Treatment-Effect Heterogeneity and a Simple Framework for Estimating Survival Benefit Associated With Local Therapy in Nonmetastatic Advanced Prostate Cancer: A Multicenter KAMU-ON Cohort Study.

OBJECTIVES

To evaluate treatment-effect heterogeneity associated with local therapy in nonmetastatic advanced prostate cancer and develop a simple hypothesis-generating framework to estimate the expected absolute survival benefit associated with local therapy.

METHODS

This multicentre retrospective cohort study included 1198 patients with clinical T3-4N0M0 or cN1M0 prostate cancer from the Kaga Million Uro-Oncology Network (KAMU-ON). Multivariable Cox models evaluated the association between local therapy and overall survival. Treatment-effect heterogeneity was assessed using interaction analyses. Adjusted 5 year overall survival and absolute survival benefit were estimated by g-computation. A simple clinical framework based on clinicopathological variables was derived from the subgroup analyses.

RESULTS

Local therapy was independently associated with improved overall survival (adjusted hazard ratio 0.32, 95% confidence interval 0.21-0.48; p < 0.001). Significant treatment-effect heterogeneity was observed according to clinical T stage (p for interaction = 0.030). The greatest absolute survival benefits were observed in patients with clinical T3b/T4 disease (20.3 percentage points) and Grade Group 4-5 disease (17.2 percentage points). The framework stratified the estimated absolute survival benefit from 0.1 to 19.5 percentage points across score groups (p for interaction = 0.041; p for trend = 0.023).

CONCLUSIONS

The survival benefit associated with local therapy varied substantially across patients with nonmetastatic advanced prostate cancer. This hypothesis-generating framework may provide estimates of expected treatment benefit and support individualized treatment discussions.