Abstract
The study offers a retrospective analysis of the positive predictive value (PPV) of several variables, i.e. digital rectal examination (DRE), transrectal ultrasonography (TRUS), PSA value, PSA density (PSAD), and free/total PSA ratio (F/T), for the histologic outcome of 179 prostate biopsies performed within a population-based screening trial. The ratio of spared benign biopsies to missed cancers (SBB/MC) if biopsy results had been decided on the basis of single variables was also evaluated. PPV was 82.9% for DRE, 56.3% for TRUS, 26.6% for PSA (cutoff ≥4 ng/mL), 47.4% for PSA (cutoff ≥10 ng/mL), 42.0% for PSAD (cutoff 0.15), 59.2% for PSAD (cutoff 0.20), 34.9% for F/T (cutoff 0.20) and 40.0% for F/T (cutoff 0.15). SBB/MC was 121/23 for DRE, 96/12 for TRUS, 11/10 for PSA (cutoff ≥4 ng/mL), 107/34 for PSA (cutoff ≥10 ng/mL), 87/23 for PSAD (cutoff 0.15), 109/26 for PSAD (cutoff 0.20), 45/8 for F/T (cutoff 0.20) and 70/14 for F/T (cutoff 0.15). Multivariate analysis of the association with biopsy outcome showed the highest odds ratio for TRUS (13.24, 95% CI=4.4–30.7), and considerably lower values for DRE (4.17, 95% CI=2.0–8–9), PSAD (cutoff 0.20: 3.24, 95% CI=–1.8–5.7) and F/T (cutoff <0.15: 3.16, 95% CI =1.7–1.8). None of the possible variable combinations was clinically useful: the highest PPV (83.3%) was obtained with a combination of suspicious DRE/TRUS, PSAD >0.20 and F/T <0.15, which nevertheless missed 20 of 52 cancers.