Clinico-genomic risk stratification for biomarker-enriched trial design in poor-prognosis metastatic castrate-resistant prostate cancer (mCRPC)

This is the authors' abstract. We don't add key points for this paper.

Background: An integrated prognostic risk-score (RS) based on prognostic clinical factors and plasma copy number alterations (CNAs) across independent metastatic castration-resistant prostate cancer (mCRPC) cohorts was applied as a classifier to guide development of biomarker-enriched clinical trial designs. Design: Plasma CNAs prognostic for survival were combined into a prognostic score and integrated with clinical prognostic factors to derive an integrated RS in three independent mCRPC cohorts. Biomarker-enrichment trial designs were simulated across a spectrum of enrichment fractions and enrichment thresholds for the RS to consider 1:1 randomization of high RS patients to a study treatment arm or to receive standard treatment, with either a two or a three-year follow-up. The enrichment strategy was applied to improve the trial's ability to detect survival benefit at a Hazard Ratio 0.70; 80% power. Design trade-offs were assessed between the number of "high-risk" patients screened versus enrolled under varying enrichment thresholds and degrees of enrichment of the RS. Results: Pooled mCRPC patients (N = 561) showed RS values from 0.243 to 3.924 and a RS ≥ 1 (60th percentile) was associated with shorter survival. Elastic net models for 2- and 3-year survival were developed. At a threshold ≥1 with 2-year follow-up, a hybrid biomarker-enriched design (50% unselected, 50% high-risk) required screening 768 patients to enroll 219 per arm (n = 438), while maintaining adequate power. Conclusion: An integrated clinico-genomic classifier can guide an mCRPC enrichment strategy by enriching a cohort with 50% short-survival patients balancing sample size and screening needs to achieve adequate power in a biomarker-enriched design.