Needle biopsies on autopsy prostates: sensitivity of cancer detection based on true prevalence.

BACKGROUND It is difficult to estimate the diagnostic accuracy of biopsy for prostate cancer because men with negative biopsy do not undergo radical prostatectomy and thus have no confirmation of biopsy findings. METHODS We performed 18-core needle biopsies on autopsy prostates from 164 men who had no history of prostate cancer. Six-core biopsies were taken from each of the mid peripheral zone (MPZ), the lateral peripheral zone (LPZ), and the central zone (CZ). We tested associations between age and tumor characteristics and analyzed the sensitivity of biopsies at each site. All statistical tests were two-sided. RESULTS Prostate cancer was present in 47 (29%) prostates. Of the 47 cancers detected, 20 were clinically significant according to histologic criteria. Tumor volume was associated with tumor grade (P = .012) and with age (P<.001). The biopsies from the CZ did not detect any cancer that was not present in biopsies of either the MPZ or LPZ. The sensitivity of the biopsies taken from the MPZ and LPZ together (53%, 95% confidence interval [CI] = 38% to 68%) was therefore the same as that of 18-core biopsies and was superior to that of biopsies of the MPZ alone (30%, 95% CI = 17% to 45%) (P = .003). The sensitivities of biopsies from the MPZ for clinically significant and insignificant cancer were 55% (95% CI = 32% to 77%) and 11% (95% CI = 2% to 29%), respectively, compared with 80% (95% CI = 56% to 94%) and 33% (95% CI = 17% to 54%) for those from the MPZ and LPZ combined. CONCLUSIONS The ability to detect prostate cancer was more related to the biopsy site than to the number of biopsy cores taken. The 12-core biopsies, six cores each from the MPZ and LPZ, were most likely to detect the majority of clinically significant cancers but also detected many insignificant cancers. When the six-core biopsies from the CZ were added, no increase in sensitivity was observed.