Estimating scenarios for survival time in men starting systemic therapies for castration-resistant prostate cancer: a systematic review of randomised trials

Background: We sought to estimate worst-case, typical and best-case scenarios for survival in men starting systemic therapies for castration resistant prostate cancer (CRPC).

Methods: We sought randomised phase 3 trials of systemic therapies for CRPC and recorded the following percentiles (represented scenario) from Kaplan-Meier overall survival (OS) curves: 90th (worst-case), 75th (lower-typical), 50th (median), 25th (upper-typical) and 10th (best-case). We determined the accuracy of using simple multiples of the median OS to estimate the other selected percentiles from each curve: 0.25 for 90th, 0.5 for 75th, 2 for 25th and 3 for 10th. Estimates were deemed accurate if within 0.75-1.33 times the actual value.

Findings: We reviewed 23 trials (13,909 men) with 48 treatment groups including 28 of chemotherapy, and three of novel hormonal agents. In trials of first-line docetaxel, the mean (interquartile range) for median OS was 19 months (17-20), and for each scenario was: worst-case 7 months (6-8); lower-typical 12 months (11-13); upper-typical 29 months (27-31); and best-case 40 months (34-44). For trials of novel hormonal agents after chemotherapy the mean values were: median OS 17 months, worst-case 5 months, lower-typical 9 months, upper-typical 24 months and best-case not reported. Simple multiples of the median gave accurate estimates of the worst-case scenario in 72% of OS curves, lower-typical in 89%, upper-typical in 84% and best-case in 84%.

Interpretation: Simple multiples of the median OS from randomised trials provided accurate estimates of worst-case, typical and best-case scenarios for survival time in men starting systemic therapies for CRPC.