Utility of the PROSTest, a novel blood-based molecular assay, versus PSA for prostate cancer stratification and detection of higher grade disease.

Abstract

5107 Background: Prostate cancer (PCa) is the most common solid organ cancer in men and the fifth leading cause of cancer-related deaths globally. PSA has utility for identifying men at risk but has low specificity and is associated with significant biopsy-related morbidities through over-investigation and over-diagnosis of disease. The PROSTest is a novel 27-gene mRNA expression machine learning-based liquid biopsy assay that was developed to help detect PCa. We evaluated the utility of this assay to predict prostate cancers in symptomatic men undergoing biopsy or surgery for PSA >2ng/mL. Methods: One hundred and twenty-three men were evaluated, 105 (85%) met eligibility criteria (age >55 years, PSA >2ng/ml, symptomatic) and underwent image-guided biopsy or surgery. Samples for PSA and PROSTest were collected prior to biopsy. PROSTest was measured following RNA isolation and cDNA production from RNA stabilized blood samples. The PCR results were fed into a machine learning algorithm that reports a score on a scale 0-100, and with the use of cutoff 50, the score is transformed into a binary readout; positive or negative. In addition to the 2x2 table below, the diagnostic value of the PROSTest was demonstrated by ROC curve. Separately, we examined whether scores could be used to differentiate Gleason Grade 1 vs Gleason Grade 2-5 patients. Results: The median age of the cohort was 68 years (55-86 years). Baseline PSA median was 8.2ng/mL (IQR: 7.2-92ng/mL). Sixty-five (62%) had a PCa diagnosis (27 GG1; 38 GG2-5) as confirmed by biopsy or after surgery. Baseline PROSTest results were >50 in 65 (62%). The distribution of PROSTest versus outcomes is included in Table 1. PROSTest sensitivity was 97% and specificity of 96% for detecting PCa. The AUROC for PROSTest was 0.39 higher than PSA (PROSTest: 0.99 vs. PSA: 0.61, p<0.0001). GG2-5 exhibited significantly higher (p<0.0001) PROSTest scores (92±3.8) than GG1 and BPH (41±41). PROSTest scores >93 were 79.1% accurate and 92.5% specific for detecting higher risk disease. Conclusions: In this head-to-head comparison, baseline PROSTest was a more sensitive and more specific biomarker than PSA in the diagnosis of all Gleason’s grades and may have value for differentiating GG1 from GG2-5. This requires validation in a larger, prospectively collected cohort. 2x2 table identifying PROSTest score relationship with PCa detection (tissue diagnosis). Biopsy/Surgery Prostate cancer detected(n=65) No Prostate cancer (including 5 prostatitis, 35 BPH)(n=40) PROSTest +ve (score ≥50) 63 (96.9%) 2 (5.0%) PROSTest -ve (score <50) 2 (3.1%) 38 (95%)