OPTIMIZATION OF PROSTATE BIOPSY STRATEGY USING COMPUTER-BASED ANALYSIS

Purpose: We evaluated and optimized the detection of cancer by prostat e biopsies. We developed a stochastic computer simulation model of ult rasound guided biopsies using mathematically reconstructed radical pro statectomy specimens. Use of this technique allows rapid evaluation of a variety of factors for their effect on prostate biopsy results. We used this model to analyze the effectiveness of sextant biopsies, whic h have been widely adopted in clinical practice. We also analyzed othe r biopsy schemes. Materials and Methods: A total of 607 tumor foci fro m 180 serially sectioned whole mount radical prostatectomy specimens w as mapped and digitized. The cancers had been clinically diagnosed by a variety of biopsy strategies. Simulated parasagittal sextant biopsie s were performed for each case. Forty simulation runs (each consisting of a set of 6 biopsies) were performed for each prostate, with realis tic random variations in sextant biopsy localization programmed in eac h run. Cancer detection by biopsy was considered reliable if 90% of th e simulation runs for each prostate were positive for cancer. A summar y algorithm was used to map the tumor foci. Results: Simulation of sex tant biopsies demonstrated reliably detected cancer in only 107 of 147 patients (73%) in whom total tumor volume was greater than 0.5 cc. Th ere was little correlation between total length of cancer in biopsy co res and tumor volume. Change of biopsy angle from 30 to 45 degrees did not result in significantly increased detection rates. Similarly, pla cing all biopsies more laterally did not increase overall detection ra tes. When we mapped tumor foci from the 40 cases in which sextant biop sies did not reliably detect tumor, we found that the foci were distri buted in areas not biopsied by the sextant method, that is the transit ion zone, midline peripheral zone and inferior portion of the anterior horn of the peripheral zone. A 10-core biopsy scheme incorporating th ese areas as well as the posterolateral prostate reliably detected can cer in 141 of 147 patients (96%) with total tumor volumes greater than 0.5 cc. Conclusions: Prostate cancer of significant volume can be pre sent in areas not sampled by standard sextant biopsies. Biopsies of th e transition zone, midline peripheral zone and inferior portion of the anterior horn of the peripheral zone should be considered for re-biop sy strategy after negative sextant biopsies. Sampling of these additio nal areas also can be incorporated in an initial biopsy scheme to incr ease overall initial rates of detection of prostate cancer.