Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The pharmacogenomics of vincristine-induced neurotoxicity in paediatric cancer patients with Wilms tumor… Loo, Tenneille 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_fall_loo_tenneille.pdf [ 8.88MB ]
Metadata
JSON: 24-1.0072033.json
JSON-LD: 24-1.0072033-ld.json
RDF/XML (Pretty): 24-1.0072033-rdf.xml
RDF/JSON: 24-1.0072033-rdf.json
Turtle: 24-1.0072033-turtle.txt
N-Triples: 24-1.0072033-rdf-ntriples.txt
Original Record: 24-1.0072033-source.json
Full Text
24-1.0072033-fulltext.txt
Citation
24-1.0072033.ris

Full Text

THE PHARMACOGENOMICS OF VINCRISTINE-INDUCED NEUROTOXICITY IN PAEDIATRIC CANCER PATIENTS WITH WILMS TUMOR OR RHABDOMYOSARCOMA  by Tenneille Loo    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Experimental Medicine)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2011  © Tenneille Loo, 2011  ii Abstract  Vincristine is one of the most effective and widely utilized antineoplastic agents. However, the clinical utility of this drug is limited by severely debilitating vincristine- induced neurotoxicities (VIN). Previous studies have associated VIN with genetic polymorphisms in genes involved in the metabolism and transportation of vincristine, including CYP3A4, CYP3A5, and ABCB1. However, the findings of such studies have not been consistently reproduced. This study hypothesizes that there are specific variants in genes involved in general drug absorption, metabolism, distribution, excretion, and toxicity (ADME-Tox) that affect the individual susceptibility to VIN in patients with Wilms tumor and rhabdomyosarcoma. Detailed clinical data was collected from 140 patients with Wilms tumor and rhabdomyosarcoma by retrospective chart review. VIN cases were characterized by type of neurotoxicity, and severity was evaluated using a validated clinical grading system for adverse events (NCI-CTCAE v4.03). A customized Illumina GoldenGate Panel was used to genotype 4,536 single nucleotide polymorphisms (SNPs) in candidate genes involved in the metabolism and transportation pathway of vincristine, as well as in genes broadly involved in ADME-Tox. None of the SNPs that were previously reported to be associated with VIN were found to be significantly associated (p-value < 0.05). With similar effect sizes, six novel genetic variants in five genes (PON1, ABCA4, ABCG1, CY51A1, SLCO1C1) were significantly associated with VIN in both tumor types. Whereas none of these genes have been previously associated with VIN or the biotransformation of vincristine, interestingly,  iii the biological functions of the encoded proteins have been indirectly linked to nerve function, neuropathy, or neurodegenerative diseases. Therefore, I hypothesize that the genetic basis of VIN is likely polygenic and that the six genes influence individual susceptibility to VIN by affecting: nerve regeneration (PON1, PPP1R9A), and cholesterol homeostasis and remyelination (ABCA4, ABCG1, CYP51A1), as well as the metabolism of vincristine (PON1, CYP51A1) and the transportation of lipids, vincristine, metabolites, or neuroprotectants (SLCO1C1, ABCA4, ABCG1). This study adds to the literature by identifying new potential biomarkers for VIN, providing novel hypotheses for the mechanisms underlying VIN susceptibility, and is a point of origin for replication studies.   iv Preface  The work contained in this thesis has been approved by the University of British Columbia and the Children’s & Women’s Health Centre of BC Research Ethics Board, the Ethics Certificate Numbers for this study are H10-01568 and H04-70358.  v Table of Contents  Abstract ................................................................................................................................... ii	
   Preface ..................................................................................................................................... iv	
   Table of Contents ..................................................................................................................... v	
   List of Tables ............................................................................................................................ x	
   List of Figures ......................................................................................................................... xi	
   List of Symbols and Abbreviations ..................................................................................... xii	
   Acknowledgements ............................................................................................................. xvii	
   Chapter  1: Introduction ......................................................................................................... 1	
   1.1	
   Adverse Drug Reactions ............................................................................................... 1	
   1.1.1	
   Factors that Contribute to Adverse Drug Reactions ............................................... 2	
   1.1.2	
   Adverse Drug Reactions in Children ...................................................................... 2	
   1.1.3	
   Pharmacogenomics ................................................................................................. 4	
   1.2	
   Vincristine .................................................................................................................... 5	
   1.2.1	
   The Antineoplastic Mechanism of Action .............................................................. 6	
   1.2.2	
   Metabolism and Transportation .............................................................................. 6	
   1.2.3	
   Clinical Use of Vincristine in Children .................................................................. 9	
   1.3	
   Vincristine-Induced Neurotoxicity ............................................................................. 10	
   1.3.1	
   Signs and Symptoms ............................................................................................ 10	
   1.3.2	
   Clinical Diagnosis of Vincristine-Induced Neurotoxicity .................................... 11	
   1.3.3	
   Management ......................................................................................................... 13	
   1.3.4	
   Future Management: Neuroprotectants ................................................................ 14	
    vi 1.3.5	
   Pathophysiology ................................................................................................... 15	
   1.4	
   Factors Affecting the Exposure to Vincristine ........................................................... 17	
   1.4.1	
   Dose Intensity, Frequency, and Cumulative Dose ................................................ 17	
   1.4.2	
   CYP3A Inhibitors ................................................................................................. 19	
   1.4.3	
   Genetic Variation in CYP3A4 and CYP3A5, and Ancestry ................................ 20	
   1.4.4	
   Vincristine Transporters ....................................................................................... 21	
   1.4.5	
   Previous Pharmacogenomic Studies ..................................................................... 22	
   1.5	
   Patient Cohort Selection: Wilms Tumor and Rhabdomyosarcoma ............................ 24	
   Chapter  2: Hypothesis, Project Goal and Objectives ........................................................ 28	
   2.1	
   Hypothesis .................................................................................................................. 28	
   2.2	
   Project Goal and Objectives ....................................................................................... 28	
   Chapter  3: Methodology ...................................................................................................... 29	
   3.1	
   Recruitment, Consent, and Establishing Causality ..................................................... 29	
   3.2	
   Defining Cases and Controls ...................................................................................... 31	
   3.3	
   Clinical Characterization of Vincristine-induced Neurotoxicity Cases ..................... 33	
   3.3.1	
   Classifying Each Vincristine-Induced Neurotoxicity Event by Type of Neuropathy ........................................................................................................... 33	
   3.3.2	
   Evaluating the Severity of Each Vincristine-Induced Neurotoxicity Event ......... 33	
   3.4	
   Study Inclusion and Exclusion Criteria ...................................................................... 40	
   3.5	
   Calculating Dose and Time to First Adverse Drug Reaction ..................................... 41	
   3.6	
   Association Studies ..................................................................................................... 42	
   3.6.1	
   General Drug Biotransformation Genotyping ...................................................... 42	
   3.6.2	
   Candidate Gene Analysis ...................................................................................... 43	
    vii 3.7	
   Genotyping: Illumina GoldenGate Assay ................................................................... 43	
   3.8	
   Quality Control and Assurance in Genotypic Data .................................................... 44	
   3.8.1	
   Quality of Patient Samples and SNPs ................................................................... 44	
   3.8.2	
   Data Cleaning ....................................................................................................... 45	
   3.9	
   Statistical Analyses: Logistic Regression Models and Manhattan Plot ..................... 46	
   3.10	
   Prediction Model ........................................................................................................ 48	
   3.11	
   Linkage Disequilibrium Studies ................................................................................. 49	
   3.12	
   Imputation Analyses ................................................................................................... 49	
   Chapter  4: Results ................................................................................................................ 51	
   4.1	
   Patient Cohort ............................................................................................................. 51	
   4.2	
   Patient Demographics and Clinical Factors ................................................................ 52	
   4.3	
   Assessing the Severity and Types of Vincristine-Induced Neurotoxicity and the Interventions Provided ............................................................................................... 55	
   4.4	
   Variation by Ancestry ................................................................................................. 60	
   4.5	
   Genetic Analyses ........................................................................................................ 62	
   4.5.1	
   Screening for Genes Involved in the General Drug Biotransformation Pathways  ............................................................................................................................. 63	
   4.5.2	
   Candidate Study .................................................................................................... 78	
   Chapter  5: Discussion ........................................................................................................... 84	
   5.1	
   The Incidence of Vincristine-Induced Neurotoxicity ................................................. 84	
   5.2	
   Clinical Factors ........................................................................................................... 85	
   5.2.1	
   Age ........................................................................................................................ 85	
   5.2.2	
   Tumor Type .......................................................................................................... 87	
    viii 5.2.3	
   Ancestry ................................................................................................................ 88	
   5.2.4	
   Concomitant Medications: Steroids and CYP3A4 Inhibitors ............................... 88	
   5.3	
   Vincristine-Induced Neurotoxicity: Types of Neurotoxicity and Severity ................ 89	
   5.4	
   Novel Genes and Genetic Variants Potentially Associated with Vincristine-Induced Neurotoxicity .............................................................................................................. 90	
   5.4.1	
   PON1 (Paraoxonase 1) and PPP1R9A (Neurabin 1) ............................................ 92	
   5.4.2	
   ABCA4, ABCG1, and CYP51A1 ......................................................................... 102	
   5.4.3	
   SLCO1C1 ............................................................................................................ 108	
   5.5	
   Previously Identified Polymorphisms and Candidate Genes Involved in the Biotransformation of Vincristine ............................................................................. 108	
   5.5.1	
   CYP3A5 and CYP3A4 ......................................................................................... 110	
   5.5.2	
   ABCB1 ................................................................................................................ 111	
   5.5.3	
   Other Candidate Genes ....................................................................................... 112	
   5.6	
   Inter-Ethnic Differences ........................................................................................... 113	
   5.7	
   Limitations ................................................................................................................ 114	
   5.8	
   Future Directions ...................................................................................................... 116	
   5.8.1	
   Expansion, Replication, and Validation of Novel SNPs and Genes ................... 116	
   5.8.2	
   Detailed Genetic Investigation of Novel Candidate Genes ................................ 117	
   5.8.3	
   Pharmacokinetic and Functional Validation Studies .......................................... 118	
   5.8.3.1	
   PON1 (Paraoxonase 1) and PPP1R9A (Neurabin 1) ................................... 119	
   5.8.3.2	
   ABCA4, ABCG1, and CYP51A1: Cholesterol Biosynthesis and Intracellular Transport ...................................................................................................... 120	
   5.8.3.3	
   Metabolizers and Transporters .................................................................... 121	
    ix 5.8.4	
   Investigation of Additional Genes Potentially Involved in VIN ........................ 121	
   5.8.5	
   Observational Trials ............................................................................................ 123	
   5.9	
   Future Implications ................................................................................................... 123	
   Chapter  6: Conclusion ........................................................................................................ 126	
   References ............................................................................................................................. 128	
   Appendix A ........................................................................................................................ 139	
   A.1	
   Excluded SNPs: Less than 98% Completion Rate Across Patients ..................... 139	
   A.2	
   Excluded SNPs: Hardy-Weinberg Disequilibrium .............................................. 140	
   A.3	
   Significant SNPs in the ADME-Tox Panel of General Drug Biotransformation 141	
   A.4	
   PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 Risk Alleles Increase the Chances of Developing Vincristine-Induced Neurotoxicity ............................... 146	
   A.5	
   Imputed Regions with Base Pair Limits Used ..................................................... 147	
   A.6	
   Manhattan Plot for Imputed SNPs ....................................................................... 148	
     x List of Tables Table 1 	
    Grading Vincristine-Induced Neurotoxicity with the CTCAE Criteria (v4.03) .... 34	
   Table 2 	
    Grading Vincristine-Induced Neurotoxicity by Intervention ................................ 38	
   Table 3 	
    Patient Demographics and Clinical Factors Between Cases and Controls ............ 53	
   Table 4 	
    Vincristine-Induced Neurotoxicity and the Interventions Provided to Wilms Tumor and Rhabdomyosarcoma Cases ................................................................ 56	
   Table 5 	
    The Frequency of Each Type of Vincristine-Induced Neurotoxicity Sign or Symptom ............................................................................................................... 58	
   Table 6 	
    Genetic Variants that are Significantly Associated with Vincristine-Induced Neurotoxicity ........................................................................................................ 65	
   Table 7 	
    Subgroup Analysis of Genetic Variants that are Significantly Associated with Vincristine-Induced Neurotoxicity: Stratification of Patients by European Ancestry ................................................................................................................ 69	
   Table 8 	
    Linkage Disequilibrium of SNP Pairs: PON1, PPP1R9A, PON1, and ABCA4 .... 71	
   Table 9	
    PON1 rs854549, PPP1R9A rs705377, and PON1 rs854560 and their Association with Vincristine-Induced Neurotoxicity ............................................................... 76	
   Table 10	
   The Effects of SNPs Previously Identified to be Associated with Vincristine- Induced Neurotoxicity .......................................................................................... 79	
   Table 11	
   The Effects of Previously Identified Genes Involved in the Metabolism and Transportation Pathway of Vincristine ................................................................. 82	
      xi List of Figures Figure 1 	
   The Pharmacokinetics and Pharmacodynamics of Vincristine ............................... 7	
   Figure 2 	
   Ethnic Distribution of the Patients in the Cohort .................................................. 60	
   Figure 3 	
   Manhattan Plot ...................................................................................................... 63	
   Figure 4 	
    SNP Clusters of PON1 rs854549, ABCA4 rs3789433, SLCO1C1 rs10770704, ABCG1 rs221948, CYP51A1 rs7797834, and ABCA4 rs549848 .......................... 66	
   Figure 5	
    PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 Risk Alleles Increase the Chances of Developing Vincristine-Induced Neurotoxicity ................................. 69	
   Figure 6	
    Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the European (CEU) Hapmap Populations ........................................................... 71	
   Figure 7	
    Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the African (YRI) Hapmap Populations ............................................................... 72	
   Figure 8	
    Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the Asian (CHB, JPT) Hapmap Populations ......................................................... 73	
   Figure 9	
    SNP Clusters of PPP1R9A rs705377 and PON1 rs854560 .................................. 76	
   Figure 10	
    Clustering of Candidate SNPs: CYP3A5 and CYP3A4 ........................................ 79	
   Figure 11	
    Clustering of Candidate SNPs: ABCB1 ............................................................... 80	
   Figure 12 	
   The Theorized Role of the Novel Genes in the Metabolism and Biotransformation of Vincristine .......................................................................... 90	
   Figure 13	
   Chemical Structures of Methyl Phenylacetate, Vincristine, and Clopidogrel ...... 96	
   Figure 14	
   Cholesterol Biosynthesis .................................................................................... 103	
     xii List of Symbols and Abbreviations ABCA4  ATP-binding cassette, sub-family B, member 4; multidrug resistance protein 3 ABCB1 ATP-binding cassette, sub-family B, member 1; multidrug resistance protein 1; P-glycoprotein ABCB4 ATP-binding cassette, sub-family B, member 4; multidrug resistance protein 4 ABCC1 ATP-binding cassette, sub-family C, member 1; multiple drug resistance protein 1 ABCC2 ATP-binding cassette, sub-family C, member 2; multiple drug resistance protein 2 ABCC3 ATP-binding cassette, sub-family C, member 3; multiple drug resistance protein 3 ABCC4 ATP-binding cassette, sub-family C, member 4; multiple drug resistance protein 4 ABCC5 ATP-binding cassette, sub-family C, member 5; multiple drug resistance protein 6 ABCC6 ATP-binding cassette, sub-family C, member 6; multiple drug resistance protein 6 ABCC7 ATP-binding cassette, sub-family C, member 7; multiple drug resistance protein 7 ABCC8 ATP-binding cassette, sub-family C, member 8; multiple drug resistance protein 8 ABCC10 ATP-binding cassette, sub-family C, member 10; multiple drug resistance protein 10  xiii ABCG1  ATP-binding cassette, subfamily G, member 1 ADL  Activities of Daily Living ADME Absorption, Distribution, Metabolism, and Excretion ADME-Tox Absorption, Distribution, Metabolism, Excretion, and Toxicity ADR  Adverse Drug Reaction ALL  Acute Lymphoblastic Leukemia ALS  Amyotrophic Lateral Sclerosis AN  Autonomic Neuropathy ANS  Autonomic Nervous System AUC  Area-Under-the-Concentration BSA  Body Surface Area CEU  Northern and Western European Ancestry (Utah, United States) CHB  Han Chinese Ancestry (Beijing, China) CHD  Chinese Ancestry (Denver, Colorado) CI  Confidence Interval CMT  Charcot Marie Tooth CNS  Central Nervous System COG  Children’s Oncology Group CPNDS Canadian Pharmacogenomics Network for Drug Safety CTEP  Cancer Therapy Evaluation Program CTCAE  Common Terminology Criteria for Adverse Events CYP450 Cytochromes P450 CYP3A4 Cytochrome P450, family 3, subfamily A, polypeptide 4  xiv CYP3A5 Cytochrome P450, family 3, subfamily A, polypeptide 5 CYP51A1  Cytochrome P450, family 51, subfamily A, polypeptide 1; Lanosterol; 14-α- Demethylase DNA  Deoxyribonucleic Acid DTR  Deep Tendon Reflex GWAS Genome-Wide Association Studies HDL  High-density Lipoprotein HWE  Hardy-Weinberg Equilibrium IADL  Instrumental Activities of Daily Living IBS  Identity-by-State IBD  Identity-by-Descent IGF-1  Insulin-like Growth Factor 1 JPT  Japanese Ancestry (Tokyo, Japan) LD  Linkage Disequilibrium m-ABC Motor Assessment Battery for Children MAF  Minor Allele Frequency MAPT  Microtubule-Associated Protein Tau MAP2  Microtubule-Associated Protein 2 Nb1  Neurabin 1 ncRNA  Non-Protein-Coding RNA NCS  Nerve Conduction Studies NCI-CTCAE  National Cancer Institute – Common Terminology Criteria for Adverse Events NGF Nerve Growth Factor  xv NIEHS National Institute of Environmental Health Sciences NrPE N-Retinylidene-Phosphatidylethanolamine NWTS  National Wilms Tumor Study OP  Organophosphate OR  Odds Ratio OT  Occupational Therapist PCA  Principal Component Analysis PCR  Polymerase Chain Reaction PN  Peripheral Neuropathy; Peripheral Neuritis PNET  Primative Neuroectodermal Tumor PON1   Paranoxonase 1 PPP1R9A Protein phosphatase 1, regulatory (inhibitor) subunit 9A PT  Physiotherapist Prin1  Principal Components 1 Prin2  Principal Components 2 RALBP1 RalA binding protein 1 ROM  Range of Motion SIADH Syndrome of Inappropriate Anti-Diuretic Hormone SLCO1C1  Solute carrier organic anion transporter family, member 1C1 SNP  Single Nucleotide Polymorphism TNS   The Total Neuropathy Score TPN   Total Parenteral Nutrition UGT  UDP glucuronosyltransferase  xvi UGT1A9 UDP glucuronosyltransferase 1 family, polypeptide A9 UGT2B4 UDP glucuronosyltransferase 2 family, polypeptide B4 US  United States UTR  Untranslated Region VIN  Vincristine-Induced Neurotoxicity WGA  Whole Genome Amplified YRI  Yoruba African Ancestry (Ibadan, Nigeria)  xvii Acknowledgements  First and foremost, I would like to acknowledge and thank the patients and their families for their commitment and integral role in drug safety research – without whom none of this would have been possible.  Additional thanks goes to CPNDS investigators, surveillance clinicians, and collaborators, especially Ursula Amstutz and Marie-Pierre Dubé for their constructive input and advice.  Lastly, I would like to thank Bruce Carleton, Rod Rassekh, Colin Ross, and Ran Goldman for their invaluable support and guidance, as well as for providing their collective expertise in reducing drug harm in children.    1 1 Chapter  1: Introduction  1.1 Adverse Drug Reactions An adverse drug reaction (ADR) is defined as "an appreciably harmful or unpleasant reaction, resulting from an intervention related to the [appropriate] use of a medicinal product [at appropriate doses], which predicts hazard from future administration and warrants prevention or specific treatment, or alteration of the dosage regimen, or withdrawal of the product" (EDWARDS and ARONSON 2000). Given this definition, ADRs are ranked as the fifth leading cause of death in the United States (US) (WHITE et al. 1999) and are estimated to cause over 2,000,000 hospitalizations and 100,000 fatalities, which includes over 26,500 mortalities in children, annually (LAZAROU et al. 1998). In the United Kingdom, a large prospective analysis of 18,820 hospital admissions at one hospital attributed 6.5% of these admissions to an ADR (PIRMOHAMED et al. 2004). Clearly, ADRs pose a significant cause of morbidity and mortality. However, the true incidence and impact of ADRs are unknown as it is estimated that 80 to 95% of ADRs are never reported to regulating authorities (HAZELL and SHAKIR 2006; LAZAROU et al. 1998; MOORE et al. 1998). Furthermore, the ADR reports of drug safety concerns that are actually reported are often incomplete and lack important details of clinical information (US GOVERNMENT ACCOUNTABILITY OFFICE 2006). As an example of ADR underreporting, a study showed that only 4% of cases of a life-threatening drug-induced skin reaction, toxic epidermal necrolysis, were reported to the Canadian ADR Monitoring Program (CADRMP) (MITTMANN et al. 2004). Additionally, the costs associated with ADRs can be enormous. In the US, the total estimate of ADR-related health care   2 2 morbidity and mortality expenditures ranges between $30 billion and $130 billion US dollars, annually (WHITE et al. 1999).  1.1.1 Factors that Contribute to Adverse Drug Reactions Many factors contribute to the development of an ADR, including age, weight, diet, drug use compliance, as well as the patient’s disease state, drug–drug interactions, and genetic variation. With important implications for the efficacy and toxicity of a drug, these complex parameters can influence drug absorption, distribution, metabolism, and excretion (ADME). Furthermore, in spite of the growing demand for therapeutic drug use, it is not feasible to conduct multiple clinical trials in every population to evaluate these factors; even though such work would aid in assessing the efficacy and toxicity of a drug, and would provide valuable information on appropriate drug dose and schedules. Therefore, drugs are increasingly utilized for off-label purposes, which extends the application of a therapeutic agent beyond the specific and defined population in which its safety and efficacy was originally assessed. Thus, the use of these drugs in individuals for which there is no safety and efficacy data may result in ADRs that were not identified in the original clinical trials.  1.1.2 Adverse Drug Reactions in Children ADRs occur more frequently in children than in adults (NEUBERT et al. 2004). One of the reasons for this is that ethical concerns have protected children from being included in clinical trials that would evaluate the safety and efficacy of drugs, as well as establish appropriate drug dose and schedules in the paediatric population. Consequently, there is a   3 3 lack of medications with approved drug indications for children. Without alternatives, paediatric drug prescribing relies heavily on empiric drug use and/or weight-based dosing extrapolations from adult clinical trials (STEPHENSON 2005; WONG et al. 2003), both of which increase the risk of ADRs. Dosage adjustments that are based solely on weight are often insufficient because they do not account for the extensive physiological differences between adults and children that affect the pharmacokinetics and pharmacodynamics of drugs. For example, children have thinner skin and newborns lack gastric acid, resulting in faster epidermal and gastrointestinal absorption rates, respectively. Children also have a higher percentage of body fat and water, which affects the distribution of fat- and water- soluble drugs (WOODROW 2006). Furthermore, neonates have an increased permeability of the blood-brain barrier which leads to higher drug exposure in the CNS (WOODROW 2006). Additionally, developmental changes in system functionality, organ function and gene expression levels (KEARNS et al. 2003; STRASSBURG et al. 2002) can affect the metabolism and excretion of drugs. For example, renal function is not completely developed in neonates. Moreover, genes encoding for enzymes involved in drug metabolism may be differentially regulated throughout the development of a child (WOODROW 2006). To illustrate, mRNA levels in both UDP glucuronosyltransferase 1 family, polypeptide A9 (UGT1A9) and UDP glucuronosyltransferase 2 family, polypeptide B4 (UGT2B4) are lower in children as compared to adults (STRASSBURG et al. 2002). Since these two UGT enzymes are responsible for glucuronidation, which is an important step in the inactivation of drugs, a decrease in the activity of these two enzymes can result in decreased drug detoxification and an increased potential for ADRs to occur.    4 4  1.1.3 Pharmacogenomics Combining the fields of pharmacology and human genetics, pharmacogenomics aims to discover relevant genetic variants that influence drug response and toxicity. Since genetic factors potentially account for up to 95% of inter-patient variability in drug response (KALOW et al. 1998), knowledge about genetic variation affecting drug efficacy and toxicity could be used to identify patients who are unlikely to respond to a drug or are at increased risk for ADRs. Unlike clinical variables that can change over time, the genetic code is static and thus, could provide clinicians with reliable marker(s) to identify patients with specific drug-risk variants prior to drug use. Increasingly, drug regulators such as the US Food and Drug Administration (FDA) are recommending the use of genetic tests in clinical practice prior to the implementation of drug therapy ( LOO et al. 2010), which showcases the added- value of pharmacogenomics testing. The ability to harness pharmacogenomics in the decision making process is increasingly becoming a reality; rapid technological advancements have improved assay completion times and significantly reduced cost, which is facilitating the translation of these scientific discoveries into valuable, efficient, and feasible predictive tests. Pharmacogenomics has the greatest application to medications that are traditionally dosed empirically, drugs with high inter-individual variability in response or toxicity, and drugs with narrow therapeutic windows, such as highly toxic oncology medications. Ultimately, the implementation of evidence-based pharmacogenomics research in clinical practice will serve to inform public policy and influence drug benefit-risk decision-making.     5 5 1.2 Vincristine Vincristine, the most commonly used Vinca alkaloid, is a potent anti-cancer drug that is associated with high cure rates, long-term relapse-free survival, and broad effectiveness against a myriad of cancers (GIDDING et al. 1999). Most notably, vincristine is rarely associated with myelosuppression, a property that has limited the use of many other antineoplastic agents. Together, these factors have contributed to the establishment of vincristine as a core component of most paediatric oncology treatment protocols, ever since the 1960s, when it was introduced into clinical practice (GIDDING et al. 1999). As such, vincristine is used in the treatment of numerous haematologic malignancies (i.e.: acute lymphoblastic leukemia (ALL), Hodgkin’s lymphoma, non-Hodgkin’s lymphoma) and solid tumors (i.e.: nephroblastomas, rhabdomyosarcomas, hepatoblastomas, neuroblastomas, medulloblastomas, primative neuroectodermal tumor (PNET)) (GIDDING et al. 1999). It is also a mainstay in many adult chemotherapy protocols, which, in addition to the aforementioned cancers, include breast cancer and multiple myelomas (GIDDING et al. 1999). Although vincristine was originally used as a monotherapy, its synergistic effect with other chemotherapy drugs and/or radiation therapy has resulted in dramatically improved response rates (CRIST et al. 2001). ADRs to vincristine however, are a problem for some patients. Depending on the tumor type, 4 to 42% of patients receiving vincristine will experience debilitating vincristine- induced neurotoxicity (VIN) (ARNDT et al. 1997; PORTER et al. 2009). Owing to the lack of alternative chemotherapy agents, oncologists consider the removal of vincristine from protocol to likely result in suboptimal outcomes (ASCO ONCOLOGY 2010). Thus, a risk- benefit assessment is essential to balance the devastating outcomes associated with severe or   6 6 life-threatening forms of VIN with a possible compromise on cure rates that are associated with reducing or withholding doses, as well as the discontinuation of vincristine.  1.2.1 The Antineoplastic Mechanism of Action Vincristine is a mitotic inhibitor that halts cells in the metaphase stage of mitosis. By binding to the building blocks of microtubules (α and β tubulin heterodimers), vincristine prohibits the addition and removal of microtubules to the ends of mitotic spindle microtubules. This interference in mitosis spindle formation and function impedes cell division, consequently forcing cellular destabilization and Bcl-2-mediated apoptosis (JORDAN et al. 1991) as illustrated by Figure 1 (PHARMGKB 2011).  1.2.2 Metabolism and Transportation Vincristine has been shown to be metabolized by two enzymes in the cytochrome P450 family (CYP450), CYP3A5 and CYP3A4 (Figure 1) (PHARMGKB 2011); to date, no other drug metabolizing enzymes have been shown to metabolize vincristine. As shown by studies in cDNA-expressed human P450s, CYP3A5 appears to be the more important of these two known metabolizers of vincristine. In comparison to CYP3A4, it metabolizes vincristine at a rate that is 9 to 14-times higher (DENNISON et al. 2006). Genetic polymorphisms in CYP3A5 have been shown to play a role in determining the extent of metabolism of vincristine (DENNISON et al. 2007; DENNISON et al. 2008). The hepatic clearance of vincristine in human liver microsomes has been shown to be five-times higher in individuals with a high CYP3A5 expression as compared to low expressors of CYP3A5   7 7                This diagram by PharmGKB illustrates the involvement of specific genes in the transportation ( ) and metabolism ( ) of vincristine. The parent drug (vincristine, ) is metabolized into threeproducts ( ). Vincristine prohibits the polymerization of microtubules by binding to tubulin. The biliary excretion of vincristine is mediated by ABCB1 and ABCC2, whereas vincristine is transported into the blood with ABCC1. Apoptosis is mediated by the phosphorylation of Bcl-2 and the subsequent inactivation of Bcl-2 to heterodimerize with BAX. In addition to the inactivation of Bcl-2, apoptosis is caused by increasing levels of p53 and p21. This figure is reproduced with permission from PharmGKB and Stanford University (PHARMGKB 2011). Copyright PharmGKB.  Figure 1  The Pharmacokinetics and Pharmacodynamics of Vincristine   8 8 (DENNISON et al. 2007). Pharmacokinetic studies have only recently identified one major metabolite (a secondary amine, M1), as well as two minor metabolites (M2 and M4) of vincristine (DENNISON et al. 2006). This is due to the high instability of the metabolites (DENNISON et al. 2006) and the difficulty in distinguishing them from the parent drug (GIDDING et al. 1999). The relative bioactivity of vincristine and its metabolites has yet to be determined (DENNISON et al. 2006). Pharmacokinetic studies have shown that ATP-binding cassette, sub-family C, member 1 (ABCC1) (STRIDE et al. 1997), ATP-binding cassette, sub-family C, member 2 (ABCC2) (STRIDE et al. 1997), ATP-binding cassette, sub-family C, member 10 (ABCC10), ATP-binding cassette sub-family B member 1 (ABCB1, also known as P-glycoprotein) (STRIDE et al. 1997), and ralA binding protein 1 (RALBP1) are transporters of vincristine (Figure 1) (DRAKE et al. 2007). The ABC transporter family is well known for promoting the efflux and the transportation of substrates across the cellular membrane. Drug resistance studies have demonstrated that ABCB1, ABCC1, and ABCC2 can facilitate drug resistance by increasing the efflux of vincristine (STRIDE et al. 1997), and gene expression studies have shown that ABCB1, ABCC1, ABCC2, and ATP-binding cassette sub-family C member 3 (ABCC3) are upregulated upon exposure to increasing concentrations of vincristine (STRIDE et al. 1997). One study has also investigated the interaction between vincristine and other members of the ABC transporter family (ATP-binding cassette sub-family C member 4 to 6, ABCC4 to 6 and ATP-binding cassette sub-family C member 8, ABCC8), and concluded that these transporters were not involved in the transportation of vincristine, nor did they confer drug resistance (LEVEQUE and JEHL 2007). However, with only a select few studies available, more research is required to provide a more comprehensive understanding of the effects and   9 9 the relative contribution of the transporters and metabolizers that are associated with vincristine.  1.2.3 Clinical Use of Vincristine in Children The most common childhood cancers are haematologic malignancies, where most children are afflicted with ALL (GIDDING et al. 1999). With vincristine as a mainstay in the treatment protocol, this cancer is treatable if detected early (GIDDING et al. 1999). Cancer type, protocol arms, relapse risk (low, standard, high, very high), gender, age, and response (rapid, slow) are a variety of factors that determine how vincristine is used within protocols, affecting dose intensity, dose frequency, length of duration, and cumulative dose. The dose intensity is fairly standardized across cancer types and protocols. For children over three-years of age, vincristine is dosed based on surface area at 1.5mg/m2/dose, with a maximum dose of 2mg/dose. Depending on protocol, children aged one to three-years are dosed by weight at 0.5mg/kg/dose or 1.5mg/m2/dose, where both of these weight-based and body surface area-based doses are considered as approximately equivalent. Dose recommendations for infants less than one-year of age vary by tumor type and protocol, ranging from weight-based dosing of 0.17mg/kg/dose to 0.05mg/kg/dose (e.g.: 0.17mg/kg/dose for neuroblastomas, 0.025mg/kg/dose for Wilms tumor and rhabdomyosarcoma, 0.03 to 0.05mg/kg/dose for infant ALL), BSA-based dosing (e.g.: 0.6mg/m2/dose for Ewings sarcoma), to dose escalation (e.g.: first dose at 50%, second dose at 75%, third dose at 100% for Ewings sarcoma). The frequency of vincristine doses can be highly variable. Within one protocol, a patient could receive doses that can change over time, ranging from weekly doses, bi-   10 10 monthly doses, to doses every two months. Depending on the phase of protocol, patients with leukemia or lymphoma receive vincristine once every seven to 60 days, for a total of 1.5 to three years. Patients with solid tumors (e.g.: Wilms tumor, rhabdomyosarcoma, Ewing sarcoma, hepatoblastomas, hepatoblastomas, and hepatic cellular carcinomas) have less variable vincristine doses that range from once every week to every six weeks, for a total of six to nine months.  1.3 Vincristine-Induced Neurotoxicity 1.3.1 Signs and Symptoms Despite its high clinical utility, the use of vincristine in clinical practice is limited by its debilitating neurotoxicities. These iatrogenic toxicities predominantly manifest as peripheral neuropathy (PN, also known as peripheral neuritis), autonomic neuropathy (AN), and central nervous system (CNS) effects. VIN mainly occurs as a mix of polyneuropathies affecting both the sensory and motor system as a result of length-dependent axonopathy and myelinopathy (GIDDING et al. 1999; HANKEY GJ 2002). Although the development and severity of neuropathies varies between individuals, VIN usually progresses in a relatively similar pattern. Early neurotoxic events (ranging from hours after the first dose to after a few doses of vincristine) include jaw pain and loss of ankle-jerk reflexes as well as distal paresthesias. These events progress to include the development of pain in other locations (e.g.: bone, muscle, joints) of the body that typically distributes in a stocking-glove fashion. Delayed neurotoxic events manifest as specific effects on motor function such as foot drop (typically characterized by the observation of foot slapping), which is caused by the lack of toe and foot extensor function and loss of stretch reflexes. Additionally, increasing muscle   11 11 weakness of the limbs (particularly the legs), begins distally, with the potential to extend to the proximal regions. Late-occurring events, includes cranial nerve involvement (ptosis, hoarseness, vocal cord paralysis, dysphagia) and AN (constipation, ileus, rarely orthostatic hypotension and urinary retention) typically develop after the earlier sensory, motor and muscle weakness symptoms; however, these events can also occur independently. Vincristine-induced CNS effects (Syndrome of Inappropriate Anti-Diuretic Hormone (SIADH) and subsequent seizures) occur infrequently and are more often observed in individuals with disrupted function of the blood-brain barrier (GIDDING et al. 1999). The neuropathic symptoms that are distinctly unique to VIN include: severe jaw pain, ptosis, vocal cord paralysis, muscle weakness, foot drop, and reduced foot dorsiflexion (GIDDING et al. 1999). Steroid use can cause myopathy and soft tissue or joint pain, which can be mistaken for vincristine-induced muscle weakness. Additionally, steroids can induce CYP3A enzymes, which could potentially increase the metabolism of vincristine, decrease the incidence and severity of other neurotoxicites, and therefore, impede the detection of genetic effects (KISHI et al. 2004). Thus, since concomitant use of these drugs could cause inherent biases in the study (KISHI et al. 2004), studies evaluating the causes of VIN will need to consider the influence of steroid use.  1.3.2 Clinical Diagnosis of Vincristine-Induced Neurotoxicity To conclusively confirm the neurologic symptoms associated with VIN, a variety of neurological exams may be conducted. Nerve conduction studies (NCS) and electromyograms (EMGs) are tests that determine the speed of signal transmission along the   12 12 nerves and record muscle electrical activity, respectively. However, given the high prevalence of VIN, these resource-intensive tests are rarely used in clinical practice to definitively identify patients with VIN. Pathologic studies on patients with neuropathy have confirmed that slow NCS speeds are indicative of axonal neuropathy ( BRADLEY et al. 1970), and that patients with muscle weakness show denervation and smaller amplitudes of motor nerve action potential in their EMGs. Together, NCS and The Total Neuropathy Score (TNS, a grading tool used to assess chemotherapy-induced neurotoxicity) are considered as the gold standard of comprehensively evaluating the subjective and objective components of neuropathy (ENGLAND et al. 2005). However, the TNS is highly intensive and time consuming scoring tool to perform. Therefore, again, owing to the high prevalence of VIN and the impracticality of using TNS to evaluate neuropathy in the clinic, the majority of these patients do not receive these specialized tests. Instead, the most widely utilized tool to assess neuropathy is the Common Terminology Criteria for Adverse Events (CTCAE), published by the Cancer Therapy Evaluation Program (CTEP) (CANCER THERAPY EVALUATION PROGRAM - NATIONAL CANCER INSTITUTE 2010). The criterion within this composite measurement tool allows VIN to be graded in a standardized manner on a scale from 0 (no toxicity) to 5 (lethal adverse event). In addition to correlating well with the findings of TNS, the universality of this validated grading tool that is often utilized in clinical practice allows for inter-study comparisons. Furthermore, this test is less intensive and also comprehensively standardizes the classification and severity of adverse events.      13 13 1.3.3 Management Generally, the current practice for vincristine regimens is to provide interventions (i.e.: rehabilitation therapy, neurology consults, pharmacotherapy to treat neurotoxicity symptoms, hospitalization and surgery) for VIN symptoms that are moderate or worse, and to modify the protocol once the patient exhibits severe VIN symptoms. However, all dose modifications are made in consideration of individual cases, cancer type, protocol, and aggressiveness of cancer. This empirical approach allows for chemotherapy treatment to be individualized in order to maximize benefit while avoiding risk of ADRs. Nevertheless, missed doses of vincristine in ALL patients and reduction in doses for non-Hodgkin lymphoma patients are associated with increased frequencies of relapses (GIDDING et al. 1999). Severe forms of VIN negatively affect a patient’s quality of life and promote reliance on others. For example, VIN can result in excruciating pain that requires narcotic therapy as a short-term treatment. Additionally, mobilization difficulties of the limbs may affect walking (legs) and the ability to feed oneself (arms), thereby requiring assistive devices as well as extensive therapy and assistance to maintain functionality. Furthermore, VIN can lead to life-threatening conditions that include the development of paralytic ileus or severe constipation, requiring colostomy to treat perforated bowels, as well as unilateral or bilateral vocal cord paralysis that may require tracheotomy or a bilevel positive-pressure ventilation (TOGHILL and BURKE 1970). The cost of managing VIN is high. For instance, patients with foot drop and reduced dorsiflexion may require wheelchairs or costly custom-fitted ankle foot orthotics, which range in price from $2,000 to $2,400. These devices serve to assist with walking and to   14 14 prevent long-term hip pain. Furthermore, ongoing therapy sessions with physical and occupational therapists (PT, OT) are needed to provide exercises and support to improve strength and fine movement, as well as to monitor the use of assistive devices for mobility. Additional costs include increased hospitalization time (e.g.: SIADH, vocal cord paralysis), as well as more drastic interventions such as surgery (e.g.: constipation, ileus). Preventative measures to reduce the incidence of these ADRs would alleviate these costly burdens. Although it was originally thought that VIN is reversible over time, there is increasing evidence suggesting that this is not necessarily true for all types of VIN. In fact, severely affected patients are unlikely to return to their baseline level of function, especially in regards to fine motor movements, likely resulting from the inability to regenerate nerves (GIDDING et al. 1999). The length of recovery time from neurotoxicity depends on the severity and type of VIN. For example, deep tendon reflex (DTR) abnormalities and gross and fine motor disorders can last for two to seven years after the end of treatment (GIDDING et al. 1999). The reversibility of symptoms depends on their severity, when the VIN was identified and treated, type of neurotoxicity, the treatments provided, and the patient’s genetic make up (GIDDING et al. 1999). For example, genetic factors are thought to be responsible for the wide inter-patient variability in recovery time from vocal cord paralysis (KURUVILLA et al. 2009).  1.3.4 Future Management: Neuroprotectants Nerve growth factors (NGFs) that promote lengthening of regenerated neurites are being studied as neuroprotectants against VIN (e.g.: lipoic acid and glutamic acid) (HAYAKAWA et al. 1994). Recent studies have shown that VIN does not develop in mice when motor and sensory neuron regeneration is promoted with insulin-like growth factor 1   15 15 (IGF-1) (CONTRERAS et al. 1997). However, the proteins affecting the distribution, regulation, and associated transporters of these neuroprotectants have not been identified. The current work in neuroprotectants is limited to the limited knowledge of the functionality and susceptibility of nerve proteins to VIN.  1.3.5 Pathophysiology Axon degeneration, demyelination, and the prevention of nerve regeneration after vincristine-induced damage are theorized to be the main three causes of VIN (GIDDING et al. 1999; PAN et al. 2003). Studies have shown that vincristine modulates its toxic effect through length-dependent axonal degeneration, corresponding with length-dependent sensorimotor neuropathy manifesting in a distal to proximal fashion (WANG et al. 2000). The speed of vincristine-induced axonal degeneration is dependent on the concentration of vincristine, which explains why higher dose intensities are much more likely to incur more severe and frequent types of neuropathy (WANG et al. 2000). Axonal transport is important for signaling and for providing nutrition to the neuronal cell body; permanent damage would likely be sustained once the perikaryon is reached (GIDDING et al. 1999). Additionally, vincristine causes shortening of neurite length, accompanied with dose-dependent limits on axon growth (HAYAKAWA et al. 1994; HOUI et al. 1993). Furthermore, vincristine causes demyelination of the myelin sheaths (GIDDING et al. 1999). Vincristine also causes rapid and temporary blockage of nerve regeneration; an observation validated by the reversal of such blockage with a neuroactive compound (PAN et al. 2003). The exact pathophysiologies of these morphological alterations in relationship with vincristine are largely unknown.   16 16 Vincristine’s indiscriminant interference in microtubule polymerization is key in the inhibition of cancerous cell proliferation. However, vincristine also has the potential to cause cellular death or impair normal cell function in non-dividing cells without affecting mitotic arrest (HERRUP and YANG 2007). In addition to being a core component of cellular division, microtubulin is important for maintaining cellular shape, polarity, motility, and in providing a scaffold for cellular trafficking of proteins and organelles. In particular, tubulin comprises over 20% of the soluble protein in neural cells; since these cells rely on tubulin for proper intracellular transport and synaptic function, these cells are also highly sensitive to tubulin levels (BAAS 1997; LAFERRIERE et al. 1997). Microtubule-associated protein τ (MAPT) promotes assembly and stabilization of microtubules. Dysregulation of this protein has been associated with functional abnormalities, including neurodegenerative disorders (Parkinson’s disease, Alzheimers disease, and progressive supranuclear palsy). The ability of vincristine to cause death in somatic nerve cells has been attributed to the inhibition of axonal microtubules and subsequent axonal degeneration, which negatively affects axonal transport and neuron transmission. Recently, it has been confirmed that microtubule-targeting agents (taxanes, Vinca alkaloids, and epothilones) cause in vitro microtubule destabilization in neural, non- neural, and cancer cells (HUFF et al. 2010). In addition to altering the structure of axonal microtubules which lead to abnormalities in fast axonal transport (GREEN et al. 1975; SAHENK et al. 1987; TANNER et al. 1998), other vincristine-induced alterations include changes the structure and shape of axonal microtubules (TANNER et al. 1998), shortening of microtubule lengths (SAHENK et al. 1987), and changes in microtubule distribution within the axon (GREEN et al. 1975). Recently, it was confirmed that axoplasmatic transport damage was neural-specific, and that rapid   17 17 proteasome-mediated tubulin degradation was limited to neuronal cells only, leaving neural and cancer cell types unchanged (HUFF et al. 2010). The addition of a vincristine dose (comparable to therapeutic levels) for 8 and 24 hours to neuronal cells significantly decreased signal transmission in neuronal cells by decreasing the tubulin levels to 20% and 9.8%, respectively (HUFF et al. 2010). This suggests that vincristine acts directly on neuronal cells to cause VIN, although further investigation is required to rule out other target sites (HUFF et al. 2010).  1.4 Factors Affecting the Exposure to Vincristine 1.4.1 Dose Intensity, Frequency, and Cumulative Dose The dose-limiting toxicities of vincristine are well known. Thus cancer treatment protocols generally recommend a maximum of 2mg/dose of vincristine, and have developed guidelines for interventions and dose modification in the event of severe ADR(s). Dose intensity and frequency of vincristine have been reliably shown to be strong determinants of VIN. As expected, doses of 2.5mg/m2/dose have been shown to be more toxic than 1.5mg/m2/dose (GIDDING et al. 1999; HOUGHTON et al. 1987). Recently, an interim analysis of a study by the Children’s Oncology Group (COG) showed that 34.6% of children over 13 years of age who were randomized to receive 2mg/m2/dose (maximum 2.5mg/dose) of vincristine had experienced grade three PN as compared 4.7% who received 1.5mg/m2/dose (maximum 2mg/dose) (RASSEKH 2010). Designed to determine if higher doses improved cure rates for intermediate-risk relapse patients with childhood B-precursor ALL, this interim finding provides further support for a potential relationship between dose intensity and VIN (RASSEKH 2010). The effect of dose frequency is most predominantly   18 18 observed in patients with Wilms tumor, as the signs and symptoms of VIN dramatically improve when the frequency of vincristine is reduced from weekly to every three weeks of vincristine dosing. Additionally, a study in 2006 showed a dramatic increase in toxicity with when vincristine was given more than one dose per week at doses higher than 2mg/m2/dose (HARRIS and BLANCHAERT 2006). This is supported with functional studies that show that the cytotoxicity of vincristine against tumor cells is more severe when dosed at a 7-day interval versus a 21-day interval (GIDDING et al. 1999; HOUGHTON et al. 1987). On the other hand, even though vincristine has been in use in clinical practice since the 1960s, it is still unclear whether cumulative dose is an important contributing factor to the development of VIN. One study suggests that the frequency or severity of VIN is not increased in patients who are subjected to over 20 weeks of weekly doses of 1.5mg/m2 (HOUGHTON et al. 1987). Similarly, a second study showed that the frequency and severity of motor dysfunction, as assessed with the Motor Assessment Battery for Children (m-ABC), was not significantly associated with the cumulative vincristine dose, even for doses over 40mg/m2 (HARTMAN et al. 2010). Together, these studies suggest that there are other differences in the patient population that are more important than the slow buildup of drug and/or metabolites in causing VIN. Conversely, a third publication suggests that cumulative dose may be indicative of the likelihood of loss of range of motion (ROM) and reduced ankle strength, neuropathies that were not assessed by the previous studies on motor dysfunction (APLENC et al. 2003). Thus, potentially, any effects of cumulative dose on VIN may be specific to certain types of neuropathy or age groups, since the m-ABC is only valid for individuals over four years old and is not completely comprehensive (e.g.: excludes ROM).   19 19 Therefore, comparisons of studies are difficult, and the variability in assessment and evaluation tools makes it difficult to compare the few reports in the literature.  1.4.2 CYP3A Inhibitors The CYP3A family of enzymes are involved in the metabolism of many drugs. Therefore, concomitantly administered medications that compete for CYP3A metabolism can reduce the hepatic clearance of vincristine, and thus increase a patient’s exposure to vincristine. Such drugs have thus been broadly classified as CYP3A inhibitors or potentiators of VIN (KIVISTO et al. 1995). Similarly, the CYP3A4 inhibitors that negatively affect the hepatic microsomal enzyme activity include antifungal azoles (fluconazole, itraconazole, voriconazole, ketoconazole) (VAN SCHIE et al. 2011;  JENG and FEUSNER 2001; KAMALUDDIN et al. 2001;  SATHIAPALAN and EL-SOLH 2001;  SATHIAPALAN et al. 2002), isoniazid (CHAN 1998), and erythromycin (CHAN 1998). In addition to being an inhibitor of CYP3A4, cyclosporin is also blocker of ABCB1, which further decreases the clearance of vincristine (CHAN 1998). Other drugs with unknown mechanisms of action but have been observed to cause increases in VIN, are nifedipine, acyclovir, and verapamil (CHAN 1998). Therefore, patients who are receiving vincristine and who are utilizing these competing drugs need to be monitored carefully for ADRs, as they may be at an increased risk for VIN (HARNICAR et al. 2009). Similarly, such factors potentially contributing to VIN need to be taken into account in pharmacogenomic studies aiming to discover genomic biomarkers for VIN, as they may act as confounders of genetic effects.     20 20 1.4.3 Genetic Variation in CYP3A4 and CYP3A5, and Ancestry The discovery of the relationship between of CYP3A inhibitors and VIN has lead to further studies on the effects of genetic variation in CYP3A enzymes on VIN. Unlike CYP3A4, which is constitutively expressed in most individuals (THOMPSON et al. 2004), CYP3A5 is a highly polymorphic gene that causes individuals to differ in its expression, affecting metabolism rates, and likely the exposure to vincristine (EGBELAKIN et al. 2011). Compared to CYP3A5 expressers, non-expressers have lower plasma concentrations of M1, the main metabolite of vincristine, and are more likely to report experiencing neuropathy (EGBELAKIN et al. 2011). In particular, the CYP3A5 gene contains three functional variants that are associated with rapid CYP3A5 mRNA degradation (BUSI and CRESTEIL 2005): CYP3A5*3, an intronic sequence variant that generates a cryptic splice site, which causes a pre-mature stop codon to be included in the mRNA, resulting a truncated protein and decreased protein functionality; CYP3A5*6, an alternative splice variant resulting from a single nucleotide transition within exon 7 (G-to-A); and CYP3A5*7, a variant that causes a reading frame shift in codons 345 and 346, introduces a premature termination codon at codon 348 (KUEHL et al. 2001). CYP3A5 expressers carry at least one CYP3A5*1 allele, which is the dominant wildtype allele that produces fully functional protein from full-length CYP3A5 mRNA (KUEHL et al. 2001). The protein content of CYP3A5 and the catalytic activity level has been found to be 18 times and 8 times higher in CYP3A5*1/*1 and CYP3A5*1/*3 genotype carriers, respectively, as compared to CYP3A5*3/*3 carriers (MACPHEE et al. 2005; THERVET et al. 2005; ZHAO et al. 2005; ZHENG et al. 2005). In agreement with this, CYP3A5 non-expressers have been shown to have lower plasma concentrations of M1, the main metabolite of vincristine, and have been reported to be more   21 21 likely to report experiencing neuropathy, as compared to CYP3A5 expressers (EGBELAKIN et al. 2011). Due to inter-ethnic differences in the allele frequencies of these variants, the expression of functional CYP3A5 varies by ethnicity. Only 10 to 20% of Caucasians and 75 to 80% of African Americans express a functional CYP3A5 enzyme (KUEHL et al. 2001). In agreement with this, Caucasians have been shown to be seven-times more likely to incur VIN, whereas African Americans suffered from higher relapse rates, suggesting a reduced efficacy of vincristine at standard doses in this population is due to the increased metabolism of the drug (GHAFOOR et al. 2002; JEMAL et al. 2002; POLLOCK et al. 2000). It is thought that CYP3A5 is the largest contributor to inter-individual and inter- ethnic differences in CYP3A-dependent drug clearance and response (KUEHL et al. 2001). This enzyme contributes to over 50% of CYP3A protein content in the body (THOMPSON et al. 2004), and is the most predominant known metabolizer of vincristine. Given the inter- ethnic differences in vincristine efficacy and toxicity and their correlation with the expresser genotype of CYP3A5, it is therefore possible that CYP3A5 is a main variable influencing the occurrence of VIN.  1.4.4 Vincristine Transporters Enzymes suggested to be involved in the transport of vincristine have also been implicated in the potential to contribute to VIN. For example, genetic variation in ABCB1 shows inter-ethnic allele frequency differences that also correlate with the incidence of VIN: Approximately 20% of Caucasians, but only 1% of Africans, are homozygous for haplotype TTT (genotype TTT/TTT) in ABCB1 causing reduced ABCB1 transporter activity (FUNG and   22 22 GOTTESMAN 2009). It could therefore also be theorized that a decrease in functionality or a lack of these transporters would cause the accumulation of vincristine in the body or in certain tissues or prevent neuroprotectants from reaching the damaged site, therefore, causing vincristine-associated toxicity.  1.4.5 Previous Pharmacogenomic Studies Only selected genes and genetic variants have been previously studied in the context of VIN. These include CYP3A4*1B and CYP3A5 variants (*3, *6, and *7), as well as three major variants in the ABCB1 transporter gene (ABCB1 rs1128503, rs2032582, rs1045642). The findings involving these gene variants and their impact on VIN have been the subject of some controversy in the literature. In a study of 107 patients, the CTCAE grading scale was used to compare the severity and frequency of VIN between CYP3A5 expressors and non- expressors (EGBELAKIN et al. 2011). This study showed the frequency of VIN (grades one to four) to be significantly higher in CYP3A5 non-expressers, and no significant difference for the more severe grades of VIN (grades three and four) (EGBELAKIN et al. 2011). In a second study with similar number of patients, the same authors have demonstrated that VIN occurs more frequently in Caucasians compared to African Americans (RENBARGER et al. 2008). While this study confirms the correlation between ethnicity and the incidence of VIN, the authors inferred that this association was related to inter-ethnic differences in CYP3A5 expression, given the preferential metabolism of vincristine by CYP3A5 over CYP3A4 and the differential minor allele frequencies (MAF) between Caucasians and Africans. Contrary to these findings, a recent study by a different group focused on individual single nucleotide polymorphisms (SNP) and haplotype analyses of the common allelic variants in CYP3A5   23 23 (HARTMAN et al. 2010). Using m-ABC to grade the severity of motor PN, this study of 34 patients observed that CYP3A5 *3/*3 and CYP3A5 *1/*3 haplotypes did not significantly affect motor performance (HARTMAN et al. 2010). The same study also did not observe an effect of ABCB1 polymorphisms rs1128503, rs2032582, rs1045642 on motor performance. A third group performed single SNP and haplotype analyses of ABCB1 rs2032582 (G2677A/T) and rs1045642 (C3435T) to determine the influence of these ABCB1 gene variants on vincristine-induced constipation and vincristine pharmacokinetics (PLASSCHAERT et al. 2004). This study concluded that only a specific ABCB1 haplotype (3435C/2677G) may be involved in increasing the elimination half-life of vincristine, although the significance was lost after correction for multiple testing was applied (PLASSCHAERT et al. 2004). The observed effect of this haplotype, however, did not extend to other pharmacokinetic parameters, which included clearance, area-under-the-concentration (AUC)-time curve, half- life, and volume of distribution at steady state. The same study also observed no connection between vincristine exposure (total body clearance, AUC, elimination time) and severity of constipation (CTCAE grades three to four as compared to one to two) in ALL patients (PLASSCHAERT et al. 2004). Given the inconsistency in pharmacokinetic effects and the lack of association with toxicity, the authors concluded that the effect of the ABCB1 haplotype on elimination half-life is likely limited in its relevance to VIN (PLASSCHAERT et al. 2004). A fourth group evaluated whether peripheral neuropathy was affected in patients with the CYP3A4*1B and CYP3A5*3, and CYP3A5*6 genotypes (APLENC et al. 2003). This study showed that the patients with CYP3A4*1B and CYP3A5*3, had a decreased risk of peripheral neuropathy. These findings were statistically significant prior to correction for multiple comparisons, and approached statistical significance (APLENC et al. 2003).   24 24 Given the limited number of studies conducted on the genetic basis of VIN, the select few genes studied, as well as the contradictory results between studies, more pharmacogenomic research in the context of VIN is needed. A fulsome investigation into the role of these proteins and their relative contributions in affecting the ADME of vincristine and its metabolites has not been conducted, nor has the role of different transporters and metabolizers of vincristine have been explicitly determined. Therefore, variation in genes other than those previously studied may be of importance for the development of VIN.  1.5 Patient Cohort Selection: Wilms Tumor and Rhabdomyosarcoma The aim of this study was to investigate the genetic basis of VIN in paediatric patients using a broad genotyping panel that comprehensively covered genetic variation in 315 genes involved in general drug ADME, in order to identify novel genes potentially associated with VIN, as well as to assess the association of previously studied candidate genes. As the first discovery project within a larger study, this study on the pharmacogenomics of VIN in paediatric Wilms tumor and rhabdomyosarcoma patients is a starting point for further studies on the genetic basis of VIN in paediatric cancer patients. Wilms tumor (nephroblastoma) is the fifth most common paediatric cancer and the most common renal cancer with up to 500 new cases that are diagnosed annually (eight patients per million children) in the US (BRESLOW et al. 1988). With a median age at diagnosis of 36.5 to 42.5 months, the majority of Wilms tumor patients are diagnosed under the age of 5 (BRESLOW et al. 1988), with an average diagnosis at 3.5 years. Considered a highly curable disease, the diagnosis is typically accompanied by symptoms of abdominal mass, pain, hypertension, fever, hematuria, and anemia. As reported by The National Wilms   25 25 Tumor Study (NWTS) Group, the five-year survival rate remains above 90%, starting in the 1980s when adjuvant radiotherapy, chemotherapy, and collaboration between interdisciplinary health professionals became standard treatment (SMITH et al. 2010). In general, Wilms tumor is treated with combination chemotherapy of vincristine, actinomycin- D, and doxorubicin, as recommended by NWTS-5. This favorable outcome has been maintained over the years, even with follow-up clinical trials that attempted to balance toxicity with efficacy with experimental reductions in the length of therapy, radiation doses, and extent of fields irradiated (GREEN et al. 1998). Rhabdomyosarcoma is the most common soft tissue sarcoma, occurring in various part of the body. The annual incidence is 4.3 patients per million in children (ARNDT et al. 1997). Rhabdomyosarcoma occurs mainly in children two to six years (OGNJANOVIC et al. 2009) and 14 to 18 years of age (WEXLER LH 2002). Patients with localized rhabdomyosarcoma have greater than 70% of a five-year recurrence-free survival rate, although this prognosis is dependent on age and disease characteristics (histology, primary anatomic site, disease extent, therapy) (MEZA et al. 2006). Approximately 42% of the paediatric rhabdomyosarcoma population develops some form of VIN as graded by the CTCAE scale (ARNDT et al. 1997). In order to standardize the comparisons between patients and to facilitate the detection of genetic effects, this study focused on patients with Wilms tumor and rhabdomyosarcoma, since these two tumor types are fairly homeogenous for the factors that could potentially affect the outcome of VIN. Firstly, standardizing the protocols by dose intensity and frequency of vincristine is critical, given their strong relationship with VIN. Protocols within each tumor type varied very little; patients with rhabdomyosarcoma and   26 26 Wilms tumor receive the same dose intensity of vincristine at 1.5mg/m2 for children aged over three years, 0.05mg/kg/dose for children aged one to three years, and 0.025mg/kg/dose for children under one year. Vincristine is received every seven days for ten weeks, a key period of susceptibility to VIN, and transitioned to every seven to 21 days for the next 4 to 9.5 months. Although Wilms tumor patients receive 15 doses over 25 weeks and rhabdomyosarcoma patients typically receive 30 to 36 doses over a time period of 36 to 48 weeks, the reported time to VIN occurs after 12 doses (ARNDT et al. 1997), which is close to the end of the weekly scheduled doses and is when the dosing schedules are similar between both tumor types. Therefore, this demonstrates that the difference in duration of therapy and cumulative dose likely does not affect VIN, at least in patients with Wilms tumor and rhabdomyosarcoma. Secondly, the concomitant medications that may be associated with neurotoxicity or neuropathy should be minimized. It is also essential to choose a group whose protocol required minimal exposure to corticosteroids and neuropathy-causing antineoplastic agents. To illustrate, in addition to steroids, a variety of chemotherapy agents are associated with neuropathy: platinum compounds (e.g.: cisplatin, carboplatin, oxaliplatin), taxanes (e.g.: docetaxel, paclitaxel), epothilones (e.g.: epothilone-B), bortezomib (CAROZZI et al. 2010; DE GEEST et al. 2010). Wilms tumor is treated with combination chemotherapy of vincristine, actinomycin-D, doxorubicin, and in some protocols, cyclophosphamide as well. Although the concomitant drugs vary depending on the chemotherapy protocol used in rhabdomyosarcoma patients, the protocol is fairly consistent: vincristine, actinomycin-D, cyclophosphamide (VAC), vincristine, actinomycin-D, ifosfamide (VAI), or vincristine, ifosfamide, etoposide (VIE) (CRIST et al. 2001). All of these   27 27 concomitant chemotherapy agents are not associated with neurotoxicity. With the exception of rare types of CNS tumors, patients with solid tumors do not receive steroids as a part of protocol. Therefore, Wilms tumor and rhabdomyosarcoma patients would not receive steroids unless enhancement of the effects of their antiemetic drugs is needed. Thirdly, the location of the tumor may affect the observed neurotoxicity (e.g.: patients with CNS tumors). This effect was minimized in the selection of patients with Wilms tumor and rhabdomyosarcoma. Lastly, the time to VIN should be fairly comparable to minimize the differences between cancer types; this has been reported to be similar between the two tumor types. With these criteria in mind, patients with Wilms tumor and rhabdomyosarcoma have been chosen as the main focus of this discovery research.   28 28 Chapter  2: Hypothesis, Project Goal and Objectives  2.1  Hypothesis This study hypothesizes that there are specific genetic variants that are responsible for augmented susceptibility to VIN in patients with Wilms tumor and rhabdomyosarcoma.  2.2  Project Goal and Objectives The primary goal of this project is to identify the SNPs that are associated with VIN in patients with Wilms tumor or rhabdomyosarcoma. The specific objectives are to:  [1]  Accurately grade the severity and types of VIN experienced by Wilms tumor or rhabdomyosarcoma patients using a validated clinical grading system for adverse events.  [2]  Genotype DNA samples from patients with VIN and controls to identify genetic variants that may be associated with VIN in both vincristine-specific and general drug biotransformation and transportation pathways.     29 29 Chapter  3: Methodology  3.1 Recruitment, Consent, and Establishing Causality The Canadian Pharmacogenomics Network for Drug Safety (CPNDS) is an active and targeted ADR surveillance network that serves to identify and recruit patients who been affected by ADRs. Additionally, patients who receive pharmacotherapy treatment but who do not experience ADRs are recruited as study controls. CPNDS is a multidisciplinary consortium comprised of researchers and clinicians from 13 large paediatric tertiary care centres, institutes and academic centres across Canada. It is also expanding with a growing number of adult centres. At each of the surveillance sites, clinician surveillors identify ADR cases, enroll patients, report relevant clinical data, and establish causality for each ADR. Surveillance clinicians are highly trained medical professionals that include physicians, clinical pharmacists, and nurses. Retrospective information was obtained from patient charts, and where possible, supplementary details (such as frequency and severity of events) were acquired through discussions with the involved health care professionals and/or from interviews with parents or family members. The information collected included: patient demographics (date of birth, gender, ancestry, body surface area (BSA), height, weight, tumor type, treatment protocol), information on vincristine and concomitant drugs (indication; treatment duration and frequency; dose; number of doses taken, reduced, and omitted; cumulative dosage; drug route given), drug reaction (type, location, date, and age of reaction; interventions required and outcomes; recovery with or without repeated exposure of suspected drug; recovery status; reappearance or abated reaction; likelihood the ADR is due to vincristine (NARANJO et al.   30 30 1981)), and treatment and test information associated with interventions, and the patient’s medical conditions (personal and family). The Naranjo algorithm evaluated factors such as: the temporal relationship between drug administration and the event date(s) of each ADR; improvements in the ADR(s) following the discontinuance of vincristine; recovery from the ADR(s) following dose modifications; reoccurrence of the ADR(s) after restarting vincristine; and a lack of plausible alternative causes (other than VIN) for each observed neurotoxicity event. Based on the Naranjo scale, only patients whose adverse event(s) were identified as “probable,” “likely,” or “definitely” due to vincristine were preliminarily classified as a case with VIN (NARANJO et al. 1981). Patients who did not experience any VIN were classified as controls. Appropriate institutional review board approval was obtained from each of the surveillance sites according to the requirements and regulations set out by each institution. Informed consent was obtained from each patient’s parent or guardian, and depending on the age of the patient, assent was attained when appropriate. This consent provided the permission to obtain biomaterial from the child for genotyping purposes, access to data in their medical charts, and to publish research results. To protect patient privacy, information was collected without patient-identifying information (with the exception of birth date). All clinical research conducted by CPNDS has been publicly registered with the National Clinical Trial Registry under the identification number NCT00414115. This study was approved by the University of British Columbia and the Children’s & Women’s Health Centre of British Columbia Research Ethics Board.     31 31 3.2 Defining Cases and Controls Once preliminarily identified as a VIN case or control by the site surveillor, each patient was independently reviewed to verify his or her case/control status. This secondary assessment process focused on carefully reviewing the parameters and circumstances that could affect each individual’s eligibility as a true case or the interpretation of VIN. These potentially confounding factors included: patient’s medical history (e.g.: surgery, tumor location), pharmacotherapy history (e.g.: VIN-potentiating drugs, steroids), and relevant neuropath(ies) that existed prior to the administration of vincristine (e.g.: constipation, hereditary neuropathic disorders, injury- or disease-induced neuropathies). In evaluating these patients, age-related issues in the presentation of the ADR were also taken into consideration. For example, the mobility of a one-year-old patient would differ from the mobility of a five-year-old. The location of the surgical site and tumor were both considered in relation to the neuropathy observed. Patients who had experienced VIN and had surgery and tumor locations that were independent of the neuropathy (i.e.: a facial rhabdomyosarcoma and constipation) were classified as cases. However, in patients with brain or solid tumors that were located close to nerves (e.g.: the lumbosacral plexus), they would require the observation of an ADR that was separate from foot drop in order to be classified as a case. Patients were assessed for drug-drug-interactions between vincristine and concomitant CYP3A4 inhibitors that could alter the clinical activity of vincristine and potentiate VIN by reducing the hepatic clearance of vincristine. Drugs that were given concomitantly with vincristine and administered prior to the occurrence of the ADR were considered as potential interacting medications. Each patient’s medication record was screened for drugs implicated   32 32 with potentiating VIN, including: antifungal azoles (e.g.: fluconazole, itraconazole, ketoconazole), cardiovascular agents (e.g.: nifedipine, verapamil), macrolide antibiotics (e.g.: erythromycin), antivirals (e.g.: acyclovir), and antidepressants (e.g.: fluvoxamine, nefazodone). To avoid incorrectly grading steroid-induced myopathy as VIN or muscle weakness, patients who were given steroids were assessed on other presenting symptoms that were separate from their muscle symptomology. For example, adverse events such as jaw pain and/or constipation were gradable symptoms that were caused independently from steroids. Both steroid and CYP3A4 use were recorded. Patients were also screened for hereditary neuropathy disorders (such as Charcot Marie Tooth disease (CMT), Friedreich’s ataxia, and Guillian-Barré syndrome), pre-disposing neuropathic diseases (AIDS, diabetes), or physical injuries that might overlap with VIN or make it appear more severe. In cases where grading was complicated by conflicting data, a committee of experts was used (paediatric oncologist and a paediatric clinical pharmacologist) to differentiate whether the cause of neuropathy was attributable to vincristine or other factors. If this was indeterminable and no other VIN existed independently from these potential confounders, then the patient was removed from study. Individuals were only defined as cases if they did not have hereditary-, injury-, or tumor-induced neuropathy, or had one or more forms of VIN that could be graded independently from the aforementioned confounding factors. Controls were defined as patients without hereditary-, injury-, or tumor-induced neuropathy who did not experience VIN during their chemotherapy treatment. Based on this assessment, some patients were reclassified from the original Naranjo classification scale, or were excluded from the study.    33 33  3.3 Clinical Characterization of Vincristine-induced Neurotoxicity Cases This clinical assessment of ADR grading and type of neuropathy was made independently from the patient’s genotype. The case and control classification and grading of each patient was initially made by the clinical surveillor and later confirmed by an expert committee to ensure consistency and accuracy of the phenotyping process.  3.3.1 Classifying Each Vincristine-Induced Neurotoxicity Event by Type of Neuropathy Once the patient’s case or control status was established, each VIN case was categorized by type of neuropathy. In patients with poly-neuropathies, each individual neuropathy was coded. PN, sensory and/or motor included: foot drop, gait disturbance, ataxia, dysphagia, decreased joint ROM, muscle weakness, ptosis, vocal cord paralysis, hoarseness, dysarthia, aphonia, pain (jaw, limbs, and joint pain), dysesthesia, and paresthesia. Ileus and constipation were categorized as AN events. Incidences of SIADH (as an endocrine disorder) and headaches were classified as CNS effects.  3.3.2 Evaluating the Severity of Each Vincristine-Induced Neurotoxicity Event The NCI-CTCAE (CANCER THERAPY EVALUATION PROGRAM - NATIONAL CANCER INSTITUTE 2010) is a standardized and validated tool that is commonly utilized by oncologists to evaluate chemotherapy-induced neurotoxicity in a variety of oncology drugs, including VIN (CAVALETTI et al. 2006; CAVALETTI et al. 2007). The severity of each VIN event was graded according to the criteria in CTCAE version 4.03 (Table 1). With this tool,  34 Table 1   Grading Vincristine-Induced Neurotoxicity with the CTCAE Criteria (v4.03) Specific Neuropathic Event Grade: 0 Grade: 1 Grade: 2 Grade: 3 Grade: 4 Grade: 5  Peripheral Neuropathy (Motor) Peripheral motor neuropathy A disorder characterized by inflammation or degeneration of the peripheral motor nerves.  Asymptomatic   *Clinical or diagnostic observations only; intervention not indicated  Moderate symptoms; limiting IADL   Severe symptoms; limiting self care ADL; assistive device indicated  Life-threatening consequences; urgent intervention indicated  Death  Gait disturbance A disorder characterized by walking difficulties.   Asymptomatic   Mild change in gait (e.g.: wide- based, limping or hobbling)  Moderate change in gait (e.g.: wide- based, limping or hobbling); assistive device indicated; limiting IADL  Disabling; limiting self care ADL   -   -  Ataxia A disorder characterized by lack of coordination of muscle movements resulting in the impairment or inability to perform voluntary activities.  Asymptomatic   *Clinical or diagnostic observations only; intervention not indicated  Moderate symptoms; limiting IADL   Severe symptoms; limiting self care ADL; mechanical assistance indicated  -   -  Dysphagia A disorder characterized by difficulty in swallowing.   Asymptomatic   Symptomatic, able to eat regular diet  Symptomatic and altered eating/swallowing  Severely altered eating/ swallowing; tube feeding or TPN or hospitalization indicated  Life-threatening consequences; urgent intervention indicated  Death  Joint ROM decreased A disorder characterized by a decrease in joint flexibility of any joint.  Asymptomatic  <=25% loss of ROM; or decreased ROM limiting athletic activity  >25 - 50% decrease in ROM; or limiting IADL  >50% decrease in ROM; or limiting self care ADL; disabling  -   - Muscle weakness A disorder characterized by a reduction in the strength of the muscles  Asymptomatic   Symptomatic; perceived by patient but not evident on physical exam  Symptomatic; evident on physical exam; limiting IADL  Limiting self care ADL; disabling   -   -  Facial muscle weakness A disorder characterized by a reduction in the strength of the facial muscles.  Asymptomatic   *Clinical or diagnostic observations only; intervention not indicated  Moderate symptoms; limiting IADL  Severe symptoms; limiting self care ADL  -  - Hoarseness A disorder characterized by harsh and raspy voice arising from or spreading to the larynx.  Asymptomatic     Mild or intermittent voice change; fully understandable; self-resolves  Moderate or persistent voice changes; may require occasional repetition but understandable on telephone; medical evaluation indicated  Severe voice changes including predominantly whispered speech   -  -  Dysarthria A disorder characterized by slow and slurred speech resulting from an inability to coordinate the muscles used in speech.  Asymptomatic   Mild slurred speech   Moderate impairment of articulation or slurred speech  Severe impairment of articulation or slurred speech  -   -  Aphonia A disorder characterized by the inability to speak. It may result from injuries to the vocal cords or may be functional (psychogenic).  Asymptomatic   -  -  Voicelessness; unable to speak  -   -    35 Specific Neuropathic Event Grade: 0 Grade: 1 Grade: 2 Grade: 3 Grade: 4 Grade: 5 Peripheral Neuropathy (Sensory) Peripheral sensory neuropathy A disorder characterized by inflammation or degeneration of the peripheral sensory nerves  Asymptomatic   *Loss of deep tendon reflexes or paresthesia  Moderate symptoms; limiting IADL   Severe symptoms; limiting self care ADL  Life-threatening consequences; urgent intervention indicated  Death  Pain  Asymptomatic Mild pain Moderate pain; limiting IADL Severe pain; limiting self care ADL - - Arthralgia A disorder characterized by a sensation of marked discomfort in a joint.  Asymptomatic  Mild pain  Moderate pain; limiting IADL  Severe pain; limiting self care ADL  -  - Myalgia A disorder characterized by marked discomfort sensation originating from a muscle or group of muscles.  Asymptomatic  Mild pain  Moderate pain; limiting IADL  Severe pain; limiting self care ADL  -  - Neuralgia A disorder characterized by intense painful sensation along a nerve or group of nerves.  Asymptomatic   Mild pain   Moderate pain; limiting IADL   Severe pain; limiting self care ADL   -   - Dysesthesia A disorder characterized by distortion of sensory perception, resulting in an abnormal and unpleasant sensation.  Asymptomatic   Mild sensory alteration   Moderate sensory alteration; limiting IADL   Severe sensory alteration; limiting self care ADL  -   -  Paresthesia A disorder characterized by functional disturbances of sensory neurons resulting in abnormal cutaneous sensations of tingling, numbness, pressure, cold, and warmth that are experienced in the absence of a stimulus.  Asymptomatic  Mild symptoms   Moderate symptoms; limiting IADL  Severe symptoms; limiting self care ADL   -   -    36 Specific Neuropathic Event Grade: 0 Grade: 1 Grade: 2 Grade: 3 Grade: 4 Grade: 5 Autonomic Neuropathy Ileus A disorder characterized by failure of the ileum to transport intestinal contents.  Asymptomatic  -   Symptomatic; altered GI function; bowel rest indicated  Severely altered GI function; TPN indicated  Life-threatening consequences; urgent intervention indicated  Death Constipation A disorder characterized by irregular and infrequent or difficult evacuation of the bowels.  Asymptomatic  Occasional or intermittent symptoms; occasional use of stool softeners, laxatives, dietary modification, or enema  Persistent symptoms with regular use of laxatives or enemas; limiting IADL    Obstipation with manual evacuation indicated; limiting self care ADL    Life-threatening consequences; urgent intervention indicated   Death Central Nervous System Effects Headache A disorder characterized by a sensation of marked discomfort in various parts of the head, not confined to the area of distribution of any nerve.  Asymptomatic  Mild pain   Moderate pain; limiting IADL  Severe pain; limiting self care ADL  -  - Endocrine disorders - Other, specify (e.g.: SIADH)  Asymptomatic  Mild symptoms; clinical or diagnostic observations only; intervention not indicated  Moderate; minimal, local or noninvasive intervention indicated; limiting age appropriate IADL  Severe or medically significant but not immediately life threatening; hospitalization or prolongation of existing hospitalization indicated; disabling; limiting self care ADL  Life-threatening consequences; urgent intervention indicated  Death *”Asymptomatic” was removed from grade one   37 patients were graded with the broad definitions of PN (sensory, motor), AN (constipation, ileus), and CNS effect (endocrine disorders), that were supplemented with specific definitions of neuropathic events (muscle weakness, recurrent laryngeal nerve palsy, dysarthia, and pain) (Table 1). Patients who did not experience neuropathy were termed as asymptomatic patients were considered to be the controls (grade zero) in this study. Since the term “asymptomatic” did not commensurate with the other criteria of mild neuropathy within the grade one category, for clarification purposes, the term “asymptomatic” was moved from the grade one to the grade zero category for clarification purposes. In the CTCAE grading tool, cases were graded by limitations on activities of daily living (ADL): mild (grade one), moderate ADL and instrumental activities of daily living (IADL) (grade two), severe and self-limiting ADL (grade three), life-threatening (grade four), and causing death (grade five). IADLs and self-care ADLs were distinguished by whether the ADR affects the person's ability to be independent. In the context of VIN, affected IADLs included difficulties in writing (muscle weakness), playing sports (gait disturbance), climbing stairs (gait disturbance), or running (mild paresis). Affected self-care ADLs included the inability to: ambulate (muscle weakness, gait disturbance), feed oneself or sit up (muscle weakness, pain), or eat (pain). In addition, vincristine-specific interventions were utilized to supplement the grading of neuropathy in the context of vincristine (Table 2). This definition was developed and based on vincristine-specific treatment modification guidelines that are typically provided in standard treatment protocols (i.e.: “stop vincristine if patient experiences grade three to four toxicity”) and allowed for a grade to be assigned in the context of typical vincristine-specific  38 Table 2   Grading Vincristine-Induced Neurotoxicity by Intervention   VIN Cases   Description of Neurotoxicity VIN-specific Interventions Grade 0 No neurotoxicity Asymptomatic None Grade 1 Mild neurotoxicity Observations and monitoring. Interventions not required, except for intermittent laxative use if necessary Grade 2 Moderate neurotoxicity requiring non-invasive interventions  Interventions required: Rehabilitation therapy (PT, OT, neurologist, physiatrist), assistive devices (i.e.: splints/braces, ankle foot orthosis, wheelchair), neurology consult (electromyography, NCS), pharmacotherapy to treat neurotoxicity symptoms (i.e.: narcotic analgesics, non-narcotic analgesics, regular laxative use), and hospitalization and surgery (i.e.: Achilles tendon lengthening, nasogastric tube). Grade 3 Severe neurotoxicity requiring invasive interventions (in addition to previous interventions) Vincristine dose modifications: hold or reduction (in addition to previous interventions) Grade 4 Life-threatening neurotoxicity requiring urgent intervention Discontinue vincristine Grade 5 Death Discontinue vincristine   39 interventions in therapy. For example, the grade two interventions were categorized into the following subgroups: rehabilitation therapy (PT, OT, neurologist, physiatrist), assistive devices (i.e.: splints, ankle foot orthosis, wheelchair), neurology consult (electromyography, NCS), pharmacotherapy to treat neurotoxicity symptoms (i.e.: narcotic analgesics, non- narcotic analgesics, laxatives), and hospitalization and surgery (i.e.: Achilles tendon lengthening, nasogastric tube) (Table 2). VIN events were given a grade three if they necessitated a reduction and/or hold in subsequent vincristine doses, where dose reductions due to VIN were considered to be any amount less than 100% of the original dose. Only interventions that were provided specifically in response to vincristine-related neurotoxicity events were graded. For example, analgesics provided to Wilms tumor patients for abdominal pain were not recorded as a vincristine-related intervention; however, it would be coded as an intervention if the same patient was prescribed analgesics that were specific for vincristine- induced-jaw, -limb, and/or -joint pain. Prophylactic interventions (i.e.: laxatives) and non- VCR related interventions (i.e.: analgesics for surgery, reduction of vincristine doses due to hepatic toxicity or decreased hepatic clearance) were not coded as VIN-related interventions. Hepatic toxicity was defined as hyperbilirubinemia, hepatic necrosis, increased transaminase levels, and hepatic veno-occlusive disease. Both Tables 1 and 2 were used to assign an overall severity grade for each VIN event. Complex cases were reviewed by the expert committee who collectively assigned a severity rating. The date to the first ADR of CTCAE grade two or three was recorded and defined as the date to the “first major” VIN event. The event with the highest VIN grade was assigned as the patient’s most severe grade of VIN and type of neuropathy. Patients who had equal grades in multiple neuropathy categories were classified as having equally severe  40 neuropathic symptoms from all applicable neuropathy categories. For example, a patient with a grade three PN event and a grade three AN event would be classified as having grade three peripheral and autonomic neuropathy.  3.4 Study Inclusion and Exclusion Criteria Between February 2004 and April 2011, children who were less than 18 years of age at the start of vincristine therapy were recruited and enrolled from paediatric tertiary referral centres across Canada. Patients were recruited using active surveillance methods developed by the Canadian Pharmacogenomic Network for Drug Safety (CARLETON et al. 2009). The recruitment included patients with a variety of cancer types including: haematological malignancies (leukemias, lymphomas) and solid tumors (rhabodomyosarcoma, Wilms tumor, neuroblastoma, PNET, Ewings sarcoma, other sarcoma, germline cell tumors, hepatoblastoma, and hepatocellular carcinoma). Patients with Wilms tumor and rhabdomyosarcoma were chosen as the initial cohort for genotyping. This group of patients had tumors in locations that were unlikely to influence the ADRs of interest, and were also on similar chemotherapy regimens, which included: dosing frequencies, no corticosteroids as a component of their chemotherapy protocol, concomitant medications that are not associated with causing neuropathy, and standardized dosing for children under one-year-old. Furthermore, the length of their chemotherapy treatment allowed the maximal inclusion of patients who had finished their chemotherapy treatment, which allowed for all the VIN and the assignment of the most appropriate grade for each patient (according to their most severe event) to be captured.  41 Only patients with CTCAE grades zero and two to five VIN were included for the purposes of studying individuals with definitive and extreme phenotypes of VIN. Patients were excluded if they had confounding conditions (e.g.: known confounders include surgery that may affect motor function) that would affect the assessment of their level and/or type of neuropathy, or whose biological samples failed genotyping, or had a CTCAE grade one of VIN as their most severe grade. Patients with a CTCAE grade one of VIN were excluded to compare of the most extreme phenotypes of VIN (grade two and higher) with the controls (grade zero). Therefore, this study included patients: who had received vincristine as a part of their primary chemotherapy protocol (both modified or standardized oncology treatment protocols), with a confirmed diagnosis of Wilms tumor or rhabdomyosarcoma, who had finished their chemotherapy treatment, and met the definition of case and control.  3.5 Calculating Dose and Time to First Adverse Drug Reaction Depending on age and protocol, patients received vincristine doses in the units of mg/kg or mg/m2. Since 30kg and 1m2 are considered to be equivalent measurements (FERRARI et al. 2003), weight-based doses (mg/kg) were converted to mg/m2 by multiplying by 30. To minimize miscalculations in rapidly growing children, this direct conversion was only applicable for patients over one-year-old. For patients who were less than 12 months of age, dose conversions using age, and the weight of the child were calculated every three months to improve the accuracy of the conversion. No conversions were necessary for patients who were given BSA-based dosages.  42 The time to the first ADR was measured from day one of vincristine therapy to the date of the first ADR. The loss of DTR and paresthesia was also excluded, as they would be graded as one. Furthermore, these specific VIN events occur very frequently; therefore, as common ADRs during vincristine therapy, they are occasionally recorded in the charts and not necessarily reported on a reliable basis.  3.6 Association Studies All work involved in genotyping preparations, data cleaning and data analysis were conducted independently from the identity of the SNPs and the case/control status of samples.  3.6.1 General Drug Biotransformation Genotyping A custom-made Illumina GoldenGate® (Illumina, San Diego, CA, USA) SNP genotyping assay panel was designed to capture the genetic variation of key genes affecting general drug disposition, including genes involved in ADME. Spanning all 23 chromosomes, the customized ADME-Tox panel consisted of 4,536 single nucleotide polymorphisms and 315 genes. To maximally cover the genes of interest, functional and tagging SNPs in the exons, introns, and intergenic regions were utilized. The genes represented on the panel include: ion channels (e.g.: SCN5A, KCNH2, KCNQ1), phase I drug-metabolizing enzymes (e.g.: CYP1A1, CYP2B6, ALDH2), phase II drug-metabolizing enzymes (e.g.: UGT2B7, GSTM1, NAT1, COMT), drug transporters (e.g.: ABCB1, ABCC1, ABCC2), transcription factors (e.g.: HNF4A, STAT3, NR1I2), drug receptors and drug targets (e.g.: VDR, PPARG, CETP), and others (e.g.: EPHX1, FMO1, PTGS1).  43  3.6.2 Candidate Gene Analysis A candidate gene approach was taken to address the hypothesis-driving approach that postulated the involvement of genetic variants in specific genes in causing VIN. Eight candidate genes specifically associated with metabolism and transportation of vincristine were targeted, including SNPs that are reportedly associated with VIN (candidate SNPs). These included: CYP3A4*1B rs2740574 ( Felix et al. 1998; KUEHL et al. 2001); CYP3A5 rs10264272 and CYP3A5*3 rs776746 (APLENC et al. 2003; D. M. TE LOO 2009); ABCB1 rs1128503, rs2032582, and rs1045642 (D. M. TE LOO 2009; PLASSCHAERT et al. 2004;  SONG et al. 2006); ABCC1; ABCC2; ABCC3; ABCC10; and RALBP1. On the custom-made panel, the eight candidate genes were covered by 210 functional and tagging SNPs.  3.7 Genotyping: Illumina GoldenGate Assay Depending on the age of the patient, biological samples were collected in the form of saliva or blood (Oragene®, DNA Genotek, Ottawa, Canada) for children, and sponges (DNA Genotek, Ottawa, Canada) for infants. Buccal swab (Catch-AllTM Sample Collection Swabs, Epicentre®,Wisconsin, USA) samples were collected from patients in all age groups. DNA was extracted using a Qiagen DNA purification kit (QIAamp®, Qiagen Inc., CA, USA) and Qiagen DNA purifier platform (Qiasymphony SP®, Qiagen Inc., CA, USA). DNA quantification was performed using a PicoGreen fluorometric assay (Quant-iT™ PicoGreen® dsDNA reagent) with a Biomek® NXP Laboratory Automation Workstation (Beckman Coulter). Extracted DNA samples were stored at -20oC. A minimum of 20ng/µl of genomic DNA was required for genotyping. Samples that did not meet this concentration were whole  44 genome amplified (WGA) with multiple displacement amplification (Qiagen® Inc.: CA, USA). DNA and WGA samples were normalized to a concentration of 20-50ng/µl and 80- 100ng/µl, respectively. After polymerase chain reaction (PCR) amplification of DNA, several products were randomly selected and run on a PCR gel for quality control purposes. Once quality control was assured, all samples were genotyped on the Illumina Golden Gate iScan Platform with the 32-sample Universal BeadChip (Illumina, San Diego, CA, USA, a high-throughput genotyping platform. The first plate set included negative (1x Tris-EDTA buffer) and positive (one sample with its genotype known) controls, while subsequent plates had three duplicate samples from the plate prior added to control for plate mishandling. Other quality control measures included having duplicate SNPs on the panel and testing for reproducibility errors.  3.8 Quality Control and Assurance in Genotypic Data 3.8.1 Quality of Patient Samples and SNPs The quality assurance of each sample and SNP clustering were preliminarily evaluated with the GenomeStudio Data Analysis Software version 2010 (Illumina, San Diego, CA, USA) (Figures 9 to 11). The missing call rate per sample is an informative indicator of sample quality. Samples with a call rate <0.95 were re-genotyped, and DNA samples with threeor more consecutive genotyping failures were excluded from the study due to low quality samples. All SNPs were manually re-clustered into three distinct clusters. SNPs were removed if there were ambiguous clustering between the AA, AB, and BB, had low intensities and low clustering scores of <0.25, and/or had high replicate errors.   45 3.8.2 Data Cleaning The data was imported into a matrix format and evaluated using SAS® version 9.2 (SAS Institute, Inc, Cary, NC). As internal controls, duplicate SNPs were cleaned and analyzed alongside the other SNPs for concordance studies. The following cleaning was performed in a specific order to maximize the data quality. Firstly, for bi-allelic SNPs, SNPs with monomorphic genotypes such as only heterozygote genotypes called or with no heterozygote genotypes called (deviation from Hardy-Weinberg equilibrium, HWE) were removed. Secondly, samples with less than 90% genotyping call rates were excluded. The genome-wide genotyping success rate for each SNP was defined as: the number of SNPs genotyped for a given patient x100 total number of SNPs  Thirdly, to remove low-quality SNPs, the SNPs that were successfully genotyped for under 98% of the patient population were removed. The genotyping success rate was defined as: the number of patients genotyped for a given SNPx100 total number of patients  The fourth step ensured the consistency of genome-wide genotyping success rate for each SNP and removed samples with <95% successful call rates. The fifth step identified unexpected relatedness and ensured that that all observations were independent with identity- by-descent (IBD) analysis tests. This was conducted with identity-by-state (IBS) estimations between all pairwise combinations of samples as an average of IBS between each pair of individuals. Samples were removed if the pairwise IBD test showed 86% to 97% IBS identity (suggesting close familial relationship) or ≥98% IBS identity (suggesting duplicate samples). The sixth step was to evaluate each SNP for HWE with an exact test (GUO and THOMPSON  46 1992) to determine external influences that may affect the inheritance of SNPs through generations. The simpleM method was employed to correct for multiple-testing and adjusted the HWE tests by taking the number of independent SNP grouping (on the full set of SNPs) into consideration. The number of independent tests is calculated by utilizing the composite linkage disequilibrium (LD) to create the correlation matrix of SNPs and MeffG  (GAO et al. 2008). Each SNP in the control patients were evaluated for HWE and excluded based on a p- value adjusted for multiple testing with the simpleM method (p-value = 0.000368). All intensity cluster plots of SNPs were visually inspected for: an association for overdispersion of the clusters, biased no calling, or erroneous genotype assignment. SNPs with any of these aforementioned characteristics were discarded.  3.9 Statistical Analyses: Logistic Regression Models and Manhattan Plot Statistical analyses were conducted with SAS® 9.2 and SAS/Genetics. Patient demographics were evaluated with the non-parametric Kruskal-Wallis Test and the Fisher Exact Test for continuous and categorical variables, respectively. Principal component analyses (PCA) were performed to assess the population structure. The 140 patients in this study were compared against the principal components 1 (prin1) and 2 (prin2) in 405 individuals of representative ancestries within the Hapmap project against 2,327 matching SNPs. These representative ancestries included 116 individuals with Northern and Western European ancestry (CEU group from Utah, United States); 86 individuals with Han Chinese ancestry (CHB group from Beijing, China); 85 individuals with Chinese ancestry (CHD group from Denver, Colorado); and 118 individuals with Yoruba ancestry (YRI group from Ibadan, Nigeria). The three major ethnic groups were roughly defined by separating the  47 scatterplot into four quadrants: European (principal component 1 > 0, principal component 2 > 0); Asian (principal component 1 < 0, principal component 2 > 0); and African (principal component 1 > 0, principal component 2 < 0) (Figure 2). The logistic regression model (p1-p = β0 + β1x + β2prin1 + β3prin2) analysis was performed separately for each tumor type and for the combined cohort, where: p = probability of each patient as a case, x = additive effect (defined as: -1 if the patient had two frequent alleles; 0 if the patient had one frequent allele and one rare allele; and one if the patient had two rare alleles), β0 = mean effect and the p-value of the genetic variant in association with VIN with adjustment for the covariates present in the model, β1 = coefficient of x, β2 = coefficient of principal component 1, and β3 = coefficient of principal component 2. To adjust for population stratification, principal components 1 and 2 were added to the additive genetic model to determine the association between the genetic variants and VIN. Genetic associations were considered for significance as part of a multivariable model where the SNPs with a global significance of the model was <0.05 within the combined cohort and <0.20 within each of the tumor types.  A 2nd additive genetic model (p1-p = β0 + β1x + β2prin1 + β3prin2 + β4age + β5tumor type) analysis was performed to evaluate the association of genetic determinants with VIN and adjusted for population stratification, age, and tumor type, where: β5 = coefficient of age, and β6 = coefficient of tumor type. A Manhattan plot of the logistic regression model adjusted for population stratification in the combined cohort was created in Excel (Windows 2009), with the SNPs organized by position and chromosome (Figure 3).  To determine the need for PCA adjustment and whether the SNPs could be replicated in the Caucasian population, a subgroup analysis was performed, and generated a 3rd and 4th  48 additive genetic model analysis. These twoanalyses evaluated the association of genetic determinants with VIN in a Caucasian-only population (p1-p = β0 + β1x), and a Caucasian- only population adjusted for age and tumor type (p1-p = β0 + β1x + β2age + β3tumor type), respectively, where: β2 = coefficient of age, and β3 = coefficient of tumor type. Both a multidimensional scaling analysis using the 140 patients and Hapmap populations and self- reported ethnicity of 87 Caucasians (Canada, Europe, Canada/Europe) were separately used to select the Caucasians.  3.10 Prediction Model A prediction model was created using a Kaplan-Meier curve that compared the percentage of individuals with VIN over the days post initiation of vincristine in patients was compared between individuals with a low, intermediate, or high risk for developing VIN (Figure 3). The total number of risk alleles was calculated with patient genotypes, defined by: 0 for a homozygous individual with no risk alleles, 1 for a heterozygous individual with one risk allele, and 2 for a homozygous individual with two risk alleles. With 12 risk alleles as the maximum number of total risk alleles, this was performed for each of the following SNPs: PON1rs854549 (risk allele: A), ABCA4 rs3789433 (risk allele: A) and rs549848 (risk allele: A), SLCO1C1 rs10770704 (risk allele: G), ABCG1 rs221948 (risk allele: A), and CYP51A1 rs7797834 (risk allele: G). The average number of risk alleles in the controls was four. The average number of risk alleles in the cases (CTCAE grades two and three) was six. The low risk group was defined as individuals with four or less risk alleles, whereas the intermediate risk group was defined as individuals with five to six risk alleles, and the high risk group was defined as individuals with seven and more risk alleles. The curve was  49 generated in SPSS® version 19, where proportional hazard ratios and p-values were calculated with Cox regression model.  3.11 Linkage Disequilibrium Studies Haploview 4.2 (BARRETT et al. 2005) and HapMap Genome Browser (release #24, Phase 1 and 2 dataset) were used to create the LD plots containing PON1 SNPs (PON1 rs854549, PON1 rs854560, and PPP1R9A rs705377) in the Hapmap population (CEU, YRI, CHB+JPT), as well as to calculate the LD of SNPs within the aforementioned Hapmap population. PLINK version 1.07 (Harvard University, MA, USA) was used to calculate two SNP pairwise LD for specific SNPs in this population, including; PON1 rs854549 and PON1 rs854560, PPP1R9A rs705377 and PON1 rs854549, PPP1R9A rs705377 and PON1 rs854560, and ABCA4 rs3789433 and ABCA4 rs549848.  3.12 Imputation Analyses The genotypes and frequencies of the patients within this cohort and the individuals were merged together with the Human Build 36. Since Impute2 uses position as SNP identifier, each SNP was checked for duplicated position in the map file. Also, since Impute2 uses segments up to 5 Mb, the SNPs in four regions of interest (ABCA4; CYP51A1, PON1, PPP1R9A; SLCO1C1; ABCG1) were grouped by location and analyses were run for all segments across chromosomes 1, 7, 12, and 21 (Appendix A.5). All segments contained enough SNPs in every panel to be imputed. When combining datasets (study and reference panels), all SNPs were aligned to the same strand convention for imputation. Impute2 (HOWIE et al. 2009) was used to change non-ambiguous SNPs (A/C, G/T, A/G, C/T) from  50 the study panel as based on reference panels. For the remaining ambiguous SNPs (A/T, and C/G), a MAF comparison between the study panel and the reference panel was conducted. Since SNPs with MAF that are near 0.5 cannot be determined for sure, they were considered absent from the study panel and imputed. With this method, the original genotypes with the imputed genotypes for each ambiguous SNP were compared. A cutoff of 0.90 was utilized in Impute2 as the final call for one of the three-allele combination (AA, AB, or BB) for the 140 individuals.       51 Chapter  4: Results 4.1 Patient Cohort Of the 1,416 patients who received vincristine as a core component of their chemotherapy treatment, 154 patients were diagnosed with Wilms tumor or rhabdomyosarcoma. Fourteen of these patients were excluded from the study for the following reasons: ten cases who were characterized with a CTCAE grade one VIN were considered too mild for inclusion in this study, two patients with low quality DNA samples, and two patients whose adverse event could not be graded independently from their tumor location and surgery site, which therefore affects the observed expression (phenotype) and severity of their VIN event (e.g.: foot drop may present itself after surgery in the pelvic region to remove the tumor. It cannot be determined whether the occurrence and the severity of this ADR is due to the surgery or VIN). The abdominal surgery of one patient could not be excluded as the cause or contributor to post-surgery complaints of constipation. Since it was not possible to determine the relative contribution of vincristine to the constipation and this patient did not have other ADRs that could be separately graded, this patient could not be established as a true case. The second patient had a rhabdomyosarcoma that was compressing on their spinal cord. Surgical attempts to debulk the lesions in the thoracic region resulted in paralysis below the thoracic cavity. Due to the paralysis, it was not possible to determine whether this patient would have developed VIN and therefore, this patient could not be established as a true control. One hundred and forty patients met the requirements for this study, including tumour type, and had CTCAE grades of zero (controls) or two to five (cases).   52 4.2 Patient Demographics and Clinical Factors In the combined cohort of patients who met the eligibility requirements, 90 patients were diagnosed with Wilms tumor and 50 patients with rhabdomyosarcoma. In total, there were 45 VIN cases, where 24 (53.33%) cases were patients with Wilms tumor and 21 (46.67%) cases were patients with rhabdomyosarcoma. In the 95 controls, there were 66 (69.47%) patients with Wilms tumor and 29 (30.53%) patients with rhabdomyosarcoma. There were no significant differences in the proportion of cases and controls within the Wilms tumor and rhabdomyosarcoma populations (p-value = 0.0889). In the combined group and the rhabdomyosarcoma cohort, there was a moderate association of age with neurotoxicity where cases were found to be significantly older than the controls (p-value = 0.0190 and 0.0400, respectively). In the combined cohort, the mean age of all cases with VIN was significantly older at 6.07 years (ranging from 1.20 to 15.70 years old), as compared to the average age of 4.42 years (ranging from 0.71 to 17.89 years old) in controls (Table 3). Similarly, cases in the rhabdomyosarcoma cohort were significantly older than controls; the average age of the cases was 7.97 years (ranging from 1.84 to 15.14 years old) and the average age of the controls was 6.12 years (ranging from 0.71 to 17.89 years old). Although there was no significant difference in the age of cases (4.41 years old, ranging 1.20 to 15.70 years old) and controls (3.67 years old, ranging 0.77 to 9.21 years old) in patients with Wilms tumor (p-value = 0.877), the trend of VIN cases being older than controls was still observed in this tumor group (Table 3). Between overall tumor type cohorts, the rhabdomyosarcoma cohort was older than the Wilms tumor cohort. Within each tumor type, the individual treatment protocols varied slightly but did not differ substantially. Some patients had relapsed in their treatment, which required additional  53 Table 3   Patient Demographics and Clinical Factors Between Cases and Controls  Wilms Tumor  Rhabdomyosarcoma  Combined Cohort  VIN Cases (n=24)‡ Controls (n=66) P-value  VIN Cases (n=21)‡ Controls (n=29) P-value  VIN Cases (n=45)‡ Controls (n=95) P-value Age (years) 4.41 (1.20,15.70) 3.67 (0.77,9.21) 0.877  7.97 (1.84,15.14) 6.12 (0.71,17.89) 0.0400*  6.07 (1.20,15.70) 4.42 (0.71,17.89) 0.0190* Male n (%) 12 (50.00) 30 (45.45) 0.812  11 (52.38) 16 (55.17) 1.000  23 (51.11) 46 (48.42) 0.857 Doses, Total Number§ 15.63 (7,42) 15.53 (5,3) 0.277  30.81 (10,61) 24.34 (4,53) 0.0623  22.71 (7,61) 18.22 (4,53) 0.0848 Dose, Cumulative (mg/m2) 23.98 (9.25,63.00) 22.77 (7.50,49.00) 0.674  43.00 (15.00,91.50) 34.90 (2.70,79.50) 0.257  32.86 (9.25,91.50) 26.47 (2.7,79.50) 0.0636 Duration of Vincristine Treatment, Days 280.63 (115,2489) 249.39 (63,827) 0.269  317.76 (98,635) 321.21 (81- 1194) 0.128  297.96 (98,2489) 271.32 (63,1194) 0.960 Concomitant medication n (%)    CYP3A4 Inhibitors, 1 or more n (%) 2 (8.33) 8 (12.12) 1.000  1 (4.76) 6 (20.69) 0.215  3 (6.67) 14 (14.74) 0.268    Steroids n (%) 1 (4.17) 6 (9.09) 0.670  8 (38.10) 5 (17.24) 0.116  9 (20.00) 11 (11.58) 0.202 Solid Tumor Type    Wilms Tumor n (%)         24 (53.33) 66 (69.47) 0.0886     Rhabdomyosarcoma n (%)         21 (46.67) 29 (30.53) Values are represented as mean (min, max) or numbers (percentage). For continuous and categorical variables, the p-values were obtained by the non-parametric Kruskal-Wallis test and the Fisher exact test, respectively. *Statistically significant at 0.05 type I error rate. ‡Grades two or three of VIN according to CTCAE v4.03. §Includes reduced doses of vincristine and excludes withheld doses of vincristine.      54 doses and increased the length of treatment time on vincristine. The majority of the Wilms patients received one of three chemotherapy protocols (I, DD4, or UH-2) of the NWTS protocols. In these protocols, the number of doses of vincristine ranged from 14 to 19 doses. Excluding withheld doses and including reduced doses, Wilms tumor cases and controls were given an average of 15.63 and 15.53 doses throughout their treatment, respectively (p-value = 0.277) (Table 3). With D9803 and D9602 as the two main rhabdomyosarcoma protocols, rhabdomyosarcoma patients were expected to receive 30 to 36 doses. Excluding withheld doses and including reduced doses, 30.81 and 24.34 doses were the average number of doses given to rhabdomyosarcoma cases and control patients, respectively (Table 3). Even though cases received a higher average number of doses, this difference was not statistically significant (p-value = 0.0628). Similarly, across the combined cohort and both tumor types, the total number of doses given to cases was consistently higher than those given to controls, whereas the combined cohort cases were given an average of 22.71 doses and an average of 18.22 doses for controls. However, also this difference was not statistically significant (p- value = 0.0848) (Table 3). This observation was closely tied to the cumulative dose, where the dosage was higher in cases than controls in the Wilms tumor (23.98 mg/m2 and 22.77 mg/m2, respectively, p-value = 0.674), rhabdomyosarcoma (43.00 mg/m2, 34.90 mg/m2, respectively, p-value = 0.257), and combined cohorts (32.86 mg/m2, 26.47 mg/m2, respectively, p-value = 0.0636) (Table 3). Due to the longer duration of treatment, the total number of doses and cumulative dose was higher in the rhabdomyosarcoma cohort (27.06 doses and 38.30mg/m2) as compared to the Wilms tumor cohort (15.56 doses and 23.09 mg/m2).  55 There was no significant gender difference in the combined cohort where there were 23 males (51.11%) and 46 females (48.42%); nor were there gender differences in the individual tumor groups (Table 3). With the exception of rhabdomyosarcoma cases who used more steroids than controls, the concomitant use of CYP3A4 inhibitors and steroids were lower in cases than controls across both tumor types and the combined cohort. The use of either drug type was not statistically significant within each tumor group and within the combined cohort. In summary, there were no significant differences in gender, total number of vincristine doses received, cumulative vincristine dose, or duration of vincristine treatment, or use of CYP3A4 inhibitors and steroids between cases and controls in the combined cohort, Wilms tumor, or rhabdomyosarcoma cohort (Table 3).  4.3 Assessing the Severity and Types of Vincristine-Induced Neurotoxicity and the Interventions Provided All 45 cases of VIN had poly-neuropathies and there were no individuals with a mono-neuropathy. Wilms tumor cases were significantly younger (4.41 years old) than rhabdomyosarcoma cases (7.97 years old, p-value = 0.00183) (Table 4). Similarly, due to the differences in protocols, in comparing the cases between patients with Wilms tumor and rhabdomyosarcoma, there was a significant difference in the total number of vincristine doses (15.63 doses and 30.81 doses, p-value = 2.60x10-6), cumulative dose (23.98 mg/m2 and 43.00 mg/m2, p-value = 1.87x10-5), and duration of vincristine treatment (280.63 days compared to 317.76 days, p-value = 1.58x10-4) (Table 4). Although patients with Wilms tumor received fewer doses, a smaller cumulative dose, and a shorter duration of vincristine treatment, the mean onset to the first VIN event was not significantly different (p-value =  56 Table 4   Vincristine-Induced Neurotoxicity and the Interventions Provided to Wilms Tumor and Rhabdomyosarcoma Cases   VIN Cases‡   Wilms Tumor (n=24) Rhabdomyosarcoma (n=21) P-value Combined Cohort (n=45) Age (years) 4.41 (1.20,15.70) 7.97 (1.84,15.14) 0.00183*  6.07 (1.20,15.70) Male n (%) 12 (50.00) 11 (52.38) 1.000  23 (51.11) Doses, Total Number§ n 15.63 (7,42) 30.81 (10,61) 0.00000260*  22.71 (7,61) Dose, Cumulative (mg/m2) 23.98 (9.25,63.00) 43.00 (15.00,91.50) 0.0000187*  32.86 (9.25,91.50) Time to First Major VIN Event, Days 81.96 (2,335) 73.76 (0,188) 0.810  78.13 (0,335) Duration of Vincristine Treatment, Days 280.63 (115,2489) 317.76 (98,635) 0.000158*  297.96 (98,2489) Overall Type of Neurotoxicity    Peripheral Neuropathy only n (%) 18 (75.00) 16 (76.19) 0.575     34 (75.56)    Autonomic Neuropathy only n (%) 2 (8.33) 0 (0.00)  2 (4.44) Peripheral and Autonomic Neuropathy n (%) 4 (16.67) 4 (19.05)  8 (17.78) Peripheral and Autonomic Neuropathy, and Central Nervous System Effects n (%) 0 (0.00) 1 (4.76)  1 (2.22) CTCAE Grade (VIN Severity)    Grade 2, Total Number n (%) 15 (62.50) 9 (42.86) 0.238   24 (53.33)    Grade 3, Total Number n (%) 9 (37.50) 12 (57.14)  21 (46.67)    Overall CTCAE Grade 2.38 (2,3) 2.57 (2,3) 0.240  2.47 (2,3) VCR-Specific Interventions Withheld Doses, Total Number of Patients n (%) 6 (25.00) 9 (42.86) 0.226  15 (33.33) Reduced Doses, Total Number of Patients n (%) 5 (20.83) 10 (47.62) 0.112  15 (33.33) Withheld, Number of Doses 0.93 (0,3) 5.95 (0,20) 0.0404*  1.78 (0,20) Reduced, Number of Doses 0.79 (0,9) 3.24 (0,12) 0.0900  1.93 (0,12) Other Interventions Rehabilitation Therapy n (%) 10 (41.67) 15 (71.43) 0.0715  25 (55.56) Assistive Device(s) n (%) 4 (16.67) 13 (61.90) 0.00247*  17 (37.78) Neurology Consult n (%) 4 (16.67) 4 (19.05) 1.000  8 (17.78) Pharmacotherapy Intervention(s) to Treat Neurotoxicity Symptoms n (%) 14 (58.33) 4 (19.05) 0.0138*  18 (40.00) Hospitalization / Surgery n (%) 1 (4.17) 1 (4.76) 1.000  2 (4.44) Values are represented as mean (min, max) or numbers (percentage). For continuous and categorical variables, the p-values were obtained by the non-parametric Kruskal-Wallis test and the Fisher exact test, respectively.  ‡Grades two or three of VIN according to CTCAE v4.03.  §Includes reduced doses of vincristine and excludes withheld doses of vincristine.  *Statistically significant at 0.05 type I error rate   57 0.810) between cases with Wilms tumor (11.7 weeks) and rhabdomyosarcoma (10.53 weeks). In the combined cohort, the average time to the first major VIN event was 11.16 weeks. Twenty-four (53.33%) children suffered from CTCAE grade two VIN, and 21 (46.67%) of children suffered from CTCAE grade three VIN required vincristine-specific medical interventions. The severity of VIN was not significantly affected by tumor type (p- value = 0.238). No patients had life-threatening (grade four) or death-causing (grade five) VIN. This toxicity resulted in 30 (66.66%) cases having doses withheld and/or reduced. This proportion was not significantly different between tumor types or by type of intervention; doses were withheld in six Wilms tumor cases and nine rhabdomyosarcoma cases (p-value = 0.226), and doses were reduced in five Wilms tumor cases and ten rhabdomyosarcoma cases (p-value = 0.112) (Table 4). The average number of doses reduced was also not significantly different between tumor types (p-value = 0.0900); however, the average number of doses withheld in patients with Wilms tumor (0.93 doses) and rhabdomyosarcoma (5.95 doses) differed significantly between tumor groups at p-value of 0.0404 (Table 4). PN was the most severe form of neurotoxicity (n =34, 76%) (Table 4). A combination of peripheral and autonomic neuropathy was the second most common type of neuropathy in eight (18%) individuals, followed by AN in two (4%) patients. One patient (2%) experienced vincristine-induced CNS effects, peripheral and autonomic neuropathy (Table 4). The most frequent poly-neuropathy complaints were foot drop, muscle weakness, and limb and joint pain, which occurred at a frequency of 42% (n = 19), 40% (n = 18), and 40% (n = 18), respectively (Table 5). The less common VIN symptoms included: constipation and ileus (n = 11, 24%), ptosis (n = 8, 18%), jaw pain (n = 7, 16%), vocal cord paralysis (n = 6, 13%), and SIADH (n = 1, 2%) (Table 5). With 10 out of 11 patients experiencing constipation, it  58 Table 5   The Frequency of Each Type of Vincristine-Induced Neurotoxicity Sign or Symptom   VIN Cases‡  VIN Sign or Symptom Wilms Tumor (n=24) Rhabdomyosarcoma (n=21) P-value  Combined Cohort (n=45) Peripheral Neuropathy (Sensory)    Jaw Pain n (%) 4 (16.67) 3 (14.29) 1.000  7 (15.56)    Limb and Joint Pain n (%) 11 (45.83) 7 (33.33) 0.543  1eight (40.00) Peripheral Neuropathy (Motor)    Foot Drop n (%) 8 (33.33) 11 (52.38) 0.237  19 (42.22)    Ptosis n (%) 4 (16.67) 4 (19.05) 1.000  8 (17.78)    Vocal Cord Paralysis n (%) 3 (12.50) 3 (14.29) 1.000  6 (13.33)    Muscle Weakness n (%) 4 (16.67) 14 (66.67) 0.00088*  18 (40.00) Autonomic Neuropathy    Constipation and Ileus n (%) 6 (25.00) 5 (23.81) 1.000  11 (24.44) Central Nervous System Effects    SIADH n (%) 0 (0.00) 1 (4.76) 0.467  1 (2.22) Values are represented as numbers (percentage). For categorical variables, the p-values were obtained by the Fisher exact test, respectively ‡grades two or three of VIN according to CTCAE v4.0 *statistically significant at 0.05 type I error rate    59 was the most frequent manifestation of AN, with ileus being observed in only one patient. Rhabdomyosarcoma patients were significantly more likely to experience muscle weakness (n = 14, 66.67%, p-value = 0.00088) than Wilms patients (Table 5). No significant differences between tumor types were seen for the other typical VIN symptoms. The majority of VIN patients received rehabilitation therapy (n = 25, 56%), assistive devices (n = 17, 38%), or pharmacotherapy interventions to treat neurotoxicity symptoms (n = 18, 40%) (Table 4). Hospitalization and surgery was required for two (4%) patients to monitor their SIADH condition and lengthen their Achilles tendon, respectively. Rhabdomyosarcoma cases (n = 13, 61.90%) were more likely to require assistive devices (p- value = 0.00247) (Table 4) than Wilms patients (n = 4, 16.67%). This observation was unrelated to the location of the tumor as the majority of rhabdomyosarcoma patients had superficial tumors that were located in the nasopharyngeal, facial, testicular, or axilla region (n = 13, 56.52%). The remaining ten (43.48%) cases who had the tumor located in their forearm and pelvic region were graded and classified on jaw pain, ptosis, or constipation, which were signs and symptoms of VIN that were unrelated to the mobility of the extremities. Seven patients with rhabdomyosarcoma (53.8%) required splints or braces for leg or wrists, four patients (30.8%) required ankle foot orthosis (AFO) and two (15.4%) needed a walker or wheelchair. Only two (50%) Wilms tumor patients required a splint or brace, and two (50%) required AFOs. Conversely, Wilms tumor patients were significantly more likely to require pharmacotherapy-based interventions to relieve vincristine-induced pain (jaw, limbs, joints) or vincristine-induced constipation (p-value = 0.0138) (Table 4). Out of all patients with Wilms tumor who received pharmacotherapy-based interventions, 11 (79%) patients received analgesics and three (21%) patients received regular laxative use,  60 whereas all rhabdomyosarcoma patients with pharmacotherapy interventions (n = 4) required analgesics. Overall, there were no significant differences in gender, frequency of each grade, VIN severity, and number of patients with doses of vincristine withheld and reduced between the tumor groups (Table 4).  4.4 Variation by Ancestry Patient ethnicity in the study population was highly variable. The self-reported ancestry indicated that there were 91 individuals who had all four grandparents from one country: Europe (n = 45), Canada (n = 24), Caribbean (n = 4), East Asia (n = 12), South Asia (n = 4), Middle East (n = 1), and South America (n = 1). There were 33 individuals with grandparents from two or more countries, which included: Europe/Canada (n = 18), Aboriginal/Canada (n = 3), Europe/Aboriginal/Canada (n = 1), Aboriginal/Europe (n = 2) Europe/Latin America (n = 2), Africa/Canada (n = 1), Africa/Europe (n = 1), Africa/South Asia (n = 1), Caribbean/Canada (n = 1), Europe/Canada/Latin America (n = 1), Europe/South Asia (n = 1), and Europe/South America/Caribbean (n = 1). No self-reported ancestry information was available for 16 individuals. In comparison to a scatterplot of the first two principal components (Figure 2), the self-reported ancestry corresponded fairly well with the patient’s genetic ancestry. Three major ethnic groups (European, Asian, African) were roughly defined by separating the scatterplot into four quadrants where there were approximately: 40 cases and 79 controls with European descent; 4 cases and 11 controls with Asian descent; and 1 case and 5 controls in the African descent (Figure 2). The grandparents of the one case in the African population  61              Based on the genotypes of patients with Wilms tumor and rhabdomyosarcoma, this scatter plot shows the variation of cases ( ) and controls ( ) by ethnicity. The principal components of patients are compared to the reference samples in the European (CEU, ), Asian (CHB and CHD, ), and African (YRI, ) Hapmap populations.  Figure 2  Ethnic Distribution of the Patients in the Cohort Asian African European Study Controls Study Cases Hapmap African (YRI) Hapmap Asian (CHB, CHD) Hapmap European (CEU)  62 were reported to be from Jamaica, England, France, and El Salvador. Therefore, this patient was one-quarter African, which was confirmed genetically in Figure 2 with the principal components 1 and 2 between Africa and Europe. With the exception of one control patient (whose genetic ancestry was found between the Asia and Europe and self-reported grandparental ancestry as three from Guyana and one from South America), the self-reported ancestry corresponded well with the genetic ancestry and was able to identify the ancestry of the 16 patients with unknown self-reported ancestry.  4.5 Genetic Analyses  Beginning with 4,536 SNPs, 403 SNPs were manually removed SNPs because of low signal intensities, low clustering scores, and high replication errors. This included the removal of one candidate SNP, ABCB1 rs1045642 that consistently failed genotyping for most DNA samples. As a duplicate SNP on the panel, ABCB1 rs1045642 was twice removed from the analysis due to the failure of two different probes. In a second step, an additional 155 SNPs were removed due to: SNP homozygosity (n = 97), genotyping completion rates of less than 98% across patients (n = 33) (Appendix A.1), and Hardy Weinberg disequilibrium (n = 25) (Appendix A.2). No patients were removed from the cohort because of IBD or low SNP genotyping completion rates. After data cleaning, a total of 3,978 SNPs remained to be evaluated in 140 patients.  The imputation analyses identified 26 SNPs that were duplicated by position and were merged together. The average concordance were always higher than 84.91% for all chromosomes, meaning that in average, at least 97.15% of imputed genotypes correspond to the original ones. Only 15 of 69 ambiguous SNPs have been imputed, whereas the 54  63 remaining ambiguous SNPs are not present in the Hapmap panel. There were 9,397 SNPs with a completion rate higher than or equal to 98%, with a cut off of 0.90. In the 140 individuals, there were 615 SNPs with a p-value < 0.05, when ancestry, age, and tumor type were controlled. The most significant SNP was PON1 rs854549 on chromosome 7 at a p- value of 0.0006103.  4.5.1 Screening for Genes Involved in the General Drug Biotransformation Pathways  Utilizing the simpleM method, 136 independent SNP groups were derived from the 3,978 SNPs and were used to correct for multiple testing, resulting in a significance cutoff at p-value = 0.00037. Given the wide ethnic variation in the population and the potential differences in allele frequencies by ancestry, it was necessary to adjust the additive logistic regression model for ancestry. In the first analysis adjusting for population stratification, 37 SNPs were significant at the level of p-value < 0.05 in the combined cohort prior to multiple testing correction, but none remained significant after this correction was applied (Figure 3). In the separate logistic regression models for each tumor type, the genetic variants in PON1 rs854549, ABCA4 rs3789433 and rs549848, SLCO1C1 rs10770704, CYP51A1 rs7797834, and ABCG1 rs221948 were significant in the Wilms tumor and rhabdomyosarcoma groups as well as the combined cohort with an individual p-value of <0.05 for the gene variant. PON1 rs854549, ABCA4 rs3789433, SLCO1C1 rs10770704 were also the most strongly associated SNPs in the combined cohort (Appendix A.3). These six SNPs showed a consistent association with VIN in both tumor types and in the combined cohort (Table 6). The effect sizes for these six SNPs were similar with odds ratios (ORs) ranging from 2.39 to 3.65 in the combined cohort for risk alleles, and 0.35 to 0.37 for protective variants (Table 6).  64  This Manhattan plot contains the p-values adjusted for ancestry, age, and tumor type for all the SNPs in ADME-Tox panel. SNPs are ordered by position and chromosome. P-values <0.05 are statistically significant at type I error rate, unadjusted for multiple testing. P-values <0.00037 are statistically significant at type I error rate, adjusted for multiple testing. !" !#$" %" %#$" &" &#$" '" '#$" !" $!!" %!!!" %$!!" &!!!" &$!!" '!!!" '$!!" (!!!" !" # $ % & '( )* +,-#.#/#.0* % "& ""' "( ""$ ") "* "+ ", "%! "%% "%& "%' "%( "%$ "%) "%*"%+"%,""""&!""&%"&&"&'" !"#$%-/123234* &'("$($%-/%&55&5&3* ()!*$+$%-/5545163* +,(+-%-/6514366* 7!89":0** ;*&<&&&65* 7!89":0** ;*&<&2* +,(.$%-/==%431* +,(+-%-/234131* Figure 3  Manhattan Plot  65 Table 6   Genetic Variants that are Significantly Associated with Vincristine-Induced Neurotoxicity      Wilms Tumor (n= 90)  Rhabdomyosarcoma (n=50)  Combined Cohort (n=140)  Combined Cohort (n=140) Gene SNP rsID SNP Location & Effects Gene Function  OR (95% CI)† P-value†  OR (95% CI)† P-value†  OR (95% CI)† P-value†  OR (95% CI)† P-value‡ PON1 rs854549 flanking 3’-UTR Antioxidant  3.49 (1.44-8.47) 0.00564*  3.02 (1.03-8.84) 0.0436*  3.04 (1.57-5.89) 0.00092*  3.34 (1.67-6.66) 0.00061* ABCA4 rs3789433 intron Transporter  2.97 (1.27-6.95) 0.0119*  3.33 (1.16-9.57) 0.0250*  2.90 (1.53-5.49) 0.00109*  3.11 (1.60-6.02) 0.00075* SLCO1C1 rs10770704 intron Transporter  0.39 (0.17-0.90) 0.0282*  0.28 (0.09-0.85) 0.0256*  0.35 (0.18-0.66) 0.00134*  0.31 (0.16-0.62) 0.00098* ABCG1 rs221948 intron Transporter  3.06 (1.01-9.32) 0.0480*  5.08 (1.07-23.9) 0.0397*  3.65 (1.54-8.68) 0.00328*  3.57 (1.47-8.63) 0.00467* CYP51A1 rs7797834 synonymous H453H Metabolizing enzyme  2.09 (1.01-4.34) 0.0467*  2.77 (1.00-7.61) 0.0483*  2.39 (1.33-4.28) 0.00332*  2.32 (1.27-4.21) 0.00559* ABCA4 rs549848 intron Transporter  0.26 (0.09-0.74) 0.0119*  0.32 (0.11-0.92) 0.0361*  0.37 (0.18-0.74) 0.00533*  0.29 (0.13-0.61) 0.00128* The shown p-value is the individual p-value of the genetic component of the logistic regression model: †Values derived from a logistic regression model adjusted for principal components. ‡Values derived from a logistic regression model adjusted for principal components, age, and tumor type. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing.   66  Significant differences in age between tumor types and between cases and controls, required an additional adjustment for age. In this second analysis, age, tumor type, and principal components 1 and 2 were added as covariates in the logistic regression model. None of the 3,978 SNPs were significantly associated with VIN after correction for multiple tests as demonstrated by the Manhattan Plot (Figure 3). Prior to correcting for multiple tests, a total of 173 SNPs were statistically significant at the level of p-value < 0.05, where 69 (39.9%) SNPs had a postulated and/or confirmed effect on the gene (Appendix A.3). Compared to the initial ancestry-only adjustment, the six SNPs in PON1, ABCA4, SLCO1C1, ABCG1, and CYP51A1 were more significant (Table 6). In the Wilms tumor, rhabdomyosarcoma patients, and the total cohort, the PON1 rs854549 risk allele (A) was present in 19 (79.2%), 13 (61.9%), and 32 (71.1%) of the individuals (Table 6) with VIN, respectively, conferring a combined cohort OR of 3.34 (95% confidence interval (CI): 1.67- 6.66) (Table 6). For ABCA4 rs3789433 (minor risk allele: A), CYP51A1 rs7797834 (minor risk allele: G), ABCG1 rs221948 (minor risk allele: A), ABCA4 rs549848 (minor protective allele: A), and SLCO1C1 rs10770704 (minor protective allele: G), the adjusted combined OR and 95% CI: were 3.90 (1.63-9.34), 2.75 (1.38-5.47), 2.41 (0.76-7.62), 0.36 (0.15-0.85), 0.36 (0.15-0.85), respectively (Table 6). All six SNPs genotyped very well and showed clearly distinct clusters for the three genotype groups (Figure 4). ABCG1 rs221948 was the only SNP out of the six candidates for which no homozygous carriers of the minor allele were observed (Table 6).  In a sub-analysis of the 87 self-reported Caucasian patients (Europe, Canada, or Europe/Canada) adjusted for tumor type and age, five of the six previously identified SNPs were still significant at the level of <0.05: PON1 rs854549 (OR 2.30, 95% CI: 1.06-5.00, p-  67   SNP clusters are generated with GenomeStudio Data Analysis Software v2010 in a Cartesian plot with normalized intensity !"#$%!"#$%$%&' Norm Intensity (A) N o rm  I n te n s it y  ( B ) &'(&)%!"()#&%((' *+("$($%!"()#&%((' &'(&)%!"$%&#%#' &'(,$%!"**+&%#' (-!.$&$%!"))&)#(%' Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Figure 4   SNP Clusters of PON1 rs854549, ABCA4 rs3789433, SLCO1C1 rs10770704, ABCG1 rs221948, CYP51A1 rs7797834, and ABCA4 rs549848  68 value = 0.0350), ABCA4 rs3789433 (OR 3.90, 95% CI: 1.63-9.34, p-value = 0.00222), SLCO1C1 rs10770704 (OR 0.36, 95% CI: 0.15-0.85, p-value = 0.0194), CYP51A1 rs7797834 (OR 2.75, 95% CI: 1.38-5.47, p-value = 0.00383), and ABCA4 rs849848 (OR 0.36, 95% CI: 0.15-0.85, p-value = 0.0204) risk alleles (Table 7). ABCG1 rs221948 was the only SNP that was no longer significant (OR 2.41, 95% CI: 0.76-7.62, p-value = 0.133) (Table 7). Carriers of an increasing number of putative risk alleles in the PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 SNPs had a higher likelihood of incurring CTCAE grades two or three of VIN during chemotherapy treatment with vincristine (p-value = 4.08x10-10 and 1.76x10-7, respectively) (Figure 5). The two ABCA4 SNPs were not in strong LD and were independently associated with VIN, where the p-value was <0.05 in a multivariate logistic regression model that included both SNPs (p-value = 0.0252). With a hazard ratio of 18.20 (95% CI: 6.93-47.83, p-value = 3.39x10-9), 14 out of 18 (77.78%) patients with seven or more putative risk alleles (considered high risk) developed VIN during treatment, whereas only 6 out of 71 (8.45%) patients with four or less of these variants (considered low risk) developed VIN during chemotherapy treatment (Figure 5). Similarly, in comparing the intermediate risk to the low risk patients, there were 25 out of 51 (49.02%) patients with five to six putative risk alleles who experienced VIN, with a hazard ratio of 7.78 (95% CI: 3.19- 19.00, p-value = 6.67x10-6) (Figure 5). Upon further examination of the gene, PON1 rs854549 was found to be in high LD with two other SNPs that were genotyped in this study, PPP1R9A rs705377 and PON155 rs854560 (Figure 6 to 8). In this study, the LD between PPP1R9A rs705377 and PON1 rs854549, and PON155 rs854560 and PON1 rs854549 was r2 = 0.773 and 0.559 respectively  69 Table 7   Subgroup Analysis of Genetic Variants that are Significantly Associated with Vincristine-Induced Neurotoxicity: Stratification of Patients by European Ancestry     Combined Cohort  Gene SNP rsID  OR (95% CI)† P-value† 1 PON1 rs854549  2.30 (1.06-5.00) 0.0350*  2 ABCA4 rs3789433  3.90 (1.63-9.34) 0.00222* 3 SLCO1C1 rs10770704  0.36 (0.15-0.85) 0.0194* 4 ABCG1 rs221948  2.41 (0.76-7.62) 0.132 5 CYP51A1 rs7797834  2.75 (1.38-5.47) 0.00383* 6 ABCA4 rs549848  0.36 (0.15-0.85) 0.0204* †Values derived from logistic regression model adjusted for age, and tumor type. The shown p-value is the individual p-value of the genetic component of the logistic regression model. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing.     70             This curve demonstrates that the increase in number of PON1 rs854549, ABCA4 rs3789433, SLCO1C1 rs10770704, ABCG1 rs221948, CYP51A1 rs7797834, and ABCA4 rs849848 risk alleles is associated with earlier onset and increased likelihood of VIN, when the number of risk alleles are segregated into low, intermediate, and high risk of developing VIN during chemotherapy treatment. Values are represented as numbers (percentage) and the p-values were obtained by the Fisher exact test. †P-value comparing CTCAE v4.03 grade zero and grade two. ‡P-value comparing CTCAE v4.03 grade zero and grade three. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing. CTCAE v4.03 Grade  Low Risk (n=71) Intermediate Risk (n=51) High Risk (n=18)  P-value   ≤4 Risk Alleles 56 Risk Alleles ≥7 Risk Alleles 0  65 (91.54%) 26 (50.98%) 4 (22.22%) 2  2 (2.816%) 15 (29.41%) 7 (38.88%)  4.08 x 10-10* † 3  4 (5.633%) 10 (19.60%) 7 (38.88%)  1.76 x 10-7* ‡   Low risk (65 controls, 6 cases) !4 risk alleles High risk (4 controls, 14 cases) "7 risk alleles Intermediate risk (26 controls, 25 cases) 5-6 risk alleles Days Post Initiation of Vincristine P a ti e n ts  w it h  V in c ri s ti n e -I n d u c e d  N e u ro to x ic it y  ( C a s e s , % ) HR, 95% CI: 18.20 (6.93-47.83) P-value: 3.39x10-9 HR, 95% CI: 7.78 (3.19-19.00) P-value: 6.67x10-6 HR: 1 Figure 5 PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 Risk Alleles Increase the Chances of Developing Vincristine-Induced Neurotoxicity  71  Table 8   Linkage Disequilibrium of SNP Pairs: PON1, PPP1R9A, PON1, and ABCA4 SNP Pair  r2 PON1 rs854549 PON1 rs854560  0.559 PPP1R9A rs705377 PON1 rs854549  0.773 PPP1R9A rs705377 PON1 rs854560  0.430     ABCA4 rs3789433 ABCA4 rs549848  0.023 LD in specific SNPs in the patients in this study calculated from PLINK version 1.07.    72           Physical location of PON1 rs854549 and two flanking SNPs in surrounding genes. The LD between PON1 rs854549 and PPP1R9A rs705377 is r2=0.71. The LD between PON1 rs854549 and PON1 rs854560 is r2=0.50.  !!!"#$%!"#$%&'$$! !&'"("#(&)&)*!+,-./012! !&'"))("#(&)&3%! 456!7-89-8!1:8;<-=:>! Figure 6 Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the European (CEU) Hapmap Populations  73 Physical location of PON1 rs854549 and two flanking SNPs in surrounding genes. The LD between PON1 rs854549 and PPP1R9A rs705377 is r2=1.0. The LD between PON1 rs854549 and PON1 rs854560 is r2=0.04. !!!"#$%!"#$%&'$$! !&'"("#(&)&)*!+,-./012! !&'"))("#(&)&3%! 456!7-89-8!1:8;<-=:>! Figure 7 Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the African (YRI) Hapmap Populations  74 Physical location of PON1 rs854549 and two flanking SNPs in surrounding genes. The LD between PON1 rs854549 and PPP1R9A rs705377 is r2=1.0. The LD between PON1 rs854549 and PON1 rs854560 is r2=0.04. !!!"#$%!"#$%&'$$! !&'"("#(&)&)*!+,-./012! !&'"))("#(&)&3%! 4567819!5-:;-:!1<:=>-?<@! Figure 8 Linkage Disequilibrium Plot of PON1, PPP1R9A, and the Intergenic Region in the Asian (CHB, JPT) Hapmap Populations  75 (Table 8). These LD values are similar, but are moderately stronger than the European (CEU) Hapmap populations at r2 = 0.71 and r2 = 0.50, respectively (Figure 6). The LD between PPP1R9A rs705377 and PON1 rs854549 is stronger in individuals in the African (YRI) and Asian (CHB+JPT) Hapmap populations, where both have r2 = 1.0 (Figure 7 and 8). However, as a strong contrast, the LD between PON155 rs854560 and PON1 rs854549 is much weaker in the African (YRI) and Asian (CHB+JPT) Hapmap populations, at r2 = 0.04 and r2 = 0.16, respectively (Figure 7 and 8). In comparing the MAF of the combined cohort with the MAF of the European, African, and Asian Hapmap populations in PON1 rs854549 (0.286, 0.381, 0.092, 0.056, respectively), PPP1R9A rs705377 (0.307, 0.424, 0.092, 0.942, respectively) and PON155 rs854560 (0.332, 0.400, 0.140, 0.944, respectively), our study population showed the most similar allele frequencies to the European Hapmap population. Both of these SNPs linked to PON1 rs854549 had ORs greater than one in the combined cohort, suggesting a similar association with VIN, although only PPP1R9A rs705377 was significantly associated at the level of p-value < 0.05 in the logistic regression model adjusted for ancestry (Table 9). However, both were significant in the principal component-, age-, and tumor type-adjusted model (Table 9). In the Wilms tumor, rhabdomyosarcoma and the combined cohorts, the risk allele for PPP1R9A rs705377 was present in 18 (75.0%), 13 (61.9%), and 21 (46.7%) of the individuals with VIN, conferring an OR of 2.21 (95% CI: 1.12-4.37), 1.78 (95% CI: 0.72-4.40), and 1.91 (95% CI: 1.08-3.36), respectively (Table 9). For PON155 rs854560, the risk allele was present in 17 (70.8%), 12 (57.1%), and 22 (48.9%) of the individuals with VIN, conferring an OR of 1.84 (95% CI: 0.88-3.82), 1.92 (95% CI: 0.75-4.96), and 1.61 (95% CI: 0.89-2.90) in the Wilms tumor, rhabdomyosarcoma, and the combined cohorts, respectively (Table 9). Whereas PON1 rs854549 and PPP1R9A rs705377  76 Table 9  PON1 rs854549, PPP1R9A rs705377, and PON1 rs854560 and their Association with Vincristine-Induced Neurotoxicity    Wilms Tumor  Rhabdomyosarcoma  Combined Cohort  Combined Cohort Gene & SNP rsID Alleles & Genotype  VIN Cases (n=24) Controls (n=66) OR (95% CI)† P-value†  VIN Cases (n=21) Controls (n=29) OR (95% CI)† P-value†  VIN Cases (n=45) Controls (n=95) OR (95% CI)† P-value†  OR (95% CI)‡ P-value‡ PON1 rs854549 A - minor allele  23 32 3.49 (1.44-8.48) 0.00565*  15 10 3.02 (1.03-8.84) 0.0436*  38 42 3.05 (1.58-5.90) 0.00092*  3.34 (1.67-6.66) 0.00061*  C - major allele  25 100    27 48    52 148  AA  4 (16.66) 3 (4.54)    2 (9.52) 1 (3.45)    6 (13.33) 4 (4.21)  AC  15 (62.50) 26 (39.39)    11 (52.38) 8 (27.58)    26 (57.77) 34 (35.78)  CC  5 (20.83) 37 (56.06)    8 (38.09) 20 (68.96)    13 (28.88) 57 (60.00) PPP1R9 A  rs705377 A - minor allele  24 40 2.15 (1.02-4.52) 0.0430*  15 14 1.85 (0.71-4.81) 0.207  34 52 1.91 (1.08-3.36) 0.02584*  1.87 (1.00-3.50) 0.0491*  G - major allele  24 92    27 44    56 138  AA  6 (25.00) 7 (10.60)    2 (9.52) 2 (6.90)    5 (11.11) 7 (7.37)  AG  12 (50.00) 26 (39.39)    11 (52.38) 10 (34.48)    16 (35.55) 50 (52.63)   GG  6 (25.00) 33 (50.00)    8 (38.09) 17 (58.62)    24 (53.33) 38 (40.00) PON1 rs854560 T - minor allele  21 41 1.71 (0.80-3.66) 0.163  13 11 2.02 (0.70-5.85) 0.194  39 54 1.60 (0.89-2.90) 0.116  2.09 (1.15-3.76) 0.0142*  A - major allele  27 91    29 47    51 136  TT  4 (16.66) 5 (7.575)    1 (4.76) 2 (6.90)    8 (17.77) 9 (9.47)  TA  13 (54.16) 31 (46.96)    11 (52.38) 7 (24.13)    14 (31.11) 50 (52.63)  AA  7 (29.16) 30 (45.45)    9 (42.85) 20 (68.96)    23 (51.11) 36 (37.89) Values are represented as numbers (percentage) or numbers (95% CI). The shown p-value is the individual p-value of the genetic component of the logistic regression model: †Values derived from a logistic regression model adjusted for principal components. ‡Values derived from a logistic regression model adjusted for principal components, age, and tumor type. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing.  77          SNP clusters are generated with GenomeStudio Data Analysis Software v2010 in a Cartesian plot with normalized intensity !"#$%!"#$%$%&' !!!$&'('!"()$*((' !"#$%!"#$%$+)' Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Figure 9 SNP Clusters of PPP1R9A rs705377 and PON1 rs854560  78 genotyped well on the GoldenGate platform, PON155 rs854560 had ambiguous sample clustering in GenomeStudio (Figure 9), which may have reduced the association of this SNP with VIN.  4.5.2 Candidate Study In the a priori SNP list of candidate SNPs that were previously reported to be associated with VIN, when adjusting for ancestry, none were found to be significantly associated (p-value < 0.05) with VIN in either tumor group or in the combined cohort: CYP3A5*3 rs776746 (OR 0.71, p-value = 0.32), CYP3A4*1B rs2740574 (OR 0.85, p-value = 0.73), ABCB1 G2677GT/A rs2032582 (OR 1.37, p-value = 0.22), and ABCB1 C1236T rs1128503 (OR 1.08, p-value = 0.75) (Table 10). For CYP3A5*6 rs10264272, no rare variants were observed in the rhabdomyosarcoma cohort, and this SNP had an overall OR of 1.41 (p-value = 0.79). Due to the observation of only one variant carrier in the Wilms tumor cohort, the association of CYP3A4*7 rs41303343 with VIN could not be evaluated (Table 10). Also, in the analysis adjusted for age, tumor type, and ancestry, none of the candidate SNPs were significant (Table 10). Whereas CYP3A5*3 rs776746, CYP3A4*1B rs2740574, ABCB1 G2677T/A rs203582, and ABCB1 C1236T rs1128503 genotyped fairly well on the GoldenGate platform, both ABCB1 C3435T rs1045642 probes had ambiguous sample clustering in GenomeStudio (Figure 10 and 11), which may have reduced the association of this SNP with VIN.  Within the candidate genes, ABCC1 rs35593 (OR 2.36, 95% CI: 1.05-5.30, p-value = 0.0367), ABCC2 rs7476245 (OR 4.11, 95% CI: 1.20-14.1, p-value = 0.0244), and ABCB1  79 Table 10 The Effects of SNPs Previously Identified to be Associated with Vincristine-Induced Neurotoxicity     Wilms Tumor  Rhabdomyosarcoma  Combined Cohort  Combined Cohort Gene & SNP rsID Alleles & Genotype Functional Effects  VIN Cases (n=24) Controls (n=66) OR (95% CI)† P-value†  VIN Cases (n=21) Controls (n=29) OR (95% CI)† P- value†  VIN Cases (n=45) Controls (n=95) OR (95% CI)† P- value†  OR (95% CI)‡ P- value‡ CYP3A5 rs776746 A Functional  7 21 1.36 (0.46-4.02) 0.58  3 9 0.50 (0.13-1.93) 0.31  10 30 0.79 (0.35-1.79) 0.57  0.86 (0.37-2.0) 0.718 (*3) G Non-functional  41 111    39 49    80 160  AA Functional  0 (0) 4 (6.06)    1 (4.76) 1 (3.45)    1 (2.22) 5 (5.263)  AG Functional  7 (29.16) 13 (19.69)    1 (4.76) 7 (24.13)    8 (17.77) 20 (21.05)  GG Non-functional  17 (70.83) 49 (74.24)    19 (90.47) 21 (72.41)    36 (80.00) 70 (73.68) rs10264272 A Functional  1 1 6.28 (0.18-214.18) 0.31  0 1 -- N/A  1 2 1.41 (0.11-18.0) 0.79  1.13 (0.075-16.98) 0.929 (*6) G Non-functional  47 131    42 57    89 188  AA Functional  0 (0) 0 (0)    0 (0) 0 (0)    0 (0) 0 (0)  AG Functional  1 (4.166) 1 (1.52)    0 (0) 1 (3.45)    1 (2.22) 2 (2.11)  GG Non-functional  23 (95.83) 65 (98.48)    21 (100) 28 (96.55)    44 (97.77) 93 (97.89) rs41303343 A Non-functional  0 1 -- N/A  0 0 -- N/A  0 1 -- N/A  -- N/A (*7) T Functional  48 131    42 58    90 189 CYP3A4*1B rs2740574 G   5 12 1.80 (0.48-6.78) 0.39  1 3 0.49 (0.04-5.76) 0.57  6 15 1.07 (0.35-3.25) 0.91  1.42 (0.45-4.48) 0.548  A   43 120    41 55    84 175  GG   0 (0) 2 (3.03)    0 (0) 0 (0)    0 (0) 2 (2.11)  GA   5 (20.83) 8 (12.12)    1 (4.76) 3 (10.34)    6 (13.33) 11 (11.57)  AA   19 (79.16) 56 (84.84)    20 (95.23) 26 (89.65)    39 (86.66) 82 (86.31) ABCB1 rs2032582 A   27 55 1.63 (0.85-3.11) 0.14  17 23 1.0 (0.36-2.76) 1.0  44 78 1.33 (0.79-2.24) 0.28  1.40 (0.81-2.41) 0.222 (G2677T/A) C Functional  21 77    25 35    46 112  AA   9 (37.50) 14 (21.21)    2 (9.52) 2 (6.90)    11 (24.44) 16 (16.84)  AC   9 (37.50) 27 (40.90)    13 (61.90) 19 (65.51)    22 (48.88) 46 (48.42)  CC   6 (25.00) 25 (37.87)    6 (28.57) 8 (27.58)    12 (26.66) 33 (34.73) rs1128503 A   25 60 1.29 (0.67-2.48) 0.44  18 26 0.85 (0.34-2.10) 0.72  43 87 1.07 (0.64-1.78) 0.81  1.06 (0.62-1.81) 0.827 (C1236T) G Functional  23 70    24 32    47 103  AA   9 (37.50) 15 (22.72)    3 (14.28) 5 (17.24)    12 (26.66) 20 (21.05)  AG   7 (29.16) 31 (46.96)    12 (57.14) 16 (55.17)    19 (42.22) 47 (49.47)  GG   8 (33.33) 20 (30.30)    6 (28.57) 8 (27.58)    14 (31.11) 28 (29.47) Values are represented as numbers (percentage) or numbers (95% CI). CYP3A5 rs776746 (6986G>A) is an alternative splice variant (missense) in the intron. CYP3A5 rs10264272 (14690G>A) is an alternative splice variant (truncation) in the exon. CYP3A5 rs41303343 has a T insertion in 27132, causing a frameshift. CYP3A4 rs2740574 occurs in the Flanking 5’-UTR region (-392A>G). ABCB1 rs2032582 is a non-synonymous change in the exon (2677A>G). ABCB1 rs1128503 is a synonymous change in the exon (1236A>G). The shown p-value is the individual p-value of the genetic component of the logistic regression model: †Values derived from a logistic regression model adjusted for principal components. ‡Values derived from a logistic regression model adjusted for principal components, age, and tumor type. N/A = Not applicable   80              SNP clusters are generated with GenomeStudio Data Analysis Software v2010 in a Cartesian plot with normalized intensity !"#$%&'!"##$#%$&(!"#$%&)$*' !"#$%&'!"'()$%)#)&(!"#$%&)+*' !"#$%&'!"%'*(**%*&(!"#$%&),*' !"#$%-'!")#%(+#%&(!"#$%-)./*' '0' 000000 0012310 000000 000000 000000 000000 000000 000000 000000 000000 000000 000000 00000' Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Figure 10  Clustering of Candidate SNPs: CYP3A5 and CYP3A4  81             There are two different probes for ABCB1 rs1045642 that failed. SNP clusters are generated with GenomeStudio Data Analysis Software v2010 in a Cartesian plot with normalized intensity   !"#"$%!"#$%#&'#()*#+,,-.( !"#"$%!"//#'&$%()*/#%+-.( !"#"$%!"/$0&+0#()1%0%&2.( %&% %&% !"#"$%!"/$0&+0#()1%0%&2.( Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Norm Intensity (A) N o rm  I n te n s it y  ( B ) Figure 11  Clustering of Candidate SNPs: ABCB1  82 Table 11 The Effects of Previously Identified Genes Involved in the Metabolism and Transportation Pathway of Vincristine     Wilms Tumor  Rhabdomyosarcoma  Combined Cohort  Combined Cohort Gene SNP rsID SNP location & Postulated SNP Function§  OR (95% CI)† P-value†  OR (95% CI)† P-value†  OR (95% CI)† P-value†  OR (95% CI)‡ P-value‡ ABCC1 rs35593 Intron, regulatory potential  2.53 (0.87-7.38) 0.0886  2.36 (0.71-7.86) 0.161  2.51 (1.15-5.50) 0.0210*  2.36 (1.05-5.30) 0.0367* ABCC2 rs7476245 intron  5.30 (1.15-24.3) 0.115  2.99 (0.37-23.8) 0.682  4.16 (1.25-13.8) 0.089  4.11 (1.20-14.1) 0.0244* ABCB1 rs6979885 intron  0.36 (0.13-0.96) 0.112  0.58 (0.19-1.75) 0.722  0.48 (0.24-0.96) 0.142  0.43 (0.21-0.90) 0.0262* ABCC10 rs1214763 flanking_5UTR  0.44 (0.11-1.67) 0.226  0.28 (0.06-1.42) 0.124  0.37 (0.13-1.02) 0.0551  0.37 (0.13-1.02) 0.0564 ABCC3 rs11651060 flanking_5UTR  2.26 (1.02-5.03) 0.145  1.08 (0.49-2.37) 0.941  1.63 (0.94-2.83) 0.258  1.65 (0.93-2.90) 0.0816 CYP3A5 rs776746 intron  1.36 (0.46-4.01) 0.576  0.50 (0.13-1.93) 0.313  0.79 (0.35-1.79) 0.568  0.86 (0.37-2.0) 0.718 RALBP1 rs3322 UTR  0.77 (0.24-2.50) 0.668  0.89 (0.22-3.52) 0.862  0.88 (0.37-2.10) 0.776  0.73 (0.29-1.82) 0.502 CYP3A4 rs2740574 flanking_5UTR  1.79 (0.47-6.77) 0.593  0.49 (0.04-5.75) 0.874  1.06 (0.34-3.25) 0.824  1.42 (0.45-4.48) 0.548 The SNP with the lowest p-value in each candidate gene (in the logistic regression model adjusted for ancestry, or adjusted for ancestry, age, and tumor type) is listed. §Postulated SNP function on gene from National Institute of Environmental Health Sciences (NIEHS) (XU 2009). The shown p-value is the individual p-value of the genetic component of the logistic regression model. †Values derived from a logistic regression model adjusted for principal components. ‡Values derived from a logistic regression model adjusted for principal components, age, and tumor type. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing.  83 rs6979885 (OR 0.43, 95% CI: 0.21-0.90, p-value = 0.0262) were found to be significant prior to correction of multiple testing in the combined cohort adjusted for ancestry, age, and tumor type (Table 11). All three SNPs are located in the intronic regions of the respective genes, and only ABCC1 was postulated to have a regulatory function. No SNPs were found to be significant (prior to multiple testing correction) in the remaining candidate genes, RALBP1, ABCC10, ABCC3, CYP3A5, and CYP3A4 (Table 11).  84 Chapter  5: Discussion As one of the largest paediatric patient cohorts assessed for VIN to date, this study is the first to investigate the pharmacogenomics of VIN with a broad panel of 315 genes that are involved in the absorption, distribution, metabolism, and excretion of prescription drugs. This study has also been the first to investigate the importance of genetic variation in the development of VIN in paediatric patients with Wilms tumor and rhabdomyosarcoma. Previous studies have only been conducted in paediatric patients with ALL, one adult study in multiple myeloma patients. Additionally, at this point in time, this research paper is the only study that has established a phenotyping process in which potential confounding effects of concomitant drugs (e.g.: steroids, CYP3A inhibitors) are taken into consideration.  5.1 The Incidence of Vincristine-Induced Neurotoxicity In this study, VIN occurred in 26% of patients with Wilms tumor and 42% of patients with rhabdomyosarcoma. This finding is in agreement with the different incidence rates of VIN that has been reported for each of the tumor types in the literature. For example, the frequency of VIN has been observed to range from 4 to 28% in ALL patients (PORTER et al. 2009) and has been reported to occur in 42% of patients with rhabdomyosarcoma (ARNDT et al. 1997). Since clinical trials have mainly focused on hepatic or haematologic toxicities from drug-combination use within the protocol, no studies have specifically investigated the incidence of VIN in patients with Wilms tumor (GREEN et al. 1998; MORGAN et al. 1988). However, in this study, the incidence of VIN observed in Wilms tumor patients falls within the range of toxicity that has been reported for ALL, the cancer type for which most of the VIN studies have been conducted.  85 5.2 Clinical Factors 5.2.1 Age A significant association was observed between VIN and age at time of diagnosis, where there was a higher incidence of VIN in older patients. This age-related association is in agreement with previous observations about VIN (TABORI et al. 2005), including the aforementioned COG study that was temporarily and partially suspended due to VIN (RASSEKH 2010). In this COG study, suspension of patient accrual was only applicable to children in the age group of 13-years or older for which an increased incidence of VIN was observed, whereas patients less than 13-years of age were continued to be enrolled into the study (RASSEKH 2010). Furthermore, confirming previous literature reports, the average age of the patients diagnosed with Wilms tumor was younger than those with diagnosed with rhabdomyosarcoma (BRESLOW et al. 1988; WEXLER LH 2002). Therefore, the increased incidence of VIN in rhabdomyosarcoma patients may be related to the older age of patients with this tumor type as compared to patients with Wilms tumor. Age is thus an important factor that potentially influenced the significant difference in incidence rates of VIN observed between tumor types. The diagnosis of VIN may be a key factor contributing to the association between age and VIN in paediatric patients, which also affects the intervention and the treatment provided. Older children allow for a more objective confirmation of VIN over younger children, by virtue of their ability to verbally express having VIN symptoms (i.e.: muscle weakness), or by physically showing signs of VIN. For example, foot drop is more easily assessed in a child that can walk as compared to one that crawls. Therefore, the increased incidence of VIN observed for older patients may at least be partially attributed to an  86 improved diagnosis of VIN in older children. In agreement with this, rhabdomyosarcoma cases in this study were significantly more likely to experience muscle weakness, which is one form of VIN that cannot be easily assessed by observation alone. Therefore, rhabdomyosarcoma patients may have been able to better communicate or confirm their symptoms of muscle weakness, as well as cooperate with resistance tests, whereas such specifics are more difficult to elicit from a younger age group of children. Patient age may also have affected the interventions used for the treatment of VIN. In this study, rhabdomyosarcoma patients required more assistive devices and rehabilitation therapy, whereas patients with Wilms tumor were more likely to utilize narcotic and non- narcotic analgesics to alleviate pain. Due to ease of cooperation, children in the 8-year-old age group (patients with rhabdomyosarcoma) may have been more likely to be prescribed with splints and AFOs by their physiotherapists as compared to the 4-year-old age group (patients with Wilms tumor). Furthermore, older children may be more likely to wear such assistive devices on a regular basis, and are easier to follow and observe for improvement over time, resulting in an increased use of this intervention in older children. At this point in time, there is no rationale that may explain for the observed increased use of analgesics in Wilms tumor patients. Thus, this tentatively suggests that the most severe form of the neuropathy experienced in this tumor group may be more sensory than motor, however, surveillance studies in these patients will be required to determine whether this is a true observation or merely an artifact. Furthermore, in addition to the age-related differences in the ability to assess VIN, the development of neuropathy itself may be affected by age. Demyelination is a hallmark of neurodegenerative diseases (i.e.: CMT), and has also been observed to increase with age (as demonstrated by its progressive development in adults)  87 (GIDDING et al. 1999). Given the potential influence of age on the diagnosis and incidence of VIN, the adjustment for age was included as a clinical covariate in the logistic regression analysis of genetic associations, in order to avoid potential confounding of genetic associations by patient age.  5.2.2 Tumor Type In addition to age at time of diagnosis, the differences in chemotherapy protocols between Wilms tumor and rhabdomyosarcoma patients resulted in significant differences in the total number of vincristine doses received, cumulative dose, and treatment length between (Table 4). Therefore, this study adjusted for tumor type to account for the association(s) that may have been present in one of the tumour types only. In evaluating the differences between the two tumor types, this study showed that even though the mean number of doses withheld per patient was significantly higher in the rhabdomyosarcoma patients, there was no significant difference in the number of patients who had vincristine doses withheld or reduced due to VIN between the two tumor groups (Table 4). Since rhabdomyosarcoma patients were treated with vincristine for a longer time period, this observation likely resulted from the increased opportunities for additional vincristine-specific interventions in these patients who already required interventions earlier in their treatment. More importantly, however, was the fact that these differences were independent from and did not affect the severity and frequency of each type of VIN. Furthermore, the mean onset to the most severe VIN event between tumor types occurred at similar time points. Occurring at approximately 11.5 and 10.5 weeks for patients with Wilms tumor and rhabdomyosarcoma, respectively (Table 4), the mean onset of VIN in this study was comparable to observations  88 by other studies at approximately 12 doses (ARNDT et al. 1997). These findings thus suggests that the frequency of vincristine doses administered, which is the same at one dose per week during the first ten weeks of treatment for both tumor types investigated in this study, is a much more important contributing factor to VIN than the overall cumulative dose, which is confirmed with Figure 3.  5.2.3 Ancestry  Ethnicity has previously been reported to affect the occurrence of VIN (RENBARGER et al. 2008). The ethnicities of both case and control patients in this cohort were distributed between the three major ancestries of European, Asian, and African. This indicated that an adjustment for ancestry was required in this study. Furthermore, in the six individuals of African descent, there was only one individual (one-quarter African) who experienced VIN. No VIN was experienced in the five other individuals who had at least two or more of their grandparents with African descent. Although there are few individuals of African descent in this study, these findings are in agreement with incidence of VIN in the African population the literature. Conversely, this observation was not seen in the European or Asian population.  5.2.4 Concomitant Medications: Steroids and CYP3A4 Inhibitors As the more aggressive malignancy, rhabdomyosarcomas require a more intensive chemotherapy regimen than Wilms tumors. Consequently, these patients are more likely to experience nausea, requiring the use of steroids (serotonin antagonists such as dexamethasone) to enhance the effect and efficacy of antiemetic medications. Thus, although  89 steroids are not in the protocols for Wilms tumor and rhabdomyosarcoma patients, this higher utilization of steroids by this tumor group may be a contributing factor in the increased incidence of muscle weakness in rhabdomyosarcoma cases as compared to Wilms tumor cases. In contrast to the observed difference in steroid use between tumor types, there was no significant difference in steroid use between the case and control populations. With steroid and CYP3A4 inhibitor use shown to be insignificant contributors in influencing outcomes in both patient populations in this study, no adjustment for concomitant medications was required.  5.3 Vincristine-Induced Neurotoxicity: Types of Neurotoxicity and Severity Similarly, the types of neurotoxicity and their progression in this study is similar to that reported in the literature (GIDDING et al. 1999). PN was the most frequently observed form of neuropathy, followed by AN and CNS effects. In both cohorts, the initial development of VIN was typically sensory in nature (jaw pain, sensory loss, and paresthesias), whereas severe motor neuropathy problems developed more gradually over time (MCCARTHY and SKILLINGS 1992). Foot drop, muscle weakness, and both limb and joint pain were the most frequent complaints of VIN, followed by symptoms of ptosis, jaw pain, and vocal cord paralysis (Table 5). In this active surveillance-based assessment of VIN, 4.29% of the children with VIN complained of some form of vocal cord paralysis and hoarseness (Table 5). The observation of vocal cord paralysis and hoarseness is much higher than the literature review-based incidence rate of 1.36% and may be explained by under diagnosis of these adverse effects of vincristine (KURUVILLA et al. 2009). For example, hoarseness may be attributed to an upper tract infection, croup, or laryngitis, and may not be  90 considered as the starting point of vincristine-induced vocal cord paralysis (KURUVILLA et al. 2009). This suggests that vincristine-induced vocal cord paralysis may be underrepresented and may be up to three times more prevalent than previously reported. Close monitoring would enable early detection and prevent the life-threatening outcomes associated with vocal cord paralysis.  5.4 Novel Genes and Genetic Variants Potentially Associated with Vincristine- Induced Neurotoxicity  Out of the 4,536 SNPs capturing genetic variation in 315 genes, this study identified six novel candidate SNPs that may predict the development of CTCAE grades two and three VIN in paediatric patients with Wilms tumor or rhabdomyosarcoma. With results independently associated in the two tumor cohorts, this evidence suggests an association between VIN and PON1 rs854549 (paranoxonase 1), ABCA4 rs3789433 and rs549848 (ATP- binding cassette, sub-family B, member 4, multidrug resistance protein 3), SLCO1C1 rs10770704 (solute carrier organic anion transporter family, member 1C1), ABCG1 rs221948 (ATP-binding cassette, subfamily G, member 1), and CYP51A1 rs7797834 (cytochrome P450, family 51, subfamily A, polypeptide 1, also known as lanosterol or 14-α-demethylase). None of the five candidate genes, as well as PPP1R9A, have been previously associated with vincristine or VIN. I examined the current knowledge of the biological function of these genes to generate supporting hypotheses about their prospective role in the development of VIN and the potential effect of genetic variation in these genes in modulating individual susceptibility to VIN.  Highlighted in Figure 12 is a summary of how these identified genes could be integrated into the current knowledge base.  91            This modified diagram by PharmGKB illustrates the potential involvement of the novel genes in the transportation ( ) and metabolism ( ) of vincristine as directed by the black arrows. Vincristine is metabolized into its product(s) ( ). The modified components are denoted with an asterisk. This original figure was modified with permission from PharmGKB and Stanford University.  !"#$%&'(#")* +,-./0/)* -12/)* 34+1/+/)* 05+6/)* 05+07)* Figure 12  The Theorized Role of the Novel Genes in the Metabolism and Biotransformation of Vincristine  92 5.4.1 PON1 (Paraoxonase 1) and PPP1R9A (Neurabin 1) PON1 rs854549 is a non-coding genetic variant located in the intergenic region between PON1 and a second flanking gene, PPP1R9A (protein phosphatase 1, regulatory (inhibitor) subunit 9A). In addition to being in strong LD with both genes, PON1 rs854549 is also located in close proximity to both genes; it is 403 base pairs beyond the 3’-untranslated region (UTR) region of PPP1R9A, and 858 base pairs beyond the 3’-UTR region of PON1. The gene products of PON1 (26 kb, 9 exons) and PPP1R9A (386.5 kb, 15 exons) are paraonxonase 1 (PON1) and Neurabin 1 (Nb1, neural tissue specific F-actin-binding protein I), respectively. Comprised of 355 amino acids, PON1 is a hydrolytic enzyme that is produced in the liver and is transported in the serum while being bound with high-density lipoprotein (HDL) (HASSETT et al. 1991). As part of the cellular antioxidant defence system, PON1 is known for its role in the prevention of xenobiotic-generated oxidative stress, lipid peroxidation, and in the reduction of cardiovascular disease, likely due to its association with HDL (AVIRAM et al. 2000). PON1 is known as a detoxifier of exogenous substrates, including highly neurotoxic pesticides such as organophosphates (OP) and oxon metabolites, as well as biochemical warfare nerve agents such as sarin and soman (DAVIES et al. 1996). Interestingly, the aforementioned xenobiotic chemicals have demonstrated to be capable of affecting neurobehavioural activities, including motor coordination and balance, movement, and paralysis, as well as causing neurotoxicity, specifically motor and sensory PN presenting as muscle wasting, areflexia, and cranial nerve palsies (MISRA and KALITA 2009). In addition to eliciting similar neurotoxic effects to vincristine, overexposure to these industrial pesticides (e.g.: farming exposure, trace contaminations in food) have shown to cause similar  93 pathological outcomes to vincristine; both cause chemically-induced axonal degeneration and the inhibition of neurite outgrowth (COSTA et al. 2005; LI et al. 2003). This mechanism of action is potentially mediated by the lipid peroxidation of nerve membranes, which lead to peripheral nerve ischemia and hypoxia (ANDRONE et al. 2000) and ultimately axonal degeneration and neurite outgrowth inhibition (COSTA et al. 2005; LI et al. 2003). It has been speculated that decreased PON1 activity hinders the body’s ability to protect against environmental toxins and may therefore contribute to an increased susceptibility to neurotoxicity (ABBOTT et al. 1995; COSTA et al. 2005; LI et al. 2003; MACKNESS et al. 1998). Alternatively, two studies have reported decreased non-specific esterase activity in vincristine-treated animals (KOZIK and MAZIARZ 1983; KOZIK and SZCZECH 1983). Since PON1 has esterase-like and peroxidase-like activities, this observation may therefore be related to an effect of vincristine on PON1.  Interestingly, this theory is being tested in the context of the development of neuropathy alongside other disorders. PON1 has also been associated with an increased susceptibility to PN in diabetic patients (ABBOTT et al. 1995), as well as sporadic amyotrophic lateral sclerosis (ALS), a neurodegenerative disease that primary affects the motor neurons (CRONIN et al. 2007; SLOWIK et al. 2006). In comparison to Type 1 and Type 2 diabetes patients without neuropathy, PON1 activity has been observed to be dramatically lower in the corresponding Type 1 and Type 2 diabetes patients who have sensory neuropathy and decreased reflexes (MACKNESS et al. 1998). This has prompted speculation that genetic variation in PON1 could be key in influencing the development of diabetic neuropathy (MACKNESS et al. 1998). My study is the first to suggest the involvement of PON1 with VIN, and additionally, the first to link motor neuropathy with PON1. This  94 relationship between PON1 and VIN is supported by studies that show the improvement of diabetic neuropathy with lipoic acid, which is an inhibitor of peroxidation and reactive oxygen species (ANDRONE et al. 2000), as well as a putative neuroprotectant and has been shown to improve chemotherapy-induced neuropathy (MELLI et al. 2008). In vivo studies on the activity of paraoxonase in mice has demonstrated that in comparison to those fed a regular diet, there was a trend of increased paraoxonase activity when mice were provided a diet rich in α-lipoic acid (YI and MAEDA 2006). In 2006, a multicenter, randomized, double- blind, placebo-controlled trial, demonstrated the improvement of neuropathic symptoms when patients with diabetic polyneuropathy were provided with oral doses of α-lipoic acid (ZIEGLER et al. 2006).  In addition to the theory that lipid peroxidation of nerve membranes is an indirect cause of diabetic neuropathy (ANDRONE et al. 2000), the findings in my study indicates that a similar mechanism may be involved in the development of VIN. Therefore, it can be inferred that lipoic acid may also have therapeutic potential in acting as a useful protective agent in the treatment of VIN. Similar to the studies linking PON1 and diabetes-induced neuropathy, some studies have implicated genetic variants in PON1 as significant contributors to susceptibility to ALS, alongside mutated forms of superoxide dismutase that results in additional loss of antioxidant function (CRONIN et al. 2007; SLOWIK et al. 2006). As a progressive neurodegenerative disease, the first few symptoms of ALS present similarly to vincristine-induced PN, including: foot drop, the inability to conduct fine motor tasks such as buttoning or writing, slurred speech, difficulty swallowing (dysphagia), and the involvement of the intercostal muscles in prompting respiratory distress.  95  Given the protective function of PON1 as an antioxidant, the majority of PON1 studies have investigated its role in the areas of toxicology and cardiology. However, there is also evidence supporting a role of PON1 in the metabolism and biotransformation of drugs (COSTA et al. 2005). PON1 has been shown to have the capability to hydrolyze substrates (aromatic esters such as methyl phenylacetate side groups, aromatic and aliphatic lactones, and cyclic carbonates) and metabolize a variety of drugs, including prulifloxacin and clopidogrel (BOUMAN et al. 2011). A methyl phenylacetate group is present in the chemical structure of clopidogrel, and intriguingly, also in vincristine (Figure 13). Thus, by identifying a potential PON1-mediated cleavage site in vincristine, it is possible that vincristine is metabolized by PON1. Its relative contribution to the metabolism of vincristine and whether PON1 functions independently from CYP3A5-mediated metabolism is currently unknown, and requires further investigation.  PON1 is located in a cluster of genes encoding for the PON proteins, where PON2 and PON3 are clustered upstream and in close proximity to PON1. There is 60% sequence homology between all three PON proteins (PRIMO-PARMO et al. 1996). Like PON1, PON2 and PON3 also have lactonase activity, as demonstrated by their ability to metabolize statin lactones (e.g.: simvastatin, lovastatin) (DRAGANOV and LA DU 2004). Although these two enzymes are speculated to be also involved in drug metabolism (COSTA et al. 2005), their paraoxonase activity has been suggested to be lower than PON1 (DRAGANOV and LA DU 2004). Including PON1 rs854549, the ADME-Tox panel identified six PON1 and an additional six PON2/PON3 tagging SNPs that were significantly associated with VIN at a level of 0.05 (Appendix A.3). Although the linkage between PON1 rs854549 and these 12  96 Figure 13 Chemical Structures of Methyl Phenylacetate, Vincristine, and Clopidogrel    PON1 is known to hydrolyze methyl phenylacetate as well as clopidogrel. Vincristine also contains a methyl phenylacetate structure. Therefore, it is possible that PON1 metabolizes vincristine. Methyl phenylacetate Vincristine Methyl phenylacetate Clopidogrel Methyl phenylacetate  97 SNPs was not very strong, it is remarkable that given the small size of PON1, PON2 (30 kb, 9 exons), and PON3 (36.5 kb, 9 exons), there was a relatively large number of polymorphisms within this PON gene cluster that were associated with VIN. Therefore, it is possible that PON2 and PON3 may also have supporting roles in protecting against VIN.  With evidence supporting that functional PON1 has neuroprotective properties that potentially prevent xenobiotic-induced axonal degeneration and may also be involved in the metabolism of vincristine, I speculate that PON1 may be involved in the protection against vincristine-induced iatrogenic axonal degeneration and neurotoxicity. A greater understanding of the environmental and genetic factors affecting the expression levels and functionality of PON1 and their relevance in the context of the development of VIN will be key in determining if the loss of neuroprotective properties (through PON1) can weaken the body’s response against chronic exposure to environmental or medical toxins such as vincristine.  Interestingly, Nb1, the protein encoded by PPP1R9A, is also known to be involved in the function of neuronal cells (NAKABAYASHI et al. 2004). Nb1 is a F-actin binding protein that is mostly concentrated in brain and synapse of developed neurons, whereas with lower levels of Nb1 mRNA are present in other tissues (NAKABAYASHI et al. 2004). By binding to actin, Nb1 moves from the neuron cell body to localize in the synapses of both rudimentary and established nerve endings. Postulated to improve synapse formation and function, Nb1 is theorized to be involved in neurite (axon and dendrite) outgrowth and regeneration (BURNETT et al. 1998). Thus, Nb1 could play a role in repairing the damage caused by iatrogenic toxins such as vincristine or neurotoxic chemicals (e.g.: OP, nerve agents) that shorten neurite length and limit growth in a dose-dependent manner (HAYAKAWA et al. 1994; HOUI et al.  98 1993). Genetic variations in PPP1R9A could thus potentially increase an individual’s susceptibility to VIN, specifically by affecting the body’s ability to regenerate axons and neurons from chemically-induced injury. Interestingly, the production of Nb1 is stimulated by NGFs (HIER et al. 1972) that are currently being tested as neurotrophic agents and neuroprotectants against chemotherapy-induced neurotoxicity, such as IGF-I (CONTRERAS et al. 1997). Since NGFs promote the survival, function, and development of neurons, one could infer that Nb1 functions to promote the same outcome.  With the most significant SNP in this study located in close proximity to both the PPP1R9A and PON1 genes, it is possible that genetic variants in either gene that are linked to PON1 rs854549 may contribute to VIN. There are limited genetic studies on PPP1R9A, and to my knowledge, no functional SNPs have been identified in this gene. However, as a fairly large gene (386.5 kb) encompassing 15 exons, it is likely that further genetic studies would reveal functional variants in PPP1R9A. Since PPP1R9A does not have any known drug ADME function or involvement in general drug-related toxicity, this gene was not comprehensively covered in this study with tagging SNPs and will require further studies to determine whether genetic variants in PPP1R9A are associated with VIN.  For PON1, four functional SNPs have been described: two non-synonymous SNPs (PON155 rs854560 and PON1192 rs662) in exons 3 and 6, respectively; as well as two SNPs in the promoter region of PON1 (PON1-108 rs705379, PON1-162 rs705381). PON155 rs854560 results in a codon change from leucine (TTG) to methionine (ATG), and PON1192 rs662 changes from glutamine (CAA) to arginine (CGA). Single SNP analyses have also shown PON1192 rs662 to have a greater effect on PON1 enzymatic activity as compared to PON155 rs854560, which appears to have an effect on both paraoxonase activity and protein levels.  99 This stronger effect of PON1192 rs662 on enzyme activity has been attributed to the location of the 192-residue in the active site of the protein. PON155L carriers have been shown to have a higher enzyme activity and higher PON1 levels than carriers of the PON155M/PON155M genotype (BROPHY et al. 2000). The PON155 rs854560 variant appears to mainly affect levels of PON1 levels over activity in the serum (GARIN et al. 1997). This effect by PON155 rs854560 has been related to this SNP’s strong LD to variants in the promoter region of PON1, specifically PON1-108 rs705379 (BROPHY et al. 2001). However, previous work has not fully established the individual effect that these SNPs may have on function, thus, further studies are needed to verify the relative contributory effect of these non-synonymous SNPs, the genetic variants in the PON1 promoter region, or by other functional mutations in LD on modified enzymatic activity and levels of PON1.  Since the ADME-Tox genotyping panel was developed using tagging SNPs that covered variation in the targeted genes, follow-up LD and imputation studies were performed to provide a more comprehensive understanding of the SNPs in these genes. LD studies were performed to determine if PON1 rs854549 was in high LD with other SNPs in PON1 or PPP1R9A that were also genotyped as part of the ADME-Tox panel. In addition to identifying linkage blocks, there was a particular focus on determining the linkage between PON1 rs854549 and the four known functional variants of PON1. Of the four SNPs known to be involved in altering the activity of PON1, only PON155 rs854560 was present in the ADME-Tox panel. Indeed, PON1 rs854549 was in strong LD with PON155 rs854560 within the population in this study at r2 = 0.559 (Table 8), as well as within the Hapmap population (r2 = 0.5). Although the individual cluster result showed that the high throughput genotyping assay was not optimal for PON1 rs854560, this SNP was still significantly associated with  100 VIN in the combined cohort in the analysis adjusted for age, tumor type, and ancestry. The strength of this association may increase further when genotyped separately on a more robust genotyping platform such as a Taqman® assay (Applied Biosystems, Foster City, CA, USA). Of the three other functional PON1 variants not included in the ADME-Tox panel, PON1192 rs662 is not in high LD (r2 = 0.1) with PON1 rs854549 in the Hapmap population (Table 8), whereas the LD for the two other functional variants in the promoter region (PON1-108 and PON1-162) were not characterized in the Hapmap database and are currently unknown.  Imputation studies are in silico experiments that infer genotypes for SNPs that were not directly genotyped in the study patients, and were used to increase the number of SNPs tested in specific gene regions, thereby obtaining a more comprehensive genotype coverage of the gene. This imputation analysis of the PON1 and PPP1R9A cluster region using LD data from the Hapmap Project and the Human Build Project revealed that PON1 rs854549 was still the most significant SNP, when controlling for age, tumor type, and ancestry (Appendix A.6). Again, due to their absence from the Hapmap reference panel, the functional promoter SNPs (PON1-108 rs705379, PON1-162 rs705381) were not imputed. Additionally, the original PON155 rs854560 genotype could not be compared against the imputed genotype, given the ambiguity of the alleles (A/T) in this SNP and the close MAF value to 0.50. This ambiguity prevented the SNP from being aligned to the same strand convention when combining the study and the Hapmap reference panels; therefore, this SNP was excluded from the imputation analysis. As expected, given the low LD with PON1 rs854549, the imputation-based genetic association p-value for PON1192 rs662 was p-value = 0.28, confirming the lack of association of this functional SNP with VIN. PPP1R9A rs705377, a SNP in the PPP1R9A gene that was included in the genotyping panel as a tagging SNP for  101 PON1, was in LD with PON1 rs854549 within the study population at 0.773 (Table 8). In the imputed studies, 11 SNPs in the intronic regions of PPP1R9A were significant at a level of p- value < 0.05 (prior to multiple testing correction), of which five SNPs were postulated to have a regulatory function.  At this stage, is not known as to whether the newly identified PON1 rs854549 is associated with a modified enzymatic activity or protein levels of PON1 or Nb1, or mRNA levels of PON1 and PPP1R9A. Although it is located in the intergenic region at the 3’-end of two closely located genes and is not transcribed, a functional effect of PON1 rs854549 itself cannot be excluded, as it could control PON1 and PPP1R9A as nearby genes. For example, this SNP could be the binding site of regulatory proteins that can modulate the transcription termination for either gene. Given the propensity of DNA to form larger loops and coils, these effects of these regulatory elements are not limited to the 3’-end of the genes, but can also affect the 5’-end, which can affect mRNA transcription rates and protein production. To illustrate, a recent study selected the intronic and intergenic regions of transporters to show the regulatory effect of these genetic variations in the 3’ and 5’ regions on enhancers on drug-associated genes (KIM et al. 2011). In demonstrating an effect on mRNA expression and activity, this illuminates an example in which genetic variations in non-coding regions can affect drug efficacy and toxicity (KIM et al. 2011). An increasing number of studies have shown that gene function can be affected by SNPs in non-coding gene regions by affecting: the regulation of transcription activity (FISHER et al. 2000), mRNA stability and microRNA binding (MISHRA et al. 2007), as well as translation efficiency and protein folding (FUNG and GOTTESMAN 2009). Furthermore, PON1 rs854549 could be part of a region that is transcribed as an ncRNA (non-protein-coding RNA). Although there are limited studies, it is noted  102 ncRNAs facilitate normal development and physiology, and given this regulatory role, it can be inferred that damage to ncRNAs can cause disease (TAFT et al. 2010). Finally, an effect on enzymatic activity or protein levels could be caused by any SNP (a non-synonymous variant or a functional mutation with a regulatory effect) in PON1 or Nb1 that is in LD with PON1 rs854549, and has not been identified through this a priori selection of SNPs or limited imputation coverage. To locate such functional variants associated with PON1 rs854549, further studies to identify additional polymorphisms in PON1 and PPP1R9A and their LD with PON1 rs854549 are needed.  5.4.2 ABCA4, ABCG1, and CYP51A1 As a major component of the myelin sheath, cholesterol is a key component in maintaining proper nerve function within the nervous system (SAHER et al.). Lipid generation and transport is fundamental in maintaining the structure, function and survival of cells. Consequently, the cholesterol and phosphatidylcholine content in myelin sheath is tightly regulated, and indicates the potential importance of regulating cholesterol levels in the context of VIN. Indeed, the enzyme encoded by one of the genes identified to be associated with VIN in this study, CYP51A1 (22.4 kb), is involved in sterol biosynthesis. In demethylating lanosterol, this enzyme performs the first of many conversion steps to produce cholesterol (Figure 14). Desmosterol is one of the metabolites produced by the conversion of lanosterol to brain cholesterol (by CYP51A1), and is not consumed in the diet (BAE and PAIK 1997; LUTJOHANN et al. 2002). As the immediate precursor to the cholesterol in the brain, desmosterol is found in the PNS and in myelin (BOURRE et al. 1990). Furthermore, nerve regeneration requires local production of cholesterol, and since CYP51A1 is located in the  103 Figure 14 Cholesterol Biosynthesis  CYP51A1 is an enzyme involved in the conversion of lanosterol to desmosterol and cholesterol   lanosterol 4,4-dimethyl-5!- cholesta-8,24-dien-36-ol desmosterol 24,25-dihydrosterols cholesterol CYP51A1  104 sciatic nerve, it is reasonable to hypothesize that CYP51A1 may be responsible for providing axons with the cholesterol needed to myelinate and remyelinate axons (JUREVICS et al. 1998). It was only recently that CYP51A1 was associated with in developmental myelination and remyelination after injury (SONG 2010). Therefore, due to this requirement of local cholesterol production, I hypothesize that genetic variants in CYP51A1 affect normal CYP51A1 function, and that non-functional or decreased activity of CYP51A1 may consequently affect cholesterol homeostasis and regulation, and well as compromise the regeneration of myelin as a protective mechanism against iatrogenic-induced demyelination. Perhaps, in addition to age-related demyelination, defects to this lipid-mediated process are the reason why older teens and adults experience more severe symptoms of VIN (TABORI et al. 2005). Additionally, as part of the cytochrome P450 family of key enzymes involved in the metabolism of numerous drugs, CYP51A1 could also be a possible metabolizer of vincristine. The SNP in CYP51A1 identified in this study is synonymous for the amino acid histidine (at position 453) and is a possible exonic splicing enhancer or silencer (XU AND TAYLOR 2009). Functional or regulatory effects have also been described for synonymous gene variants, such as altering mRNA splicing or translation (XU AND TAYLOR 2009). Alternatively, like PON1 rs854549, this candidate SNP could also be in LD with another variant with a functional effect located elsewhere in the gene. Interestingly, also two other candidate genes identified in this study, ABCA4 (128.3 kb) and ABCG1 (78.0 kb) are potentially involved in lipid and cholesterol biotransformation. Both are a part of the lipid transporter family, where ABCA4 has been suggested to have flippase activity (POLLOCK and CALLAGHAN 2011), and ABCG1 has been shown to function  105 as a lipid exporter (KOBAYASHI et al. 2006). The functions of both proteins are involved in the direct mediation of lipid trafficking: flippases in biological membranes are essential for the maintenance of transbilayer lipid asymmetry, and lipid importers or exporters are crucial for the maintenance of lipid homeostasis throughout the body. Encompassing a genomic region of 128 kb, ABCA4 is a large gene that does not contain any strong LD blocks, and has been extensively covered in the genotyping panel used in this study. Besides ABCA4 rs3789433 and rs549848, 14 additional SNPs in this gene were associated with VIN with a p-value of <0.05 (Supplementary Table 1). ABCA4 is postulated to function as a flippase that maintains the proportion of phospholipids and lipids in cells (POLLOCK and CALLAGHAN 2011). Loss of function of ABCA4 in retinal cells causes Stargardt disease, an autosomal recessive disease causing macular degeneration. Predominantly found and studied in retinal cells, this translocase is specifically localized to the outer segment of the disk edges of rods and cones that are embedded in the photoreceptor cells, and is theorized in its function as an inward-directed retinoid flippase of N- retinylidene-phosphatidylethanolamine (NrPE). With the majority of studies being conducted in retinal cells, and with a clear outcome of Stargardt disease with the loss of ABCA4 function, it had been presumed that ABCA4 was mainly expressed in retinal cells. However, the location of ABCA4 has not been exhaustively studied. Indeed, expressed sequence tags (short segments of the cDNA that are utilized to identify the transcriptionally active expression of certain genes) experiments suggest that while ABCA4 is highly expressed in the eye, the gene is also expressed in the muscle and brain as shown in the NCBI EST database (http://www.ncbi.nlm.nih.gov/dbEST). Furthermore, ABCA4 was recently shown to be expressed in chorioid plexus membranes, which can affect the distribution of drugs in the  106 blood-brain barrier, blood-cerebrospinal fluid barrier, and CSF conditions (BHONGSATIERN et al. 2005). The function of ABCA4 is thought to be similar to that of ABCB4, which is also a lipid transporter and phosphatidylcholine flippase. Deficiency in ABCB4 has been implicated with severe liver disease (progressive familial intrahepatic cholestasis), whereas remarkably, an over-expression study in ABCB4 transgenic mice resulted in the development of PN in the hind legs (SMIT et al. 1996). The authors of this study postulate the observed PN to be caused by the addition of extra ABCB4 protein in the highly organized myelin sheaths. Disrupting the regulation of phosphatidylcholine in the outer leaflet of the myelin sheaths also disturbs the highly organized structure and composition of protein and high lipid content in myelin. As the potential start of the demyelination process, this may be exacerbated with exposure to vincristine. Further studies are required to assess whether ABCA4 is also expressed in the myelin sheaths, and whether a similar functional outcome of VIN will be observed with the overexpression of ABCA4. Additionally, ABCA4 may affect the development of VIN in an alternative process. By testing ABCA4 activity using an ATP-hydrolysis assay across several ligands, including retinoids and all-trans retinal (which have similar structures to NrPE), it has been suggested that ABCA4 mediates retinoid transport (SUN et al. 1999). Retinoids, including retinoid acid, are important signaling molecules that facilitate the survival and function of neurons, as well as regulate neurite length, growth, and number (LATASA and COSGAYA 2011). Furthermore, retinoids are considered to be integral regulators of myelination, and are also theorized to regulate Schwann cell physiology in both development and in pathological conditions (degeneration and remyelination) (LATASA and COSGAYA 2011). This regulation is important,  107 given that patients with motor neuron disease (i.e.: ALS) have a defect leading to lower in retinoid levels; and similarly, withholding retinoids in adult rats causes them to develop motor neuron diseases (CORCORAN et al. 2002). This suggests that tight regulation of retinoids is important for normal neuron function. In the context of VIN, I hypothesize that nucleotide variations in ABCA4 may lead to affect the susceptibility to VIN. Similar to ABCA4, also ABCG1 has been shown to be involved in cholesterol homeostasis and transport (KOBAYASHI et al. 2006), as well as the secretion and efflux of lipids and sphingomyelin, a component of the myelin sheath (KOBAYASHI et al. 2006). Furthermore, AGCG1 is involved in the secretion and regulation of the removal of cholesterol and phospholipids from peripheral cells (KOBAYASHI et al. 2006) and is found in high abundance in neurons (KIM et al. 2008), suggesting a potential importance in neuronal cell function. Finally, ABCA4 and ABCG1 are both part of the ABC transporter family, and therefore, in addition to being known as lipid transporters, it is possible that they also transport toxins, drugs, and metabolites, including vincristine and its metabolites. Besides the integral role that transporters play in the bioavailability of drugs and potentially resulting toxicity, the same genes could also be associated with limiting the effectiveness, activeness, functionality, and transportation of naturally protective agents (i.e.: neuroprotectants such as calpain and NGFs) or other damaging agents (i.e.: vincristine and OP). No studies so far have investigated such a transporter function of ABCA4 and ABCG1 in the context of vincristine, its metabolites, or neuroprotective agents.     108 5.4.3 SLCO1C1 SLCO1C1 (57.9 kb), also known as OATP14, is part of the organic anion transporter family. Mutations in the gene of this transmembrane receptor has been shown to affect the uptake of thyroid hormones (e.g.: thyroxine, tri-iodothyronine and reverse tri-iodothyronine) in the brain tissue and affects the development of endocrine disorders associated with thyroid dysfunction. Therefore, changes to the function of SLCO1C1 may potentially exacerbate the symptoms of neuropathy through hormonal imbalances that affect metabolic processes. Alternatively, it is possible that as a transporter, SLCO1C1 is involved in the transport of vincristine, vincristine metabolites, or neuroprotectants, and changes to SLCO1C1 could affect VIN through altered transport of these compounds. Like ABCA4 and ABCG1, no studies so far have assessed the capability of SLCO1C1 to transport vincristine, its metabolites, or neuroprotectants. Further studies are thus required to elucidate this potential mechanistic link between variants in this gene, as well as the other genes in affecting one’s susceptibility to VIN (Figure 4).  5.5 Previously Identified Polymorphisms and Candidate Genes Involved in the Biotransformation of Vincristine It is theorized that a decrease in expression or functionality of the transporters of vincristine (ABCB1, ABCC1, ABCC2, ABCC3, ABCC10, and RALBP1) and its metabolites or reduced activity of enzymes that metabolize vincristine (CYP3A4 and CYP3A5) may increase individual exposure to vincristine, potentially causing toxicity due to the inability of the body to tolerate such increased exposure. Based on this theory, several studies have tested genetic variants in these genes that are involved in the biotransformation of vincristine and  109 their effects on susceptibility to VIN. A positive association was found for two studies prior to multiple correction, where the association between the genotypes of CYP3A4 and CYP3A5 variants approached, but did not reach statistical significance (EGBELAKIN et al. 2011; PLASSCHAERT et al. 2004; APLENC et al. 2003). The extent of the positive association, as well as the methodology and study parameters utilized in the third study is unknown, since only an abstract is available (D. M. TE LOO 2009). The findings of the aforementioned studies have not been consistently reproduced; no evidence for a strong association of any of the previously reported candidate genes or variants with VIN was observed this present study, as well as two other studies (PLASSCHAERT et al. 2004; HARTMAN et al. 2010). Variable phenotyping approaches between studies, as well as and lack of consideration of the contributing effects of concomitant medications to VIN may have contributed to these discrepancies between studies. For example, one study utilized the Movement Assessment Battery for Children (m-ABC) to assess the effect of VIN on individual motor skills (HARTMAN et al. 2010). Similar to the TNS, which is considered to be the gold standard approach to comprehensively evaluate the subjective and objective components of neuropathy (ENGLAND et al. 2005), the m-ABC scale is an extensive grading tool that is too impractical to be employed in clinical practice. Cross-study comparisons of findings are hindered by studies that utilize different phenotyping definitions of VIN severity. Additionally, other studies may have overestimated the effect of non-functional CYP3A5 and reduced ABCB1 activity in the ALL population. For example, concomitant use of steroids (which induce CYP3A enzymes) in the study populations and CYP3A4 inhibitors were not taken into consideration. By adding the use of steroid and CYP3A4 inhibitors as part of a study’s comprehensive phenotyping profile, the effect of these parameters on the  110 significance of each genetic variant can be accounted, although this was not the case in this study, but may be of greater importance in other studies with cancer types that utilize steroids as a part of the chemotherapy regime. Furthermore, drug disposition and the metabolism of vincristine may not necessarily be solely dependent on CYP3A5 and the transporters that had been shown to be involved in these processes from previous pharmacokinetic and functional studies. These studies were not exhaustive, as not all potential transporters or metabolizing enzymes of vincristine were investigated. Therefore, the investigated enzymes may not necessarily be the major biological transporters or metabolizers of vincristine, and the relative activity and effects are unknown. In particular, my findings suggest that other metabolizing enzymes, such as PON1 or CYP51A1, and other transporters, such as ABCA4, ABCG1, and SLCO1C1 may significantly contribute to the metabolism and transportation of vincristine, respectively.  5.5.1 CYP3A5 and CYP3A4 Functional polymorphisms in CYP3A5 (CYP3A5*3, CYP3A5*6) and CYP3A4 (CYP3A4*1B) were not significantly associated with VIN in my study. However, a trend in the same direction as reported in the literature (EGBELAKIN et al. 2011) was observed for all of the priority SNPs. As previously reported, the wild type protective allele (A) of the CYP3A5 genetic variants was observed more frequently in controls, even though this difference was not significant. Although the effect of the protective allele was expected to be dominant and result in a fully functional CYP3A5 enzyme, seven carriers of this suggested protective allele still experienced VIN. These cases were re-evaluated to ensure that they were classified and graded appropriately. None of these patients were given high doses or  111 more frequent doses of vincristine, and none had a history of neurotoxicity, yet all of them had documented neurotoxic event(s) due to vincristine. My findings thus suggest that the potential effect and involvement of this SNP in individual susceptibility to VIN is limited and possibly not as influential as previously postulated. Without adjusting for steroid use in ALL patients, Egbelakin et al. demonstrated that all CTCAE grades (grades one to four) of VIN did not occur as frequently in CYP3A5 expressers as compared to non-expressers (p-value = 0.032 and p-value = 0.0007). Intriguingly, however, this association with VIN was not significant when considering only more severe CTCAE grades of three and four VIN (p- value = 0.11) (EGBELAKIN et al. 2011). Similarly, also in their study, 89% of CYP3A5 expressers still experienced neurotoxicity, indicating that the CYP3A5 genotype status was not the only contributor to VIN (EGBELAKIN et al. 2011). The same study showed that the duration of VIN was shorter in CYP3A5 expressors (p-value = 0.0002) (EGBELAKIN et al. 2011). CYP3A5 expressers experienced neurotoxicity during 16% of their treatment months on vincristine, whereas VIN symptoms were present during 27% of their treatment months on vincristine for the non-expressers (EGBELAKIN et al. 2011). The potential effect of CYP3A5 on the duration of neurotoxicity was not assessed in the study. Therefore, it cannot be excluded that CYP3A5 may be a stronger indicator of the duration of neurotoxicity, in comparison to its predictive potential for the susceptibility to VIN.  5.5.2 ABCB1 In the investigation of genetic variation in ABCB1, the effects of ABCB1 rs1045642 (C3435T), rs2032582 (G2677T), and rs1128503 (C1236T) on VIN have only been studied in  112 the context of constipation and specific motor dysfunction in ALL patients (HARTMAN et al. 2010; PLASSCHAERT et al. 2004). As a haplotype, ABCB1 rs2032582 and rs1045642 have been associated with marginal increases in the elimination time of vincristine (PLASSCHAERT et al. 2004), whereas no associations of single variants were observed. With the conclusion that the single SNPs of ABCB1 rs1128503 and rs2032582 were unlikely to have an effect on VIN, as a haplotype, they have been is associated with marginal increases in the elimination time of vincristine (PLASSCHAERT et al. 2004). In this study of Wilms tumor and rhabdomyosarcoma patients, there was no observed effect of ABCB1 rs2032582 or rs1128503 on VIN, which was extended to all forms of peripheral and autonomic neuropathy, as well as CNS effects. Since ABCB1 rs1128503 is in high LD with the other two genotyped variants, this genetic variant is unlikely to be associated with VIN. However, since a probe failure prevented this ABCB1 variant from being genotyped, no fulsome conclusions about the potential impact of ABCB1 rs1128503 on the susceptibility to VIN can be derived from this study. At the same time, an effect of ABCB1 rs1128503 on VIN as a haplotype or single SNP cannot be excluded. Further studies are required with a different SNP probe or using a separate genotyping method, such as Taqman® genotyping, to further elucidate the relevance of this variant in the context of VIN.  5.5.3 Other Candidate Genes In the ADME-Tox panel, there were 234 SNPs that captured the genetic variation in the eight candidate genes. There were only three transporters, for which at least one gene variant was associated with VIN at a p-value of < 0.05; ABCC1 (three variants), ABCC2 (one variant), and ABCB1 (one variant). Interestingly, two of the three ABCC1 variants are  113 postulated to have a regulatory function, which may have implications for the regulation of the expression and production of the ABCC1 transporter. However, more studies are needed further evaluate such regulatory effects. The investigated SNPs in the other candidate genes were not found to have an effect on the susceptibility to VIN. Even though this indicates that some of these polymorphisms do have an effect on the transportation of vincristine and may be involved in VIN, this effect does not appear to be strong. Whereas these genes have previously been implicated in the biotranformation of vincristine, none of these SNPs have been explicitly associated with VIN.  5.6 Inter-Ethnic Differences  Other than the functional genetic variants in CYP3A5 and ABCB1, there were no other known SNPs that showed inter-ethnic variability in the frequency of the minor risk allele that corresponded with the incidence of VIN (RENBARGER et al. 2008). This inter-ethnic difference in functional CYP3A5 was proven in a study that showed that VIN was more prevalent in the Caucasian population (RENBARGER et al. 2008). This observation was also supported by my study, where even though the number of patients is limited, it appeared that individuals with African ancestry were less likely to experience VIN. Since the functionality and MAF of CYP3A5 correlated well with the incidence of VIN and cancer cure rates between different ethnicities, it was theorized that VIN was mainly attributable to CYP3A5 (RENBARGER et al. 2008) and therefore, ethnicity was utilized (Caucasians and Africans) as a surrogate marker for CYP3A5 expression and genotype.  However, the lack of association of CYP3A5 polymorphisms with VIN observed in the study suggests that the ancestry-related differences in the incidence of VIN may be  114 caused by other SNPs with inter-ethnic allele frequency differences. Indeed, for the most strongly associated SNP in the study, PON1 rs854549, and the two variants in LD with this SNP, PON1 rs854560 and PPP1R9A rs705377, a higher frequency of the risk allele (minor allele) was observed in the Caucasian population, compared to the African and Asian populations, in the Hapmap populations. The reported lower incidence of VIN in African populations is thus in accordance with allele frequency differences in PON1/PPP1R9A variants between different ethnicities, providing an alternative explanation for ancestry- related differences in the occurrence of VIN. In order to assess whether the ethnic diversity of the patient population had an effect on the observed association of PON1/PPP1R9A with VIN in my study, the genetic association analysis was repeated including only patient of Caucasian origin. In this analysis, all three PON1/PPP1R9A variants were still found to be significantly associated with VIN. Although the ethnic difference in susceptibility to VIN has previously been ascribed to the differential MAF of CYP3A5, this provides new evidence supporting the role of PON1 and PPP1R9A in VIN.  5.7 Limitations Like other retrospective studies, this study is limited by its reliance on the written documents in patient medical records. There are some VIN symptoms that are not necessarily recorded in the charts, such as the loss of DTRs, which are symptoms that affect almost every patient treated with vincristine. However, because of the active surveillance approach used here, these limitations were minimized through close discussions with the family members of the patients to provide a more fulsome perspective of the experience of VIN.  115 Furthermore, the severity grading of VIN used in this study was strongly dependent on the use of interventions to treat VIN. A physician’s practice of providing interventions for VIN, and their decision to reduce or hold doses of vincristine, may vary between different physicians or different hospitals, which may have affected the identification of VIN cases and controls in the study. However, in order to avoid erroneous classification of a VIN cases as controls and to maximize phenotypic differences in VIN between cases and controls, patients with grade one VIN were not included in the study. To standardize the units when calculating the total cumulative dose, this study also utilized the accepted conversion of 30kg to 1m2 (FERRARI et al. 2003). It has been noted however, that body weight-based doses are approximately 30% less than the doses given by surface area (FERRARI et al. 2003). However, this dose conversion was also recommended and outlined in rhabdomyosarcoma protocols (MORGAN et al. 1988). Additionally, CMT and other hereditary neuropathy disorders were excluded from this patient cohort as they would alter the interpretation of the patient phenotype of type and level of neurotoxicity. However, depending on age, CMT can be difficult to diagnose and may not present at time of the study. Furthermore, CMT is very rare (global prevalence between 4.7 to 36 out of 100,000) and therefore is likely to not have an effect on the results of this study (VALLAT and FUNALOT 2010). Additionally, there were a few SNPs that were excluded due to genotyping failure. The reasons for failure are many, but are likely related to the specific properties of the DNA sequence in the target region, which may have resulted in suboptimal temperatures during the genotyping process, secondary structure formation, or interaction with genotyping probes for other SNPs. Although imputation studies were conducted to negate these issues, the extent of  116 the depth of imputation relies on the number of SNPs that were originally genotyped. This SNP panel had sufficient coverage to conduct imputation analysis with SNPs from the Hapmap reference panel, but more SNPs will be required to infer the genotypes of SNPs from the 1000 Genomes project (VIA et al. 2010). Lastly, it is much more likely for associations to be identified in a study with a relatively small number of samples for SNPs with a relatively high MAF. Due to the limited sample size of this study, it is possible that potential associations of rare alleles with VIN were not identified.  5.8 Future Directions This explorative work in the pharmacogenomics of VIN has been valuable as it has identified genes of focus for future studies, developed an important phenotyping and genotyping framework, as well as generated novel alternative postulates for the pathophysiology of VIN. In addition to furthering the understanding of VIN in paediatric cancer patients with Wilms tumor or rhabdomyosarcoma, this work will serve as a starting point for the study of other cohorts in replication studies.  5.8.1 Expansion, Replication, and Validation of Novel SNPs and Genes As the next step, expansion and replication studies are required to determine whether the association between VIN and these new candidate SNPs and genes can be independently replicated in additional cohorts, including: additional patients with Wilms tumor and rhabdomyosarcoma, different cancer types (e.g.: leukemias, lymphomas, and brain tumors), stratified ethnic groups, and also within the adult population where VIN tends to be more  117 severe. Except for the adult population, all of these patient types have already been recruited and phenotyped, and can be found in the 1,262 paediatric patients who were not part of this study. As a valuable tool, the high throughput screen of SNPs allowed for a multiple genetic variants to be simultaneously genotyped. However, this technique was unable to achieve optimal reaction conditions for each of the individual SNP reactions, leading to some suboptimal results or individual SNP failure. In order to obtain genotype information on variants that are of particular interest, and to minimize the costs of re-running the experiment, individual SNPs can be genotyped separately with an alternative genotyping platform. For example, it would be of interest to complete separate studies on PON1 rs854560, a novel candidate SNP that showed ambiguities in genotype clustering, and ABCB1 rs1045642, a variant that was previously associated with VIN and failed to be genotyped in this study. Additional SNPs of interest include all the known functional variants in PON1, PON1192 rs662 to verify the imputation results in this study, PON155 rs854560, which had ambiguous sample clustering, but specifically PON1-108 rs705379, PON1-162 rs705381 that were not genotyped and could not be imputed in this study.  5.8.2 Detailed Genetic Investigation of Novel Candidate Genes After replicating the findings, the next step will require a more detailed investigation of the genetic variation in these candidate genes. These studies would extend this work beyond the a priori selection of SNPs that were included in the ADME-Tox panel and the imputation analyses. This is particularly important since the selection of the original genes and SNPs in the ADME-Tox panel were based on the limited information in the genetics of  118 VIN, where this original genotyping quality and coverage also influenced the depth and accuracy of imputation. By re-sequencing the genes that are potentially implicated with VIN, this analysis will facilitate the identification of other genetic variants that may be associated with VIN, whether it is an indirect association (in LD with the identified SNPs), or directly (unknown variants within the same genes may be independently associated with VIN). At the same time, this analysis will provide genotype information and the potential association of known functional SNPs within the candidate genes, such as the PON1-108 rs705379, PON1-162 rs705381 variants that were neither part of the original genotyping panel or the Hapmap project. 5.8.3 Pharmacokinetic and Functional Validation Studies Following the identification and validation of the most likely causal SNPs within the candidate genes, it is important to perform functional validation studies to investigate the mechanisms that underlie the associations between these novel genes and VIN. Similarly, pharmacokinetic studies (in patients who are stratified by their VIN-associated genotypes) that compare the concentrations of vincristine and its metabolites will determine how these identified gene variants impact individual susceptibility to VIN. For example, this may provide information that relate drug concentrations and exposure to VIN. Expression of the different genetic variants in both cell culture and animal models could be utilized to investigate the impact of this genetic variation on enzyme activity and gene expression levels, as well as the effects on the cytotoxicity of vincristine and its metabolism. Cell culture models to study the effect of vincristine on chick embryo neurons (ENGLAND et al. 1973) and human Schwann cells (LEHMANN et al. 2011) have already been developed. Additionally, animal models of PON1 (FURLONG et al. 2000) and ABCA4  119 knockouts (SMIT et al. 1996) are available, as well as and PON1 transgenic mice. These animal models can be directly investigated or developed into cancer models (solid tumor implantation) as described in the texts to evaluate the balance between the development of VIN and survival rates. Furthermore, animal histology studies would determine the effects of these SNPs on the neuron, axon, and myelin during stages of damage and recovery from vincristine-induced neurodegeneration. These studies will provide a deeper insight into the role of each of these genes in the development of VIN and provide more information to supplement the postulated mechanism of action. Finally, another important step would be to investigate the expression of these proteins in different tissues, such as liver or neuronal cells, in order to provide further support for the postulated underlying mechanisms.  5.8.3.1 PON1 (Paraoxonase 1) and PPP1R9A (Neurabin 1) Genetically modified mice have been utilized to determine the effects of PON1 in the context of OP-induced neurotoxicity (FURLONG et al. 2000). Similarly, these PON1-/- and transgenic mice models could be utilized to study the importance of PON1 as a protectant against the development of VIN when they are administered clinically relevant doses of vincristine. For comparison purposes, these models could also be tested with other neurotoxins, such as OP (e.g.: parathion, malathion, chlorpyrifos, diazinon) and cuprizone, a neurotoxin that induces demyelination and remyelination in the CNS (MATSUSHIMA AND MORELL 2001). If the association of PON1 rs854549 with VIN is replicated, then further studies investigating the effect of this SNP on mRNA and protein levels will determine whether this SNP has regulatory effects. If this is established, protein-DNA binding studies can be  120 performed to evaluate whether a regulatory protein interacts within this potential site of regulation. For example, an electrophoretic mobility shift assay (EMSA) will determine if protein(s) bind to this DNA sequence and chromatin immunoprecipitation (chIP) studies will identify the binding complex of proteins that mediate this interaction with the DNA.  5.8.3.2 ABCA4, ABCG1, and CYP51A1: Cholesterol Biosynthesis and Intracellular Transport Similar to the in vivo overexpression of ABCB4 in mice (SMIT et al. 1996), the overexpression of ABCA4 in transgenic mice could be engineered to study its involvement and potential impact on the development of vincristine-induced PN. Additionally, in vitro retinoid studies, as outlined by Zhong et al., can be utilized to determine the importance of the transport of retinoids in the development of neurotoxicity (ZHONG and MOLDAY 2010). Moreover, to study the effects of the wildtype proteins and their respective polymorphisms on the myelin sheath during regeneration from vincristine-induced axon degeneration in these model systems, the myelin lipids could be labeled with an intraneural injection of 3H-acetate (GOODRUM et al. 1994), or 1-14C acetate (YAO 1988). Electron microscopy autoradiography studies would follow the distribution of these pre-labeled myelin lipids to determine whether polymorphisms in these proteins would affect the distribution of these lipids during reformation of the myelin sheath. In addition, morphometric analyses and assessment of mRNA levels of myelin protein P0 would provide further information about the impact of the candidate genes on the neural cells’ ability to re- myelinate and to regenerate axons. Finally, since CYP51A1 is known to be involved in  121 cholesterol biosynthesis, the impact of genetic variation on the functionality of this protein could be determined by comparing the ratio of cholesterol to both demosterol and lanosterol.  5.8.3.3 Metabolizers and Transporters The potential of PON1 and CYP51A1 to metabolize vincristine, and of SLCO1C1, ABCB4 and ABCG1 to transport vincristine has not been investigated so far, and thus requires investigation using functional studies. Over-expression and siRNA knockdown studies could be utilized to characterize the effects of these genes and their variants on in vivo vincristine metabolism and transport. Additionally, evaluating parent drug levels and the formation of metabolites will also identify major oxidative metabolites, and possible substrates for transportation. The variable genetic polymorphisms in the expression of these genes of interest may explain the inter-individual variability in clinical pharmacokinetic studies. Further studies could be conducted to determine the relative contribution of wildtype and variant proteins to vincristine metabolism in comparison to CYP3A5 and CYP3A4, and their relative transportation efficiency as compared to other vincristine transporters (ABCB1, ABCC1, ABCC2, ABCC10, RALBP1).  5.8.4 Investigation of Additional Genes Potentially Involved in VIN  The genotyping approach of this study was limited to the genes involved in the general ADME and toxicity of drugs, as well as several vincristine-specific genes of interest when the study was designed in 2009. However, an increasing number of other genes have been recently identified for their potential involvement in the pathophysiology of vincristine neurotoxicity. Thus, such genes could be genotyped in an additional study to determine their  122 role in VIN. Possible candidate genes involved in the mechanism of action of vincristine include the MAPT, microtubule-associated protein 2 (MAP2), and the various tubulin isoforms. To balance price, depth and spread of coverage, polymorphisms in these genes could be screened using a high throughput auxiliary genotyping panel, similar to the genotyping method as used for the ADME-Tox genotyping panel. Analogous to the genotyping panel used in this study, polymorphisms in the additional candidate genes would be selected based on their known functional effect as well as for use as HapMap derived haplotype tagging SNPs that would provide comprehensive coverage of the genetic variation in the genes of interest. SNPs with altered enzyme activity common non-synonymous, and literature validated rare non-synonymous, synonymous coding SNPs samples will be prioritized within these tagging SNPs.  Finally, as a complement to this targeted in-depth genotyping approach, a genome- wide association study (GWAS) could be performed to investigate the impact of genetic variation on the susceptibility to VIN on a genome wide scale. With a GWAS, the association between genetic variation and VIN can be investigated in a hypothesis-free manner using more than five million SNPs across the genome in each patient sample. However, due to the large number of gene variants analysed, relatively large sample sizes are required to identify true-positive results with sufficient statistical power. Although this will be possible to perform for the more frequently occurring cancers such as ALL, the collaboration of many centres and sites will be required to study cancers that occur less frequently (like Wilms tumor and rhabdomyosarcoma). Furthermore, exome sequencing could be performed in small numbers of patients with extreme phenotypes to identify previously unknown rare genetic variants that are potentially associated with VIN.  123  5.8.5 Observational Trials  This study utilized medical charts to retrospectively collect information about the patients’ medical history and vincristine-related adverse events. Prospective studies would enable following the development of VIN in more detail (e.g.: by grading the severity of VIN at each visit to characterize the development of neurotoxicity), as well as assessing the length of VIN and the recovery period. Such studies could thus provide detailed and nuanced phenotyping information on VIN, where these specifics could provide important information to facilitate further refinement of the identified genetic associations with VIN.  5.9 Future Implications The results from this preliminary study provide further evidence for the potential of utilizing genetic markers as a means to assess an individual patient’s risk of developing VIN. Despite an increased risk for VIN, the lack of alternative effective chemotherapy agents means that vincristine will still be used, but likely at a more individual basis. For example, this may differentiate who would benefit more from closer monitoring for early diagnosis and treatment. Similarly, the use of early preventative therapies may prevent the progression of VIN into more severe forms and could thus maximize a patient’s daily life function and independency. For example, this study identified vincristine-induced vocal cord paralysis as occurring at a higher incidence rate than previously thought. Therefore, this provides evidence that closer monitoring of patients is warranted, and that visualization of the airway is important for the diagnosis of vincristine-induced vocal cord paralysis. To illustrate, a larynscopy procedures may enable early detection and prevention of this adverse effect,  124 allowing for conservative treatment or a change in vincristine treatment if deemed necessary. While an endoscopic inspection to investigate laryngeal paralysis has been recommended to prevent misdiagnosis (KURUVILLA et al. 2009), and increased awareness and monitoring for vincristine-induced vocal cord paralysis will also be helpful. These interventions are useful because they are low-risk and high-yield, particularly in life-threatening situations such as vocal cord paralysis. This work will serve to provide information to aid the development of a simple, rapid predictive biomarker test. However, once developed, it will require further studies to validate the findings as well as to determine the cost-effectiveness in clinical practice. Individual risk predictions using genomic biomarkers for VIN may facilitate the appropriate allocation of scarce resources such as PT or OT referrals to those patients who are most likely to gain the greatest benefit from such interventions, as well as identifying patients who may benefit from counselling and alternative treatment options. Additionally, to minimize their risk of developing VIN, these options may include doses at lower frequencies or reduced doses of vincristine, as well as implementing standard dose escalation protocols that will minimize these dose-limiting toxicities. As proposed in a paper by Wittes et al., trials could be put in place with the goal to achieve a higher overall anti-tumor effect, the trials would test the best dosage and schedule in drugs with lower toxicities to reduce dose-limiting toxicities (WITTES and GOLDIN 1986). This will likely require a separate clinical trials per cancer type, due to the differential aggressiveness of the cancers and also because they are dosed differentially. The other patients who are not at risk for developing VIN can be given the regular protocol of vincristine, according to their cancer type. Pending further research, future treatment options may include the use of neuroprotectants such as glutamate or lipoic acid (ANDRONE  125 et al. 2000), where this research is supported by this discovery of the association between an antioxidant pathway (PON1) and VIN. Ultimately, such alterations in vincristine therapy need to be balanced between the risk of VIN and survival outcomes (maximize cell killing and minimize toxicity), and will require clinical testing to determine treatments with an optimized risk-benefit ratio for the different predicted risk groups. These results have the potential to change the dosing techniques of vincristine in cancer therapy and provide decision makers and clinicians with one step closer to personalizing medicine.     126 Chapter  6: Conclusion As one of the few antineoplastic agents that does not cause myelosuppression and its breadth as an effective chemotherapy agent, the use of vincristine has been limited by VIN. This study was the first to screen a broad panel of genes involved in drug ADME and toxicity for their involvement in VIN, for hypothesis-generating purposes of discovering novel associations of SNPs with VIN. Given the incomplete understanding of the biotransformation of vincristine in the body and the inability of previously identified polymorphisms in the genes in the metabolism and transportation pathway of vincristine to accurately explain the occurrence of VIN, this study adds important new evidence by identifying new potential biomarkers for VIN and providing novel hypotheses for the mechanisms underlying the susceptibility to VIN, as well as the development of VIN. The exact mechanism underlying a potential interaction between the newly identified genes and vincristine, and their relationship with VIN is currently unknown and requires further investigation. This is the first time PON1, PPP1R9A, ABCA4, CYP51A1, ABCG1, SLCO1C1 have been associated with VIN. Interestingly, each new candidate gene identified in this study has been indirectly linked to neuropathy, neurodegenerative diseases, remyelination and regeneration of the axon, neural cell function, neuroprotectants, metabolism, and transportation – all of which are areas that strengthen the hypothesis of an involvement of these genes in the susceptibility to VIN. All candidate SNPs identified in this study had effects of similar size and no study so far has identified a single strong genetic predictor for VIN. This suggests that the genetic basis of VIN is likely polygenic, and that a combination of risk variants from multiple genes may be required to accurately identify patients at high risk for VIN. At this stage, I hypothesize that the function of these genes could be involved in  127 VIN in the context of nerve regeneration (PON1, PPP1R9A), cholesterol homeostasis and myelin regeneration (ABCA4, ABCG1, CYP51A1), metabolism of vincristine (PON1, CYP51A1), and transportation of lipids, vincristine, metabolites, and neuroprotectants (SLCO1C1, ABCA4, ABCG1). I also hypothesize that polymorphisms in the six genes may affect the activity and functionality of each gene, leaving at-risk individuals inherently susceptible and unable to compensate and/or protect against the unexpected toxic effects of vincristine and other chemical stressors, prompting VIN when vincristine is given at standard doses. As a discovery study, this pilot work will require a further investigation of the identified candidate biomarkers. Nevertheless, this study provides an important first step towards more individualized therapies using vincristine, with the ultimate goal of improving outcomes by reducing the debilitating consequences of VIN.     128 References  ABBOTT, C. A., M. I. MACKNESS, S. KUMAR, A. J. BOULTON and P. N. DURRINGTON, 1995 Serum paraoxonase activity, concentration, and phenotype distribution in diabetes mellitus and its relationship to serum lipids and lipoproteins. Arteriosclerosis, thrombosis, and vascular biology 15: 1812-1818. ANDRONE, L., N. A. GAVAN, I. A. VERESIU and R. ORASAN, 2000 In vivo effect of lipoic acid on lipid peroxidation in patients with diabetic neuropathy. In vivo 14: 327-330. APLENC, R., W. GLATFELTER, P. HAN, E. RAPPAPORT, M. LA et al., 2003 CYP3A genotypes and treatment response in paediatric acute lymphoblastic leukaemia. British journal of haematology 122: 240-244. ARNDT, C., M. TEFFT, E. GEHAN, J. ANDERSON, M. JENSON et al., 1997 A feasibility, toxicity, and early response study of etoposide, ifosfamide, and vincristine for the treatment of children with rhabdomyosarcoma: a report from the Intergroup Rhabdomyosarcoma Study (IRS) IV pilot study. Journal of pediatric hematology/oncology : official journal of the American Society of Pediatric Hematology/Oncology 19: 124-129. ASCO ONCOLOGY, 2010 ASCO (American Society of Clinical Oncology) in Action Alert: Interview with Dr. Michael Link, http://www.asco.org/ascov2/Practice+&+Guidelines/ASCO+in+Action/ASCO+in+A ction+Alerts/ASCO+in+Action+Alert+:+Interview+with+Dr.+Michael+Link AVIRAM, M., E. HARDAK, J. VAYA, S. MAHMOOD, S. MILO et al., 2000 Human serum paraoxonases (PON1) Q and R selectively decrease lipid peroxides in human coronary and carotid atherosclerotic lesions: PON1 esterase and peroxidase-like activities. Circulation 101: 2510-2517. BAAS, P. W., 1997 Microtubules and axonal growth. Current opinion in cell biology 9: 29-36. BAE, S. H., and Y. K. PAIK, 1997 Cholesterol biosynthesis from lanosterol: development of a novel assay method and characterization of rat liver microsomal lanosterol delta 24- reductase. The Biochemical journal 326 ( Pt 2): 609-616. BARRETT, J. C., B. FRY, J. MALLER and M. J. DALY, 2005 Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21: 263-265. BHONGSATIERN, J., S. OHTSUKI, M. TACHIKAWA, S. HORI and T. TERASAKI, 2005 Retinal- specific ATP-binding cassette transporter (ABCR/ABCA4) is expressed at the choroid plexus in rat brain. Journal of neurochemistry 92: 1277-1280. BOUMAN, H. J., E. SCHOMIG, J. W. VAN WERKUM, J. VELDER, C. M. HACKENG et al., 2011 Paraoxonase-1 is a major determinant of clopidogrel efficacy. Nature medicine 17: 110-116. BOURRE, J. M., M. CLEMENT, D. GERARD, R. LEGRAND and J. CHAUDIERE, 1990 Precursors for cholesterol synthesis (7-dehydrocholesterol, 7-dehydrodesmosterol, and desmosterol): cholesterol/7-dehydrocholesterol ratio as an index of development and aging in PNS but not in CNS. J Neurochem 54: 1196-1199. BRADLEY, W. G., L. P. LASSMAN, G. W. PEARCE and J. N. WALTON, 1970 The neuromyopathy of vincristine in man. Clinical, electrophysiological and pathological studies. Journal of the neurological sciences 10: 107-131.  129 BRESLOW, N., J. B. BECKWITH, M. CIOL and K. SHARPLES, 1988 Age distribution of Wilms' tumor: report from the National Wilms' Tumor Study. Cancer research 48: 1653- 1657. BROPHY, V. H., R. L. JAMPSA, J. B. CLENDENNING, L. A. MCKINSTRY, G. P. JARVIK et al., 2001 Effects of 5' regulatory-region polymorphisms on paraoxonase-gene (PON1) expression. American journal of human genetics 68: 1428-1436. BROPHY, V. H., G. P. JARVIK, R. J. RICHTER, L. S. ROZEK, G. D. SCHELLENBERG et al., 2000 Analysis of paraoxonase (PON1) L55M status requires both genotype and phenotype. Pharmacogenetics 10: 453-460. BURNETT, P. E., S. BLACKSHAW, M. M. LAI, I. A. QURESHI, A. F. BURNETT et al., 1998 Neurabin is a synaptic protein linking p70 S6 kinase and the neuronal cytoskeleton. Proceedings of the National Academy of Sciences of the United States of America 95: 8351-8356. BUSI, F., and T. CRESTEIL, 2005 CYP3A5 mRNA degradation by nonsense-mediated mRNA decay. Molecular pharmacology 68: 808-815. CANCER THERAPY EVALUATION PROGRAM (NATIONAL CANCER INSTITUTE, N., 2010 Common Terminology Criteria for Adverse Events (CTCAE) Version 4.0 CARLETON, B., R. POOLE, M. SMITH, J. LEEDER, R. GHANNADAN et al., 2009 Adverse drug reaction active surveillance: developing a national network in Canada's children's hospitals. Pharmacoepidemiology and drug safety 18: 713-721. CAROZZI, V. A., A. CANTA, N. OGGIONI, B. SALA, A. CHIORAZZI et al., 2010 Neurophysiological and neuropathological characterization of new murine models of chemotherapy-induced chronic peripheral neuropathies. Experimental neurology 226: 301-309. CAVALETTI, G., B. FRIGENI, F. LANZANI, M. PIATTI, S. ROTA et al., 2007 The Total Neuropathy Score as an assessment tool for grading the course of chemotherapy- induced peripheral neurotoxicity: comparison with the National Cancer Institute- Common Toxicity Scale. Journal of the peripheral nervous system : JPNS 12: 210- 215. CAVALETTI, G., S. JANN, A. PACE, R. PLASMATI, G. SICILIANO et al., 2006 Multi-center assessment of the Total Neuropathy Score for chemotherapy-induced peripheral neurotoxicity. Journal of the peripheral nervous system : JPNS 11: 135-141. CHAN, J. D., 1998 Pharmacokinetic drug interactions of vinca alkaloids: summary of case reports. Pharmacotherapy 18: 1304-1307. CONTRERAS, P. C., J. L. VAUGHT, J. A. GRUNER, C. BROSNAN, C. STEFFLER et al., 1997 Insulin-like growth factor-I prevents development of a vincristine neuropathy in mice. Brain research 774: 20-26. CORCORAN, J., P. L. SO and M. MADEN, 2002 Absence of retinoids can induce motoneuron disease in the adult rat and a retinoid defect is present in motoneuron disease patients. Journal of cell science 115: 4735-4741. COSTA, L. G., T. B. COLE and C. E. FURLONG, 2005 Paraoxonase (PON1): from toxicology to cardiovascular medicine. Acta bio-medica : Atenei Parmensis 76 Suppl 2: 50-57. CRIST, W. M., J. R. ANDERSON, J. L. MEZA, C. FRYER, R. B. RANEY et al., 2001 Intergroup rhabdomyosarcoma study-IV: results for patients with nonmetastatic disease. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 19: 3091-3102.  130 CRONIN, S., M. J. GREENWAY, J. H. PREHN and O. HARDIMAN, 2007 Paraoxonase promoter and intronic variants modify risk of sporadic amyotrophic lateral sclerosis. Journal of neurology, neurosurgery, and psychiatry 78: 984-986. D. M. TE LOO, R. M. V. S., P. M. HOOGERBRUGGE, 2009 Effect of azole antifungal therapy on vincristine toxicity in childhood acute lymphoblastic leukemia. Journal of Clinical Oncology 27: 15. DAVIES, H. G., R. J. RICHTER, M. KEIFER, C. A. BROOMFIELD, J. SOWALLA et al., 1996 The effect of the human serum paraoxonase polymorphism is reversed with diazoxon, soman and sarin. Nature genetics 14: 334-336. DE GEEST, K., J. A. BLESSING, R. T. MORRIS, S. D. YAMADA, B. J. MONK et al., 2010 Phase II clinical trial of ixabepilone in patients with recurrent or persistent platinum- and taxane-resistant ovarian or primary peritoneal cancer: a gynecologic oncology group study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 28: 149-153. DENNISON, J. B., D. R. JONES, J. L. RENBARGER and S. D. HALL, 2007 Effect of CYP3A5 expression on vincristine metabolism with human liver microsomes. The Journal of pharmacology and experimental therapeutics 321: 553-563. DENNISON, J. B., P. KULANTHAIVEL, R. J. BARBUCH, J. L. RENBARGER, W. J. EHLHARDT et al., 2006 Selective metabolism of vincristine in vitro by CYP3A5. Drug metabolism and disposition: the biological fate of chemicals 34: 1317-1327. DENNISON, J. B., M. A. MOHUTSKY, R. J. BARBUCH, S. A. WRIGHTON and S. D. HALL, 2008 Apparent high CYP3A5 expression is required for significant metabolism of vincristine by human cryopreserved hepatocytes. The Journal of pharmacology and experimental therapeutics 327: 248-257. DRAGANOV, D. I., and B. N. LA DU, 2004 Pharmacogenetics of paraoxonases: a brief review. Naunyn-Schmiedeberg's archives of pharmacology 369: 78-88. DRAKE, K. J., J. SINGHAL, S. YADAV, A. NADKAR, C. PUNGALIYA et al., 2007 RALBP1/RLIP76 mediates multidrug resistance. International journal of oncology 30: 139-144. EDWARDS, I. R., and J. K. ARONSON, 2000 Adverse drug reactions: definitions, diagnosis, and management. Lancet 356: 1255-1259. EGBELAKIN, A., M. J. FERGUSON, E. A. MACGILL, A. S. LEHMANN, A. R. TOPLETZ et al., 2011 Increased risk of vincristine neurotoxicity associated with low CYP3A5 expression genotype in children with acute lymphoblastic leukemia. Pediatric blood & cancer 56: 361-367. ENGLAND, J. D., G. S. GRONSETH, G. FRANKLIN, R. G. MILLER, A. K. ASBURY et al., 2005 Distal symmetric polyneuropathy: a definition for clinical research: report of the American Academy of Neurology, the American Association of Electrodiagnostic Medicine, and the American Academy of Physical Medicine and Rehabilitation. Neurology 64: 199-207. ENGLAND, J. M., M. E. KADIN and M. N. GOLDSTEIN, 1973 The effect of vincristine sulphate on the axoplasmic flow of proteins in cultured sympathetic neurons. J Cell Sci 12: 549-565. FELIX, C. A., A. H. WALKER, B. J. LANGE, T. M. WILLIAMS, N. J. WINICK et al., 1998 Association of CYP3A4 genotype with treatment-related leukemia. Proceedings of the National Academy of Sciences of the United States of America 95: 13176-13181.  131 FERRARI, A., M. CASANOVA, G. BISOGNO, I. ZANETTI, G. CECCHETTO et al., 2003 Rhabdomyosarcoma in infants younger than one year old: a report from the Italian Cooperative Group. Cancer 97: 2597-2604. FISHER, M. B., M. VANDENBRANDEN, K. FINDLAY, B. BURCHELL, K. E. THUMMEL et al., 2000 Tissue distribution and interindividual variation in human UDP- glucuronosyltransferase activity: relationship between UGT1A1 promoter genotype and variability in a liver bank. Pharmacogenetics 10: 727-739. FUNG, K. L., and M. M. GOTTESMAN, 2009 A synonymous polymorphism in a common MDR1 (ABCB1) haplotype shapes protein function. Biochimica et biophysica acta 1794: 860-871. FURLONG, C. E., W. F. LI, V. H. BROPHY, G. P. JARVIK, R. J. RICHTER et al., 2000 The PON1 gene and detoxication. Neurotoxicology 21: 581-587. GAO, X., J. STARMER and E. R. MARTIN, 2008 A multiple testing correction method for genetic association studies using correlated single nucleotide polymorphisms. Genetic epidemiology 32: 361-369. GARIN, M. C., R. W. JAMES, P. DUSSOIX, H. BLANCHE, P. PASSA et al., 1997 Paraoxonase polymorphism Met-Leu54 is associated with modified serum concentrations of the enzyme. A possible link between the paraoxonase gene and increased risk of cardiovascular disease in diabetes. The Journal of clinical investigation 99: 62-66. GHAFOOR, A., A. JEMAL, V. COKKINIDES, C. CARDINEZ, T. MURRAY et al., 2002 Cancer statistics for African Americans. CA: a cancer journal for clinicians 52: 326-341. GIDDING, C. E., S. J. KELLIE, W. A. KAMPS and S. S. DE GRAAF, 1999 Vincristine revisited. Critical reviews in oncology/hematology 29: 267-287. GOODRUM, J. F., T. EARNHARDT, N. GOINES and T. W. BOULDIN, 1994 Fate of myelin lipids during degeneration and regeneration of peripheral nerve: an autoradiographic study. J Neurosci 14: 357-367. GREEN, D. M., N. E. BRESLOW, J. B. BECKWITH, J. Z. FINKLESTEIN, P. GRUNDY et al., 1998 Effect of duration of treatment on treatment outcome and cost of treatment for Wilms' tumor: a report from the National Wilms' Tumor Study Group. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 16: 3744- 3751. GREEN, D. M., P. NORKOOL, N. E. BRESLOW, J. Z. FINKLESTEIN and G. J. D'ANGIO, 1990 Severe hepatic toxicity after treatment with vincristine and dactinomycin using single-dose or divided-dose schedules: a report from the National Wilms' Tumor Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 8: 1525-1530. GREEN, L. S., J. A. DONOSO, I. E. HELLER-BETTINGER and F. E. SAMSON, 1975 Axonal transport disturbances in the vincristine-induced peripheral neuropathy. Transactions of the American Neurological Association 100: 195-196. GUO, S. W., and E. A. THOMPSON, 1992 Performing the exact test of Hardy-Weinberg proportion for multiple alleles. Biometrics 48: 361-372. HANKEY GJ, W. J., 2002 Diseases of the Peripheral Nerve, pp. 579-592 in Clinical Neurology, edited by J. NORTHCOTT. Demos Medical Publishing, London. HARNICAR, S., N. ADEL and J. JURCIC, 2009 Modification of vincristine dosing during concomitant azole therapy in adult acute lymphoblastic leukemia patients. Journal of  132 oncology pharmacy practice : official publication of the International Society of Oncology Pharmacy Practitioners 15: 175-182. HARRIS, C. M., and R. H. BLANCHAERT, 2006 Bilateral recurrent laryngeal nerve palsy resulting from treatment with vincristine. Journal of oral and maxillofacial surgery : official journal of the American Association of Oral and Maxillofacial Surgeons 64: 738-739. HARTMAN, A., R. H. VAN SCHAIK, I. P. VAN DER HEIDEN, M. J. BROEKHUIS, M. MEIER et al., 2010 Polymorphisms in genes involved in vincristine pharmacokinetics or pharmacodynamics are not related to impaired motor performance in children with leukemia. Leukemia research 34: 154-159. HASSETT, C., R. J. RICHTER, R. HUMBERT, C. CHAPLINE, J. W. CRABB et al., 1991 Characterization of cDNA clones encoding rabbit and human serum paraoxonase: the mature protein retains its signal sequence. Biochemistry 30: 10141-10149. HAYAKAWA, K., G. SOBUE, T. ITOH and T. MITSUMA, 1994 Nerve growth factor prevents neurotoxic effects of cisplatin, vincristine and taxol, on adult rat sympathetic ganglion explants in vitro. Life sciences 55: 519-525. HAZELL, L., and S. A. SHAKIR, 2006 Under-reporting of adverse drug reactions : a systematic review. Drug safety : an international journal of medical toxicology and drug experience 29: 385-396. HERRUP, K., and Y. YANG, 2007 Cell cycle regulation in the postmitotic neuron: oxymoron or new biology? Nature reviews. Neuroscience 8: 368-378. HIER, D. B., B. G. ARNASON and M. YOUNG, 1972 Studies on the mechanism of action of nerve growth factor. Proceedings of the National Academy of Sciences of the United States of America 69: 2268-2272. HOUGHTON, J. A., W. H. MEYER and P. J. HOUGHTON, 1987 Scheduling of vincristine: drug accumulation and response of xenografts of childhood rhabdomyosarcoma determined by frequency of administration. Cancer treatment reports 71: 717-721. HOUI, H., S. MOCHIO and T. KOBAYASHI, 1993 Gangliosides attenuate vincristine neurotoxicity on dorsal root ganglion cells. Muscle & nerve 16: 11-14. HOWIE, B. N., P. DONNELLY and J. MARCHINI, 2009 A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS genetics 5: e1000529. HUFF, L. M., D. L. SACKETT, M. S. PORUCHYNSKY and T. FOJO, 2010 Microtubule-disrupting chemotherapeutics result in enhanced proteasome-mediated degradation and disappearance of tubulin in neural cells. Cancer research 70: 5870-5879. JEMAL, A., A. THOMAS, T. MURRAY and M. THUN, 2002 Cancer statistics, 2002. CA: a cancer journal for clinicians 52: 23-47. JENG, M. R., and J. FEUSNER, 2001 Itraconazole-enhanced vincristine neurotoxicity in a child with acute lymphoblastic leukemia. Pediatric hematology and oncology 18: 137-142. JORDAN, M. A., D. THROWER and L. WILSON, 1991 Mechanism of inhibition of cell proliferation by Vinca alkaloids. Cancer research 51: 2212-2222. JUREVICS, H., T. W. BOULDIN, A. D. TOEWS and P. MORELL, 1998 Regenerating sciatic nerve does not utilize circulating cholesterol. Neurochemical research 23: 401-406. KALOW, W., B. K. TANG and L. ENDRENYI, 1998 Hypothesis: comparisons of inter- and intra- individual variations can substitute for twin studies in drug research. Pharmacogenetics 8: 283-289.  133 KAMALUDDIN, M., P. MCNALLY, F. BREATNACH, A. O'MARCAIGH, D. WEBB et al., 2001 Potentiation of vincristine toxicity by itraconazole in children with lymphoid malignancies. Acta paediatrica 90: 1204-1207. KEARNS, G. L., S. M. ABDEL-RAHMAN, S. W. ALANDER, D. L. BLOWEY, J. S. LEEDER et al., 2003 Developmental pharmacology--drug disposition, action, and therapy in infants and children. The New England journal of medicine 349: 1157-1167. KIM, M. J., P. SKEWES-COX, H. FUKUSHIMA, S. HESSELSON, S. W. YEE et al., 2011 Functional characterization of liver enhancers that regulate drug-associated transporters. Clin Pharmacol Ther 89: 571-578. KIM, W. S., C. S. WEICKERT and B. GARNER, 2008 Role of ATP-binding cassette transporters in brain lipid transport and neurological disease. Journal of neurochemistry 104: 1145-1166. KISHI, S., W. YANG, B. BOUREAU, S. MORAND, S. DAS et al., 2004 Effects of prednisone and genetic polymorphisms on etoposide disposition in children with acute lymphoblastic leukemia. Blood 103: 67-72. KIVISTO, K. T., H. K. KROEMER and M. EICHELBAUM, 1995 The role of human cytochrome P450 enzymes in the metabolism of anticancer agents: implications for drug interactions. British journal of clinical pharmacology 40: 523-530. KOBAYASHI, A., Y. TAKANEZAWA, T. HIRATA, Y. SHIMIZU, K. MISASA et al., 2006 Efflux of sphingomyelin, cholesterol, and phosphatidylcholine by ABCG1. Journal of lipid research 47: 1791-1802. KOZIK, M. B., and L. MAZIARZ, 1983 Morphological and histochemical changes in rats brain induced by vincristine administered in early periods of postnatal development. Folia morphologica 42: 199-216. KOZIK, M. B., and J. SZCZECH, 1983 Histoenzymatic and morphometric changes occurring in nerve cells of gyrus cinguli following treatment of young rats with vincristine. Experimental pathology 24: 243-252. KUEHL, P., J. ZHANG, Y. LIN, J. LAMBA, M. ASSEM et al., 2001 Sequence diversity in CYP3A promoters and characterization of the genetic basis of polymorphic CYP3A5 expression. Nature genetics 27: 383-391. KURUVILLA, G., S. PERRY, B. WILSON and H. EL-HAKIM, 2009 The natural history of vincristine-induced laryngeal paralysis in children. Archives of otolaryngology--head & neck surgery 135: 101-105. LAFERRIERE, N. B., T. H. MACRAE and D. L. BROWN, 1997 Tubulin synthesis and assembly in differentiating neurons. Biochemistry and cell biology = Biochimie et biologie cellulaire 75: 103-117. LATASA, M. J., and J. M. COSGAYA, 2011 Regulation of retinoid receptors by retinoic acid and axonal contact in Schwann cells. PLoS One 6: e17023. LAZAROU, J., B. H. POMERANZ and P. N. COREY, 1998 Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA : the journal of the American Medical Association 279: 1200-1205. LEHMANN, H. C., W. CHEN, R. MI, S. WANG, Y. LIU et al., 2011 Human Schwann Cells Retain Essential Phenotype Characteristics After Immortalization. Stem Cells Dev. LEVEQUE, D., and F. JEHL, 2007 Molecular pharmacokinetics of catharanthus (vinca) alkaloids. Journal of clinical pharmacology 47: 579-588.  134 LI, H. L., D. P. LIU and C. C. LIANG, 2003 Paraoxonase gene polymorphisms, oxidative stress, and diseases. Journal of molecular medicine 81: 766-779. LOO, T. T., C. J. ROSS, J. SISTONEN, H. VISSCHER, P. MADADI et al., 2010 Pharmacogenomics and active surveillance for serious adverse drug reactions in children. Pharmacogenomics 11: 1269-1285. LUTJOHANN, D., A. BRZEZINKA, E. BARTH, D. ABRAMOWSKI, M. STAUFENBIEL et al., 2002 Profile of cholesterol-related sterols in aged amyloid precursor protein transgenic mouse brain. Journal of lipid research 43: 1078-1085. MACKNESS, B., M. I. MACKNESS, S. ARROL, W. TURKIE, K. JULIER et al., 1998 Serum paraoxonase (PON1) 55 and 192 polymorphism and paraoxonase activity and concentration in non-insulin dependent diabetes mellitus. Atherosclerosis 139: 341- 349. MACPHEE, I. A., S. FREDERICKS, M. MOHAMED, M. MORETON, N. D. CARTER et al., 2005 Tacrolimus pharmacogenetics: the CYP3A5*1 allele predicts low dose-normalized tacrolimus blood concentrations in whites and South Asians. Transplantation 79: 499- 502. MATSUSHIMA, G. K., and P. MORELL, 2001 The neurotoxicant, cuprizone, as a model to study demyelination and remyelination in the central nervous system. Brain pathology 11: 107-116. MCCARTHY, G. M., and J. R. SKILLINGS, 1992 Jaw and other orofacial pain in patients receiving vincristine for the treatment of cancer. Oral surgery, oral medicine, and oral pathology 74: 299-304. MELLI, G., M. TAIANA, F. CAMOZZI, D. TRIOLO, P. PODINI et al., 2008 Alpha-lipoic acid prevents mitochondrial damage and neurotoxicity in experimental chemotherapy neuropathy. Experimental neurology 214: 276-284. MEZA, J. L., J. ANDERSON, A. S. PAPPO and W. H. MEYER, 2006 Analysis of prognostic factors in patients with nonmetastatic rhabdomyosarcoma treated on intergroup rhabdomyosarcoma studies III and IV: the Children's Oncology Group. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 24: 3844-3851. MISHRA, P. J., R. HUMENIUK, G. S. LONGO-SORBELLO, D. BANERJEE and J. R. BERTINO, 2007 A miR-24 microRNA binding-site polymorphism in dihydrofolate reductase gene leads to methotrexate resistance. Proceedings of the National Academy of Sciences of the United States of America 104: 13513-13518. MISRA, U. K., and J. KALITA, 2009 Toxic neuropathies. Neurology India 57: 697-705. MITTMANN, N., S. R. KNOWLES, M. GOMEZ, J. S. FISH, R. CARTOTTO et al., 2004 Evaluation of the extent of under-reporting of serious adverse drug reactions: the case of toxic epidermal necrolysis. Drug safety : an international journal of medical toxicology and drug experience 27: 477-487. MOORE, T. J., B. M. PSATY and C. D. FURBERG, 1998 Time to act on drug safety. JAMA : the journal of the American Medical Association 279: 1571-1573. MORGAN, E., E. BAUM, N. BRESLOW, J. TAKASHIMA and G. D'ANGIO, 1988 Chemotherapy- related toxicity in infants treated according to the Second National Wilms' Tumor Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 6: 51-55.  135 NAKABAYASHI, K., S. MAKINO, S. MINAGAWA, A. C. SMITH, J. S. BAMFORTH et al., 2004 Genomic imprinting of PPP1R9A encoding neurabin I in skeletal muscle and extra- embryonic tissues. Journal of medical genetics 41: 601-608. NARANJO, C. A., U. BUSTO, E. M. SELLERS, P. SANDOR, I. RUIZ et al., 1981 A method for estimating the probability of adverse drug reactions. Clinical pharmacology and therapeutics 30: 239-245. NEUBERT, A., H. DORMANN, J. WEISS, T. EGGER, M. CRIEGEE-RIECK et al., 2004 The impact of unlicensed and off-label drug use on adverse drug reactions in paediatric patients. Drug safety : an international journal of medical toxicology and drug experience 27: 1059-1067. OGNJANOVIC, S., A. M. LINABERY, B. CHARBONNEAU and J. A. ROSS, 2009 Trends in childhood rhabdomyosarcoma incidence and survival in the United States, 1975- 2005. Cancer 115: 4218-4226. PAN, Y. A., T. MISGELD, J. W. LICHTMAN and J. R. SANES, 2003 Effects of neurotoxic and neuroprotective agents on peripheral nerve regeneration assayed by time-lapse imaging in vivo. The Journal of neuroscience : the official journal of the Society for Neuroscience 23: 11479-11488. PHARMGKB, 2011 Pathway:Vinka Alkaloid Pathway, Pharmacokinetics. http://www.pharmgkb.org/do/serve?objId=PA150981002&objCls=Pathway PLASSCHAERT, S. L., E. GRONINGER, M. BOEZEN, I. KEMA, E. G. DE VRIES et al., 2004 Influence of functional polymorphisms of the MDR1 gene on vincristine pharmacokinetics in childhood acute lymphoblastic leukemia. Clinical pharmacology and therapeutics 76: 220-229. POLLOCK, B. H., M. R. DEBAUN, B. M. CAMITTA, J. J. SHUSTER, Y. RAVINDRANATH et al., 2000 Racial differences in the survival of childhood B-precursor acute lymphoblastic leukemia: a Pediatric Oncology Group Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 18: 813-823. POLLOCK, N. L., and R. CALLAGHAN, 2011 The lipid translocase, ABCA4: seeing is believing. FEBS J. PORTER, C. C., A. E. CARVER and E. A. ALBANO, 2009 Vincristine induced peripheral neuropathy potentiated by voriconazole in a patient with previously undiagnosed CMT1X. Pediatr Blood Cancer 52: 298-300. PRIMO-PARMO, S. L., R. C. SORENSON, J. TEIBER and B. N. LA DU, 1996 The human serum paraoxonase/arylesterase gene (PON1) is one member of a multigene family. Genomics 33: 498-507. RASSEKH, R., 2010 Children's Oncology Group (COG) Safety Alert, pp., edited by T. LOO, Vancouver. RENBARGER, J. L., K. C. MCCAMMACK, C. E. ROUSE and S. D. HALL, 2008 Effect of race on vincristine-associated neurotoxicity in pediatric acute lymphoblastic leukemia patients. Pediatric blood & cancer 50: 769-771. SAHENK, Z., S. T. BRADY and J. R. MENDELL, 1987 Studies on the pathogenesis of vincristine-induced neuropathy. Muscle & nerve 10: 80-84. SAHER, G., S. QUINTES and K. A. NAVE, Cholesterol: a novel regulatory role in myelin formation. Neuroscientist 17: 79-93. SATHIAPALAN, R. K., A. AL-NASSER, H. EL-SOLH, I. AL-MOHSEN and S. AL-JUMAAH, 2002 Vincristine-itraconazole interaction: cause for increasing concern. Journal of pediatric  136 hematology/oncology : official journal of the American Society of Pediatric Hematology/Oncology 24: 591. SATHIAPALAN, R. K., and H. EL-SOLH, 2001 Enhanced vincristine neurotoxicity from drug interactions: case report and review of literature. Pediatric hematology and oncology 18: 543-546. SLOWIK, A., B. TOMIK, P. P. WOLKOW, D. PARTYKA, W. TURAJ et al., 2006 Paraoxonase gene polymorphisms and sporadic ALS. Neurology 67: 766-770. SMIT, J. J., F. BAAS, J. E. HOOGENDIJK, G. H. JANSEN, M. A. VAN DER VALK et al., 1996 Peripheral neuropathy in mice transgenic for a human MDR3 P-glycoprotein mini- gene. The Journal of neuroscience : the official journal of the Society for Neuroscience 16: 6386-6393. SMITH, M. A., N. L. SEIBEL, S. F. ALTEKRUSE, L. A. RIES, D. L. MELBERT et al., 2010 Outcomes for children and adolescents with cancer: challenges for the twenty-first century. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 28: 2625-2634. SONG, O., FUJII, KATO, NAKASHIMA, 2010 Expression of LDM in the nervous system and its changes in the processes of myelination and remyelination, pp., Amsterdam. SONG, P., J. K. LAMBA, L. ZHANG, E. SCHUETZ, N. SHUKLA et al., 2006 G2677T and C3435T genotype and haplotype are associated with hepatic ABCB1 (MDR1) expression. Journal of clinical pharmacology 46: 373-379. STEPHENSON, T., 2005 How children's responses to drugs differ from adults. British journal of clinical pharmacology 59: 670-673. STRASSBURG, C. P., A. STRASSBURG, S. KNEIP, A. BARUT, R. H. TUKEY et al., 2002 Developmental aspects of human hepatic drug glucuronidation in young children and adults. Gut 50: 259-265. STRIDE, B. D., C. E. GRANT, D. W. LOE, D. R. HIPFNER, S. P. COLE et al., 1997 Pharmacological characterization of the murine and human orthologs of multidrug- resistance protein in transfected human embryonic kidney cells. Molecular pharmacology 52: 344-353. SUN, H., R. S. MOLDAY and J. NATHANS, 1999 Retinal stimulates ATP hydrolysis by purified and reconstituted ABCR, the photoreceptor-specific ATP-binding cassette transporter responsible for Stargardt disease. The Journal of biological chemistry 274: 8269- 8281. TABORI, U., L. SUNG, J. HUKIN, N. LAPERRIERE, B. CROOKS et al., 2005 Medulloblastoma in the second decade of life: a specific group with respect to toxicity and management: a Canadian Pediatric Brain Tumor Consortium Study. Cancer 103: 1874-1880. TAFT, R. J., K. C. PANG, T. R. MERCER, M. DINGER and J. S. MATTICK, 2010 Non-coding RNAs: regulators of disease. The Journal of pathology 220: 126-139. TANNER, K. D., J. D. LEVINE and K. S. TOPP, 1998 Microtubule disorientation and axonal swelling in unmyelinated sensory axons during vincristine-induced painful neuropathy in rat. The Journal of comparative neurology 395: 481-492. THERVET, E., C. LEGENDRE, P. BEAUNE and D. ANGLICHEAU, 2005 Cytochrome P450 3A polymorphisms and immunosuppressive drugs. Pharmacogenomics 6: 37-47. THOMPSON, E. E., H. KUTTAB-BOULOS, D. WITONSKY, L. YANG, B. A. ROE et al., 2004 CYP3A variation and the evolution of salt-sensitivity variants. American journal of human genetics 75: 1059-1069.  137 TOGHILL, P. J., and J. D. BURKE, 1970 Death from paralytic ileus following vincristine therapy. Postgraduate medical journal 46: 330-331. US GOVERNMENT ACCOUNTABILITY OFFICE, U. S., 2006 Drug Safety: Improvements Needed in FDA’s Postmarket Decision-making and Oversight Process. Publication GAO-06- 402 VALLAT, J. M., and B. FUNALOT, 2010 [Charcot-Marie-Tooth (CMT) disease: an update]. Medecine sciences : M/S 26: 842-847. VAN SCHIE, R. M., R. J. BRUGGEMANN, P. M. HOOGERBRUGGE and D. M. TE LOO, 2011 Effect of azole antifungal therapy on vincristine toxicity in childhood acute lymphoblastic leukaemia. The Journal of antimicrobial chemotherapy. VIA, M., C. GIGNOUX and E. G. BURCHARD, 2010 The 1000 Genomes Project: new opportunities for research and social challenges. Genome Med 2: 3. WANG, M. S., Y. WU, D. G. CULVER and J. D. GLASS, 2000 Pathogenesis of axonal degeneration: parallels between Wallerian degeneration and vincristine neuropathy. Journal of neuropathology and experimental neurology 59: 599-606. WEXLER LH, C. W., HELMAN LJ. , 2002 Rhabdomyosarcoma and the undifferentiated sarcomas. In: Pizzo PA, Poplack DG, editors. Principles and practice of pediatric oncology. Philedalphia: Lippincott Wiliams and Wilkins. 4 Edition 939-971. WHITE, T. J., A. ARAKELIAN and J. P. RHO, 1999 Counting the costs of drug-related adverse events. PharmacoEconomics 15: 445-458. WITTES, R. E., and A. GOLDIN, 1986 Unresolved issues in combination chemotherapy. Cancer treatment reports 70: 105-125. WONG, I., D. SWEIS, J. COPE and A. FLORENCE, 2003 Paediatric medicines research in the UK: how to move forward? Drug safety : an international journal of medical toxicology and drug experience 26: 529-537. WOODROW, R., 2006 Essentials of pharmacology for health occupations. 1: 80. XU, Z., and J. A. TAYLOR, 2009 SNPinfo: integrating GWAS and candidate gene information into functional SNP selection for genetic association studies. Nucleic acids research 37: W600-605. XU, Z. T., JA, 2009 SNPinfo: Integrating GWAS and Candidate Gene Information into Functional SNP Selection for Genetic Association Studies. Nucleic Acids Research. YAO, J. K., 1988 Biosynthesis of membrane cholesterol during peripheral nerve development, degeneration and regeneration. Lipids 23: 857-862. YI, X., and N. MAEDA, 2006 alpha-Lipoic acid prevents the increase in atherosclerosis induced by diabetes in apolipoprotein E-deficient mice fed high-fat/low-cholesterol diet. Diabetes 55: 2238-2244. ZHAO, Y., M. SONG, D. GUAN, S. BI, J. MENG et al., 2005 Genetic polymorphisms of CYP3A5 genes and concentration of the cyclosporine and tacrolimus. Transplantation proceedings 37: 178-181. ZHENG, H. X., A. ZEEVI, K. MCCURRY, E. SCHUETZ, S. WEBBER et al., 2005 The impact of pharmacogenomic factors on acute persistent rejection in adult lung transplant patients. Transplant immunology 14: 37-42. ZHONG, M., and R. S. MOLDAY, 2010 Binding of retinoids to ABCA4, the photoreceptor ABC transporter associated with Stargardt macular degeneration. Methods Mol Biol 652: 163-176.  138 ZIEGLER, D., A. AMETOV, A. BARINOV, P. J. DYCK, I. GURIEVA et al., 2006 Oral treatment with alpha-lipoic acid improves symptomatic diabetic polyneuropathy: the SYDNEY 2 trial. Diabetes care 29: 2365-2370.    139 Appendix  Appendix A  A.1 Excluded SNPs: Less than 98% Completion Rate Across Patients # Gene SNP rsID Patients genotyped (%) Patients not genotyped (%) 1 DPYD rs11165783 97.14 2.86 2 EPHX1 rs3738043 91.43 8.57 3 XDH rs6718606 92.86 7.14 4 UGT2B28 rs4694709 96.43 3.57 5 UGT2B28 rs10002503 96.43 3.57 6 UGT2B28 rs4580710 96.43 3.57 7 LOC642496 rs4356971 95.71 4.29 8 UGT2A1 rs7671313 96.43 3.57 9 SLCO6A1 rs13358778 90.71 9.29 10 SLC22A4/5 rs162892 95.71 4.29 11 SLC29A1 rs693955 97.14 2.86 12 GSTA1/2/3/4/5 rs9395826 97.14 2.86 13 NOS3 rs3793342 72.86 27.14 14 CYP11B2 rs3097 92.86 7.14 15 RXRA rs10881582 75.00 25.00 16 RXRA rs1536474 96.43 3.57 17 RXRA rs1805352 97.14 2.86 18 SLC22A18 rs2283231 94.29 5.71 19 SLC22A18 rs367035 72.86 27.14 20 SLC22A18 rs413781 75.00 25.00 21 ALDH3B2 rs11603033 96.43 3.57 22 SLCO1B1 rs2306283 92.86 7.14 23 SLCO1B1/1A2 rs12371604 97.14 2.86 24 SLC28A1 rs7166433 97.14 2.86 25 SLCO3A1 rs4932599 92.86 7.14 26 SLC7A5 rs12931876 93.57 6.43 27 PNMT rs2934963 97.14 2.86 28 CYP2B6 rs2279341 89.29 10.71 29 APOE rs429358 90.71 9.29 30 SLC19A1 rs7278425 87.86 12.14 31 TXNRD2 rs6518591 95.71 4.29 32 COMT rs740603 90.71 9.29 33 COMT rs4818 96.43 3.57   140 A.2 Excluded SNPs: Hardy-Weinberg Disequilibrium # Gene SNP rsID P-value* Exact P-value* 1 POP rs2106806 6.2104E-08 0 2 CYBB rs7059081 2.34122E-14 0 3 CYBB rs5917471 0.00013826 8 0.0001 4 DYNLT3 rs5963339 3.39494E-07 0.0002 5 POP rs1927286 2.35327E-10 0.0001 6 POP rs6625400 4.44066E-20 0 7 ABCB7 rs5937937 1.45219E-09 0 8 ATP7A rs2227291 8.56886E-07 0 9 POP rs1023465 5.31666E-12 0 10 POP rs2499043 4.68751E-06 0 11 POP rs1541341 1.77809E-07 0 12 POP rs6637326 1.27807E-12 0 13 DPYD rs2027056 1.93496E-08 0 14 UGT2A1 rs2163659 1.71369E-07 0 15 PPARD rs2267664 1.04848E-08 0.0001 16 PPARD rs3798343 1.04848E-08 0 17 SLC22A7 rs2254303 3.2693E-08 0 18 SLCO5A1 rs2380566 1.43313E-20 0 19 SLCO1A2 rs10770800 7.06918E-08 0 20 SLCO1B1/1A 2 rs10841798 7.06918E-08 0.0001 21 SLC15A1 rs1289393 2.70866E-05 0 22 SPG7 rs3935627 0.00032291 1 0.0003 23 SLC13A2 rs12453704 1.3313E-06 0 24 SULT2B1 rs12611137 0.000 0881 3 0 25 ABCG1 rs7277991 2.99378E-11 0 *Exact p-values obtained by Monte-Carlo procedure with 10,000 permutations. The significance threshold is p- value = 0.000368  141 A.3 Significant SNPs in the ADME-Tox Panel of General Drug Biotransformation  Gene SNP rsID Postulated SNP Function§ Gene Location and Function OR (95% CI)† P-value† 1 PON1 rs854549 unknown flanking_3UTR 3.34 (1.67-6.66) 0.00061* 2 ABCA4 rs3789433 unknown intron 3.11 (1.60-6.02) 0.00075* 3 SLCO1B3/1C1 rs10770704 unknown intron 0.31 (0.16-0.62) 0.00098* 4 ABCA4 rs549848 unknown intron 0.29 (0.13-0.61) 0.00128* 5 ABCA4 rs1889404 regulatory potential intron 2.95 (1.51-5.76) 0.00150* 6 ABCA4 rs4147815 regulatory potential intron 2.94 (1.48-5.82) 0.00192* 7 ALDH9A1 rs7527279 unknown intron 0.29 (0.13-0.64) 0.00210* 8 ABCA4 rs1007347 unknown intron 2.62 (1.40-4.90) 0.00237* 9 ABCB5 rs1029595 unknown flanking_3UTR 2.39 (1.36-4.20) 0.00240* 10 SLC22A16 rs7765849 unknown intron 3.75 (1.55-9.07) 0.00323* 11 ABCA4 rs3789421 unknown intron 2.66 (1.35-5.24) 0.00457* 12 GPX3 rs2161359 unknown intron 2.22 (1.28-3.87) 0.00458* 13 ABCG1 rs221948 unknown intron 3.57 (1.47-8.63) 0.00467* 14 PON2/3 rs10487132 unknown intron 2.34 (1.28-4.25) 0.00518* 15 SLC15A1 rs3782993 regulatory potential intron 3.77 (1.48-9.59) 0.00531* 16 CYP51A1 rs7797834 ESE or ESS splicing, regulatory potential coding (SYNONYMOUS) H453H 2.32 (1.27-4.21) 0.00559* 17 ADH1A/B/C rs9307239 TFBS flanking_5UTR 2.29 (1.26-4.15) 0.00618* 18 CYP51A1 rs7793861 miRNA 3UTR 2.25 (1.25-4.05) 0.00652* 19 ABCC4 rs9524855 unknown intron 2.36 (1.27-4.38) 0.00657* 20 SULT2A1 rs296364 unknown intron 0.51 (0.31-0.83) 0.00740* 21 ADH4/5 rs1312200 TFBS flanking_5UTR 3.05 (1.34-6.94) 0.00754* 22 ABCC4 rs7324283 regulatory potential intron 0.40 (0.21-0.79) 0.00799* 23 ABCC4 rs4148494 unknown intron 0.44 (0.24-0.81) 0.00819* 24 TPMT rs1142345 nsSNP, probably damaging coding (NON-SYNONYMOUS) Y240C 9.30 (1.76-49.1) 0.00857* 25 SLC22A6/8 rs3809069 TFBS, regulatory potential flanking_5UTR 2.56 (1.27-5.19) 0.00864* 26 POR rs11764251 unknown intron 2.05 (1.19-3.51) 0.00890* 27 SLC28A3 rs7029691 unknown intron 0.39 (0.19-0.79) 0.00947* 28 ABCC4 rs10508021 regulatory potential intron 2.19 (1.20-4.00) 0.00987* 29 GPX3 rs3828599 regulatory potential intron 0.40 (0.20-0.80) 0.0102* 30 SLC7A7 rs12884337 regulatory potential intron 2.05 (1.18-3.58) 0.0106* 31 EPHX1 rs2854450 TFBS, regulatory potential flanking_5UTR 2.34 (1.21-4.50) 0.0107* 32 PON2/3 rs6977389 regulatory potential intron 0.48 (0.27-0.84) 0.0110* 33 SLC28A3 rs4877842 unknown intron 0.44 (0.23-0.83) 0.0118* 34 SLCO1B1/1A2 rs7137060 unknown intron 0.29 (0.11-0.76) 0.0124* 35 UGT2B11 rs4400059 unknown intron 0.32 (0.13-0.78) 0.0124*  142  Gene SNP rsID Postulated SNP Function§ Gene Location and Function OR (95% CI)† P-value† 36 ABCC4 rs9634642 unknown intron 1.98 (1.15-3.40) 0.0124* 37 ABCC4 rs9524848 unknown intron 1.91 (1.14-3.19) 0.0127* 38 SLCO4A1 rs2427377 unknown flanking_3UTR 0.42 (0.21-0.83) 0.0131* 39 PON2/3 rs1053275 regulatory potential coding (SYNONYMOUS) A99A 0.49 (0.28-0.86) 0.0133* 40 ABCA4 rs35146614 regulatory potential intron 0.14 (0.03-0.67) 0.0136* 41 ABCA4 rs4147807 unknown intron 2.10 (1.16-3.80) 0.0137* 42 PON1 rs705377 regulatory potential intron 2.09 (1.15-3.76) 0.0142* 43 AOX1 rs4674311 unknown intron 0.48 (0.27-0.86) 0.0146* 44 ABCA1 rs2740486 regulatory potential intron 1.99 (1.14-3.46) 0.0147* 45 SLCO2A1 rs11710021 unknown intron 0.23 (0.07-0.75) 0.0153* 46 SLCO1B1 rs11045819 ESE or ESS splicing, nsSNP, benign coding (NON-SYNONYMOUS) P155T 0.27 (0.09-0.78) 0.0154* 47 UGT1A1-9 rs17863783 regulatory potential coding 4.91 (1.35-17.8) 0.0154* 48 ABCA4 rs950283 regulatory potential intron 0.49 (0.28-0.87) 0.0155* 49 ABCC4 rs17268163 unknown intron 2.36 (1.17-4.75) 0.0156* 50 SLC22A16 rs10214672 unknown intron 2.53 (1.18-5.41) 0.0165* 51 PON1/3 rs13228784 unknown flanking_5UTR 0.41 (0.19-0.85) 0.0166* 52 SLC22A16 rs221712 unknown intron 2.16 (1.14-4.06) 0.0167* 53 AOX1 rs2540064 regulatory potential flanking_3UTR 0.47 (0.25-0.87) 0.0174* 54 SLC31A1 rs10981704 unknown flanking_5UTR 0.48 (0.27-0.88) 0.0177* 55 NNMT rs2301128 unknown intron 2.53 (1.17-5.49) 0.0178* 56 SLCO1A2 rs10841795 ESE or ESS splicing, nsSNP, benign, regulatory potential coding (NON-SYNONYMOUS) I13T 0.25 (0.08-0.79) 0.0180* 57 GPX3 rs10463312 unknown intron 0.49 (0.27-0.88) 0.0184* 58 EPHX1 rs2740168 unknown intron 1.94 (1.11-3.39) 0.0189* 59 SLCO1B3/1C1 rs11045399 unknown intron 0.46 (0.24-0.88) 0.0190* 60 GPX3 rs8177447 regulatory potential intron 0.34 (0.14-0.84) 0.0201* 61 SLC7A7 rs12891079 unknown intron 2.13 (1.12-4.05) 0.0202* 62 CYP7A1 rs10087499 unknown flanking_3UTR 2.12 (1.12-4.02) 0.0203* 63 AOX1 rs2241080 unknown intron 2.07 (1.11-3.85) 0.0204* 64 ABCA1 rs2297406 regulatory potential intron 0.47 (0.24-0.89) 0.0206* 65 SLC22A16 rs2207356 TFBS flanking_3UTR 2.38 (1.14-4.96) 0.0207* 66 ABCA1 rs2740484 regulatory potential intron 2.00 (1.11-3.62) 0.0210* 67 NNMT rs2852432 TFBS flanking_5UTR 0.47 (0.25-0.89) 0.0211* 68 ABCA1 rs2482424 regulatory potential intron 0.34 (0.14-0.85) 0.0215* 69 ALDH7A1 rs4836272 unknown intron 1.87 (1.09-3.21) 0.0226* 70 FMO1/2 rs16864296 unknown intron 4.18 (1.22-14.3) 0.0228* 71 SLCO2A1 rs6439445 unknown flanking_3UTR 0.49 (0.27-0.90) 0.0229* 72 CYP17A1 rs10509762 TFBS, regulatory potential flanking_5UTR 5.57 (1.26-24.5) 0.0233* 73 TRPV1 rs3826501 regulatory potential intron 0.47 (0.24-0.90) 0.0233*  143  Gene SNP rsID Postulated SNP Function§ Gene Location and Function OR (95% CI)† P-value† 74 TPMT rs1800460 ESE or ESS splicing, nsSNP, possibly damaging, regulatory potential coding (NON-SYNONYMOUS) A154T 6.96 (1.29-37.4) 0.0237* 75 ABCC1/6 rs8187843 unknown intron 3.69 (1.18-11.4) 0.0238* 76 CYP26A1 rs2068888 flanking_3UTR 0.53 (0.31-0.92) 0.0241* 77 ABCC2 rs7476245 unknown intron 4.11 (1.20-14.1) 0.0243* 78 ABCC4 rs7330673 unknown intron 0.39 (0.17-0.88) 0.0245* 79 CHST3 rs11000122 TFBS flanking_5UTR 0.49 (0.26-0.91) 0.0248* 80 PON2/3 rs11977702 TFBS flanking_5UTR 0.44 (0.21-0.90) 0.0252* 81 PON1 rs757158 TFBS flanking_5UTR 0.51 (0.28-0.92) 0.0255* 82 SLC28A1 rs12910476 unknown intron 0.41 (0.19-0.90) 0.0260* 83 ABCB1 rs6979885 unknown intron 0.43 (0.21-0.90) 0.0262* 84 FGFR4 rs351855 nsSNP, possibly damaging, regulatory potential coding (NON-SYNONYMOUS) G388R 1.95 (1.07-3.53) 0.0270* 85 TRPV1 rs17632843 regulatory potential intron 0.52 (0.29-0.93) 0.0272* 86 SLCO3A1 rs2238359 regulatory potential intron 0.48 (0.25-0.92) 0.0276* 87 CYP46A1 rs943884 regulatory potential intron 0.28 (0.09-0.87) 0.0281* 88 CYP2C8 rs11572181 unknown flanking_3UTR 5.29 (1.19-23.4) 0.0281* 89 ABCA4 rs4147839 unknown intron 2.01 (1.07-3.76) 0.0286* 90 ALDH1B1 rs2228093 nsSNP, possibly damaging, regulatory potential coding (NON-SYNONYMOUS) A86V 0.40 (0.18-0.91) 0.0288* 91 UGT1A1-9 rs17862838 unknown flanking_5UTR 3.75 (1.14-12.3) 0.0292* 92 SLCO3A1 rs12913189 unknown intron 2.05 (1.07-3.93) 0.0293* 93 AHR rs4236290 unknown intron 0.30 (0.10-0.88) 0.0294* 94 CYP4F12 rs609290 ESE or ESS splicing, nsSNP, benign, regulatory potential coding (NON-SYNONYMOUS) I90V 3.14 (1.12-8.80) 0.0294* 95 ACO1 rs13292540 unknown intron 0.49 (0.26-0.93) 0.0296* 96 ABCG2 rs9999111 unknown intron 0.22 (0.06-0.86) 0.0301* 97 ABCA4 rs3789407 unknown intron 2.05 (1.07-3.94) 0.0303* 98 SLC31A1 rs10817465 unknown flanking_5UTR 1.83 (1.05-3.19) 0.0305* 99 ABCC4 rs9524849 unknown intron 1.96 (1.06-3.62) 0.0306* 100 SERPINA6 rs7141205 unknown flanking_3UTR 0.46 (0.23-0.93) 0.0308* 101 XDH rs635581 unknown flanking_3UTR 1.87 (1.05-3.34) 0.0320* 102 ABCA4 rs17111122 unknown flanking_5UTR 1.82 (1.05-3.16) 0.0325* 103 ABCC4 rs4773850 unknown intron 0.54 (0.30-0.95) 0.0328* 104 AOX1 rs2540066 unknown intron 0.52 (0.29-0.94) 0.0329* 105 ADH4 rs3805322 unknown intron 8.58 (1.18-62.0) 0.0331* 106 ABCC8 rs2355017 unknown intron 1.97 (1.05-3.68) 0.0331* 107 SLCO1B3/1C1 rs4581504 unknown intron 2.00 (1.05-3.80) 0.0334* 108 CYP24A1 rs3787555 unknown intron 1.96 (1.05-3.68) 0.0336* 109 GSTA1/2/3/4/5 rs2608615 TFBS flanking_5UTR 0.18 (0.03-0.88) 0.0341* 110 SLC13A3 rs2425884 unknown intron 1.84 (1.04-3.24) 0.0343* 111 ABCA4 rs2297634 unknown intron 1.78 (1.04-3.05) 0.0345*  144  Gene SNP rsID Postulated SNP Function§ Gene Location and Function OR (95% CI)† P-value† 112 SLCO2A1 rs6763132 unknown flanking_3UTR 1.98 (1.05-3.75) 0.0345* 113 EPHX1 rs4653436 unknown flanking_3UTR 0.49 (0.26-0.95) 0.0346* 114 PON2/3 rs2237585 unknown intron 1.82 (1.04-3.20) 0.0350* 115 NAT1 rs6586714 unknown intron 0.25 (0.07-0.91) 0.0355* 116 AOX1 rs2348025 unknown intron 1.82 (1.04-3.19) 0.0355* 117 DPYD rs1023244 unknown intron 0.18 (0.03-0.89) 0.0362* 118 SLC13A3 rs4810535 unknown intron 1.94 (1.04-3.64) 0.0367* 119 ABCC1/6 rs35593 regulatory potential intron 2.36 (1.05-5.30) 0.0367* 120 SLC13A3 rs761218 unknown intron 0.51 (0.27-0.96) 0.0372* 121 ABCC9 rs704179 unknown intron 0.45 (0.21-0.95) 0.0373* 122 CYP2C19 rs4304697 unknown intron 0.20 (0.04-0.91) 0.0374* 123 SLC22A13/14 rs697331 TFBS flanking_5UTR 2.78 (1.06-7.30) 0.0374* 124 ABCB5 rs12669866 regulatory potential intron 1.75 (1.03-2.97) 0.0376* 125 ABCA4 rs4147798 unknown intron 1.83 (1.03-3.26) 0.0377* 126 ERCC2_KLC2L rs10853773 regulatory potential intron 0.48 (0.24-0.96) 0.0380* 127 ACO1 rs10970969 unknown intron 1.97 (1.03-3.74) 0.0380* 128 SLC7A7 rs2281678 TFBS intron 1.75 (1.03-2.98) 0.0384* 129 SLC22A13/14 rs818816 unknown intron 1.90 (1.03-3.52) 0.0386* 130 ABCC5 rs939335 TFBS flanking_5UTR 1.77 (1.02-3.06) 0.0393* 131 ALDH9A1 rs12408101 regulatory potential 5UTR 0.46 (0.22-0.96) 0.0394* 132 ABCC5 rs3817404 unknown intron 0.50 (0.26-0.96) 0.0395* 133 CYP46A1 rs3783321 regulatory potential intron 1.82 (1.02-3.22) 0.0395* 134 ADH1A/B/C rs1614972 unknown intron 1.87 (1.02-3.41) 0.0397* 135 CHST3 rs7081747 unknown flanking_5UTR 1.73 (1.02-2.93) 0.0400* 136 NOX3 rs6919626 regulatory potential intron 1.73 (1.02-2.94) 0.0400* 137 ABCC8 rs11603988 unknown flanking_3UTR 2.02 (1.03-3.98) 0.0401* 138 ABCC4 rs1887162 unknown intron 0.50 (0.26-0.97) 0.0406* 139 SLC7A7 rs2281677 TFBS, ESE or ESS splicing, regulatory potential 5UTR 1.77 (1.02-3.06) 0.0408* 140 ABCG1 rs4148117 unknown intron 0.57 (0.34-0.97) 0.0408* 141 DPYD rs11587873 unknown intron 2.06 (1.02-4.13) 0.0411* 142 CYP2C18 rs1010570 unknown intron 0.51 (0.27-0.97) 0.0411* 143 CYP24A1 rs2426498 unknown flanking_5UTR 2.14 (1.03-4.47) 0.0413* 144 PON1 rs2074351 unknown intron 0.54 (0.30-0.97) 0.0414* 145 ABCB5 rs10230205 unknown intron 1.98 (1.02-3.84) 0.0416* 146 SLC7A7 rs1805059 regulatory potential coding (SYNONYMOUS) S53S 1.83 (1.02-3.28) 0.0418* 147 CHST3 rs4319439 unknown flanking_5UTR 1.75 (1.02-3.00) 0.0418* 148 ABCB4 rs31652 unknown flanking_3UTR 2.08 (1.02-4.23) 0.0423* 149 COL1A2 rs42524 ESE or ESS splicing, nsSNP, benign, regulatory potential coding (NON-SYNONYMOUS) P549A 1.80 (1.02-3.20) 0.0424*  145  Gene SNP rsID Postulated SNP Function§ Gene Location and Function OR (95% CI)† P-value† 150 FMO4 rs11576306 unknown flanking_3UTR 0.10 (0.01-0.93) 0.0428* 151 SLC22A13/14 rs169196 unknown flanking_3UTR 1.81 (1.01-3.22) 0.0439* 152 SLC22A13/14 rs4679029 TFBS flanking_5UTR 2.67 (1.02-6.99) 0.0440* 153 SLC22A13/14 rs4679032 TFBS flanking_5UTR 2.67 (1.02-6.99) 0.0440* 154 TBXAS1 rs1557967 unknown intron 0.49 (0.24-0.98) 0.0448* 155 GSTO1/O2 rs10491045 regulatory potential intron 3.30 (1.02-10.7) 0.0458* 156 ABCG2 rs2622604 unknown intron 1.84 (1.01-3.38) 0.0459* 157 ABCA4 rs554931 regulatory potential intron 0.48 (0.23-0.98) 0.0462* 158 SLCO1B1/1A2 rs10743413 unknown flanking_5UTR 1.99 (1.01-3.92) 0.0463* 159 FMO1/2 rs12567338 miRNA 3UTR 1.82 (1.00-3.31) 0.0464* 160 SLC22A1/2 rs2450975 unknown intron 0.44 (0.20-0.98) 0.0464* 161 ABCA4 rs6658767 unknown intron 3.25 (1.01-10.4) 0.0465* 162 PON2/3 rs13226149 ESE or ESS splicing, regulatory potential coding (SYNONYMOUS) F21F 0.50 (0.25-0.99) 0.0470* 163 ERCC2 rs3916874 regulatory potential intron 0.50 (0.25-0.99) 0.0478* 164 ALDH4A1 rs9117 miRNA, regulatory potential 3UTR 1.86 (1.00-3.47) 0.0483* 165 ABCC8 rs8192690 nsSNP, benign, regulatory potential coding (NON-SYNONYMOUS) V1572I 0.26 (0.07-0.99) 0.0483* 166 CAT rs7118388 unknown flanking_5UTR 0.55 (0.30-0.99) 0.0484* 167 CYP4Z2P rs12759314 TFBS flanking_5UTR 0.32 (0.10-0.99) 0.0484* 168 RAC2 rs5750401 unknown flanking_5UTR 0.57 (0.33-0.99) 0.0486* 169 ABCC1/6 rs9924755 regulatory potential coding (SYNONYMOUS) A830A 2.05 (1.00-4.19) 0.0486* 170 PON1 rs854560 nsSNP, possibly damaging, regulatory potential coding (NON-SYNONYMOUS) L55M 1.87 (1.00-3.50) 0.0491* 171 SLC7A7 rs5027249 unknown intron 1.93 (1.00-3.71) 0.0494* 172 EPHX1 rs1051740 ESE or ESS splicing, nsSNP, possibly damaging, regulatory potential coding (NON-SYNONYMOUS) Y113H 1.70 (1.00-2.90) 0.0495* 173 RAC2 rs7290922 unknown flanking_5UTR 1.97 (1.00-3.90) 0.0497* The functional effects of the SNP could include: transcriptional regulatio  (transcription factor binding site, TFBS), modulation of splicing pattern or efficiency (exonic splicing enhancers, ESE; exonic splicing silencers, ESS), protein translation effects (miRNA binding sites), regulatory potential, amino acid change (nsSNP), and damaging or benign SNPs. SNPs that tag to 2 or more genes have both genes listed. §Postulated SNP function on gene from NIEHS (XU 2009). †Values derived from a logistic regression model adjusted for principal components, age, and tumor type. The shown p-value is the individual p-value of the genetic component of the logistic regression model. *Statistically significant at 0.05 type I error rate, unadjusted for multiple testing.    146 A.4 PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 Risk Alleles Increase the Chances of Developing Vincristine-Induced Neurotoxicity   Number of Risk Alleles CTCAE v4.03 Grade  2 3 4 5 6 7 8 9  P-value 0  10 (100%) 31 (88.57%) 24 (92.30%) 15 (65.21%) 11 (39.28%) 4 (33.33%) 0 (0%) 0 (0%)  7.13 x 10-10* 2  0 (0%) 1 (2.85%) 1 (3.84%) 4 (17.39%) 11 (39.28%) 3 (25%) 2 (66.66%) 0 (0%) 3  0 (0%) 3 (8.57%) 1 (3.84%) 4 (17.39%) 6 (21.42%) 5 (41.66%) 1 (33.33%) 3 (100%) PON1, ABCA4, SLCO1C1, CYP51A1, and ABCG1 risk alleles increase the chances of developing vincristine-induced neurotoxicity The increase in number of PON1 rs854549, ABCA4 rs3789433, SLCO1C1 rs10770704, ABCG1 rs221948, CYP51A1 rs7797834, and ABCA4 rs849848 risk alleles is associated increased likelihood of developing VIN during chemotherapy treatment.  147 A.5 Imputed Regions with Base Pair Limits Used  Gene Chromosome SNP Position (base pairs) Segment for Imputation (base pairs) ABCA4 1 rs3789433 94,348,028 94,000,001 - 99,000,000 ABCA4 1 rs549848 94,297,444 CYP51A1 7 rs7797834 91,581,086 91,500,001 - 96,500,000 PON1 7 rs854549 94,764,521 PON1 7 rs854560 94,784,020 PPP1R9A 7 rs705377 94,752,793 SLCO1C1 12 rs10770704 20,747,028 20,000,001 - 25,000,000 ABCG1 21 rs221948 42,506,262 41,000,001 - 46,000,000  148 A.6 Manhattan Plot for Imputed SNPs   Manhattan plot for case/control (n = 140) with imputed and not imputed SNPs on regions of chromosomes 1, 7, 12, and 21 (n = 9,397) controlling for prin1, prin2, age, and tumor type with a threshold of 3.68x10-4 !"#$%&'( )*)))+,( !"#!$% #&'()!)*%'+,)*%% -./%''')01!% 23+#)#)% !"#4)%

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072033/manifest

Comment

Related Items