Value of prostate multiparametric magnetic resonance imaging for predicting biopsy results in first or repeat biopsy

To assess multiparametric magnetic resonance imaging (mp-MRI) in predicting prostate biopsy results.

Patients who underwent mp-MRI prior to prostate biopsy were prospectively included. The prostate was subdivided into 14 sectors and mp-MRI findings assessed using a five-level subjective suspicion score (SSS). Biopsy included targeted samples of abnormal sectors and systematic samples of normal peripheral zone sectors.

Two hundred and eighty-eight patients were included [153 biopsy naïve, 135 with negative (n = 51) or positive (n = 84) prior biopsy]. Biopsy was positive in 168 patients. mp-MRI area under the receiver operating characteristic (ROC) curve (AUC) was 69.1% (95% CI: 67.1-70.9%), 72.5% (95% CI: 69.5-76%), and 73.8% (95% CI: 68.3-79.3%) at per sector, per lobe, and per patient analysis, respectively. At the per sector level, the AUC was significantly larger if detection was limited to cancers with a Gleason score of ≥7 (72.6%; 95% CI: 69.8-75.8%; p < 0.01) or ≥8 (87.1%; 95% CI: 78.3-95.7%; p < 0.01). mp-MRI performance was significantly influenced by prostate volume (p = 0.02), the presence of a concordant hypoechoic area (p < 0.001), but not by prostate-specific antigen (PSA) value, status of prior biopsy, or radiologists' experience. SSS was significantly associated with the Gleason score in true-positive lobes and patients (p < 0.0001). Using a SSS threshold of ≥3, cancer was missed in 13/102 lobes and 4/72 patients with cancers of Gleason score ≥7.

mp-MRI provides a good detection of cancers with a Gleason score of ≥7 in candidates suitable for prostate biopsy.