning
Mean (SD) 73.89 (19.77) 74.12 (19.93)
Median (range) 75 (0-100) 75 (35-100)
• Social Functioning
Mean (SD) 84.11 (16.92) 80.95 (19.02)
Median (range) 90 (30-100) 85 (35-100)
• School Functioning
Mean (SD) 70.43 (20.30) 68.22 (22.20)
Median (range) 75 (5-100) 70 (0-100)
Abbreviations: SD, standard deviation

Varni and colleagues have previously explored one standard deviation (SD)
below the population mean as a meaningful cut-point for at-risk status for impaired
HRQOL relative to a population sample (Varni et al., 2007b). In this study (Table 4.4)
subsets with increased risk for impaired HRQOL ranged from 12.86% to 17.14% of the
study sample by patient self-report, and from 14.52% to 22.22% by parent proxy-report.
There was a greater proportion in the at-risk group for the physical health category by
parent proxy-rating than by participant self-rating. Within the psychosocial health
subscales, the category with the largest at-risk group was social functioning according to
both patient self-ratings and parent proxy-ratings. The second highest at-risk subset by
participant self-rating was school functioning, while emotional functioning was second
highest by parent proxy-rating.

100

Table 4.4.
PedsQL
TM
Generic Cores Scales: At Risk Status for Impaired HRQOL
Scale Number Mean SD >1 SD
Cut-off
Values
Number %
Patient self-report 70
Total score 76.42 16.54 59.88 9 12.86
Physical Health 76.97 19.71 57.26 9 12.86
Psychosocial Health 76.14 16.72 59.42 10 14.29
• Emotional
functioning
73.89 19.77 54.12 9 12.86
• Social
Functioning
84.11 16.92 67.19 12 17.14
• School
Functioning
70.43 20.30 50.13 10 14.29

Parent Proxy Report 63
Total score 74.76 16.99 57.77 10 15.87
Physical Health 75.10 21.58 53.52 12 19.05
Psychosocial Health 74.59 16.94 57.65 12 19.05
• Emotional
functioning
74.12 19.93 54.19 11 17.46
• Social
Functioning
80.95 19.02 61.93 14 22.22
• School
Functioning
68.22 22.20 46.02 9 14.52
Abbreviations: SD, standard deviation

Study participants’ mean HRQOL total scores, physical health and psychosocial
health summary scores were compared with published scores on the same scales
previously self-reported by populations of children/adolescents with various health
conditions and to self-reports by healthy children/adolescents (Varni et al., 2007b). At a
median time of 4.2 years after treatment completion childhood cancer survivors in this
study had scores in that fell approximately midway between those of patients with
cancer in active treatment and those of healthy children (Table 4.5).
Table 4.5.
PedsQL™ Self-Reported Mean PedsQL™ Total Scores, Physical Health,
and Psychosocial Health Summary Scores in Various Samples
Sample N Age
Range
(Years)
Mean (SD)
Total Health
Scores*
Mean (SD)
Physical Health
Summary
Scores*
Mean (SD)
Psychosocial
Health Summary
Scores*
1. Current study
sample
70 8-25.56 76.42 (16.54) 76.97 (19.71) 76.14(16.54)
2. Pediatric cancer
sample
393 5-19 71.97 (16.12) 71.97 (21.37) 72.1 (16.3)
3. Pediatric asthma
sample
165 7-12 74.85 (16,52) 76.51 (18.01) 74.0 (18.4)
101

4. Pediatric cardiac
sample
426 5-19 77.47 (14.51) 82.28 (15.68) 75.0 (16.1)
5. Pediatric diabetes
sample
331 5-19 80.35 (12.89) 85.89 (13.33) 77.3 (14.6)
6. Healthy children
sample
5480 5-18.1 83.84 (12.65) 87.53 (13.50) 81.9 (14.1)
Samples 2-6 from Varni, J. W., Limbers, C. A., & Burwinkle, T. M. (2007). Impaired health-related quality of
life in children and adolescents with chronic conditions: a comparative analysis of 10 disease clusters and
33 disease categories/severities utilizing the PedsQL 4.0 Generic Core Scales. Health Qual Life Outcomes,
5, 43.

*Acute version- 30 day recall

The PedsQL™ Multidimensional Fatigue Scale includes a general fatigue score
(6 items), sleep/rest fatigue score (6 items), and cognitive fatigue score (6 items). The
measure has the same format, instructions, Likert response scale and scoring method
as the PedsQL™ 4.0 Generic Core Scales (Varni et al., 2002). Items are reverse scored
and linearly transformed to a 0-100 scale with higher scores indicating fewer symptoms
of fatigue. As shown in Table 4.6, the fatigue total score was slightly lower (i.e.,
endorsing more symptoms of fatigue) when self-rated by subjects than when rated by
parent proxy-report. Both study subjects’ and parent proxy mean scores on subscales
within the fatigue scale indicated the category with the most symptoms was cognitive
fatigue, followed by sleep/rest fatigue, and then by general fatigue.
Table 4.6.
Fatigue by Patient and Parent-Proxy Report:
PedsQL
TM
Multidimensional Fatigue Scale
Patient Self-Report (n=71) Parent Proxy-Report (n=63)
Measure Score Score
Total Score
Mean (SD), Median (Range) 70.89 (15.35), 70.83 (43.06-100) 72.16 (19.24), 76.39 (19.44-100)

General Fatigue
Mean (SD), Median (Range) 75.65 (15.57), 79.17 (41.67-100) 74.11 (20.50), 75 (8.33-100)

Sleep/rest Fatigue
Mean (SD), Median (Range) 72.16 (19.68), 75 (29.17-100) 72.16 (19.68), 75 (29.17-100)

Cognitive Fatigue
Mean (SD), Median (Range) 69.78 (23.32), 75 (0-100) 70.25 (24.61), 75 (0-100)

102

Study participants’ mean self-reported fatigue scale scores were compared with
published scores on the same scales self-reported by populations of
children/adolescents with cancer and by healthy children/adolescents (Varni et al.,
2002). As shown in Table 4.7, childhood cancer survivors in this study had total fatigue
scores, general fatigue scores, sleep/rest fatigue scores, and cognitive fatigue scores
that were very close to those of the published pediatric cancer sample, which included
patients on active treatment and in the post-treatment survivorship phase. The study
sample endorsed more fatigue symptoms on every scale than did the sample of healthy
children.
Table 4.7.
PedsQL™ Self-Reported Mean PedsQL™
Multidimensional Fatigue Scale Scores in Various Samples
Sample N Age
Range
(Years)
Mean (SD)
Total
Fatigue
Score
Mean (SD)
General
Fatigue
Score*
Mean (SD)
Sleep/Rest
Fatigue Score *
Mean (SD)
Cognitive
Fatigue Score*
1. Current
study
sample
70 8-25.56 70.89 (15.35) 75.65 (15.57) 72.16 (19.68) 69.78 (23.32)
2. Pediatric
cancer
sample
220 5-19 70.98 (18.20) 74.99 (19.59) 67.03 (23.08) 70.92 (22.35)
3.Healthy
children
sample
52 5-18.1 80.49 (13.33) 85.34 (14.95) 75.00 (18.76) 81.14 (17.43)
Samples 2-3 from Varni, J. W., Burwinkle, T. M., Katz, E. R., Meeske, K., & Dickinson, P. (2002). The
PedsQL in pediatric cancer: reliability and validity of the Pediatric Quality of Life Inventory Generic Core
Scales, Multidimensional Fatigue Scale, and Cancer Module. Cancer, 94(7), 2090-2106.

*Acute version- 30 day recall

To address the study’s exploratory aim, patient self-report and parent proxy-
report data for HRQOL and fatigue were assessed for concordance. As noted
previously, when compared to parent proxy-reports, study subjects had slightly higher
(i.e., better) mean scores on all aspects of HRQOL, except emotional functioning (Table
4.3 and Figure 4.1). School functioning was rated lowest and social functioning was
rated highest by both subjects and parents. For all HRQOL categories, parent proxy-
ratings resulted in a somewhat greater proportion of individuals meeting the at- risk
103

criterion of being >1SD below the mean than was the case for subject self-rated HRQOL
(Table 4.4). Study subjects’
mean total scores for self-rated
fatigue were slightly lower (i.e.,
they endorsed more symptoms
of fatigue), than the parent
proxy-rated mean total scores
for fatigue (Table 4.6 and Figure
4.2). On subscales within the fatigue scale, both subject self-ratings and parent-proxy
ratings endorsed the fewest symptoms
of general fatigue, followed by
sleep/rest fatigue; they reported the
most symptoms of cognitive fatigue.
Study subjects endorsed fewer
symptoms of general fatigue than did
their parents’ proxy-reports.
To further explore relation ships between patient self-reported and parent proxy-
reported HRQOL and fatigue, intraclass correlations (ICC) and Spearman rank
correlations were computed (see Appendix).

The Spearman rank correlation was used to
determine whether there was a linear relationship between study subjects’ and parents’
scores, while the ICC was used to assess the degree of agreement of the observations
from each patient-parent pair. As shown in Table 4.8, there was a statistically significant
positive linear relationship (Spearman rank correlation) between patient self-report and
parent proxy-report on measures of HRQOL and fatigue; as patient scores increased,
parent scores also increased. However, the ICC results indicate that patient-parent
dyads did not assign the same scores to individual items on the measures. The ICCs
104

range from 0.3 to 0.5, which suggests poor agreement between patient-parent dyads;
agreement is indicated when ICCs are close to 1. Figures 4.3, 4.4, and 4.5 show the
fitted lines for the ICC results for psychosocial health, physical health, and fatigue,
illustrating low intraclass correlation, since fitted lines for highly correlated ICC’s would
fall on or close to the solid gray line (y=x) and the value would be close to 1.

Table 4.8.
HRQOL by Patient and Parent-Proxy Report: PedsQL
TM
Total Score, Physical Health, and
Psychosocial Health
Measure One-way ANOVA Spearman Rank
Correlation
ICC Rho P value

Physical Health^ 0.46 0.26 0.05
Psychosocial Health^ 0.35 0.01
• Emotional Functioning 0.36 0.34 0.01
• Social Functioning 0.40 0.31 0.02
• School Functioning 0.47 0.39 0.002

Fatigue Total Score* 0.52 0.49 0.0001
• General Fatigue 0.42 0.42 0.0007
• Sleep/Rest Fatigue 0.49 0.50 < 0.0001
• Cognitive Fatigue 0.48 0.46 0.0002
^ n=60 pairs; * n=61 pairs

105

Study subjects completed a brief 6-point Likert scale self-report form (Cardiac
Health Form), which was adapted from one used in Stanford’s Pediatric
106

Cardiopulmonary Exercise Laboratory that was also used in a study of subclinical
cardiac toxicity in childhood cancer survivors (Cox et al., 2008). The cardiac symptom
measure was designed for self-rating of fatigue, palpitations, and chest pain in the
context of running, walking, climbing stairs, and school/work. In practice, its matrix
format was found to be confusing for patients as a self-report measure and it remained
difficult to complete when administered by the study team. These procedural difficulties
and the unvalidated nature of the measure raised concern about interpreting results, and
consequently data from this measure were not included in univariate/multivariate
analyses. Study participants did not have difficulty completing the single overarching
PRO item in the measure (“In general, do you
consider your health to be: excellent, very good,
good, fair, or poor?”), and their responses indicated
that the majority of study participants rated their
overall health as very good or excellent (Table 4.9).
Univariate and multivariate analyses were performed with psychosocial HRQOL
summary score as the outcome variable, based on research indicating that it is the
perception of psychosocial well-being that influences physical functioning and treatment
outcomes (Wood, Enriquez, Ghugre, Otto-Duessel, et al., 2005). Results of univariate
analyses, which included relevant demographic and clinical characteristics as predictors,
are shown in Table 4.10. Variables statistically significantly associated with
psychosocial quality of life were physical health (p< 0.0001) and fatigue (p< 0.0001).
Additional variables identified as candidates for inclusion in multivariate analyses were
female sex (p=0.07), tumor resection (p=0.14), and cumulative red blood cell transfusion
volume (cPRBCv) (p=0.12); although they did not meet the criterion for statistical
significance, two variables (age at study evaluation, duration of follow-up) were retained
for inclusion in the multivariate analysis for a priori reasons.
Table 4.9.
Patient-Reported Global Health Score
Score No. (%)
Excellent 26 (38)
Very good 23 (33)
Good 14 (20)
Poor/Fair 6 (9)
Total 69 (100)
107

Table 4.10.
Sample Characteristics and Their Association with Patient-Reported
Psychosocial Health-Related Quality of Life (n=70)
Characteristic Coef (SD) Ref Coef (SD) F-test p-
value
MV
candidate
Sex (%), (Ref: Male)
Female -7.28 (3.93) 79.68 (2.74) 0.07 *
Race/ethnicity (%), (Ref: Hispanic/Latino) 74.713 (2.542)
0.65

Caucasian non-Hispanic 3.46(4.82)
Other 4.55 (6.17)
Diagnosis (%), (Ref: ALL) 77.97 (3.69)
0.72

Acute myeloid leukemia -2.05 (6.74)
Bone tumor 0.70 (5.72)
Soft tissue tumor -4.81 (5.01)
Age at diagnosis (years), continuous 0.35 (0.41) 72.97 (4.26) 0.40
Age at study evaluation (years),
continuous
0.49 (0.42) 69.03 (6.40)
0.25
*
Duration of follow up (years),
continuous^
1.40 (1.20) 70.05 (5.46)
0.25
*
Tumor resection (%), (Ref: No) 64.29 (6.23)
0.14
*
NA 13.07 (6.9)
Yes 13.26 (6.9)
Anthracycline chemotherapy, continuous 0.006 (0.01) 75.26 (2.63) 0.60
Cumulative PRBC transfusion volume,
continuous
-0.04 (0.02) 79.56 (2.93)
0.12
*
Treatment intensity (levels) -0.95 (2.60) 77.70 (4.76) 0.72
Predicted LIC^ -0.27 (0.53) 76.79 (2.44) 0.61
LVEF, continuous 0.50 (0.37) 45.39 (22.75) 0.18
PedsQL Physical Health mean score 0.61 (0.07) 29.15 (5.67) 0.0000 *
PedsQL Fatigue Total mean score 0.81 (0.10) 18.82 (6.61) 0.0000 *
^ n=69
Abbreviations: coef, coefficient; ref coef, reference coefficient; MV, multivariate; ALL, acute lymphoblastic
leukemia

For the multivariate analysis (Table 4.11), predictor variables with p<0.15 via
likelihood ratio test on univariate analysis were included in stepwise model building until
the multivariate LRT p-value was <0.05 for remaining covariates, while retaining age at
diagnosis and duration of follow-up in the model. In the final reduced multivariate model,
higher psychosocial HRQOL was associated with older age at study evaluation
(p=0.0003), better self-perceived physical health (p<0.0001), and fewer symptoms of
fatigue (p<0.0001). This model explained 71% of the variance in LVEF among the
sample of childhood cancer survivors who completed patient-reported outcomes
measures.

108

Table 4.11.
Multiple Regression Analysis of Variables Predicting Psychosocial HRQOL
Predictors Coef (SE) LRT p-value Reference Coef (SD)
Age at evaluation 0.88 (0.24) 0.0003
-2.52 (6.86) PedsQL Physical Health 0.43 (0.08) < 0.0001
Total Fatigue score 0.46 (0.11) < 0.0001
F(3, 64) = 52.56
R-squared = 0.71
F-test p-value < 0.0001

4.5 Discussion
Patient-reported outcomes are increasingly recognized and required as a key
aspect of clinical research. For example, the U.S. Food and Drug Administration has
released guidance concerning the use of PROs in trials to support drug labeling,
including clear language requiring PROs in pediatrics and not parent or clinical proxy
reports (Lipscomb et al., 2007; U.S. Department of Health and Human Services, 2009).
Another example of the current emphasis on PROs is the Patient-Centered Outcomes
Research Institute (PCORI), which was established by legislation in 2010 to fund a
broad range of research studies that seek the patient’s voice in terms of symptom and
HRQOL experiences, measured by validated instruments, so that threats to their
functional status and HRQOL can be adequately addressed.
This study incorporated validated PRO measures of HRQOL and fatigue to be
able to thoughtfully link them with critical biological/clinical observations collected on the
same day, and to acquire data that could provide a useful reference for potential later
assessment points in longitudinal follow up of the CHLA cancer survivor cohort. A total of
70 participants were included in analyses (for fatigue, n=71). On average, at a median
four years after treatment completion – a relatively early point in the survivorship
trajectory – study participants’ rated their overall HRQOL as good. Subjects’ mean self-
reported HRQOL total score, as well as their physical health and psychosocial health
summary scores, fell between the mean scores reported for healthy children and those
reported for children receiving cancer treatment. Perceived general well-being was
109

echoed in participants’ responses to a broad question on a Cardiac Health Form,
wherein the majority rated their overall health as very good or excellent. The finding of
childhood cancer survivors’ predominantly positive appraisal of their status is consistent
with previous reports of equal or better self-reported HRQOL among off-therapy
survivors of non-neurologic malignancies compared to healthy controls and population
norms (Shankar, 2005; Zeltzer et al., 2008), and may be related to the potential
influence of repressive adaptation, response shift, and/or post-traumatic growth from
their cancer experience (Arpawong, 2013; Elkin, Phipps, Mulhern, & Fairclough, 1997;
O'Leary, Diller, & Recklitis, 2007; Phipps & Srivastava, 1997; Phipps, Steele, Hall, &
Leigh, 2001).
While these findings regarding generally good self-appraised overall HRQOL are
reassuring, the range of scores was wide and there were subsets of participants whose
scores were greater than one standard deviation from the mean, a clinically meaningful
indicator for risk of impaired HRQOL (Varni, Burwinkle, Seid, & Skarr, 2003). In this
study 13% to17% of participants were found to be at risk for decrements in various
aspects of HRQOL by self-report (and 15% to 22% were at risk by parent proxy-report).
Individuals at increased risk for diminished HRQOL may benefit from identification of,
and attention to, modifiable risk factors for impaired HRQOL. Smith and colleagues
recently reported that more than one-third of the participants in their study of on-
treatment adolescents and young adult cancer survivors had at least one unmet service
need, and that having unmet service needs was associated with worse overall HRQOL,
fatigue, physical, emotional, social, and school/work functioning, and mental health
(Smith, 2013). In the dissertation study, subjects’ median duration of follow up was
about 4 years, a point in early survivorship when childhood cancer survivors have less
frequent contact with their treatment center, but this is also a time when they are likely to
be navigating developmental and health care transitions that could be facilitated with
110

targeted support. Incorporating regular systematic HRQOL assessment using validated
PRO measures into long-term follow-up visits could help identify survivors at risk for
diminished HRQOL, discover unmet service needs, and facilitate interventions that are
appropriate, available and accessible, including mental health, physical/occupational
therapy, and pain management services. At the same time, there may be childhood
cancer survivors within the upper end of the range who report doing well, but whose
well-being should be validated to be sure they are not overestimating their functioning
and underestimating their need for support and/or information (Engelen et al., 2011), as
well as to better understand the factors that promote positive adaptation.
A positive association between older age at evaluation and higher self-rated
psychosocial HRQOL was seen in both univariate and multivariate analysis. Findings
from previous studies of childhood cancer survivors regarding age and HRQOL have
been inconsistent, owing at least partly to the use of different measures and study
designs (Cantrell, 2007). Prior studies have reported that older age is associated with
having greater perceived control in coping with cancer, and a larger repertoire of coping
strategies; in contrast, older age has also been associated with poorer HRQOL and
higher intensity of symptoms during treatment, possibly because of greater cognitive
ability to comprehend life threat and other health status information, resulting in a net
increase in external stressors (Evan & Zeltzer, 2006). A plausible explanation for the
association between older age and better psychosocial functioning in the dissertation
study may be offered by recent neuroimaging research, which has shown that brain
development continues after puberty and that among the last regions of the brain to
achieve full functional maturation is the prefrontal cortex, an area crucial to self-control
and affect regulation (Giedd et al., 1999)
.
According to Thompson, affect regulation
refers to the extrinsic and intrinsic processes responsible for monitoring, evaluating, and
modifying emotional reactions, especially their intensive and temporal features, to
111

accomplish one’s goals (Thompson, 1994). To the extent that these processes are more
developed in older children/adolescents, they may facilitate adaptation and benefit
finding, resulting in higher self-rated psychosocial HRQOL.
In both univariate and multivariate analyses, there was a statistically significant
positive association between physical health mean scores and psychosocial HRQOL
(p<0.0001). Items in the physical health measure include ratings of how hard it is to walk
more than one block, run, do sports activity or exercise, lift something heavy, take a bath
or shower unaided, and do chores around the house; as well as whether the individual
has aches or pains, or low energy. Study participants’ mean physical health summary
score was higher than that of a pediatric cancer sample that included patients on active
treatment, nearly the same as that of a sample of pediatric asthma patients, but lower
than summary scores from samples of pediatric cardiac and diabetes patients, and
healthy children (Table 4.5). It is of note that the range was wide (12.5 to 100), and that
a subset of nearly 13% of study participants had physical health scores that were greater
than one standard deviation below the mean according to participant self-ratings. To
further investigate factors related to physical health, additional analyses were performed
(data not shown), which indicated that increased risk of lower patient-reported physical
health scores was associated with female sex (p=0.005), and lower left ventricular
ejection fraction (LVEF) (p=0.01). The association between female sex and poorer
physical functioning confirms findings from previous reports (Shankar, 2005; Zeltzer et
al., 2008), although the reason for the association is unclear. It is possible that females
are socialized to have greater ease in endorsing statements that describe problems with
physical functioning and/or that there are other unidentified factors confounding the
association between female sex and diminished self-rated health such as increased
susceptibility to acute and long-term treatment-related toxicities (e.g., female sex is an
established risk factor for cardiomyopathy in childhood cancer survivors). The finding of
112

a statistically significant association between lower LVEF, which reflects impaired
cardiac function, and poorer self-rated physical health is in line with findings from a study
of childhood cancer survivors which reported that systolic function and cardiac risk
factors were linked to lower self-reported health even in the absence of clinically evident
cardiotoxicity (Cox et al., 2008). In the latter study, investigators concluded that health
perceptions, together with treatment history and cardiac risk assessment, may be early
indicators of previously unidentified cardiotoxicity.
There was a statistically significant association between participants’ mean
scores on the fatigue scale and psychosocial HRQOL in univariate and multivariate
analyses. On all fatigue dimensions (general fatigue, sleep/rest fatigue, cognitive
fatigue), study participants self-reported greater fatigue than is reported by healthy
children. Subjects’ lowest fatigue scores were seen on the cognitive fatigue subscale.
Published reports have had conflicting findings about the prevalence of fatigue among
adult and childhood cancer survivors (Botner et al., 1997; Dirksen, Friis, Olesen,
Skovgaard, & Sorensen, 1997; Jorgensen & Sorensen, 1997; K. K. Sorensen, Dahl, &
Smedsrod, 1997; Steenland et al., 1997), but a recent report from the Childhood Cancer
Survivor Study found that neurocognitive function in childhood cancer survivors
appeared to be particularly vulnerable to the effects of fatigue and sleep disruption, and
that the effect of fatigue, sleepiness, sleep quality, and vitality on neurocognitive
outcomes was independent of the effects of cranial radiation, steroids and antimetabolite
chemotherapy, sex, and current age (Clanton et al., 2011). Given the deleterious effects
of fatigue, it is not surprising that dissertation study participants’ lowest mean
psychosocial health subscale scores were on the school functioning subscale since
fatigue in general, and cognitive fatigue in particular, can contribute to poor school
functioning. These findings suggest the potential value of identifying childhood cancer
survivors who have post-treatment fatigue and who might benefit from interventions
113

aimed at (1) adapting activities of daily living to conserve energy, improving sleep
hygiene, and (2) ensuring accommodations in the school setting such as untimed tests,
rest periods, and modified expectations for the amount of time spent each day on
homework.
Exploratory comparisons of patient self-report and parent proxy-report in this
study showed a statistically significant positive linear relationship between patient self-
report and parent proxy report on all aspects of HRQOL and fatigue; that is, patient and
parent scores increased or decreased in the same direction. In general, compared to
the parent proxy-reports, study participants rated their HRQOL slightly higher, but they
endorsed more fatigue symptoms. In addition, parent proxy-ratings of HRQOL and
fatigue showed a perception of a somewhat larger at-risk subset than did participant self-
ratings. Results of the ICC analysis indicated differences between patient-parent dyads
in their ratings of individual items on the measures. Cross informant variance, defined as
discrepancies between self- and proxy-report of HRQOL, has been referenced in the
literature as the “proxy problem” (Eiser & Morse, 2001; Theunissen et al., 1998; Upton,
Lawford, & Eiser, 2008). The magnitude of the proxy problem was investigated in a
recent study of HRQOL reports from adolescents with a variety of chronic conditions and
their parents, which found that the adolescent and parent dyads agreed in 43% to 51%
of the cases, and most disagreement tended to be minor (Sattoe, van Staa, & Moll,
2012). However, even minor differences suggest that self-reports and parent-proxy
reports are not inter-changeable. It is increasingly understood that the child’s perspective
is valid and comprehensive assessment of HRQOL for research and clinical purposes
should include information from both parent and child whenever possible (Eiser & Morse,
2001; Sattoe et al., 2012).
A limitation of the study is its cross-sectional design, which constrains inferences
about causal relationships and cannot provide information about possible changes in
114

HRQOL and fatigue over time. A volunteer bias is recognized as a threat to the
generalizability of the study’s findings because there may have been undetected
differences between participants and nonparticipants. There are many intrinsic and
extrinsic factors, known and unknown, that could affect an individual’s perception/self-
rating of their HRQOL and fatigue, including but not limited to sex, age, socioeconomic
status, race/ethnicity, acculturation, additional symptoms (e.g., pain), and mood state
(e.g., anxiety, depression). Variability between individuals and groups and for the same
individual at different times of measurement is acknowledged. In addition, because
survivors of central nervous system tumors and those who received chest irradiation
were not eligible for the study, survivors with more diminished HRQOL were likely to
have been under-represented in our sample. Also, individuals who could not tolerate the
MRI examination were excluded, which may have inadvertently introduced bias if
patients who were more or less likely to have higher HRQOL scores and/or less fatigue
were more apt to participate. Other recognized threats to generalizability include survivor
bias (i.e., individuals had to have survived to be included in the study), and the exclusion
of children less than seven years old and those individuals who received chest
irradiation. It is not known to what extent the study sample is representative of all
childhood cancer survivors with the same diagnoses and a similar length of follow up
since treatment completion, which limits the generalizability of study findings.
This study is the first to investigate HRQOL, as determined by well-validated
PRO measures, in conjunction with ascertainment of organ iron content and cardiac
function in a cohort of pediatric cancer survivors with various malignancies representing
a spectrum of clinical, demographic, and treatment characteristics. Consistent with
previous reports, study participants generally rated their HRQOL as good, but the finding
that a subset ranging from 13% to 17% of participants was at risk for decrements in
various aspects of HRQOL by self-report (and an even larger proportion was found to be
115

at risk by parent proxy-report) suggests that greater attention is warranted to assess
post-treatment HRQOL in clinical practice. Systematic assessment of HRQOL is not yet
routine in survivorship care, and incorporating regular HRQOL assessment using
validated PRO measures into long-term follow-up visits is recommended, along with
research to determine best practices for administering these measures in childhood
cancer survivor populations (e.g., web-based and tablet-based assessments, mobile
telephone applications), including evaluation of user acceptability and the impact of
systematic assessment and personalized interventions on HRQOL outcomes. The study
finding that older age was a protective factor for psychosocial HRQOL merits further
investigation, since findings from other studies have been inconsistent. Prospective,
longitudinal research is needed to determine whether neurobiological development
affects childhood cancer survivors’ understanding and perception of their cancer
experience over time in ways that affect the likelihood of higher self-ratings of
psychosocial HRQOL.
This study found that perceived better physical health was associated with higher
self-rated psychosocial HRQOL, and that risk factors associated with worse self-rated
physical health were female sex and lower LVEF. These findings suggest that regular
assessment of perceived physical health using validated PRO measures may be a
useful predictor of preclinical disease when combined with recommended surveillance in
childhood cancer survivors at risk for cardiomyopathy. Childhood cancer survivors who
are female and those with lower LVEF values may represent vulnerable individuals who
need careful monitoring because of increased risk for impaired physical health that
ultimately diminishes their HRQOL.
In this study, fatigue was a found to be a risk factor for diminished psychosocial
HRQOL. In addition, participants’ school functioning subscale scores were the lowest
116

among the subscales in the psychosocial health domain. Based on these findings, it is
recommended that a validated measure of fatigue be incorporated into medical follow-up
visits. Systematic assessment could assist in identifying childhood cancer survivors
whose symptoms of fatigue may otherwise be overlooked since survivors themselves
may not volunteer this information, especially if they have acclimated themselves to
fatigue over time. Additional assessment of sleep quality may be particularly useful for
adolescents since even most healthy teens get less sleep than they require for optimal
academic performance, mental and physical health; and inadequate sleep has been
linked to lowered resilience and an increased risk of cardiovascular and metabolic
diseases (George & Davis, 2013). Several valid and reliable sleep assessment tools are
available for use with adolescents, and adolescents with sleep problems not improved
by basic sleep hygiene interventions may benefit from referral to a sleep disorders clinic
(George & Davis, 2013). Early intervention to minimize detrimental effects of fatigue may
avert or minimize problems associated with school performance that may have lifelong
effects on academic and vocational achievement, attainment of life goals, and positive
self-concept.
A statistically significant positive linear relationship was observed between
patient self-report and parent proxy report on all aspects of HRQOL and fatigue in this
study, although there was cross-informant variance in ratings of individual items on study
measures. These findings underscore the importance of eliciting both parent and child
reports of HRQOL and symptoms such as fatigue. In clinical practice, the availability of
child and parent perspectives from valid, easily administered HRQOL and symptom
measures could be the springboard for discussion with a health care professional to
address issues of concern and improve mutual understanding of the impact of cancer
and its treatment. Further study is needed to investigate specific areas of cross
117

informant variance, as well as to determine whether clinical and demographic factors
(such as patients’ age, parents’ educational level, primary language) affect agreement
between participant and parent scores, and whether there are shifts in child-parent
concordance over time.
4.6 Conclusion
Although most study participants rated their psychosocial health as good, a
subset ranging from 13% to 17% may have increased risk for impaired HRQOL.
Protective factors for psychosocial HRQOL included older age, endorsement of fewer
symptoms of fatigue, and better perceived physical health. Risk factors for poorer
physical health were female sex and lower LVEF. Taken together, study findings point
to the importance of systematic assessment of HRQOL, fatigue and other symptoms
using validated PRO measures and obtaining information from both patient and parent
whenever possible during survivorship care, when there is a window of opportunity for
targeted interventions to mitigate adverse psychosocial effects of the cancer experience,
and identify symptoms and risk factors amenable to intervention to improve long-range
HRQOL. As the initial assessment point in longitudinal follow-up of the study sample,
the study provides a baseline for future comparisons that may help to better understand
the dynamic nature of the psychosocial impact of cancer during the first years of post-
treatment survivorship. Longitudinal studies are needed to further explore risk and
protective factors, as well as to identify and evaluate targeted interventions aimed at
minimizing adverse effects of cancer treatment such as fatigue, and maximizing healthy
survivorship. Ultimately, what is learned from research focused on psychosocial
outcomes for survivors of childhood cancer will help drive evidence-based interventions
that support successful transitions into adulthood and long-term survivorship for this
large and growing population.
118

CHAPTER 5: CONCLUSION

Chronic illnesses among long-term survivors, if not prevented or at least managed well,
could…become a national health problem in nations like the United States. It is thus
important to make every effort to ensure that the cured child of today does not survive
only to become the disabled, incapacitated adult of tomorrow. To quote the motto of
pediatric oncology, “Cure is not enough.” -- Guilio D’Angio, MD (D'Angio, 2012)

5.1 Conclusion from three studies
Modern pediatric oncology began 50 years ago when Dr. Sidney Farber and his
team first administered the chemotherapeutic agent aminopterin to children with
leukemia (Mukherjee, 2010, p.32-36). At that time and as survival rates began to
increase, health complications related to treatment were observed, but were generally
accepted as part of the cost of curing diseases that had previously been uniformly fatal.
Myron Karon, MD, one of the pioneers of pediatric oncology, noted that the prevailing
view at the time was that “If someone lived long enough to develop complications, we
would jump for joy and treat the complications” (Karon, 1973). Today, given estimates
that one in 750 20-year-olds in the U.S. is a cancer survivor and given the accumulated
weight of data characterizing late and lingering effects of cancer and its treatment, no
one is sanguine about accepting the “costs of cure” (D'Angio, 2012). Especially in the
past two decades, attention to survivorship has accelerated with numerous publications
about survivors’ health and premature morbidity/mortality (many emanating from the
large federally-funded Childhood Cancer Survivor Study cohort), several Institute of
Medicine reports, special journal issues devoted to survivorship, consensus- and
evidence-based long-term follow-up guidelines, and books about survivorship care for
health care professionals and lay audiences. Survivorship research is essential to
achieve a major goal of pediatric oncology, which is to secure as much as possible a
119

healthy and productive future for the adult who has been cured of cancer as a child
(D'Angio, 2012). Beyond that, much of what is learned through research focused on
childhood cancer survivors may have applicability to survivors of cancers in adulthood –
a group now numbering approximately 11 million. According to another pediatric
oncology pioneer, Anna Meadows, MD, the “recent emphasis on survivorship research
and outcomes-based clinical care began with pediatric oncology” (Meadows, 2006, p.
5160). Childhood cancer survivorship research that may be applicable across the age
spectrum is in keeping with the accomplishments of pediatric cancer research that have
benefited adults, such as discoveries in cancer genetics and multimodal therapeutic
strategies.
The three studies described here addressed several areas of knowledge and
practice that are of high priority in pediatric oncology survivorship care. First, although a
substantial proportion of childhood cancer patients receive cPRBCv in the range known
to be associated with iron overload, there have been no previous systematic studies of a
large cohort with a variety of malignant diagnoses in whom iron burden in heart, liver,
and pancreas is quantitated using MRI technology. Second, despite recognition that
anthracycline chemotherapeutic agents used to treat many childhood cancers are
associated with the development of cardiomyopathy in dose-dependent fashion, the
existence of experimental evidence that anthracylines increase cardiac iron deposition,
and the possibility that tissue-laden iron may moderate damage to myocardial cells
caused by anthracyclines, cardiac function in the context of both anthracycline exposure
and cardiac iron overload has not been investigated. Third, although there is increasing
awareness of the need to ascertain patient-reported outcomes, comparatively few
studies have sought to link PROs with pertinent biological outcomes; without eliciting the
childhood cancer survivor’s voice in terms of symptoms and HRQOL experience, threats
to functional status and HRQOL cannot be adequately addressed.
120

Utilizing a large, established, single institution cohort of childhood cancer
survivors treated at CHLA between June 1, 2004 and December 31, 2009, and
representing a spectrum of diagnoses and ranges of treatment intensity, the three
interdigitated studies of this dissertation sought to (1) determine the prevalence,
distribution, and severity of transfusional iron deposition in the heart, liver, and pancreas;
(2) characterize the prevalence, distribution, and severity of cardiac dysfunction, and (3)
assess childhood cancer survivors’ associated HRQOL and fatigue using validated
patient-reported outcome measures. The studies were performed during a single call-
back visit to CHLA, at which time an MRI was obtained, blood was drawn for analysis,
and PRO measures were administered in the context of an interview with each study
participant and (if available) their parent. Seventy-five participants were enrolled; their
clinical/demographic characteristics are shown in Table 1 in Chapters 2, 3, and 4.

Study 1 focused on iron measurement during MRI evaluation. Key findings included the
following:
• The absence of detectable cardiac iron was a pertinent negative finding.
• Liver iron concentration and pancreas R2* (reflecting iron content) were
abnormal in 36 (58.9%) and 34 (47.2%) of study subjects, respectively.
Statistically significant correlations between liver and pancreatic R2* were found,
and both hepatic and pancreatic R2* were positively correlated with serum iron,
iron binding capacity, serum ferritin, and cPRBCv.
• Predictors of increased liver iron concentration found to be significant in
univariate analyses were diagnosis of AML or solid tumor, older age at
diagnosis/age at MRI, greater treatment intensity, higher cumulative
anthracycline dose, and higher adjusted cPRBCv.
121

• The final reduced multivariate model included statistically significant positive
associations between higher cPRBCv (p<0.0001) and older age at diagnosis
(p<0.0001). This model explained 52% of the variance in LIC among the sample
of 73 childhood cancer survivors who underwent liver MRI examination.

Study 2 focused on assessment of cardiac structure/function by MRI. Key findings
included the following:
• Evidence of clinically significant impaired left ventricular function was not found in
the majority of study subjects. Two males had LVEF ≤50%, a level which would
trigger clinical cardiology assessment and consideration of medical intervention.
• Predictors of lower LVEF found to be significant in univariate analyses were
race/ethnicity (being Caucasian/non-Hispanic or other), diagnosis of AML or solid
tumor, older age at diagnosis/MRI evaluation, greater treatment intensity, higher
cumulative anthracycline dose, use of dexrozoxane, larger body surface area,
taller height at MRI evaluation, higher serum ferritin, and higher serum iron.
• Two reduced multivariate models had nearly identical model statistics, while
suggesting somewhat different implications. In one model, there were significant
associations between decreased LVEF and the variables of higher cumulative
anthracycline dose (p=0.003) and older age at MRI (p=0.009); in the other model
there were statistically significant associations between decreased LVEF and the
variables of higher cumulative anthracycline dose (p=0.005) and taller height at
MRI (p=0.006). The models explained 18% to19% of the variance in LVEF
among the sample of 74 childhood cancer survivors who underwent cardiac MRI
examination.

122

Study 3 focused on assessment of study subjects’ HRQOL and fatigue, using validated
patient-reported outcome measures. Key findings included:
• Most study subjects self-rated their HRQOL as good. Mean HRQOL total scores,
physical health summary scores and psychosocial health summary scores of the
study sample were approximately midway between those of pediatric patients
with cancer in active treatment and those of healthy children. There were
subsets ranging from 13% to 17% of the sample, however, whose scores were
greater than one standard deviation below the mean, indicating at-risk status for
diminished HRQOL.
• Study participants endorsed more fatigue symptoms on every scale than did
healthy children. On subscales within the fatigue measure, subjects reported the
fewest symptoms of general fatigue, followed by sleep-rest fatigue, and they
reported the most symptoms of cognitive fatigue.
• In univariate analyses, predictors of higher self-rated psychosocial HRQOL
included endorsement of fewer symptoms of fatigue and higher self-rated
physical health. Risk factors for lower self-rated physical health were female sex
and lower LVEF.
• In multivariate analysis older age at evaluation (p=0.0003), higher self-rated
physical health (p<0.0001) and fewer symptoms of fatigue (p<0.0001) were
predictive of higher self-rated psychosocial HRQOL.
• A statistically significant positive linear relationship was observed between
patient self-report and parent proxy report on all aspects of HRQOL and fatigue
in this study, although there was cross-informant variance in ratings of individual
items on study measures. Participants self-rated their HRQOL slightly better, but
endorsed more fatigue symptoms than did their parents’ proxy-reports.
123

5.2 Limitations
Limitations of this three-aim study include its cross-sectional design, which
constrains inferences about causal relationships and cannot provide information about
possible changes in variables such as HRQOL and fatigue over time. By the cross-
sectional nature of the study, we do not have pre-MRI values for iron or cardiac
biomarkers, nor do we have longitudinal information to portray trends in the natural
history of study subjects’ biomarkers, which were largely within normal limits at the time
of MRI evaluation, but may have fluctuated during and after cancer treatment. We do not
have longitudinal information to portray trends in organ iron uptake and disposition. Nine
(12%) study subjects received treatment with dexrazoxane, a chelating agent used as a
cardioprotectant during the administration of cardiotoxic chemotherapy, and it is
acknowledged that the extent to which dexrazoxane administration may have affected
organ iron concentration is unknown. However, this agent is not used as a chelator in
the management of iron overload and it is administered only during anthracycline
chemotherapy; thus, it would not be expected to have had much effect on organ iron
concentration (Coates, 2013). It is recognized that the extent to which dexrazoxane
administration may have affected LVEF results is also unknown.
There are other unstudied and therefore undetected factors that may have
affected LVEF results, such as genetic predisposition and lipid profiles. It is recognized
that LVEF in normal subjects is probably age-dependent in the pediatric age group, such
that some of the association of age and height might not reflect vulnerability to
cardiotoxicity, but simply normal physiologic adaptation not adequately captured by
published norms.
A volunteer bias is recognized as a threat to the generalizability of study findings
because there may have been undetected differences between participants and
nonparticipants. There are many intrinsic and extrinsic factors, known and unknown,
124

that could affect an individual’s perception/self-rating of their HRQOL and fatigue,
including but not limited to sex, age, socioeconomic status, race/ethnicity, acculturation,
additional symptoms (e.g., pain), and mood state (e.g., anxiety, depression). Variability
between individuals and groups and for the same individual at different times of
measurement is acknowledged. In addition, because survivors of central nervous
system tumors and those who received chest irradiation were not eligible for the study,
survivors with more diminished HRQOL may have been under-represented in our
sample. Also, individuals who could not tolerate the MRI examination were excluded,
which may have inadvertently introduced bias if patients who were more or less likely to
have organ iron loading, better cardiac function, higher HRQOL scores and/or less
fatigue were more apt to participate. Other recognized threats to generalizability include
survivor bias (i.e., individuals had to have survived to be included in the study), and the
exclusion of children less than seven years old. In addition, it is not known to what
extent the study sample is representative of all childhood cancer survivors with the same
diagnoses and a similar length of follow up since treatment completion, which limits the
generalizability of study findings.
It is acknowledged that there may have been individual observations that exerted
particular influence on the coefficients in the data analysis; while influence diagnostics
were not performed due to time constraints, the one outlier ferritin value was removed
from the analysis and this was believed to have taken care of the most extreme
influence (Sposto, 2013). For the final manuscripts to be submitted for publication, all
analyses will be re-evaluated, including looking at the scale used (log versus natural) for
both the independent and dependent variables, and reviewing some influence statistics.

125

5.3 Strengths
The three related studies presented here are the first to examine the prevalence
of transfusional iron deposition and cardiac dysfunction by MRI, and to evaluate
associated health-related quality of life among a comparatively large sample of
childhood cancer survivors representing a spectrum of clinical, demographic, and
treatment factors.
In Study 1, although the principal organ of interest for iron quantitation was the
heart, we found no evidence of iron loading by cardiac MRI. This pertinent negative
finding is somewhat reassuring, but must be tempered by the realization that further
research is needed to more fully understand the fate of transfusional iron in children
treated for cancer. Open questions include how normal iron homeostasis might be
modified through suppression of erythropoiesis by chemotherapy, inflammation
associated with infections and tissue breakdown, altered nutrition, and circulating factors
arising from malignant cells. Importantly, there is a theoretical possibility that childhood
cancer survivors who receive high cPRBCv are exposed to cardiotoxic circulating NTBI
that may result in organ damage that was not detectable among this sample whose
median time of 4.2 years since treatment completion was relatively early in the
survivorship trajectory.
The findings that fully 58.9% of study subjects had elevated liver iron
concentration and that nearly half (47.2%) had abnormal pancreatic R2* results are
clinically significant because of the related health risks of liver fibrosis/cirrhosis and
hepatocellular carcinoma, as well as glucose intolerance and diabetes mellitus. Higher
cumulative PRBC transfusion volume and older age at diagnosis were risk factors for
elevated LIC. Findings from Study 1 have implications for the development of tailored
monitoring/management guidelines for childhood cancer patients/survivors at risk of
liver/pancreatic damage due to iron overload. For example, systematic tracking of
126

cumulative transfusion volumes – not routine in clinical practice at present – could flag
patients whose management requires higher PRBC volumes and who therefore should
be monitored with serum iron biomarkers (such as ferritin) and MRI evaluation to detect
subclinical iron loading. Staging algorithms to classify patients into risk categories could
be developed collaboratively by the hematology and oncology subspecialties for cost-
effective follow-up and conservation of magnet time. Study findings provide evidence for
adding the clinical problem of iron overload to existing late effects surveillance guidelines
for childhood cancer survivors, and point to the need for exploration of the additive risk
of liver/pancreatic damage from specific chemotherapeutic exposures, as well as for the
design of health education interventions to minimize further liver/pancreatic damage.
This study demonstrated the feasibility of using non-invasive MRI assessment of iron
deposition in several organ systems in one MRI session. Further research is needed to
advance understanding of the biologic mechanisms of iron overload, including
genetic/genomic/epigenetic parameters that may mediate organ iron uptake. In addition,
prospective, longitudinal studies that link assessment of organ iron content to organ
function and to the development of associated morbidity/mortality are needed for better
understanding of risks related to organ iron overload in childhood cancer survivors.
Among the strengths of Study 2 is its demonstration of the feasibility of acquiring
MRI images to assess cardiac structure and function in an imaging session that also
included iron quantitation in the heart, liver, and pancreas. This study is the first to
confirm the association between higher cumulative anthracycline dose and decline in
LVEF through the use of non-invasive cardiac MRI examination in early survivorship
phase. Although there was a low yield of clinically significant abnormal LVEF values in
the study sample, the progressive and asymptomatic nature of anthracycline-induced
cardiomyopathy underscores the need for ongoing clinical monitoring. Further study of
data from a panel of cardiac parameters may contribute to the determination of the most
127

useful and cost-effective tools for such monitoring. In contrast to most previous reports
of anthracycline cardiotoxicity, risk factors associated with declines in LVEF in this study
included older age at diagnosis, and taller height, factors that suggest interplay among
pubertal status, diagnosis, treatment, traditional cardiovascular risk factors, and genetic
polymorphisms that may be fruitful avenues for further research. Although study findings
did not support the hypothesis that higher cPRBCv was associated with declines in
LVEF, a new finding was that increased serum ferritin and serum iron were associated
with declines in LVEF. As noted in regard to Study 1 above, this Study 2 finding
suggests the possibility that study subjects, a significant subset of whom were shown in
Study 1 to have liver and pancreatic iron loading, may have sustained NTBI-induced
damage to cardiac myocytes during critical windows of exposure while receiving cancer
treatment and PRBC transfusions.
Study findings regarding heart-related health risks, as well as the diabetes risk
associated with the increased incidence of liver/pancreatic iron concentrations among
the study sample (Study 1, above), point to the need for health education interventions
aimed at modifiable risk factors to reduce anthracycline-induced and traditional
cardiovascular risks; the health education curriculum should also help ensure that all
childhood cancer survivors have a survivorship care plan and are familiar with the
evidence- and consensus-based guidelines for lifetime monitoring for risks associated
with their specific treatment history (Children's Oncology Group, 2008). Complementary
to the research recommendations stemming from Study 1, future research building on
Study 2 findings should employ prospective, longitudinal designs that begin at the
inception of cancer treatment and that include large sample sizes as well as additional
relevant biological measures such as pubertal status, traditional cardiovascular risk
assessment, and cardiotoxicity-related genetic analyses.
128

Study 3 is the first to investigate HRQOL and fatigue, as determined by well-
validated PRO measures, in conjunction with ascertainment of organ iron content and
cardiac function in a large cohort of pediatric cancer survivors. Study participants
generally rated their HRQOL as good, but the finding that a subset ranging from 13% to
17% of participants was at risk for diminished HRQOL by self-report (and an even larger
proportion was found to be at risk by parent proxy-report) suggests that greater attention
is warranted to assess post-treatment HRQOL in clinical practice. Protective factors for
psychosocial HRQOL included older age, endorsement of fewer symptoms of fatigue,
and higher self-rated physical health. The study finding that older age was a protective
factor for psychosocial HRQOL merits further investigation to explore whether
neurobiological development affects childhood cancer survivors’ understanding and
perception of their cancer experience over time in ways that influence self-rated HRQOL.
Endorsement of more symptoms of fatigue was associated with diminished psychosocial
HRQOL in Study 3. In addition, participants’ mean school functioning subscale scores
were the lowest among the subscales in the psychosocial health domain. These
findings point to the need for early identification of fatigue and timely intervention to
minimize barriers associated with school performance that may have lifelong effects on
academic and vocational achievement, attainment of life goals, and positive self-
concept.
In Study 3, higher self-rated physical health was associated with higher self-rated
psychosocial HRQOL. Risk factors for lower self-rated physical health were female sex
and lower LVEF. These findings, along with findings from Study 2, suggest that regular
assessment of perceived physical health using validated PRO measures of HRQOL may
be a useful predictor of preclinical disease when combined with recommended
surveillance in childhood cancer survivors at risk for cardiomyopathy. Childhood cancer
129

survivors who are female and those with lower LVEF values may represent vulnerable
individuals who need especially careful monitoring because of increased risk for
impaired physical health that also ultimately diminishes their HRQOL.
Study 3 explored correlations between patient self-report and parent proxy-report
of HRQOL and fatigue, and found a statistically significant positive linear relationship
between patient self-report and parent proxy-report. In general, participants’ self-rated
their HRQOL slightly better, but endorsed more fatigue symptoms than did their parents’
proxy-reports. Cross-informant variance was seen in ratings of individual items on study
measures. Further research is needed to analyze specific areas of cross-informant
variance, as well as to determine whether there are clinical or demographic factors that
influence agreement between participant and parent scores, and whether/in what ways
child-parent concordance may shift over time.
Taken together, findings from Study 3 point to the importance of a practice
change in survivorship care: incorporating systematic assessment of HRQOL, fatigue
and other symptoms using validated PRO measures and obtaining information from both
patient and parent whenever possible. Routine assessment when there is a window of
opportunity for targeted interventions to mitigate adverse psychosocial effects of the
cancer experience could contribute to improved long-range HRQOL. As the initial
assessment point in longitudinal follow-up of the CHLA childhood cancer survivor cohort,
Study 3 provides a baseline for future comparisons that may help to better understand
the dynamic nature of the psychosocial impact of cancer during the first years of post-
treatment survivorship.
Ultimately, it is hoped that relevant findings from the three studies described here
will be translated into evidence-based clinical practice, while stimulating further research
to explore health risk and protective factors, as well as to identify and evaluate targeted
130

interventions aimed at minimizing adverse effects of cancer treatment, and maximizing
healthy survivorship. The significance of these studies to individual childhood cancer
survivors is summarized in the following concluding statement by Tavo Zavala, a long-
term survivor of childhood acute lymphoblastic leukemia:

For many of us diagnosed with childhood cancer, being a "survivor" means much more
than being alive and disease free, it means emerging from our cancer experience as a
whole person and not allowing it to consume our identity and existence. Many of the
childhood cancer survivors I've encountered agree that finding meaning in our cancer
experience and learning about our long-term medical needs are among the most
important steps in the process of emerging from cancer as a whole person. Research in
late effects of childhood cancer treatments is critical because it establishes the
knowledge base we survivors need to address our long-term medical needs and
maintain our health. By participating in research as subjects, such as was the case in the
“Ironman” (dissertation) study, childhood cancer survivors can find meaning in their
experience by contributing to knowledge that will benefit many people.
– Tavo Zavala (Zavala, 2013)
131

REFERENCES

Abosoudah, I., Greenberg, M. L., Ness, K. K., Benson, L., & Nathan, P. C. (2011).
Echocardiographic surveillance for asymptomatic late-onset anthracycline
cardiomyopathy in childhood cancer survivors. Pediatr Blood Cancer, 57(3), 467-
472. doi: 10.1002/pbc.22989
Adams, M. J., & Lipshultz, S. E. (2005). Pathophysiology of anthracycline- and radiation-
associated cardiomyopathies: implications for screening and prevention. Pediatr
Blood Cancer, 44(7), 600-606. doi: 10.1002/pbc.20352
Adams, M.J., Duffy, S.A., Constine, L.S., & Lipshultz, S.E. (2005). Cardiovascular effects
of cancer therapy. In C. L. Schwartz, Hobbie, W.L., Constine, L.S. & Ruccione,
K.S. (Ed.), Survivors of Childhood and Adolescent Cancer: A Multidisciplinary
Approach (pp. 133-159). Berlin: Springer.
Adamson, P.C., Bagatell, R., Balis, F.M., Blaney, S.M. (2011). General principles of
chemotherapy. In D. G. P. Pizzo, P.A. (Ed.), Principles and Practice of Pediatric
Oncology (6th ed., pp. 279-355). Philadelphia: Lippincott, Williams & Wilkins.
Agrawal, A.K., Hastings, C.A., Feusner, J. (2011). Hematologic supportive care for
children with cancer. In D. G. P. Pizzo, P.A. (Ed.), Principles and Practice of
Pediatric Oncology (6th ed., pp. 1152-1189). Philadelphia: Lippincott, Williams &
Wilkins.
Anderson, B. (2005). Dexrazoxane for the prevention of cardiomyopathy in anthracycline
treated pediatric cancer patients. Pediatr Blood Cancer, 44(7), 584-588. doi:
10.1002/pbc.20358
Anderson, Gregory J., McLaren, Gordon D., Skoien, Richard, & Powell, LawrieW.
(2011). HFE-Associated Hereditary Hemochromatosis Iron Physiology and
Pathophysiology in Humans (pp. 385-398): Humana Press.
Andrews, N. C. (1999). Disorders of iron metabolism. N Engl J Med, 341(26), 1986-
1995.
Andrews, N. C. (2008). Forging a field: the golden age of iron biology. Blood, 112(2),
219-230. doi: 112/2/219 [pii] 10.1182/blood-2007-12-077388
Andrews, N. C. (2009). Iron deficiency and related disorders. In J. P. Greer, Foerster, J.,
Rodgers, G.M., Paraskevas, F., Glader, B., Arber, D.A., Means, R.T., Jr. (Ed.),
Wintrobe's Clinical Hematology (12th ed., Vol. 1, pp. 810-835). Philadelphia, PA:
Lippincott, Williams & Wilkins.
Andrews, N. C., & Schmidt, P. J. (2007). Iron homeostasis. Annu.Rev.Physiol, 69, 69-85.
Angelucci, E., Brittenham, G. M., McLaren, C. E., Ripalti, M., Baronciani, D., Giardini, C.,
. . . Lucarelli, G. (2000). Hepatic iron concentration and total body iron stores in
thalassemia major. N Engl J Med, 343(5), 327-331. doi:
10.1056/NEJM200008033430503
Armstrong, G. T., Plana, J. C., Zhang, N., Srivastava, D., Green, D. M., Ness, K. K., . . .
Flamm, S. D. (2012). Screening adult survivors of childhood cancer for
cardiomyopathy: comparison of echocardiography and cardiac magnetic
resonance imaging. J Clin Oncol, 30(23), 2876-2884. doi: JCO.2011.40.3584 [pii]
10.1200/JCO.2011.40.3584
Arpawong, T.E., Oland, A., Milam, J.E., Ruccione, K., Meeske, K. (2013). Post-traumatic
growth among an ethnically diverse sample of adolescent and young adult
cancer survivors. Psycho-Oncology. doi: 10.1002/pon.3286
Bacon, B. R., Adams, P. C., Kowdley, K. V., Powell, L. W., & Tavill, A. S. (2011).
Diagnosis and management of hemochromatosis: 2011 practice guideline by the
132

American Association for the Study of Liver Diseases. Hepatology, 54(1), 328-
343. doi: 10.1002/hep.24330
Barnard, D.R., Rogers, Z.R. (2004). Blood component therapy. In A. J. Altman (Ed.),
Supportive Care of Children with Cancer: Current Therapy and Guidelines from
the Children's Oncology Group (3rd ed., pp. 39-57). Baltimore: The Johns
Hopkins University Press.
Barry, E., Alvarez, J. A., Scully, R. E., Miller, T. L., & Lipshultz, S. E. (2007).
Anthracycline-induced cardiotoxicity: course, pathophysiology, prevention and
management. Expert Opin Pharmacother, 8(8), 1039-1058. doi:
10.1517/14656566.8.8.1039
Benetatos, L., Alymara, V., Vassou, A., & Bourantas, K. L. (2008). Malignancies in beta-
thalassemia patients: a single-center experience and a concise review of the
literature. Int.J Lab Hematol., 30(2), 167-172.
Berry, G. J., & Jorden, M. (2005). Pathology of radiation and anthracycline cardiotoxicity.
Pediatr Blood Cancer, 44(7), 630-637. doi: 10.1002/pbc.20346
Blair, A.B., Tarango, C., Bowers, D., Leavey, P., Journeycake, J. (2009). Transfusional
iron overload in acute lymphoblastic leukemia. Paper presented at the American
Society of Pediatric Hematology Oncology, San Diego, CA.
Blanco, J. G., Sun, C. L., Landier, W., Chen, L., Esparza-Duran, D., Leisenring, W., . . .
Bhatia, S. (2011). Anthracycline-related cardiomyopathy after childhood cancer:
role of polymorphisms in carbonyl reductase genes--a report from the Children's
Oncology Group. J Clin Oncol, 30(13), 1415-1421. doi: JCO.2011.34.8987 [pii]
10.1200/JCO.2011.34.8987
Bonkovsky, H. L. (1991). Iron and the liver. Am J Med Sci, 301(1), 32-43.
Borgna-Pignatti, C., Rugolotto, S., De Stefano, P., Zhao, H., Cappellini, M. D., Del
Vecchio, G. C., . . . Cnaan, A. (2004). Survival and complications in patients with
thalassemia major treated with transfusion and deferoxamine. Haematologica,
89(10), 1187-1193.
Botner, A., Strandbygaard, B., Sorensen, K. J., Have, P., Madsen, K. G., Madsen, E. S.,
& Alexandersen, S. (1997). Appearance of acute PRRS-like symptoms in sow
herds after vaccination with a modified live PRRS vaccine. Vet Rec, 141(19),
497-499.
Braunwald, E. (2008). Biomarkers in heart failure. N Engl J Med, 358(20), 2148-2159.
doi: 358/20/2148 [pii] 10.1056/NEJMra0800239
Brittenham, G. M. (2011). Iron-chelating therapy for transfusional iron overload. N Engl J
Med, 364(2), 146-156. doi: 10.1056/NEJMct1004810
Brittenham, G. M., Sheth, S., Allen, C. J., & Farrell, D. E. (2001). Noninvasive methods
for quantitative assessment of transfusional iron overload in sickle cell disease.
Semin Hematol, 38(1 Suppl 1), 37-56.
Bryant, J., Picot, J., Levitt, G., Sullivan, I., Baxter, L., & Clegg, A. (2007).
Cardioprotection against the toxic effects of anthracyclines given to children with
cancer: a systematic review. Health Technol Assess, 11(27), iii, ix-x, 1-84. doi:
05/34/01 [pii]
Buchanon, G.R. (2005). Commentary Blood transfusions in children with cancer and
hematologic disorders: Why, when, and how? Pediatr Blood Cancer, 44, 114-
116.
Cain, P. A., Ahl, R., Hedstrom, E., Ugander, M., Allansdotter-Johnsson, A., Friberg, P., &
Arheden, H. (2009). Age and gender specific normal values of left ventricular
mass, volume and function for gradient echo magnetic resonance imaging: a
cross sectional study. BMC Med Imaging, 9, 2. doi: 10.1186/1471-2342-9-2
133

Cantrell, M.A. (2007). Health-related quality of life in childhood cancer: State of the
science. Oncol Nurs Forum, 34(1), 103-111.
Cascales, A., Sanchez-Vega, B., Navarro, N., Pastor-Quirante, F., Corral, J., Vicente, V.,
& de la Pena, F. A. (2010). Clinical and genetic determinants of anthracycline-
induced cardiac iron accumulation. Int J Cardiol. doi: S0167-5273(10)00759-X
[pii] 10.1016/j.ijcard.2010.09.046
Chatterjee, R., & Katz, M. (2000). Reversible hypogonadotrophic hypogonadism in
sexually infantile male thalassaemic patients with transfusional iron overload.
Clin Endocrinol (Oxf), 53(1), 33-42. doi: cen962 [pii]
Children's Oncology Group. (2008). Long-term Follow-up Guidelines for Survivors of
Childhood, Adolescent, and Young Adult Cancers Version 3.0 Retrieved from
www.survivorshipguidelines.org
Clanton, N. R., Klosky, J. L., Li, C., Jain, N., Srivastava, D. K., Mulrooney, D., . . . Krull,
K. R. (2011). Fatigue, vitality, sleep, and neurocognitive functioning in adult
survivors of childhood Cancer: A report from the childhood cancer survivor study.
Cancer. doi: 10.1002/cncr.25797
Coates, T. D. (2013, April 29. 2013).
Cohn, J. N., Ferrari, R., & Sharpe, N. (2000). Cardiac remodeling--concepts and clinical
implications: a consensus paper from an international forum on cardiac
remodeling. Behalf of an International Forum on Cardiac Remodeling. J Am Coll
Cardiol, 35(3), 569-582. doi: S0735109799006300 [pii]
Cook, L. S. (2010). Therapeutic phlebotomy: a review of diagnoses and treatment
considerations. J Infus Nurs, 33(2), 81-88. doi: 10.1097/NAN.0b013e3181d00010
00129804-201003000-00003 [pii]
Costin, G., Kogut, M. D., Hyman, C. B., & Ortega, J. A. (1979). Endocrine abnormalities
in thalassemia major. Am J Dis Child, 133(5), 497-502.
Cox, C. L. (2003). Online exclusive: a model of health behavior to guide studies of
childhood cancer survivors. Oncol Nurs Forum, 30(5), E92-99. doi:
10.1188/03.ONF.E92-E99
Cox, C. L., Rai, S. N., Rosenthal, D., Phipps, S., & Hudson, M. M. (2008). Subclinical
late cardiac toxicity in childhood cancer survivors: impact on self-reported health.
Cancer, 112(8), 1835-1844. doi: 10.1002/cncr.23378
D'Angio, G. (2012). Foreword. In N. Keene, Hobbie, W., Ruccione, K. (Ed.), Childhood
Cancer Survivors: A Practical Guide to Your Future (Third ed., pp. xiii-xv).
Bellingham, WA: Childhood Cancer Guides.
Dirksen, A., Friis, M., Olesen, K. P., Skovgaard, L. T., & Sorensen, K. (1997). Progress
of emphysema in severe alpha 1-antitrypsin deficiency as assessed by annual
CT. Acta Radiol, 38(5), 826-832.
Eiser, C., & Morse, R. (2001). Can parents rate their child's health-related quality of life?
Results of a systematic review. Qual Life Res, 10(4), 347-357.
Elkin, T. D., Phipps, S., Mulhern, R. K., & Fairclough, D. (1997). Psychological
functioning of adolescent and young adult survivors of pediatric malignancy. Med
Pediatr Oncol, 29(6), 582-588.
Encyclopaedia Britannica. (2013). Heart. Retrieved from Encyclopaedia Britannica
Online Academic Edition website:
http://www.britannica.com/EBchecked/topic/258344/heart
Eng, J., & Fish, J. D. (2011). Insidious iron burden in pediatric patients with acute
lymphoblastic leukemia. Pediatr Blood Cancer, 56(3), 368-371. doi:
10.1002/pbc.22851
Engelen, V., Koopman, H. M., Detmar, S. B., Raat, H., van de Wetering, M. D., Brons,
P., . . . Grootenhuis, M. A. (2011). Health-related quality of life after completion of
134

successful treatment for childhood cancer. Pediatr Blood Cancer, 56(4), 646-653.
doi: 10.1002/pbc.22795
Evan, E. E., & Zeltzer, L. K. (2006). Psychosocial dimensions of cancer in adolescents
and young adults. Cancer, 107(7 Suppl), 1663-1671. doi: 10.1002/cncr.22107
Fischer, R., & Harmatz, P. R. (2009). Non-invasive assessment of tissue iron overload.
Hematology Am Soc Hematol Educ Program, 215-221. doi: 2009/1/215 [pii]
10.1182/asheducation-2009.1.215
Fleming, R. E., & Ponka, P. (2012). Iron overload in human disease. N Engl J Med,
366(4), 348-359. doi: 10.1056/NEJMra1004967
Ganz, T., & Nemeth, E. (2011a). The hepcidin-ferroportin system as a therapeutic target
in anemias and iron overload disorders. Hematology Am Soc Hematol Educ
Program, 2011, 538-542. doi: 2011/1/538 [pii] 10.1182/asheducation-2011.1.538
Ganz, T., & Nemeth, E. (2011b). Hepcidin and disorders of iron metabolism. Annu Rev
Med, 62, 347-360. doi: 10.1146/annurev-med-050109-142444
Gattermann, N., & Rachmilewitz, E. A. (2011). Iron overload in MDS-pathophysiology,
diagnosis, and complications. Ann Hematol, 90(1), 1-10. doi: 10.1007/s00277-
010-1091-1
George, N. M., & Davis, J. E. (2013). Assessing Sleep in Adolescents Through a Better
Understanding of Sleep Physiology. Am J Nurs. doi:
10.1097/01.NAJ.0000430921.99915.24
Gianni, L., Herman, E. H., Lipshultz, S. E., Minotti, G., Sarvazyan, N., & Sawyer, D. B.
(2008). Anthracycline cardiotoxicity: from bench to bedside. J Clin Oncol, 26(22),
3777-3784. doi: 26/22/3777 [pii] 10.1200/JCO.2007.14.9401
Giedd, J. N., Blumenthal, J., Jeffries, N. O., Castellanos, F. X., Liu, H., Zijdenbos, A., . . .
Rapoport, J. L. (1999). Brain development during childhood and adolescence: a
longitudinal MRI study. Nat Neurosci, 2(10), 861-863. doi: 10.1038/13158
Gujja, P., Rosing, D. R., Tripodi, D. J., & Shizukuda, Y. (2010). Iron overload
cardiomyopathy: better understanding of an increasing disorder. J Am Coll
Cardiol, 56(13), 1001-1012. doi: S0735-1097(10)02572-6 [pii]
10.1016/j.jacc.2010.03.083
Gurram, M. K., Newman, W., & Kobrinsky, N. (2012). Prevalence of iron overload in
pediatric oncology patients after blood transfusion. Clin Adv Hematol Oncol,
10(6), 363-365.
Haase, J. E., Heiney, S. P., Ruccione, K. S., & Stutzer, C. (1999). Research triangulation
to derive meaning-based quality-of-life theory: adolescent resilience model and
instrument development. Int J Cancer Suppl, 12, 125-131. doi:
10.1002/(SICI)1097-0215(1999)83:12+<125::AID-IJC22>3.0.CO;2-7 [pii]
Halonen, P., Mattila, J., Suominen, P., Ruuska, T., Salo, M. K., & Makipernaa, A. (2003).
Iron overload in children who are treated for acute lymphoblastic leukemia
estimated by liver siderosis and serum iron parameters. Pediatrics, 111(1), 91-
96.
Hershko, C., Pinson, A., & Link, G. (1996). Prevention of anthracycline cardiotoxicity by
iron chelation. Acta Haematol, 95(1), 87-92.
Hinds, P. S. (2010). Progress in quality of life in children and adolescents with cancer.
Semin Oncol Nurs, 26(1), 18-25. doi: S0749-2081(09)00111-9 [pii]
10.1016/j.soncn.2009.11.004
Hinds, P. S., Burghen, E. A., Haase, J. E., & Phillips, C. R. (2006). Advances in defining,
conceptualizing, and measuring quality of life in pediatric patients with cancer.
Oncol Nurs Forum, 33(1 Suppl), 23-29. doi: 5255043272405122 [pii]
10.1188/06.ONF.S1.23-29
135

Hinds, P. S., Gattuso, J. S., Fletcher, A., Baker, E., Coleman, B., Jackson, T., . . . Pui, C.
H. (2004). Quality of life as conveyed by pediatric patients with cancer. Qual Life
Res, 13(4), 761-772.
Hinds, P. S., & Martin, J. (1988). Hopefulness and the self-sustaining process in
adolescents with cancer. Nurs Res, 37(6), 336-340.
Hinds, P. S., Oakes, L. L., Hicks, J., & Anghelescu, D. L. (2005). End-of-life care for
children and adolescents. Semin Oncol Nurs, 21(1), 53-62.
Horacek, J. M., Tichy, M., Pudil, R., Jebavy, L., Zak, P., Ulrychova, M., . . . Maly, J.
(2008). Multimarker approach to evaluation of cardiac toxicity during preparative
regimen and hematopoietic cell transplantation. Neoplasma, 55(6), 532-537.
Howlader, N., Noone, A.M., Krapcho, M., Neyman, N., Aminou, R., Altekruse, S.F.,
Kosary, C.L., Ruhl, J., Tatalovich, Z., Cho, H., Mariotto, A., Eisner, M.P., Lewis,
D.R., Chen, H.S., Feuer, E.J., Cronin, K.A. (eds). . (2012). SEER Cancer
Statistics Review, 1975-2009 (Vintage 2009 Populations), based on November
2011 SEER data submission. Bethesda, MD: National Cancer Institute Retrieved
from http://seer.cancer.gov/csr/1975_2009_pops09/.
Huang, X. (2003). Iron overload and its association with cancer risk in humans: evidence
for iron as a carcinogenic metal. Mutat Res, 533(1-2), 153-171. doi:
S0027510703002148 [pii]
Jaffer, A., Barty, R., Jamula, E., Wang, G., Liu, Y., Heddle, N., Athale, U. (2012). Are
children with cancer receiving myelosuppressive therapy at risk for transfusion-
related iron overload? Paper presented at the 54th ASH Annual Meeting and
Exposition, Atlanta, GA.
https://ash.confex/ash/2012/webprogram/Paper53551.html
Jastaniah, W., Harmatz, P., Pakbaz, Z, Fischer, R., Vichinsky, E., Walters, M.C. (2008).
Transfusional iron burden and liver toxicity after bone marrow transplantation for
acute myelogenous leukemia and hemoglobinopathies. Pediatr Blood Cancer,
50, 319-324.
Jorgensen, P. L., & Sorensen, K. (1997). [Education--in disability evaluation as nurse
instructor]. Sygeplejersken, 97(50), 28-30.
Kaiserova, H., Simunek, T., Sterba, M., den Hartog, G. J., Schroterova, L., Popelova, O.,
. . . Bast, A. (2007). New iron chelators in anthracycline-induced cardiotoxicity.
Cardiovasc Toxicol, 7(2), 145-150. doi: CT:7:2:145 [pii] 10.1007/s12012-007-
0020-6
Karon, M. (1973). Problems in the evaluation of long-term results. In G. Mathe, Pouillart,
P., Schwarzenberg, L. (Ed.), Recent Results in Cancer Research (pp. 113-120).
New York: Springer-Verlag.
Kazak, A. E., Alderfer, M., Rourke, M. T., Simms, S., Streisand, R., & Grossman, J. R.
(2004). Posttraumatic stress disorder (PTSD) and posttraumatic stress
symptoms (PTSS) in families of adolescent childhood cancer survivors. J Pediatr
Psychol, 29(3), 211-219.
Kirk, P., Roughton, M., Porter, J. B., Walker, J. M., Tanner, M. A., Patel, J., . . . Pennell,
D. J. (2009). Cardiac T2* magnetic resonance for prediction of cardiac
complications in thalassemia major. Circulation, 120(20), 1961-1968. doi:
CIRCULATIONAHA.109.874487 [pii] 10.1161/CIRCULATIONAHA.109.874487
Kletzky, O. A., Costin, G., Marrs, R. P., Bernstein, G., March, C. M., & Mishell, D. R., Jr.
(1979). Gonadotropin insufficiency in patients with thalassemia major. J Clin
Endocrinol Metab, 48(6), 901-905.
Kohgo, Y., Ikuta, K., Ohtake, T., Torimoto, Y., & Kato, J. (2008). Body iron metabolism
and pathophysiology of iron overload. Int J Hematol, 88(1), 7-15. doi:
10.1007/s12185-008-0120-5
136

Kremer, L. C., & Caron, H. N. (2004). Anthracycline cardiotoxicity in children. N Engl J
Med, 351(2), 120-121. doi: 10.1056/NEJMp048113 351/2/120 [pii]
Kremer, L. C., van der Pal, H. J., Offringa, M., van Dalen, E. C., & Voute, P. A. (2002).
Frequency and risk factors of subclinical cardiotoxicity after anthracycline therapy
in children: a systematic review. Ann Oncol, 13(6), 819-829.
Krischer, J.P., Epstein, S., Cuthbertson, D.D., Goorin, A.M., Epstein, M.L., Lipshultz,
S.E. (1997). Clinical cardiotoxicity following anthracycline treatment for childhood
cancer: the Pediatric Oncology Group experience. Journal of Clinical Oncology,
15, 1544-1552.
Kwok, J. C., & Richardson, D. R. (2003). Anthracyclines induce accumulation of iron in
ferritin in myocardial and neoplastic cells: inhibition of the ferritin iron mobilization
pathway. Mol Pharmacol, 63(4), 849-861.
Kwok, J. C., & Richardson, D. R. (2004). Examination of the mechanism(s) involved in
doxorubicin-mediated iron accumulation in ferritin: studies using metabolic
inhibitors, protein synthesis inhibitors, and lysosomotropic agents. Mol
Pharmacol, 65(1), 181-195. doi: 10.1124/mol.65.1.181 65/1/181 [pii]
Lauret, E., Rodriguez, M., Gonzalez, S., Linares, A., Lopez-Vazquez, A., Martinez-Borra,
J., . . . Lopez-Larrea, C. (2002). HFE gene mutations in alcoholic and virus-
related cirrhotic patients with hepatocellular carcinoma. Am J Gastroenterol,
97(4), 1016-1021. doi: 10.1111/j.1572-0241.2002.05553.x
Link, G., Tirosh, R., Pinson, A., & Hershko, C. (1996). Role of iron in the potentiation of
anthracycline cardiotoxicity: identification of heart cell mitochondria as a major
site of iron-anthracycline interaction. J Lab Clin Med, 127(3), 272-278. doi:
S0022-2143(96)90095-5 [pii]
Lipscomb, J., Gotay, C. C., & Snyder, C. F. (2007). Patient-reported outcomes in cancer:
a review of recent research and policy initiatives. CA Cancer J Clin, 57(5), 278-
300. doi: 57/5/278 [pii] 10.3322/CA.57.5.278
Lipshultz, S. E., & Adams, M. J. (2010). Cardiotoxicity after childhood cancer: beginning
with the end in mind. J Clin Oncol, 28(8), 1276-1281. doi: JCO.2009.26.5751 [pii]
10.1200/JCO.2009.26.5751
Lipshultz, S. E., Alvarez, J. A., & Scully, R. E. (2008). Anthracycline associated
cardiotoxicity in survivors of childhood cancer. Heart, 94(4), 525-533. doi:
94/4/525 [pii] 10.1136/hrt.2007.136093
Lipshultz, S. E., & Colan, S. D. (2004). Cardiovascular trials in long-term survivors of
childhood cancer. J Clin Oncol, 22(5), 769-773. doi: 10.1200/JCO.2004.12.937
JCO.2004.12.937 [pii]
Lipshultz, S. E., Colan, S. D., Gelber, R. D., Perez-Atayde, A. R., Sallan, S. E., &
Sanders, S. P. (1991). Late cardiac effects of doxorubicin therapy for acute
lymphoblastic leukemia in childhood. N Engl J Med, 324(12), 808-815. doi:
10.1056/NEJM199103213241205
Lipshultz, S. E., Lipsitz, S. R., Sallan, S. E., Dalton, V. M., Mone, S. M., Gelber, R. D., &
Colan, S. D. (2005). Chronic progressive cardiac dysfunction years after
doxorubicin therapy for childhood acute lymphoblastic leukemia. J Clin Oncol,
23(12), 2629-2636. doi: 10.1200/JCO.2005.12.121
Lipshultz, S. E., Miller, T. L., Scully, R. E., Lipsitz, S. R., Rifai, N., Silverman, L. B., . . .
Sallan, S. E. (2012). Changes in cardiac biomarkers during doxorubicin treatment
of pediatric patients with high-risk acute lymphoblastic leukemia: associations
with long-term echocardiographic outcomes. J Clin Oncol, 30(10), 1042-1049.
doi: JCO.2010.30.3404 [pii] 10.1200/JCO.2010.30.3404
Lipshultz, S. E., Scully, R. E., Lipsitz, S. R., Sallan, S. E., Silverman, L. B., Miller, T. L., .
. . Colan, S. D. (2010). Assessment of dexrazoxane as a cardioprotectant in
137

doxorubicin-treated children with high-risk acute lymphoblastic leukaemia: long-
term follow-up of a prospective, randomised, multicentre trial. Lancet Oncol,
11(10), 950-961. doi: S1470-2045(10)70204-7 [pii] 10.1016/S1470-
2045(10)70204-7
Lutz, K., von Komorowski, G., Durken, M., Engelhardt, R., & Dinter, D. J. (2008).
Myocardial iron overload in transfusion-dependent pediatric patients with acute
leukemia. Pediatr Blood Cancer, 51(5), 691-693. doi: 10.1002/pbc.21663
MacKenzie, E. L., Iwasaki, K., & Tsuji, Y. (2008). Intracellular iron transport and storage:
from molecular mechanisms to health implications. Antioxid Redox Signal, 10(6),
997-1030. doi: 10.1089/ars.2007.1893
Mayo Clinic. (2011a). 5 Medication-free strategies to help prevent heart disease. Heart
disease. http://www.mayoclinic.com/health/heart-disease-prevention/WO00041
Mayo Clinic. (2011b). Liver problems prevention. Retrieved from Mayo Clinic website:
http://www.mayoclinic.com/health/liver-
problems/DS01133/DSECTION=prevention
Meadows, A. T. (2006). Pediatric cancer survivorship: research and clinical care. J Clin
Oncol, 24(32), 5160-5165. doi: 10.1200/JCO.2006.07.3114
Mertens, A. C., Liu, Q., Neglia, J. P., Wasilewski, K., Leisenring, W., Armstrong, G. T., . .
. Yasui, Y. (2008). Cause-specific late mortality among 5-year survivors of
childhood cancer: the Childhood Cancer Survivor Study. J Natl Cancer Inst,
100(19), 1368-1379. doi: djn310 [pii] 10.1093/jnci/djn310
Mertens, A. C., Yasui, Y., Neglia, J. P., Potter, J. D., Nesbit, M. E., Jr., Ruccione, K., . . .
Robison, L. L. (2001). Late mortality experience in five-year survivors of
childhood and adolescent cancer: the Childhood Cancer Survivor Study. J Clin
Oncol, 19(13), 3163-3172.
Meyer, R. A. (2004). Aerobic performance and the function of myoglobin in human
skeletal muscle. Am J Physiol Regul Integr Comp Physiol, 287(6), R1304-1305.
doi: 287/6/R1304 [pii] 10.1152/ajpregu.00541.2004
Minotti, G., Cairo, G., & Monti, E. (1999). Role of iron in anthracycline cardiotoxicity: new
tunes for an old song? FASEB J, 13(2), 199-212.
Minotti, G., Ronchi, R., Salvatorelli, E., Menna, P., & Cairo, G. (2001). Doxorubicin
irreversibly inactivates iron regulatory proteins 1 and 2 in cardiomyocytes:
evidence for distinct metabolic pathways and implications for iron-mediated
cardiotoxicity of antitumor therapy. Cancer Res, 61(23), 8422-8428.
Miranda, C. J., Makui, H., Soares, R. J., Bilodeau, M., Mui, J., Vali, H., . . . Santos, M. M.
(2003). Hfe deficiency increases susceptibility to cardiotoxicity and exacerbates
changes in iron metabolism induced by doxorubicin. Blood, 102(7), 2574-2580.
doi: 10.1182/blood-2003-03-0869 2003-03-0869 [pii]
Mukherjee, S. (2010). The Emperor of All Maladies: A Biography of Cancer. New York:
Scribner.
Mulrooney, D. A., Yeazel, M. W., Kawashima, T., Mertens, A. C., Mitby, P., Stovall, M., .
. . Leisenring, W. M. (2009). Cardiac outcomes in a cohort of adult survivors of
childhood and adolescent cancer: retrospective analysis of the Childhood Cancer
Survivor Study cohort. BMJ, 339, b4606.
Myers, C. E., Gianni, L., Simone, C. B., Klecker, R., & Greene, R. (1982). Oxidative
destruction of erythrocyte ghost membranes catalyzed by the doxorubicin-iron
complex. Biochemistry, 21(8), 1707-1712.
Nelson, S.C., Bendel, A.E. & Perkins, J.L. (2007). Transfusional Iron Overload in
Pediatric Oncology. Paper presented at the American Society of Pediatric
Hematology/Oncology Abstracts, Toronto, Canada.
138

Noetzli, L. J., Mittelman, S. D., Watanabe, R. M., Coates, T. D., & Wood, J. C. (2011).
Pancreatic iron and glucose dysregulation in thalassemia major. Am J Hematol.
doi: 10.1002/ajh.22223
Noetzli, L. J., Panigrahy, A., Hyderi, A., Dongelyan, A., Coates, T. D., & Wood, J. C.
(2012). Pituitary Iron and Volume Imaging in Healthy Controls. AJNR Am J
Neuroradiol. doi: ajnr.A2788 [pii] 10.3174/ajnr.A2788
Noetzli, L. J., Papudesi, J., Coates, T. D., & Wood, J. C. (2009). Pancreatic iron loading
predicts cardiac iron loading in thalassemia major. Blood, 114(19), 4021-4026.
doi: blood-2009-06-225615 [pii] 10.1182/blood-2009-06-225615
Nottage, K., Gurney, J. G., Smeltzer, M., Castellanos, M., Hudson, M. M., & Hankins, J.
S. (2012). Trends in transfusion burden among long-term survivors of childhood
hematological malignancies. Leuk Lymphoma. doi:
10.3109/10428194.2012.750724
Nuss, S. L., Hinds, P. S., & LaFond, D. A. (2005). Collaborative clinical research on end-
of-life care in pediatric oncology. Semin Oncol Nurs, 21(2), 125-134; discussion
134-144.
O'Leary, T. E., Diller, L., & Recklitis, C. J. (2007). The effects of response bias on self-
reported quality of life among childhood cancer survivors. Qual Life Res, 16(7),
1211-1220. doi: 10.1007/s11136-007-9231-3
Oeffinger, K. C., Mertens, A. C., Sklar, C. A., Kawashima, T., Hudson, M. M., Meadows,
A. T., . . . Robison, L. L. (2006). Chronic health conditions in adult survivors of
childhood cancer. N Engl J Med, 355(15), 1572-1582. doi: 355/15/1572 [pii]
10.1056/NEJMsa060185
Orgel, E., Zung, L., Ji, L., Finklestein, J., Feusner, J., & Freyer, D. R. (2013). Early
Cardiac Outcomes Following Contemporary Treatment for Childhood Acute
Myeloid Leukemia: A North American Perspective. Pediatr Blood Cancer. doi:
10.1002/pbc.24498
Oudit, G. Y., Trivieri, M. G., Khaper, N., Liu, P. P., & Backx, P. H. (2006). Role of L-type
Ca2+ channels in iron transport and iron-overload cardiomyopathy. J Mol Med
(Berl), 84(5), 349-364. doi: 10.1007/s00109-005-0029-x
Paajanen, T.A., Oksala, N.K.J., Kuukasjarvi, P., Karhunen, P.J. (2010). Short stature is
associated with coronary heart disease: a systematic review of the literature and
a meta-analysis. European Heart Journal, 31, 1802-1809. doi:
10.1093/eurheartj/ehq155
Padilla, G.V., Frank-Stromberg, M., Koresawa, S. (2004). Single Instruments for
Measuring Quality of Life. In M. Frank-Stromberg, Olsen, S.J. (Ed.), Instruments
for Clinical Health-Care Research (pp. 128-149). Boston: Jones and Bartlett
Publishers.
Panjrath, G. S., Patel, V., Valdiviezo, C. I., Narula, N., Narula, J., & Jain, D. (2007).
Potentiation of Doxorubicin cardiotoxicity by iron loading in a rodent model. J Am
Coll Cardiol, 49(25), 2457-2464. doi: S0735-1097(07)01221-1 [pii]
10.1016/j.jacc.2007.02.060
Patrick, D.L., Erickson, P. (1993). Health Status and Health Policy: Quality of Life in
Health Care Evaluation and Resource Allocation. New York: Oxford University
Press.
Pearson, T. A., Mensah, G. A., Alexander, R. W., Anderson, J. L., Cannon, R. O., 3rd,
Criqui, M., . . . Vinicor, F. (2003). Markers of inflammation and cardiovascular
disease: application to clinical and public health practice: A statement for
healthcare professionals from the Centers for Disease Control and Prevention
and the American Heart Association. Circulation, 107(3), 499-511.
139

Phipps, S., & Srivastava, D. K. (1997). Repressive adaptation in children with cancer.
Health Psychol, 16(6), 521-528.
Phipps, S., Steele, R. G., Hall, K., & Leigh, L. (2001). Repressive adaptation in children
with cancer: a replication and extension. Health Psychol, 20(6), 445-451.
Pincott, E.S., Burch, M. (2011). New biomarkers in heart failure. Progress in Pediatric
Cardiology, 31, 49-52. doi: 10.1016/j.ppedcard.2010.121.009
Plein, S., Greenwood, J. P., Ridgway, J. P., Cranny, G., Ball, S. G., & Sivananthan, M.
U. (2004). Assessment of non-ST-segment elevation acute coronary syndromes
with cardiac magnetic resonance imaging. J Am Coll Cardiol, 44(11), 2173-2181.
doi: S0735-1097(04)01746-2 [pii] 10.1016/j.jacc.2004.08.056
Plein, S., Ridgway, J. P., Jones, T. R., Bloomer, T. N., & Sivananthan, M. U. (2002).
Coronary artery disease: assessment with a comprehensive MR imaging
protocol--initial results. Radiology, 225(1), 300-307.
Porter, J. B. (2001). Practical management of iron overload. Br J Haematol, 115(2), 239-
252. doi: 3195 [pii]
Reddy, K., Eng, J., Carlson, C., Ginsberg, J., Fish, J. (2011). Impact of transfusional
support for childhood cancer: Peering behind the iron curtain. Pediatr Blood
Cancer, 56(6), 908.
Reddy, K., Eng, J., Carlson, C., Ginsberg, J., Fish, J. (2012). Iron overload in childhood
cancer survivors. Paper presented at the 12th International Conference on Long-
Term Complications of Treatment of Children and Adolescents for Cancer,
Williamsburg, Virginia.
Ries, L.A.G., Melbert, D., Krapcho, M., Stinchcomb, D.G., Howlader, N., Horner, M.J.,
Mariotto, A., Miller, B.A., Feuer, E.J., Altekruse, S.F., Lewis, D.R., Clegg, L.,
Eisner, M.P., Reichman, M., Edwards, B.K. (eds). . (2008). SEER Cancer
Statistics Review, 1975-2005, based on November 2007 SEER data submission
Retrieved from http://seer.cancer.gov/csr/1975_2005/
Rouault, T. A. (2006). The role of iron regulatory proteins in mammalian iron
homeostasis and disease. Nat Chem Biol, 2(8), 406-414. doi: nchembio807 [pii]
10.1038/nchembio807
Ruccione, K. S., Mudambi, K., Sposto, R., Fridey, J., Ghazarossian, S., & Freyer, D. R.
(2012). Association of projected transfusional iron burden with treatment intensity
in childhood cancer survivors. Pediatr Blood Cancer, 59(4), 697-702. doi:
10.1002/pbc.24046
Santacroce, S. J. (2003). Parental uncertainty and posttraumatic stress in serious
childhood illness. J Nurs Scholarsh, 35(1), 45-51.
Sarikouch, S., Peters, B., Gutberlet, M., Leismann, B., Kelter-Kloepping, A., Koerperich,
H., Kuehne, T., Beerbaum, P. (2010). Sex-specific pediatric percentiles for
ventricular size and mass as reference values for cardiac MRI: assessment by
steady-state free precession and phase-contrast MRI flow. Circ Cardiovasc
Imaging, 3(1), 65-76. doi: 109.859074
Sattoe, J. N., van Staa, A., & Moll, H. A. (2012). The proxy problem anatomized: child-
parent disagreement in health related quality of life reports of chronically ill
adolescents. Health Qual Life Outcomes, 10, 10. doi: 1477-7525-10-10 [pii]
10.1186/1477-7525-10-10
Sawyer, D. B. (2013). Anthracyclines and heart failure. N Engl J Med, 368(12), 1154-
1156. doi: 10.1056/NEJMcibr1214975
Sawyer, D. B., Peng, X., Chen, B., Pentassuglia, L., & Lim, C. C. (2010). Mechanisms of
anthracycline cardiac injury: can we identify strategies for cardioprotection? Prog
Cardiovasc Dis, 53(2), 105-113. doi: S0033-0620(10)00119-2 [pii]
10.1016/j.pcad.2010.06.007
140

Schafer, A. I., Cheron, R. G., Dluhy, R., Cooper, B., Gleason, R. E., Soeldner, J. S., &
Bunn, H. F. (1981). Clinical consequences of acquired transfusional iron overload
in adults. N Engl J Med, 304(6), 319-324. doi: 10.1056/NEJM198102053040603
Schechter, A. N. (2008). Hemoglobin research and the origins of molecular medicine.
Blood, 112(10), 3927-3938. doi: 112/10/3927 [pii] 10.1182/blood-2008-04-
078188
Shankar, S., Robison, L., Jenney, M.E.M., Rockwood, T.H., Wu, E., Feusner, J.,
Friedman, D., Kane, R.L., Bhatia, S. (2005). Health-related quality of life in young
survivors of childhood cancer using the Minneapolis-Manchester Quality of Life
Youth Form. Pediatrics, 115, 435-442. doi: 10.1542/peds.2004-0649
Siah, C. W., Ombiga, J., Adams, L. A., Trinder, D., & Olynyk, J. K. (2006). Normal iron
metabolism and the pathophysiology of iron overload disorders. Clin Biochem
Rev, 27(1), 5-16.
Simunek, T., Sterba, M., Popelova, O., Kaiserova, H., Adamcova, M., Hroch, M., . . .
Gersl, V. (2008). Anthracycline toxicity to cardiomyocytes or cancer cells is
differently affected by iron chelation with salicylaldehyde isonicotinoyl hydrazone.
Br J Pharmacol, 155(1), 138-148. doi: bjp2008236 [pii] 10.1038/bjp.2008.236
Smith, A.W., Parsons, H.M., Kent, E.E., Bellizzi, K., Zebrack, B.J., Keel, G., Lynch, C.E.,
Rubenstein, M.B., Keegan, T.H.M., AYA HOPE Study Collaborative Group,.
(2013). Unmet support service needs and health-related quality of life among
adolescents and young adults with cancer: the AYA HOPE study. Frontiers in
Oncology, 3. doi: 10.3389/fonc.2013.00075
Sorensen, K. K., Dahl, L. B., & Smedsrod, B. (1997). Role of endocardial endothelial
cells in the turnover of hyaluronan in Atlantic cod (Gadus morhua). Cell Tissue
Res, 290(1), 101-109.
Sorensen, K., Levitt, G. A., Bull, C., Dorup, I., & Sullivan, I. D. (2003). Late anthracycline
cardiotoxicity after childhood cancer: a prospective longitudinal study. Cancer,
97(8), 1991-1998. doi: 10.1002/cncr.11274
Sposto, R. (2013, May 6, 2013).
Steenland, K., Cedillo, L., Tucker, J., Hines, C., Sorensen, K., Deddens, J., & Cruz, V.
(1997). Thyroid hormones and cytogenetic outcomes in backpack sprayers using
ethylenebis(dithiocarbamate) (EBDC) fungicides in Mexico. Environ Health
Perspect, 105(10), 1126-1130.
Steinherz, L. J., Steinherz, P. G., Tan, C. T., Heller, G., & Murphy, M. L. (1991). Cardiac
toxicity 4 to 20 years after completing anthracycline therapy. JAMA, 266(12),
1672-1677.
Theunissen, N. C., Vogels, T. G., Koopman, H. M., Verrips, G. H., Zwinderman, K. A.,
Verloove-Vanhorick, S. P., & Wit, J. M. (1998). The proxy problem: child report
versus parent report in health-related quality of life research. Qual Life Res, 7(5),
387-397.
Thompson, R. A. (1994). Emotion regulation: a theme in search of definition. Monogr
Soc Res Child Dev, 59(2-3), 25-52.
Trachtenberg, B. H., Landy, D. C., Franco, V. I., Henkel, J. M., Pearson, E. J., Miller, T.
L., & Lipshultz, S. E. (2011). Anthracycline-Associated Cardiotoxicity in Survivors
of Childhood Cancer. Pediatr Cardiol. doi: 10.1007/s00246-010-9878-3
Tukenova, M., Guibout, C., Oberlin, O., Doyon, F., Mousannif, A., Haddy, N., . . . de
Vathaire, F. (2010). Role of cancer treatment in long-term overall and
cardiovascular mortality after childhood cancer. J Clin Oncol, 28(8), 1308-1315.
doi: JCO.2008.20.2267 [pii] 10.1200/JCO.2008.20.2267
141

U.S. Department of Health and Human Services, Food Drug Administration, et al.
(2009). Guidance for Industry Patient-Reported Outcome Measures: Use in
Medical Product Development to Support Labeling Claims. Rockville, MD.
Upton, P., Lawford, J., & Eiser, C. (2008). Parent-child agreement across child health-
related quality of life instruments: a review of the literature. Qual Life Res, 17(6),
895-913. doi: 10.1007/s11136-008-9350-5
Vag, T., Kentouche, K. Krumbein, I. Reichenbach, J.R., Lopatta, E., Renz, D.M.,
Stenzel, M., Beck, J., Kaiser, W.A., Mentzel, H.-J. (2011). Noninvasive
measurement of liver iron concentration at MRI in children with acute leukemia:
initial results. Pediatr Radiol. doi: 10.1007/s00247-011-2122-3
VanderHeide, R.S., L'Ecuyer, T.J. (2007). Molecular basis of anthracycline-induced
chemotherapy. Heart Metabolism, 35, 1-4.
Varni, J. W., Burwinkle, T. M., Katz, E. R., Meeske, K., & Dickinson, P. (2002). The
PedsQL in pediatric cancer: reliability and validity of the Pediatric Quality of Life
Inventory Generic Core Scales, Multidimensional Fatigue Scale, and Cancer
Module. Cancer, 94(7), 2090-2106. doi: 10.1002/cncr.10428 [pii]
Varni, J. W., Burwinkle, T. M., & Lane, M. M. (2005). Health-related quality of life
measurement in pediatric clinical practice: an appraisal and precept for future
research and application. Health Qual Life Outcomes, 3, 34. doi: 1477-7525-3-34
[pii] 10.1186/1477-7525-3-34
Varni, J. W., Burwinkle, T. M., & Seid, M. (2005). The PedsQL as a pediatric patient-
reported outcome: reliability and validity of the PedsQL Measurement Model in
25,000 children. Expert Rev Pharmacoecon Outcomes Res, 5(6), 705-719. doi:
10.1586/14737167.5.6.705
Varni, J. W., Burwinkle, T. M., Seid, M., & Skarr, D. (2003). The PedsQL 4.0 as a
pediatric population health measure: feasibility, reliability, and validity. Ambul
Pediatr, 3(6), 329-341. doi: 03-035R [pii]
Varni, J. W., Limbers, C. A., & Burwinkle, T. M. (2007a). How young can children reliably
and validly self-report their health-related quality of life?: an analysis of 8,591
children across age subgroups with the PedsQL 4.0 Generic Core Scales. Health
Qual Life Outcomes, 5, 1. doi: 1477-7525-5-1 [pii]
10.1186/1477-7525-5-1
Varni, J. W., Limbers, C. A., & Burwinkle, T. M. (2007b). Impaired health-related quality
of life in children and adolescents with chronic conditions: a comparative analysis
of 10 disease clusters and 33 disease categories/severities utilizing the PedsQL
4.0 Generic Core Scales. Health Qual Life Outcomes, 5, 43. doi: 1477-7525-5-43
[pii] 10.1186/1477-7525-5-43
Varni, J. W., Seid, M., & Kurtin, P. S. (2001). PedsQL 4.0: reliability and validity of the
Pediatric Quality of Life Inventory version 4.0 generic core scales in healthy and
patient populations. Med Care, 39(8), 800-812.
Varni, J. W., Seid, M., & Rode, C. A. (1999). The PedsQL: measurement model for the
pediatric quality of life inventory. Med Care, 37(2), 126-139.
Vergely, C., Delemasure, S., Cottin, Y., Rochette, L. (2007). Preventing the cardiotoxic
effects of anthracyclines: from basic concepts to clinical data. Heart Metabolism,
35, 1-7.
Vermylen, C. (2008). What is new in iron overload? Eur J Pediatr, 167(4), 377-381. doi:
10.1007/s00431-007-0604-y
Visscher, H., Ross, C. J., Rassekh, S. R., Barhdadi, A., Dube, M. P., Al-Saloos, H., . . .
Hayden, M. R. (2011). Pharmacogenomic prediction of anthracycline-induced
cardiotoxicity in children. J Clin Oncol, 30(13), 1422-1428. doi:
JCO.2010.34.3467 [pii] 10.1200/JCO.2010.34.3467
142

Vogiatzi, M. G., Macklin, E. A., Trachtenberg, F. L., Fung, E. B., Cheung, A. M.,
Vichinsky, E., . . . Giardina, P. J. (2009). Differences in the prevalence of growth,
endocrine and vitamin D abnormalities among the various thalassaemia
syndromes in North America. Br J Haematol, 146(5), 546-556. doi: BJH7793 [pii]
10.1111/j.1365-2141.2009.07793.x
Ward, K. M., Binns, H., Chin, C., Webber, S. A., Canter, C. E., & Pahl, E. (2004).
Pediatric heart transplantation for anthracycline cardiomyopathy: cancer
recurrence is rare. J Heart Lung Transplant, 23(9), 1040-1045. doi:
10.1016/j.healun.2003.08.014 S105324980300408X [pii]
Werba, B. E., Hobbie, W., Kazak, A. E., Ittenbach, R. F., Reilly, A. F., & Meadows, A. T.
(2007). Classifying the intensity of pediatric cancer treatment protocols: the
intensity of treatment rating scale 2.0 (ITR-2). Pediatr Blood Cancer, 48(7), 673-
677. doi: 10.1002/pbc.21184
Wood, J. C. (2007). Magnetic resonance imaging measurement of iron overload. Curr
Opin Hematol, 14(3), 183-190. doi: 10.1097/MOH.0b013e3280d2b76b
00062752-200705000-00002 [pii]
Wood, J. C. (2008). Cardiac iron across different transfusion-dependent diseases. Blood
Rev, 22 Suppl 2, S14-21. doi: S0268-960X(08)70004-3 [pii]
10.1016/S0268-960X(08)70004-3
Wood, J. C. (2009). History and current impact of cardiac magnetic resonance imaging
on the management of iron overload. Circulation, 120(20), 1937-1939. doi:
CIRCULATIONAHA.109.907196 [pii] 10.1161/CIRCULATIONAHA.109.907196
Wood, J. C. (2011). Impact of iron assessment by MRI. Hematology Am Soc Hematol
Educ Program, 2011, 443-450. doi: 2011/1/443 [pii] 10.1182/asheducation-
2011.1.443
Wood, J. C., Enriquez, C., Ghugre, N., Otto-Duessel, M., Aguilar, M., Nelson, M. D., . . .
Coates, T. D. (2005). Physiology and pathophysiology of iron cardiomyopathy in
thalassemia. Ann N Y Acad Sci, 1054, 386-395. doi: 1054/1/386 [pii]
10.1196/annals.1345.047
Wood, J. C., Enriquez, C., Ghugre, N., Tyzka, J. M., Carson, S., Nelson, M. D., &
Coates, T. D. (2005). MRI R2 and R2* mapping accurately estimates hepatic iron
concentration in transfusion-dependent thalassemia and sickle cell disease
patients. Blood, 106(4), 1460-1465. doi: 2004-10-3982 [pii] 10.1182/blood-2004-
10-3982
Wood, J. C., Noetzl, L., Hyderi, A., Joukar, M., Coates, T., & Mittelman, S. (2010).
Predicting pituitary iron and endocrine dysfunction. Ann N Y Acad Sci, 1202, 123-
128. doi: NYAS5545 [pii] 10.1111/j.1749-6632.2010.05545.x
Woodgate, R. L. (2005). A different way of being: adolescents' experiences with cancer.
Cancer Nurs, 28(1), 8-15. doi: 00002820-200501000-00002 [pii]
Woodgate, R. L., & Degner, L. F. (2003). A substantive theory of Keeping the Spirit
Alive: the Spirit Within children with cancer and their families. J Pediatr Oncol
Nurs, 20(3), 103-119. doi: 10.1053/jpon.2003.75 S1043454203000936 [pii]
Woodgate, R. L., & Degner, L. F. (2004). Cancer symptom transition periods of children
and families. J Adv Nurs, 46(4), 358-368. doi: 10.1111/j.1365-2648.2004.03002.x
JAN3002 [pii]
World Health Organization. (1993). Study protocol for the World Health Organization
project to develop a Quality of Life assessment instrument (WHOQOL). Qual Life
Res, 2(2), 153-159.
Wouters, K. A., Kremer, L. C., Miller, T. L., Herman, E. H., & Lipshultz, S. E. (2005).
Protecting against anthracycline-induced myocardial damage: a review of the
most promising strategies. Br J Haematol, 131(5), 561-578. doi: BJH5759 [pii]
143

10.1111/j.1365-2141.2005.05759.x
Xu, X., Persson, H. L., & Richardson, D. R. (2005). Molecular pharmacology of the
interaction of anthracyclines with iron. Mol Pharmacol, 68(2), 261-271. doi:
mol.105.013383 [pii] 10.1124/mol.105.013383
Zacharski, L. R. (2010). Hemochromatosis, iron toxicity and disease. J Intern Med,
268(3), 246-248. doi: 10.1111/j.1365-2796.2010.02233.x
Zacharski, L. R., Chow, B. K., Howes, P. S., Shamayeva, G., Baron, J. A., Dalman, R.
L., . . . Lavori, P. W. (2008). Decreased cancer risk after iron reduction in patients
with peripheral arterial disease: results from a randomized trial. J Natl Cancer
Inst, 100(14), 996-1002. doi: djn209 [pii] 10.1093/jnci/djn209
Zavala, O. (2013).
Zeltzer, L. K., Lu, Q., Leisenring, W., Tsao, J. C., Recklitis, C., Armstrong, G., . . . Ness,
K. K. (2008). Psychosocial outcomes and health-related quality of life in adult
childhood cancer survivors: a report from the childhood cancer survivor study.
Cancer Epidemiol Biomarkers Prev, 17(2), 435-446. doi: 17/2/435 [pii]
10.1158/1055-9965.EPI-07-2541
Zethelius, B., Berglund, L., Sundstrom, J., Ingelsson, E., Basu, S., Larsson, A., . . .
Arnlov, J. (2008). Use of multiple biomarkers to improve the prediction of death
from cardiovascular causes. N Engl J Med, 358(20), 2107-2116. doi:
358/20/2107 [pii] 10.1056/NEJMoa0707064
Zhang, A. S., & Enns, C. A. (2009). Molecular mechanisms of normal iron homeostasis.
Hematology Am Soc Hematol Educ Program, 207-214. doi: 2009/1/207 [pii]
10.1182/asheducation-2009.1.207

144

APPENDIX

1. Literature reports of transfusion-related iron overload among childhood cancer
patients
2. Manuscript: Ruccione, et al. (2012). Association of projected transfusional iron
burden with treatment intensity in childhood cancer survivors. Pediatr Blood Cancer,
59(4), 697-702.
3. Funding secured for dissertation study
4. IRB approval
5. Study consent/permission/assent forms (English)
6. Computation of intraclass correlations and Spearman rank correlations for patient
self-reported and parent proxy-reported HRQOL and fatigue

Asset Metadata

Creator Ruccione, Kathleen S. (author)

Core Title Transfusional iron, anthracyclines and cardiac outcomes in childhood cancer survivors

Contributor Electronically uploaded by the author (provenance)

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Preventive Medicine (Health Behavior)

Publication Date 07/23/2013

Defense Date 04/02/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag childhood cancer survivors,iron overload,OAI-PMH Harvest,transfusional iron

Format application/pdf (imt)

Language English

Advisor RIchardson, Jean (committee chair), Jordan-Marsh, Maryalice (committee member), Krailo, Mark (committee member), Rohrbach, Luanne (committee member), Siegel, Stuart E. (committee member), Wood, John C. (committee member)

Creator Email kruccione@chla.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-297327

Unique identifier UC11293180

Identifier etd-RuccioneKa-1824.pdf (filename),usctheses-c3-297327 (legacy record id)

Legacy Identifier etd-RuccioneKa-1824-0.pdf

Dmrecord 297327

Document Type Dissertation

Format application/pdf (imt)

Rights Ruccione, Kathleen S.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Abstract (if available)

Abstract Background: Childhood cancer survivors receiving multiple packed red blood cell (PRBC) transfusions may be at increased risk for iron deposition in vital organs such as the heart, thereby increasing risk of serious late effects due to organ dysfunction with associated symptoms, which may diminish their health-related quality of life (HRQOL). ❧ Purpose: To determine the prevalence, distribution and severity of transfusional iron deposition in heart, liver and pancreas among a cohort of childhood cancer survivors, and to characterize the prevalence, distribution and severity of cardiac dysfunction and pathological remodeling among a cohort of childhood cancer survivors