Phase I study design of indisulam: evaluation and optimization
Anthe S. Zandvliet (1), Jan H.M. Schellens (2,3), William Copalu (4), Jantien Wanders (4), Mats O. Karlsson (5), Jos H. Beijnen (1,2), Alwin D.R. Huitema (1)
1 Department of Pharmacy & Pharmacology, The Netherlands Cancer Institute/Slotervaart Hospital, Amsterdam, The Netherlands, 2 Department of Biomedical Analysis, Section of Drug Toxicology, Utrecht University, Utrecht, The Netherlands, 3 Department of Clinical Pharmacology, Division of Medical Oncology, The Netherlands Cancer Institute/Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands, 4 Eisai Ltd., London, UK and 5 Department of Pharmaceutical Biosciences, Division of Pharmacokinetics and Drug Therapy, Uppsala University
Objectives: Phase I dose escalation studies were previously performed to define a safe dose of the anticancer drug indisulam. Multiple dosing regimens were evaluated in four parallel phase I studies. The dose limiting toxicity was neutropenia. We hypothesized that PK-PD data from an initial phase I study can be used to determine an optimal starting dose for subsequent studies. The aims of this analysis were 1) to retrospectively evaluate the phase I study design of indisulam, 2) to optimize this design and 3) to verify the advantages of the optimized design.
Methods: PK-PD models were developed for each phase I study separately. These models were subsequently used for trial simulation of the three other phase I studies to predict the recommended dose for each administration regimen. NONMEM V was used throughout the analysis. We took into account interpatient variability (random effects model) and parameter uncertainty (covariance matrix). The 5th percentile of the simulated recommended dose was considered as a safe starting dose for a clinical dose escalation study. We investigated the advantages of the optimized design regarding the total number of patients and the number of patients that was treated at a subtherapeutic dose level.
Results: For each phase I study, the structural pharmacokinetic model comprised 3 compartments, with saturable distribution and elimination. Indisulam-induced myelosuppression was described by a semi-physiological model.[2,3] The neutropenic effect could be better assessed for 3-weekly dosing regimens than for a weekly regimen. Consequently, simulations from the weekly regimen were the least predictive. Predicted starting doses were between 25 and 104% of the clinically determined recommended dose and could therefore all be considered safe. The optimized study design required fewer patients for study completion (-17 to -29%). Using the optimized design, 21-31% of the patients could have been treated with a higher indisulam dose.
Conclusions: This study demonstrated that PK-PD modeling and simulation may assist in the optimization of phase I dose escalation studies of cytotoxic anticancer agents. Initial evaluation of one regimen, followed by model development, trial simulation and clinical evaluation of the other three regimens could have reduced the number of patients required for the phase I clinical program of indisulam. This strategy would also minimize treatment at subtherapeutic dose levels.
 Van Kesteren C et al. J Clin Oncol 2002; 20: 4065-73
 Friberg LE et al. J Clin Oncol 2002; 20: 4713-21
 Van Kesteren C et al. Invest New Drugs 2005; 23: 225-34