Adjuvant cisplatin chemotherapy following cystectomy for bladder cancer: results of a prospective randomized trial

Between April 1984 and May 1989, 77 eligible patients with invasive, nonmetastatic (stage M0) transitional cell carcinoma of the bladder were stratified after radical cystectomy and pelvic lymph node dissection on the basis of nodal status (stage pN0 versus pN1-2) and were randomly assigned to either observation or postoperative cisplatin chemotherapy (3 courses of 90 mg./m.2 cisplatin given for 3 consecutive days at monthly intervals). Patient eligibility included a creatinine clearance of greater than 60 ml. per minute. There were 40 eligible patients in the control group (median age 61 years) and 37 in the cisplatin group (median age 64 years). In regard to postoperative tumor stage and nodal status, there was no statistical difference between the 2 patient groups. In the cisplatin group 21 patients received the full dose, 9 required dose reduction and 7 refused treatment. Median followup was 5 years 9 months (range 3 to 8 years). Survival analysis showed no significant difference (log rank p = 0.65) between the 40 patients in the control group and the 37 in the cisplatin group. The survival rate at 5 years was 54% (95% confidence interval 39 to 69%) in the control group and 57% (95% confidence interval 40 to 74%) in the treatment group. Patients with cancer confined to the bladder wall (stage pT3a or less) had a 5-year overall survival rate of 70% and those with tumor growth in the perivesical fat or into the prostate (stages pT3b plus pT4a) had a 5-year overall survival rate of 40%. This difference in survival between the low stage subgroup (stages pT3a or less) and the high stage subgroup (pT3b plus pT4a) is highly significant (p = 0.0043). However, no difference between the controls and the cisplatin group was found within either the low or high stage subgroups. The reasons for failing to show a survival benefit from adjuvant high dose cisplatin monotherapy after radical cystectomy are discussed.