Administration of troxacitabine by continuous infusion achieved the prospectively defined target plasma concentration. Pharmacokinetics (PK) modeling coupled with real-time PK assessment was an efficient approach to conduct hypothesis-driven phase I trials.

Purpose Troxacitabine (TROX) is a L‐cytidine analogue anticancer agent currently in phase II/III trials. The study's objective is to develop & validate a population pharmacokinetic (PPK) model for TROX. Methods Plasma samples from 111 cancer patients receiving IV doses of 0.12 – 12.5 mg/m2 were used to develop the PPK model with NONMEM. About 13 samples per patient were obtained from the 1st dose. 2 covariate groups (I: BSA, SEX, AGE, SCR; II: WT, HT, SEX, AGE, SCR) & PK parameters were evaluated by linear multiple regression. The 2 final PPK models were validated by internal & external methods. Results TROX PPK was characterized by a 3‐compartment model, exponential interpatient variability (IPV) error model, combination residual error model, & FOCE INTER estimator method. Clearance was influenced by BSA (27% decrease IPV) or WT (20% decrease IPV). Central compartmental volume was influenced by BSA (12% decrease IPV). Model validations reveal both final models accurate in predicting plasma TROX concentrations with improved PK parameter predictions with the addition of covariates. Conclusion Covariate modeling supports the use of BSA in current dosing strategies for TROX. Clinical Pharmacology & Therapeutics (2004) 75, P82–P82; doi: