The Pharmacogenomics Journal (2015) 15, 505–512 © 2015 Macmillan Publishers Limited All rights reserved 1470-269X/15 www.nature.com/tpj ORIGINAL ARTICLE Genetic variants in DNA repair genes as potential predictive markers for oxaliplatin chemotherapy in colorectal cancer EJ Kap1, P Seibold1, S Richter1, D Scherer2, N Habermann2, Y Balavarca2, L Jansen3, N Becker4, K Pfütze5,6, O Popanda7, M Hoffmeister3, A Ulrich8, A Benner4, CM Ulrich2,9, B Burwinkel5,6, H Brenner3,10 and J Chang-Claude1 Oxaliplatin-based chemotherapy exerts its effects through generating DNA damage. Hence, genetic variants in DNA repair pathways could modulate treatment response. We used a prospective cohort of 623 colorectal cancer patients with stage II–IV disease treated with adjuvant/palliative chemotherapy to comprehensively investigate 1727 genetic variants in the DNA repair pathways as potential predictive markers for oxaliplatin treatment. Single nucleotide polymorphisms (SNP) associations with overall survival and recurrence-free survival were assessed using a Cox regression model. Pathway analysis was performed using the gamma method. Patients carrying variant alleles of rs3783819 (MNAT1) and rs1043953 (XPC) experienced a longer overall survival after treatment with oxaliplatin than patients who did not carry the variant allele, while the opposite association was found in patients who were not treated with oxaliplatin (false discovery rate-adjusted P-values for heterogeneity 0.0047 and 0.0237, respectively). The nucleotide excision repair (NER) pathway was found to be most likely associated with overall survival in patients who received oxaliplatin (P-value = 0.002). Our data show that genetic variants in the NER pathway are potentially predictive of treatment response to oxaliplatin. The Pharmacogenomics Journal (2015) 15, 505–512; doi:10.1038/tpj.2015.8; published online 17 March 2015 INTRODUCTION oxaliplatin.15 The NER pathway is responsible for removing bulky Colorectal cancer (CRC) is the fourth most common cancer in DNA lesions and involves the incision of multiple nucleotides. western countries, with ~ 65 000 new diagnoses per year and a Genetic variants in multiple DNA repair genes have been inves- 5-year relative survival of 63% in Germany.1,2 The use of oxaliplatin tigated in association with FOLFOX chemotherapy in epidemio- fl 9,10,16–18 in the treatment of stage III and IV CRC has been recommended in logical and in vitro studies, yielding con icting results. the German guideline for CRC treatment since 2004.3 The current However, these studies considered only a small proportion of guideline also recommends the use of adjuvant chemotherapy in genes involved in DNA repair. Therefore, we carried out a compre- high-risk stage II patients. Although the objective response rate of hensive investigation of the DNA repair pathway to identify poten- tial predictive markers for oxaliplatin treatment in a prospective the addition of oxaliplatin to 5-fluorouracil (FOLFOX) is estimated to cohort of CRC patients treated with adjuvant/palliative chemo- increase to ca 50%, not all CRC patients benefitfromoxaliplatin therapy. treatment, and a substantial number of patients experience gastrointestinal, hematological and neurological adverse effects.4–7 Currently, there are no markers used in clinical practice to identify PATIENTS AND METHODS patients who are likely to benefit from oxaliplatin treatment. Study population Oxaliplatin exerts its effects by binding to DNA and causing inter- and intrastrand crosslinks and bulky adducts. These DNA Our study population comprised CRC patients recruited for the DACHS (Darmkrebs: Chancen der Verhütung durch Screening) study,19,20 an modifications will disrupt cellular replication, causing the cell to – 8,9 ongoing population-based case control study with additional long-term undergo apoptosis. Genetic variants in different biological follow-up of CRC patients. Inclusion criteria for the DACHS study are: pathways involved with DNA damage repair, drug transport, resident of the Rhein-Neckar-Odenwald region in Germany, age 30 years or metabolism and cell cycle regulation have been proposed as older at diagnosis, speaking German and the physical and mental ability to predictive markers.10,11 Genetic variants could reduce the func- participate. For a total of 1749 histological confirmed cases diagnosed tional activity of a DNA repair pathway for repairing DNA damage between 2003 and 2007, both genotype and complete follow-up data until 31 December 2010 (for either 3 or 5 years) were available. The study was caused by oxaliplatin. This would lead to an increased number of fi fi approved by the ethics committees of the Medical Faculty of the University DNA modi cations and increased ef cacy of the drug. Six different of Heidelberg and the State Medical Boards of Baden-Wuerttemberg and 12–14 DNA repair pathways exist whereby mainly the nucleotide Rhineland-Palatinate. All patients gave their written informed consent. The excision repair (NER) pathway repairs the DNA damage caused by informed consent does not cover making the individual lifestyle data 1Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany; 2Division of Preventive Oncology, National Center for Tumor Diseases (NCT) and DKFZ, Heidelberg, Germany; 3Division of Clinical Epidemiology and Aging Research, DKFZ, Heidelberg, Germany; 4Division of Biostatistics, DKFZ, Heidelberg, Germany; 5Division of Molecular Epidemiology, DKFZ, Heidelberg, Germany; 6Division Molecular Biology of Breast Cancer, Department of Gynecology and Obstetrics, University of Heidelberg, Heidelberg, Germany; 7Division of Epigenomics and Cancer Risk Factors, DKFZ, Heidelberg, Germany; 8Department of General, Visceral and Transplantation Surgery, University Hospital Heidelberg, University of Heidelberg, Heidelberg, Germany; 9Cancer Prevention Program, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA and 10German Cancer Consortium (DKTK), Heidelberg, Germany. Correspondence: Professor J Chang-Claude, Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 581, Heidelberg D-69120, Germany. E-mail: [email protected] Received 29 September 2014; revised 21 November 2014; accepted 28 January 2015; published online 17 March 2015 DNA repair gene variants as predictive markers EJ Kap et al 506 Figure 1. Inclusion of CRC patients from the DACHS study. publically available. However, genotype data will be deposited in the used for selection of tagging SNPs, with a pairwise tagging approach database of Genotypes and Phenotypes of the NIH (dbGaP, http://www. based on the reference data from the HapMap project (Central Europe ncbi.nlm.nih.gov/gap). (CEU) population, phase II/release 24) using r2 ⩾ 0.8 as cutoff and excluding At baseline, trained interviewers collected sociodemographic, medical SNPs with a minor allele frequency o0.05. Genotyping of the 688 SNPs and lifestyle data via a standardized questionnaire. Furthermore, partici- was performed on a Illumina GoldenGate assay (Illumina, San Diego, CA, pants provided a blood sample (499% of the analyzed patients) or a USA).21 Nine SNPs that failed Illumina GoldenGate genotyping were mouthwash at recruitment. At 3 and 5 years of follow-up, information on genotyped using the iPLEX assay (Sequenom, Hamburg, Germany) for the individual patient therapy was collected from the treating physician. Vital MassArray system.22 status, date and cause of death were collected through the local SNPs with a success rate below 95% or poor quality of the cluster plot population registries. and samples with a call rate below 96% were excluded. Quality assurance Nearly, all the patients received 5-fluorouracil or capecitabine and only a also comprised 156 internal duplicate samples (concordance 499.7%) and minority of the patients received in addition irinotecan, bevacizumab or 74 external control samples from Centre d'Etude du Polymorphisme cetuximab. Of those patients who received oxaliplatin almost 90% received Humain (CEPH) (concordance 498%). SNPs with a minor allele frequency oxaliplatin as first-line treatment. Less than 20% of all patients received o0.01 or a deviation from Hardy-Weinberg equilibrium in additionally second-line treatment. As shown in Figure 1, we excluded patients who did genotyped controls (Po0.001) were also excluded, after which 654 SNPs not receive adjuvant chemotherapy or received both adjuvant and neo- in the DNA repair pathways from the Illumina GoldenGate assay were adjuvant chemotherapy, patients with unknown start date of chemother- available for analysis. apy or who died within 30 days of the start of chemotherapy. We defined Additional genotype data were available from the whole-genome patients receiving oxaliplatin as those who received at least four cycles. For Illumina CytoSNP v12.2.1 assay (4299 000 SNPs, Illumina, San Diego, CA, this analysis, 623 CRC patients (stages II–IV) were available of whom 201 USA) performed in collaboration with the Genetics and Epidemiology of received oxaliplatin. Colorectal Cancer Consortium (GECCO).23 Quality control for GECCO included exclusion based on call rate (o98%), lack of Hardy-Weinberg Genotyping equilibrium in controls and low minor allele frequency. Using the aforementioned criteria, 1073 additional SNPs from 109 DNA repair genes DNA was extracted from Ethylenediaminetetraacetic acid (EDTA) blood as well as 52 additional genes were selected. Supplementary Table S1 and mouthwash samples using the