A novel liquid biopsy aid for identifying clinically significant prostate cancer in at-risk men: Combining artificial intelligence with flow cytometry-based immunophenotyping.

Abstract

e17508 Background: Current screening methods for prostate cancer (PCa) result in a large number of false positives making it difficult for clinicians to assess disease status thus warranting advancements in screening and early detection methods. The goal of this study was to design a liquid biopsy assay that uses flow cytometry-based immunophenotyping and artificial intelligence (AI) analysis to detect clinically significant PCa in at-risk men. Methods: Standard multiparametric flow cytometry techniques were used to immunophenotype 27 different MDSC, myeloid, and lymphocyte cell populations found in the peripheral blood of 114 biopsy-confirmed PCa (GS6 = 54; ≥ GS7 = 60) and 58 biopsy-confirmed benign prostatic hyperplasia (BPH) patients. Pattern recognition neural networks (PRNNs), a type of AI, were trained with flow cytometry data from two data sets: patients with confirmed clinically significant PCa (≥ GS7) versus patients with BPH or clinically insignificant PCa (GS6). Predictions were evaluated using the performance of the trained PRNNs on 37 known ≥ GS7 PCa patients and 40 known BPH/ < GS7 PCa patients not used in the PRNN training set (holdout samples). Results: In the holdout data set, 21 out of the 23 patients with clinically significant PCa were identified correctly (Sens = 91.3%, 95%CI = 72.0%-98.9%; PPV = 33.3%, 95%CI = 28.4%-38.7%) while 30 out of the 72 patients at low-risk or no-risk were identified correctly (Spec = 41.7%, 95%CI = 30.2%-53.9%; NPV = 93.8%, 95%CI = 79.5%-98.3%) for an accuracy of 53.7% (95%CI = 43.2%-64.0%). For all patients in this study, our data suggests that approximately 47 out of the 112 patients with BPH or GS6 PCa could have avoided an unnecessary prostate biopsy procedure. Conclusions: Pairing our PRNN with surface biomarker blood-based analyses of various leukocyte populations has enabled the ability to objectively identify patterns indicative for the existence of clinically significant PCa (≥ GS7). In a clinical setting, we believe that this technology, in use with other known clinical factors, would allow for clinicians to have a more informed decision when recommending their patients for a prostate biopsy procedure.