Our purpose was to determine the maximum tolerated dose of, and the minimum interval between treatments with, multiple cycles of carboplatin (CBDCA) rescued with peripheral blood progenitors and filgrastim. Eligible patients had advanced cancers without prior chemotherapy or radiotherapy. The study design involved a sequential cross-over in which patients initially received two or three courses of cyclophosphamide (CPA) at a dose of 3.0 g/m2, supported by filgrastim. Multiple leukaphereses were then performed during the rebound phase of hematological recovery following each CPA-induced nadir to harvest peripheral blood progenitors, which were then reinfused as rescue following each of four courses of CBDCA. We attempted to administer the CBDCA at 14-day intervals. The CBDCA dose (mg/m2/course) was escalated as follows in successive cohorts of patients: Level I, 500; Level II, 800; Level III, 1200; Level IIIa, 1000. Following determination of the maximum tolerated dose of CBDCA administered in this fashion, a subsequent cohort of patients (Level IV) were treated with two courses of high-dose CPA and four courses of the combination of CBDCA (1000 mg/m2) plus CPA (1500 mg/m2). Thirty-one patients were enrolled in the trial. Five patients were removed from study prior to completion of protocol therapy, three due to toxicity and two who developed progressive cancer while on study. The maximum tolerated dose of CBDCA was 1000 mg/m2, with dose-limiting ototoxicity occurring at 1200 mg/m2. The median inter-treatment interval for all cycles was 15 days (range, 12-30). The median intervals between CBDCA courses for each dose level were: Level I, 17 days; Level II, 17 days; Level III, 14 days; Level IIIa, 15 days; Level IV, 16 days. The median dose intensity of the CPA phase was 1493 mg/m2/week. The median (and range) CBDCA dose intensities (measured from the start of CBDCA) for each dose level were: I, 185 (151-222); II, 328 (305-380); III, 567 (512-646); IIIa, 465 (363-481); Level IV, 468 (333-500). Neutropenic fever complicated 35 of 113 CBDCA or CBDCA/CPA courses. Platelet transfusion was required in 51 of 113 courses. One patient had severe epistaxis. There were no treatment-related deaths. Among 27 patients with ovarian cancer who were evaluable for response, there were 5 pathologically documented complete (including 3 of 10 at Level IV) and 16 partial responses. We concluded that peripheral blood progenitors facilitate the simultaneous dose escalation and schedule intensification of carboplatin chemotherapy. The effect is sustained over four courses of treatment.(ABSTRACT TRUNCATED AT 400 WORDS)