A composite multigene risk score derived from tumor-biology alterations specific to metastatic castrate-resistant prostate cancer (mCRPC) state was evaluated as a classifier to design biomarker-based enrichment clinical trials. A plasma cell-free DNA copy number alteration risk score based on alterations in 24 genes was simulated to develop a biomarker classifier-based clinical trial design enriched for high-risk patients to detect a survival advantage of a novel treatment (hazard ratio of 0.70 with 80% power). We determined the design trade-offs between the number of patients screened and enrolled when varying the type of patients to enrich and the extent of enrichment needed. For a 2-year overall survival end point in mCRPC state, fully enriching patients with mCRPC having a high-risk score of 3 or more (the 95th percentile of a range of risk scores in patients with mCRPC) was determined to require screening to a maximum of 4,149 patients to enroll 259 patients for the targeted effect size. A nonenriched trial was determined to require enrolling 689 patients to be equivalently powered. We identified a pragmatic alternative, which is to enrich patients with mCRPC with a risk score of 1 or more (the 67th percentile) and an enrichment fraction of 0.25. This would require screening 658 patients to enroll 584 patients, and it maximizes the ability to detect a difference in treatment effect by risk score. A plasma multi-CNA risk score classifier can feasibly be leveraged to design an enrichment trial in mCRPC. Enriching 25% of patients screened with a risk score >1 was observed to be optimal for obtaining an adequately powered, biomarker-based mCRPC-enriched clinical trial.
Read full abstract