Abstract 1768P
Background
5-FU and its oral pro-drug capecitabine are widely used chemotherapy but associated toxicities are common and potentially fatal. The current strategy of DPYD testing is limited by low sensitivity, possibly due to causal variants lying outside the fluoropyrimidine pathway. We performed a genome wide association study (GWAS) to identify novel genetic markers that can predict serious early capecitabine toxicity.
Methods
Genotyping and toxicity data for 930 colorectal cancer patients treated with adjuvant capecitabine in the QUASAR2 trial were analysed. Per cycle Common Terminology Criteria for Adverse Events (CTCAE) graded toxicity data was used to identify cases with ≥grade 3 toxicities in the first two treatment cycles. Binary coded variables were generated for: diarrhoea, haematological, hand-foot syndrome, mucositis, vomiting and global toxicity. Association analysis was performed using SNPTEST. Both trial arms were analysed separately and then meta-analysed using META. The study was powered at 80% to detect variants with minor allele frequency (MAF)>10% having effect sizes ≥2. Outputs were filtered according to: INFO>=0.8, Hardy-Weinberg equilibrium<1x10-6, MAF>1%, presence in both arms, GWAS significance(P<5X10-8) and P heterogeneity>0.05. Annotation of the outputs were performed using Variant Effect Predictor. Top independent variants were identified from conditional and joint association analysis (COJO).
Results
Whilst the number of cases experiencing early ≥ grade 3 toxicity for each individual toxicity variable was small (0.9-8.3%), 15.3% experienced one or more ≥ grade 3 event. 14 independent variants (MAF>5%) were associated with early toxicity (P<5X10-8). In addition, two rare variants (MAF<2%) were associated with global toxicity at genome-wide significance with large effect sizes. Two DPYD variants (c.2846A>T, c.1129–5923C>G) from current tests were significantly associated with early global toxicity (P=7.33x10-5, 0.0005 respectively).
Conclusions
Our GWAS found novel rare variants with large effect sizes associated with early capecitabine toxicity. These variants require further validation and could contribute to future genetic panels to predict risk of capecitabine toxicity.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Cancer Research UK Birmingham Centre.
Disclosure
I.S. Chin: Other, Institutional, Research Grant: Cancer Research UK Birmingham Centre. D. Church: Other, Personal, Advisory Board, Unrelated to the abstract: MSD. D.J. Kerr: Financial Interests, Personal, Funding, Unrelated to abstract: Oxford Cancer Biomarkers; Non-Financial Interests, Personal, Other, Unrelated to abstract: My-BioMed Biotechnology Co., Ltd., China; Other, Personal and Institutional, Research Grant, Unrelated to abstract: Bayer. C. Palles: Other, Personal, Advisory Role, Unrelated to the abstract: Oxford Cancer Biomarkers. All other authors have declared no conflicts of interest.