Longitudinal analysis of serial measurements of free and total PSA among men with and without prostatic cancer

Abstract

Objectives Evaluation of free and total serum prostate specific antigen (PSA) levels before diagnosis of prostate cancer. Methods Free and total PSA levels were measured on frozen sera samples of 26 men with no history of prostate disease (controls), 29 men with a histologic diagnosis of benign prostatic hyperplasia (BPH) made at simple prostatectomy (BPH cases), and 23 men with a histologic diagnosis of prostatic cancer (cancer cases). Longitudinal regression analysis was used to evaluate PSA levels as a function of years before diagnosis of prostate disease. Results On average, mean total serum PSA was statistically significantly greater for subjects with cancer (5.0 ng/mL ± 0.9) versus BPH (2.8 ng/mL ± 0.3) and control subjects (0.8 ng/mL ± 0.1) by 4 years before diagnosis, whereas free PSA levels were similar among groups at 4 years before diagnosis. The ratio of free to total serum PSA continuously decreased among cancer cases over the decade before cancer diagnosis. At a time when mean total and free PSA levels were similar among groups (8 years before diagnosis), the ratio of free to total PSA was statistically significantly lower for cancer cases (0.13 ± 0.01) compared with BPH (0.17 ± 0.01) and control cases (0.21 ± 0.02). Use of a free to total PSA ratio of ≤0.12 when total PSA was between 4.0 and 10.0 ng/mL resulted in the highest sensitivity (76%) and specificity (94%) for diagnosis among subjects with and without cancer. Lowering the reflex range to 2.5 ng/mL increased false positive tests more than it increased sensitivity. Conclusions The ratio of free to total PSA is the earliest serum marker predicting a subsequent diagnosis of prostate cancer. Measurement of the free to total serum PSA ratio would appear to reduce false positive results among men without prostate cancer.