Abstract
Thiotepa is an alkylating agent widely used in high-dose chemotherapy. The pharmacokinetics of thiotepa and its main metabolite tepa show a wide interpatient variability, which may be responsible for the interpatient variability in toxicity. The aim of this study was to develop and validate a pharmacokinetically guided dosing strategy with the sum of the thiotepa and tepa area under the concentration-time curve (AUC) as the target parameter. A total of 46 patients received 77 courses of chemotherapy with thiotepa (80-120 mg/m(2) per day) divided into two daily 30-minute infusions in combination with cyclophosphamide and carboplatin. Patients received up to three courses of chemotherapy. The interpatient, course-to-course, day-to-day, and residual variability in the pharmacokinetics of thiotepa and tepa were estimated with a population analysis with the software program NONMEM. The planned strategy consisted of the collection of blood samples on day 1 and either day 3 or day 4 of each 4-day course. The thiotepa dose was planned to be adjusted on day 3 of each course and before the start of a new course on the basis of Bayesian predictions of the pharmacokinetics with data of day 1 and/or the possible previous course. The prediction procedure was validated by dividing the dataset into an index and validation set. The Bayesian predictions of the validation set were compared with true AUC values generated with individual fits of each course. The performance of the complete strategy was tested with a simulation procedure in 1,000 patients. Interpatient variability and course-to-course variability were in the same order (+/-20%); day-to-day variability was less (+/-15%). The sampling strategy resulted in predictions of the AUC without bias with acceptable precision (+/-20%). The simulation showed that variability in exposure was effectively decreased by the dosing strategy. This strategy resulted in a reduction in the variability of the exposure to thiotepa and tepa and can be implemented in a clinical study.