162 O Randomized study of cyclophosphamide, doxorubicin & cisplatin (CAP) vs single agent carboplatin in ovarian cancer patients requiring chemotherapy: Interim results of ICON2

Abstract

In 1991, a large international trial, ICON2, was launched to compare the CAP regimen with optimal dose carboplatin with the aim of accruing a maximum of 2000 patients worldwide. Parallel trials were coordinated under the same protocol in the UK, Italy and Switzerland. Both arms consisted of 6 cycles of chemotherapy at three weekly intervals. The CAP arm gave cyclophosphamide 500 mg/m², doxorubicin 50 mg/m² & cisplatin 50 mg/m². The carboplatin arm gave a minimum recommended dose of 5(GFR+25) mg, using the AUC method of Calvert et al. (JCO 1987; 7: 1748–1756). An independent Data Monitoring Committee reviewed interim analyses of the data at regular intervals and gave advice on whether the data, together with results from other relevant trials, justified continuing the recruitment of further patients. By the end of September 1995, 1377 petients had been accrued from 140 centres in Italy, the UK, Switzerland, Eire, Poland, Australia, Greece, Singapore and Brazil. The two arms were well balanced regarding baseline characteristics. Compliance with the treatment protocol was good: in both arms 75% of patients received at least 75% of the planned dose intensity and at ieast 75% of the planned total dose. Patients treated with CAP experienced more severe toxicity overall (WHO grade III or IV): leucopenia 34% vs 10%; thrombocylopenia 7% vs 16%; nausea & vomiting 20% vs 9%; mucositis 21% vs 0%; alopecia 70% vs 3%; cardiac 2% vs 0%; other 2% vs 3%. A total of 526 patients progressed or died, with an 11% reduction in the risk ot progression or death in favour of CAP, (95% confidence interval=-6%, 24%, p=0.16). A total of 391 patients have died, with reduction in the risk of death in favour of CAP of 12% (-7%, 28%; p=0.20). Thus currently although there is a slight trend in favour of CAP there is no clear evidence of a difference in effectiveness between the two treatments. The data will be analysed again in October 1996 by which time we expect approximately twice as many events to have occurred.