Introduction

Multiparametric MRI (mpMRI) improves the detection of significant prostate cancer (PC) and extraprostatic extension (EPE). We combined pre-biopsy mpMRI data and clinical parameters to develop a risk model (RM) to predict individual side-specific risk of EPE on radical prostatectomy (RP).

Methods

MRI and clinical parameters of 132 men who underwent mpMRI fusion-biopsy and RP were analysed as training set. The RM was validated prospectively in 132 consecutive patients. Multivariate regression analysis was used to determine EPE predictors for RM development. The calibration of the RM was analysed using a calibration plot. The accuracy was compared to digital rectal examination (DRE), ESUR MRI criteria for EPE alone and the nomogram for side-specific EPE prediction of Steuber et al., using receiver operating characteristics (ROC) in training and validation set. Differences between the ROC curves were analysed using Likelihood ratio tests.

Results

Primary Gleason pattern on biopsy on specific side, ESUR MRI criteria of side-specific lesion, PSA-density, clinical T-stage, lesion volume in milliliter and capsule contact length in millimeter on MRI were significant EPE-predictors and were included in the RM (Figure a). The calibration plot of the RM showed that predicted and actual probabilities were close (slope 1.12)(Figure b). ROC area under the curve (AUC) for the RM was significantly larger in both sets (0.88 and 0.84), compared to DRE (0.69, p=0.004, 0.66, p<0.001) and the risk model of Steuber et al. (0.77, p=0.009, 0.71, p=0.006). Compared to ESUR criteria (AUC 0.87 and 0.80), the AUC was only significant larger in the validation set (p=0.03) (Figure c/d).

Conclusions

The RM, incorporating clinical and standardized MRI parameters performed significantly better compared to a renowned risk model, ESUR MRI criteria and clinical parameters alone. Thus, it provides accurate individual risk stratification of side-specific EPE of prostate cancers prior to RP.

Funding

None