To evaluate the performance of published guidelines compared with that of current practice for radiographic staging of men with newly diagnosed prostate cancer.

Using data from the Michigan Urological Surgery Improvement Collaborative clinical registry, we identified 1509 men diagnosed with prostate cancer from March 2012 through June 2013. Clinical data included age, prostate-specific antigen (PSA) level, Gleason score (GS), clinical trial stage, number of biopsy cores, and bone scan (BS) results. We then fit a multivariate logistic regression model to examine the association between clinical variables and the occurrence of bone metastases. Because some patients did not undergo BS, we used established methods to correct for verification bias and estimate the diagnostic accuracy of published guidelines.

Among 416 men who received a BS, 48 (11.5%) had evidence of bone metastases. Patients with bone metastases were older, with higher PSA levels and GS (all P <.05). In multivariate analyses, PSA (P <.001) and GS (P = .004) were the only independent predictors of positive BS. Guidelines from the American Urological Association and the National Comprehensive Cancer Network demonstrated similar performance in detecting bone metastases in our population, with fewer negative study results than those of the European Association of Urology guideline. Applying the American Urological Association recommendations (ie, image when PSA level >20 ng/mL or GS ≥8) to current clinical practice, we estimate that <1% of positive study results would be missed, whereas the number of negative study results would be reduced by 38%.

Based on current practice patterns, more uniform application of existing guidelines would ensure that BS is performed for almost all men with bone metastases, while avoiding many negative imaging studies.