Kim, Yohan

ABSTRACTBackgroundCommercial biomarkers and multiparametric MRI (mpMRI) have been utilized to triage men with elevated prostate specific antigen (PSA) and determine patients most likely to harbor clinically‐significant prostate cancer (csPCa). We studied combinations of mpMRI, PSA‐based and novel extracellular vesicle (EV)‐based biomarkers to determine the optimal pre‐biopsy testing to predict csPCa at biopsy.MethodsMen presenting with elevated PSA (≥ 2 ng/mL) were prospectively enrolled and all men underwent clinically indicated mpMRI and blinded study blood draws to determine PSA, prostate health index (PHI) scoring, and EV serum levels. MRI‐fusion transperineal prostate biopsy was performed in all patients. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated, and receiver operator characteristic (ROC) curves were constructed. Bootstrapping analysis was performed to provide more accurate assessment of predictive abilities.ResultsUltimately, 175 consecutive men were prospectively enrolled. Median age in the study population was 65 years. Combinations of biomarkers and MRI demonstrated better predictive ability for csPCa on biopsy than individual modalities. Predictive ability was greatest for PHI density (PHID)‐mpMRI with an AUC of 0.86 (95% CI: 0.80–0.92) while combinations of PSA density ‐mpMRI (AUC: 0.81; 95% CI: 0.73–0.89) and STEAP1‐mpMRI (AUC: 0.77; 95% CI: 0.70–0.86) demonstrated similar ability to predict csPCa. The study is limited by small, predominantly white patient cohort and requires external validation.ConclusionsInclusion of prostate density with biomarkers increases prognostic ability for detecting csPCa. EV density can refine prediction of csPCa and in combination with PHID‐mpMRI leads to superior specificity, thereby decreasing unnecessary biopsies.

337 Background: Prostate cancer remains the most common solid organ malignancy among males in the United States. Commercial biomarkers and multiparametric MRI have been utilized to better triage men with elevated PSA and determine patients most likely to have clinically significant prostate cancer (csPCa) upon prostate biopsy. We studied the combination of MRI with PSA-based and novel extracellular vesicle (EV)-based biomarkers to determine the optimal pre-biopsy testing to reduce the number of unnecessary biopsies. Methods: Men presenting to our clinic with elevated PSA (≥ 2 ng/ml) were prospectively enrolled from October 2019 to March 2022. All men underwent multiparametric MRI and blood draws to determine prostate health index (PHI) scoring, EV density, and PSA density. MRI-fusion biopsy was performed with both systematic and targeted prostate sampling, with clinicians blinded to prior blood testing results. Blood concentrations of prostate EV (ProsEVs) were measured by nanoscale flow cytometry using antibodies against PSMA and STEAP1. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated, and receiver operator characteristic (ROC) curves were constructed separately for all patients, those without prior diagnosis of PCa, and biopsy-naïve patients. Bootstrapping analysis was performed to provide more accurate assessment of the predictive ability of all combinations. Results: Ultimately, 175 men enrolled and underwent MRI, blood testing, and prostate biopsy. Among the 175 patients included, 101 (58%) had never undergone biopsy, 38 (22%) had a previous negative biopsy, and 36 (21%) were on active surveillance (AS) for GS6 disease. AUC of mpMRI (PI-RADS≥3) to predict clinically significant PCa was 0.66 with a sensitivity, specificity, NPV and PPV of 95.2%, 37.6%, 76.0% and 79.26%. MRI combined with STEAP1-EV density had an AUC of 0.77 with a sensitivity, specificity, NPV and PPV of 90.1%, 46.7%, 64.4% and 80.9%. A multivariable model including MRI, STEAP1-EV density, PSA density and PHI density outperformed other combinations with an AUC, sensitivity, specificity, NPV and PPV of 0.90, 89.2%, 68.7%, 72.8% and 88.1%. Conclusions: Combination of mpMRI with ProsEV and PSA derivatives demonstrated better predictive ability for csPCa on biopsy compared to imaging or serum testing alone. Inclusion of prostate volume with biomarkers further increased predictive ability which could lead to a decrease in number of unnecessary biopsies for men with benign conditions and indolent cancer.

334 Background: Prostate cancer (PCa) screening has relied on prostate-specific antigen (PSA) levels to screen patients for biopsy. However, the limited specificity of serum PSA leads to high rates of negative biopsies and detection of clinically insignificant cancers. Tumor-derived extracellular vesicles, microscopic particles released by tumor cells and ubiquitously found in all bodily fluids, have emerged as novel biomarkers for a diverse set of malignancies. This study explored the utility of urine prostate-specific extracellular vesicles (ProsEVs) identified by unique surface protein markers with novel flow cytometry-based techniques as biomarkers for PCa screening. Methods: This prospective study recruited at a tertiary care center between Oct 2019 and Mar 2022. Patients with PSA≥2 ng/ml underwent prebiopsy mpMRI followed by transperineal prostate biopsy (PI-RADS 1-2) or combined MRI-guided biopsy (PI-RADS ≥ 3). Patients with metastatic PCa or prior systemic therapy were excluded. Urine samples were collected at MRI using Colli-Pee. ProsEVs were labeled using monoclonal antibodies for PSMA and STEAP1 and ProsEV concentrations were determined using nanoscale flow cytometry. EV densities were calculated by normalizing EV counts to prostate volume. Multivariable logistic regression models were constructed and model performance was compared used ROC curve analysis. Results: This study had a total of 179 patients prospectively recruited, 95 were diagnosed with Gleason score (GS) ≥ 7 prostate cancer (csPCa), while the remaining had GS <7 or benign disease. Analysis of urinary ProsEV densities revealed a significant dose-response relationship between benign, GS <7 PCa and GS ≥ 7 PCa groups and ProsEV densities (p=0.01 for PSMA, and <0.001 for STEAP1). The log sum of the PSMA and STEAP1 EV densities exhibited the strongest correlation with other ProsEV variables and was selected for model development. ROC curve analysis showed that the final model including Log Sum ProsEVs, PSA, and PI-RADS outperformed one including PSA and PI-RADS (0.733 vs 0.651, p=0.02). Specificity for detecting csPCa at a sensitivity of 90% was 39.5% in the final model compared to 37% for PSA and PI-RADS (Table). Conclusions: This study highlights the potential of urine ProsEVs as novel biomarkers for improving screening specificity for csPCa detection. Future works should explore analysis of intra-ProsEV content to further improve specificity and advance liquid biopsy techniques. [Table: see text]