The combination of cisplatin-based chemotherapy with interleukin-2 (IL-2) and interferon, referred to as biochemotherapy, has shown encouraging results in patients with advanced melanoma. Toxicity is high, however and no objective parameters exist to distinguish between patients who are likely to respond and those who are not. The purpose of this pilot study was to determine whether in vitro cisplatin-induced damage to the glutathione S-transferase-pi (GST-pi) gene in peripheral blood mononuclear cells (PBMCs) before therapy correlated with the histological response in melanoma patients with local-regional metastases who received concurrent biochemotherapy before definitive surgery. Before therapy, PBMCs from 16 patients were exposed to cisplatin at concentrations of 25, 50 or 100 microM for 3 h and the extent of damage to the GST-pi gene was quantitated by polymerase chain reaction (PCR). Patients were subsequently treated on a biochemotherapy regimen consisting of cisplatin 20 mg/m2 intravenously (i.v.) on days 1-4, vinblastine 1.5 mg/m2 i.v. on days 1-4, dacarbazine 800 mg/m2 i.v. on day 1, IL-2 9 MIU/m2 per day i.v. by continuous infusion on days 1-4 (total of 96 h), and interferon alpha2a 5 MU/m2 subcutaneously on days 1-5. The 16 patients were categorized into two groups: major responders (n = 7) and non-major responders (n = 9). Although we observed a wide interpatient variation, a statistically significant correlation existed between the histological response and the degree of DNA damage caused in the PBMCs at all three cisplatin concentrations tested (P = 0.024 for 25 microM; P = 0.036 for 50 microM; P = 0.007 for 100 microM). Our pilot study suggests that determination of in vitro cisplatin-induced DNA damage using a gene-specific PCR assay may be useful in predicting the histological response to biochemotherapy.