Abstract

Purpose: The anticancer drug 5-fluorouracile (5-FU) which is indicated for the treatment of a variety of solid malignancies such as colorectal, breast, head and neck neoplasms is extensively biotransformed to 5fluoro-5,6- dihydrouracil (5-FDHU) by the dihydropyrimidine deshydrogenase enzyme (DPD). DPD deficiency is recognized as an important risk factor, predisposing patient to undergo severe/lethal toxicities. To date, relationships between 5-FU, 5- FDHU and toxicity following i.v. bolus administration has not been studied using the population pharmacokinetics approach.

Methods: Retrospective pharmacokinetic data of 5-FU and 5-FDHU from 127 colorectal cancer patients were used for the population pharmacokinetic analysis. Treatment schedule consisted of an adjuvant therapy with 5-FU plus leucovorin. 5- FU and 5-FDHU complete plasma profiles recorded on day-1 of the first chemotherapy cycle were modeled simultaneously using NONMEM software. Gastro-intestinal adverse events graded according to the WHO criteria were recorded after the first cycle. A population logistic regression model was developed to identify predictive factors of these adverse events.

Results: A three-compartment pharmacokinetic mixture model best described 5-FU and 5-FDHU kinetics profiles. Linear and saturated elimination from the central compartment of 5-FU and a linear elimination from the 5-FDHU compartment were used. A bimodal distribution of the inter-compartmental clearance was observed allowing two subpopulation with high (17 L/h) and low values (3.35 L/h). DPD-phenotype is suspected to explain this mixture. No covariates were introduced in the final model. Also, no relationship was found between maximal metabolism rate and DPD-phenotype. Predictive factors associated with occurrence of high grade gastro-intestinal adverse events were gender, dose and lean body mass suggesting serious cautions with the BSA-weighted dose for women. For the low-grade toxicities, 5-FU area under curve was predictive for woman and 5-FDHU area under curve for men.

Conclusion: A population pharmacokinetic mixture model was developed to describe kinetic profiles of 5-FU and its major metabolite. This model has significant implications, to identify patients with potentially low DPD phenotype requiring earlier adjustment of the 5-FU dose. Also this analysis highlights the need for developing alternative dosing-scheme for women.

Keywords: Population analysis, modeling, mixture model, logistic regression, 5-FU, 5-FDHU, tolerability