Relationship Between Metastasis-free Survival and Overall Survival in Patients With Nonmetastatic Castration-resistant Prostate Cancer

Abstract

Background: Metastasis-free survival (MFS) has been shown to be predictive of overall survival (OS) in hormone-sensitive localized prostate cancer. We evaluated the relationship between MFS and OS in nonmetastatic castration-resistant prostate cancer (nmCRPC).

Patients and methods: A retrospective analysis of 1207 high-risk patients with nmCRPC from the SPARTAN study (clinicaltrials.gov, NCT01946204) was undertaken. Landmark analyses of MFS status at several time points from randomization were performed to minimize guarantee-time bias. Hazard ratio (HR) of death as a function of MFS status was estimated based on a Cox proportional hazards model with 2-sided 95% confidence interval (CI). Estimated HRs were adjusted for stratification factors. Correlation analysis was performed using the Fleischer method.

Results: At all time points, MFS status strongly predicted OS. At landmark time points of 6, 9, and 12 months, risk of death was significantly higher for patients with metastases versus those without (adjusted HR at 6 months = 4.12; 95% CI, 2.60-6.54; P < .0001). MFS was positively correlated with OS based on the Fleischer method (HR, 0.69; 95% CI, 0.69-0.70; P < .0001). Approximately one-half of the variability in OS can be explained by MFS.

Conclusion: Metastasis development, regardless of time point, is associated with significantly greater risk of death in men with high-risk nmCRPC; hence, MFS is predictive of OS.

Keywords: Antiandrogen; Correlation; Endpoint; Landmark Analyses; Surrogate.