A candidate gene study of capecitabine-related toxicity in colorectal cancer identifies new toxicity variants at DPYD and a putative role for ENOSF1 rather than TYMS

OBJECTIVE: Capecitabine is an oral 5-fluorouracil (5-FU) pro-drug commonly used to treat colorectal carcinoma and other tumours. About 35% of patients experience dose-limiting toxicity. The few proven genetic biomarkers of 5-FU toxicity are rare variants and polymorphisms, respectively, at candidate loci dihydropyrimidine dehydrogenase (DPYD) and thymidylate synthase (TYMS).

DESIGN: We investigated 1456 polymorphisms and rare coding variants near 25 candidate 5-FU pathway genes in 968 UK patients from the QUASAR2 clinical trial.

RESULTS: We identified the first common DPYD polymorphisms to be consistently associated with capecitabine toxicity, rs12132152 (toxicity allele frequency (TAF)=0.031, OR=3.83, p=4.31×10^-6) and rs12022243 (TAF=0.196, OR=1.69, p=2.55×10^-5). rs12132152 was particularly strongly associated with hand-foot syndrome (OR=6.1, p=3.6×10^-8). The rs12132152 and rs12022243 associations were independent of each other and of previously reported DPYD toxicity variants. Next-generation sequencing additionally identified rare DPYD variant p.Ala551Thr in one patient with severe toxicity. Using functional predictions and published data, we assigned p.Ala551Thr as causal for toxicity. We found that polymorphism rs2612091, which lies within an intron of ENOSF1, was also associated with capecitabine toxicity (TAF=0.532, OR=1.59, p=5.28×10^6). ENSOF1 is adjacent to TYMS and there is a poorly characterised regulatory interaction between the two genes/proteins. Unexpectedly, rs2612091 fully explained the previously reported associations between capecitabine toxicity and the supposedly functional TYMS variants, 5'VNTR 2R/3R and 3'UTR 6 bp ins-del. rs2612091 genotypes were, moreover, consistently associated with ENOSF1 mRNA levels, but not with TYMS expression.

CONCLUSIONS: DPYD harbours rare and common capecitabine toxicity variants. The toxicity polymorphism in the TYMS region may actually act through ENOSF1.