Introduction: The dihydropyrimidine dehydrogenase gene (DPYD)*9A (c.85T>C) genotype is relatively common. The correlation between DPYD*9A genotype and dihydropyrimidine dehydrogenase (DPD) deficiency phenotype is controversial. In a cohort of 28 patients, DPYD*9A was the most commonly diagnosed variant (13 patients [46%]) and there was a noticeable genotype-phenotype correlation. In this study we genotyped a larger cohort of a mixed racial background to explore the prevalence of DPYD*9A variant and to confirm the genotype-phenotype correlation. Patients and Methods: Between 2011 and 2018, in addition to genotyping for high-risk DPYD variants (DPYD*2A, DPYD*13 and DPYD*9B), genotyping for DPYD*9A variant was performed on 113 patients with gastrointestinal malignancies treated with fluoropyrimidines. Fluoropyrimidines-associated toxicity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 5.0). Fisher exact test was used for statistical analysis. Results: Heterozygous and homozygous DPYD*9A genotypes were identified in 46 (41%) and 11 (10%) patients, respectively. Among patients with DPYD*9A genotypes (n = 57), men and women represented 30 (53%) and 27 (47%) patients, respectively. Caucasian, African American, and other ethnicities represented 29 (50.9%), 26 (45.6%), and 2 (3.5%) patients, respectively. Grade 3/4 toxicities were experienced in 26 patients with DPYD*9A genotype (3 patients had homozygous status) and in 20 patients with wild type DPYD*9A (P =.4405). In patients who received full-dose fluoropyrimidines (n = 85), Grade 3/4 toxicities were experienced in 22 patients with DPYD*9A genotype (2 patients had homozygous status), and in 17 patients with wild type DPYD (P =.8275). Conclusion: In our updated analysis, the prevalence of heterozygous and homozygous DPYD*9A genotypes were 41% and 10%, respectively. The correlation between DPYD*9A genotype and DPD clinical phenotype was not reproduced. The noticeable correlation that we previously reported is likely because of small sample size and selection bias.
