Several non-randomised trials have shown that transitional-cell carcinoma of the bladder is a moderately chemosensitive tumour. We investigated whether the addition of neoadjuvant cisplatin-based chemotherapy to radical surgery or radiotherapy would improve survival.

Patients with T2 G3, T3, T4a, N0-NX, or M0 transitional-cell carcinoma of the bladder undergoing curative cystectomy or full-dose external-beam radiotherapy were randomly assigned three cycles of neoadjuvant chemotherapy (cisplatin, methotrexate, and vinblastine, with folinic acid rescue, n=491) or no chemotherapy (n=485). When possible, clinical tumour response was assessed cytoscopically after completion of chemotherapy but before cystectomy or radiotherapy; histopathologically assessed response was on cystectomy samples. We recorded every 6 months locoregional persistence or relapse of tumour, appearance of distant metastases, survival, and cause of death.

Median follow-up of patients still alive was 4.0 years. 485 patients died, and 78.6% of deaths were due to transitional-cell carcinoma. Chemotherapy mortality was 1% and operative (cystectomy) mortality was 3.7%. Kaplan-Meier curves compared by means of the log-rank test gave a calculated absolute difference between groups in 3-year survival of 5.5% (95% CI -0.5 to 11.0, p=0.075; 55.5% for chemotherapy, 50.0% for no chemotherapy). Median survival in the chemotherapy group was 44 months compared with 37.5 months for the no-chemotherapy group. 32.5% of cystectomy samples contained no tumour after neoadjuvant chemotherapy.

Three cycles of neoadjuvant chemotherapy before cystectomy or radiotherapy did not give the 10% improvement in 3-year survival that was judged to be necessary for introduction into routine use. The chemotherapy regimen was associated with a higher pathological complete-response rate in primary tumours, but there was no clear evidence that it would increase survival.