Clinical implications of introducing a new PSA assay

Abstract

A number of different prostate-specific antigen (PSA) assays are in common use. There has been little consideration of the possible clinical implications of interassay variation. The availability of two assays in the same laboratory provided an opportunity to audit the clinical implications of the interassay variation in PSA levels. The same serum samples from patients with prostate cancer on follow-up were analyzed for PSA by the Abbott AxSYM assay and by the Abbott ARCHITECT assays. To assess within-patient reproducibility of the interassay variation, repeat analysis of PSA by both assays was conducted in a second sample obtained at least 1 month after the first. Samples from 156 cases were analyzed. The mean ratio of serum PSA values by the two assays (AxSYM assay/ARCHITECT assay) was 0.89 (range 0.5-2.27). The interassay coefficient of variation was 20%. In a subgroup of 50 cases with repeat samples available, the correlation coefficient, r, of the interassay variation in PSA between the first and second samples was 0.441. Interassay variation in serum PSA is clinically significant, both between patients and on repeated measurement within the same patient. Clinicians should be aware that simple correction factors may not accurately control for variation between PSA assays. Ideally, patients on follow-up for prostate cancer should be monitored using a single PSA assay.