Background/Objectives: Early detection of prostate cancer (PCa) enables timely therapeutic intervention and improved clinical outcomes. Screening strategies are increasingly individualized and now incorporate multiparametric MRI findings, reported using the Prostate Imaging Reporting and Data System (PI-RADS), to refine biopsy decision-making. PROSTest is a novel machine learning (ML)-enhanced, 30-gene mRNA liquid biopsy assay developed to detect PCa from whole blood. In this prospective study (NCT06872619), we evaluated whether PROSTest could function as a pre-biopsy triage tool to inform biopsy decisions while preserving sensitivity for clinically significant prostate cancer (csPCa). Methods: Of 121 men evaluated, 111 (91.7%) completed the full diagnostic work-up—including PSA testing, PROSTest analysis, and PI-RADS assessment—and subsequently underwent image-guided biopsy. Peripheral blood samples for PROSTest were collected prior to biopsy. RNA-stabilized samples underwent RNA isolation followed by reverse transcription and quantitative PCR. Gene expression data were processed using a proprietary machine learning algorithm to generate a continuous range from 0 to 100. A clinically validated cut-off ≥50 was applied to produce a binary (positive/negative) result. The diagnostic accuracy of PROSTest was assessed against histology-confirmed prostate cancer. Results: The median age of participants was 69 years (47–83 years) and the median PSA was 7.5 ng/mL (IQR: 5.8–11.4 ng/mL); most patients (104 of 111; 93.7%) had a PI-RADS score of three to five. PCa was diagnosed in 97 men (87.4%) including eight in ISUP Grade Group (GG) 1, 46 in GG2, 33 in GG3, three in GG4 and seven in GG5. PROSTest was positive in 102/111 (91.9%). Among men with biopsy-confirmed PCa, diagnostic accuracy was 99% (93/94). Of the 17 men without histologic evidence of disease, eight (47%) were PROSTest-negative. The overall accuracy was 91% (84.1–95.6%) with an NPV of 89% (51.6–98.4%). Among the nine patients with positive PROSTest but negative biopsy, PI-RADS scores were 4 (n = 6), 3 (n = 1), and 2 (n = 2). Conclusions: PROSTest demonstrated an overall accuracy of 91% (95% CI: 84.1–95.6%) with an NPV of 89%. Among men without a detectable prostate cancer on biopsy, 47% (8/17) were PROSTest-negative. These results suggest that PROSTest may serve as a useful pre-biopsy triage assay.
Rahbar et al. (Sun,) studied this question.