Internal Validation and Correlation of 2 Different Prostate Specific Antigen Assays for Prostate Cancer Screening in Community Based Men.

Abstract

There are more than 20 prostate specific antigen assays in use in the United States and studies have been mixed regarding correlations among them. The Abbott ARCHITECT Total PSA and the Roche Modular Analytics E170 Total PSA have been used in 2 local hospital systems that recently merged. We defined the comparability of these assays and present it from a clinical perspective. A prospective review of men 45 years old or older who underwent serum prostate specific antigen testing at our facilities was performed. Specimens were processed using both assays. Inclusion criteria were prostate specific antigen values obtained from March 1 to June 1, 2014. Prostate specific antigen values that were undetectable by either assay were excluded from study. The data were explored using regression analysis. Of 441 serum samples collected 414 fit the inclusion criteria. The data were stratified by prostate specific antigen 0 to 4 (367), 1.5 to 4.0 (107), 2.5 to 4.0 (48), 4.01 to 10 (33) and greater than 10 ng/ml (14). Regression analysis of the overall data yielded an R2 of 0.997 (SE 1.657). The subgroup analysis for each respective range showed R2 values of 0.974 (SE 0.15), 0.889 (SE 0.25), 0.677 (SE 0.25), 0.852 (SE 0.65) and 0.986 (SE 2.22). In general the prostate specific antigen range of most concern in prostate cancer screening is 4.01 to 10 ng/ml. Our data show an 85% correlation between the assays within this range. This relationship is too weak to recommend interchanging the tests with confidence. Additionally, we describe a successful model when combining laboratories using disparate assays.