Diagnosis of prostate cancer in patients with persistently elevated PSA and tumor-negative biopsy in ambulatory care: performance of MR imaging in a multi-reader environment.

PURPOSE: False-negative results are obtained in approx. 20 % of prostate cancer (PCa) patients (pts) at initial systematic transrectal biopsy (Bx), in particular when digital rectal examination (DRE) or transrectal ultrasound (TRUS) is negative. The aim of this study was to assess whether MR endorectal imaging of the prostate in a multi-reader ambulatory care setting may assist in patient selection for re-biopsy. MATERIALS AND METHODS: 115 consecutive pts with persistent PSA elevation, negative Bx, DRE and TRUS were examined using T2w axial and coronal and T1w axial sequences for tumor diagnosis. MR images were prospectively read as tumor-suspicious or tumor-negative by the MR radiologist on duty. Additionally, a retrospective readout of a prostate MR expert and an abdominal imaging fellowship-trained radiologist was performed to evaluate the effect of the reader's experience on tumor detection. Imaging findings were compared to the results of the repeat Bx (61 pts) or the clinical course of at least two years. RESULTS: For the prospective reading, the sensitivity of MRI was 83 %, the specificity was 69 %, the PPV was 33 % and the NPV was 96 %. ROC analysis revealed a significantly better performance of the prostate MR imaging expert compared to the abdominal imaging radiologist (area under ROC 0.88 vs. 0.66, p < 0.001). Based on the prospective reading, a pre-test probability for PCa of 17.4 % as in our study can be reduced to 5 % when obtaining a tumor-negative result in MRI. CONCLUSION: MR imaging in a multi-reader ambulatory care setting assists in patient selection for re-biopsy. Reducing the post-test probability for PCa to 5 % allows for further follow-up instead of re-biopsy in MR tumor-negative patients. Specific training and experience improve tumor detection in prostate MR imaging.