Background: Despite biomarker development advances, early detection of aggressive prostate cancer (PCa) remains challenging. We previously developed a clinical-grade urine test (Michigan Prostate Score [MiPS]) for individualized aggressive PCa risk prediction. MiPS combines serum prostate-specific antigen (PSA), the TMPRSS2:ERG (T2:ERG) gene fusion, and PCA3 lncRNA in whole urine after digital rectal examination (DRE).
Objective: To improve on MiPS with a novel next-generation sequencing (NGS) multibiomarker urine assay for early detection of aggressive PCa.
Design, setting, and participants: Preclinical development and validation of a post-DRE urine RNA NGS assay (Urine Prostate Seq [UPSeq]) assessing 84 PCa transcriptomic biomarkers, including T2:ERG, PCA3, additional PCa fusions/isoforms, mRNAs, lncRNAs, and expressed mutations. Our UPSeq model was trained on 73 patients and validated on a held-out set of 36 patients representing the spectrum of disease (benign to grade group [GG] 5 PCa).
Outcome measurements and statistical analysis: The area under the receiver operating characteristic curve (AUC) of UPSeq was compared with PSA, MiPS, and other existing models/biomarkers for predicting GG ≥3 PCa.
Results and limitations: UPSeq demonstrated high analytical accuracy and concordance with MiPS, and was able to detect expressed germline HOXB13 and somatic SPOP mutations. In an extreme design cohort (n = 109; benign/GG 1 vs GG ≥3 PCa, stratified to exclude GG 2 cancer in order to capture signal difference between extreme ends of disease), UPSeq showed differential expression for T2:ERG.T1E4 (1.2 vs 78.8 median normalized reads, p < 0.00001) and PCA3 (1024 vs 2521, p = 0.02), additional T2:ERG splice isoforms, and other candidate biomarkers. Using machine learning, we developed a 15-transcript model on the training set (n = 73) that outperformed serum PSA and sequencing-derived MiPS in predicting GG ≥3 PCa in the held-out validation set (n = 36; AUC 0.82 vs 0.69 and 0.69, respectively).
Conclusions: These results support the potential utility of our novel urine-based RNA NGS assay to supplement PSA for improved early detection of aggressive PCa.
Patient summary: We have developed a new urine-based test for the detection of aggressive prostate cancer, which promises improvement upon current biomarker tests.
Keywords: Algorithm; Biomarkers; Early detection; Machine learning; Mutations; Next-generation sequencing; Prostate cancer; RNA; Transcriptome; Urine detection.
Copyright © 2021 European Association of Urology. All rights reserved.