Characterizing population-wide genomic risk distribution for development of a novel clinical-genomic risk system for prognostication in patients with clinically localized prostate cancer

作者：Udit Singhal, Ralph Jiang, James A. Proudfoot, Elizabeth C. Chase, Krithika Suresh, Elai Davicioni, Tudor Borza, Michael J. Zeléfsky, Brian J. Davis, B.J. Stish, R. Jeffrey Karnes, Stephen J. Freedland, Martha K. Terris, William J. Aronson, Matthew R. Cooperberg, Fábio Ynoe de Moraes, Alejandro Berlín, Curtiland Deville, Nicholas G. Zaorsky, Soumyajit Roy, Angela Y. Jia, Jonathan Shoag, Will Jackson, Daniel E. Spratt, Matthew J. Schipper, Todd M. Morgan, Robert T. Dess · 发表于：Prostate Cancer and Prostatic Diseases · 年份：2025 · DOI：10.1038/s41391-025-01062-8 · 被引用次数：2 · 研究领域：Prostate Cancer Diagnosis and Treatment、Prostate Cancer Treatment and Research、Cancer Genomics and Diagnostics

PURPOSE: Genomic classifiers are endorsed by guidelines and commonly used to inform prognosis in prostate cancer. We aimed to understand the distribution of genomic risk within the validated staging collaboration for cancer of the prostate (STAR-CAP) and propose a system integrating genomic and clinicopathologic risk. We hypothesized that genomic heterogeneity would have implications on risk estimates and may inform treatment decisions. MATERIALS AND METHODS: Genomic risk was assessed using the Decipher genomic classifier in two separate multi-institutional, prospectively collected population-based registries: (1) Decipher Genomics Resource for Intelligent Discovery (GRID) [n = 50,891] and (2) Michigan Urological Surgery Improvement Collaborative (MUSIC-Decipher) [n = 1602]. The primary endpoint was estimated prostate cancer-specific mortality (PCSM), and secondary endpoint was distant metastasis (DM). Marginal risk estimates provided by STAR-CAP were combined with hazard ratios of Decipher to calculate integrated risk estimates. RESULTS: Median age and PSA was 68 years and 6.2 ng/mL in GRID, and 66 years and 10.5 ng/mL in MUSIC. The GRID population had 50.2%, 18.5%, and 31.4% with low-, intermediate-, and high-Decipher risk, compared to 48.0%, 16.2%, and 35.8% in MUSIC. Decipher-based genomic risk varied across STAR-CAP stages in both registries. Estimates of 10-year PCSM (0.1% to 48.8%) and DM (0.3%-72.9%) varied widely after integration of clinical-genomic risk. Use of an ...