UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The pharmacogenomics of cisplatin-induced hearing loss Pussegoda, Kusala 2012

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

Item Metadata

Download

Media
24-ubc_2012_spring_pussegoda_kusala.pdf [ 1.43MB ]
Metadata
JSON: 24-1.0103451.json
JSON-LD: 24-1.0103451-ld.json
RDF/XML (Pretty): 24-1.0103451-rdf.xml
RDF/JSON: 24-1.0103451-rdf.json
Turtle: 24-1.0103451-turtle.txt
N-Triples: 24-1.0103451-rdf-ntriples.txt
Original Record: 24-1.0103451-source.json
Full Text
24-1.0103451-fulltext.txt
Citation
24-1.0103451.ris

Full Text

THE PHARMACOGENOMICS OF CISPLATIN-INDUCED HEARING LOSS by Kusala Pussegoda B.Sc.H, University of Ottawa, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Medical Genetics)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2012  ©Kusala Pussegoda, 2012  Abstract Cisplatin is a widely used chemotherapeutic agent for the treatment of solid tumours. A serious complication of cisplatin treatment is permanent hearing loss. The study hypothesis is that genetic variants in genes involved in drug metabolism and transport can contribute to increased susceptibility to hearing loss in pediatric oncology patients treated with cisplatin. Patients were recruited from across Canada through the Canadian Pharmacogenomics Network for Drug Safety (CPNDS). Recently, our group identified several predictive genetic variants that were highly associated with cisplatin-induced hearing loss in children. We evaluated whether we could replicate these findings in a new independent cohort of 155 pediatric patients. Associations were replicated for genetic variants in TPMT (rs12201199, P=0.0013, Odds Ratio, OR 6.1) and ABCC3 (rs1051640, P=0.036, OR 1.8). A predictive model combining variants in TPMT, ABCC3 and COMT with clinical variables significantly improved the prediction of risk of developing hearing loss compared to clinical risk factors alone (P=0.00048). We next evaluated whether we could identify additional genetic variants that confer susceptibility to cisplatininduced hearing loss. We identified novel variants in ABCB5 (rs10950831, P=1.06×10-6, OR 2.0) and DPYD (rs6667550, P=0.0047, OR 1.9) that were significantly associated with cisplatininduced hearing loss. We included these variants into the initial genetic model that consists of variants in TPMT, ABCC3 and COMT to evaluate whether we could improve the prediction of risk. We demonstrate that the risk of prediction of hearing loss significantly improves by including genetic variants in ABCB5 and DPYD (P=0.0023). We also demonstrate that by combining the clinical and genetic factors we can significantly improve the prediction of risk of hearing loss compared to clinical factors alone (P=2.63x10-7). We were able to replicate previously described findings and also provide evidence for novel genetic variants in the ii  prediction of cisplatin-induced hearing loss in children. Furthermore, this study demonstrates that predictive models can classify patients based on predicted risk of cisplatin-induced hearing loss. These findings have the potential to influence treatment modifications for cisplatin therapy and may improve safety in children.  iii  Preface This dissertation is comprised of both published and unpublished material as follows: This study was approved by the University of British Columbia Clinical Research Ethics Board (certificate number H04-70358).  iv  Table of Contents  Abstract .......................................................................................................................................... ii Table of Contents .......................................................................................................................... v List of Tables .............................................................................................................................. viii List of Figures ............................................................................................................................... ix Acknowledgments ....................................................................................................................... xii Chapter 1: Introduction ............................................................................................................... 1 1.1 Adverse Drug Reactions........................................................................................................ 1 1.1.1 The Genetic Contribution to Drug Toxicity ................................................................... 3 1.2 Pharmacogenomics................................................................................................................ 4 1.2.1 From Scientific Discoveries to the Clinic....................................................................... 6 1.2.2 Pharmacogenomic Studies in Children........................................................................... 7 1.3 Cisplatin ................................................................................................................................ 8 1.3.1 Cisplatin Use in Children ............................................................................................... 9 1.3.2 Mechanisms of Cisplatin Action .................................................................................... 9 1.4 Cisplatin-Induced ADRs ..................................................................................................... 10 1.4.2 Pathophysiology of Cisplatin-Induced Hearing Loss ................................................... 11 1.4.3 Cisplatin-Induced Hearing Loss in Children ................................................................ 12 1.4.4 Mechanism of Cisplatin-Induced Hearing Loss ........................................................... 13 1.4.5 Clinical Risk Factors .................................................................................................... 14 1.5 Cisplatin Transport – Uptake and Efflux ............................................................................ 15 1.6 Pharmacogenomic Studies of Cisplatin-Induced Hearing Loss .......................................... 16 1.6.1 Polymorphisms with Megalin and Cisplatin-Induced Hearing Loss ............................ 16 1.6.2 Polymorphisms with Glutathione S-Transferases (GSTs) and Cisplatin-Induced Hearing Loss .......................................................................................................................... 18 1.6.2 Variations in Genes in DNA Repair Pathways and Cisplatin-Induced Hearing Loss .. 21 1.6.3 Genes Associated with Hereditary Deafness and Cisplatin-Induced Hearing Loss ..... 23 1.7 Predictive Modelling in Pharmacogenomics....................................................................... 24 1.7.1 Measures of Predictive Tests ........................................................................................ 24 v  1.7.2 Genomic Profiling ........................................................................................................ 25 1.8 Potential Confounders ......................................................................................................... 26 1.8.1 Role of Ancestry in Drug Response and Toxicity ........................................................ 26 1.8.1.1 Ancestry as Confounder in Pharmacogenomic Studies ......................................... 26 1.8.1.2 Controlling for Differences in Ancestry ................................................................ 27 1.8.2 The Importance of Clinical Characterization ............................................................... 28 1.9 Importance of Replication Studies ...................................................................................... 29 1.10 Hypothesis and Thesis Objectives..................................................................................... 30 1  Chapter 2: Replication of TPMT and ABCC3 genetic variants highly associated with cisplatin-induced hearing loss in children ................................................................................ 43 2.1 Introduction ......................................................................................................................... 43 2.2 Material and Methods.......................................................................................................... 45 2.2.1 Patients.......................................................................................................................... 45 2.2.2 Genotyping ................................................................................................................... 46 2.2.3 Statistical Analysis ....................................................................................................... 47 2.2 Results ................................................................................................................................. 50 2.2.1 Patient Characteristics .................................................................................................. 50 2.2.2 Genetic Results ............................................................................................................. 51 2.4 Discussion ........................................................................................................................... 54 1  Chapter 3: Genetic Variants in ABCB5 and DPYD are Associated with Cisplatin-Induced Hearing Loss in Children ........................................................................................................... 69 3.1 Introduction ......................................................................................................................... 69 3.2 Material and Methods ...................................................................................................... 70 3.2.1 Patients.......................................................................................................................... 70 3.2.2 Genotyping ................................................................................................................... 71 3.2.3 Statistical Analysis ....................................................................................................... 72 3.3 Results ................................................................................................................................. 75 3.3.1 Patients.......................................................................................................................... 75 3.2.2 Genetic Results ............................................................................................................. 75 3.3 Discussion ........................................................................................................................... 78 Chapter 4: Discussion and Future Directions .......................................................................... 91  vi  4.1 Strengths and Limitations ................................................................................................ 91 4.2 Replication and Functional Studies ................................................................................. 92 4.3 Investigation of Additional Gene and Gene Variants Involved in Cisplatin-Induced Hearing Loss .......................................................................................................................... 94 4.4 Implementing Genetic Testing ........................................................................................ 94 4.5 Conclusions ..................................................................................................................... 95 References .................................................................................................................................... 97  vii  List of Tables Table 1.1: Pharmacogenomic biomarkers in FDA drug labels ..................................................... 33 Table 1.2: Criteria used to determine the grade of cisplatin-induced hearing loss in patients ..... 34 Table 1.3: Select studies describing cisplatin-induced hearing loss clinical risk factors ............ 35 Table 1.4: Studies associated with megalin (LRP2) and cisplatin-induced hearing loss ............. 36 Table 1.5: Studies associated with GSTM3*B and cisplatin-induced hearing loss...................... 37 Table 1.6: Studies associated with GSTP1 and cisplatin-induced hearing loss ........................... 38 Table 1.7: Studies associated with GSTM1 or GSTT1 and cisplatin-induced hearing loss ......... 39 Table 1.8: Studies associated with DNA repair pathway and cisplatin-induced hearing loss ..... 40 Table 1.9: Studies associated with hereditary deafness genes ..................................................... 41 Table 1.10: Calculation of sensitivity, specificity, positive and negative predictive value .......... 42 Table 2.1:Patient demographics .................................................................................................... 59 Table 2.2 Genetic variants associated with cisplatin-induced hearing loss .................................. 60 Table 2.3 Association between additional top ranked SNPs and cisplatin-induced hearing loss . 61 Table 2.4 Genetic variants associated with cisplatin-induced hearing loss in patients of European ancestry ......................................................................................................................................... 62 Table 2.5: Risk group comparisons and grade of hearing loss .................................................... 63 Table 2.6: Comparison of risk groups.......................................................................................... 64 Table 3.1 Patient demographics of the combined cohort .............................................................. 83 Table 3.2 Association between SNPs and cisplatin-induced hearing loss .................................... 84 Table 3.3: Comparison of risk groups........................................................................................... 85  viii  List of Figures Figure 1.1: Phase I and II drug metabolizing enzymes ................................................................. 31 Figure 1.2: Candidate genes involved in the metabolism of platinum drugs................................ 32 Figure 2.1: Distribution of ancestry of patients treated with cisplatin ......................................... 65 Figure 2.2: Receiver operating characteristic (ROC) curves of clinical and genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort ..................................... 66 Figure 2.3: Kaplan-Meier curve of cisplatin-induced hearing loss in three different risk groups (Table 2.6) combining genetic factors. ......................................................................................... 67 Figure 2.4: Kaplan-Meier curve of cisplatin-induced hearing loss in three different risk groups (Table 2.6) combining both clinical and genetic information....................................................... 68 Figure 3.1: ROC curves of genetic variables in the initial model for the prediction of cisplatininduced hearing loss in the combined cohort ................................................................................ 86 Figure 3.2: ROC curves of new genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort .......................................................................................................... 87 Figure 3.3: ROC curves of clinical and newly identified genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort ................................................................. 88 Figure 3.4: Kaplan-Meier curve of cisplatin-induced hearing loss in different risk groups (Table 3.3) combining genetic factors. ..................................................................................................... 89 Figure 3.5: Kaplan-Meier curve of cisplatin-induced hearing loss in different risk groups (Table 3.3) combining clinical and genetic factors. ................................................................................. 90  ix  List of Symbols, Abbreviations SNP  Single nucleotide polymorphism  COMT  Catechol O-methyltransferase  TPMT  Thiopurine S-methyltransferase  ABCC2  ATP-binding cassette transporter, subfamily C, member 2; multiple drug resistance protein 2 (ABCC2)  ABCC3  ATP-binding cassette transporter, subfamily C, member 3; multiple drug resistance protein 3 (ABCC3)  ABCB5  ATP-binding cassette transporter, subfamily B, member 5; multiple drug resistance protein 5 (ABCB5)  ABCG2  ATP-binding cassette transporter, subfamily B, member 2; multiple drug resistance protein 2 (ABCG2)  ADME  Absorption, Distribution, Metabolism, Excretion, and Toxicity  ADR  Adverse Drug Reaction  AUC  Area Under the Curve  CADRMP  Canadian Adverse Drug Reaction Monitoring Program  CI  Confidence Interval  CPNDS  Canadian Pharmacogenomics Network for Drug Safety  CTCAE  Common Terminology Criteria for Adverse Events  CTR1  Solute Carrier 31, Member 1  CYP  Cytochrome P450  DYPD  Dihydropyrimidine Dehydrogenase  FDA  U.S. Food and Drug Administration  GST  Glutathione S-Transferase  LD  Linkage Disequilibrium  MAF  Minor Allele Frequency x  MAPK  Mitogen-Actived Protein Kinase  NOX-3  NADPH oxidase 3  NPV  Negative Predictive Value  OR  Odds Ratio  PC  Principal Component  PD  Pharmacodynamics  PK  Pharmacokinetics  PPV  Positive Predictive Value  ROC  Receiver Operating Characteristics  SJS  Stevens-Johnson Syndrome  SLC31A1  Solute Carrier 31, Member 1  TEN  Toxic Epidermal Necrolysis  UGT  Uridine Diphosphate Glucuronosyltransferase  VKORC1  Vitamin K Epoxide Reductase  xi  Acknowledgments First and foremost, I am deeply grateful to my research supervisor Dr. Michael Hayden for his supervision, guidance and support. He has given me the opportunity to be involved in a variety of different aspects of scientific research which I have learned from and gained much knowledge. Through this experience I have learned to challenge myself, to become better and to reach higher. I would like to sincerely thank my advisory committee, Dr. Colin Ross, Dr. Rod Rassekh and Dr. Jan Friedman for their expertise and insightful comments. They have provided me with support, wisdom, encouragement and guidance throughout my thesis work. I sincerely thank Dr. Bruce Carleton for his advice, expertise and discussions on science and life in general. I would also like to thank other members of the lab. In particular, Johanna Sistonen, Mojgan Yazdanpanah, Ursula Amstutz, Henk Visscher, Tenneille Loo and Kaityln Shaw for their friendship, expertise, constructive advice, encouragement, and inspiring discussions. This thesis would not be possible without the patients and their families participating in this study and their dedication to supporting research in drug safety. I thank members of the Canadian Pharmacogenomic Network for Drug Safety including investigators, technicians, surveillors, and collaborators for their support. A special thanks to my friends for sticking by me during times of both great joy and despair and reminding me that there is life outside of the laboratory. Finally, I would like to send my deepest gratitude to my parents and brother for their unconditional love, encouragement, support and belief in me through this journey.  xii  Chapter 1: Introduction 1.1 Adverse Drug Reactions In modern drug development, clinical trials focus on the safety and efficacy of medications at standard doses. The inherent differences within the population are not taken into account or investigated in the drug development pipeline. It is now well recognized individual variability in response to medications range from a lack of therapeutic benefit to life-threatening adverse drug reactions(ADRs)[1]. An ADR is defined as a harmful, undesirable reaction to a medication which justifies methods for prevention, dose modifications or cessation of therapy[2].  ADRs can vary from mild to severe reactions, those that are permanent to ones that reverse once medication is discontinued. The variability in drug response has a major impact on health care costs and quality of life. In some cases, severe ADRs can result in hospitalizations and even put patients at risk of morbidity and mortality. Each year greater than 2 million severe ADRs claim over 100,000 lives in the United States, ranking them as the 5th leading cause of death[3-5]. In the United States alone, ADRs are estimated to cost between 30 to 130 billion dollars annually[3, 4, 6]. The economic burden associated with ADRs can be higher than for conditions such as obesity, where the annual costs ranges from 24 to 70 billion dollars[7]. A more recent prospective study conducted in the United Kingdom to address the current burden of ADRs through an analysis of 18,820 hospital admissions, attribute 1225 admissions (6.5%) to an ADR[8]. The costs associated with ADRs can pose a significant burden on the healthcare system not only by the treatment required for the ADR itself but also by the monitoring required to screen all patients for potential ADRs. Clozapine-induced agranulocytosis is an example in  1  which patients are frequently monitored for their white blood cell count through frequent blood tests[6, 9].  ADRs can range from minor reactions to severe-life threatening events such as heart failure. These reactions can occur within minutes after treatment or years after first exposure to the medication. For example, most carbamazepine-induced Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) cases occur within the first 8 weeks of exposure to the drug[10, 11]. Severe adverse effects may not even be observed until the drug is on the market as clinical trials are limited by study size. It may take millions of individuals exposed to the drug before unexpected adverse events emerge[12]. When regulatory authorities deem that the risk of these adverse events outweigh the benefit, medications are forced to be removed from the market.[12] However, drug withdrawals potentially leave patients that benefitted from the therapeutic effect of the medication without access to treatment. An example of this is Vioxx, a potent arthritis pain killer. Once released into the market, reports of heart attack and stroke resulted in the drug being pulled from the shelves[12].  This highlights the importance of post-market surveillance of drugs once available to the general public in order to identify and assess of adverse events of medications. Health Canada encourages health professionals to report ADRs to identify unknown adverse events and to assess the balance between risks and benefits of medications to improve safety once on the market. However, despite the obvious benefit of reporting ADRs, it is estimated that approximately 90% of all ADRs are never reported to regulatory agencies[13]. There are several studies that assess reasons of under-reporting which include doubt of association with an ADR,  2  the reaction is too well known to report, failure to recognize an adverse event, lack of time, and not knowing how to report an ADR to regulatory agencies[14-16]. A study evaluating TEN highlights how even life-threatening ADRs are under-reported in Canada. They identified that only 4-10% of cases of TEN were reported to the Canadian Adverse Drug Reaction Monitoring Program (CADRMP)[17].  ADRs can significantly impact patient health and result in morbidity and mortality, permanent disability and treatment cessation. It is suggested that many ADRs can be prevented; suggesting that increased effort should be placed on monitoring and reporting of adverse events[5]. Although many risk factors can contribute to the development of an ADR, in most cases it remains unclear how they influence the efficacy and toxicity of the drug.  1.1.1 The Genetic Contribution to Drug Toxicity Drug toxicity is often associated with drug concentrations both in the plasma and target location. Drug concentrations are dependent on key processes involved in absorption, distribution, metabolism and excretion (ADME). Drug metabolism and excretion are processes that inactivate and detoxify drugs to protect the human body from potential harmful adverse effects. Metabolism can be divided into phase I and phase II reactions which convert the parent compound into metabolites that are usually more water soluble for excretion through the kidney. Phase I metabolism involves functional group modifications through oxidation, reduction and hydrolysis reactions. A majority of Phase I metabolism enzymes belong to the cytochrome P450 (CYP) family of enzymes (Figure 1.1)[18]. Phase II enzymes are involved in conjugation reactions and include enzymes such as uridine diphosphate glucuronosyltransferases (UGTs), 3  thiopurine S-methyltransferase (TPMT) and glutathione S-tranferases (GSTs)[18]. These enzymes are known to exhibit genetic polymorphisms that may contribute to variability in drug response and lead to clinical consequences including ineffective drug treatment and druginduced toxicity.  1.2 Pharmacogenomics Pharmacogenomics is the study of the influence of individual genetic variation on drug response. The field of study dates back to the 1900‟s when individuals such Sir Archibald Garrod observed growing clinical evidence that individuals responded differently to standard doses of medications usually accompanied by variations in the concentrations of drug metabolites[1]. It was not until the 1950s that Arno Motulsky wrote an article on the genetic basis of adverse drug reactions[19]. In 1959, Friedrich Vogel introduced the field as “pharmacogenomics”[19]. Research is now focused on the potential to shape the practice of medicine to treat patients at the individual level. In current practice patients are dosed based on standard protocols, adjusting for body surface area[20]. Many non-genetic factors can contribute to large population differences observed with response to therapy such as age, concomitant medications, drug interactions, organ function, patient health and environment[20]. Unlike non-genetic factors that change throughout a patient‟s lifetime, clinicians can utilize genetic variants that remain stable to reliably identify patients at risk of adverse events[21]. Genetic factors such as single nucleotide polymorphisms (SNPs) are known to play a role in the incidence and severity of ADRs[1, 22]. For example, clinical factors such as age, body surface area, gender, and dietary intake of vitamin K account for approximately 12% of the dose variability in patients receiving warfarin[23, 24]. However,  4  one study has described that 25% of the variability in warfarin dose is due to variation in vitamin K epoxide reductase (VKORC1)[25]. Genetic variation plays a role in drug response through pharmacokinetics (PK) and pharmacodynamics (PD). PD is the study of the effect of drugs on the body. This involves the interaction of the drug with its target site and is responsible for the relationship between drug concentration and effect[26]. In contrast, PK is the study of how the body affects the extent and rate of the ADME of drugs[27]. Differences in the genetic variants in genes encoding drugmetabolizing enzymes and transporters can substantially change the PK and PD pathways involved in drug metabolism[28]. For example, a study examined the role of CYP2C19 polymorphisms on the PK of clopidogrel and evaluated whether genotypic differences influenced the variability of response to the drug. It is estimated that up to 4-30% of patients do not show an adequate anti-platelet response with clopidogrel[28]. This study was able to show that individuals that were poor metabolizers (i.e., CYP2C19*2/*2 or CYP2C19*2/*3) responded poorly to the anti-platelet effect of clopidgrel[28]. This highlights how genetic variants can influence drug response, particularly through changes in the expression and activity of genes that influence PK and drug ADME. Genetic variation may also explain why some individuals experience ADRs whereas the drug is safe and effective in other individuals[29]. A major goal of pharmacogenomic studies is to identify genetic factors involved to predict and prevent severe ADRs from occurring after treatment. With the sequencing of the human genome and advancement in genomic technology, it is becoming increasingly more possible to screen individuals for key variants involved in drug metabolism and identifying patients at risk prior to implementing therapy[30]. Pharmacogenomics research has advanced our understanding of the mechanisms of actions of drugs including warfarin and clopidogrel[31]. Pharmacogenomics not 5  only has the potential to advance our knowledge of the PK and PD of medications, but also may improve the management and safety of these medications in patients[31]. 1.2.1 From Scientific Discoveries to the Clinic The variability in drug response has led to the increasing concerns by patients, clinicians, and policy makers about the safety of medications. The withdrawal of medications from the market is not an ideal solution as it leaves individuals that benefit from treatment without access to an effective therapy. By identifying patients that are at high risk of adverse events prior to treatment we can improve the safety of medications. Furthermore, these medications remain accessible to patients for whom there is a therapeutic benefit.  The translation of pharmacogenomic scientific discoveries are of importance not only to patients and medical research centers, but also within the pharmaceutical industry governed by drug regulatory agencies including Health Canada and the U.S. Food and Drug Administration (FDA). The release of the FDAs‟ pharmacogenomic guidelines in 2004 highlighted the importance of genetic testing for medications. The FDA has established a list of approximately thirty different pharmacogenomic markers that are mentioned in approved drug labels[32]. However, of these drugs, only four have pharmacogenomic tests available with a level of evidence where a recommendation can be made for their use in practice. Currently, FDA requires genetic testing mainly for oncology related drugs that affect efficacy and response to therapy such as Her2/neu for Trastuzmab (Table 1.1)[32]. The FDA has also recommended pharmacogenomic testing in their label for five drugs including carbamazepine, where HLA-B*1502 is recommended prior to therapy to prevent life-threatening reactions such as SJS and TEN (Table 1.1)[32]. The overarching goal of pharmacogenomics, is to implement pharmacogenomic evidence-based criteria to 6  help guide clinicians to better evaluate the risk-benefit profile of medications in order to provide optimal treatments for patients.  1.2.2 Pharmacogenomic Studies in Children ADRs are of particular concern in the pediatric population because less than 25% of approved pharmaceuticals have been tested in children. As a result, there is a gross lack of guidelines for safety and efficacy in children[3]. In drug development, clinical trials that evaluate toxicity and efficacy of medications are often not carried out in children due to ethical concerns or challenges such as informed consent[33]. Children have a limited capacity to understand the risks associated with participating in research which justifies why a child under the age of 18 cannot legally give consent to participate in a clinical trial. A guardian or parent has the ability to consent the child; however there has been debate as to whether the parent is acting only in the child‟s best interest[33]. Consequently, dosing guidelines in children are frequently extrapolated from adult clinical data and adjusted based on weight and/or body surface area[34]. We now understand that this simplified dosing approach is not an adequate long-term solution. Children cannot be treated as small adults because age-related differences in the expression and activity of drug metabolizing enzymes and transporters contribute to the altered metabolism and excretion of drugs in children compared to adults. For example, the level of activity of some enzymes in the CYP family increase within the first three months of development[35]. These changes can explain why drug clearance of carbamazepine for example, is greater in children than adults[35]. The Phase II enzyme UGT2B7 is another example that highlights the difference in activity of enzymes during development. The increase in activity of UGT2B7 with age increases the clearance of morphine from plasma[35]. Furthermore, developmental changes in renal function can also alter clearance of drugs from the plasma resulting in differences in 7  elimination of compounds[35]. For this reason, inferring the PK and PD behavior of a drug extrapolated from adult data may not provide accurate dosing guidelines in children. Furthermore, these changes can significantly impact drug concentrations and can result in the exposure of infants and children to toxic drug levels.  Safe and effective drug therapy in pediatric patients requires an understanding of the developmental differences in drug metabolism and clearance in children. The contribution of genetic variants to the variability in drug response and ADRs has been investigated primarily in adult patients. Pharmacogenomic studies in children can help elucidate the mechanisms that affect the variability of drug response and toxicity in this unique population.  1.3 Cisplatin Cisplatin (PLATINOL®) is among the most potent anti-cancer drugs. It is an effective and widely used chemotherapeutic agent for the treatment of solid tumours. In adult patients, cisplatin is used to treat several cancers including ovarian, testicular, lung and head and neck tumours[36, 37]. Cisplatin is currently used as a standard of care treatment for many types of cancer in children, including neuroblastoma, osteosarcoma, germ cell tumours, brain tumours and hepatoblastoma[38]. In fact, cisplatin has shown such an efficacy in standard risk hepatoblastoma that it can be used as monotherapy with an over 80% 3-year event-free survival[39]. Cisplatin is often incorporated in combination with other chemotherapeutic agents such as vincristine, etoposide and doxorubicin. Combination therapies have resulted in high cure rates including testicular cancer patients where the overall cure rate is over 90%[40].  8  1.3.1 Cisplatin Use in Children There are approximately 16 new children treated with cisplatin annually in British Columbia alone which translates to an estimated 136 children annually treated with cisplatin in Canada. A number of different factors are used to determine how cisplatin should be administered in children including the tumour type, the stage of the tumour and the treatment protocol. Overall, children are treated with a bolus of cisplatin for most tumour types. However, the treatment of all extra-cranial germ cell tumours is unique as these patients are treated at low individual doses of 20mg/m2 over five days[41]. These differences in how cisplatin is delivered to patients can affect duration of treatment, dose intensity and dose frequency and can influence the toxicity associated with treatment. 1.3.2 Mechanisms of Cisplatin Action Cisplatin is a cell-cycle non-specific antineoplastic agent which is part of a class of antitumour agents that act at all stages in cell cycle[42]. Cisplatin forms covalent bonds to the N7 positions of purine bases in DNA forming intra-strand and inter-strand crosslinks[43]. These modifications damage and distort the structure of the double-stranded DNA. This affects proteins including hMSH2, a component of the mismatch repair complex and transcription binding factors such as the TATA binding protein which are associated with cellular processes including cell-cycle arrest, transcription, and DNA repair[44, 45]. Cisplatin activates signal transduction pathways including those involved in p53, p73 and mitogen-actived protein kinase (MAPK) which then activate programmed cell death (apoptosis) in tumour cells[44, 45]. However, when this signalling pathway is altered, resistance may occur, a limitation for cisplatin chemotherapy[44].  9  1.4 Cisplatin-Induced ADRs Cisplatin is limited in its‟ clinical application due to serious adverse effects including nephrotoxicity, neurotoxicity, and ototoxicity. Although nephrotoxicity has been reported, it can be prevented by saline hydration therapy or diuresis[46, 47]. However, there are no proven therapeutic treatment options available for neurotoxicity and ototoxicity[47]. Cisplatin is one of the most ototoxic drugs currently in the market. Cisplatin is responsible for the severe irreversible hearing loss, which affects 10-25% of adults and 26-90% of children depending on dose, treatment regimen and degree of hearing loss measured[48-53]. Furthermore, patients show no improvement in hearing and often the hearing loss continues to progress and increase in severity after the end of therapy[54]. Current clinical practice involves dosing to toxicity. Dose reductions or termination of cisplatin treatment are suggested in standard treatment protocols once signs of significant hearing loss (grade 2 or greater) are observed, thereby significantly limiting the use of cisplatin therapy in susceptible patients[55]. Although alternative chemotherapy agents exist that are less ototoxic, such as carboplatin and oxaliplatin, these treatments may not be as effective in treating certain tumours[41]. For example, cisplatin is more effective to treat adult extracranial germ cell tumours than carboplatin, and therefore is the therapy of choice[41]. Furthermore, other adverse effects including neurotoxicity and nephrotoxicity may be more prominent in these other chemotherapeutic agents. Therefore, discontinuation of cisplatin is often not a simple solution to prevent severe ototoxicity from occurring. However, there are cases where alternative medications for certain tumour types have been shown to be equally effective, less ototoxic, and are options within the current standards of care. For example, in Europe, germ cell tumours are treated using carboplatin, etoposide and 10  bleomycin, a treatment regimen that has similar outcomes to treatment with cisplatin, etoposide and bleomycin but with less ototoxicity[56]. There are also significant costs associated with cisplatin-induced hearing loss. Health-care costs for patients that develop severe hearing loss are mainly due to education and development, hearing aids and/or cochlear implants[57]. An economic cost analysis estimated lifetime cost of serious cisplatin-induced hearing impairment is between $445,447 and $562,198 per patient[57]. Assuming that an alternative medication to cisplatin exists that is as effective for treatment, the costs avoided due to cisplatin therapy in Canada are estimated to be $19.6 million by administering a genetic test for cisplatin-induced hearing loss[57].  1.4.2 Pathophysiology of Cisplatin-Induced Hearing Loss Pure-tone audiometry measurements are used to determine the degree of hearing loss in patients. Toxicity can result in significant disability, highlighting the importance of careful and frequent monitoring of auditory function in patients. In patients receiving the same dose of cisplatin, the inter-individual variability of cisplatin-induced effects on hearing is profound, ranging from no hearing loss, to high frequency hearing loss and progresses to severe hearing impairment in the speech frequencies[58, 59]. It is estimated that 50% of consonants in the English language are pronounced in the high frequency range including the sounds “th”, “f”, “k”, and “s”[60]. With signs of hearing loss above 4000Hz, these sounds become indistinguishable to the affected individual[60]. In clinical practice, it is recommended that audiograms be obtained prior to therapy, prior to each round of treatment, and upon any signs of hearing loss. These are usually offered as standard care by medical service plans. However, follow-up tests may be missed by patients in rural areas, until severe hearing loss occurs, at which time there are no preventative 11  options to consider. There are currently no guidelines as to when audiogram hearing tests should be conducted post therapy. Several different grading criteria for hearing loss have been developed over the years. The most commonly used methods are the Common Terminology Criteria for Adverse Events (CTCAE)[61] and the Brock scale[49] (Table 1.2). However, various other grading schemes exist such as the Muenster classification for early detection of hearing loss[62] and the Chang scale (Table 1.2)[55]. Patients are classified into different categories from individuals who do not develop hearing loss (Grade 0) to individuals who develop mild to severe hearing loss (Grade 1-4). Individuals diagnosed with hearing impairment of grade 3 or grade 4 generally require intervention including hearing aids and in extreme cases, a cochlear implant. Furthermore, patients show no improvement in hearing and often the progression and severity of hearing loss continues long after the end of therapy[54].  1.4.3 Cisplatin-Induced Hearing Loss in Children It is well known that children are at the greatest risk of developing hearing loss from cisplatin treatment [48]. This is because cisplatin-induced hearing loss is dependent on age. Studies have shown that hearing loss with therapy is more prevalent in younger children[55, 63]. It is particularly devastating in children as they are at a critical stage of development. In children, even mild losses in hearing can significantly influence the overall quality of life including speech and language development, social-emotional development and increase the risk of learning difficulties[64, 65]. Furthermore, once detected, children show no improvement in hearing and often hearing loss progresses in severity long after treatment has ended[58, 59]. In patients  12  experiencing severe hearing loss, hearing aids and other assistive devices are recommended. The use of hearing aids may be delayed until hearing loss has stabilized and treatment is complete. Although these devices can improve a child‟s hearing, they do not completely correct hearing loss. This suggests that by detecting hearing loss early or prior to treatment will allow children at risk of cisplatin-induced hearing loss to potentially have improved quality of life.  1.4.4 Mechanism of Cisplatin-Induced Hearing Loss Cisplatin damage occurs by apoptosis at three sites in the cochlea: the outer hair cells in the organ of Corti, the spiral ganglion and the stria vascularis[45, 66]. Cisplatin-induced hearing loss is difficult to treat, primarily as the mechanisms involved in this process are not well understood. The generation of reactive oxygen species (ROS) are involved in the toxicity associated with cisplatin therapy[67, 68]. Damage to the hair cells occurs as a consequence of antioxidant depletion, increased lipid peroxidation and an increase in ROS such as superoxide and hydrogen peroxide in the cochlea[69]. This can lead to calcium influx and activation of apoptotic pathways[69]. Rybak et al have demonstrated that NADPH oxidase 3 (NOX-3), expressed in the inner ear generates low levels of ROS species[69]. When activated, NOX-3 increases the production of ROS in the cochlea leading cell death[69]. A study has shown that nitroxide can react with ROS leading to the activation of caspase-3[70]. Studies in cisplatin treated guinea pigs have shown that activation of caspase-3 can lead to apoptosis in cells of the stria vascularis and spiral ligament in the cochlea[70]. The increase in free radicals can lead to morphological and functional changes in the organ of Corti[71]. The production of ROS changes acoustic  13  transduction by modulating the outer hair cell motility in the organ of Corti leading to cell death and consequently leading to hearing loss[72].  1.4.5 Clinical Risk Factors Several studies have suggested that several clinical risk factors contribute to cisplatin-induced hearing loss. Non-genetic factors such as high cumulative cisplatin dose, younger age at time of treatment, concomitant use of aminoglycosides, pre-existing inner hearing impairment, and cranial irradiation are known to influence cisplatin-induced ototoxicity (Table 1.3)[59].  High cumulative dose has been described as a risk factor for cisplatin-induced hearing loss(Table 1.3)[48, 73, 74]. Studies have shown cumulative doses greater than 400mg/m2 in pediatric patients[48, 75]. Many studies have confirmed that younger patients are at increased risk of developing hearing loss[48, 55, 73]. One study has demonstrated that pediatric patients less than 5 years of age were at significantly higher risk of developing hearing loss with cisplatin therapy than patients between 15 or 20 years of age (P=0.001) (Table 1.3)[48]. They found that 40% of children younger than 5 years would develop severe hearing loss, compared to 5% of patients between 15 to 20 years[48]. Bokemeyer et al has also observed that patients treated with concomitant vincristine were at significantly higher risk of developing hearing loss (P=0.001)[73]. Several studies have also shown aminoglycoside antibiotics, which are widely used for the treatment of gram negative bacterial infections, can cause dose-related hearing loss[76, 77]. Aminoglycosides are also suggested to affect sensory inner ear hair cells, and thus may be a confounding risk factor to cisplatin ototoxicity[77]. In children, radiotherapy is often part of the treatment protocol for brain tumours, such as medulloblastoma, because of the  14  positive progression free survival (86% at 3 years and 79% at 5 years)[78]. A serious adverse event associated with radiotherapy is sensorineural hearing loss that affects the inner ear and cochlea. Studies have observed that the incidence of hearing loss with concurrent cisplatin therapy and cranial irradiation increases with increasing radiation dose (Table 1.3)[74]. In fact hearing loss may be observed at doses as low as 45Gy at frequencies greater than 2000Hz[74, 79]. Other potential risk factors have been evaluated, such as gender[74], but have not been significantly associated with cisplatin-induced hearing loss.  1.5 Cisplatin Transport – Uptake and Efflux Early studies have suggested that cisplatin uptake occurs by passive diffusion. However, more recent studies suggest a number of active uptake and efflux mechanisms play a role in drug resistance and transport of cisplatin[80]. The ATP-binding cassette (ABC) family of genes are well known for their ability to transport drugs across the cell membrane. Studies have shown ATP-binding cassette transporter C 2 (ABCC2) overexpression enhance the resistance to cisplatin and therefore can decrease the intracellular drug concentrations, suggesting it may play a role in promoting cisplatin efflux (Figure 1.2)[81, 82]. There have been some controversial results about whether ATP-binding cassette transporter G 2 (ABCG2) is a transporter for platinum compounds. Several studies have shown that ABCG2 does not transport cisplatin[83]. However, a recent study with oxaliplatin has reported that with increased expression of ABCG2, there is increased resistance to chemotherapy, suggesting it may be a transporter of platinum compounds[84]. Copper efflux transporters ATP7A and ATP7B seem to be involved in the accumulation of cisplatin in tumour cell lines[85]. Recent studies have demonstrated that copper transporter 1 (CTR1; also known as solute carrier 31 (copper transporters), member 1,  15  SLC31A1) mediates the influx of cisplatin and is responsible for the accumulation of the drug intracellularly[86]. With controversial results and the limited number of studies investigating cisplatin transport, more research is required to fully understand the transporters involved in cisplatin influx and efflux.  1.6 Pharmacogenomic Studies of Cisplatin-Induced Hearing Loss Clinical criteria are not currently used to identify individuals who are at an increased risk of developing hearing loss. Genetic variation in the genes involved in ADME have been recognized to influence patient drug response and susceptibility to adverse drug events, including ototoxicity[29]. The following studies have described genetic variants that were associated with cisplatin-induced hearing loss.  1.6.1 Polymorphisms with Megalin and Cisplatin-Induced Hearing Loss There is a three to five-fold increased accumulation of platinum-DNA adducts in the marginal cells of the stria vascularis compared to the organ of Corti[87]. This suggests that a membrane pump expressed in the stria vascularis is responsible for the high accumulation of these platinumDNA adducts. Megalin is an endocytic receptor expressed in marginal cells of the stria vascularis of the inner ear and renal proximal tubular cells[88]. Megalin has also been shown to transport aminoglycosides which are known ototoxic drugs[89]. Several studies have been carried out to determine whether polymorphisms in megalin are associated with cisplatininduced ototoxicity.  16  Riedemann et al graded patients based on the Muenster classification for early detection of cisplatin-induced hearing loss[90] (Table 1.4). Seventy-four pediatric patients treated with cisplatin were recruited. Patients with risk factors (prior hearing loss, <5 years of age, cranial irradiation, ototoxic medications and renal impairment) were excluded from the study. The study cohort consisted of 50 patients; 25 controls and 25 cases treated for a variety of malignancies. The controls consisted of patients with no hearing loss (Grade 0) and mild hearing loss (Grade 1). Patients in the study ranged from children to young adults. Two nonsynonymous SNPs (rs2075252 and rs4668123) in the megalin gene (LRP2) that occur with a high frequency in the Caucasian population were evaluated. A significant association was discovered between the A-allele of LRP2 rs2075252 (Glu4094Lys) and cisplatin-induced hearing loss. The risk „A‟ allele of rs2075252 was carried by 13 (52%) cases and 6 (24%) controls.  A subsequent independent study carried out in children did not show an association with the Megalin rs2075252 variant and cisplatin-induced hearing loss[64]. Several differences in methodology and phenotyping may explain the discrepant results between studies. Unlike Riedmann et al, patients with confounding risk factors were not excluded in the latter study. Rather the authors corrected the genetic associations for these clinical cofounders in the analysis. Furthermore, patients were phenotyped using a different grading scheme (CTCAE) and those with mild hearing loss (Grade 1) were excluded in these subsequent studies to evaluate more extreme phenotypes. Cisplatin-induced hearing loss often increases in severity over time, even after completion of cisplatin treatment (i.e. patients with Grade 1 hearing loss progress to Grade 2 to 4)[91, 92]. Therefore, patients with Grade 1 hearing loss may have been misclassified in the Riedmann et al cohort.  17  Overall, given the lack of replication and the low number of patients (n=25 ototoxicity cases) in the Riedmann et al study, it is difficult to determine whether the association of rs2075252 with cisplatin ototoxicity is a true association or whether it has been over-estimated resulting in a false-positive finding.  1.6.2 Polymorphisms with Glutathione S-Transferases (GSTs) and Cisplatin-Induced Hearing Loss GSTs play a role in the amount of free platinum compounds that bind to DNA by catalyzing conjugation of these compounds to glutathione. GSTs have been found to play a significant role in the detoxification of platinum drugs through conjugation of the active metabolite to glutathione and exporting it out of the cell[93, 94]. GSTs are expressed in mammalian cochlea and animal studies have suggested that GSTs can also protect against oxidative damage in the cochlea that leads to ototoxicity[95]. This suggests GSTs may play a role in the development of cisplatin ototoxicity. Several studies have therefore examined whether polymorphisms that decrease expression in GST genes such GSTM1 and GSTT1 are associated with cisplatin-induced hearing loss (Tables 1.5-1.9) [96-99]  1.6.2.1 GSTM3*B Two studies have found associations between cisplatin-induced hearing loss and genetic variants in GST genes. Peters et al identified an association between cisplatin-induced hearing loss and the GSTM3*B variant, a three-base pair deletion that creates a recognition site for a transcription factor[96]. GSTM3*B was carried at a higher frequency in control patients and thus proposed to  18  be protective against cisplatin-induced hearing loss (OR 5.37, P=0.02). This study was carried out in a relatively small cohort of pediatric patients (n=39) treated for a variety of tumour types in children and did not correct the genetic findings for multiple testing and clinical cofounders (Table 1.5). Conversely, in a second study carried out in 100 adult patients treated for ovarian cancer, there was no significant association with GSTM3*B or any other GST genes in the context of cisplatin-induced ototoxicity[99]. The lack of replication of GSTM3*B with cisplatin ototoxicity may be due to differences between cohorts including age of patients and the ototoxicity grading criteria. Furthermore, Khrunin et al did not evaluate ototoxicity by audiometric tests prior to start of treatment, suggesting that individuals may have had undetected pre-existing hearing loss which could compromise results. Other studies in both children and adults have not investigated the GSTM3*B deletion. Therefore, it is difficult to conclude that GSTM3*B is important in identifying patients at risk of hearing loss due to these contradictory results. However, there is a possibility that this variant may be important in children and not in adults. Further studies need to be conducted to investigate whether this is a finding is a true association.  1.6.1.2 GSTP1 Oldenburg et al. identified an association of GSTP1 with cisplatin-induced hearing loss in a cohort of 173 adult testicular cancer patient (OR 4.21, P<0.001) (Table 6) [97]. This study used significantly different ototoxicity grading criteria than most studies of cisplatin-associated ototoxicity. They ranked patients based on the threshold of hearing at 4000Hz to separate patients into specific percentile groups rather than using a grading scheme. An additional followup study by Oldenburg et al did not find an association with GSTP1 in an extended cohort of  19  adult testicular cancer patients[98]. However, as all the patients from the previous study were included in this extended cohort, this cannot be considered an independent replication study. The effect of GSTP1 rs1695 was in the same direction in all studies. These studies have demonstrated a similar direction of association with the GSTP1 rs1695 variant and cisplatinototoxicity. However, this association only achieved statistical significance in one study of adult patients (p < 0.001)[97], and trended towards significance in two other studies (P-values= 0.055 and 0.16) [98] (Table 1.6). In these studies the association of the GSTP1 rs1695 A-allele has shown a similar trend in association with cisplatin-induced ototoxicity (odds ratios for protective GG allele in pediatric studies: no association and 0.78; and in adult studies: 0.23, 0.71, and no association). However, this association only achieved statistical significance (p < 0.05) in one study of adult patients[97]. It is possible that the relatively small effect size of the GSTP1 rs1695 variant has been undetectable in the underpowered studies to date. Additional studies should further investigate the potential role of GSTP1 variants in cisplatin-induced hearing loss. The small effect size will require large studies to confirm or refute.  1.6.1.3 GSTM1 and GSTT1 Oldenburg et al. also identified an association of GSTM1 with cisplatin-induced hearing loss and developed two multi-marker models consisting of GSTM1, GSTP1, and GSTT1 in testicular cancer patients (Table 1.7)[97]. In the first cohort, the GSTM1 positive odds ratio was 2.36 (p = 0.022, n = 173), and in the second study that combined the first cohort with a second cohort, the odds ratio was 1.81 (p = 0.025, n = 238) [98]. The second association observed with GSTM1 may therefore have been largely due to the effect from the original 173 testicular cancer patients. The 20  weaker association in the new patient cohort could also be due to the fact that hearing loss of some patients in the follow-up study was self-reported, as opposed to audiometric grading. Other studies that investigated GSTM1 and GSTT1 have found no significant association with hearing loss after cisplatin therapy including three independent studies in children[64, 96] treated for various malignancies, as well as an adult study in ovarian cancer patients[99] (Table 4.6). Furthermore, the effect of GSTM1 was observed in opposite directions. Therefore, further replication of the original findings in an independent patient cohort would significantly strengthen this association.  At this time, there is no consistent evidence of a strong association between genetic variants in, GSTM1, GSTT1 and cisplatin-induced hearing loss. However, it cannot be excluded that the effect of some of these variants may apply only to adult testicular cancer patients, in whom they were identified. The treatment protocols for this group are often unique, with longer and slower infusions of cisplatin. Replication of this finding in another independent testicular patient cohort is required prior to use in the clinic to predict risk.  1.6.2 Variations in Genes in DNA Repair Pathways and Cisplatin-Induced Hearing Loss The cross-links that form between cisplatin and DNA impede DNA replication. A major pathway involved in DNA repair is the nucleotide excision repair (NER) pathway which functions to remove platinum adducts. Polymorphisms in genes involved in NER (ERCC1, ERCC2, ERCC4, ERCC5, XPA and XPC) have been investigated in the context of susceptibility to cisplatin-induced ototoxicity[99, 100]. Other DNA repair genes that have been investigated include XPD and XRCC1[99].  21  Caronia et al reported 32 patients (adults and children) diagnosed with osteosarcoma (Table 1.8) [100]. Several genetic variants in the NER pathway were investigated and a weak association of a non-synonymous SNP in XPC rs2228001 (Lys939Gln) with cisplatin-mediated hearing loss was observed (P=0.042) [100]. The polymorphism is hypothesized to reduce the activity of XPC, and thus affect the capacity for DNA repair. This decrease in DNA repair capacity may increase apoptosis in the cochlea resulting in ototoxicity[100]. However, no multiple testing corrections were applied in this study and the number of patients in the study was limited (n=32). Furthermore, the study did not evaluate differences in clinical factors between cases and controls, and therefore did not take potential confounding factors into account.  Functional studies evaluating DNA repair capacity in XPC rs2228001 have also shown inconsistent results. For example, one study observed that individuals that carry the C-risk allele had significantly higher DNA damage compared to the A-allele[101], where as Khan et al found similar DNA repair capacity between the two alleles[102]. There are currently no replication studies to confirm this finding and due to inconsistent findings from functional studies, it is difficult to draw any conclusions about an effect of rs2228001 on cisplatin ototoxicity at this time and the use in clinical practice would require additional supporting evidence.  None of the variants in other genes in the DNA repair pathway investigated in the aforementioned studies were associated with cisplatin-induced hearing loss, including polymorphisms in ERCC1, ERCC2, ERCC4, ERCC5 and XPA. Another independent study by  22  Khrunin et al carried out in 100 adult ovarian cancer patients (Table 1.8) also did not observe any association of variants in ERCC1, XPD and XRCC1 with cisplatin ototoxicity[99].  1.6.3 Genes Associated with Hereditary Deafness and Cisplatin-Induced Hearing Loss Genes involved in hereditary deafness have also been studied in the context of cisplatin ototoxicity. Mutations in the mitochondrial genome are thought to be responsible for a number of clinical abnormalities including hearing loss. Previously, variants in GJB2, SLC26A4 and mitochondrial polymorphisms in MTRNR1 (A1555G), MTTL1 (A3243G) and MTTS1 (A7445G) have been shown to be associated with aminoglycoside- mediated ototoxicity and progressive nonsyndromic high frequency hearing loss (hereditary deafness). As the molecular basis of cisplatin-induced ototoxicity remains unclear, Knoll et al hypothesized that genes associated with hereditary deafness may also be associated with cisplatin-induced ototoxicity. Knoll et al carried out mutation screening in the GJB2 and SLC26A3 genes by sequencing and genotyping of the mitochondrial DNA (mtDNA) polymorphisms in 11 children that were selected based on selfreport of clinically apparent hearing loss (Table 1.9). No evidence for a contribution of these variants to the susceptibility to cisplatin-induced hearing loss was detected. Given the very small cohort of patients in this study, it may have been underpowered to find any association. However, a separate independent study in 39 patients also investigated the role of mtDNA polymorphisms and also did not observe any association with cisplatin-induced hearing loss. Therefore, there is currently no evidence to suggest an association of genes involved in hereditary deafness with cisplatin-induced hearing loss. However, a larger study would be required to adequately assess whether these variants are potential predictors of cisplatin-induced ototoxicity.  23  1.7 Predictive Modelling in Pharmacogenomics 1.7.1 Measures of Predictive Tests Both sensitivity and specificity are measures for assessing the results of diagnostic and screening tests (Table 1.10). Sensitivity represents the proportion of truly diagnosed people in a screened population who are identified as being affected and therefore, it is a measure of the probability of correctly diagnosing a disease or condition[103]. Specificity is the proportion of truly nonaffected individuals identified. Thus, specificity measures the probability of correctly identifying a person without the disease or condition[103]. Generally there are trade-offs between specificity and sensitivity. Ideally, a clinical test would have high specificity and high sensitivity. An ideal clinical test will be accurate, producing a positive result in the presence of the disease and a negative result in the absence of the disease. A test may produce false positive and false negative results (Table 1.10). Positive and negative predictive values represent such errors. Positive predictive value (PPV) is defined as the proportion of individuals with a positive test result who truly have the disease. Negative predictive value (NPV) is defined as the proportion of individuals with a negative diagnostic test being truly disease-free. These measures are used in pharmacogenomics to identify whether a genetic test can accurately predict adverse events in patients. Furthermore, they allow clinicians to evaluate the risks and benefits associated with the use of a genetic test in identifying patients at risk of adverse drug reactions.  24  1.7.2 Genomic Profiling Among the pharmacogenomic biomarkers currently in drug labels, FDA recommends testing HLA-B*1502 for carbamazepine, HLA-B*5701 for abacavir, G6PD for rasburicase and several variants in TPMT including TPMT*2, *3A,*3B, and *4 for azathioprine and 6mercaptopurine.(Table 1.1)[32]. With studies identifying multiple variants associated with a disease it suggests that more than one genetic variant may contribute to a complex disease or condition. Genetic profiling, which evaluates genetic risk by interpreting the state of multiple variants and/or multiple genes, can be a useful in clinical care[104]. Genetic profiles can provide useful risk classifications by identifying patients as high, intermediate and low risk[103]. The receiver operating characteristic (ROC) curve is a method that has been established in the context of genomic profiling[105, 106]. The curve is constructed using a combination of sensitivity and specificity for each threshold of a continuous test result[104]. The magnitude of the area under the ROC curve (AUC) is a measure of the proportion of variation that contributes to the risk of a disease or condition. It remains undetermined whether combining a high number (>20 susceptibility genes) of weak predictors that generate a high discriminative value will be useful as a predictive test[104]. Determining the level of discrimination for clinical use depends on a variety of factors such as the goal of testing and the prevalence of the condition among others[104]. Screening programs usually seek a discriminative value (AUC) of greater than 0.80[104, 107]. The benefit of using genetic profiling in pharmacogenomics is to identify patients at high versus low risk of ADRs to select appropriate treatment procedures such as increased follow-up and alternative treatments. The balance between risk and benefit of a test is measured by evaluating sensitivity, specificity, positive and negative predictive values. Genomic profiling may have the 25  potential to convince FDA to include pharmacogenomic testing for multiple variants that have an additive effect. Furthermore, by including clinical risk factors into this risk prediction model combining genetic factors, it is possible to potentially improve the discriminative value to identify patients at greatest risk of adverse events.  1.8 Potential Confounders 1.8.1 Role of Ancestry in Drug Response and Toxicity It is well recognized that ancestry differences influence drug response, efficacy and toxicity[108]. There are several examples of drugs, such as warfarin, which exhibit different responses in patients of different ancestries because of genetic differences[109]. The four major distinct ancestral populations are European, Indian, African and Asian. Given these differences, it is not surprising that genetic polymorphisms in genes involved in drug metabolism and drug transporters vary significantly in allele frequencies between different ethnicities. Genetics may contribute to the understanding of how ethnic differences affect drug response. This leads to questions as to whether ancestry-specific variants are associated with severe ADRs. Several studies have suggested that there are cases in which this occurs. For example, the variant HLA-B*1502 is associated with carbamazepine-induced SJS only in Asian populations[110]. It is evident that ethnicity can play a significant role in pharmacogenomic findings.  1.8.1.1 Ancestry as Confounder in Pharmacogenomic Studies Most studies are carried out using a homogenous cohort of patients that consist of individuals that are of similar ancestral origins. The reason to select a homogenous cohort for association 26  studies is to limit the amount of population stratification in the study. Population stratification is the differences in allele frequencies between cases and controls due to ancestry[111]. Population stratification is suggested to lead to false positive associations as the effect observed is due to allele frequency differences between ethnic groups rather than with disease or drug toxicity, confounding the results. In addition, by combining multiple different subpopulations into a study the effect of a true association is masked, leading to false negative findings[112]. It has been proposed that ethnic differences must be sufficiently controlled for in genetic association studies.  1.8.1.2 Controlling for Differences in Ancestry Many genetic association studies aim to maximize power by including as many patient samples as possible into the analysis. This may mean that patients that are included in association studies are not strictly from one ethnic group. The level of ethnic diversity and the number of interethnic unions in Canada showcases this important fact. Patient self-reported ethnicity is insufficient in determining genetic ancestry, especially in populations with inter-ethnic admixture[113]. Several methods have been described to identify and correct for population stratification. Principal component analysis (PCA) is being increasingly used to detect and correct for stratification in genetic association studies[112]. The method requires genotyping patient samples for a large number of genetic markers. This data is then translated into individual principal components that capture a portion of variability in the population. The first principal component (PC1) explains the greatest amount of variability in data and distinguishes between individuals of European and East-Asian ancestry. The second principal component (PC2)  27  distinguishes between individuals of Africans and non-African ancestry. By reliably determining genetic ancestry of patients, we can control and correct for these ethnic differences, thereby improving power of an association study and minimizing false positive results.  1.8.2 The Importance of Clinical Characterization Phenotype is defined as the observable characteristics or traits of an individual such as physiological or biological characteristics. Phenotyping patients are a key step in classifying individuals as cases or controls for genetic association studies. In particular, accurate and consistent phenotyping of patients is crucial to obtain reproducible findings in pharmacogenetic studies. In the context of cisplatin-induced hearing loss, patients are phenotyped based on clinical criteria (grading schemes) which are used by clinicians to identify the severity or degree of hearing loss. There are limitations in the use of each grading scheme in terms of being able to objectively distinguish between the different degrees of hearing loss among patients. In fact, there is no consensus on which grading scheme is optimal. Challenges exist in identifying a grading scheme that is used consistently which often leads to discrepancies in patient phenotyping. Consequently, many genetic studies have used different definitions of cases and controls, making it difficult to compare findings and also results in different reported incidences of hearing loss[97, 99, 100]. Audiologic grading schemes need to be standardized so that reproducible grading classifications can be used across multiple sites in order to produce comparable results. It is particularly important that the system used to classify hearing loss be sensitive, reliable, feasible and  28  applicable across all ages of pediatric patients. Accurate and consistent phenotyping of patients play an important role in genetic association studies.  1.9 Importance of Replication Studies A single association study is often inadequate to prove that a true association exists between causative genetic variants and a disease or condition. Replication of findings in multiple independent cohorts of patients is required for validation to minimize false positive results (type 1 errors) and to ensure findings are consistent and precise[114]. Replication is defined as the ability of another study to statistically confirm the same hypothesis[115]. It is now well recognized that replication of genetic associations are required before any causal inferences can be drawn[114]. An increasing number of studies attempting to replicate findings have raised awareness that consistent replication is challenging and difficult to achieve. Nevertheless, there is significant lack of replication of research findings, particularly in the field of genetic association studies. Lack of replication can be due to several factors including Type I errors, lack of statistical power, population stratification, population heterogeneity[114].  In studying/investigating the pharmacogenomics of ADRs, replication of findings in multiple independent cohorts increases the confidence that these genetic variants are truly associated with adverse events. This is important when implementing genetic findings into a predictive model to identify patients at risk of ADRs, especially if these findings will then be used in clinical diagnosis and treatment.  29  1.10 Hypothesis and Thesis Objectives The study hypothesizes that genetic variants contribute to increased susceptibility to cisplatininduced hearing loss in pediatric oncology patients treated for a variety of tumours. The aim of this study was to identify genetic factors that confer susceptibility to cisplatin-induced hearing loss to develop pharmacogenetic predictive tests to identify individuals at risk of developing hearing loss.  Specific Aims of the study were as follows: 1. To investigate the replication of the association of prior findings of genetic variants with cisplatin-induced hearing loss in a new independent cohort of well characterized pediatric patients 2. To genotype patients using an enhanced candidate gene panel to identify novel genetic variants that confer susceptibility to cisplatin-induced hearing loss  30  Figure 1.1: Phase I and II drug metabolizing enzymes Drug-metabolizing enzymes are classified into phase I and phase II reactions. Phase I reactions involves functional group modifications through oxidation, reduction and hydrolysis reactions. Phase II metabolism are involved in conjugation reactions. The relative contribution of each enzyme to metabolism is observed by the size of each section of the pie chart. ALDH, aldehyde dehydrogenase; ADH, alcohol dehydrogenase; CYP, cytochrome P45; DPD, dihydropyrimidine dehydrogenase, NQO1, NADPH: quinine oxidoreductase or DT diaphorase; COMT, catechol O-methyltransferase; GST, glutathione-S- transferase; HMT, histamine methyltransferase; NAT, N-acetyltransferase; STs, sulfotransferases; TPMT, thipurine methyltransferase; UGTs, uridine 5‟-triphosphate glucuronosyltransferases. From Evans, W.E. and M.V. Relling, Pharmacogenomics: translating functional genomics into rational therapeutics. Science,1999. 286(5439): p. 487-91[18]. Reprinted with permission from AAAS.  31  Figure 1.2: Candidate genes involved in the metabolism of platinum drugs Several genes are involved in the transportation of cisplatin. The influx of cisplatin is suggested to be mediated by SLC31A1, whereas efflux is mediated by several drug transporters including ATP7A, ATP7B, ABCC2 and ABCG2. This figure is reproduced with permission from PharmGKB and Stanford University[116]. Copyright PharmGKB.  32  Table 1.1: Pharmacogenomic biomarkers in FDA drug labels Drug (s)  Genetic Biomarker(s)  Trastuzumab  Her2/neu  Maraviroc  CCR5  Dasatinib  Philadelphia Chromosome  Vemurafenib  BRAF V600E  Carbamazepine  HLA-B*1502  Abacavir  HLA-B*5701  Use in Clinic  Required for Testing: Efficacy for breast cancer treatment Efficacy for CCR5-tropic HIV-1 infection Efficacy for acute lymphoblastic leukemia Efficacy for treating metastatic melanoma  Recommended for Testing:  Azathioprine, 6-mercaptopurine Rasburicase  TPMT G6PD  Risk of Stevens-Johnson Syndrome/Toxic Epidermal Necrolysis Risk of severe hypersensitivity reactions Risk of severe, life-threatening myelosuppression Risk of severe hemolysis  Adapted from U.S. Food and Drug Administration. Table of Pharmacogenomic Biomarkers in Drug Labels. http://www.fda.gov/Drugs/ScienceResearch/ResearchAreas/Pharmacogenetics/ucm083378.htm Accessed 04/03/2012.  33  Table 1.2: Criteria used to determine the grade of cisplatin-induced hearing loss in patients Grading Scheme  Grade 0 Normal Hearing  Grade 1 Mild Hearing Loss  Grade 2 Moderate Hearing Loss  Grade 3 Severe Hearing Loss  Grade 4 Severe Hearing Loss  CTCAE Criteria[61]  <20 dB at all frequencies  Hearing loss of 20-25 dB at 4 - 8 kHz  Hearing loss of > 25 dB at 4 - 8 kHz  Hearing loss of > 25 dB at 2 - 8 kHz  Hearing loss of > 40 dB at 1 - 8 kHz  Brock Criteria[49]  < 40 dB at all frequencies  Hearing loss of ≥ 40 dB at 8 kHz  Hearing loss of ≥ 40 dB at 4 kHz and above  Hearing loss of ≥ 40 dB at 2 kHz and above  Hearing loss of ≥ 40 dB at 1 kHz and above  Muenster Criteria[62]  <10 dB at all frequencies  Hearing loss of >10 to <20 dB at all frequencies  Hearing loss ≥4 kHz, >20 dB 2a: >20 to 40 dB 2b: >40 to 60 dB 2c: >60 dB  Hearing loss <4 kHz; >20 dB 3a: >20 to 40 dB 3b: >40 to 60 dB 3c: >60 dB  Mean hearing loss <4 kHz; ≥ 80 Db  Hearing loss of 1a: ≥ 40 dB at 6-12 kHz 1b: >20 to< 40 dB at 4kHz  Hearing loss of 2a: ≥ 40 dB at 4 kHz and above 2b: >20 to< 40 dB below 4kHz  Hearing loss of ≥ 40 dB at 2 or 3 kHz  Hearing loss of ≥ 40 dB at 1 kHz  Chang Criteria[55]  <20 dB at 1, 2 and 4kHz  34  Table 1.3: Select studies describing cisplatin-induced hearing loss clinical risk factors Risk Factors  High cumulative dose (P<0.0001) History of noise exposure (P=0.006) Cumulative vincristine dose (P=0.001)  Grading scheme  Age (years)  Tumour Type  Refs.  Grading according to guidelines at Department for Otorhinolaryngology at the Hannover University Medical School  31 (21-53)  Testicular Cancer Patients  Bokemeyer (1997) [73]  Brock criteria  unknown (0-20)  Hepatoblastoma, neuroblastoma, and osteosarcoma  Li et al (2004) [48]  Head and neck tumours  Bhandare et al (2007) [74]  Medulloblastoma, osteosarcoma, germ cell tumours  Chang et al (2010) [55]  Age (P=0.001) - Children <5 years are at higher risk Higher individual dose (P<0.0001) Cumulative dose (P<0.005) - Patients with >400mg/m2 are at higher risk Age (P=0.018) Dose (P<0.0001) Cranial irradiation (P=0.028) Younger age and smaller body surface area (P=0.0078)  Based on their own specific grading criteria  Chang criteria  11.3 (0.33-24)  35  Table 1.4: Studies associated with megalin (LRP2) and cisplatin-induced hearing loss Patients (n) Association LRP2: rs2075252 A-allele; OR 3.45, P=0.016 LRP2: rs46681234 T allele; OR2.32, P=0.087  LRP2: rs2075252 A-allele; OR 1.2, P=0.55  Cases  Controls  25  25  Grade 2-4  Grade 0-1  106  56  Grade 2-4  Grade 0  Grading scheme  Muenster  CTCAE v3  Age (years) Cases  12.6 (6-22)  6 (0-16)  Tumour Type  Refs.  13.6 (5-19)  Osteosarcoma, neuroblastoma, medulloblastoma, germ cell tumour, teratoma, testicular cancer  Riedemann et al (2008)  9 (0-19)  Osteosarcoma, neuroblastoma, brain tumour, germ cell tumour, hepatoblastoma, lymphoma, etc.  Ross et al (2009)  Controls  36  Table 1.5: Studies associated with GSTM3*B and cisplatin-induced hearing loss Association GSTM3*B: OR 5.37, P=0.02 Carried at a higher frequency in controls (protective)  No association with variants in: GSTM3*B  Patients (n) Cases Controls  Grading scheme  20 Grade 2-3  19 Grade: 0  Modified version of the Muenster Criteria  38 Grade 1-4  62 Grade: 0  National Cancer Institution Criteria No baseline audiometric tests  Age (years) Cases Controls 11.3 (3-22)  13.6 (7-19)  Overall age: 52 (23-65)  Tumour Type  Refs.  Osteosarcoma, neuroblastoma, medulloblastoma, germ cell tumour  Peters et al (2000)  Ovarian cancer  Khrunin et al (2010)  37  Table 1.6: Studies associated with GSTP1 and cisplatin-induced hearing loss Association  No association with GSTP1  Patients (n) Cases Controls 20 Grade 2-3  19 Grade: 0  Total number: 173 GSTP1 rs1695:AA & AG risk genotype: OR 4.21, P<0.001  Number of cases and controls unknown  GSTP1 rs1695: GG protective genotype; OR 0.81, P=0.055 Note: 173 patients were included in previous study (not an independent study)  Total: 238 (173 were included in previous study)  Grading scheme Modified version of the Muenster Criteria  Age (years) Cases Contro ls 11.3 (3-22)  13.6 (7-19)  Tumour Type  Refs.  Osteosarcoma, neuroblastoma, medulloblastoma, germ cell tumour  Peters et al (2000)  Ranked patients by 10th, 25th, 75th and 90th percentiles at 4000Hz to phenotype cases and controls  Adults Overall age: 42 (24-73)  Testicular cancer  Oldenburg et al (Jan 2007)  Self-reported hearing loss  Adults Overall age: 29 (15-64)  Testicular cancer  Oldenburg et al (Feb 2007)  Osteosarcoma, neuroblastoma, brain tumour, germ cell tumour, etc.  Ross et al (Dec 2009)  Ovarian cancer  Khrunin et al (2010)  GSTP1 rs1695: GG genotype: OR 0.71, P=0.61  106 Grade 2-4  56 Grade 0  CTCAE v3  No association with variants in GSTP1  38 Grade 1-4  62 Grade: 0  National Cancer Institution Criteria No baseline audiometric tests  6 (0-16)  9 (0-19)  Overall age: 52 (23-65)  38  Table 1.7: Studies associated with GSTM1 or GSTT1 and cisplatin-induced hearing loss Association  No association with GSTT1, GSTM1 GSTM1 positivitya (+): OR 2.36, P=0.022 GSTT1 positivity (+): OR 1.23, P=0.64 Model 1: (GSTT1 +, GSTM1 + and AA genotype GSTP1) OR 2.76, P=0.005 Model 2:(GSTT1 +, GSTM1 + and GG genotype GSTP1) OR 5.35, P<0.001 GSTM1 positivity: OR 1.81, P=0.025 GSTT1 positivitya: OR 1.20, P=0.60 Note: 173 patients were included in previous study (not an independent study)  Patients (n) Cases Controls 20 Grade 2-3  19 Grade: 0  Total number: 173 Number of cases and controls unknown  Total: 238 (173 were included in previous study)  Grading scheme Modified version of the Muenster Criteria  13.6 (7-19)  Refs.  Osteosarcoma, neuroblastoma, medulloblastoma, germ cell tumour  Peters et al (2000)  Adults Overall age: 42 (24-73)  Testicular cancer  Oldenburg et al (Jan 2007)  Self-reported hearing loss  Adults Overall age: 29 (15-64)  Testicular cancer  Oldenburg et al (Feb 2007)  CTCAE v3  Overall age: 6.8 (1.6-18)  Medulloblastoma  Barahmani et al (June 2009)  Osteosarcoma, neuroblastoma, brain tumour, germ cell tumour, etc.  Ross et al (Dec 2009)  Ovarian cancer  Khrunin et al (2010)  15 Grade unknown  GSTM1 positivity : OR 0.78, P=0.51  106 Grade 2-4  56 Grade 0  CTCAE v3  No association with variants in GSTM1 and GSTT1  38 Grade 1-4  62 Grade: 0  National Cancer Institution Criteria No baseline audiometric tests  a  11.3 (3-22)  Tumour Type  Ranked patients by 10th, 25th, 75th and 90th percentiles at 4000Hz to phenotype cases and controls  19 Grade 3-4  No association with GSTM1, GSTT1,or GSTM1GSTT1 combined model  Age (years) Cases Controls  6 (0-16)  9 (0-19)  Overall age: 52 (23-65)  a  Positive indicates if an individual carries the GSTM1 or GSTT1 gene or not. A patient carries at least on copy of the gene. Effect is calculated comparing positive vs. negative.  39  Table 1.8: Studies associated with DNA repair pathway and cisplatin-induced hearing loss Patients (n) Association  XPC rs2228001: CC genotype: OR 17.2, P=0.042 No association with variants in: ERCC1, ERCC2, ERCC4, ERCC5, XPA  No association with variants in: ERCC1, XPD, XRCC1  Cases  Controls  15  17  Grade unknown  Grade unknown  Grading scheme  Unknown  38 Grade 1-4  62 Grade: 0  National Cancer Institution Criteria No baseline audiometric tests  Age (years) Tumour Type  Refs.  Overall 15 (4-34)  Osteosarcoma  Caronia et al  Overall age: 52 (23-65)  Ovarian cancer  Khrunin et al (2010)  Cases  Controls  40  Table 1.9: Studies associated with hereditary deafness genes Patients (n) Association  Cases  Controls  20  19  Modified version of the Muenster Criteria  11  0  Self reported  No association with:  MTRNR, MTTS1, 7472insC, variations in the D-loop region No association with: GJB, SLC26A4, MTRNR1, MTTL1,  MTTS1  Grading scheme  Age (years) Cases  Controls  11.3 (3-22)  13.6 (7-19)  8.7 (1-16)  Tumour Type  Refs.  Osteosarcoma, neuroblastoma, medulloblastoma, germ cell tumour  Peters U et al (2003)  Osteosarcoma, soft tissue sarcoma, CNS tumour  Knoll et al (2006)  41  Table 1.10: Calculation of sensitivity, specificity, positive and negative predictive value Condition Present  Condition Absent  Positive outcome  True Positive (TP)  False Positive (FP)  Positive Predictive Value (PPV) = TP/(TP+FP)  Negative outcome  False Negativeb (FN)  True Negative (TN)  Negative Predictive Value (NPV) = TN/(TP+FP)  Sensitivity = TP/(TP +FN)  Specificity = TN/(FP+TN)  a  a  False positive rate represents the type I error False negative rate represents the type II error  b  42  1  Chapter 2: Replication of TPMT and ABCC3 genetic variants highly associated with cisplatin-induced hearing loss in children  2.1 Introduction Cisplatin is a widely used chemotherapeutic agent for the standard treatment of a variety of solid tumours. It is one of the most effective chemotherapeutic agents for children and has contributed to the dramatic increase in survival of many solid tumors including hepatoblastoma, brain tumors, neuroblastoma, osteosarcoma, and germ cell tumors. Cisplatin has shown efficacy in standard risk hepatoblastoma such that it can be used as monotherapy with >80% 3-year eventfree survival[39]. A major complication that limits the use of cisplatin is the risk of druginduced ototoxicity that can result in life-long disability[117]. Cisplatin is known to cause severe ototoxicity that manifests as permanent, bilateral sensorineural hearing loss, which affects 1025% of adults and 26-90% of children[48-53]. In children in particular, even mild losses in hearing can significantly influence speech and language development, social-emotional development and increase the risk of learning difficulties[58, 64]. In adults, the rate of hearing loss may be higher than that reported due to a lack of baseline and follow-up audiometry in cisplatin protocols.  In patients receiving the same dose of cisplatin, the inter-individual variability of cisplatininduced effects on hearing is profound, from no hearing loss, to high frequency hearing loss which progresses to severe hearing impairment in the speech frequencies[58, 59, 91].  1  A version of this chapter will be submitted for publication. Pussegoda K, Ross C.J, Visscher H, Yazdanpanah M, Brooks B, Rassekh S.R, Feroz Zada Y, Dubé M.P, Carleton B.C, Hayden M.R. Replication of TPMT and ABCC3 genetic variants highly associated with cisplatin-induced hearing loss in children. 43  Furthermore, patients show no improvement in hearing and often the progression and severity of hearing loss continues long after the end of therapy[54]. Higher cumulative cisplatin dose[48, 73, 74], younger age[48, 55, 73], cranial irradiation[74, 79], and concomitant use of aminoglycosides and vincristine [73, 76, 77] are known to influence cisplatin-induced ototoxicity[59]. In standard treatment protocols, the occurrence of moderate to severe cisplatin ototoxicity leads to dose reduction or the termination of cisplatin treatment, which may affect survival rates[64]. Interindividual variability in hearing loss suggests that clinical risk factors alone are insufficient predictors of safety.  There is evidence that mechanisms such as the generation of reactive oxygen species are involved in the occurrence ototoxicity associated with cisplatin treatment[67]. Currently there are no means to identify individuals who are at an increased risk of developing hearing loss. Genetic variation in the genes involved in drug biotransformation, transport, and receptors have been recognized to influence patient drug response and susceptibility to adverse drug events, including ototoxicity[29]. The identification of genetic markers that confer increased susceptibility to cisplatin-induced ototoxicity has important implications for improving patient care during cisplatin treatment.  Recently, a candidate gene study in children receiving cisplatin identified genetic variants in thiopurine S-methyltransferase (TPMT) (rs12201199, rs1800460, rs1142345), catechol Omethyltransferase (COMT) (rs9332377, rs4646316) and several other variants including in the ATP-Binding Cassette transporter C3 (ABCC3) (rs1051640) conferring increased risk of developing cisplatin-induced hearing loss[64].  44  For clinical application, it is essential to replicate these genetic findings in multiple independent cohorts of patients to reduce the number of false positive results and to ensure that the genetic risk prediction is consistent[114]. It is now well recognized that replication of genetic associations are required before any causal inferences can be drawn[114]. Increasing number of studies attempting to replicate findings raises awareness that consistent replication is challenging and difficult to achieve. The aim of this study was to investigate replication of genetic risk factors for cisplatin-induced hearing loss in an independent cohort of patients recruited from across Canada and to assess a predictive multi-SNP model. We also report the results of combined analyses summarizing both the replication and initial findings.  2.2 Material and Methods 2.2.1 Patients Study participants were recruited through the CPNDS, a national multicenter active surveillance consortium for studying adverse drug reactions[118]. Between June 2008 and March 2011, a new independent replication cohort of 155 pediatric oncology patients with cisplatin-induced hearing loss and drug-matched control patients were recruited at pediatric oncology units across Canada: BC Children‟s Hospital (Vancouver), Alberta Children‟s Hospital (Calgary), Stollery Children‟s Hospital (Edmonton), Winnipeg Health Sciences Centre (Winnipeg), Children‟s Hospital at the London Health Sciences Centre (London), McMaster Children‟s Hospital (Hamilton), Hospital for Sick Children (Toronto), Children‟s Hospital of Eastern Ontario (Ottawa), Hôpital Sainte-Justine (Montreal) and IWK Health Centre (Halifax). The previous  45  cohort (discovery and replication) included in the combined analyses has been described previously[64]. Cisplatin-induced hearing loss was diagnosed based on audiometric findings using criteria described by the CTCAEv3 (Common Terminology Criteria for Adverse Events)[61]. To better differentiate between cases and controls, subjects with serious cisplatin-induced hearing loss were defined as those with grade 2 or higher hearing impairment. Control patients were defined as children who received cisplatin with normal audiometric data who did not develop hearing loss (grade 0). Patients with grade 1 (n=8) hearing loss were excluded from the analysis. Serious hearing impairment was defined at the point at which cisplatin chemotherapy protocols recommend reducing or terminating cisplatin treatment which occurs at grades 2-4 hearing loss. Baseline and the most recent audiological assessment were used. Informed written consent or assent was obtained from each subject or their parents or legal guardians. The study was approved by the University of British Columbia / Children & Women‟s Health Centre of British Columbia Research Ethics Board (H04-70358) and all participants provided informed consent.  2.2.2 Genotyping DNA of patients was extracted from blood, saliva or buccal swabs using the QIAmp DNA purification system (Qiagen, Toronto, ON, Canada) according to the manufacturer‟s protocol. We selected 11 single nucleotide polymorphisms (SNPs) to assess whether prediction of risk of hearing loss can be improved. We selected genetic variants for genotyping based on evidence of a significant association (P<0.01) in the initial combined cohort[64] as well as P<0.01 in either the initial discovery or replication cohort[64]. We first assessed whether variants in TPMT (rs12201199, rs1800460, rs1142345) and COMT (rs9332377, rs4646316) could be replicated in 46  the current replication cohort. We then assessed whether other variants in ABCC3 (rs1051640), MTHFR (rs373964), VKORC1 (rs17884333, rs8050894), and SLCO1A2 (rs4115170, rs2306231) were replicated. Patient DNA was genotyped using a custom Illumina GoldenGate SNP genotyping assay (Illumina, San Diego, CA, USA). This assay included additional non-study SNPs for principal component analysis (PCA) to determine patient ancestry and for quality control purposes. All SNP genotype data were clustered manually using GenomeStudio software (Illumina, San Diego, CA, USA). Samples with a call rate below 95% were excluded. The average genotyping call rate for included samples was 98.3%. The overall genotype call rate of the 11 study SNPs was 99.9% and were all in Hardy-Weinberg equilibrium (P>0.05).  2.2.3 Statistical Analysis Clinical characteristics of patients with and without cisplatin-induced hearing loss were compared using the Wilcoxon-Mann-Whitney U-test for continuous variables, and Fisher‟s exact test for categorical variables. Hardy-Weinberg equilibrium tests were conducted in controls using the permutation version of the exact test of Hardy-Weinberg of Guo and Thompson[119]. The study was calculated to have a statistical power between 65-99% (mean 86%) based on the previous effect sizes and genotype frequencies in order to replicate previously associated polymorphisms with cisplatin-induced hearing loss at P=0.05. Two genetic variants in SLCO1A2 (rs11045913 and rs11045912) were excluded prior to genotyping due to inadequate power (<10%) to detect an association. Homozygous and heterozygous odds ratios (ORs) were calculated using the homozygous genotype of the protective allele as identified in the initial cohort as a reference[64]. Calculation of ORs in the presence of empty cells was done by adding 0.5 to the empty cells. Association between genetic polymorphisms and cisplatin-induced 47  hearing loss was assessed by computing Fisher‟s exact test for unadjusted (non-regression) pvalues and adjusted p-values determined by logistic regression. Independent factors for the prediction of cisplatin-induced hearing loss in the combined cohort were identified by forward logistic regression analysis. In the forward logistic regression model the following factors were included: the first principal component, age, cisplatin dose, treatment duration, vincristine treatment and germ cell tumour. None of the variables were retained in combined cohort at a threshold of 0.10 except concomitant vincristine treatment (P=1.4×10-9). However, we also included variables that are known to increase risk of hearing loss that were significantly different (P=0.05) between cases and controls. Therefore, age, concomitant vincristine treatment, germ cell tumour, and cranial irradiation were included as covariates. The treatment of all non-CNS germ cell tumours is unique as these patients are treated at low individual doses of 20mg/m2 over five days and are known to be at less risk of hearing loss[41]. In an additional analysis, we combined the initial cohort[64] and the current cisplatin replication cohort of patients to determine the overall significance of associations. Variants were considered to be statistically significant, if they had P<0.05 in the new replication cohort. Potential population stratification in the dataset was assessed by PCA using SVS/HelixTree software[112, 120]. Patients of African ancestry (n=8), as determined by PCA, were excluded in the analysis to match the patient ancestry of the previous cohort[64, 120]. Graphical display of principal components was constructed with GraphPad Prism 5 software (GraphPad Software Inc., California, USA). A secondary stratified analysis of individuals of only European ancestry was based on the first two principal components. Genetic variants were further evaluated in a multivariate logistic regression model including clinical variables using an additive model. A genetic risk score was calculated by multiplying 48  each variable with the estimated beta (log odds ratio) from the current combined cohort. The genetic only model includes variants that were selected based on statistical evidence from a forward logistic regression model in the combined cohort. Variants that were retained in this model include TPMT rs12201199, ABCC3 rs1051650 and COMT rs4646316. Association between the genetic risk score and cisplatin-induced hearing loss was assessed using logistic regression. Three prediction models were investigated: (i) clinical variables only; (ii) genetic variables only; and (iii) clinical and genetic variables. The contribution of the genetic and/or clinical risk scores to the prediction of cisplatin-induced hearing loss was investigated by comparing the area under the ROC curves of the prediction models. AUC estimates were obtained using the ROC plot function on the basis of the linear predictors obtained from the logistic regression model for clinical factors and genotypic scores. The statistical difference between the curves was calculated using DeLong‟s method[120] implemented in the R-package pROC[121]. Two-sided p-values <0.05 were considered statistically significant. Risk groups for cisplatin-induced hearing loss were defined based on the predictive values for each patient sample from the multi-SNP logistic regression model that included three SNPs (TPMT rs12201199, ABCC3 rs1051640, COMT rs4646316). The threshold for genetic variables used to determine lower risk was the median predicted value of controls minus one standard deviation (predictive value <0.4) and high risk was defined as the median predicted value of cases plus one standard deviation (>0.8). Intermediate risk was defined as a predictive value between 0.4-0.8. To assess the threshold for the combination of clinical and genetic variables for lower risk, median predicted value of controls (predictive value <0.45) and high risk was defined as the median predicted value of cases (>0.8). Intermediate risk was defined as a predictive value between 0.45-0.8. To assess the predictive performance of risk groups for cisplatin49  induced hearing loss, we calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for each defined threshold. Kaplan-Meier curves for cisplatin-induced hearing loss were generated for each risk group using time from start of treatment to the time of first evidence of toxicity or last audiogram. Log-rank test was used to compare the trend of survival curves. Hazard ratios for each curve were calculated using the Cox regression model considering the lower risk group as the reference. Statistical genetic analyses were conducted using SNP and Variation Suite 7.4.5 (Golden Helix, Bozeman, USA), SPSS Statistics 19 (SPSS Inc., Chicago, IL, USA) and R 2.13.0 (R Development Core Team).  2.2 Results 2.2.1 Patient Characteristics Patient characteristics of the initial cohort[64], current replication cohort and the combined cohort are listed in Table 2.1. Canadian Pharmacogenomics Network for Drug Safety (CPNDS) adverse drug reaction (ADR) surveillance clinicians routinely update patient information in the CPNDS patient database as new information became available (i.e., relapase, death and additional data on concomitant medication use). In the current replication cohort 87 (56%) of the 155 pediatric oncology patients developed hearing loss. Concomitant vincristine treatment was significantly higher in cases than controls in the replication cohort (66.7% compared to 27.9%, P=2.1x10-6), similar to the initial cohort (50.9% vs. 17.9%, P = 4.1 x 10-5; combined cohort P=1.1x10-9). Fewer osteosarcoma patients in the replication cohort developed hearing loss (11.5% in cases vs. 29.4% in controls, P=0.0073), whereas this was not significantly different in the initial cohort (22.6% in cases vs. 28.6% in controls, P=0.45). Furthermore, follow-up after 50  therapy was longer in cases than controls in the replication cohort (5.0 years vs. 2.0 years, P=2.1x10-4) but was not significantly different in the initial cohort (3.0 years vs. 2.0 years, P=0.10). 2.2.2 Genetic Results In the replication cohort, all genetic variants assessed in TPMT (rs12201199, rs1142345, rs180046) showed a significant association with cisplatin-induced hearing loss (Table 2.2). TPMT rs12201199 was the most strongly associated variant (P=0.0013, OR 6.1). The risk allele (A) was observed in 21 (12%) cases and 3 (2%) controls. The combined analysis of all 317 patients showed a stronger association with rs12201199 (P=8.7x10-7, OR 8.9) than in the initial cohort alone[64]. The other TPMT variants, rs1142345 and rs1800460 were also stronger in the combined cohort than in the initial cohort (Table 2.2). All three TPMT variants remained significantly associated in the combined cohort after correcting for clinical factors including age, vincristine treatment, germ cell tumour and cranial irradiation.  The protective „A‟ allele of COMT rs4646316 was carried by 33 (19%) cases and 32 (24%) controls in the replication cohort. Even though this variant showed an effect in the same direction as in the initial cohort, it was smaller (OR 1.3) and was not significantly associated with hearing loss in the replication cohort (P=0.33). A similar effect was seen for COMT rs9332377 (P=0.28, OR 1.4).  Next, we assessed whether other genetic variants that previously were associated with cisplatininduced hearing loss (P<0.01) could be replicated in the new cohort of patients. Additional variants could improve a model that predicts the risk of cisplatin-induced hearing loss. In total, 3  51  of the 6 variants showed an effect in the same direction as the initial cohort (Table 2.2, Table 2.3). One synonymous variant in ABCC3 (rs1051640) (E1503E) was significantly associated with cisplatin-induced ototoxicity in the current replication cohort (P=0.036, OR 1.8), and was more strongly associated in the combined cohort (P=7.8x10-4, OR 2.0; Table 2.2). The protective „A‟allele of ABCC3 rs1051640 was carried by 28 (16%) cases and 35 (26%) controls in the replication cohort. The association of ABCC3 rs1051640 genotype with cisplatin-induced hearing loss remained significant in the combined cohort after adjusting for clinical factors (P=0.0033).  Using principal component analysis, it was observed that the majority (80%) of patients had European genetic ancestry[112] (Figure 2.1). To reduce potential bias caused by population stratification, we assessed whether associations in the replication cohort remained significant in a more homogenous subset of patients of only European ancestry (n=124). In the European ancestry subset, similar results were observed, though with larger effect sizes for the TPMT and ABCC3 variants (Table 2.4).  A model that combined the effects of the associated risk genotypes and clinical risk factors was generated and Receiver Operating Characteristic (ROC) analyses were performed. The geneticonly model was constructed to include the three independent SNPs (TPMT rs12201199, ABCC3 rs1051640, COMT rs4646316) that were selected based on forward logistic regression. Although there was a lack of significant association with variants in COMT in the current replication cohort; the combined cohort logistic regression retained COMT rs4646316 because it significantly added to the prediction model. The clinical-only model included age, vincristine  52  treatment, germ cell tumour and cranial irradiation. Combining both the clinical and genetic variables significantly improved the ability to predict cisplatin ototoxicity (Figure 2.2) resulting in an area under the curve (AUC) of 0.786 compared to the clinical-only model (AUC 0.708, P=0.00048).  A predictive multi-marker model using these genetic variants could stratify patients by risk of ototoxicity. Risk groups were defined based on predictive values from the model as lower (predictive value <0.4), intermediate (0.4-0.8) and high risk (>0.8). Based on these groups, there were 30 (9.5%) individuals classified as lower risk, 245 (77.3%) as intermediate and 42 (13.2%) as high risk (Table 2.5). In the high risk group 39 (92.9%) individuals developed hearing loss (positive predictive value, 92.9%) compared to 3 (7.1%) controls conferring a specificity of 97.6% (Table 2.6). In the lower risk group 8 (26.7%) individuals developed hearing loss (positive predictive value, 73.3%) compared to 22 (73.3%) controls conferring a sensitivity of 95.9% (Table 2.6).  Individuals in the high risk group also had a significantly higher risk of cisplatin-induced ototoxicity when compared to lower to intermediate risk group (P=1.3x10-4, OR 11.0; Table 2.6). Severity of hearing loss also increased with higher risk group classifications (Table 2.5). A Kaplan-Meier plot illustrates that the intermediate and high risk groups have significantly increased risk of hearing loss over time (Ptrend=5.3x10-9; Figure 2.3). After 1 year of treatment, 71% of patients developed significant hearing loss in the high risk group compared to 27% of patients in the lower risk group (Figure 2.3). The risk of hearing loss in the high risk group was 5.9 times higher compared to the lower risk group (Figure 2.3). 53  A model combining the effect of TPMT, ABCC3 and COMT was able to discriminate between patients that are at lower versus high risk compared to a model with TPMT alone (Table 2.6). By including both ABCC3 and COMT into the model, patients at lower risk could be identified with more certainty. However, the ability to identify patients at high risk similar in the TPMT only model and the combined model that includes TPMT, ABCC3, and COMT (PPV 91.5% vs PPV 92.7%) (Table 2.6). A Kaplan-Meier plot combining both the clinical and genetic variables illustrate that we can identify 97 (50.3%) patients at high risk compared to 39 (20.2%) patients using genetics alone (Figure 2.3, Table 2.6). In the high risk group 97 (91.5%) of individuals developed hearing loss (positive predictive value, 91.5%) compared to 9 (8.5%) controls conferring a specificity of 92.7% and a sensitivity of 50.3% (Table 2.6).  2.4 Discussion The development of hearing loss after treatment with cisplatin is known to cause serious lifelong disability, particularly in children[117]. A proven approach to identify those patients at high risk of developing serious cisplatin-induced hearing loss would significantly improve the safety and efficacy of pediatric cancer therapy.  Recently, we identified genetic variants that were associated with cisplatin-induced hearing loss in children after cancer therapy[64]. Replication of genetic associations for validation in independent populations is critically important before clinical implementation and development of guidelines. In this replication study, we recruited a large independent cohort of pediatric oncology patients from across Canada who received cisplatin therapy. We confirmed the  54  associations of genetic variants in TPMT and ABCC3 with moderate to severe cisplatin-induced hearing loss. These results strengthen the evidence for the association between TPMT and ABCC3 variants with cisplatin-induced hearing loss. Variants in COMT exhibited smaller effect sizes in the replication cohort (OR: 2.5 vs. 1.3 and 5.5 vs. 1.4; previous vs. new cohorts). We generated a novel predictive model combining TPMT, ABCC3 and COMT genetic risk factors with clinical risk factors. The combined model of genetic plus clinical risk factors was more predictive than clinical risk factors alone (AUC 0.786 vs. 0.708; P= 0.00048). A predictive model that incorporates the overall most highly associated COMT variant was included as it added significantly to a predictive model with TPMT and ABCC3. The COMT variants do not contribute to the prediction of patients at high risk of cisplatin-induced ototoxicity, instead they stratify patients between intermediate and lower risk. The model was based on results from the combined cohort and requires further studies in children to validate these findings and assess the clinical utility.  Our study is the first to replicate the reported association between TPMT (rs12201199, rs1142345, rs1800460) and cisplatin-induced hearing loss in a large independent cohort of pediatric patients (n=155) treated for a variety of malignancies. Evidence indicates that patients that are heterozygous or homozygous for these TPMT variants have an increased risk of cisplatin-induced hearing loss, likely through reduced activity of the gene[64]. TPMT is likely to increase cisplatin toxicity by inactivating the binding of the compound to purines in DNA, thereby regulating cisplatin cross-linking. Loss of function mutations in TPMT can increase the efficiency of cisplatin cross-linking leading to increased toxicity[64]. Another potential  55  mechanism is that cisplatin toxicity is due to the accumulation of S-adenosylmethionine due to reduced activity of TPMT[64, 122-124].  The association between the synonymous variant in ABCC3 rs1051640 (E1503E) and hearing loss was also significantly replicated. The minor allele frequency (MAF) of rs1051640 ranges from 20% in the European population to <1% in the Nigerian Yoruban population[125]. Patients carrying the G-allele of rs1051640 were at increased risk of developing hearing loss after cisplatin treatment. This is the first study to describe the role of genetic variation in ABCC3 in the context of cisplatin-ototoxicity. ABCC3 is a transporter that mediates the efflux of organic anions, xenobiotics and glutathione S-conjugates[126, 127]. One of the mechanisms by which cancer chemotherapies, including platinum drugs are detoxified is through conjugation of the active metabolite to glutathione making the compound more anionic[93, 94]. This enables the compounds to be more readily exported from cells through an ATP-dependent pump. Studies in rat hepatocyte cell lines have shown that both ABCC2 and ABCC3 protein levels and mRNA expression increased after treatment with cisplatin[128, 129]. Other studies carried out in lung cancer cells lines have also shown that ABCC3 mRNA expression levels are significantly correlated to resistance to cisplatin and other platinum drugs[130, 131]. Reduced activity of ABCC3 can affect the detoxification pathway resulting in ineffective transport of toxic compounds out of the cell leading to toxicity. This suggests that ABCC3 levels might affect cisplatin transport. In turn, polymorphisms may regulate ABCC3 levels or affect function of the transporter. Further functional studies are required to assess the exact mechanisms by which variants in ABCC3 affect cisplatin-induced hearing loss.  56  Based on our a priori power calculations considering the effect sizes and genotype frequencies of the original studies, we had sufficient power for all polymorphisms to find similar effect sizes in the replication cohort. However, several SNPs, that were previously reported to be associated, could not be replicated, limiting the use of these variants in the prediction of risk of developing hearing loss. This is not surprising as many original studies report stronger genetic associations with larger effect sizes than in follow-up studies[132, 133]. There are several other potential reasons for the failure to replicate previous findings. Significant inter-study heterogeneity can arise due to differences in the cohorts such as population diversity, gender and treatment protocols [132, 133]. This could lead to differences in the effects of variants in specific subpopulations or lead to masking of the effects by factors that have not been sufficiently controlled for. We aimed to control for these factors by first evaluating power to ensure that the sample size is large enough to detect an association, then adjusting for clinical variables in the cohort and finally carrying out a subgroup analysis in patients of only European ancestry. Nevertheless, a combination of these factors may explain why we could not significantly replicate the association between COMT as well as other polymorphisms and cisplatin-induced hearing loss.  Clinical factors alone cannot accurately predict the risk of developing hearing loss. Combining the effect of TPMT, ABCC3 and COMT with clinical factors significantly improved the ability to predict risk compared to clinical factors alone (AUC 0.786 vs. 0.708, P=0.00048). Including the clinical risk factors with these genetic markers we can identify a greater number of patients at high risk compared to using genetic risk factors alone (50.3% for combined vs. 20.2% for genetic alone). This is of particular importance in children, for whom even mild losses in hearing can  57  cause difficulties in school performance[65, 134]. Several strategies can be proposed to prevent hearing loss in individuals at high risk of adverse effects from cisplatin. Patients might be placed on alternative medications (e.g. carboplatin) which are less ototoxic[55], receive increased monitoring for hearing loss to minimize the onset of severe stages of hearing loss, or receive otoprotective agents although concerns about the possibility of compromise of antitumour activity have been raised[135, 136]. The placement of hearing aids or cochlear implantation are also options used to manage cisplatin-induced hearing loss.  In conclusion, this study confirms previous findings and provides further evidence in an independent patient cohort for the associations between TPMT and ABCC3 in cisplatin-induced hearing loss and with COMT in the combined cohort. A combination of TPMT, ABCC3 and COMT with clinical variables provides a novel tool to improve the risk prediction of patients with hearing loss from cisplatin therapy. The discovery of additional variants through genomewide studies may also improve and refine the current predictive model. The combination of clinical and genetic risk factors can potentially improve risk classifications for hearing loss and may allow for individualized treatment.  58  Table 2.1:Patient demographics Initial Cohorta (n=162) Cases Controls (n= 106) (n= 56) p-value Age, years (median (min, max)) Dose, cumulative mg/m2 (median (min, max)) Treatment duration, weeks (median (min, max)) Gender (Male n, (%)) Caucasian ethnicity (n (%)) Concomitant medication (n, (%)) Tobramycin Vancomycin Vincristine Gentamicin Tumor type (n, (%)) brain tumor endodermal sinus tumor of thymus germ cell tumor hepatoblastoma lymphoma nasopharyngeal carcinoma neuroblastoma osteosarcoma sarcoma carcinoma retinoblastoma mesencymal tumour of liver Follow-up, years (median (min, max)) Cranial irradiation (n, (%))  Replication Cohort (n=155) Cases Controls (n= 87) (n= 68) p-value  Combined Cohort (n=317) Cases Controls (n= 193) (n=124) p-value  6.0 (0, 16)  9.0 (0, 19)  0.086  6.0 (1, 25)  11.0 (0, 18)  0.050  6.0 (0, 25)  10.0 (0, 19)  0.012  400 (120, 720)  400 (100, 720)  0.61  400 (92, 800)  400 (20, 768)  0.36  400 (92, 800)  400 (20, 768)  0.71  2.4 (1,11)  2.3 (1, 6)  0.96  2.1 (0, 8)  2.4 (0, 9)  0.65  2.2 (0, 11)  2.3 (0, 9)  0.90  71 (67.0%) 80 (75.5%)  28 (50.0%) 48 (85.7%)  0.043 0.16  43 (49.4%) 70 (80.5%)  34 (50.0%) 54 (79.4%)  1.00 1.00  114 (59.1%) 150 (77.7%)  62 (50.0%) 102 (82.3%)  0.13 0.39  32 (30.2%) 25 (23.6%) 54 (50.9%) 21 (19.8%)  15 (26.8%) 11 (19.6%) 10 (17.9%) 7 (12.5%)  0.72 0.69 4.1×10-5 0.28  23 (26.4%) 26 (29.9%) 58 (66.7%) 21 (24.1%)  15 (22.1%) 14 (20.6%) 19 (27.9%) 19 (27.9%)  0.58 0.20 2.1×10-6 0.71  55 (28.5%) 51 (26.4%) 112 (58.0%) 42 (21.8%)  30 (24.2%) 25 (20.2%) 29 (23.4%) 26 (21.0%)  0.44 0.23 1.1×10-9 0.89  25 (23.6%)  8 (14.3%)  0.22  26 (29.9%)  11 (16.2%)  0.058  51 (26.4%)  19 (15.3%)  0.026  0  1 (1.8%)  0.35  0  1 (1.5%)  0.44  0  2 (1.6%)  0.15  7 (6.6%) 22 (20.8%) 0  15 (26.8%) 5 (8.9%) 1 (1.8%)  0.00063 0.075 0.35  7 (8.0%) 17 (19.5%) 1 (1.1%)  11 (16.2%) 7 (10.3%) 2 (2.9%)  0.14 0.12 0.58  14 (7.3%) 39 (20.2%) 1 (0.5%)  26 (21.0%) 12 (9.7%) 3 (2.4%)  0.00046 0.013 0.30  1 (0.9%)  0  1.00  0  2 (2.9%)  0.19  1 (0.5%)  2 (1.6%)  0.56  26 (24.5%) 24 (22.6%) 1 (0.9%) 0 0 0  9 (16.1%) 16 (28.6%) 1 (1.8%) 0 0 0  0.23 0.45 1.00  24 (27.6%) 10 (11.5%) 0 0 1 (1.1%) 1 (1.1%)  12 (17.6%) 20 (29.4%) 1 (1.5%) 1 (1.5%) 0 0  0.18 0.0073 0.44 0.44 1.00 1.00  50 (25.9%) 34 (17.6%) 1 (0.5%) 0 1 (0.5%) 1 (0.5%)  21 (16.9%) 36 (29.0%) 2 (1.6%) 1 (0.8%) 0 0  0.073 0.019 0.56 0.39 1.00 1.00  3 (0, 18)  2 (0, 15)  0.10  5 (0, 25)  2 (0, 16)  0.00021  4 (0, 25)  2 (0, 16)  9.3×10-5  23 (21.7%)  7 (12.5%)  0.20  20 (23.0%)  8 (11.8%)  0.093  43 (22.3%)  15 (12.1%)  0.025  For age, dose, treatment duration and follow-up, the Wilcoxon-Mann-Whitney test with normal approximation was used. For gender, ethnicity, concomitant medication, tumor type, cranial irradiation the Fisher exact test was used. In bold are statistically significant values at p<0.05 type-I error rate. a Results from the initial combined cohort [64] compared to the current replication cohort as well as all cohorts combined; bCaucasian ethnicity assessed by principal component analysis  59  Table 2.2 Genetic variants associated with cisplatin-induced hearing loss Initial Cohorta (n=162)  SNP  Allele  106 cases, 56 controls OR Case Ctrl (95%CI)  pvalue  Replication Cohort (n=155)  Combined Cohort(n=317)  87 cases, 68 controls OR Case Ctrl (95%CI)  193 cases, 124 controls OR Case Ctrl (95%CI)  p-value  pvalue  pvalueb  8.9 (3.2-24.9)  8.7×10-7  4.0×10-5  6.1 (2.1-17.3)  0.00014  0.00039  6.6 (2.0-21.8)  0.00043  0.00073  1.8 (1.2-2.6)  0.0021  0.0068  1.9 (1.2-3.1)  0.0054  0.043  2.0 (1.3-2.9)  0.00078  0.0033  TPMT rs12201199  rs1142345  rs1800460  A  28  1  T  184  111  G  19  1  A  193  111  A  16  0  G  196  110  G  176  74  A  36  38  A  36  4  G  176  108  G  182  83  A  30  29  16.9 (2.3-125.9)  10.9 (1.4-82.7)  18.0 (1.1-302.7)  0.00022  0.0017  0.0031  21  3  153  133  16  3  158  133  13  3  161  133  141  104  33  32  38  23  136  113  146  101  28  35  6.1 (1.8-20.9)  4.5 (1.3-15.7)  3.6 (1.0-12.8)  0.0013  0.011  0.038  49  4  337  244  35  4  351  244  29  3  357  243  317  178  69  70  74  27  312  221  328  184  58  64  COMT rs4646316  rs9332377  2.5 (1.5-4.3)  5.5 (1.9-16.0)  0.00055  0.00018  1.3 (0.8-2.3)  1.4 (0.8-2.4)  0.33  0.28  ABCC3 rs1051640  2.1 (1.2-3.8)  0.0092  1.8 (1.0-3.3)  0.036  a  Results from the initial combined cohort [64] compared to the current replication cohort as well as all cohorts combined Adjusted for age, vincristine treatment, germ cell tumor and cranial irradiation In bold indicates results more significant in current combined cohort than in previous combined cohort[64]. SNP, Single Nucleotide Polymorphism; Ctrl, Control; OR, Odds Ratio; CI, Confidence Interval. b  60  Table 2.3 Association between additional top ranked SNPs and cisplatin-induced hearing loss Previous Cohorta (n=162)  Replication (n=155)  106 ototox., 56 controls  87 ototox., 68 controls  Combined (n=317) 193 ototox., 124 controls  Gene  SNP  Alleleb  OR (95% CI)  p-valueb  OR (95% CI)  p-valueb  OR (95% CI)  pvalueb  MTHFR_CLCN6  rs3737964  A/G  2.64 (1.40, 4.98)  0.0022  0.82 (0.46, 1.46)  0.50  1.00 (0.73, 1.37)  1.00  VKORC1  rs17884333  A/G  2.07 (1.28, 3.35)  0.0027  1.30 (0.83, 2.05)  0.25  1.61 (1.16, 2.23)  0.0041  VKORC1  rs8050894  C/G  2.07 (1.28, 3.35)  0.0027  1.10 (0.70, 1.73)  0.68  1.47 (1.06, 2.03)  0.020  SLCO1A2  rs4115170  G/A  2.09 (1.28, 3.41)  0.0028  0.93 (0.58, 1.48)  0.75  1.40 (1.00, 1.96)  0.048  SLCO1A2  rs2306231  G/A  2.08 (1.27, 3.36)  0.0034  0.93 (0.58, 1.48)  0.75  1.39 (0.99, 1.95)  0.053  a  Results from discovery and replication cohorts from previous study[64] compared to the current replication cohort as well as all cohorts combined SNP alleles assayed; minor allele is mentioned first SNP, Single Nucleotide Polymorphism; Ototox, Ototoxicity; OR, Odds Ratio; CI, Confidence Interval. b  .  61  Table 2.4 Genetic variants associated with cisplatin-induced hearing loss in patients of European ancestry  Gene  Allele  b  Previous Cohorta (n=127)  Replication (n=124)  Combined (n=252)  80 ototox., 48 controls  70 ototox., 54 controls  150 ototox., 102 controls  OR (95% CI)  p-value  OR (95% CI)  p-value  OR (95% CI)  p-value  p-valuec  3.95E-04  6.8 (1.5, 30.4)  0.0039  10.0 (3.1, 32.9)  4.29E-06  1.18x10-4  0.00045  0.0010  TPMT rs12201199 rs1142345 rs1800460 COMT rs4646316  rs9332377 ABCC3 rs1051640  A  16.0 (2.1, 120.1)  T  1  G  9.1 (1.2, 70.4)  1 0.011  5.4 (1.2, 24.6)  1 0.015  6.6 (2.0, 22.2)  A  1  A G  17.7 (1.0, 300.6) 1  0.0065  5.0 (1.1, 22.79) 1  1 0.023  8.7 (2.0, 37.2) 1  1 5.10E-04  0.0016  G  2.5 (1.4, 4.4)  0.0022  1.3 (0.7, 2.5)  0.38  1.8 (1.2, 2.8)  0.0057  0.010  A  1  A  7.5 (2.2, 25.1)  0.0027  0.017  G  1  A G  2.3 (1.2, 4.4) 1  8.31E-04  0.0059  1 0.00021  1.4 (0.7, 2.6)  1 0.35  1 0.0081  1.9 (1.0, 3.4) 1  2.2 (1.3, 3.8) 1  0.043  2.1 (1.4, 3.2) 1  a  Results from discovery and replication cohorts from previous study[64] compared to the current replication cohort as well as all cohorts combined SNP alleles assayed; minor allele is mentioned first c Adjusted for age, vincristine treatment, germ cell tumours and cranial irradiation In bold indicates results more significant in current combined cohort than in previous cohort[64]. SNP, Single Nucleotide Polymorphism; OR, Odds Ratio; CI, Confidence Interval. b  62  Table 2.5: Risk group comparisons and grade of hearing loss  Lower Risk (<0.4) Intermediate Risk (0.4-0.8) High Risk (>0.8) a  Patients with Hearing Loss Number (%a)  Normal Hearing Controls Number (%a)  Number (%b)  Grade of Hearing Loss Mean ± SEM  8 (26.7%)  22 (73.3%)  30 (9.5%)  0.77 ± 0.24  146 (59.6%)  99 (40.4%)  245 (77.3%)  1.69 ± 0.092  39 (92.9%)  3 (7.1%)  42 (13.2%)  2.71 ± 0.13  Total  Percentage in risk group Percentage of total  b  63  Table 2.6: Comparison of risk groups Model  Risk Group  Ototox. (Number, %)  Controls (Number, %)  TPMT only  Risk haplotype carriersc Non carriers  43 (91.5%) 150 (55.6%)  4 (8.5%) 120 (44.4%)  9.3 (3.1, 27.4)  TPMT, ABCC3, COMT  High (>0.8) Low plus Intermediate (<0.8)  39 (92.9%)  3 (7.1%)  11.0 (3.2, 37.6)  154 (56.0%)  121 (44.0%)  TPMT, ABCC3, COMT  High plus Intermediate(>0.4)  185 (64.5%)  102 (35.5%)  Low (<0.4)  8 (26.7%)  22 (73.3%)  Clinical + TPMT, ABCC3, COMT  High (>0.80) Low plus Intermediate (<0.8)  97 (91.5%)  9 (8.5%)  96 (45.5%)  115 (54.5%)  Clinical + TPMT, ABCC3, COMT  High plus Intermediate(>0.45) Low (<0.45)  161 (73.2%)  59 (26.8%)  32 (33.0%)  65 (67.0%)  OR (95% CI)a  4.8 (1.9, 11.9)  P-valuea 5.5x10-5  1.3x10-4  7.1x10-4  Sens  Spec  PPV  NPV  22.3%  96.8%  91.5%  44.4%  20.2%  97.6%  92.9%  44.0%  95.9%  17.7%  64.5%  73.3%  8.8 (4.0, 19.2)  5.5x10-8  50.3%  92.7%  91.5%  54.5%  3.1 (1.6, 5.9)  4.3x10-4  83.4%  52.4%  73.2%  67.0%  Ototox, Ototoxicity; OR, Odds Ratio; Sens, Sensitivity; Spec, Specificity; PPV, Positive Predictive Value; NPV, Negative Predictive Value a Adjusted for age, vincristine treatment, germ cell tumor and cranial irradiation b Percentage of total c Risk haplotype carriers are individuals that are heterozygous or homozygous for TPMTrs12201199 risk variant  64  Figure 2.1: Distribution of ancestry of patients treated with cisplatin The first two principal components were generated using SVS/Helixtree to visualize the distribution of population ancestry. The majority (80%) of the patients in replication cohort were of European ancestry after the removal of patients of African ancestry (n=8). A subgroup analysis of these patients (n=124), TPMT remained highly associated with cisplatin-induced hearing loss  65  A  B Model  AUC (95% CI)  p-value  Clinical variables  0.708 (0.651-0.765)  -  Genetic variables (3 SNPs)  0.696 (0.638-0.754)  0.77  Clinical and genetic variables  0.786 (0.736-0.836)  0.00048  Figure 2.2: Receiver operating characteristic (ROC) curves of clinical and genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort  (a) The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation while the genetic variables combine the effect of TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640 (b) The area under the curve (AUC) for the combined cohort for each model. The p-values indicate the statistical significance between the curves for the combination of genetic and clinical variables compared to the clinical variables alone. 66  HR (95% CI): 5.9 (2.7-12.8)  HR (95% CI): 2.3 (1.1-4.8)  Figure 2.3: Kaplan-Meier curve of cisplatin-induced hearing loss in three different risk groups (Table 2.6) combining genetic factors. The genetic variables combine the effect of TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640. The Kaplan-Meier curve shows that the incidence of hearing loss increases with increasing risk group status. Hazard ratios (HR) were used to compare curves to the lower risk group and were adjusted for clinical variables (Ptrend=5.3x10-9).  67  HR (95% CI): 3.9 (2.4-6.5)  HR (95% CI): 1.6 (0.99-2.6)  Figure 2.4: Kaplan-Meier curve of cisplatin-induced hearing loss in three different risk groups (Table 2.6) combining both clinical and genetic information. The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation while the genetic variables combine the effect of TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640. The curves show that the incidence of hearing loss increases with increasing risk group status. Hazard ratios (HR) were used to compare curves to the low risk group and were adjusted for clinical variables (Ptrend=3.4x10-19).  68  1  Chapter 3: Genetic Variants in ABCB5 and DPYD are Associated with Cisplatin-Induced Hearing Loss in Children  3.1 Introduction Cisplatin is a chemotherapeutic agent that is effective and widely used to treat a variety of solid tumours. A major limitation of its use is the risk of drug-induced ototoxicity that can result in life-long disability[117]. Cisplatin-induced ototoxicity presents itself as hearing loss and once detected, no improvement in hearing is observed and often hearing loss progress in severity long after treatment has ended[58, 59]. Cisplatin is responsible for the severe irreversible hearing loss which affects 10-25% of adults and 26-90% of children[48-53]. It is particularly devastating in children as they are at a critical stage of development. Even mild losses in hearing can significantly influence the overall quality of life including speech and language development, social-emotional development and increase the risk of learning difficulties[64, 65]. In patients experiencing severe hearing loss, hearing aids and other assistive devices are recommended. The use of hearing aids may be delayed until hearing loss has stabilized and treatment is complete. Although these devices can improve a child‟s hearing, they do not completely correct hearing loss. This suggests that by detecting hearing loss early or prior to treatment will allow children at risk of cisplatin-induced hearing loss to potentially have improved quality of life.  Clinical risk factors of cisplatin-induced hearing loss include high cumulative cisplatin dose[48, 73, 74], younger patient age at treatment[48, 55, 73], cranial irradiation[74, 79], and concomitant use of aminoglycosides[77]and vincristine[73]. It is now recognized that there is significant 1  A version of this chapter will be submitted for publication. Pussegoda K, Ross C.J, Yazdanpanah M, Brooks B, Rassekh S.R, Carleton B.C, Hayden M.R. Genetic variants in ABCB5 and DPYD are associated with cisplatin-induced hearing loss in children. 69  inter-individual variability in hearing loss with patients receiving similar doses of cisplatin suggesting that clinical risk factors alone are insufficient predictors of safety[58, 64]. Genetic variation in the genes involved in drug biotransformation, transport, and drug targets have been recognized to influence patient drug response and susceptibility to adverse drug events, including ototoxicity[29]. Genetic markers may help to identify patients that confer increased susceptibility to cisplatin-induced ototoxicity.  A recent genetic study in children receiving cisplatin developed a multi-marker prediction model that include genetic variants (TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640) conferring increased risk of developing cisplatin-induced hearing loss[64]. The genetic model was able to identify patients at high risk with a specificity of 97.6% and a sensitivity of 20.2% (Chapter 2). However, 79.8% of patients with hearing loss are not classified as high risk and wrongly predicted 2.4% of patients that developed hearing loss using this model. The aim of this study was to investigate whether additional genetic risk factors improve the prediction of risk for cisplatin-induced hearing loss in pediatric patients and to refine the current predictive multi-SNP model.  3.2 Material and Methods 3.2.1 Patients Study participants were recruited through the Canadian Pharmacogenomics Network for Drug Safety (CPNDS)[118]. This cohort of children was recruited at pediatric oncology units across Canada: BC Children‟s Hospital (Vancouver), Alberta Children‟s Hospital (Calgary), Stollery Children‟s Hospital (Edmonton), Winnipeg Health Sciences Centre (Winnipeg), Children‟s  70  Hospital at the London Health Sciences Centre (London), McMaster Children‟s Hospital (Hamilton), Hospital for Sick Children (Toronto), Children‟s Hospital of Eastern Ontario (Ottawa), Hôpital Sainte-Justine (Montreal) and IWK Health Centre (Halifax). The initial cohort and replication cohort that are included in the combined analyses have been described previously[64]. Cisplatin-induced hearing loss was defined based on audiometric findings using CTCAEv3 (Common Terminology Criteria for Adverse Events) criteria as described previously[61, 64]. Briefly, to better differentiate between cases and controls, subjects with serious cisplatin hearing loss were defined as those with CTCAE grade 2 or higher hearing impairment after treatment with cisplatin. Control patients were defined as children who received cisplatin with normal audiometric data who did not develop hearing loss (grade 0). Patients with grade 1 hearing loss were excluded. Written informed consent or assent was obtained from each subject or their parents or legal guardians. The study was approved by the ethics committees of all participating universities and hospitals. 3.2.2 Genotyping Genomic DNA was extracted from blood, saliva or buccal swabs using the QIAmp DNA purification system (Qiagen, Toronto, ON, Canada) according to the manufacturer‟s protocol. DNA samples were genotyped for 4536 single nucleotide polymorphisms (SNPs) using a custom Illumina GoldenGate (SNP) genotyping assay (Illumina, San Diego, CA, USA). The panel was designed to include the genetic variants of the original custom ADME panel, increase the coverage of genetic variants and cover additional genes involved in drug biotransformation[64].  71  While the original panel captures the genetic variation of approximately 220 key genes, this enhanced custom panel covers 315 key genes involved in drug biotransformation including phase I and II drug metabolizing enzymes, drug transporters, drug receptors, ion channels, drug targets, transcription factors and other disease-specific genes related to the physiological pathway of cisplatin. Furthermore, the enhanced panel was optimized to avoid genomic interferences such as regions of homology to reduce the number of failed SNPs. All SNP genotype data were clustered manually using GenomeStudio software (Illumina, San Diego, CA, USA). SNPs that could not be clustered or were nonpolymorphic were excluded from further analysis. In total 4067 SNPs were available for analysis. Samples with a call rate below 95% were excluded. Two cases from the initial cohort with a call rate <95% were removed. The remaining 315 samples had an average genotyping call rate for included samples were 99.9%.  3.2.3 Statistical Analysis Statistical genetic analyses were conducted using SNP and Variation Suite 7.4.5 (Golden Helix, Bozeman, USA), SPSS Statistics 19 (SPSS Inc., Chicago, IL, USA) and R 2.13.0 (R Development Core Team). Clinical characteristics of patients with and without cisplatin-induced hearing loss were compared using the Wilcoxon-Mann-Whitney U-test for continuous variables, and Fisher‟s exact test for categorical variables. We adjusted for multiple testing using the simpleM correction and calculated the effective number of independent tests (MeffG) at 308 for a significance threshold of 0.00016. Hardy-Weinberg equilibrium tests were conducted in controls using the permutation version of the exact test of Hardy-Weinberg of Guo and Thompson[119]. Forty-six SNPs were removed due to Hardy-Weinberg disequilibrium and 71 SNPs with a completion rate less than 72  90%. In total, 3943 SNPs were available for analysis. To assess population stratification in the data, principal component analysis (PCA) was conducted using SVS/HelixTree software[112, 137]. Patients of African ancestry (n=8), as determined by PCA, were excluded in the analysis. Homozygous and heterozygous odds ratios (ORs) using the homozygous genotype of the protective allele was used as a reference. Association between genetic polymorphisms and cisplatin-induced hearing loss was assessed by computing Fisher‟s exact test for unadjusted (non-regression) p-values and adjusted p-values determined by logistic regression. Independent factors for the prediction of cisplatin-induced hearing loss in the combined cohort were identified by forward logistic regression analysis. We included variables that were retained in the combined cohort and variables that are known to increase risk of hearing loss. Therefore, age, concomitant vincristine treatment, germ cell tumour, and cranial irradiation were included as covariates. In an additional analysis, we combined the initial cohort[64] and the replication cohort of patients to determine the overall significance of associations. Genetic variants that were significant (P<0.05) in the initial cohort that were also significant (P<0.05) in the replication cohort were further evaluated in a multivariate logistic regression model including clinical variables. A genetic risk score was calculated by multiplying each variable with the estimated beta (log odds ratio) from the current combined cohort. The initial genetic model (Chapter 2) was constructed based on statistical evidence from a logistic regression model in the combined cohort, where TPMT rs12201199, ABCC3 rs1051650 and COMT rs4646316 were retained. A revised model was constructed based on SNPs from the initial model (Chapter 2) as well as SNPs from the current study that when combined with the initial multi-marker model, independently increased the prediction of hearing loss. Genetic variants that significantly increased (P<0.05) the area under the curve (AUC) of the initial model 73  include ATP-Binding Cassette transporter B 5 (ABCB5) rs10950831 (P=0.020) and dihydropyrimidine dehydrogenase (DPYD) rs6667550 (P=0.046) and were therefore included into a new genetic prediction model. The final model combined the effects of TPMT rs12201199, COMT rs4646316, ABCC3 rs1051640, ABCB5 rs10950831 and DPYD rs6667550. Association between the genetic risk score and cisplatin-induced hearing loss was assessed using logistic regression. Three prediction models were investigated: (i) clinical variables; (ii) genetic variables; and (iii) clinical and genetic variables. The contribution of the genetic and or clinical risk scores to the prediction of cisplatin-induced hearing loss was investigated by comparing the area under the receiver-operating-characteristic (ROC) curves of the prediction models. AUC estimates were obtained using the ROC plot function on the basis of the linear predictors obtained from the logistic regression analyses. The statistical difference between the curves was calculated using DeLong‟s method[120] implemented in the R-package pROC[121]. Two-sided p-values <0.05 were considered statistically significant. Risk groups for cisplatin-induced hearing loss were defined based on the predictive values for each patient sample from the multi-SNP logistic regression model that included five SNPs (TPMT rs12201199, ABCC3 rs1051640, COMT 4646316, ABCB5 rs10950831 and DPYD rs6667550). The threshold used to determine low risk was the median predicted value of controls (predictive value <0.45) and high risk was defined as the mean predicted value of cases (>0.75). Intermediate risk was defined as a predictive value between 0.45-0.75. To assess the predictive performance of risk groups for cisplatin-induced hearing loss, we calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for each defined threshold. Kaplan-Meier curves for cisplatin-induced hearing loss were generated for each risk group using time from start of treatment to the time of first evidence of toxicity or last 74  audiogram. Log-rank test was used to compare the trend of survival curves. Hazard ratios for each curve were calculated using the Cox regression model considering the low risk group as the reference.  3.3 Results 3.3.1 Patients Patient characteristics of the initial cohort replication cohort have been described previously ([64], Chapter 2). Two cases in the initial cohort were excluded due to a call rate of <95%. Therefore, in the combined cohort 191 (61%) of the 315 pediatric oncology patients developed hearing loss (Table 3.1). Concomitant vincristine treatment was significantly higher in cases than controls in the combined cohort (58.6% compared to 27.9%, P=5.5×10-10). Fewer patients with germ cell tumour developed hearing loss (6.8% vs. 21.0%, P=0.00035). Furthermore, follow-up after therapy was longer in cases than controls (4.0 years vs. 2.0 years, P=2.1x10-4). Overall median follow up time of patients was 3 years.  3.2.2 Genetic Results We identified an intronic SNP (rs10950831) in ABCB5 that was highly associated with cisplatininduced hearing loss in both the initial cohort (P=0.013) and the replication cohort (P=0.0012) (Table 3.2). ABCB5 rs10950831 was the most strongly associated variant that remained significant after multiple testing in the combined cohort (P=1.06x10-6, OR 2.0). The risk allele (A) was observed in 154 (80.6%) cases and 62 (50.0%) controls in the combined cohort. The association of ABCB5 rs10950831 genotype with cisplatin-induced hearing loss remained significant in the combined cohort after adjusting for clinical factors (P=3.67x10-5). 75  In addition to ABCB5, we also found suggestive evidence for several additional genes associated with cisplatin-induced hearing loss (Table 3.2). SLC28A3 rs17087056 was significantly associated after multiple-testing correction in the combined cohort (P=0.00011, OR 3.2). When the association with SCL28A3 rs17087056 was adjusted for clinical factors, the significance of the effect decreased (P=0.00011 before adjustment vs. P=0.0029 after adjustment). Another genetic variant of interest that was close to the multiple testing threshold was DPYD rs6667550; however, this variant did not pass the threshold for multiple testing correction (P=0.00047, OR 1.9). Interestingly, the effect of DPYD rs6667550 became more significant in the combined cohort after adjusting for clinical factors (P=2.12x10-4).  The initial predictive model which combined the effect of TPMT, ABCC3 and COMT was able to predict 69.6% of cases (P= 7.16×10-5) (Figure 3.1). We selected additional variants to include into a multi-marker regression model based on evidence of a significant increase (P<0.05) in prediction of hearing loss from the initial model (TPMT, ABCC3, and COMT). ABCB5 rs10950831 and DPYD rs6667550 significantly improved (P<0.05) the prediction of hearing loss (AUC) from the initial multi-marker model and was therefore were included in the final model. In contrast, ROC analysis showed that by including SLC28A3 rs17087056 for the prediction of cisplatin-induced hearing loss (AUC) did not significantly improve (P>0.05) the initial model. Therefore, the ABCB5 rs10950831 and DPYD rs6667550 variants were selected for inclusion in the predictive model (Figure 3.2).  76  An ROC analysis that combined the effects of the associated risk genotypes and clinical risk factors was performed. The clinical-only model includes age, vincristine treatment, germ cell tumour and cranial irradiation. By including ABCB5 and DPYD to the initial predictive model we can significantly increase the ability to predict cisplatin-induced hearing loss (AUC 0.775 vs AUC 0.696, P=0.0023) (Figure 3.2). Therefore the final genetic-only model was constructed to include five independent SNPs (TPMT rs12201199, ABCC3 rs1051640, COMT 4646316, ABCB5 rs10950831 and DPYD rs6667550). Furthermore, combining both the clinical and genetic variables significantly increased the ability to predict cisplatin ototoxicity (Figure 3.3) resulting in an area under the curve (AUC) of 0.841 compared to the clinical-only model (AUC 0.712, P=2.63x10-7).  A predictive multi-marker model using these genetic variants could stratify patients by risk of ototoxicity. Risk groups were defined based on predictive values from the model as low (predictive value <0.45), intermediate (0.45-0.75) and high risk (>0.75). Based on these groups, there were 95 (30.2%) individuals classified as low risk, 122 (38.7 %) as intermediate and 98 (31.1%) as high risk. The initial model combining TPMT, ABCC3 and COMT conferred a specificity of 97.6% and sensitivity of 20.2% (Table 2.6). In the new model, 87 (88.8%) of individuals in the high risk group developed hearing loss (positive predictive value, 88.8%) compared to only 11 (11.2%) controls conferring a specificity of 91.1% and a sensitivity of 45.5% (Table 3.3). Although the specificity of the test decreases slightly compared to the initial model (91.1% vs. 97.6%), the sensitivity of the new model is two-fold higher (45.5% vs. 20.2%).  77  Individuals in the high risk group also had a significantly higher risk of cisplatin-induced ototoxicity when compared to low to intermediate risk group (P=1.72x10-9, OR 9.1; Table 3.3). A Kaplan-Meier plot illustrates that the intermediate and high risk groups have significantly increased risk of hearing loss over time (Ptrend=5.7x10-15; Figure 3.4). After 1 year of treatment, 61% of patients developed significant hearing loss in the high risk group compared to 22% of patients in the low risk group (Figure 3.3). A Kaplan-Meier plot combining both the clinical and genetic variables illustrates that we can identify 97 (50.8%) patients at high risk compared to 87 (45.5%) patients using genetics alone (Figure 3.5, Table 3.3). In the high risk group 97 (94.2%) of individuals developed hearing loss (positive predictive value, 94.2%) compared to 6 (5.8%) controls conferring a specificity of 95.2% and a sensitivity of 50.8% (Table 3.3). The initial model combining TPMT, ABCC3, COMT and clinical factors conferred a specificity of 92.7% and sensitivity of 50.3% (Table 2.6). Overall, by including ABCB5 and DPYD to the initial model (TPMT, ABCC3, and COMT) the specificity of the test increases from 92.7% to 95.2%, with a small increase in sensitivity (50.8% vs. 50.3%).  3.3 Discussion The ability to predict patients at risk of developing cisplatin-induced hearing loss would significantly improve the safety and efficacy of pediatric cancer therapy. The inter-individual variability in hearing loss even with treatment at similar doses suggests that there may be a genetic component. We identified genetic variants in ABCB5 and DPYD that were associated with cisplatin-induced hearing loss in children treated for a variety of malignancies. Combining these genetic variants into a predictive model could improve the ability to discriminate more patients that are at high and low risk for cisplatin-induced hearing loss.  78  The strongest associated SNP was rs10905831, an intronic variant located in ABCB5. Patients carrying the A-allele of rs10905831 were at increased risk of developing hearing loss after cisplatin treatment. This is the first study to describe the role of genetic variation in ABCB5 in the context of cisplatin-ototoxicity. ABC transporters are known to confer the efflux of a variety of drugs including cisplatin[81, 82]. Polymorphisms in ABCB5 may regulate expression levels or affect function of the transporter which can lead to intracellular accumulation of cisplatin and toxicity.  Other SNPs that were associated occurred in genes encoding for proteins that may be involved in processes that affect cisplatin absorption, distribution, metabolism and excretion. Genetic variants in these genes can influence the pharmacokinetics of cisplatin and affect toxicity and survival[29]. Another intronic variant in DPYD (rs6667550) was also associated with cisplatininduced hearing loss. One study has found that the activity DYPD decreased in patients treated with a combination of cisplatin-5-fluorouracil therapy[138]. This indicates that cisplatin may interact with DPYD and affect the activity of the enzyme. Furthermore the study suggests that individuals with low DPYD activity may be at increased risk of toxicity when receiving platinum compounds[138]. How intronic regions affect susceptibility to cisplatin-induced hearing loss remains unclear. Intronic SNPS can be linked to functional SNPs within the coding area of the gene or regulate function and expression. Further functional studies will be required to assess the exact mechanisms by which these variants affect cisplatin-induced hearing loss. We also found no significant associations with the previously reported variants in megalin (LRP) rs2075252, GSTP1 rs1695, presence of GSTM1, and XPC rs2228001 (P=0.97, P=0.16, P=0.12, P=0.49).  79  The differences in the clinical characteristics between cases and controls were corrected for by including them as covariates in the logistic regression analyses. High cumulative cisplatin dose, younger patient age at treatment, concomitant use of aminoglycosides and cranial irradiation are known to influence cisplatin-induced ototoxicity[59]. Logistic regression analyses suggest that there was little to no effect of population stratification within our cohort of patients. Inter-study heterogeneity can lead to differences in the effects of variants in specific subpopulations or lead to masking of the effects by factors that were not controlled for. We aimed to control for these factors by adjusting for clinical variables in the cohort and identifying potential population stratification.  Although clinical factors have been identified, they alone cannot reliably predict the risk of developing hearing loss. We generated a refined predictive model by incorporating genetic risk in ABCB5 and DPYD to the initial genetic model which combines the effect of TPMT, ABCC3 and COMT. We were able to improve the genetic component of the risk prediction model by including ABCB5 rs10905831 and DPYD rs6667550 into the initial genetic model (AUC 0.775 vs. 0.696; P=0.0023). Furthermore, by combining genetic and clinical risk factors were able better discriminate between patients at high and low risk. The combined model of genetic plus clinical risk factors was more predictive than clinical risk factors alone (AUC 0.841 vs. 0.712; P=2.63x10-7). In the high risk group 88.8% of cases were predicted to develop hearing loss while only 11.2% of controls were predicted to be at high risk conferring a sensitivity of 45.5%. The increase in the sensitivity from the initial model is significantly improved from the initial model (20.2%) (Chapter 2). We are able to better identify patients at high risk with the new refined predictive model. Furthermore, including clinical and genetic risk factors we can improve the  80  specificity of the test (95.2% in the current model vs. 92.7% in the initial model) while maintaining a sensitivity of 50%. Whether patients at low risk require less intensive monitoring or can be treated at higher doses to improve survival without significantly increasing cisplatininduced hearing loss needs to be further investigated. The model was based on results from the combined cohort and requires further studies in children and adults to validate these findings to assess the clinical utility.  There are an increasing number of children surviving cancer treatment. Given that cisplatininduced hearing loss is permanent and progressive, it is increasing important to reduce or prevent hearing loss, particularly in high risk groups. In children this is of particular importance as even mild losses in hearing can cause difficulties in learning in school[65, 134]. Several preventative measures have been proposed in patients that are high risk of adverse effects from cisplatin. Carboplatin which is less ototoxic agent is suggested as alternative medication for treatment [55]. Other alternatives include increased monitoring, lowering dose and/or otoprotective agents[135, 136]. However, there are concerns about the possibility of compromise of anti-tumour activity have been raised with these alternatives strategies.  This study demonstrates that it is possible to accurately discriminate patients at higher and lower risk for cisplatin-induced hearing loss based on a patient‟s genetic profile in combination with clinical risk factors. This can influence treatment decisions and improve monitoring of patients at risk of adverse events. Sequencing of ABCB5 and DPYD may identify functional variants that may affect expression and activity which can lead to a mechanism by which these genes influence cisplatin toxicity. Further validation of the prediction model in an independent cohort  81  of patients is required prior to use in clinic. Because this study includes mainly patients of European origin, it will also be important to assess these variants and the risk of cisplatininduced hearing loss in other ethnic cohorts where allele frequencies and linkage disequilbrium may differ. In addition, functional studies can be conducted to assess the exact mechanisms by which variants in ABCB5 and DPYD affect cisplatin-induced hearing loss. Identifying genetic risk factors can potentially improve risk classifications for hearing loss and may allow for individualized treatment and improve safety in children.  82  Table 3.1 Patient demographics of the combined cohort Combined Cohort (n=315) Cases Controls (n= 191) (n=124) p-value Age, years (median (min, max)) Dose, cumulative mg/m2 (median (min, max)) Treatment duration, weeks (median (min, max)) Gender (Male n, (%)) Caucasian ethnicitya (n (%)) Concomitant medication (n, (%)) Tobramycin Vancomycin Vincristine Gentamicin Tumor type (n, (%)) brain tumor endodermal sinus tumor of thymus germ cell tumor hepatoblastoma lymphoma nasopharyngeal carcinoma neuroblastoma osteosarcoma sarcoma carcinoma retinoblastoma mesencymal tumour of liver Follow-up, years (median (min, max)) Cranial irradiation (n, (%))  6.0 (0, 25)  10.0 (0, 19)  0.012  400 (92, 800)  400 (20, 768)  0.80  2.2 (0, 11.4)  2.3 (0, 8.5)  0.82  112 (58.6%) 149 (78.0%)  62 (50.0%) 102 (82.3%)  0.16 0.39  54 (28.3%) 50 (26.2%) 112 (58.6%) 42 (22.0%)  30 (24.2%) 25 (20.2%) 29 (23.4%) 26 (21.0%)  0.44 0.28 5.5×10-10 0.89  51 (26.4%)  19 (15.3%)  0.019  0  2 (1.6%)  0.15  13 (6.8%) 39 (20.4%) 1 (0.5%)  26 (21.0%) 12 (9.7%) 3 (2.4%)  0.00035 0.012 0.30  1 (0.5%)  2 (1.6%)  0.39  50 (26.2%) 33 (17.3%) 1 (0.5%) 0 1 (0.5%) 1 (0.5%)  21 (16.9%) 36 (29.0%) 2 (1.6%) 1 (0.8%) 0 0  0.072 0.018 0.56 0.39 1.00 1.00  4 (0, 25)  2 (0, 16)  6.4×10-5  43 (22.5%)  15 (12.1%)  0.025  For age, dose, treatment duration and follow-up, the Wilcoxon-Mann-Whitney test with normal approximation was used. For gender, ethnicity, concomitant medication, tumor type, cranial irradiation the Fisher exact test was used. In bold are statistically significant values at p<0.05 type-I error rate. a Caucasian ethnicity assessed by principal component analysis  83  Table 3.2 Association between SNPs and cisplatin-induced hearing loss Previous Cohorta (n=160) 104 ototox., 56 controls  Replication (n=155) 87 ototox., 68 controls  Combined (n=315) 191 ototox., 124 controls  Gene  SNP  Alleleb  OR (95% CI)  p-value  OR (95% CI)  pvalue  OR (95% CI)  pvalue  p-valuec  ABCB5  rs10950831  A/G  1.9 (1.2, 3.2)  0.013  2.2 (1.4, 3.6)  0.0012  2.0 (1.4, 2.9)  1.06x10-6  3.67x10-5  SLC28A3 DPYD  rs1087056 rs6667550  A/T G/A  2.6 (1.1, 5.8) 1.7 (1.0, 2.9)  0.026 0.037  4.3 (1.7, 11.1) 2.3 (1.3, 3.9)  0.028 0.0036  3.2 (1.7, 5.9) 1.9 (1.4, 2.8)  0.00011 0.00047  0.0029 2.12x10-4  XDH XDH RALBP1  rs6710015 rs207425 rs3322  G/A A/G A/G  1.9 (1.2, 3.2) 1.8 (1.1, 3.2) 2.5 (1.1, 5.6)  0.010 0.033 0.023  1.7 (1.0, 2.7) 1.7 (1.0, 2.7) 3.5 (1.3, 9.2)  0.045 0.045 0.012  1.8 (1.3, 2.5) 1.7 (1.2, 2.4) 2.8 (1.5, 5.1)  0.0011 0.0052 0.0064  0.0043 0.035 0.0058  PPARG  rs1152003  G/A  1.8 (1.0, 3.1)  0.042  2.2 (1.2, 4.0)  0.013  1.8 (1.3, 2.5)  0.0012  0.0052  a  Results from discovery and replication cohorts from previous study[64] compared to the replication cohort as well as all cohorts combined SNP alleles assayed; minor allele is mentioned first; P-values in bold are significant after multiple testing correction in combined cohort c Adjusted for age, vincristine treatment, germ cell tumor and cranial irradiation SNP, Single Nucleotide Polymorphism; Ototox, Ototoxicity; OR, Odds Ratio; CI, Confidence Interval. b  .  84  Table 3.3: Comparison of risk groups Modeld  Risk Group  Genetic only  High (>0.75) Low plus Intermediate (<0.75)  Genetic only  High plus Intermediate(>0.45) Low (<0.45)  Clinical + Genetic  High (>0.82) Low plus Intermediate (<0.82)  Clinical + Genetic  High plus Intermediate(>0.34) Low (<0.34)  Ototox. (Number, %)  Controls (Number, %)  OR (95% CI)a  P-valuea  Sens  Spec  PPV  NPV  87 (88.8%)  11 (11.2%)  9.1 (4.4, 18.6)  1.72x10-9  45.5%  91.1%  88.8%  52.1%  104 (47.9%)  113 (52.1%)  159 (72.3%)  61 (27.7%)  6.3 (3.5, 11.3)  5.73x10-10  83.2%  50.8%  72.3%  66.3%  32 (33.7%)  63 (66.3%)  97 (94.2%)  6 (5.8%)  8.8 (4.0, 19.2)  5.5x10-8  50.8%  95.2%  94.2%  55.7%  94 (44.3%)  118 (55.7%)  177 (74.1%)  62 (25.9%)  3.1 (1.6, 5.9)  4.3x10-4  92.7%  50.0%  74.1%  81.6%  14 (18.4%)  62 (81.6%)  Ototox, Ototoxicity; OR, Odds Ratio; Sens, Sensitivity; Spec, Specificity; PPV, Positive Predictive Value; NPV, Negative Predictive Value a Adjusted for age, vincristine treatment, germ cell tumor and cranial irradiation b Percentage of total c Risk haplotype carriers are individuals that are heterozygous or homozygous for TPMTrs12201199 risk variant d Clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation while the genetic variables include TPMT rs12201199, COMT rs4646316, ABCC3 rs1051640, ABCB5 rs10950831 and DPYD rs6667550  85  A  B Model  AUC (95% CI)  p-value  Clinical Variables  0.713 (0.656, 0.770)  -  TPMT  0.594 (0.532-0.656)  -  TPMT and ABCC3  0.661 (0.601-0.722)  0.0030  Initial model (TPMT, ABCC3 and COMT)  0.696 (0.638-0.755)  7.16×10-5  Figure 3.1: ROC curves of genetic variables in the initial model for the prediction of cisplatin-induced hearing loss in the combined cohort (a) The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation. The initial model combines the effect of TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640 as described previously in a step-wise manner (b) The area under the curve (AUC) for the combined cohort for each model. The p-values indicate the statistical significance between each genetic marker in the initial model compared to TPMT rs12201199 alone.  86  A  B Model  AUC (95% CI)  p-value  Clinical Variables Initial Model (TPMT, ABCC3, COMT) Initial Model + ABCB5 Initial Model + ABCB5 and DPYD  0.712 (0.654, 0.769) 0.696 (0.638-0.755) 0.752 (0.698-0.806) 0.777 (0.726-0.828)  0.020 0.0023  Figure 3.2: ROC curves of new genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort (a) The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation. The initial model combines the effect of TPMT rs12201199, COMT rs4646316 and ABCC3 rs1051640 as described previously. New genetic variants that statistically improve the initial model were included in a step-wise manner. The final model combines the effect of the initial variants and ABCB5 rs10950831 and DPYD rs6667550 (b) The area under the curve (AUC) for the combined cohort for each model. The p-values indicate the statistical significance between the new genetic models compared to the initial genetic model. 87  A  B  Model  AUC (95% CI)  p-value  Clinical Variables  0.712 (0.654, 0.769)  -  Genetic Variables (5 SNPs)  0.777 (0.726-0.828)  0.11  Clinical and New Genetic Variables  0.841 (0.799-0.884)  2.63x10-7  Figure 3.3: ROC curves of clinical and newly identified genetic variables for the prediction of cisplatin-induced hearing loss in the combined cohort (a) The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation while the genetic variables include TPMT rs12201199, COMT rs4646316, ABCC3 rs1051640, ABCB5 rs10950831 and DPYD rs6667550 (b) The area under the curve (AUC) for the combined cohort for each model. The p-values indicate the statistical significance between the curves for the combination of genetic and clinical variables compared to the clinical variables and the genetic variables alone.  88  HR (95% CI): 4.5 (3.0-6.8)  HR (95% CI): 2.2 (1.5-3.3)  Figure 3.4: Kaplan-Meier curve of cisplatin-induced hearing loss in different risk groups (Table 3.3) combining genetic factors. The genetic variables combine the effect of TPMT rs12201199, COMT rs4646316, ABCC3 rs1051640, ABCB5 rs10950831 and DPYD rs6667550. The Kaplan-Meier curve shows that the incidence of hearing loss increases with increasing risk group status. Hazard ratios (HR) were used to compare curves to the low risk group and were adjusted for clinical variables (Ptrend=5.7x10-15).  89  HR (95% CI): 9.1 (4.9-17.0)  HR (95% CI): 4.2 (2.3-7.6)  Figure 3.5: Kaplan-Meier curve of cisplatin-induced hearing loss in different risk groups (Table 3.3) combining clinical and genetic factors. The clinical variables include age, vincristine treatment, germ cell tumours and cranial irradiation while the genetic variables include TPMT rs12201199, COMT rs4646316, ABCC3 rs1051640, ABCB5 rs10950831 and DPYD rs6667550. The Kaplan-Meier curve shows that the incidence of hearing loss increases with increasing risk group status. Hazard ratios (HR) were used to compare curves to the low risk group and were adjusted for clinical variables (Ptrend=1.7 x10-23).  90  Chapter 4: Discussion and Future Directions  Cisplatin is of the most effective and widely used chemotherapeutic agents for the treatment of solid tumours; yet, it‟s application is limited by cisplatin-induced hearing loss and other ADRs. The studies conducted as part of this thesis employed a candidate gene approach to screen genes involved in drug ADME to identify novel genetic variants associated with cisplatin-induced hearing loss. This was the first study to replicate previous associations and develop predictive models to discriminate between pediatric patients that are at high versus low risk of cisplatininduced hearing loss using both genetic and clinical factors. 4.1 Strengths and Limitations This study has the largest cohort of well phenotyped pediatric patients on cisplatin therapy. Patients were phenotyped using CTCAE grading scheme and those with mild hearing loss (Grade 1) were excluded to evaluate more extreme phenotypes. Cisplatin-induced hearing loss often increases in severity over time, even after completion of cisplatin treatment (i.e. patients with Grade 1 hearing loss progress to Grade 2 to 4). Furthermore, we aimed to control for confounding factors by adjusting for clinical variables in the cohort and carrying out a subgroup analyses in patients of European ancestry. For cisplatin, the currently identified genetic variants do not independently improve the risk of prediction over clinical factors. This highlights a key limitation in the predictive models that have been generated. Ideally we would have liked to identify genetic predictors that have a greater impact on drug effects than clinical parameters alone. One possible explanation may be that additional novel genetic risk factors may be associated with cisplatin-induced hearing loss, which may increase the genetic contribution to this ADR. However, it is also possible that other 91  factors such as environment and clinical parameters contribute significantly to the risk of cisplatin-induced hearing loss. The clinical risk factors that have been identified in this study are currently not used to modify treatment-decisions. This study has shown that clinical factors indeed do contribute to the risk and susceptibility of hearing loss with cisplatin treatment and therefore, should be taken into account in clinical treatment decision-making. However, clinical factors alone are not adequate to predict hearing loss and by adding genetic risk factors we have shown that we can improve prediction of risk and to better classify patients into different risk groups. Another limitation of this study is that the prediction models were generated were based on the results from the combined cohort and have not yet been replicated. Therefore, it is possible that we are over-fitting the model. In addition we have selected variants that are known to affect patheways involved in ADME. Other genetic variants outside drug metabolism pathways may also affect cisplatin-induced hearing loss, but were not investigated in this study.  4.2 Replication and Functional Studies For clinical application, it is essential to validate genetic findings in additional independent cohorts of patients to reduce the number of false positive results and to ensure that the genetic risk prediction is consistent and precise. A standardized phenotyping method would significantly facilitate accurate replication studies, and enable cross-study comparison of research findings. Many current studies have a small number of patients in which an effect was detected, but with little or no independent replication of these findings it is not possible to draw conclusions for clinical application. Furthermore, genetic risk variants may also differ with treatment for different malignancies because of differences in treatment protocols. However, due to low sample sizes many studies have found it difficult to carry out subgroup analyses to determine 92  whether identified variants are associated with specific treatment protocols. In addition, a majority of studies have been carried out in populations of only European ancestry. Whether these same genetic variants are also associated in different patient populations with other ancestries still needs to be investigated.  As a next step, replication of the novel variants that were identified in this study and the predictive models should be evaluated in additional independent cohorts. The replication of these findings in both pediatric and adult patients treated with cisplatin is required for validation to ensure findings are consistent and precise and to provide an unbiased selection of SNPs for the predictive model. It is important to recognize that children and adults metabolize drugs differently, which suggests that different genetic variants may be responsible for cisplatininduced toxicity between these two groups.  Another method to validate findings is through functional validation studies to investigate the mechanisms that underlie the associations between cisplatin-induced hearing loss and the identified genes. One possible method would be to evaluate expression and activity of functional genetic variants in the genes that were identified in this study in cell culture in the presence of cisplatin. This method may confirm whether certain genotypes are associated with increased toxicity.  93  4.3 Investigation of Additional Gene and Gene Variants Involved in Cisplatin-Induced Hearing Loss The identification of additional genetic risk variants for cisplatin ototoxicity may explain cases that cannot currently be predicted using known risk factors. Furthermore, these variants may have the potential to improve the positive predictive power and/or the sensitivity of genetic testing. As a next step, re-sequencing the genes that were associated with cisplatin-induced hearing loss would facilitate the identification of additional variants that may be implicated. This may include variants in linkage with the newly identified SNPs as well as unknown variants within the same gene that are independently associated. To complement this targeted candidate gene approach, a genome-wide association study (GWAS) could also be performed to investigate cisplatin-induced hearing loss at the genome-wide level. A GWAS is a hypothesis-free approach that can screen for an association with cisplatin-induced hearing loss in more than one million SNPS across the human genome. The benefit of performing a GWAS is to identify novel genes we would not suspect to contribute to cisplatin-induced hearing loss. These studies usually require large samples sizes for sufficient power to detect a true positive association due to the number of variants that are analyzed.  4.4 Implementing Genetic Testing Pharmacogenomic testing has the potential to improve the safety and efficacy of medications. With well-established clinical applications and the variety of genotyping methods that have been established, pharmacogenomic testing is still rarely used in clinical practice. It has become widely accepted for clinicians to treat patients based on a „trial and error‟ dosing approach. Therefore, there is resistance to using tests to influence medical decisions, particularly when it 94  comes to pharmacogenetic testing because it requires interpretation of genotypes and additional training in genetics and pharmacology to incorporate it into clinical decision-making. The association studies for cisplatin-induced hearing loss thus far have been retrospective. To assess the real-world clinical utility of predictive pharmacogenetic testing will require prospective studies to assess safety and efficacy. This is a challenge because it is not well-known whether alternative therapies have similar efficacies in specific tumor types. Cisplatin prospective trials will allow clinicians to develop strategies to manage hearing loss such as dose modifications, use of alternate medications that are just as effective, or through the use of otoprotective agents. Whether we need to wait for prospective trials prior to implementing genetic testing for cisplatin risk variants is still under debate. However, with the information and understanding we have gained from studying and replicating genetic factors associated with cisplatin-induced hearing loss, it is difficult to not address these concerns. Patients and parents of patients undergoing cisplatin therapy should be aware of these risk factors in order to have the choice to be genotyped for these genetic risk markers. At least patients identified to be at lower risk may have one less thing to worry about with chemotherapy treatment, while those patients at high risk can monitor their hearing more closely. Knowing the risk status for cisplatin-induced hearing loss may impact the patients‟ quality of life, such as the social-emotional development in children. Although challenging, pharmacogenomics has the potential to translate knowledge of genetic variability into better therapeutics and improve quality of life.  4.5 Conclusions This thesis makes important contributions to the pharmacogenomics of cisplatin. This study adds important new evidence to understanding the mechanisms behind cisplatin-induced hearing loss 95  given that cisplatin biotransformation in the body is not well understood. Currently in the area of cisplatin-induced hearing loss, there is a lack of replication of genetic associations, a lack of prediction models to identify patients at risk of toxicity and an absence of guidelines for pharmacogenomic testing. We were able to replicate previous associations in a large independent cohort of pediatric patients. Furthermore, this study shows that it may be possible to classify individuals at risk of cisplatin-induced hearing loss based on a patients‟ genetic profile in combination with clinical risk factors. It also provides new hypotheses for the mechanisms underlying the susceptibility for cisplatin-induced hearing loss in patients treated with cisplatin. These results have the potential to influence modifications to treatment decisions for cisplatin therapy to improve safety and efficacy of medications, bringing personalized medicine one step closer to reach.  96  References 1.  2. 3.  4. 5. 6. 7. 8. 9. 10. 11. 12.  13. 14. 15. 16.  17. 18. 19.  Kalow, W., B.K. Tang, and L. Endrenyi, Hypothesis: comparisons of inter- and intraindividual variations can substitute for twin studies in drug research. Pharmacogenetics, 1998. 8(4): p. 283-9. Edwards, I.R. and J.K. Aronson, Adverse drug reactions: definitions, diagnosis, and management. Lancet, 2000. 356(9237): p. 1255-9. Lazarou, J., B.H. Pomeranz, and P.N. Corey, Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA, 1998. 279(15): p. 1200-5. Ernst, F.R. and A.J. Grizzle, Drug-related morbidity and mortality: updating the cost-ofillness model. J Am Pharm Assoc (Wash), 2001. 41(2): p. 192-9. White, T.J., A. Arakelian, and J.P. Rho, Counting the costs of drug-related adverse events. Pharmacoeconomics, 1999. 15(5): p. 445-58. Crilly, J., The history of clozapine and its emergence in the US market: a review and analysis. Hist Psychiatry, 2007. 18(1): p. 39-60. Colditz, G.A., Economic costs of obesity and inactivity. Med Sci Sports Exerc, 1999. 31(11 Suppl): p. S663-7. Pirmohamed, M., et al., Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ, 2004. 329(7456): p. 15-9. Alvir, J.M., et al., Clozapine-induced agranulocytosis. Incidence and risk factors in the United States. N Engl J Med, 1993. 329(3): p. 162-7. Pirmohamed, M., et al., Phenotype standardization for immune-mediated drug-induced skin injury. Clin Pharmacol Ther, 2011. 89(6): p. 896-901. Revuz, J.E. and J.C. Roujeau, Advances in toxic epidermal necrolysis. Semin Cutan Med Surg, 1996. 15(4): p. 258-66. Greener, M., Drug safety on trial. Last year's withdrawal of the anti-arthritis drug Vioxx triggered a debate about how to better monitor drug safety even after approval. EMBO Rep, 2005. 6(3): p. 202-4. Hazell, L. and S.A. Shakir, Under-reporting of adverse drug reactions : a systematic review. Drug Saf, 2006. 29(5): p. 385-96. Belton, K.J., et al., Attitudinal survey of adverse drug reaction reporting by medical practitioners in the United Kingdom. Br J Clin Pharmacol, 1995. 39(3): p. 223-6. Eland, I.A., et al., Attitudinal survey of voluntary reporting of adverse drug reactions. Br J Clin Pharmacol, 1999. 48(4): p. 623-7. Belton, K.J., Attitude survey of adverse drug-reaction reporting by health care professionals across the European Union. The European Pharmacovigilance Research Group. Eur J Clin Pharmacol, 1997. 52(6): p. 423-7. Mittmann, N., et al., Evaluation of the extent of under-reporting of serious adverse drug reactions: the case of toxic epidermal necrolysis. Drug Saf, 2004. 27(7): p. 477-87. Evans, W.E. and M.V. Relling, Pharmacogenomics: translating functional genomics into rational therapeutics. Science, 1999. 286(5439): p. 487-91. Kalow, W., Pharmacogenetics and pharmacogenomics: origin, status, and the hope for personalized medicine. Pharmacogenomics J, 2006. 6(3): p. 162-5. 97  20.  21. 22. 23.  24. 25. 26. 27. 28.  29. 30. 31. 32. 33. 34.  35. 36. 37. 38.  39.  Mathijssen, R.H., et al., Flat-fixed dosing versus body surface area based dosing of anticancer drugs in adults: does it make a difference? Oncologist, 2007. 12(8): p. 91323. Belle, D.J. and H. Singh, Genetic factors in drug metabolism. Am Fam Physician, 2008. 77(11): p. 1553-60. Nelson, M.R., et al., Genome-wide approaches to identify pharmacogenetic contributions to adverse drug reactions. Pharmacogenomics J, 2009. 9(1): p. 23-33. Carlquist, J.F., et al., Genotypes of the cytochrome p450 isoform, CYP2C9, and the vitamin K epoxide reductase complex subunit 1 conjointly determine stable warfarin dose: a prospective study. J Thromb Thrombolysis, 2006. 22(3): p. 191-7. Kim, M.J., et al., A regulatory science perspective on warfarin therapy: a pharmacogenetic opportunity. J Clin Pharmacol, 2009. 49(2): p. 138-46. Rieder, M.J., et al., Effect of VKORC1 haplotypes on transcriptional regulation and warfarin dose. N Engl J Med, 2005. 352(22): p. 2285-93. Moroney, A. Pharmacodynamics: Merck Manual 2007 [cited March 2, 2012; Available from: http://www.merck.com/mmpe/sec20/ch304/ch304a.html. Kopacek, K.B. Pharmacokinetics: Merck Manual Professional. 2007 [cited March 2, 2012; Available from: http://www.merck.com/mmpe/sec20/ch303/ch303a.html Kim, K.A., et al., The effect of CYP2C19 polymorphism on the pharmacokinetics and pharmacodynamics of clopidogrel: a possible mechanism for clopidogrel resistance. Clin Pharmacol Ther, 2008. 84(2): p. 236-42. Wang, L., H.L. McLeod, and R.M. Weinshilboum, Genomics and drug response. N Engl J Med, 2011. 364(12): p. 1144-53. Chapman, M.A., et al., Initial genome sequencing and analysis of multiple myeloma. Nature. 471(7339): p. 467-72. Johnson, J.A., et al., Pharmacogenomics: application to the management of cardiovascular disease. Clin Pharmacol Ther, 2011. 90(4): p. 519-31. Administration, U.S.F.a.D. Table of Pharmacogenomic Biomarkers in Drug Labels. 2012 [cited March 5, 2012. Kauffman, R.E., Clinical trials in children: problems and pitfalls. Paediatr Drugs, 2000. 2(6): p. 411-8. Leeder, J.S., et al., Understanding the relative roles of pharmacogenetics and ontogeny in pediatric drug development and regulatory science. J Clin Pharmacol. 50(12): p. 137787. Kearns, G.L., et al., Developmental pharmacology--drug disposition, action, and therapy in infants and children. N Engl J Med, 2003. 349(12): p. 1157-67. Helm, C.W. and J.C. States, Enhancing the efficacy of cisplatin in ovarian cancer treatment - could arsenic have a role. J Ovarian Res, 2009. 2: p. 2. Hohnloser, J.H., et al., Cisplatin based chemotherapy in testicular cancer patients: long term platinum excretion and clinical effects. Eur J Med Res, 1996. 1(11): p. 509-14. Lanvers-Kaminsky, C., et al., Continuous or repeated prolonged cisplatin infusions in children: a prospective study on ototoxicity, platinum concentrations, and standard serum parameters. Pediatr Blood Cancer, 2006. 47(2): p. 183-93. Perilongo, G., et al., Cisplatin versus cisplatin plus doxorubicin for standard-risk hepatoblastoma. N Engl J Med, 2009. 361(17): p. 1662-70.  98  40.  41.  42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53.  54.  55. 56.  57. 58. 59.  Bosl GJ, B.D., Sheinfeld J, Cancer of the Testis. Cancer: Principles and Practice of Oncology, ed. V.T.J. DeVita, Hellman, S. & Rosenberg, S. A. 2005, Philadelphia, PA: Lippincott Williams & Wilkins. Afzal, S., et al., Challenges in management of patients with intracranial germ cell tumor and diabetes insipidus treated with cisplatin and/or ifosfamide based chemotherapy. J Neurooncol, 2009. 97(3): p. 393-9. Ozdogan, O., et al., Does cisplatin chemotherapy decrease the MDP uptake of normal bone? An experimental study. Ann Nucl Med, 2008. 22(5): p. 357-62. Jamieson, E.R. and S.J. Lippard, Structure, Recognition, and Processing of CisplatinDNA Adducts. Chem Rev, 1999. 99(9): p. 2467-98. Siddik, Z.H., Cisplatin: mode of cytotoxic action and molecular basis of resistance. Oncogene, 2003. 22(47): p. 7265-79. Rybak, L.P. and V. Ramkumar, Ototoxicity. Kidney Int, 2007. 72(8): p. 931-5. Santoso, J.T., et al., Saline, mannitol, and furosemide hydration in acute cisplatin nephrotoxicity: a randomized trial. Cancer Chemother Pharmacol, 2003. 52(1): p. 13-8. Mukherjea, D. and L.P. Rybak, Pharmacogenomics of cisplatin-induced ototoxicity. Pharmacogenomics. 12(7): p. 1039-50. Li, Y., R.B. Womer, and J.H. Silber, Predicting cisplatin ototoxicity in children: the influence of age and the cumulative dose. Eur J Cancer, 2004. 40(16): p. 2445-51. Brock, P.R., et al., Cisplatin ototoxicity in children: a practical grading system. Med Pediatr Oncol, 1991. 19(4): p. 295-300. Blakley, B.W. and S.F. Myers, Patterns of hearing loss resulting from cis-platinum therapy. Otolaryngol Head Neck Surg, 1993. 109(3 Pt 1): p. 385-91. Ilveskoski, I., et al., Ototoxicity in children with malignant brain tumors treated with the "8 in 1" chemotherapy protocol. Med Pediatr Oncol, 1996. 27(1): p. 26-31. Skinner, R., et al., Ototoxicity of cisplatinum in children and adolescents. Br J Cancer, 1990. 61(6): p. 927-31. Montaguti, M., et al., [Cisplatin and carboplatin-induced ototoxicity in children: clinical aspects and perspectives for prevention]. Acta Otorhinolaryngol Ital, 2002. 22(1): p. 148. Bertolini, P., et al., Platinum compound-related ototoxicity in children: long-term followup reveals continuous worsening of hearing loss. J Pediatr Hematol Oncol, 2004. 26(10): p. 649-55. Chang, K.W. and N. Chinosornvatana, Practical grading system for evaluating cisplatin ototoxicity in children. J Clin Oncol, 2010. 28(10): p. 1788-95. Mann, J.R., et al., The United Kingdom Children's Cancer Study Group's second germ cell tumor study: carboplatin, etoposide, and bleomycin are effective treatment for children with malignant extracranial germ cell tumors, with acceptable toxicity. J Clin Oncol, 2000. 18(22): p. 3809-18. Dionne, F., et al., Economic impact of a genetic test for cisplatin-induced ototoxicity. Pharmacogenomics J, 2011. Ekborn, A., et al., Cisplatin-induced hearing loss: influence of the mode of drug administration in the guinea pig. Hear Res, 2000. 140(1-2): p. 38-44. Lewis, M.J., et al., Ototoxicity in children treated for osteosarcoma. Pediatr Blood Cancer, 2009. 52(3): p. 387-91.  99  60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70.  71. 72.  73. 74. 75. 76.  77. 78.  79. 80.  Grewal, S., et al., Auditory late effects of childhood cancer therapy: a report from the Children's Oncology Group. Pediatrics, 2010. 125(4): p. e938-50. Cancer Therapy Evalution Program-Common Terminology Criteria for Adverse EventsVersion 3. 2003. Schmidt, C.M., et al., [The "Muenster classification" of high frequency hearing loss following cisplatin chemotherapy]. HNO, 2007. 55(4): p. 299-306. Schell, M.J., et al., Hearing loss in children and young adults receiving cisplatin with or without prior cranial irradiation. J Clin Oncol, 1989. 7(6): p. 754-60. Ross, C.J., et al., Genetic variants in TPMT and COMT are associated with hearing loss in children receiving cisplatin chemotherapy. Nature genetics, 2009. 41(12): p. 1345-9. Blair J.C., P.M.E., Viehweg S.H., The effects of mild sensorineural hearing loss on academic performance of young school age children. Volta Rev, 2001. 87: p. 87-93. Kim, H.J., et al., Roles of NADPH oxidases in cisplatin-induced reactive oxygen species generation and ototoxicity. J Neurosci, 2010. 30(11): p. 3933-46. Vlajkovic, S.M., G.D. Housley, and P.R. Thorne, Adenosine and the auditory system. Curr Neuropharmacol, 2009. 7(3): p. 246-56. Rybak, L.P., et al., Cisplatin ototoxicity and protection: clinical and experimental studies. Tohoku J Exp Med, 2009. 219(3): p. 177-86. Rybak, L.P., et al., Mechanisms of cisplatin-induced ototoxicity and prevention. Hear Res, 2007. 226(1-2): p. 157-67. Watanabe, K., et al., Expression of caspase-activated deoxyribonuclease (CAD) and caspase 3 (CPP32) in the cochlea of cisplatin (CDDP)-treated guinea pigs. Auris Nasus Larynx, 2003. 30(3): p. 219-25. Ravi, R., S.M. Somani, and L.P. Rybak, Mechanism of cisplatin ototoxicity: antioxidant system. Pharmacol Toxicol, 1995. 76(6): p. 386-94. van Ruijven, M.W., J.C. de Groot, and G.F. Smoorenburg, Time sequence of degeneration pattern in the guinea pig cochlea during cisplatin administration. A quantitative histological study. Hear Res, 2004. 197(1-2): p. 44-54. Bokemeyer, C., et al., Analysis of risk factors for cisplatin-induced ototoxicity in patients with testicular cancer. Br J Cancer, 1998. 77(8): p. 1355-62. Bhandare, N., et al., Ototoxicity after radiotherapy for head and neck tumors. Int J Radiat Oncol Biol Phys, 2007. 67(2): p. 469-79. Kushner, B.H., et al., Ototoxicity from high-dose use of platinum compounds in patients with neuroblastoma. Cancer, 2006. 107(2): p. 417-22. Veenstra, D.L., et al., Pharmacogenomic testing to prevent aminoglycoside-induced hearing loss in cystic fibrosis patients: potential impact on clinical, patient, and economic outcomes. Genet Med, 2007. 9(10): p. 695-704. Waguespack, J.R. and A.J. Ricci, Aminoglycoside ototoxicity: permeant drugs cause permanent hair cell loss. J Physiol, 2005. 567(Pt 2): p. 359-60. Packer, R.J., et al., Treatment of children with medulloblastomas with reduced-dose craniospinal radiation therapy and adjuvant chemotherapy: A Children's Cancer Group Study. J Clin Oncol, 1999. 17(7): p. 2127-36. Pan, C.C., et al., Prospective study of inner ear radiation dose and hearing loss in headand-neck cancer patients. Int J Radiat Oncol Biol Phys, 2005. 61(5): p. 1393-402. Ceckova, M., et al., Effect of ABCG2 on cytotoxicity of platinum drugs: interference of EGFP. Toxicol In Vitro, 2008. 22(8): p. 1846-52. 100  81.  82.  83.  84.  85.  86.  87.  88. 89. 90. 91.  92.  93. 94.  95.  96.  Cui, Y., et al., Drug resistance and ATP-dependent conjugate transport mediated by the apical multidrug resistance protein, MRP2, permanently expressed in human and canine cells. Mol Pharmacol, 1999. 55(5): p. 929-37. Liedert, B., et al., Overexpression of cMOAT (MRP2/ABCC2) is associated with decreased formation of platinum-DNA adducts and decreased G2-arrest in melanoma cells resistant to cisplatin. J Invest Dermatol, 2003. 121(1): p. 172-6. Alqawi, O., S. Bates, and E. Georges, Arginine482 to threonine mutation in the breast cancer resistance protein ABCG2 inhibits rhodamine 123 transport while increasing binding. Biochem J, 2004. 382(Pt 2): p. 711-6. Boyer, J., et al., Characterization of p53 wild-type and null isogenic colorectal cancer cell lines resistant to 5-fluorouracil, oxaliplatin, and irinotecan. Clin Cancer Res, 2004. 10(6): p. 2158-67. Samimi, G., et al., Increased expression of the copper efflux transporter ATP7A mediates resistance to cisplatin, carboplatin, and oxaliplatin in ovarian cancer cells. Clin Cancer Res, 2004. 10(14): p. 4661-9. Ishida, S., et al., Uptake of the anticancer drug cisplatin mediated by the copper transporter Ctr1 in yeast and mammals. Proc Natl Acad Sci U S A, 2002. 99(22): p. 14298-302. Thomas, J.P., et al., High accumulation of platinum-DNA adducts in strial marginal cells of the cochlea is an early event in cisplatin but not carboplatin ototoxicity. Mol Pharmacol, 2006. 70(1): p. 23-9. Mizuta, K., et al., Ultrastructural localization of megalin in the rat cochlear duct. Hear Res, 1999. 129(1-2): p. 83-91. Schmitz, C., et al., Megalin deficiency offers protection from renal aminoglycoside accumulation. J Biol Chem, 2002. 277(1): p. 618-22. Riedemann, L., et al., Megalin genetic polymorphisms and individual sensitivity to the ototoxic effect of cisplatin. Pharmacogenomics J, 2008. 8(1): p. 23-8. Knight, K.R., D.F. Kraemer, and E.A. Neuwelt, Ototoxicity in children receiving platinum chemotherapy: underestimating a commonly occurring toxicity that may influence academic and social development. J Clin Oncol, 2005. 23(34): p. 8588-96. Weissenstein, A., et al., Progressive hearing loss after completion of cisplatin chemotherapy is common and more pronounced in children without spontaneous otoacoustic emissions before chemotherapy. Int J Pediatr Otorhinolaryngol, 2012. 76(1): p. 131-6. Yang, P., et al., Role of the glutathione metabolic pathway in lung cancer treatment and prognosis: a review. J Clin Oncol, 2006. 24(11): p. 1761-9. Leitner, H.M., R. Kachadourian, and B.J. Day, Harnessing drug resistance: using ABC transporter proteins to target cancer cells. Biochem Pharmacol, 2007. 74(12): p. 167785. el Barbary, A., R.A. Altschuler, and J. Schacht, Glutathione S-transferases in the organ of Corti of the rat: enzymatic activity, subunit composition and immunohistochemical localization. Hear Res, 1993. 71(1-2): p. 80-90. Peters, U., et al., Glutathione S-transferase genetic polymorphisms and individual sensitivity to the ototoxic effect of cisplatin. Anticancer Drugs, 2000. 11(8): p. 639-43.  101  97.  98.  99.  100.  101. 102. 103. 104. 105. 106.  107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117.  Oldenburg, J., et al., Cisplatin-induced long-term hearing impairment is associated with specific glutathione s-transferase genotypes in testicular cancer survivors. J Clin Oncol, 2007. 25(6): p. 708-14. Oldenburg, J., et al., Association between long-term neuro-toxicities in testicular cancer survivors and polymorphisms in glutathione-s-transferase-P1 and -M1, a retrospective cross sectional study. J Transl Med, 2007. 5: p. 70. Khrunin, A.V., et al., Genetic polymorphisms and the efficacy and toxicity of cisplatinbased chemotherapy in ovarian cancer patients. Pharmacogenomics J, 2010. 10(1): p. 54-61. Caronia, D., et al., Common variations in ERCC2 are associated with response to cisplatin chemotherapy and clinical outcome in osteosarcoma patients. Pharmacogenomics J, 2009. 9(5): p. 347-53. Zhu, Y., et al., Modulation of DNA damage/DNA repair capacity by XPC polymorphisms. DNA Repair (Amst), 2008. 7(2): p. 141-8. Khan, S.G., et al., A new xeroderma pigmentosum group C poly(AT) insertion/deletion polymorphism. Carcinogenesis, 2000. 21(10): p. 1821-5. Kraft, P., et al., Beyond odds ratios--communicating disease risk based on genetic profiles. Nat Rev Genet, 2009. 10(4): p. 264-9. Janssens, A.C., et al., Predictive testing for complex diseases using multiple genes: fact or fiction? Genet Med, 2006. 8(7): p. 395-400. Metz, C.E., Basic principles of ROC analysis. Semin Nucl Med, 1978. 8(4): p. 283-98. Lu, Q. and R.C. Elston, Using the optimal receiver operating characteristic curve to design a predictive genetic test, exemplified with type 2 diabetes. Am J Hum Genet, 2008. 82(3): p. 641-51. Wilson, P.W., et al., Prediction of coronary heart disease using risk factor categories. Circulation, 1998. 97(18): p. 1837-47. Gurwitz, D. and A.G. Motulsky, 'Drug reactions, enzymes, and biochemical genetics': 50 years later. Pharmacogenomics, 2007. 8(11): p. 1479-84. Johnson, J.A., Ethnic differences in cardiovascular drug response: potential contribution of pharmacogenetics. Circulation, 2008. 118(13): p. 1383-93. Chung, W.H., et al., Medical genetics: a marker for Stevens-Johnson syndrome. Nature, 2004. 428(6982): p. 486. Freedman, M.L., et al., Assessing the impact of population stratification on genetic association studies. Nat Genet, 2004. 36(4): p. 388-93. Visscher, H., et al., Application of principal component analysis to pharmacogenomic studies in Canada. Pharmacogenomics J, 2009. 9(6): p. 362-72. Suarez-Kurtz, G., et al., Self-reported skin color, genomic ancestry and the distribution of GST polymorphisms. Pharmacogenet Genomics, 2007. 17(9): p. 765-71. Hirschhorn, J.N. and D. Altshuler, Once and again-issues surrounding replication in genetic association studies. J Clin Endocrinol Metab, 2002. 87(10): p. 4438-41. Moonesinghe, R., M.J. Khoury, and A.C. Janssens, Most published research findings are false-but a little replication goes a long way. PLoS Med, 2007. 4(2): p. e28. Marsh, S., et al., Platinum pathway. Pharmacogenet Genomics, 2009. 19(7): p. 563-4. Kling, J., US FDA contemplates collection of pharmacogenomic data. Nat Biotechnol, 2003. 21(6): p. 590.  102  118.  119. 120.  121. 122. 123.  124. 125. 126.  127.  128. 129.  130. 131. 132. 133. 134.  135. 136.  Ross, C.J., et al., Genotypic approaches to therapy in children: a national active surveillance network (GATC) to study the pharmacogenomics of severe adverse drug reactions in children. Ann N Y Acad Sci, 2007. 1110: p. 177-92. Guo, S.W. and E.A. Thompson, Performing the exact test of Hardy-Weinberg proportion for multiple alleles. Biometrics, 1992. 48(2): p. 361-72. DeLong, E.R., D.M. DeLong, and D.L. Clarke-Pearson, Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics, 1988. 44(3): p. 837-45. Robin, X., et al., pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics, 2011. 12: p. 77-85. Weinshilboum, R.M., Pharmacogenomics: catechol O-methyltransferase to thiopurine Smethyltransferase. Cell Mol Neurobiol, 2006. 26(4-6): p. 539-61. Weinshilboum, R.M. and S.L. Sladek, Mercaptopurine pharmacogenetics: monogenic inheritance of erythrocyte thiopurine methyltransferase activity. Am J Hum Genet, 1980. 32(5): p. 651-62. Ochoa, B., et al., S-Adenosyl-L-methionine increases serum BUN and creatinine in cisplatin-treated mice. Arch Med Res, 2009. 40(1): p. 54-8. A haplotype map of the human genome. Nature, 2005. 437(7063): p. 1299-320. Ballatori, N., et al., Molecular mechanisms of reduced glutathione transport: role of the MRP/CFTR/ABCC and OATP/SLC21A families of membrane proteins. Toxicol Appl Pharmacol, 2005. 204(3): p. 238-55. Lee, Y.M., et al., Identification and functional characterization of the natural variant MRP3-Arg1297His of human multidrug resistance protein 3 (MRP3/ABCC3). Pharmacogenetics, 2004. 14(4): p. 213-23. Schrenk, D., et al., Up-regulation of transporters of the MRP family by drugs and toxins. Toxicol Lett, 2001. 120(1-3): p. 51-7. Young, L.C., et al., Multidrug resistance proteins MRP3, MRP1, and MRP2 in lung cancer: correlation of protein levels with drug response and messenger RNA levels. Clin Cancer Res, 2001. 7(6): p. 1798-804. Young, L.C., et al., Expression of multidrug resistance protein-related genes in lung cancer: correlation with drug response. Clin Cancer Res, 1999. 5(3): p. 673-80. Oguri, T., et al., Association between expression of the MRP3 gene and exposure to platinum drugs in lung cancer. Int J Cancer, 2001. 93(4): p. 584-9. Chanock, S.J., et al., Replicating genotype-phenotype associations. Nature, 2007. 447(7145): p. 655-60. Ioannidis, J.P., et al., Replication validity of genetic association studies. Nat Genet, 2001. 29(3): p. 306-9. Bess, F.H., J. Dodd-Murphy, and R.A. Parker, Children with minimal sensorineural hearing loss: prevalence, educational performance, and functional status. Ear Hear, 1998. 19(5): p. 339-54. Orgel, E., et al., Hearing loss among survivors of childhood brain tumors treated with an irradiation-sparing approach. Pediatr Blood Cancer, 2011. Muldoon, L.L., et al., Delayed administration of sodium thiosulfate in animal models reduces platinum ototoxicity without reduction of antitumor activity. Clin Cancer Res, 2000. 6(1): p. 309-15.  103  137. 138.  Price, A.L., et al., Principal components analysis corrects for stratification in genomewide association studies. Nature genetics, 2006. 38(8): p. 904-9. Magne, N., et al., Impact of platinum complexes on dihydropyrimidine dehydrogenase activity in 5-fluorouracil-treated patients. Ann Oncol, 2003. 14(8): p. 1329-30.  104  

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:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0103451/manifest

Comment

Related Items