Abstract: PD43-04
Sources of Funding: None.

Introduction

To evaluate the auxiliary function of periprostatic fat thickness (PPFT) on MRI to Prostate Imaging Reporting and Data System (PI-RADS) in predicting the presence of prostate cancer (PCa) and high grade prostate cancer (HGPCa).

Methods

The demographic data and the clinical information of 683 patients received transrectal ultrasound- (TRUS-) guided biopsy and multi-parametric magnetic resonance imaging (mp-MRI) were retrospectively reviewed. In addition, the PPFT was measured as the shortest perpendicular distance from the pubic symphysis to prostate on midsagittal T1-weighted MR images. The univariate and multivariate analyses were performed for determing independent predictors of PCa and HGPCa in whole study cohort and subgroups according to PI-RADS score. We also constructed two nomograms for predicting PCa and HGPCa based on binary logistic regression results.

Results

Overall, there were 371 patients (54.3%) having PCa and 292 patients (42.8%) having HGPCa. The mean value of PPFT was 4.04mm. Multivariate analysis revealed that age, PSA, TPV, PI-RADS score, PPFT were independent predictors of PCa. All factors plus DRE were independent predictors for HGPCa. The PPFT was the independent predictors of PCa (OR 2.56, p = 0.004) and HGPCa (OR 2.70, p = 0.014) for subjects with the PI-RADS score of 3. The present two nomograms based on multivariate analysis outperformed the single PI-RADS on aspects of predicting accuracy for PCa (aurea under the curve [AUC]: 0.922 vs 0.883, p= 0.029) and HGPCa (0.919 vs 0.873, p = 0.007). Decision-curve analysis also indicated superior net benefits and wide predicting ranges of the present two nomograms.

Conclusions

The PPFT on mp-MRI is an independent predictor of PCa and HGPCa, especially for patients with the PI-RADS score of 3. The nomograms incorporated predictors of PPFT and PI-RADS demonstrate good performance in predicting the prsence of PCa and HGPCa.

Funding

None.