Imaging meets genomics in treatment-naïve prostate cancer: SUVmax on prostate-specific membrane antigen (PSMA) PET as a biomarker of genomic classifier risk.

Abstract

267 Background: Genomic Classifiers (GC) are increasingly used at the initial staging of prostate cancer to refine prognosis and guide treatment planning. The Decipher genomic classifier is a validated prognostic tool in prostate cancer, that predicts the risk of metastasis and prostate cancer-specific mortality. In parallel, PSMA PET/CT has emerged as a highly sensitive imaging modality for prostate cancer detection and staging, yet its relationship with genomic risk remains unclear. Establishing whether GC risk groups correspond with intraprostatic SUVmax on PSMA PET/CT could provide a non-invasive imaging biomarker and help identify SUV thresholds for discriminating high-risk disease. Methods: We retrospectively analyzed treatment-naïve men with histologically confirmed prostate cancer who had available Decipher genomic classifier testing, multiparametric prostate MRI and PSMA PET/CT obtained prior to or within 35 days following biopsy. Correlation between SUVmax and Decipher score was assessed using Spearman’s rho. Group comparisons were tested with Kruskal-Wallis and Mann-Whitney U tests, including low/intermediate vs high-risk. Diagnostic performance of SUVmax thresholds was evaluated using ROC analysis and Youden’s J statistic. Results: A total of 103 patients met eligibility criteria. Median age was 70 years (range, 46–84), median PSA was 6.7 ng/mL (range, 1.42–88.5), and median intraprostatic SUVmax was 11.45 (range, 2.4–81.8). Decipher risk groups were distributed as low (n=28), intermediate (n=25), and high (n=50). SUVmax was positively correlated with Decipher score (Spearman’s rho = 0.30, p =0.002), exceeding correlations with PSA (rho = 0.23, p =0.018) and Pi-RADS (rho = 0.12, p =0.24). Mann-Whitney analysis showed significant SUVmax differences between High vs Intermediate ( p =0.0055) and low/intermediate vs high ( p =0.042). ROC analysis identified SUVmax ≥15 as the optimal cutoff for predicting high Decipher risk (sensitivity 0.45, specificity 0.76, Youden’s J = 0.206). Conclusions: SUVmax on PSMA PET/CT correlated positively with Decipher genomic risk, with high-risk patients showing greater intraprostatic avidity. While differences across risk groups were modest, an SUVmax ≥15 emerged as the most discriminative threshold for identifying High-risk disease. These findings suggest that PET-derived SUVmax may provide complementary, non-invasive information to genomic classifiers and support integrated risk stratification in prostate cancer.