A novel derivative of camptothecin, 9-aminocamptothecin (9-AC), is currently under Phase II evaluation in various cancers. Exceptionally mild toxicities were observed in patients with brain tumors who were treated with anticonvulsants. To investigate a pharmacokinetic interaction between 9-AC and anticonvulsants, and to evaluate the pharmacodynamics of 9-AC, we investigated the clinical pharmacology of 9-AC, administered by a 72-h infusion, in three Phase II studies. Plasma concentrations of total 9-AC (lactone plus carboxylate) at a steady state were measured in 56, 10, and 14 patients with non-small cell lung cancer, malignant glioma, and head and neck cancer, respectively. For lung cancer or glioma patients, 9-AC was infused at 45 (51 patients) or 59 (15 patients) microg/m2/h, and 9-AC was infused at 35.4 microg/m2/h in head and neck cancer patients. All glioma patients had been treated with phenytoin or carbamazepine. 9-AC clearance did not differ among the dosage rates, but differed according to the diseases (P = 0.002). Glioma patients had a higher clearance (1.0-18.0; median, 2.0 liters/h/m2) than lung cancer patients (0.3-5.1; median, 0.9 liters/h/m2). A logistic regression model described the relationship between the 9-AC concentration and the probability of grade 4 neutropenia, which was the main toxicity. Observed incidences of grade 4 neutropenia for patients with model-predicted probability of 0-20%, 20-40%, and 40-100% were 10%, 32%, and 67%, respectively, and corresponded to 9-AC concentration of <54, 54-86, and >86 ng/ml, respectively. Anticonvulsants seem to induce the clearance of 9-AC, and the concentration of 9-AC predicts the probability of grade 4 neutropenia.