Predictive value of prostate specific antigen in a European HIV-positive cohort: does one size fit all?

BACKGROUND It is common practice to use prostate specific antigen (PSA) ≥4.0 ng/ml as a clinical indicator for men at risk of prostate cancer (PCa), however, this is unverified in HIV+ men. We aimed to describe kinetics and predictive value of PSA for PCa in HIV+ men. METHODS A nested case control study of 21 men with PCa and 40 matched-controls within EuroSIDA was conducted. Prospectively stored plasma samples before PCa (or matched date in controls) were measured for the following markers: total PSA (tPSA), free PSA (fPSA), testosterone and sex hormone binding globulin (SHBG). Conditional logistic regression models investigated associations between markers and PCa. Mixed models were used to describe kinetics. Sensitivity and specificity of using tPSA >4 ng/ml to predict PCa was calculated. Receiver operating characteristic curves were used to identify optimal cutoffs in HIV+ men for total PSA. RESULTS 61 HIV+ men were included with a median 6 (IQR 2-9) years follow-up. Levels of tPSA increased by 13.7% per year (95% CI 10.3, 17.3) in cases, but was stable in controls (-0.4%; 95% CI -2.5, 1.7). Elevated PSA was associated with higher odds of PCa at first (OR for twofold higher 4.7; 95% CI 1.7, 12.9; P<0.01) and last sample (8.1; 95% CI 1.1, 58.9; P=0.04). A similar relationship was seen between fPSA and PCa. Testosterone and SHBG level were not associated with PCa. tPSA level >4 ng/ml had 99% specificity and 38% sensitivity. The optimal PSA cutoff was 1.5 ng/ml overall (specificity =84%, sensitivity =81%). CONCLUSIONS PSA was highly predictive of PCa in HIV+ men; however, the commonly used PSA>4 ng/ml to indicate high PCa risk was not sensitive in our population and use of the lower cutoff of PSA>1.5 ng/ml warrants consideration.