Genomic Classification to Predict Survival in Metastatic Prostate Cancer: Development of Somatic Tumor Risk Assessment for Overall Survival-Prostate.

Abstract

Purpose: Tumor comprehensive genomic profiling (CGP) has revolutionized cancer care and identifies patients for biomarker-specific therapy. In metastatic hormone-sensitive prostate cancer (mHSPC), although individual genes are prognostic, no comprehensive genomic classification exists using CGP that accounts for combinations of alterations to inform prognosis. We developed a DNA-based CGP classification that is prognostic for overall survival (OS) and could inform treatment.

Methods: This was a retrospective cross-sectional study using multivariable models to develop a clinicogenomic prognostic risk classification in US veterans with synchronous mHSPC. The primary outcome was OS from time of metastasis.

Results: A total of 7,201 veterans with metastatic prostate cancer and CGP were identified. There were 2,484 veterans (median [IQR] age, 72 [67-77] years) with synchronous mHSPC and tissue CGP, which were divided into training and testing data sets. Sixteen genes associated with survival were identified, and favorable, intermediate, and unfavorable genomic prognostication groups were created based on the mortality risk to generate the Somatic Tumor Risk Assessment for OS-Prostate (STRATOS-P) classification. In a multivariable model, classification into intermediate and unfavorable groups was associated with increased mortality relative to the favorable group (adjusted hazard ratio [aHR], 1.54 [95% CI, 1.33 to 1.78]; aHR, 2.37 [95% CI, 1.97 to 2.485], respectively), demonstrating an average AUC of 0.83. In an external, nonveterans validation cohort, intermediate and unfavorable classifications were associated with increased mortality (aHR, 2.45 [95% CI, 1.87 to 3.21]; aHR, 4.37 [95% CI, 3.06 to 6.22], respectively) with an AUC of 0.79. The intermediate and unfavorable genomic prognostication groups were also associated with increased mortality across multiple disease states including synchronous and metachronous diagnoses, castration resistance, and analyte type.

Conclusion: In metastatic prostate cancer, tumor DNA genomic alterations are prognostic for OS. The STRATOS-P classification is a validated prognostic tool that has the potential to guide decision making in mHSPC.