Abstract
PURPOSE: Standard dosing of infusional 5-FU results in subtherapeutic drug exposure for up to half of all patients. We evaluated plasma uracil concentration and DPYD rs4294451 genotype as candidate predictors of individual-level 5-FU exposure. METHODS: We conducted a prospective study of drug exposure in patients with gastrointestinal cancer receiving infusional 5-FU. Participants were evaluated by measurement of fasting pretreatment plasma uracil concentration, DPYD gene sequence (including genotyping of rs4294451) and 5-FU area under the curve (5-FU AUC). We used a linear mixed-effects model to evaluate the association of 5-FU AUC with uracil concentration and rs4294451 genotype, adjusting for cycle number, sex, serum creatinine, and 5-FU dose. RESULTS: There were 29 evaluable participants with a median age of 64 years (range 41-81); nine (31%) were female. The median plasma uracil concentration was 10.4 ng/mL (IQR 7.2, 12.6). Nine participants carried the DPYD rs4294451 T-allele (7 with T/A, 2 with T/T.) Among all participants the median 5-FU AUC was 22.0 mg*h/L in cycle 1 and 19.3 mg*h/L in cycle 2. In the mixed-effects model, higher rs4294451 T-allele count (0, 1, or 2) was significantly associated with lower 5-FU AUC (-4.0 mg*h/L per T-allele [95% CI -8.0, 0.0], p = 0.049), as was male sex (-7.5 mg*h/L [95% CI -13.8, -1.3], p = 0.021). Pretreatment plasma uracil concentration was not significantly associated with 5-FU AUC (p = 0.57). The subject with the highest uracil concentration (23.1 ng/mL) had a rare DPYD missense variant (c.2185G > A [p.A729T]) and experienced early 5-FU-related toxicity. CONCLUSIONS: DPYD rs4294451 T-allele count and male sex were significantly associated with reduced 5-FU drug exposure. DPYD rs4294451 and male sex merit further evaluation as candidate biomarkers to inform initial dosing of infusional 5-FU.