Evaluation of a multigenomic liquid biopsy (PROSTest) for prostate cancer detection and follow-up in a Caribbean population

Introduction: The PROSTest is a novel machine learning-based liquid biopsy assay that functions as a diagnostic and prognostic tool in prostate cancer (PCa). The algorithm outcome (scored 0-100) has a cutoff of >50 to indicate PCa. In this study, we evaluated the screening utility of the test in comparison with the commonly used PSA test. Materials and methods: 509 men (median age: 62 [40-88]), predominantly Afro-Caribbean (n=449, 88.2 %) were enrolled in this study at a referral centre for prostatic cancer screening and management. These included 417 symptomatic patients with urinary tract (UT) problems, of which 12 were diagnosed with PCa, 53 with benign prostatic hyperplasia (BPH) and 352 were free from prostatic diseases, prior to PROSTest (Cohort 1). The remaining 92 of the 509, with relatively young age (median 49 and range 40-75 years) were screened with the PROSTest first and included 28 with UT symptoms and 64 asymptomatic but with a family history of cancer (Cohort 2). PROSTest and PSA was performed on all subjects. Results: In Cohort 1, the PROSTest was positive in 11 (91.7 %) PCa, 9 (17.0 %) of BPH, and 29 (8.2 %) of subject with UT symptoms but no disease. In Cohort 2, PROSTest was positive in 4 of 28 (14.3 %) symptomatic patients and in 5 of 64 (7.8 %) asymptomatic subjects. A total of 111 individuals (from both cohorts) were identified as "at risk" (PSA >4 ng/mL). Sixteen of 111 (14.4 %) were PROSTest-positive. Thirty-nine percent (43/111) have completed a 1-year follow-up. Of the 43 with elevated PSA, 2 out of 2 (100 %) who were also PROSTest-positive, were diagnosed with PCa. Conclusion: PROSTest is a more sensitive and specific biomarker than PSA in the diagnosis and, potentially, screening for PCa. The seemingly false positive PROSTest results in BPH or symptomatic UT require a longer follow-up of these patients.