Stratification of prostate-specific antigen level and results of transrectal ultrasonography and digital rectal examination as predictors of positive prostate biopsy.

OBJECTIVE: To determine which of several variables--age, serum level of prostate-specific antigen (PSA), findings of transrectal ultrasonography (TRUS) and findings of digital rectal examination (DRE)--are the best predictors of positive prostate biopsy results.

SETTING: An urban, university-affiliated tertiary care hospital.

PATIENTS: a cohort of 1330 consecutive men referred to the diagnostic imaging department for TRUS and TRUS-guided prostate biopsy. Each patient was referred after examination by a urologist because of clinical suspicion of prostate cancer.

METHODS: All of the men had undergone prior determination of serum level of PSA. Repeat DRE was performed at the time of imaging. The variables age, PSA level, TRUS findings and DRE findings were tested aline and in combination as predictors of positive biopsy results by means of logistic regression analysis. A summary of percentage risk for positive biopsy results was constructed for each combination of statistically significant variables, stratified for age.

RESULTS: Cancer was detected in 541 men (40.7%). A strong correlation was observed between serum PSA level and the likelihood of positive biopsy result (p < 0.001). Of 402 men with normal age-specific PSA, 109 (27.1%) had positive biopsy results. Of 403 men with PSA of 10 ng/mL or more, 233 (57.8%) had cancer. The level of serum PSA was also related to the number of prostate sextants harbouring cancer (p < 0.001). TRUS findings at the time of biopsy were a strong predictor of cancer of the PSA level was abnormal (p < 0.001). DRE results alone did not correlate with positive biopsy results, regardless of age, PSA level of TRUS findings.

CONCLUSIONS: Men in whom there is a clinical concern for prostate cancer should undergo prostate biopsy if there is any elevation of age-specific PSA, particularly if the findings of TRUS are also abnormal. Because DRE alone was not predictive of biopsy outcome, algorithms for prostate cancer detection that rely on abnormal DRE results to identify men who should undergo biopsy will miss a significant number of cancers.