@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Medicine, Faculty of"@en, "Pathology and Laboratory Medicine, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Lin, David Chia-Hsiang"@en ; dcterms:issued "2013-04-30T00:00:00"@en, "2012"@en ; vivo:relatedDegree "Doctor of Philosophy - PhD"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """Background: Cardiac transplantation is considered the primary therapy for patients with end-stage heart failure. However, the detection of chronic cardiac allograft rejection (expressed as cardiac allograft vasculopathy; CAV) remains an important unsettled issue in cardiac transplantation. The current gold standards for the diagnosis and monitoring of acute rejection and CAV are invasive in nature with risk for complications. From a clinical perspective, more accurate, minimally-invasive alternatives are clearly desirable. The goal of my thesis is to identify biomarkers of human heart allograft rejection, and assess their potential clinical utility and biological implications in the respective disease contexts. Central hypothesis: Peripheral blood-derived molecular biomarker panels provide a means for sensitive and specific diagnosis of acute and chronic cardiac allograft rejection, as well as helping to gain insight into the underlying mechanisms of rejection. Methods: Genomic biomarkers of acute cardiac allograft rejection (AR) and proteomic biomarkers of cardiac allograft vasculopathy (CAV) were identified via Affymetrix microarray analysis of whole blood samples and iTRAQ proteomic analysis of plasma samples, respectively, from cardiac transplant patients. From the genes differentially expressed between AR vs. Non-rejectors (NR), and differentially expressed proteins between CAV and Non-significant CAV (Non-CAV) subjects, biomarkers panels for AR and CAV were generated using classification methods. AR and CAV biomarkers were further analyzed for their biological implications using bioinformatical tools. Results: Microarray comparison between the AR and NR subjects revealed over 1000 differentially expressed genes, many of which that were associated with cellular functions involved in innate and humoral immunity. The 12-gene biomarker panel generated based on the differentially expressed candidates demonstrated 83% sensitivity and 100% specificity. Proteomic analysis of CAV versus Non-CAV plasma samples ultimately lead to the generation of an 18-protein biomarker panel which demonstrated 80% sensitivity and 89% specificity for CAV."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/43474?expand=metadata"@en ; skos:note " BIOMARKERS OF ACUTE AND CHRONIC HUMAN HEART ALLOGRAFT REJECTION by David Chia-Hsiang Lin B.M.L.Sc., The University of British Columbia, 2006 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Pathology and Laboratory Medicine) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) October 2012 © David Chia-Hsiang Lin, 2012 ii Abstract Background: Cardiac transplantation is considered the primary therapy for patients with end- stage heart failure. However, the detection of chronic cardiac allograft rejection (expressed as cardiac allograft vasculopathy; CAV) remains an important unsettled issue in cardiac transplantation. The current gold standards for the diagnosis and monitoring of acute rejection and CAV are invasive in nature with risk for complications. From a clinical perspective, more accurate, minimally-invasive alternatives are clearly desirable. The goal of my thesis is to identify biomarkers of human heart allograft rejection, and assess their potential clinical utility and biological implications in the respective disease contexts. Central hypothesis: Peripheral blood-derived molecular biomarker panels provide a means for sensitive and specific diagnosis of acute and chronic cardiac allograft rejection, as well as helping to gain insight into the underlying mechanisms of rejection. Methods: Genomic biomarkers of acute cardiac allograft rejection (AR) and proteomic biomarkers of cardiac allograft vasculopathy (CAV) were identified via Affymetrix microarray analysis of whole blood samples and iTRAQ proteomic analysis of plasma samples, respectively, from cardiac transplant patients. From the genes differentially expressed between AR vs. Non- rejectors (NR), and differentially expressed proteins between CAV and Non-significant CAV (Non-CAV) subjects, biomarkers panels for AR and CAV were generated using classification methods. AR and CAV biomarkers were further analyzed for their biological implications using bioinformatical tools. Results: Microarray comparison between the AR and NR subjects revealed over 1000 differentially expressed genes, many of which that were associated with cellular functions involved in innate and humoral immunity. The 12-gene biomarker panel generated based on the differentially expressed candidates demonstrated 83% sensitivity and 100% specificity. Proteomic analysis of CAV versus Non-CAV plasma samples ultimately lead to the generation of an 18-protein biomarker panel which demonstrated 80% sensitivity and 89% specificity for CAV. iii Conclusion: Taken together, the work from my thesis shows the potential utility of blood derived ’omic’-based biomarker panels in a clinical setting as diagnostic and monitoring tools for key cardiac post-transplantation conditions. This body of work also demonstrates the value of using ‘omics’ technologies to gain biological insight into AR and CAV. iv Preface This dissertation contains chapters which are based on published manuscripts. Chapter 1 Portions of Chapter 1 are based on review articles published in Circulation Research [Marchant DJ, Boyd JH, Lin D, Granville DJ, Garmaroudi FS, McManus BM: Inflammation in myocardial diseases. Circ Res 2012;110:126-44.] and Canadian Journal of Cardiology [Lin D, Hollander Z, Meredith A, McManus BM: Searching for 'omic' biomarkers. Can J Cardiol 2009;25 Suppl A:9A- 14A.]. I was a co-equal contributor in the first review article; specifically, I wrote the section of the review focusing on cardiac inflammation and cardiac allograft rejection. For the latter review article, I was the primary author and was responsible for the critical review and revision of the manuscript. Chapter 2 Chapter 2 is an altered version of the manuscript published in Journal of Cardiac Failure [Lin D, Hollander Z, Meredith A, et al.: Molecular signatures of end-stage heart failure. J Card Fail 2011;17:867-74.]. This study was conceptualized by Dr. Bruce McManus, Ms. Zsuzsanna Hollander and me. With assistance from Dr. Bruce McManus and Ms. Zsuzsanna Hollander (who conducted the statistical analyses), I was responsible for subject review, sample selection and design of analyses. As well, I performed all functional enrichment analyses of the data, interpretation of the genomic and proteomic results and writing of the manuscript. Dr. Bruce McManus, Ms. Zsuzsanna Hollander and Ms. Anna Meredith assisted in editing of the manuscript. Chapter 3 Chapter 3 is an altered version of the manuscript published in the Journal of Heart and Lung Transplantation [Lin D, Hollander Z, Ng RT, et al.: Whole blood genomic biomarkers of acute cardiac allograft rejection. J Heart Lung Transplant 2009;28:927-35.]. This study was carried out under the Biomarker in Transplantation (BiT) umbrella, co-lead by Drs. Robert McMaster, Paul Keown and Bruce McManus. This study was conceptualized by Dr. Bruce McManus and me. v With assistance from Dr. Bruce McManus and Ms. Zsuzsanna Hollander (who conducted the statistical analyses on the microarray data), I was responsible for subject review, sample selection and design of analyses. I was also responsible for all Gene Ontology analysis, design, execution and data analysis of the classifier gene validation experiments, in addition to interpreting the results and writing the manuscript. Dr. Bruce McManus and Ms. Zsuzsanna Hollander made intellectual contributions to the design of the study, and assisted in revision of the manuscript. Chapter 4 Chapter 4 is based on a study that was also carried out under the BiT umbrella. This study was conceptualized by Dr. Bruce McManus and me. With assistance from Dr. Gabriela Cohen Freue (who performed the statistical analyses on the proteomics data), I was responsible for sample selection, design of the study, including all subject review, subject selection and processing of clinical coronary angiography data. I was also responsible for all functional enrichment analysis, interpretation of results, writing and editing of the manuscript. Drs. Gabriela Cohen Freue and Bruce McManus also made intellectual contributions to the design of the study, in addition to final editing of the manuscript. Ethics Approval All work described in dissertation, including samples collected and experimented performed, were conducted as part of the BiT project, which was approved by The Providence Health Care Research Ethics Board, certificate number H04-50286. vi Table of Contents Abstract ................................................................................................................................. ii Preface ................................................................................................................................. iv Table of Contents .................................................................................................................. vi List of Tables ......................................................................................................................... ix List of Figures ......................................................................................................................... x List of Abbreviations ............................................................................................................. xi Acknowledgements ............................................................................................................. xv Dedication ......................................................................................................................... xvii CHAPTER 1: Introduction ....................................................................................................... 1 1.1 “Pre-“heart transplantation ............................................................................................ 2 1.1.1 Overview of heart failure ............................................................................................ 2 1.1.2 Conditions leading to heart failure ............................................................................. 2 1.1.3 Classification and current therapies of heart failure .................................................. 4 1.2 Heart transplantation ...................................................................................................... 6 1.2.1 Overview and brief history ......................................................................................... 6 1.2.2 Immunosuppressive therapies ................................................................................... 6 1.2.3 Survival rate of heart transplantation patients .......................................................... 9 1.3 “Post-“heart transplantation hurdles ........................................................................... 10 1.3.1 Overview of cardiac allograft rejections ................................................................... 10 1.3.2 Acute cardiac allograft rejection ............................................................................... 11 1.3.2.1 Occurrence and diagnosis of acute cardiac allograft rejection ........................ 11 1.3.2.2 Pathogenesis of acute cardiac allograft rejection ............................................ 14 1.3.3 Cardiac allograft vasculopathy (CAV) ........................................................................ 18 1.3.3.1 Occurrence and diagnosis of CAV as an expression of chronic rejection ........ 18 1.3.3.2 Pathogenesis of cardiac allograft vasculopathy ............................................... 20 1.4 Overview of biomarkers ................................................................................................ 23 CHAPTER 2: Dissertation overview ....................................................................................... 25 2.1 Rationale of research proposal ..................................................................................... 26 2.2 Overview of research proposal ..................................................................................... 27 2.3 Central hypothesis ........................................................................................................ 28 2.4 Specific aims .................................................................................................................. 28 CHAPTER 3: Molecular signatures of end-stage heart failure patients prior to cardiac transplantation .................................................................................................................... 29 3.1 Background ................................................................................................................... 30 3.2 Rationale ....................................................................................................................... 30 3.3 Materials and methods ................................................................................................. 32 3.3.1 Subjects and specimens ............................................................................................ 32 vii 3.3.2 Sample and data processing ..................................................................................... 34 3.3.2.1 Genomics .......................................................................................................... 34 3.3.2.2 Proteomics ........................................................................................................ 34 3.3.3 Analysis ..................................................................................................................... 35 3.3.3.1 Genomics .......................................................................................................... 35 3.3.3.2 Proteomics ........................................................................................................ 35 3.3.3.3 Functional enrichment ..................................................................................... 35 3.4 Results ........................................................................................................................... 37 3.4.1 Ischemic heart disease (IHD) versus Non-ischemic cardiomyopathy (NICM) .......... 37 3.4.1.1 Genomics .......................................................................................................... 37 3.4.1.2 Proteomics ........................................................................................................ 37 3.4.2 Chronic heart failure (CHF) versus Normal cardiac function (NCF) .......................... 38 3.4.2.1 Genomics .......................................................................................................... 38 3.4.2.2 Proteomics ........................................................................................................ 41 3.5 Discussion ...................................................................................................................... 43 3.5.1 Integration of biological information and interpretation ......................................... 44 3.5.1.1 Response to cardiac damage/wound healing response (wound healing, extracellular matrix (ECM) remodeling, cytoskeleton regulation).................................... 44 3.5.1.2 Inflammation/immune response (IL-6 signaling, kallikrein-kinin system) ....... 45 3.5.1.3 Blood coagulation/cell adhesion (protein C signaling, platelet-endothelium- leukocyte interactions) ...................................................................................................... 45 3.5.1.4 Apoptosis/DNA damage checkpoint................................................................. 46 3.5.2 Potential applications, caveats to the study, and future direction .......................... 46 CHAPTER 4: Biomarkers of acute cardiac allograft rejection ................................................. 48 4.1 Background ................................................................................................................... 49 4.2 Rationale ....................................................................................................................... 49 4.3 Materials and methods ................................................................................................. 51 4.3.1 Subjects and specimens selection ............................................................................ 51 4.3.2 Sample and data processing ..................................................................................... 54 4.3.3 Analysis ..................................................................................................................... 54 4.3.3.1 Identification of biomarkers ............................................................................. 54 4.3.3.2 Functional enrichment analysis ........................................................................ 56 4.3.3.3 Generation and evaluation of the AR biomarker panel ................................... 56 4.4 Results ........................................................................................................................... 57 4.4.1 Differentially expressed genes in AR patients .......................................................... 57 4.4.2 Dysregulation of molecular and cellular processes in AR patients ........................... 57 4.4.2.1 Gene ontology (GO) analysis ............................................................................ 57 4.4.3 AR biomarker panel genes ........................................................................................ 59 4.4.4 Evaluation of the AR biomarker panel ...................................................................... 60 viii 4.5 Discussions .................................................................................................................... 62 4.5.1 Integration of biological information and interpretation ......................................... 62 4.5.2 Assessment and validation of the AR biomarker panel ............................................ 65 4.5.3 Current study results versus CARGO results ............................................................. 67 4.5.4 Potential applications, caveats to the study, and future directions ......................... 68 CHAPTER 5: Biomarkers of cardiac allograft vasculopathy .................................................... 70 5.1 Background ................................................................................................................... 71 5.2 Rationale ....................................................................................................................... 71 5.3 Materials and methods ................................................................................................. 73 5.3.1 Subjects and specimens ............................................................................................ 73 5.3.1.1 Screening and identification of CAV and Non-CAV patients ............................ 73 5.3.2 Sample selection and data processing ...................................................................... 74 5.3.3 Analysis ..................................................................................................................... 74 5.3.3.1 Identification of CAV biomarkers and functional enrichment ......................... 74 5.3.3.2 Validation of CAV biomarker panel .................................................................. 75 5.4 Results ........................................................................................................................... 76 5.4.1 Coronary angiography and patient characteristics ................................................... 76 5.4.2 Differentially expressed proteins and CAV biomarker panel ................................... 80 5.4.3 Principal component analysis of the proteomic CAV biomarker panel .................... 83 5.4.4 Functional enrichment of the protein CAV biomarkers ........................................... 84 5.4.5 Performance estimation of the of CAV biomarker panel ......................................... 85 5.5 Discussion ...................................................................................................................... 87 5.5.1 Establishing definition of CAV for the study ............................................................. 87 5.5.2 Integration of biological information and interpretation ......................................... 87 5.5.2.1 Complement system-mediated effects ............................................................ 90 5.5.2.2 Other immune- and inflammatory effects ....................................................... 90 5.5.2.3 Non-alloimmune specific factors, e.g., lipid and hormone transport .............. 91 5.5.2.4 Response to injury mechanisms ....................................................................... 91 5.5.3 Evaluation of the CAV biomarker panel performance .............................................. 92 5.5.4 Potential applications, caveats to the study, and future directions ......................... 93 CHAPTER 6: Conclusion ....................................................................................................... 94 6.1 Closing remarks ............................................................................................................. 95 6.2 Future opportunities ................................................................................................... 100 References ......................................................................................................................... 104 Appendices – Assay and ‘omics’ methodologies .................................................................. 118 Appendix A - Genomics technology – Affymetrix GeneChip® microarray ............................. 118 Appendix B - Proteomics technology – Mass spectrometry and iTRAQ proteomics ............. 119 ix List of Tables Table 1. Roles of immune effectors and complement system in allograft rejection .................... 15 Table 2. Demographics of subject cohorts (CHF and NCF) ............................................................ 33 Table 3. Top 10 process networks identified by MetaCore as statistically significant, based on the 7,426 differentially expressed probe sets between CHF and NCF. ......................................... 40 Table 4. Top 10 process networks identified as statistically significant by MetaCore, based on the differentially expressed proteins between CHF and NCF. ...................................................... 42 Table 5. Demographics of cardiac transplant subject cohorts. ..................................................... 53 Table 6. Relative expression levels and associative GO terms over-represented in the 1295 statistically significant probe sets. ................................................................................................. 58 Table 7. Acute cardiac allograft rejection biomarker panel. ......................................................... 59 Table 8. AR biomarker performance evaluation. .......................................................................... 61 Table 9. Summary of the biological functions of the AR biomarker panel genes based on previous literature. ........................................................................................................................ 63 Table 10. Cardiac transplant patient demographics. .................................................................... 77 Table 11. Summary of angiographically-assessed data regarding coronary artery stenosis. ....... 78 Table 12. Proteomic biomarker panel reflecting cardiac allograft vasculopathy. ......................... 81 x List of Figures Figure 1. Histological diagnosis and grading of acute cardiac rejection. ...................................... 13 Figure 2. Major contributing mechanisms of myocardial inflammation in the context of cardiac allograft transplantation. ............................................................................................................... 17 Figure 3. Percentage diameter stenosis of coronary artery .......................................................... 19 Figure 4. Major events central to the onset and development of CAV. ........................................ 22 Figure 5. Heatmap of the top 100 differentially expressed probe sets between chronic heart failure (CHF) and normal cardiac function (NCF). .......................................................................... 39 Figure 6. Top 10 GO terms based on functional enrichment analysis of the 7,426 probe sets differentially expressed between CHF and NCF. ........................................................................... 40 Figure 7. Top 10 GO terms based on functional enrichment analysis of the 71 PGs which showed differential concentrations between CHF and NCF. ...................................................................... 42 Figure 8. Division of subject samples into training and test cohorts. ........................................... 52 Figure 9. Overall workflow of the data analysis. ........................................................................... 55 Figure 10. AR biomarker expression evaluation. ........................................................................... 61 Figure 11. Classification performance of the CAV biomarker panel. ............................................ 79 Figure 12. 3D scatter plot of the principal component analysis (PCA) results. ............................. 83 Figure 13. Top 10 Gene Ontology (GO) terms based on functional enrichment analysis of the proteins on the CAV biomarker panel generated.......................................................................... 84 Figure 14. Receiver operating characteristic (ROC) curve for the CAV protein biomarker panel identified. ....................................................................................................................................... 86 Figure 15. Potential implications of biomarkers identified in the context of cardiac allograft vasculopathy (CAV). ....................................................................................................................... 89 Figure 16. Summary of potential utility of AR and CAV biomarkers identified ........................... 103 xi List of Abbreviations Ab – antibody ABI – Applied Biosystems ACC – American College of Cardiology ACE – angiotensin converting enzyme AHA – American Heart Association AlloAb – alloantibody AMR – antibody-mediated rejection AP-1 – activator protein-1 APC – antigen presenting cells AR – acute cardiac allograft rejection ARVC – arrhythmogenic right ventricular cardiomyopathy AST – aspartate transaminase AUC – area under the curve AV – atrio-ventricular AZA – azathioprine BiT – Biomarker in Transplantation initiative BNP – brain natriuretic peptide CAD – coronary artery disease CARGO – Cardiac Allograft Rejection Gene Expression Observation CAV – cardiac allograft vasculopathy CHF – chronic heart failure CI – calcineurin inhibitors CK – creatine kinase CR – chronic rejection CRP – C-reactive protein CSA – cyclosporine CTL – cytotoxic T lymphocyte cTnI – cardiac troponin I cTnT – cardiac troponin T xii CVB3 – coxsackievirus B3 DC – dendritic cells DCM – dilated cardiomyopathy DS – diameter stenosis EC – endothelial cell ECM – extracellular matrix EDTA – ethylenediaminetetraacetic acid EMB – endomyocardial biopsy ESHF – end-stage heart failure FDA – Food and Drug Administration FDR – false discovery rate GO – gene ontology GR – glucocorticoid receptor HCM – hypertrophic cardiomyopathy HF – heart failure HG – human genome I/R – ischemia/reperfusion ICAM-1 – intercellular adhesion molecule-1 ICM – ischemic cardiomyopathy IFN – interferon Ig – immunoglobulin IHD – ischemic heart disease IL – interleukin IPI – International Protein Index ISHLT – International Society for Heart and Lung Transplantation iTRAQ – isobaric tags for relative and absolute quantitation IVUS – intravascular ultrasound LAD – left anterior descending artery LCX – left circumflex LDA – linear discriminant analysis LIMMA – linear models for microarray data xiii LOOCV – leave-one-out cross-validation LV – left ventricle LVAD – left ventricular assist device MAC – membrane attack complex MALDI – matrix assisted laser desorption ionization MBL – mannose-binding lectin MeSH – medical subject heading MHC – major histocompatibility complex MLD – minimum lumen diameter MMF – mycophenolate mofetil mTOR – mammalian target of rapamycin NCF – normal cardiac function NFH – non-failing heart NFκB – nuclear factor κB NICM – non-ischemic cardiomyopathy NIH – National Institute of Health NK cells – natural killer cells NYHA – New York Heart Association PBMC – peripheral blood mononuclear cell PC – principal component PCA – principal component analysis PG – protein group QCA – quantitative coronary angiography qPCR – quantitative real-time polymerase chain reaction RCA – right coronary artery RCM – restrictive cardiomyopathy RD – reference diameter RMA – robust multi-array average ROC – receiver operating characteristic ROS – reactive oxygen species RT- IVT – reverse transcription-in vitro transcription xiv SAM – significant analysis of microarray SAS – statistical analysis system SDA – stepwise discriminant analysis SIR – sirolimus SMC – smooth muscle cell SNP – single-nucleotide polymorphism SPC – smooth muscle progenitor cell TAC – tacrolimus TNF – tumor necrosis factor TOF – time of flight VCAM-1 – vascular cell adhesion molecule-1 VO2 max – maximum oxygen (O2) uptake xv Acknowledgements First and foremost, I would like to thank my supervisor, Dr. Bruce McManus, for his patience and mentorship. Thank you Dr. McManus for your contagious enthusiasm, unwavering support, and timely encouragements. As well, thank you for believing in my abilities but at the same time, being upfront about areas I should improve on. The science and life lessons I have learned under your tutelage I will forever cherish. To my committee members, Drs. David Granville, Honglin Luo, Alice Mui and Raymond Ng, thank you for all the constructive criticisms and insightful comments during my committee meetings to help ensure I am progressing along the right track. To Dr. Honglin Luo, thank you for your support since I’ve worked in your laboratory as an undergraduate student. To Drs. Alice Mui and Raymond Ng, thank you for your encouragements and providing shining moments of clarity during times when I struggle or have self doubts. Without your support this thesis and I would not have seen the light at the end of the tunnel. During my studies I was very fortunate to have received several supporting awards and studentship from the Michael Smith Foundation for Health Research, the Canadian Institutes of Health Research, and Genome Canada. I am very grateful to be part of the Biomarker in Transplantation (BiT) initiatives and have the chance to work with many great people who helped me along the way. To the members of the Biomarkers in Transplantation project computational team, in particular, Drs. Robert Balshaw, Gabriela Cohen Freue, Oliver Günther and Ms. Zsuzsanna Hollander, thank you for being patient and taking the time to show me the ropes in your world of data, statistics, and analysis. Also, thank you Janet Wilson-McManus for the opportunities to be involved in different projects within the BiT initiatives and for believing in my abilities. Through the BiT initiative, I also had the opportunity to collaborate with and learn from researchers and technicians from local and neighboring centres and institutes. To Dr. Andrew Ming-Lum, Ms. Erin McCarrell and members from the James Hogg Research Centre, thank you for all the training, and sharing your technical expertise and knowledge with me. xvi To my lab mates and colleagues, Drs. Jon Carthy, Farshid Garmaroudi and Brian Wong, as well as Ms. Seti Boroomand, Ms. Anna Meredith and Mr. Jerry Wong, thank you for your friendship, support, pearls of wisdom, and moments of hilarity, both inside and outside of the laboratory. Laughter makes the world go around, and you make the journey to completing this degree a lot enjoyable than I could ever hoped for. Finally, I would like to thank my sister, Sophia, for her unconditional love and support. Your sympathetic ears and motivation have helped me more than you know. To my loving parents, Frank and Linda, thank you for all the sacrifices you have made to raise and teach me; and for always being there and encouraging me in my endeavors. You two have always inspired me to work harder and pursue my dreams, and for that I will be forever grateful. xvii Dedication To my ah-gong and ah-ma, And To my family members, Frank, Linda and Sophia. I hope you are as proud of me as I am of you. 1 CHAPTER 1: Introduction 2 1.1 “Pre-“heart transplantation 1.1.1 Overview of heart failure Heart failure (HF) is a progressive clinical syndrome characterized by inability of the heart to adequately pump blood to meet metabolic demands of the body.1,2 The cause of HF is diverse, and can result from a variety of cardiac disorders that affect the structure or function, i.e., contraction/relaxation, conduction, or rhythm of the heart.3,4 It is estimated that approximately 2% of the adult population will suffer from heart failure, although the prevalence is considerably higher in the older population, occurring in up to 10% of people 65 years or older.5 The lifetime risk on average, however, for a 40 year old adult, is approximately 20%.5-7 Importantly, the prognosis of HF patients also tends to worsen over time,8,9 and is associated a 30-40% mortality rate within the first year after the initial diagnosis, and a 5-year mortality rate between 48 and 70%.5,10 Not only does the development of HF greatly affect the quality of life of patients, its prevalence also has a huge economical impact on the health care system. In the United States, more than 39 billion dollars are spent yearly on the care of HF patients, including hospitalization, treatment and associated costs.5,11 1.1.2 Conditions leading to heart failure As described earlier, HF can result from virtually any form of cardiac disorders, and reflect contributions from both genetic and environmental factors.2 From an etiological point- of-view, two major categories of conditions that can ultimately lead to the development of HF are: i) ischemic heart diseases (IHD)/ischemic cardiomyopathies (ICM), and ii) non-ischemic cardiomyopathies (NICM).12 Ischemic heart diseases (IHD) generally arise from coronary artery diseases (CAD) such as atherosclerosis, which is characterized by the narrowing of coronary arteries via intimal plaque formation.3,13 This occlusive process can lead to a lack of adequate oxygenated blood supply to the heart, generating regions of ischemic heart muscle and ultimately myocardial dysfunction.3,13 Historically, the term ‘ischemic cardiomyopathy’ was first introduced in 1970 by 3 Burch and colleagues in New Orleans, to describe conditions which involve severe myocardial dysfunction thought to be the results of occlusive CAD.3,14 Non-ischemic cardiomyopathies (NICM) are a heterogeneous group of conditions, including diseases that are predominantly genetic in nature, e.g., arrhythmogenic right ventricular cardiomyopathy (ARVC) and hypertrophic cardiomyopathy (HCM), mixed/predominantly-non-genetic in nature, e.g., dilated cardiomyopathy (DCM) and restrictive cardiomyopathy (RCM), or acquired, e.g., inflammatory cardiomyopathy due to myocarditis.15 Briefly, ARVC is a relatively rare form of inheritable cardiac muscle disease (1:5000), which mainly affects the right ventricle and is characterized by the loss of muscle cells, i.e., myocytes, which are replaced by fatty and/or fibrofatty tissues.3,15 HCM is also an inheritable cardiac disease and is more prevalent than ARVC, occurring in 1:500 people in the general population.3,15 HCM is largely characterized by non-dilated hypertrophy of the left ventricle (LV), in the absence of another disease (e.g., hypertension) which may also cause hypertrophy, or thickening, of the LV wall.3,15 DCM is the most common form of NICM,16 and is characterized by the enlargement of one or both ventricles (although primarily the LV),17 impaired systolic function, and normal LV wall thickness.3,15 The origin of DCM can be can be idiopathic, genetic (e.g., familial cardiomyopathy), or environmental (e.g. alcoholic cardiomyopathy), but can all lead to ventricular dilation and associated decreased systolic function (or just systolic dysfunction).3,15 RCM is a rare form of cardiac muscle disease that can be either sporadic or familial.3,15 RCM is characterized by features such as enlargement of both atriums, i.e., biatrial enlargement, ventricular filling dysfunction or abnormal relaxation, presence of restrictive physiology, but normal thickness of LV wall and atrial ventricular (AV) valves, as well as normal (or borderline normal) systolic function.3,15 Inflammatory cardiomyopathy is considered an acquired form of NICM, and involves cardiac dysfunction as a result of myocarditis.3,15 Myocarditis can be either an acute or chronic inflammatory condition, and can be induced by various causes, such as virus (e.g., coxsackievirus CVB3), bacteria (e.g., streptococcus and meningococcus), fungus (e.g., 4 aspergillosis), parasites (e.g., toxoplasmosis), as well as toxin (e.g., cocaine), and drug hyper sensitivity reactions (e.g., antibiotics).3,15 As part of the inflammatory process, myocarditis is typically characterized by progressive and active injury (e.g., infiltration of inflammatory cells leading to myocyte necrosis in the heart), and eventual healing of the damaged heart (e.g., replacement of necrotic tissue with fibrosis).3,15 1.1.3 Classification and current therapies of heart failure Given the complexity of HF, several methods of classification and categorization have been proposed for HF. For instance, HF can be described as either systolic, i.e., involving contraction of the heart, or diastolic, i.e., involving relaxation of the heart. From a more clinical and pathophysiological perspective,10 the American College of Cardiology (ACC)/American Heart Association (AHA) has proposed a four stage system (Stages A through D), where the first two stages are considered as ‘pre-HF’, e.g., patients at risk of HF but not yet showing clinical signs of HF, and the latter two stages include patients with clinical HF presently or previously (Stage C), and those who may require more advanced therapy or end-of-life care (Stage D).18-20 From a more functional-based perspective, the New York Heart Association (NYHA) classification of HF ranges from class I to class IV and is thought to compliment the ACC/AHA system – typically used to further classify patients in stage C or D.18,19 The NYHA classes, with class I being the least severe and class IV being the most, are largely dependent on clinical symptoms and the degree of functional limitation in normal physical activities for the patients, e.g., climbing stairs.18-20 Therapy for heart failure is generally dependent on the clinical symptoms and functional limitations of the patients. The type of therapy can range from lifestyle changes, e.g., increased physical activity, decreased alcohol and salt intake, such as in the case of ACC/AHA Stage A patients, to the use of medications such as angiotensin converting enzyme (ACE) inhibitors and beta adrenergic blocking agents (beta-blockers).18 Other medications such as angiotensin II receptor blockers and aldosterone antagonists, or even procedures such as resynchronization therapy are also used in selected patients with more severe form of HF, e.g., ACC/AHA stage C.18 5 While the aforementioned therapies are commonly used in attempt to improve the patients symptoms and prevent the progression of heart failure, some patients do eventually advance to end-stage heart failure (ESHF), and will require more extraordinary measures of interventions.18 Surgical approaches such as implantation of mechanical ventricular assist devices, e.g., left ventricular assist device (LVAD) are often used to support and prolong the life of ESHF patients.18 Given that adult cardiac myocytes are largely thought to be terminally differentiated cells with limited ability for self regeneration, the human heart is, for the most part, unable to replace injured or dead myocytes to the extent that it significantly reverses the HF progression from the point of ESHF.3 As such, currently, heart transplantation remains the primary therapy of choice and the definitive long term solution for patients with end-stage vital heart failure.3,18 6 1.2 Heart transplantation 1.2.1 Overview and brief history The first successful human-to-human heart transplantation was performed on December 3rd, 1967, by Dr. Christiaan Barnard and his colleagues in Cape Town, South Africa.21,22 Nearly one month later, in January 1968, the first heart transplantation was performed in North America by Dr. Norman Shumway, at Stanford University.21,22 For the past number of decades, advances in different areas related to heart transplantation, e.g., surgical techniques, patient management and immunosuppressive therapies, have helped improved the survival rate of patients significantly and established heart transplantation as the primary therapy for end-stage vital heart failure patients. To date, it has been estimated that over 85,000 heart transplantations have been performed around the world since the early 1980’s.21- 24 According to the latest report by the International Society of Heart and Lung Transplantation (ISHLT), approximately 4000 heart transplantations were reported in 2009 alone.24,25 1.2.2 Immunosuppressive therapies After heart transplantation, recipient’s immune system can recognize the transplanted heart, i.e., cardiac allograft, as foreign and react against it.26,27 This immune (or ‘alloimmune’) response to nonself antigens expressed by the donor tissue can lead to the injury and dysfunction of the allograft, which can present as rejection episodes.26,27 During the rejection process, immune cells such as T cells, B cells, natural killer cells and macrophages can undergo activation, proliferation, and infiltrate the cardiac allograft, causing injury to myocytes and tissues via cellular and antibody-mediated mechanisms.28 The mechanistic details regarding acute and chronic cardiac allograft rejections are described in more details later in this chapter. In order to suppress and avoid the possible alloimmune response, e.g., immune cell activation and proliferation, and associated inflammation post-transplantation, cardiac transplant patients are routinely put on immunosuppressive regimes, which typically serve several major purposes: 1) to avoid early acute rejection post-transplantation by acting as a 7 prophylaxis, i.e., induction therapy, 2) to treat acute rejection episodes that does occur, i.e., anti-rejection therapy, and 3) to help maintain the allograft long term as a prophylaxis, i.e., maintenance therapy.29,30 The specific dosage, type, and duration of immunosuppression is also optimized based on individual patients in order to provide the desired outcome listed above, while minimizing the potential for adverse effects and opportunistic infections.29-31 Majority of the immunosuppressive medications used clinically today belongs to one of the following therapeutic groups: i) calcineurin inhibitors, ii) antiproliferative agents / antimetabolites, and iii) steroids.32 Selected examples of the more commonly used drug from each therapeutic class are discussed below. Briefly, calcineurin inhibitors (CI) such as cyclosporine (CSA) and tacrolimus (TAC) are able to enter the immune cells through mechanisms such as diffusion, and bind to specific immunophilins, i.e., CSA and TAC to cyclophilin and FC binding protein (FKBP-12), respectively. The resulting immunophilin-drug complex can bind and inhibit calcineurin; this in turn inhibits the transcription of a number of cytokine genes and thus prevents the formation of cytokines such as interleukin (IL)-2 and others.32,33 Antiproliferative agents, in general, inhibit the differentiation and proliferation of immunocompetent lymphocytes post-(allo)antigen recognition, thus suppressing the immune response post-transplantation.27,34 Certain antiproliferative agents can interfere with metabolic pathways or be incorporated in synthesis pathways to generate faulty molecules that are structurally similar to metabolites essential for the differentiation and division of immune cells.27,34 These specific antiproliferatives are sometimes also referred to as antimetabolites, e.g., azathioprine (AZA) and mycophenolate mofetil (MMF).27,34 In essence, AZA and MMF inhibits the production of purine and guanine, respectively, which are compounds necessary in cell proliferation.27,32,34 In blood, AZA is converted by plasma esterases or glutathione to 6- mercaptopurine (6-MP) and eventually to 6-thio-inosine-5’-monophosphate, a purine analog that once incorporated into the cellular DNA can inhibits its synthesis.27,32,34 By inhibiting DNA synthesis, AZA can effectively interfere with the proliferation of immune cells such as lymphocytes (and in fact, all rapidly dividing cells), particularly during phases of immune 8 response when these cells require proper nucleotide synthesis to undergo rapid cell division.27,32,34 Similar to AZA, MMF also interferes with purine metabolism and inhibits the synthesis of DNA, as well as RNA.27,32,34 However, unlike AZA which works more broadly on all cells undergoing rapid nucleotide synthesis and division/proliferation, MMF, acts more specifically on lymphocytes, i.e., T and B cells.27,32,34 In the body, MMF is metabolized to mycophenolic acid, which inhibits inosine monophosphate (IMP), a key enzyme involved in regulation of guanosine monophosphate (GMP) and de novo pathway for synthesis of purines, e.g., guanine.27,32,34 With the exception of lymphocytes, most cells in the human body are able to utilize two pathways, i.e., de novo and salvage, for the production of purine nucleotides.27,32,34 As such, MMF affects primarily the DNA replication and proliferation of lymphocytes.27,32,34 Another example of antiproliferative agent is sirolimus (SIR). Similar to calcineurin inhibitors (CI) such as CSA and TAC, SIR is also able to bind to immunophilins.27,32,34 In contrast to CSA and TAC which inhibits calcineurin, the immunophilin-SIR (i.e., FKBP-sirolimus) drug complex binds and inhibits the mammalian target of rapamycin (mTOR) instead.27,32,34 mTOR is a key kinase downstream of IL-2 receptor (IL2R) signaling pathway that assists in the production of proteins which regulate cell cycle progression. Thus, through inhibition of mTOR, SIR can suppress IL-2R signaling mediated activation and proliferation of immune cells, e.g., T and B cells.27,32,34 Corticosteroids (or referred to simply as steroids) are commonly given to cardiac transplant recipients because of their immunosuppressive and anti-inflammatory properties.27,32,34 One of the more frequently used types of corticosteroids, glucocorticoids (e.g., prednisolone, active metabolite of prednisone), are able to enter cells via diffusion and intracellularly bind to cytoplasmic glucocorticoid receptors (GR).27,32,34 The resulting GR- glucocorticoid complexes can translocate to the nucleus, act as a transcription factor via binding to different regulatory elements, e.g., glucocorticoid response element, and transcriptionally regulate expression of different genes that will ultimately affect the immune and inflammatory response. One example of this is the IκB gene, which can be transcriptional upregulated by the GR-glucocorticoid complex. This can eventually lead to increased production of IκB enzyme complex, which is able to inhibit nuclear factor (NF) κB and activator protein-1 (AP-1), both of which considered key players in the activation of lymphocytes.27,32,34 9 Given the diversity of mechanisms, pathways and effects associated with each type of immunosuppressive agents, a combination of drugs is typically used in the management of cardiac transplant recipients.27,30 Currently, the most commonly used immunosuppressive regime for cardiac transplant recipients post-transplantation, i.e., maintenance immunosuppression, is widely referred to as the triple drug therapy, which involves a calcineurin inhibitor, e.g., CSA, a corticosteroid, e.g., prednisone, and an antiproliferative/antimetabolite agent, e.g., MMF.27,30 1.2.3 Survival rate of heart transplantation patients Advancements in our understanding and utilization of immunosuppressive agents have helped improve the clinical outcome for heart transplant patients. In particular, the short-term survival rate (i.e., 1-year) for cardiac recipients has now surpassed 85%,24,35,36 where as the 5 year survival rate is roughly 75%.37 According to the ISHLT 2011 report, the half life, defined as time at which half of the cardiac recipients are still alive, is approximately 10 years.24 However, despite the substantial progress in allograft and patient survival, the detection of allograft rejection (described later) remains to be one of the most important yet unsettled areas of cardiac transplantation.38 10 1.3 “Post-“heart transplantation hurdles 1.3.1 Overview of cardiac allograft rejections As noted earlier, the cardiac allograft rejection process can be generally described as the result of recipient’s immune response to nonself antigens expressed by the donor tissue, i.e., cardiac allograft.39 Moreover, cardiac rejection can be further distinguished, in part based on the timeframe of events and histological evidence, into three categories: hyperacute, acute, and chronic.26,27 Hyperacute rejection refers specifically to the process in which preformed donor- specific antibodies from the recipient’s body react rapidly and intensely, usually within minutes to hours post-transplant, against the cardiac allograft.26,27 This is almost always a lethal process but fortunately, with the use of blood-type matching and crossmatch assays to ensure there are no donor-specific cytotoxic antibodies present in the recipient’s serum prior to transplantation, hyperacute rejections are largely preventable.26,27 In contrast, acute rejection is a predominantly cell-mediated process that can also involve the presence of acquired antibody-mediated response. Although acute rejection episodes typically occur within 6 months to 1 year post-transplant, it can also take place at later timepoints.26,27,37 Chronic rejection, or more specifically, development of cardiac allograft vasculopathy (CAV) as an expression of chronic rejection, is a relatively more insidious process that tends to develop over months and years after transplantation.27,40 It thought to be mediated by immune and non-immune processes that can ultimately lead to obliterative vasculopathy and failure of the cardiac allograft.41-45 In line with the premise of my research work, the following sections will further describe acute rejection and CAV as an expression of chronic rejection. 11 1.3.2 Acute cardiac allograft rejection 1.3.2.1 Occurrence and diagnosis of acute cardiac allograft rejection It has been estimated that between 20% to 50% of patients may experience acute cardiac allograft rejection (AR) at least once within one year post-surgery, despite the use of immunosuppressive therapy.46 Nevertheless, improvements in surgical management, as well as refinement in combination immunosuppressive regimes and post-transplantation patient management have helped decreased the rate of acute rejection over the last number of years.24,25 Between 1994 and 2000, approximately 59% of cardiac transplant recipients required hospitalization for rejection treatment within 5 years post-transplant. According to the most recent report by the ISHLT, this percentage has dropped to approximately 45%, i.e., incidence of rejection within 5 years post-transplant that ultimately required patient hospitalization.24,25 Clinically, cardiac recipients undergoing acute cardiac rejection may be asymptomatic or present with a range of symptoms, depending on the severity and duration of the rejection.26,27 These symptoms can be as mild as shortness of breath, to arrhythmias that could transpire to syncope and in some cases, cardiac arrest.26,27 Given the high likelihood of AR (particularly during the first year post-transplantation) and the potential severity of allograft dysfunction, routine screening for detection and diagnosis of AR remains one of the most important areas of cardiac transplantation research. The current gold standard for definitive diagnosis of acute cardiac allograft rejection relies primarily on endomyocardial biopsies (EMB).3 Routine EMBs are generally performed every week during the first month post- transplant, then once every two weeks/bi-weekly during the second month, followed by once every month until the end of year one.26 Depending on the transplantation program, the subsequent EMBs after the first year post-transplant can range from every 6 months to every year.26 EMBs are also indicated when patients are suspected of having acute cardiac rejection based on clinical symptoms. The EMB procedure is invasive, and involves the use of cardiac catheterization and a flexible bioptome, which is then guided transvenously, typically via the right internal jugular vein (sometimes the femoral vein), into the right ventricle of the 12 allograft.3,4,26 Once inside, the bioptome excise generally a minimum of 3, to preferably 4 or more, 1-mm3 pieces of the myocardium.47 The collected EMBs are then processed for the histological evaluation by pathologists, and graded based on the 2004 ISHLT grading system. Based on the grading system that was originally adopted by the ISHLT in 1990, the revised 2004 version classify acute (cellular) cardiac allograft rejection into three grades (0R to 3R, with 3R being the most severe).47 Briefly, Grade 0R represents no rejection. Grade 1R is considered mild rejection, evident by presence of immune cell infiltrates in the interstitial and/or perivascular space, with no more than 1 focus of myocyte damage.47 Grade 2R, or moderate rejection, is characterized by the presence of two or more foci of cellular infiltrate with associated myocyte damage. In grade 3R, or severe rejection, diffuse cellular infiltrate along with multiple foci of myocyte damage can be observed; vasculitis, edema or hemorrhage may also be present.47 In the same ISHLT consensus report (“Revision of the 1990 Working Formulation for the Standardization of Nomenclature in the Diagnosis of Heart Rejection”), a separate histological evaluation guideline and grading nomenclature was also recommended for the recently emphasized form of cardiac rejection, called antibody-mediated rejection (AMR).47,48 There remain some controversies with regards to the specific definition of cardiac AMR, although a consensus statement has been recently described by the ISHLT.48,49 Furthermore, the significance of mixed acute cellular rejection and AMR remains to be elucidated.49 The details regarding diagnostic challenges and management of AMR episodes have been extensively described in a recent review by Kittleson and Kobashigawa.50 Currently, however, the diagnosis of acute cellular rejection based on EMB is still considered the more effective and widely adopted surrogate for the outcome and wellbeing of transplanted hearts.49 For the purpose of this dissertation and the research projects described within, the term “acute rejection” (AR), unless stated otherwise, refers specifically to acute cellular rejection as defined by the criteria for 2004 ISHLT grade 2R and above (or grade 3A and above based on the older 1990 ISHLT grading system; see Figure 1). 13 EMB Examples ISHLT 1990 Grade ISHLT 2004 Grade Characteristics 0 0R Normal EMB showing no evidence of cellular infiltration Example: No evidence of mononuclear (lymphocytes/ macrophages) inflammation or myocyte damage. 1A 1B 2 1R Interstitial and/or perivascular infiltrate with up to 1 focus of myocyte damage Example: Diffuse mononuclear cell infiltrate with an interstitial pattern of lymphocytes between and around myocytes without associated myocyte damage. 3A 2R Two or more foci of infiltrate with associated myocyte damage Example: Low power view showing three foci of damaging mononuclear cell infiltrate with normal myocardium intervening. 4 3R Diffuse infiltrate with multifocal myocyte damage ± edema, ± hemorrhage ± vasculitis Example: Severe acute rejection with widespread myocyte damage and some necrosis. Figure 1. Histological diagnosis and grading of acute cardiac rejection. Examples of EMBs corresponding to the different ISHLT rejection grades are shown. Images shown are reproduced from the official ISHLT guideline,47 with permission from the publisher and the Journal of Heart and Lung Transplantation. All EMB images shown are based on Hematoxylin and Eosin (H&E) staining. 14 1.3.2.2 Pathogenesis of acute cardiac allograft rejection As noted earlier, histologically, acute cellular rejection, i.e., ISHLT grading 2R or above, is typically characterized by the presence of inflammatory cellular infiltrates with associated myocyte damage.47 The cellular infiltrates are generally comprised of lymphocytes, macrophages, and occasionally eosinophils.47,51 Mechanistically, acute cardiac rejection is thought to involve both cellular and humoral processes.52,53 After the heart transplantation, donor and recipient-derived antigen presenting cells (APCs; e.g., dendritic cells) can trigger direct and indirect allorecognition, respectively. In direct allorecognition, the intact foreign donor MHC antigens and peptides presented on the surface of donor APCs are recognized by recipient T cells. Specifically, the donor organ-derived APCs can migrate from the allograft to the recipient’s lymphoid tissues, where they activate, through the direct pathway, CD4+ and CD8+ T cells.45,52,53 On the other hand, in indirect allorecognition, the recipient APCs first uptake and process the donor MHC antigens, before presenting the donor-derived allopeptides to recipient T cells.45,52,53 While both direct and indirect pathways are activated as part of the alloimmune response post-transplant, it is thought that the direct pathway is primarily responsible for initiating the acute cellular rejection process, whereas the indirect pathway has been linked more so to the development of CAV and chronic rejection.45,52,53 After the initiating allorecognition event, activated T cells can undergo clonal expansion, as well as produce cytokines, e.g., IL-2, IL-4, IL-5, IFN-γ, and chemokines, thus creating an inflammatory milieu which further promotes the recruitment and activation of additional immune cells.45,52,53 Activated immune cells, e.g., T cells, B cells, macrophages and dendritic cells (DC), can all interact dynamically with the complement system.54 A brief description of the different immune effectors/regulators, including complement system, and their roles in the context of cardiac allograft rejection is summarized in Table 1. 15 Table 1. Roles of immune effectors and complement system in allograft rejection Key immune Effectors and Regulators Examples of roles and involvement in cardiac rejection T cells CD8+ cytotoxic T cells • Cytotoxic T cell-mediated direct lysis of target cardiac allograft cells via perforin and granzyme-mediated pathways • Production of inflammatory cytokines CD4+ T helper cells • Involved in activation of CD8+ T cells and B cells • Production of inflammatory cytokine, e.g., IL-4, IL-5 T helper 17 cells • Production of IL-17, as well as IL-21 and IL-22; IL-17 is considered to have proinflammatory properties Regulatory T cells • Homeostatic controllers of inflammation partly through production of anti-inflammatory cytokines, e.g., IL-10 and TGF-β B cells • Production of alloantibodies • Alloantibody(alloAb)-mediated and alloAb-independent mechanisms, e.g., regulation of T-cell mediated inflammation, cytokine production Neutrophils, Mast cells, NK cells, DC and Macrophages • Early responders as part of the innate immunity, e.g., infiltration of recipient NK cells and macrophages into cardiac allograft • Production of cytokines, e.g., IL-1, IL-6, TNFα, IFNγ • Antigen presentation and activation of other immune cells • Perforin and granzyme-mediated attack, i.e., NK cells Complement system • Brain death and ischemia/reperfusion (I/R), both part of the organ procurement/transplant process, favor a proinflammatory condition and activation of the complement system cascade • Active complement fragments can act as inflammatory mediators, e.g., C3a and C5a, and modulate immune cell response • Formation of membrane attack complex (MAC), which can trigger subsequent damage and destruction of the targeted (foreign) cells, e.g., myocytes of the cardiac allograft 16 Together, the major pro-inflammatory mechanisms and immunological events that are triggered post-transplantation can result in inflammation and immune cell infiltration in the myocardium, as observed in EMBs taken during acute rejection episodes.37,45,52,53 Consequently, insults to the local myocardial tissue, through mechanisms such as cytotoxic T cell-mediated direct lysis, complement cascade activation and B cell alloantibody production, can cause myocyte damage and necrosis, ultimately leading to functional impairment of the cardiac allograft, or “cardiac rejection” (summarized in Figure 2) . 17 Figure 2. Major contributing mechanisms of myocardial inflammation in the context of cardiac allograft transplantation. Both pre- and post-transplantation events can contribute as initiators of downstream complement-, alloantibody- and cell-mediated pathways. These major pathways can act synergistically to create an inflammatory milieu and eventually lead to the necrosis and apoptosis of cells in the cardiac allograft, e.g., endothelial and parenchymal cells. This can further exacerbate the myocardial inflammation observed in acute and chronic cardiac allograft rejections. Abbreviations: CTL=cytotoxic T lymphocytes; MBL=mannan-binding lectin; I/R=ischemia/reperfusion. 18 1.3.3 Cardiac allograft vasculopathy (CAV) 1.3.3.1 Occurrence and diagnosis of CAV as an expression of chronic rejection The long term (i.e. 10-year) survival rate of the heart recipients is only about 50%,35,55 and is largely limited by the development of cardiac allograft vasculopathy (CAV) as an expression of chronic rejection (CR). CAV affects approximately half of cardiac transplant recipients within the first several years post-transplantation, and is responsible for up to 15% of deaths in cardiac allograft recipients after they have survived the first year post-transplant.56 Given the prevalence and the significance of CAV, cardiac transplant recipients undergo routine tests, typically at least once a year, to monitor the health of the transplanted heart. Currently, the most widely used modality for the diagnosis of CAV remains coronary angiography. In this procedure, a cardiac catheter is inserted into the patient’s artery, commonly through femoral artery in the groin area, and threaded into the coronary arteries of the cardiac allograft. A radio-opaque contrast material is then injected through the catheter, and X-rays images are taken via fluoroscopy to allow visualization and evaluation of the allograft coronary blood vessels, e.g., epicardial coronary arteries and braches. The screening and diagnosis of CAV is largely based on the detection of narrowing, or stenosis, of coronary arteries, i.e., blood vessels which supply oxygenated blood to the heart.45,57,58 The severity of stenosis is often expressed as percentage diameter stenosis (%DS),59 as shown in Figure 3. In contrast to the diagnosis of acute cellular cardiac rejection, where specific histological criteria and grading guideline have been established and widely adopted internationally for years, there is relatively less consensus on the classification of CAV. A working formulation of nomenclature for CAV has recently been put forth by the ISHLT.60 However, the cut-off values used in the proposed guideline to define the mildest form of CAV are quite high, e.g., angiographic evidence of stenosis in the primary coronary arteries up to 70% diameter reduction.60 From a research perspective, a lower stenosis threshold to define ‘significant’ CAV may be required in order to identify highly sensitivity and specific biomarkers for screening and diagnostic purposes. This ‘fit-for-purpose’ approach in the context of biomarker development,61 e.g., using more ‘extreme’ patient phenotypes for biomarker discovery, is further demonstrated and explained in later research and discussion chapters throughout this dissertation. 19 Figure 3. Percentage diameter stenosis of coronary artery The illustration above represents a simplified view of a longitudinal section of a coronary artery segment. The severity of coronary artery stenosis is often measured and expressed as percentage of diameter stenosis (%DS), as defined by the following formula: %DS = [ (RD – MLD) / RD ] x 100 where RD is the reference diameter (an average of the normal region), and MLD is the minimum lumen diameter. 20 1.3.3.2 Pathogenesis of cardiac allograft vasculopathy The histopathological features of CAV were first described in human cardiac allografts in 1970 as a diffuse, occlusive vascular condition involving coronary intimal proliferation and obliterative changes in the coronary arteries.62-66 Decades later, it has been recognized that CAV is a complex, multifactorial condition involving both immunologic and non-immunologic contributing factors and mechanisms.40,45,67,68 Although the exact pathobiology remains unclear, it is widely accepted that CAV is initiated by a combination of allogeneic response to the allograft and ischemia/reperfusion (I/R) injury, resulting in damage to the endothelial cells (ECs).40,69-73 In the progression of CAV, the initial activation, injury, dysfunction or destruction of ECs is typically followed by the subsequent activation, migration and proliferation of vascular SMCs, along with elaboration of cytokines and extracellular matrix (ECM) protein.40,44,45,74 These events are thought to be central to the onset and development of progressive luminal narrowing and eventual impaired vascular function of the allograft.40,44,45,74 Multiple processes can contribute to the injury of ECs during the cardiac transplantation process, e.g., surgery-related mechanical damage, hypoxia and I/R-induced complement system mediated injury.40,45,74,75 Post-transplantation, immunologic responses to the donor vasculature, via activated immune cells and circulating inflammatory cytokines and complement fragments, can further exacerbate the damage and activation of the endothelium.40,45,74 Activated ECs can up-regulate the expression of adhesion molecules (e.g., intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1) and P-/E-selectins),74,76-81 as well as secrete cytokines such as IL-1, IL-6, and TNF-α.55,74,82,83 This can further promote the recruitment and attachment of leukocytes to ECs, and create an inflammatory milieu. The same immune mechanisms described earlier, e.g., cell-, antibody- or complement- mediated mechanisms, can also lead to the apoptosis of target cells in the allograft, including ECs.84 Apoptosis of ECs, either through acute alloimmune response (i.e., ‘acute rejection’) or low grade, persistent inflammatory response (i.e., ‘chronic rejection’), are also thought to play a pivotal role in the development of CAV.84 In human heart transplants, apoptosis of ECs are observed in graft coronary arteries showing early signs of CAV.84,85 21 Another key feature seen in the development of CAV is the progressive neointimal accumulation of vascular SMCs.74,86-88 Parallel to this, a decrease in medial SMCs can also often be observed.74,89,90 This reduction is believed to be, in part, due to the injury and/or death of SMCs through similar mechanisms that affects the ECs, e.g., I/R injury, cytotoxic soluble mediators or receptor-based apoptosis.69,74,91-95 Several processes are thought to contribute to neointimal hyperplasia: i) activation of medial SMC in response to soluble mediators such as growth factor TGF-β, and cytokines such as IFN-α and IL-1, ii) migration of SMC (in response to chemokine gradients), and iii) proliferation of SMC in the intima.74,96,97 Further, it has been suggested that SMCs undergoing response to injury can transform from a more (differentiated) contractile state to a more (dedifferentiated) synthetic one.40,44,68 The synthetic state SMCs are capable of migrating from the media to the intima, proliferate and synthesize ECM components, e.g., proteoglycans.40,73 The latter two effects are considered major contributors of obliterative intimal thickening seen in the vessels with CAV.40,67,98,99 More recently, it has been proposed that circulating smooth muscle progenitor cells can be recruited to the injured vascular sites, and are also able to contribute to the progression of CAV through similar effects.40,43,45,72 Key events central to the onset and progression of CAV based on current literature are summarized in Figure 4.40,45,74,77-81,100-105 22 Figure 4. Major events central to the onset and development of CAV. It is widely accepted that the development of CAV is contributed by both immune and non- immune factors. (I) Peri-transplantation effects, e.g., surgical-, I/R, or hypoxia-mediated injury, as well as post-transplantation effects involving cell-, antibody- and complement-mediated mechanism, can damage the cardiac allograft vasculature, e.g., endothelium, lead to the generation of an inflammatory milieu. (II) ECs that are activated injured, or undergoing apoptosis can secrete additional soluble mediators, which can act in a paracrine fashion and affect neighboring ECs, or induce neighboring medial SMCs. (III) The inflammatory environment can lead to the injury, activation or apoptosis of SMCs in the allograft. Once activated, SMCs can migrate to the intimal layer, proliferate, as well as synthesize ECM such as proteoglycans. (IV) The cumulative effects by the cells in the vasculature (e.g., activation and proliferation of ECs, fibroblasts and SMCs; excess production and deposition of ECM), as part of the response- to-injury process, help contribute to the thickening of the intima, remodeling of the vessel wall and consequently, progression of CAV. Abbreviations: EC=endothelial cell; SMC=smooth muscle cell; ECM=extracellular matrix; CK=cytokine; GF=growth factor; ROS=reactive oxygen species 23 1.4 Overview of biomarkers The term biological marker was first popularized by the United States National Library of Science in 1989 when it was introduced as a Medical Subject Heading (MeSH; such as that used by PubMed article database; www.pubmed.com) to help categorize and index journal articles.106 At the time, biological markers (or now referred to simply as ‘biomarkers’), was defined as “measurable and quantifiable biological parameters (e.g., specific enzyme concentration, specific hormone concentration, specific gene phenotype distribution in a population, presence of biological substances) which serve as indices for health- and physiology-related assessments, such as disease risk, psychiatric disorders, environmental exposure and its effects, disease diagnosis, metabolic processes, substance abuse, pregnancy, cell line development, epidemiologic studies, etc”. 106 Decades later, a similar definition was adopted by the United States Food and Drug Administration (FDA) and National Institute of Health (NIH) working group, and biomarker was defined in the pharmacogenomics guidance published in 2001 as “characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention”.107 Similarly, according to Health Canada, a biomarker is described as any “measurable characteristic that is an indicator of normal biologic processes, pathogenic processes, and/or response to therapeutic or other interventions.”108 It is interesting to note that, by the current definitions of biomarkers, it has been argued that one can approach the biomarker discovery process in an existential fashion. Such a perspective means that, to a significant extent, we do not necessarily need to know the detailed identity, function or biological relationship of a given biomarker or biomarker set to be reproducibly and objectively useful, and clinically valuable.61,109 Historically, biomarkers have been identified, sometimes serendipitously, through targeted studies of physiological processes. Many biomarkers discovered decades ago in the pre- ‘omics’ era are still being used in medical practice, such as aspartate transaminase (AST), C- reactive protein (CRP), and cardiac troponin I (cTnI).61 Presently however, given the recent 24 explosion of high-performance ‘omic’ technologies – genomics and proteomics, among others – the rate at which biomarker candidates are being discovered is now faster than ever. Specific examples of the ‘omic’ platforms utilized in the scope of this dissertation and the research projects within are described in the Appendices – Assays and ‘Omics’ Methodologies section. In this dissertation, biomarkers are simply defined as distinctive, objectively measured biological or biologically derived indicator of a process, event, or condition. 25 CHAPTER 2: Dissertation overview 26 2.1 Rationale of research proposal Currently, cardiac transplantation is considered the primary therapy for patients with end-stage vital heart failure. In the context of pre-transplantation research, studies have suggested that unique molecular expression profiles are associated with different conditions i.e., cardiomyopathies, that can lead to HF and the eventual need for a cardiac transplantation. However, it is not clear whether these molecular expression profiles remain unique at the point just before transplantation (i.e., end-stage heart failure). Once the patient has received the cardiac allograft, the detection of acute and chronic rejections (as cardiac allograft vasculopathy; CAV) remains to be some of the most important yet unsettled areas of cardiac transplantation research. As described earlier in the introduction, the current gold standards for the diagnosis and monitoring of acute rejection and CAV are invasive in nature with risk for complications. From a clinical perspective, more accurate, minimally-invasive alternatives for the diagnosis of AR and CAV are clearly desirable. While traditional (molecular), blood-based cardiac biomarkers such as Creatine Kinase / CK-MB or Troponin I and T (cTnI and cTnT) can be quantitatively measured to reflect the severity of cardiac injury,4,110,111 they are not specific to the cause of the injury. Classic markers such as troponins are also thought to lack diagnostic sensitivity during the early post-operative period.35,112,113 As such, renewed efforts to improve long-term survival through enhanced monitoring and diagnosis of AR episodes and development of CAV have directed attention towards the search of better biomarkers.114 The availability of a diverse range of high-performance ‘omic’ technologies has meant new and exciting opportunities for biomarker discovery. It has been suggested that, rather than finding a single ‘silver bullet’ biomarker of rejection, multiple biomarker may be necessary to achieve the level of diagnostic power that could potentially be clinically useful one day.36 Indeed, recent studies have shown evidence in this regard.115,116 However, more work is required to examine the use of a high-throughput, unbiased approach to identify biomarkers for the diagnosis of acute and chronic human heart cardiac allograft rejection. 27 2.2 Overview of research proposal In this dissertation I focus on the discovery and examination of biomarkers, using current state-of-the-art ‘omics’ technologies, at three specific timepoints and relevant conditions of interest in the context of heart transplantation, including ESHF, AR and CAV. Specifically, I explore the potential utility of these biomarkers from a clinical point of view, as well as investigate their biological plausibility and implications from a pathobiological perspective. I start the research by examining the molecular profile of patients at the ‘end-stage’ of their chronic heart failure, in order to determine whether there is any unique ‘omic’ profile that associates with them. In particular, I want to determine i) whether HF of non-ischemic or ischemic origin present with unique biomarker profiles and, ii) what molecular and biological processes are perturbed in these CHF patients that underlie the need for eventual heart transplantation? In my next two research aims, I transition from the pre- to the post-transplantation phase. The major portion of my work then focuses on two critical events after transplantation that are detrimental to the wellbeing and long-term survival of patients – acute rejection and CAV as an expression of chronic rejection. The main questions I attempt to answer here are: i) are there biomarkers that are significantly different between patients undergoing these conditions, versus those who are not? ii) can a specific combination or panel of biomarkers work together and be potentially utilized for the diagnosis or monitoring of AR and CAV? iii) what can the identified biomarkers inform us regarding the underlying pathobiology of these conditions? Finally, I will also examine and discuss the caveats and potential challenges of biomarker research and translation of bench findings to the clinic. 28 2.3 Central hypothesis Peripheral blood-derived molecular biomarker panels provide the means for sensitive and specific diagnosis of acute and chronic cardiac allograft rejection, as well as help gain insight into the underlying mechanisms. 2.4 Specific aims 1) To compare the blood-based molecular profiles between patients with heart failure of ischemic or non-ischemic origin, and determine the major biological processes perturbed in these patients at the ‘end-stage’ of their chronic heart failure. 2) To determine if there is a unique gene expression profile in the whole blood that is specific to the acute cardiac allograft rejection; and if such expression profile is present, determine its clinical utility and biological plausibility. 3) To determine if there is a unique proteomic signature in the plasma that is able to reflect the presence of significant coronary artery stenosis, a strong indicator of CAV development, and examine its implications for CAV screening and monitoring. 29 CHAPTER 3: Molecular signatures of end-stage heart failure patients prior to cardiac transplantation 30 3.1 Background Heart failure (HF), a progressive clinical syndrome characterized by inability of the heart to adequately pump blood to meet metabolic demands of the body,1,2 can result from virtually any form of cardiac disorders.2 Yet, despite being a multifactorial pathological condition with diverse etiologies, initial HF can persist and progress into chronic heart failure (CHF) and ultimately ‘end-stage’ heart failure (ESHF). The latter is often the final pathway leading to death or the need for cardiac transplantation.1,117 According to the 2011 report from the registry of the ISHLT (based on data between January 2005 to June 2010), the two major etiologies of heart disease that preceded the need for heart transplantation in adults are in fact, ischemic heart disease/cardiomyopathy (37.7%) and non-ischemic cardiomyopathy (53.3%).24 Conditions such as ischemic heart disease (IHD) and non-ischemic cardiomyopathy (NICM) may follow different timelines leading to the onset of HF and eventual progression to CHF.1 While CHF primarily impacts the cardiovascular system, it is often considered a multi- system disorder due to its interplay with musculoskeletal, neurohormonal, metabolic, immunological, and other systems of the body.118-120 The complexity and systemic nature of CHF has made it an intriguing and attractive candidate for ‘omic’ investigations. Since the first human microarray analysis of ESHF,121 research to date have generally involved comparing gene expression changes in failing versus non-failing hearts (NFH),122-124 pre-left ventricular assist device (pre-LVAD) versus post-LVAD implantation,125,126 or in different causes of HF (i.e., cardiomyopathies).127,128 These published work have typically been based on cardiac tissue samples.2 However, previous studies have reported high concordance between (gene) expression profiles derived from disease tissue and PBMC,129 and proposed the use of whole blood and PBMC as viable substitutes.130 3.2 Rationale It is currently unclear whether unique whole blood-derived signatures exist in patients with end-stage HF of either non-ischemic or ischemic origin. Although some pilot studies have documented significant differences in genomic signatures, albeit derived from cardiac tissues, 31 between the different etiologies of HF,128 others have had less success in similar efforts.131 It is also unclear what molecular and biological processes are perturbed in these CHF patients that underlie the need for eventual heart transplantation In this chapter, I first compare IHD and NICM using a high-throughput holistic approach and attempt to shed light on the issues noted earlier, and answer the following questions: Is there a significant difference in the peripheral blood genomic and proteomic profiles between IHD and NICM patients at the time of ESHF? As well, what major biological processes and networks are dysregulated in CHF patients in general, relative to individuals with NCF? 32 3.3 Materials and methods 3.3.1 Subjects and specimens The experiments in this chapter were conducted under the Biomarkers in Transplantation (BiT) initiative, which was approved by the Providence Health Care Research Ethics Board.132 Peripheral blood samples used in this research chapter were collected within, on average, two weeks just prior to transplantation (i.e., “end-stage”) from 29 cardiac transplant patients. Blood samples were also collected from 20 healthy individuals with normal cardiac function (NCF). The cardiac transplant patients were divided into two groups based on the original clinical/pathological diagnoses: 1) Ischemic heart disease (IHD; n=16), or 2) Non-Ischemic Cardiomyopathy NICM (n=13). The IHD and NICM patients were grouped as the CHF cohort for the analysis comparing CHF and NCF (Table 2). Whole blood and plasma samples were collected in PAXgene and EDTA tubes, and analyzed using Affymetrix microarrays and iTRAQ proteomics, respectively. Additional descriptions of each of these techniques are provided in the Appendices – Assays and ‘Omics’ Methodologies section. 33 Table 2. Demographics of subject cohorts (CHF and NCF) Patient Demographics CHF NCF IHD NICM Age Range 52-70 26-64 26-65 Average 60 46 42 Percent Men 94 62 60 Race (Percent) Caucasian 94 100 80 Asian 6 - 15 Other - - 5 Ejection Fraction (Percent) Range 13-35 15-40 - Average 20 20 - New York Heart Association Class Range 2-4 2-4 - VO2 Max Range 6-15 10-18 - Average 10.0 13.4 - Length of Time with Heart Failure (Years) Range 0.4-20 0.7-15 - Average 7.6 7.3 - Co-morbidities (Percent) Chronic Kidney Disease 31 8 - Hypertension 31 8 - Dyslipidemia 38 0 - Diabetes Mellitus 13 8 - Cirrhosis 6 8 - Others 19 15 - No co-morbidities 25 69 - 34 3.3.2 Sample and data processing 3.3.2.1 Genomics A total of 26 CHF patient PAXgene whole blood samples (i.e., 15 IHD plus 11 NICM) and 20 NCF samples were available and used for the microarray analysis. Total RNA was isolated using PAXgeneTM RNA Kits as previously described.116,132 RNA samples were processed via reverse-transcription-in vitro transcription (RT-IVT) to generate labeled cRNA which were then fragmented for hybridization on the microarray for analysis. The microarray analysis was performed at the Microarray Core Laboratory at Children’s Hospital, Los Angeles using Affymetrix Human Genome U133 Plus 2.0 chips. The microarrays were checked for quality problems using affy v1.22.0 and affyPLM v1.20.0 BioConductor packages. 3.3.2.2 Proteomics In total, 22 CHF patients’ plasma samples (14 IHD and 8 NICM) were available for proteomic analysis. Plasma samples from 16 healthy individuals were pooled and served as the normal reference sample for each iTRAQ experimental run. Sample processing, data acquisition and analysis were carried out as previously described.133 Briefly, samples were processed via immuno-affinity chromatography (Genway Biotech; San Diego, CA), to deplete the 14 most abundant plasma proteins (albumin, fibrinogen, transferrin, IgG, IgA, IgM, haptoglobin, α2- macroglobulin, α1-acid glycoprotein, α1-antitrypsin, apolipoprotein-I, apolipoprotein-II, complement C3 and apolipoprotein B).133 Depleted plasma protein samples were labelled with iTRAQ reagents according to manufacturer’s protocol (Applied Biosystems; Foster City, CA). iTRAQ labelled peptides were analyzed by a 4800 MALDI TOF/TOF mass spectrometer (Applied Biosystems; Foster City, CA). Data was analyzed using ProteinPilot™ software v2.0 with the integrated Paragon™ Search and Pro Group™Algorithms searching against the International Protein Index (IPI HUMAN v3.39) database. 35 3.3.3 Analysis The statistical analysis of the genomic and proteomic data was performed using R version 2.9.0 and BioConductor version 2.4.135 The details described below are applicable to both comparisons in the study: IHD versus NICM and CHF versus NCF. 3.3.3.1 Genomics Background correction, normalization and summarization of the microarrays corresponding to the 29 CHF and 20 NCF were performed with Robust Multi-array Average (RMA) technique (affy BioConductor package v1.22.0), including log base 2 transformation of the data. To reduce noise, probe sets with relatively constant expression values across all samples, i.e., interquantile range <0.5, were eliminated from further analysis. The remaining probe sets were analyzed using a robust moderated t-test available in the limma BioConductor package, v2.18.0. Probe sets with a False Discovery Rate (FDR) <5% were considered differentially expressed. 3.3.3.2 Proteomics The data was log base two transformed. Protein group code algorithm was used to allow subsequent comparison and analysis of proteins, as protein ‘groups’ (PGs), across different iTRAQ experiments. Only protein groups (PG) detected in at least 2/3 of the samples in each comparison group were included in the analysis, as described previously.133 In the second analysis (CHF versus NCF), since the data for each of the CHF samples represent abundance expressed relative to the pooled normal reference sample (NCF), the PGs were analyzed using a one group robust moderated t-test for the null hypothesis that the mean relative abundance equals one. Tests results with FDR <5% were considered statistically significant. 3.3.3.3 Functional enrichment Functional enrichment of the statistically significant genes, proteins and metabolites was pursued using MetaCore (GeneGo Inc., www.genego.com), a web-based suite of tools designed for functional analysis of (-omics) experimental data. Through MetaCore, Gene Ontology (GO) and GeneGo process networks-based analyses were carried out. The functional ontologies used in the two analyses are unique and complementary; the former utilizes a 36 publicly available database (www.gene-ontology.org) and GO terms, while the latter uses a manually curated and annotated GeneGo process networks database of human gene/DNA and protein interactions. These networks reflect the interplay of molecules, as a group, in particular biological/molecular processes or pathways. The GO terms, GeneGo process networks and pathways identified with FDR <5% were considered significant. 37 3.4 Results 3.4.1 Ischemic heart disease (IHD) versus Non-ischemic cardiomyopathy (NICM) 3.4.1.1 Genomics After normalization, pre-filtering was carried out to eliminate probe sets with relatively constant expression levels across all samples. A total of 14,047 probe sets, representing 7,450 genes, remained and were analyzed via robust moderated t-test. Differentially expressed probe sets were not found between IHD and NICM at FDR <5%. 3.4.1.2 Proteomics A total of 125 PGs were detected in at least 2/3 of the samples within each group and included in the subsequent robust moderated t-test analysis. None of the PGs (FDR <5%) were differentially expressed between IHD and NICM samples. 38 3.4.2 Chronic heart failure (CHF) versus Normal cardiac function (NCF) 3.4.2.1 Genomics Following normalization and pre-filtering, a total of 14,047 probe sets remained and were subjected to a robust moderated t-test analysis. 7,426 probe sets included in the moderated t-test were found to be differentially expressed between CHF and NCF samples (FDR <5%), with 2,364 being up-regulated in CHF relative to NCF. The top 100 of these probe sets, corresponding to 75 genes, were visualized using a heatmap (Figure 5). Enrichment analysis was also carried out on the 7,426 probe sets, using GO-based and GeneGo process network-based functional ontologies to uncover differentially regulated biological/molecular processes and functions, as well as networks. To gain a complete, unbiased, global perspective on the processes and networks that are significantly regulated in CHF and NCF subjects, all the genes which we found to be differentially expressed were analyzed. The top 10 significantly enriched GO terms and GeneGo process networks (FDR <5%) are summarized in Figure 6 and Table 3. 39 Figure 5. Heatmap of the top 100 differentially expressed probe sets between chronic heart failure (CHF) and normal cardiac function (NCF). Each row represents 1 probe set and each column represents 1 sample. The CHF cohort is comprised of two groups of samples: ischemic heart disease (blue) and non-ischemic cardiomyopathy (green). NCF samples are shown in black. 40 Table 3. Top 10 process networks identified by MetaCore as statistically significant, based on the 7,426 differentially expressed probe sets between CHF and NCF. Process Networks Transcription mRNA processing Immune TCR signalling Signal transduction androgen receptor nuclear signalling Cytoskeleton regulation of cytoskeleton rearrangement Proteolysis/ubiquitin-proteasomal proteolysis Cytoskeleton actin filaments Apoptosis/apoptotic nucleus Proliferation lymphocyte proliferation Cell cycle G1-S interleukin regulation DNA damage checkpoint Abbreviation: TCR = T cell receptor Figure 6. Top 10 GO terms based on functional enrichment analysis of the 7,426 probe sets differentially expressed between CHF and NCF. The x-axis represents the statistical significance of each Gene Ontology term: -log(P value) 41 3.4.2.2 Proteomics 125 PGs were detected in at least 2/3 of the CHF samples. Of these, 71 PGs had a mean ratio significantly smaller or larger than one, relative to a pooled normal sample (FDR <5%), suggesting these PGs may be differentially abundant in CHF samples. Of these, 17 were observed at higher levels of abundance in CHF relative to the pooled normal reference sample. All statistically significant PGs were subjected to the same functional enrichment analyses as in the Genomics section. The significantly enriched GO terms and GeneGo process networks have been summarized in Table 4 and Figure 7 (FDR <5%). 42 Table 4. Top 10 process networks identified as statistically significant by MetaCore, based on the differentially expressed proteins between CHF and NCF. Process Networks Blood coagulation Inflammation kallikrein-kinin system Inflammation protein C signalling Inflammation complement system Cell adhesion / platelet-endothelium-leucocyte interactions Immune phagosome in antigen presentation Proteolysis ECM remodelling Cell adhesion integrin priming Immune phagocytosis Inflammation IL-6 signalling Abbreviation: ECM=extracellular matrix. The analysis was based on 71 PGs differentially expressed between CHF and NCF Figure 7. Top 10 GO terms based on functional enrichment analysis of the 71 PGs which showed differential concentrations between CHF and NCF. The x-axis represents the -log(P value) 43 3.5 Discussion The first key finding based on results from this study is that there was a lack of evidence of a difference in ‘omic’ profiles between end-stage IHD and NICM subjects. Using genomic and proteomic platforms, no differentially expressed genes or proteins between the IHD and NICM whole blood and plasma samples were observed. These results suggest that there may be a high level of similarity between the IHD and NICM subjects from an ‘omics’ profile perspective. Previous work by others has suggested substantial similarities in transcriptomic patterns of ICM and NICM.117,131 In a study by Steenman et al., no differentially expressed genes were found between ICM and NICM tissues, although a larger sample size would have been desirable.117 A prevailing view in the literature is that, irrespective of the distinct underlying etiology that initiated HF, the expression signatures in advanced/end-stage HF may be dominated by a final common pathway.123,130,131 Thus, it is possible that what I have observed is the result of a convergence of perturbations, reflected in the peripheral blood, obscuring initial upstream differences between IHD and NICM. Although this result may not necessarily represent convergence of myocardial events, gene expression changes in peripheral blood have been shown to correlate with the histological and functional status of the heart as demonstrated in the context of cardiac transplantation.136-139 In light of such findings, the next logical steps are to further utilize the results generated from the same platforms, and examine what major biological processes and networks are perturbed in CHF patients in general, relative to individuals with NCF. In contrast to the highly similar molecular signatures between IHD and NICM during end-stage of CHF, analysis of CHF versus NCF revealed significant differences. A total of 7,426 probe sets were found to be differentially expressed in the CHF relative to the NCF blood samples in the genomics analysis. Although this result is not completely surprising, it fully illustrates molecular dysregulation present in end-stage CHF patients relative to healthy controls, at least at the genomic level. Unsupervised hierarchical clustering was performed on the top 100 of these differentially regulated probe sets, and a heat map was created to visualize relative expression levels between CHF and NCF samples. As shown in Figure 5, based on 100 44 probe sets alone, there is a greater resemblance between IHD and NICM etiologies of HF than between CHF of either etiology versus NCF. Significant differences between CHF and NCF were also observed in the proteomic results, as 71 proteins were found to be being differentially expressed between the two groups. 3.5.1 Integration of biological information and interpretation The pathophysiology of HF involves systemic disturbances in a variety of biological processes. Using multiple discovery platforms and ontological databases has provided a global perspective on the interplay of diverse pathological processes underlying CHF. When functional enrichment of ‘‘-omic’’ profiles was carried out using MetaCore, the top 10 significant GO terms identified (Figure 6) based on the 7,426 differentially expressed probe sets suggest metabolic dysregulation manifest in the peripheral blood of CHF patients. Perturbations in cardiac energy metabolism are generally accepted to play a role in the progression of HF.140,141 Other significantly regulated GO functions and GeneGo networks (Figure 7; Table 3 and Table 4) also reinforce existing knowledge regarding CHF. Specifically, the functional enrichment results demonstrated that majority of the differentially expressed genes and proteins found in the CHF samples fall within one of the following categories: 3.5.1.1 Response to cardiac damage/wound healing response (wound healing, extracellular matrix (ECM) remodeling, cytoskeleton regulation) A key feature of HF progression is adverse structural remodeling of the myocardium. I noted a prominent overrepresentation of differentially expressed genes and proteins associated with cytoskeleton regulation and the remodeling process in CHF patients (Figure 7; Table 3 and Table 4). As discussed by Liew and colleagues, damage to the cardiac ECM and cytoskeleton is common during the remodeling process.2 The differentially expressed plasma proteins detected are potentially derived from multiple organs, including the failing heart and other organs affected during the course of the HF development.142 These plasma proteins can reflect soluble factors arising from remodeling, as well as inflammation and immune responses (Table 4). Braunwald, in a recent review, suggested the use of ECM breakdown and remodeling- related molecules as biomarkers of HF.118 In fact, the linkage between CHF and matrix/ 45 cytoskeletal, as well as proteolysis/stress, genes and proteins has also been described in studies comparing failing and nonfailing human hearts.143 It is also worth noting that circulating cellular components of the peripheral blood are contributors to the expression profiles detected on microarrays. Given that inflammation and immune activation are also thought to play a major role in CHF,120 PBMC-derived gene expression changes are likely to be significant. Thus, it is possible that some of the differentially expressed genes related to cytoskeleton regulation were a consequence of immunological events underlying CHF arising from immune cell activation and cytoskeletal rearrangement in the circulating blood cells of CHF patients. 3.5.1.2 Inflammation/immune response (IL-6 signaling, kallikrein-kinin system) Involvement of immune system and inflammatory mediators in CHF has been previously suggested in the literature by Fildes et al. and others.144 The IL-6 signaling network was revealed as statistically significant in this study (Table 4). IL-6 is known to play a multitude of roles, whereas it has been associated with myocardial dysfunction and muscle wasting, it has also been shown to induce myocyte hypertrophy and inhibit cardiac myocyte apoptosis.145 Increased concentrations of IL-6 in the plasma and myocardium of CHF patients have also been observed.146 Activation of kallikrein-kinin system (Table 4) has been shown to be involved in the intramyocardial inflammation process.147 As described earlier, many of the cytoskeletal regulatory genes identified may be a result circulating immune cell activation. In the context of stroke and acute cardiac rejection, circulating blood cell-derived gene profiles have already been implicated in immune events at the systemic and organ levels.137,148 3.5.1.3 Blood coagulation/cell adhesion (protein C signaling, platelet-endothelium- leukocyte interactions) Development of CHF has been linked to endothelial and blood coagulation abnormalities,149 as well as interactions between platelets, endothelium, and leukocytes.150 Further, marked increase in plasma level of soluble adhesion molecules has been shown in CHF patients.151 Upregulation of adhesion molecules in the myocardium of failing hearts has also been associated with chronic low grade inflammation.152 The identification of protein C signaling as a significant network (Table 4) is also interesting. Protein C, an extracellular serine 46 protease important in anticoagulation, is also known to exhibit anti-inflammatory and anti- apoptotic activities through its interaction with endothelial protein C receptors, which are found on the endothelium and some white blood cells;153 however, the exact role of protein C in CHF remains largely unclear at the present time. 3.5.1.4 Apoptosis/DNA damage checkpoint During the cardiac remodeling process, a gradual but substantial reduction of myocytes can be seen after initial hypertrophy.154 The contribution of cardiomyocyte apoptosis as a mechanism of progressive myocardial dysfunction and CHF has been suggested by Narula et al. 154 However, it is also likely that the apoptosis-related events highlighted by the enrichment analysis are contributed by immune cells involved in the chronic inflammation process. Thus, it remains to be elucidated whether the observations here are due to apoptosis of myocytes, peripheral blood cells, or both, and whether they are a cause or effect of CHF. In summary, to highlight certain salient features of the CHF vs. NCF results from this chapter: 1) there is a high level of agreement in biological processes and networks between significant markers identified across multiple ‘-omics’ technologies, namely genomics and proteomics, and 2) multiple ‘-omics’ strategies uncovered a cohesive set of markers involved in themes currently accepted to be involved in CHF. The high concordance between results generated from multiple platforms reinforces our current understanding of the central mechanisms involved in CHF. Additional studies will be required to decipher the functional nuances of individual biomarkers and determine whether they are a cause or consequence of CHF. Such work will add significantly toward a better understanding of cardiac injury and repair, and its therapy. 3.5.2 Potential applications, caveats to the study, and future direction The possible impact of clinical variables such as gender and age on the present study warrants mention. Boheler et al. have demonstrated that HF gene expression profiles can differ considerably among patients of different age and sex.122 It is not clear to what extent this may have influenced the observations in the study – the most noticeable difference being the higher percentage of men and average age in the IHD group. However, it is important to note that the 47 analysis by Boheler et al. focused primarily on failing versus nonfailing hearts, and has not been extended to IHD versus NICM comparisons.122 One other caveat of the present study relates to the use of end-stage CHF samples. In contrast to an earlier study wherein Kittleson et al. identified a 90-gene expression profile that was able to differentiate ICM and NICM based on myocardial tissue-derived RNA.128 None of these 90 genes was statistically significant (FDR <0.05) in the IHD versus NICM peripheral blood data (results not shown). There are a few obvious explanations for the observed difference – one relates to sample collection time points. Kittleson et al. sampled at different disease stages (eg, at LVAD placement or cardiac transplantation, after LVAD support, or at endomyocardial biopsy in newly diagnosed HF patients), whereas all of the CHF blood samples used in this study were collected days before cardiac transplantation (i.e., end-stage). As such, the results observed are likely to specifically relate to the converging biological mechanisms at the end of IHD and NICM development. That said, had unique genomic or proteomic profile been observed with either IHD or NICM group at the end-stage of heart failure, these molecular profiles can have interesting implications for follow-up studies. For instance, one could examine whether the unique molecular profiles associated with IHD and NICM are still detectable and distinguishable in blood post-transplantation. One may also consider comparing these unique profiles with the NCF subjects pre- and post-transplantation, and examine whether the level of difference changes. In other words, from the ‘omics’ perspective – how does cardiac transplantation influence the ESHF patients? And, do the IHD or NICM patients respond similarly or differently to the cardiac transplantation process? Another factor to consider for future study design, of course, is the sampling site, blood versus heart tissue. It would be of great interest in future studies to analyze the 90 genes described by Kittleson et al., in blood samples collected at earlier time points before transplantation. An interesting question arises as to whether one can establish when the biological effects of CHF dominate the causes, and whether there is a ‘‘point-of-no-return’’ in the development of CHF. For these reasons, future studies should consider incorporating analysis of additional blood and biopsy samples from subjects at earlier stages of HF to better understand the full spectrum of mechanisms involved in the development of CHF. 48 CHAPTER 4: Biomarkers of acute cardiac allograft rejection 49 4.1 Background As described earlier, it has been estimated that up to 50% of patients will experience at least one episode of acute cardiac allograft rejection (ISHLT grade 2R or above) during the first year post-transplantation, despite the use of immunosuppressive therapies.46 Currently, the definitive diagnosis of allograft rejection relies primarily on the endomyocardial biopsy (EMB), an invasive and inconvenient procedure.35,155 EMB is also hindered by sampling errors and inter-observer variability despite the availability of international guidelines such as those set by the International Society for Heart and Lung Transplantation (ISHLT).38,115 Considering cardiac transplant recipients in many programs undergo at least 12-13 surveillance EMBs in the first year post-transplantation, and each procedure poses low but definite risks for complications to the patients – pneumothorax, cardiac perforation and even death – more accurate, precise, and less invasive alternatives are clearly desirable.35,156,157 4.2 Rationale In recent years, the availability of high-performance platforms has provided researchers an alternative unbiased approach to discovering biomarkers, and the use of ‘omics’ technology has been explored as a potential tool to help enhance the diagnosis of acute and chronic rejection.114 Of the current ‘omics’ technologies available, genomics is the relatively more recognized, studied and used platform. Rejection is a complex process that involves several critical leukocyte-mediated events (i.e., recognition of alloantigen on the allograft, release of effector molecules, initiation of the inflammatory response, and activation/recruitment of circulating immune cells). As such, numerous research groups have made efforts to examine the peripheral blood (mononuclear cells) [PBMC] expression profile in relation to allograft rejection.137,158-164 This approach has shown some promise, and recent studies involving microarray and qPCR analysis of peripheral blood gene expression profiles have provided evidence that they may be closely correlated with biopsy-proven acute cardiac allograft rejection.115,137,139 50 These early successes reported in the literature are encouraging, and demonstrate the promise in quantitative assessment of peripheral blood gene expression. In order to further elucidate the potential and possible pitfalls of ‘omics’, i.e., genomics technology, in the context of acute cardiac allograft rejection, additional studies and more evidence are clearly needed. In this chapter, I will focus on the discovery and examination of acute rejection biomarkers using a genomics-based approach, via Affymetrix microarrays. First, I will explore the peripheral blood to determine if there are any biomarkers within that are differentially expressed in patients undergoing acute cardiac allograft rejection, relatively to those who are not. If these biomarkers are indeed present, I will then investigate if i) a specific combination of them can potentially be utilized for the diagnosis or monitoring of cardiac transplant recipients, and ii) what the individual biomarkers may represent, from a pathobiological aspect, in the context of acute cardiac allograft rejection. 51 4.3 Materials and methods 4.3.1 Subjects and specimens selection The experiments in this chapter were conducted under the Biomarkers in Transplantation (BiT) initiative, which was approved by the Providence Health Care Research Ethics Board.116 Subjects who received a cardiac transplant at St. Paul’s Hospital, Vancouver, British Columbia between March 2005 and February 2008 were invited to participate. Subjects who agreed and signed consent forms were enrolled in the study. Transplant subjects received basilimax induction followed by standard triple immunosuppressive therapy (cyclosporine, prednisone, mycophenolate mofetil). Cyclosporine was replaced by tacrolimus for women and by sirolimus in the setting of renal impairment. Subjects with a 2R or 3R rejection episode within 3 months post-transplant received methylprednisolone. For the purpose of this chapter, the focus was to initially identify biomarkers of acute cardiac allograft rejection of ISHLT grade 2R or above, i.e., moderate rejection or worse (characterized by the presence of two or more foci of cellular infiltrate with associated myocyte damage based on the EMB). As such, the two groups of patients (blood samples) that were analyzed and compared were “acute rejection” (AR; defined as ISHLT grade ≥2R), and “non- rejection” (defined as ISHLT grade = 0R). A total of 28 subjects with blood samples corresponding to at least one AR (12 subjects) or one NR episode (16 subjects) were selected for microarray analysis. The subjects were divided into two independent cohorts (Figure 8). The first was a training cohort, (Figure 8, left) consisting of 6 AR and 12 NR samples collected from subjects with no serious complications (e.g., prolonged peri-transplant ischemia, infection, non-responsiveness to AR treatment). The training AR samples were collected prior to treatment for rejection and corresponded to the first AR episode of the subject. The second was a test cohort (Figure 8, right) consisting of 6 AR and 4 NR samples from subjects not included in the training set. All AR samples were collected within two days of biopsy-proven rejection episodes. All biopsies were over-read in a blinded manner by an experienced transplant cardiac pathologist using the revised ISHLT grading 52 scale.47 Patient demographics were comparable between training and test cohort subjects (Table 5). Figure 8. Division of subject samples into training and test cohorts. Subjects enrolled between January 2005 and September 2007 who satisfied the selection criteria were considered training samples and used for biomarker discovery. The inclusion and exclusion criteria are as follow: only include NR samples collected from patients with no AR episode in first 6 months post-transplantation; exclude subjects who did not respond to AR treatment and/or had major/multiple complications within the first 6 months post-transplant; exclude NR samples which were taken during an acute rejection treatment. Two NR samples were mapped to each AR sample, collected at approximately the same timepoint. One AR and 3 NR subjects who did not satisfy the aforementioned criteria were excluded from the training cohort and included in the test cohort in addition to 5 five AR subjects and 1 NR subject enrolled between October 2007 and February 2008. Four subjects, enrolled in this time-period, were excluded due to death or lack of blood sample collection at the time of biopsy-proven AR. 53 Table 5. Demographics of cardiac transplant subject cohorts. Training Cohort (n=18) Test Cohort (n=10) Age (mean, standard deviation in years) 52±15 48±13 Gender (n male) 14 7 Ethnicity (n) Caucasian 16 9 Asian 1 1 Other 1 - Primary Disease (n) Ischemic Heart Disease 9 4 Non-ischemic Cardiomyopathy 7 4 Other 2 2 54 4.3.2 Sample and data processing Blood samples were collected in PAXgene tubes. The 28 subjects blood samples selected for microarray analysis were processed to isolate total RNA, using PAXgeneTM Blood RNA Kits as previously described.116 RNA quality was checked using an Agilent BioAnalyzer. RNA samples were then processed via reverse-transcription-in vitro transcription (RT-IVT) to generate labeled cRNA which is then fragmented for hybridization on the Affymetrix GeneChip® Human Genome (HG) U133 Plus 2.0 array. Microarray analysis was performed at the Microarray Core Laboratory at Children’s Hospital in Los Angeles, California. The microarrays were checked for quality using affy (version 1.16.0) and affyPLM (version 1.14.0) BioConductor packages,165,166 and Mahalanobis Distance Quality Control (MDQC).167 4.3.3 Analysis 4.3.3.1 Identification of biomarkers Statistical analysis was performed using a “funnel” approach (Figure 9) with SAS System for Windows version 9.1.3,168 R version 2.7.0, 134 134 134 133 132 132 132 132 132 131131 and BioConductor version 2.2.135 In step 1, the Robust Multi-array Average (RMA)169 technique was used for background correction, normalization and summarization (affy BioConductor package version 1.18.1). To reduce noise, probe-sets with consistently low expression values across all samples were eliminated from further analysis. The remaining probe-sets were analyzed using three moderated t-tests (Figure 9, step 2). Significance Analysis of Microarrays (SAM)170 was performed using samr R package version 1.25 (http://cran.r- project.org/web/packages/samr/index.html). Limma BioConductor package (version 2.14.3)171 was used for performing the other two moderated t-tests.172 To ensure stringency, only probe- sets with a False Discovery Rate (FDR) <5% in all three moderated t-tests and a fold change >2 were considered statistically significant. 55 Figure 9. Overall workflow of the data analysis. DE = differentially expressed. 56 4.3.3.2 Functional enrichment analysis Functional enrichment of the differentially expressed genes (identified as described earlier) was examined using FatiGO,173 available within version 3 of Babelomics,174 a suite of web-based tools designed for functional analysis. 4.3.3.3 Generation and evaluation of the AR biomarker panel Biomarker panel genes were pinpointed using Stepwise Discriminant Analysis (SDA) with forward selection on the statistically significant probe-sets (Figure 9, step 3). The classifier was built and tested with Linear Discriminant Analysis (LDA). The biomarker panel genes were assessed by quantitative reverse-transcription- polymerase chain reaction (qRT-PCR; or simply as qPCR) using whole blood RNA from 16 of the training samples. RNA samples were first reverse transcribed to cDNA using SuperScript III First- Strand Synthesis System according to the manufacturer’s protocol. qPCR was performed using gene-specific primers and Applied Biosystems (ABI) TaqMan Gene Expression Assays, on the ABI 7900HT Fast Real-Time PCR System. The qPCR data was analyzed using qBase v1.3.4.175 Expression levels of the biomarker genes were normalized against β-actin gene. The performance of the biomarker panel was also assessed through a six-fold cross- validation and an internal validation using the test cohort (Figure 8, right column). In both validations, the threshold between AR and NR samples was chosen to maximize the negative predictive value. 57 4.4 Results 4.4.1 Differentially expressed genes in AR patients After normalization and pre-filtering, 25,082 probe-sets remained and were included in analyses (Figure 9, step 2) using the training samples (Figure 8, left side). A total of 1295 probe- sets were identified as having an FDR <5% in three moderated t-tests (SAM, robust and non- robust moderated T statistics) and a fold change >2. 4.4.2 Dysregulation of molecular and cellular processes in AR patients 4.4.2.1 Gene ontology (GO) analysis Of the 1295 biomarker candidates that were differentially expressed in the training cohort, 1208 were downregulated and 87 were upregulated in AR relative to NR samples. Using FatiGO, these 1295 candidates were mapped to gene ontology (GO) terms through functional enrichment analysis. Over-represented, statistically significant GO terms were reviewed and are summarized in Table 6. Many downregulated probe-sets found in AR were associated with molecular and cellular functions such as regulation of enzymatic activities and protein metabolic processes. Conversely, numerous upregulated probe-sets found in AR were linked to innate and humoral immunity, response to wounding, and hypoxia. These functions and cellular processes have been linked to allograft rejection.176,177 58 Table 6. Relative expression levels and associative GO terms over-represented in the 1295 statistically significant probe sets. Regulation in AR vs. NR Number of probe Sets GO Term Type Exemplary GO terms corresponding to the significant probe sets Down 1,208 Biological processes - signal transduction - biopolymer metabolic processes - cellular protein metabolic process - cellular component organization and biogenesis Molecular functions - GTPase regulator activity - RNA binding - ion binding - enzyme inhibitor activity Up 87 Biological processes - innate / humoral immune response - response to wounding - response to hypoxia - acute inflammatory response Molecular functions - creatine transporter activity - transcription factor binding / activity - tumor necrosis factor binding - actin binding 59 4.4.3 AR biomarker panel genes SDA was applied on the 1295 differentially expressed probe-sets. Twelve probe-sets (corresponding to 12 genes) were identified that, together, best differentiate between AR and NR samples (Table 7). These 12 biomarker panel genes showed a 2 to 3.3 fold change in expression levels between AR and NR samples. Ten of the 12 biomarker panel genes were downregulated in AR. Table 7. Acute cardiac allograft rejection biomarker panel. Probe Set ID Gene Symbol Gene Name Fold Change Regulation of AR versus NR 207883_s_at TFR2 Transferrin receptor 2 2.1 up 229067_at SRGAP2P1 SLIT-ROBO Rho GTPase activating protein 2 pseudogene 1 3.3 down 221841_s_at KLF4 Kruppel-like factor 4 2.7 down 214659_x_at YLPM1 YLP motif containing 1 2.0 down 204493_at BID BH3 interacting domain death agonist 2.0 down 201669_s_at MARCKS Myristoylated alanine-rich C-kinase substrate 2.8 down 1556209_at CLEC2B C-type lectin domain family 2, member B 2.3 down 235412_at ARHGEF7 Rho guanine nucleotide exchange factor (GEF) 7 2.2 down 226851_at LYPLAL1 Lysophospholipase-like 1 2.2 down 202749_at WRB Tryptophan rich basic protein 2.1 down 1556283_s_at FGFR1OP2 FGFR1 oncogene partner 2 2.6 up 209580_s_at MBD4 Methyl-CpG binding domain protein 4 2.0 down 60 4.4.4 Evaluation of the AR biomarker panel qPCR was performed for the 10 classifier genes with commercially available primers on 5 AR and 11 NR training samples. Three exemplary genes are illustrated in Figure 10. Seven of the 10 genes were consistent in the direction of regulation of AR relative to NR between the microarray and the qPCR platforms. The cross-validation sensitivity was 100% and specificity was 75%. The biomarker panel was then applied to the test cohort. One AR (out of possible 6) and one NR (out of possible 4) sample were misclassified, corresponding to a sensitivity of 83% and specificity of 75% for the 12-gene classifier. Alternatively, when the classifier was trained using all 37 AR and NR samples collected within the first five months post-transplant from the 18 training cohort subjects, the internal validation results improved to 100% specificity (Table 8). 61 Figure 10. AR biomarker expression evaluation. Average gene expression values of AR and NR samples from qPCR (top row) and microarrays (bottom row) are displayed for each gene (columns). The error bars represent the standard deviation within each group. Table 8. AR biomarker performance evaluation. Cross- Validation Internal Validation [10 Test Samples] Training Set [18 Samples] Training Set [38 Samples] Sensitivity 100% 83% 83% Specificity 75% 75% 100% During cross-validation, the data was randomly split into six parts, each part containing 1 AR and 2 NR samples. In each fold, a different part of the data served as the test set, while the remaining (five of six) parts were used as the training set. This process was repeated six times (each ‘part’ served as the test set once), and each time, analysis steps 2 and 3 (Figure 9) were performed on the training set, and the obtained biomarker panel was tested on the test set. The internal validations were performed on the test cohort samples 62 4.5 Discussions 4.5.1 Integration of biological information and interpretation With high performance platforms now available, a new door has opened to allow discovery of molecular signatures. In the previous research chapter, the ‘omics’ technology was applied with the goal in mind to gain insight into the underlying pathobiology and molecular profile disturbances in the context of end-stage heart failure. In contrast, a major focus in this chapter is identifying specific ‘panels’ of differentially expressed genes to serve as classifiers of disease presence or absence, i.e., monitoring and diagnostics. It should be noted that the main measure of a classifier’s excellence is based on classification accuracy with independent data, rather than biological plausibility.178 While the interpretation of biological context behind biomarker classifiers is not always straightforward,178 biomarkers that fit currently accepted biological and physiological paradigms are more readily accepted by the research and clinical communities.179 Therefore, interpreting the biological plausibility of classifier genes is worthwhile as it increases the value of a microarray dataset.178 In this study, 1295 probe-sets demonstrated expression levels that differ in AR versus NR patients. This is perhaps not surprising, given the profound disturbance that allograft rejection-related processes can have at a cellular and molecular level. These gene expression changes in the peripheral blood reflect responsive and adaptive mechanisms to the early events underlying rejection (i.e., inflammation, alloimmune activation/response). From the 1295 probe-sets identified as statistically significant, 12 genes were identified together as the most effective biomarker panel. I then examined the published literature for information regarding the biological functions of these 12 and found that nine have been relatively well studied. Observations of potential relevance to transplantation have been summarized in Table 9. 63 Table 9. Summary of the biological functions of the AR biomarker panel genes based on previous literature. Gene Symbol Described Biological Functions TFR2 • Transferrin receptor 2 • Involved in iron homeostasis 180, disruption in iron homeostasis observed in lung allograft patients 181 • TFR2 levels upregulated in activated T cells 182; T cell activation is an expected phenomenon during allograft rejection KLF4 • Kruppel-like factor 4 • Linked to regulation of B cells; overexpression suppresses cell proliferation 183 • B cell activation is associated with downregulation of KLF4 mRNA and protein 184 BID • BH3 interacting domain death agonist • Involved in perforin and granzyme B induced apoptosis 185,186 • Involved in hypoxia/reoxygenation induced lymphocyte apoptosis 187 • Antagonizes apoptosis in certain circumstances 187 MARCKS • Myristoylated alanine-rich C-kinase substrate • Is a major substrate of PKC 188 • MARCKS and PKC implicated in many cell growth, differentiation, metabolic and functional pathways in all immune cell types 188-199 CLEC2B • C-type lectin domain family 2, member B • Also called Activation-induced C-type lectin, a transmembrane receptor on monocytes, granulocytes, B and T cells 200,201 • Is the ligand for the NKp80 receptor on NK cells and monocytes; interaction regulates activity of these cells 201 • Reverse signalling induced by NKp80 binding of CLEC2B on monocytes result in TNFα production which has been suggested to play a role in acute and chronic lung rejection 202 ARHGEF7 • Rho guanine nucleotide exchange factor (GEF) 7 • Is a positive regulator of Rho family of small molecular weight G-proteins 203,204 • IL-2 stimulation of T cells result in upregulation of ARHGEF7 mRNA, and is thought to be involved in Rho mediated cellular changes (cytoskeletal rearrangements) 203,204 WRB • Tryptophan rich basic protein • Gene first identified as mapping to 21q22.3, a locus associated with congenital heart disease in Down’s syndrome 205,206 • WRB has been shown to be downregulated in endothelial cells in response to C- reactive protein 207 FGFR1OP2 • Fibroblast growth factor receptor oncogene partner 2 • First described as a fusion partner of fibroblast growth factor receptor 1 in the setting of 8p11 myeloproliferative syndrome (EMS), which is characterized by eosinophilia and T/B cell lymphoma prior to transformation into acute myeloid leukemia 208,209 MBD4 • Methyl-CpG binding domain protein 4 • Encodes a DNA repair protein 210,211 • Plays critical role in genome stability/integrity, repair and cell cycle response to DNA damage 212 • Found to be mutated in various carcinomas 213-215 64 The process of graft rejection involves activation and proliferation of immune and inflammatory cells. In this context, the upregulation of TFR2 and FGFR1OP2, and downregulation of KLF4 and BID in AR versus NR patients is consistent with cell activation and proliferation during an AR episode. TFR2 is a transferrin receptor important in iron uptake into cells and its levels rise in activated immune cells.182 TFR2 is also important in regulating iron homeostasis, and interestingly, dysregulated iron homeostasis has been observed in lung allograft patients.181 The FGFR10P2 gene has also been associated with proliferation as it was first identified in a myeloproliferative syndrome that involves eosinophils as well as T and B cells, 208,209 but the normal function of FGFR10P2 is not yet known.209 Some genes are known to be downregulated in cells undergoing activation or proliferation, including the transcription factor KLF4184 and the pro-apoptotic protein BID,187 so their decrease in AR versus NR patients is consistent with activation of immune cells. The other genes in the classifier panel identified have also been implicated in biological processes that may have relevance to the process of graft rejection. Additional biological studies are needed to determine which specific peripheral blood cells express these genes and why many of their levels are decreased in AR patients. The Rho guanine nucleotide exchange protein ARHGEF7 is important in regulating cytoskeletal changes necessary for maintenance of cell morphology and migration, and has been reported to be upregulated in activated T cells.204 The observed downregulation of ARHGEF7 mRNA in AR patients would seem contradictory. However, many possibilities exist to explain this observation; for example, perhaps the most activated and migratory T cells have left the circulation to enter the tissues of the graft. Similar arguments could be made for the intracellular signalling molecule MARCKS and transmembrane ligand/receptor CLEC2B. Finally, two of the genes in the biomarker panel have described functions, but their role in graft rejection will have to be defined. MBD4 is a DNA repair enzyme,210,211 while WRB is a gene that has been described to be downregulated in non-blood cells by inflammatory mediator C-reactive protein.207 65 4.5.2 Assessment and validation of the AR biomarker panel The 12 genes belonging to the biomarker panel have been subjected to preliminary biological investigation, and hypotheses regarding their involvement in acute rejection can be formulated. Three genes in the panel, SRGAP2P1, YLPM1 and LYPLAL1, have not yet been characterized in any biological studies. Their sequence, cellular location and predicted biochemical or biophysical characteristics are available in databases such as Genecards [www.genecards.org] and NCBI [www.ncbi.nlm.nih.gov]. While the identified biomarkers appear to be biologically plausible in the context of acute cardiac rejection, as described earlier, the other major focus in this study is the assessment and validation of the classifier (i.e., biomarker panel consisting of specific combination of genes) of acute rejection that has been generated based on genes that were the differentially expressed between the AR and NR subjects. In this particular study, qPCR was performed for 10 classifier genes on 16 training patient samples (5 AR and 11NR) available, using commercially available primers, i.e., Applied Biosystems (ABI) TaqMan Gene Expression Assays. The motivation for performing this set of experiment was to assess the expression levels of the classifier genes using an alternative technological platform and in turn, investigate the potential utility and transition to using qPCR as a possible clinical assay. For discovery purposes, microarrays have shown great potential as a tool for screening a large number of potential candidates to discover the key (combination of) biomarkers that is correlated with the disease of interest. From a pragmatic perspective however, it has been suggested that biomarker panel/classifier gene signature analysis that requires the simple use of qPCR may be more ideal, since such platforms are more readily available in clinical laboratories; in some cases they may already be validated for clinical diagnostic tests.178,216 Of the 10 classifier/biomarker panel genes evaluated, seven were consistent in the direction of regulation of AR relative to NR between the microarray and the qPCR platforms. Potential factors contributory to differences between qPCR and microarrays have been detailed in the work of Morey et al.217 Indeed, disagreement or lack of correlation between microarray and qPCR data is quite common and has been reported by others.217,218 Beyond the explanation 66 of others,217,218 the binding of commercially available gene-specific primers to gene regions wherein polymorphisms or splice variations may exist could differ from those regions detected by the probe-sets on a microarray. Ultimately, internally developed probe-set-specific primers may help avoid the aforementioned gene region binding discrepancy, provide better correlations between qPCR and microarray result, and address this widely recognized issue. As noted earlier, the main measure of a classifier’s excellence is based on classification accuracy with independent data.178 In this study, the subjects were first divided into two independent cohorts for this purpose (Figure 8), thus allowing internal validation of the biomarker panel, in addition to cross-validation. As shown in Table 8, the biomarker panel, which consists of 12 genes, was able to classify the internal validation samples with 83% sensitivity and 100% specificity. In comparison, the sensitivity and specificity of EMB (the current gold standard for the diagnosis of acute cardiac allograft rejection) has been reported to vary between approximately 75-90% and 80-90+%, respectively; these values are also thought to be heavily dependent on the number of (right ventricular) biopsy samples taken for evaluation.219-221 Taken together, the initial results based on this study suggest that peripheral blood genomic molecular profile, such as the 12 gene classifier panel identified in this study, hold considerable potential in discriminating acute rejection from non-rejection in heart transplant recipients. 67 4.5.3 Current study results versus CARGO results The Cardiac Allograft Rejection Gene Expression Observation (CARGO) study warrants a mention here, as it is perhaps the most widely recognized and discussed microarray study in the context of acute cardiac allograft rejection diagnosis in the recent years. The CARGO study used microarray analysis and real-time PCR to examine and validate gene expression profiles of allograft recipients’ peripheral blood mononuclear cells. Deng and colleagues115 reported 11 genes that distinguish ISHLT grade 0 rejection (quiescence) from moderate/severe rejection (1990 ISHLT grade ≥3A; see Figure 1). These 11 genes were compared to the 12-gene biomarker panel identified in this study, and no genes were found to be in-common. There are several reasons for such an apparent difference in results. First, the microarray platforms used were different between the two studies – CARGO employed a custom array with 7370 genes represented, while this particular study used Affymetrix HG U133 Plus 2.0 microarray with 47,000 transcripts which correspond to at least 25,000 human genes. A key factor in different results relates to the fact that only 3 of the 12 biomarker classifier panel genes from this study were present on the custom array used in CARGO. In other words, 9 of the biomarker panel genes found in this study had no chance being detected on the array that CARGO investigators used. Second, the diagnostic timeframe (i.e., time post-transplant) for which the CARGO biomarker panel is able to diagnose rejection is different from that observed with the biomarker panel generated in this study. In the CARGO published study, analyses were carried out on samples collected ~60+ days post-transplant. Thus, the generalizability and diagnostic utility of the classifier developed, based on the samples used in CARGO, is aimed towards samples collected after 2 months post-transplant [www.xdx.com/allomap].115 In this study, the samples analyzed were collected between week 1 and month 5 post-transplant. Thus, the classifier reflects differential gene expression detectable as early as week 1 post-transplant. Lastly, the sample sources used were different in between the two studies –the CARGO study focused on peripheral blood mononuclear cells (PBMC), while in the current study whole blood PAXgene samples were used for microarray analysis. The gene expression evaluated in 68 this study, therefore, is reflective of all peripheral blood circulating cells during acute rejection, and is not necessarily restricted to those transcripts arising from PBMCs. Regardless of particular differences in the biomarker panels that distinguish rejection from non-rejection in this initial study and that of CARGO, there was commonality at a higher, biological process level. In-common gene ontological (GO) terms (based on the AR biomarkers from both studies) include biological processes such as such as cell motility, signal transduction and immune response. 4.5.4 Potential applications, caveats to the study, and future directions Based on the internal- and cross-validation results, the classifier panel developed in this study appeared to be able to discriminate between AR and NR samples collected as early as one week post-transplant and as late as five months post-transplant. This characteristic, along with the advantage of whole blood approach that is minimally invasive, gives the biomarker panel the potential to serve as a complementary, pre-screening tool to help determine which patients really need the EMB. There are several additional considerations pertinent to these studies that deserve comment. First, larger training and testing cohorts would be desirable. However, the statistical approach chosen was designed to be sufficiently robust to accommodate a smaller sample size. Second, patients enrolled in this study were primarily of Caucasian ethnicity, from a single institution and largely on a consistent local immunosuppressant regimen. To increase the generalizability and broader applicability of the biomarker panel, inclusion and testing of independent, external cohorts would be desirable for future studies. Third, this particular study examined rejection episodes that occurred within the first 5 months post-transplant, but given that most acute rejections occur within this timeperiod (usually during the first 6 months),37 such a classifier panel can still potentially benefit the care of most cardiac transplant patients. Further, this initial study has focused on the discovery of biomarkers of acute cardiac allograft rejection defined as ISHLT grade 2R or above. It may also be of interest in future studies to include mild rejection patients (i.e., ISHLT grade 1R), and examine whether the biomarker 69 panel/molecular signature described here also work for a wider spectrum of patients and correlate with different severity of rejection. Last but not least, a universal limitation faced in biomarker studies aimed at classification of AR versus NR patients is the reliance on EMB. As noted in the introduction, the different ISHLT rejection grades are subject to variability under the eyes of expert pathologists, which creates somewhat of a predicament in trying to identify a biomarker panel that outperforms the current “gold standard”, the EMB, in the diagnosis of acute heart rejection. The primary focus of this study was to create a classifier which can help discriminate between AR and NR samples, regardless of the underlying mechanisms that cause the rejection episode. The complexity of the rejection process, including cellular and soluble factors,222 remains a challenge to better understand how to care for patients and to interpret any biomarker panel results. This intriguing problem will remain the focus of many research groups in the biomarker arena. Certainly, among variables that influence the potential applicability of molecular signatures identified in this study in guiding care, the time post-transplant is among the most important for many reasons. Immunosuppressive regimens, including the nature of transient augmentation in the face of rejection, will continue to evolve. These therapeutic changes no doubt, will modify signatures in ways yet to be discovered. The enigmas that remain about how to monitor for human cardiac allograft rejection may be partly resolved through additional work in the future and examine alternative sources for biomarkers, such as the plasma proteome and the serum and urine metabolome. 70 CHAPTER 5: Biomarkers of cardiac allograft vasculopathy 71 5.1 Background Whereas the occurrence of acute cardiac rejection is considered one of the main short term obstacles, the development of Cardiac allograft vasculopathy (CAV) is a major hurdle in the long term survival of cardiac transplant recipients.45,56,58 Each year, approximately 4000 cardiac transplantations take place around the globe.56 Of these, almost 50% will develop CAV in the first several years post-transplant.56 It has been estimated that CAV is responsible for up to 15% of deaths in cardiac allograft recipients after they have survived the first year post-transplant.56 Considering the prevalence and the significance of CAV, cardiac transplant recipients typically undergo routine tests at least once a year to monitor the health of the transplanted heart. The screening and diagnosis of CAV is largely based on detection of narrowing, or stenosis, of coronary arteries, i.e., blood vessels which supply oxygenated blood to the heart.45,57,58 Unfortunately, the most widely used modality for the diagnosis of CAV remains coronary angiography, an invasive technique that is costly and uncomfortable for patients. Further, this procedure is associated with definite risks for complications.45,57,58 5.2 Rationale Given the nature of the current gold standard, an alternative, minimally invasive method for detecting CAV that is both sensitive and specific is highly desirable. A simple blood test that is based on specific molecular biomarker signatures to help screen, diagnose or monitor CAV has the potential to alleviate discomfort of cardiac transplant patients and improve their wellbeing. Similar to the scenario in acute cardiac allograft rejection, biomarkers such as C-reactive protein (CRP) and brain natriuretic peptide (BNP), as well as other gene expression tests,223 have all been suggested as having utility in the diagnosis of CAV.45 However, with the increasing availability ‘omics’ technologies, it is now possible to examine multiple molecular biomarkers of risk or disease in a high-throughput, unbiased manner. As noted in the previous chapters, this holistic approach to biomarker discovery has yielded promising data, such as in the case of 72 acute kidney rejection diagnosis133 and acute cardiac rejection diagnosis, which resulted in identification of biomarker panels that are more sensitive and specific for the disease of interest, ultimately leading to better translation into the clinic.115 Currently, proteomics technology has yet to be examined in the context of detecting allograft coronary artery stenosis as a strong indicator of CAV development and expression of chronic rejection. As such, there were several goals in this work. First, I wanted to identify plasma-derived biomarkers that are differentially expressed between patients with and without significant CAV using an unbiased, data-driven approach. Second, I want to assess the diagnostic performance and the potential clinical utility (e.g., for CAV development screening and monitoring) of the biomarker panel generated based on the biomarker candidates identified. Lastly, I also examined the biological plausibility and the possible implications of these biomarkers in the context of CAV development. 73 5.3 Materials and methods 5.3.1 Subjects and specimens This study was conducted under the Biomarkers in Transplantation (BiT) initiative, which was approved by the Providence Health Care Research Ethics Board.132 Subjects who received a cardiac transplant at St. Paul’s Hospital, Vancouver, British Columbia, were approached by our research coordinators, and those who consented were enrolled in the study. 5.3.1.1 Screening and identification of CAV and Non-CAV patients Screening for CAV as an expression of chronic rejection was routinely performed using dobutamine stress echocardiography, coronary angiography and intravascular ultrasounds (IVUS) according to the ‘Protocol for Long-term Surveillance of Cardiac Allograft Vasculopathy’ guidelines [http://www.heartcentre.ca/CADsurveillance2007.pdf.pdf]224 as established by St. Paul’s Heart Centre. Angiograms were assessed in a core lab using quantitative coronary angiography (QCA) as previously described.59 Percentage diameter stenosis (%DS) was calculated based on the following formula: %DS = [(RD – MLD)/RD] X 100, where RD is reference diameter (an average of the normal region of the blood vessel), and MLD is the minimum lumen diameter. Whenever possible, the proximal, mid, and distal portion of the coronary arteries are assessed. Physicians, nurses and technicians who are involved in the collection of the coronary angiography data were blinded to the molecular study protocol and other data. Coronary angiographic criteria were used to characterize patients in the current study as QCA is the most widely-available and consistently measured endpoint for evaluating CAV at our institution. Presence of biologically significant cardiac allograft vasculopathy (CAV) was defined in the study as maximum percentage of diameter stenosis (Max %DS) in the left anterior descending artery (LAD) ≥ 40%. Non-significant CAV development (Non-CAV) was defined as Max %DS in LAD ≤20%. Possible CAV was defined as 20%< Max %DS <40%. 74 5.3.2 Sample selection and data processing Blood samples used for the current study were those collected in EDTA tubes at the nearest (earlier) timepoint corresponding to the date when coronary angiography was carried out – typically at least 1 year post-transplant, during routine post-transplantation check-ups. 40 cardiac transplant patients samples were selected for the proteomic analysis (10 CAV and 9 non-CAV used for generating the biomarker panel/classifier; additional 21 possible CAV for principal component analysis). Plasma samples from healthy individuals were pooled and served as the normal reference sample for each iTRAQ experimental run. Sample processing, data acquisition and analysis were carried out as described in previous studies.133 Briefly, samples were processed via immuno-affinity chromatography (Genway Biotech; San Diego, CA), to deplete the 14 most abundant plasma proteins (albumin, fibrinogen, transferrin, IgG, IgA, IgM, haptoglobin, α2-macroglobulin, α1-acid glycoprotein, α1-antitrypsin, apolipoprotein-I, apolipoprotein-II, complement C3 and apolipoprotein B).133 Depleted plasma protein samples were labelled with iTRAQ reagents according to manufacturer’s protocol (Applied Biosystems; Foster City, CA). iTRAQ labelled peptides were analyzed by a 4800 MALDI TOF/TOF mass spectrometer (Applied Biosystems; Foster City, CA). Data was analyzed using ProteinPilot™ software v2.0 with the integrated Paragon™ Search and Pro Group™ Algorithms, and subsequent searching against the International Protein Index (IPI HUMAN v3.39) database. 5.3.3 Analysis 5.3.3.1 Identification of CAV biomarkers and functional enrichment The statistical analysis was performed using R version 2.10.1.225 The data was log base two transformed. Protein groups (PG; described earlier and in the Appendices) detected in at least 75% of the CAV and Non-CAV samples were included in the subsequent analysis. A robust moderated t-test was used to find proteins with differential relative levels between CAV and Non-CAV samples. Elastic Net model226 was applied to consider the most significant proteins ( robust moderated t-test p-value <0.1) to identify the biomarker panel/classifier. 75 Functional enrichment was performed on the candidate protein biomarkers (identified by Robust-LIMMA and Elastic Net) using MetaCore (GeneGo Inc; www.genego.com). Gene Ontology (GO)-based analyses were carried out through MetaCore, using publicly available (www.geneontology.org) database, to assess the functional significance of the proteins of interest. GO terms with FDR <5% were considered statistically significant. 5.3.3.2 Validation of CAV biomarker panel Cross validation, i.e., leave-one-out cross-validation (LOOCV), was used to evaluate the performance of the analytical pipeline (and the biomarker panel). In leave-one-out cross- validation, one sample is left out as a test sample and the remaining 18 samples are used to discover and build a classifier score. The resulting score is then used to classify the sample left out. This procedure is repeated until all samples are left out once, and the performance is estimated by the average of the results. The classification of test samples based on their plasma samples was performed using the Elastic Net classifiers built based on the training set in each fold of the cross-validation. Specifically, the biomarker classifier score (i.e., a value generated based on the combined contribution of the biomarker panel proteins and their expression levels) was used to classify test patient samples from the cross-validation as either having significant CAV or not. Principal component analysis was also performed on all patients samples selected for this study, based on the biomarker panel proteins. Receiver operating characteristic (ROC) curve was also constructed for the CAV biomarker panel, based on the probabilities estimated by the cross-validation. The ROC curve and the area under the curve (AUC) were computed using the ROCR package.227 76 5.4 Results 5.4.1 Coronary angiography and patient characteristics The heart transplantation population included in the current analysis comprised of consented patients for whom both coronary angiography data and corresponding plasma samples were available (Table 10). In the 40 patients available for the study, out of the three major coronary arteries assessed (LAD, left anterior descending; LCX, left circumflex; RCA, right coronary artery), the most severely stenosed (Max %DS) vessel was typically the LAD, in approximately two-thirds of the patients (n=26; 65%) (Table 11). The proteomic biomarkers (Table 12) were identified in the training cohort, which was selected based on the more ‘extreme’ phenotype according to the CAV definition employed in the current study. In the training cohort, the Max %DS in the LAD vessel in the CAV subjects ranged from 41% to 70% and averaged around 53%, whereas the Non-CAV subjects ranged from 0% to 20% and averaged at 10% (Table 11; Figure 11). It is also interesting to note that in almost all Non- CAV subjects, all three major coronary arteries, i.e., LAD, LCX and RCA, were relatively clear compare to the CAV subjects, and generally did not show angiographic signs of moderate/severe coronary artery stenosis (Figure 11). Additional summarization of the coronary angiography data is provided in Table 11. 77 Table 10. Cardiac transplant patient demographics. Training Cohort Samples Additional Samples CAV Non-CAV Possible CAV n = 10 n = 9 n = 21 Sex Male, n (%) 9 (90%) 6 (67%) 16 (76%) Female, n (%) 1 (10%) 3 (33%) 5 (24%) Age Years, (mean ± SD) 54.5 ± 12.1 50.1 ± 13.6 52.2 ± 12.0 Ethnicity Caucasian, n 9 7 19 Asian, n 0 2 1 Others, n 1 0 0 Primary Disease Ischemic heart disease, n 3 3 8 Non-ischemic cardiomyopathy, n 2 4 8 Others*, n 5 2 5 *Others include cardiogenic shock, hypertrophic cardiomyopathy and unspecified cardiomyopathy. CAV, significant cardiac allograft vasculopathy (Max %DS ≥40%); Non-CAV, non-significant cardiac allograft vasculopathy (Max %DS ≤20%), Possible CAV (20%