The purpose of this study was to assess the prognostic value of in vitro drug chemosensitivity testing using the Hamburger-Salmon human tumor colony-forming assay (HTCA) in fresh tumor samples obtained from newly diagnosed patients with stage II-IV ovarian cancer undergoing maximum cytoreductive surgery and prior to platinum-based chemotherapy. The HTCA was performed on fresh ovarian cancers obtained from 93 patients at their initial exploratory laparotomy to evaluate in vitro sensitivity to cisplatin, carboplatin, and cyclophosphamide following a 1-hr drug exposure. Prospective clinical follow-up was performed on all patients with the primary study endpoints being pathologically proven complete response at second-look surgery and disease-free and overall survival durations. In vitro drug sensitivity was strongly dose-dependent. At a concentration of 5 micrograms/ml only 23% of tumor samples were sensitive (as defined by a > or = 50% decrease in tumor colony-forming units compared to controls) to cisplatin; 13% of tumors were sensitive to carboplatin at a concentration of 50 micrograms/ml and 11% to 4-OH-cyclophosphamide at a concentration of 1 microgram/ml. At doses which were 10 times the previously stated concentrations, the sensitivity rates to cisplatin, carboplatin, and 4-OH-cyclophosphamide increased to 72, 63, and 53%, respectively. Subjects were categorized as having drug-sensitive disease if HTCA results showed in vitro drug sensitivity to at least one of the agents used in their primary chemotherapy. Multivariate analysis failed to show any advantage in clinical response rate, progression-free interval, or survival duration for patients with drug-sensitive disease compared to drug-resistant disease; however, there was evidence of a trend toward an enhanced pathologically proven complete response rate in patients who had chemosensitive tumors in vitro. Second-look surgery was performed in 28 of 55 patients with optimal surgical resections and no clinical evidence of disease at the completion of their primary chemotherapy. Fifty percent (5/10) of patients with drug-sensitive disease achieved a pathologic complete response, while only 3/18 (17%) patients with drug-resistant tumors had a documented pathologic complete response (P = 0.13). As reported in other ovarian cancer studies, patient characteristics which were found to be significantly associated with survival were stage of disease (II-III vs IV), optimal primary surgical resection (i.e., < 1 cm2 residual tumor) vs suboptimal resection, clinical measurability of disease at initiation of chemotherapy, and response to primary chemotherapy. In conclusion, in vitro drug sensitivity, as measured by the HTCA, does not appear to be an independent prognostic factor for survival in patients with stage II-IV epithelial ovarian cancer who undergo standard treatment with tumor debulking surgery and primary platinum-based chemotherapy.(ABSTRACT TRUNCATED AT 400 WORDS)