B-077 Multicenter Comparison of PSA Immunoassays: Analytical Bias and Its Impact on Prostate Cancer Screening

Abstract

Prostate-specific antigen (PSA) testing plays a central role in prostate cancer screening and management, guiding clinical decisions from initial detection to biopsy and follow-up. However, total PSA (tPSA) results can vary across commercial immunoassays due to lack of testing harmonizations, potentially altering interpretation near decision thresholds and leading to unnecessary biopsies or missed cancers. Four academic hospitals (Corewell Health/Oakland University, University of Washington, The Ohio State University, and The University of Texas MD Anderson Cancer Center) participated in this study, each using a distinct automated tPSA platform: Abbott Architect i2000, Beckman Access DxI, Roche Cobas e801, and Siemens Atellica IM. Analytical comparison was made using 25 leftover, de-identified serum samples tested across all four platforms within the same week, percent bias was calculated relative to the peer-group mean. For clinical performance, retrospective data were analyzed across institutions by comparing biopsy-confirmed prostate cancer cases and controls without prostate cancer diagnosis, clinical sensitivity and specificity were evaluated at PSA thresholds of 3.0 and 4.0 ng/mL with age stratification (=45, =50, =55 years). Substantial inter-assay variation was observed. In pooled sample testing, Beckman DxI demonstrated the largest positive bias (average +13%) and Siemens Atellica showed the most negative bias (average -14%), with Roche Cobas mildly positive (+6%) and Abbott Architect slightly negative (-5%). These analytical patterns were reflected in clinical performance. At the 4.0 ng/mL cutoff, sensitivity across age strata was the highest for Beckman DxI (88%), followed by Roche Cobas (84–85%), Abbott Architect (80–83%), and Siemens Atellica (75–76%). Specificity at 4.0 ng/mL ranged from 92–94% (Siemens, Abbott) to 89–92% (Beckman) and 79–86% (Roche), suggesting additional influence from institutional patient populations. This multicenter study demonstrated significant analytical and clinical variation among commercial PSA assays, driven by both platform-specific bias and differences in patient populations. These findings highlight the need for improved assay standardization and for clinicians to understand inter-assay variability and clinical context (including age and institutional referral patterns) to support consistent PSA interpretation and clinical decision-making.