OBJECTIVE: High dose (HD) Ara-C combined with a single HD idarubicin dose (IDA) is an efficient and safe salvage regimen for patients with refractory or relapsed acute lymphoblastic leukemia as indicated by phase II studies. No data are available on the pharmacokinetics of IDA after a rapid HD intravenous infusion. An open phase II pharmacokinetic and clinical study was performed to evaluate antileukemic efficacy, IDA pharmacokinetics and to investigate the presence of IDA and its reduced metabolite idarubicinol (IDAol) in cerebrospinal fluid (CSF) of patients treated with HD-IDA. PATIENTS AND METHODS: Twenty-five patients with refractory or relapsed acute lymphoblastic leukemia received Ara-C 3 g/m2 from days 1-5, idarubicin (HD-IDA) 40 mg/m2 as rapid intravenous (i.v.) infusion on day 3 and subcutaneous G-CSF 5 microg/kg from day 7 until PMN recovery. Pharmacokinetics of IDA was evaluated after HD idarubicin administration in nine of these patients. CSF samples were collected in 15 patients at different times. IDA and IDAol concentrations were quantified by a validated HPLC assay described in detail elsewhere. RESULTS: Eleven patients (44%, 95% CI: 23-65%) achieved complete remission with median disease free survival for 6 months. After administration of HD-IDA i.v. bolus of 40 mg/m2, plasma level profiles of unchanged drug and IDAol were similar to those previously described after standard dose and measured with the same analytical method. The mean terminal half-life measured for IDA in this group of patients (14.9 h) was not significantly different from the mean value observed after standard dose (13.9 h, P=0.72). IDAol t1/2 was also similar after HD-IDA (46.2 h) and standard dose (39.4 h, P=0.79). Pharmacokinetic data reveal that in our series of patients IDA and IDAol clearances are significantly higher than those observed in patients treated with 12 mg/m2 of IDA but, although the administered dose (mg/m2) of the drug is 3.3 times higher, IDA exposure (measured in terms of AUC) is only 2.3 times and IDAol exposition 2.1 times greater. Furthermore, HD infusion resulted in a ratio between the AUC of parent drug and idarubicinol not different from the value observed with the standard-dose. IDA and IDAol were measurable only in 3 of the 15 cerebrospinal fluid samples collected. CONCLUSION: Responses observed in our series are comparable to those reported with other salvage regimens. The IDA exposure lower than expected may explain the safety of the single i.v. administration of 40 mg/m2 of IDA, combined with HD Ara-C, with a degree of myelosuppression equivalent to that reported with this agent administered in standard doses. Our data do not allow us to clearly attribute this behavior to a pharmacokinetic non-linearity since the baseline creatinine clearance, even within normal values, and patient age are significantly different in the two groups. Cerebrospinal fluid penetration was poor, reaching levels not considered as cytotoxic.