To compare cisplatin-cyclophosphamide versus carboplatin-cyclophosphamide as primary chemotherapy for stage III (suboptimal) and stage IV ovarian cancer.

Three hundred forty-two patients were randomly assigned to treatment with six courses of intravenous (i.v.) cisplatin 100 mg/m2 plus i.v. cyclophosphamide 600 mg/m2, or i.v. carboplatin 300 mg/m2 plus i.v. cyclophosphamide 600 mg/m2.

The estimated median survivals were 17.4 and 20.0 months for the cisplatin and carboplatin study arms, respectively. The null hypothesis of a 30% survival superiority with the cisplatin arm was rejected at the P = .02 level. Clinical response rates were 52% for the cisplatin arm and 61% for the carboplatin arm. Pathologic complete response rates were similar for both study arms. There was less thrombocytopenia on the cisplatin arm (P less than .001); however, there was less nausea and emesis (P less than or equal to .001 for courses 1 to 5), renal toxicity (P less than .001), anemia (P = .01), hearing loss (P less than .001), tinnitus (P = .01), neuromuscular toxicities (P = .001), and alopecia (P less than .001) on the carboplatin arm.

Carboplatin-cyclophosphamide proved to have a significantly better therapeutic index than cisplatin-cyclophosphamide in patients with stage III (suboptimal) and stage IV ovarian cancer.