Identification of additional DPYD polymorphisms that increase the risk of severe fluoropyrimidine toxicity and improved predictive accuracy when combined with previously validated variants.

Abstract

Fluoropyrimidine (FP) chemotherapy can cause life-threatening toxicity. Four DPYD polymorphisms (DPYD*2A, *13, p.Asp949Val, HapB3) are well-established to increase FP toxicity risk. This study aimed to identify additional DPYD polymorphisms that increase FP toxicity. Adult patients treated with standard doses of systemic FP (5-fluorouracil/capecitabine) for any cancer with available DPYD genetic data were included. The primary endpoint was a composite of CTCAE grade ≥3 toxicity or treatment modification due to toxicity in the first two FP cycles. A literature-curated list of suspected deleterious unvalidated DPYD variants was classified as uncommon (minor allele frequency <0.01) or common. The genetic association with toxicity was analyzed via multivariable logistic regression. Among 849 eligible patients, the composite toxicity endpoint occurred in 25%. Genetic data were available for five uncommon and six common suspected deleterious DPYD variants. In the primary analysis of 799 patients who did not carry a validated variant, carriers of uncommon deleterious variants (1.1% of patients) had significantly higher toxicity risk than non-carriers (67% vs. 24%; adjusted OR 7.36; 95% CI 1.75-38.20; p = 0.009). None of the common deleterious variants were associated with toxicity. Toxicity prediction in the entire cohort (n = 849) was slightly improved by testing the uncommon and validated variants vs. testing only the validated variants (positive predictive value: 44.1% vs. 40.0%). Five uncommon DPYD variants, in combination, increase FP toxicity risk and improve risk prediction. Testing these variants could identify more high-risk patients who should receive adjusted FP doses to prevent severe toxicity.