Multiparametric MRI-based nomograms in predicting positive surgical margins of prostate cancer after laparoscopic radical prostatectomy.

Background: Positive surgical margins (PSMs) are an independent risk factor of biochemical recurrence in patients with prostate cancer (PCa) after laparoscopic radical prostatectomy; however, limited MRI-based predictive tools are available. This study aimed to develop a novel nomogram combining clinical and multiparametric MRI (mpMRI) parameters to reduce PSMs by improving surgical planning. Methods: One hundred and three patients with PCa (55 patients with negative surgical margins [NSMs] and 48 patients with PSMs) were included in this retrospective study. The following parameters were obtained using GE Functool post-processing software: diffusion-weighted imaging (DWI); intravoxel incoherent motion model (IVIM); and diffusion kurtosis imaging (DKI). Patients were divided into different training sets and testing sets for different targets according to a ratio of 7:3. The least absolute shrinkage and selection operator (LASSO) regression algorithm was used to analyze the data set to select the optimal MRI predictors. Preoperatively clinical parameters used to build a clinical nomogram (C-nomogram). Multivariable logistic regression analysis was used to build an MRI nomogram (M-nomogram) by introducing the MRI parameters. Based on the MRI and clinical parameters, build an MRI combined with clinical parameters nomogram (MC-nomogram). Comparisons with the M-nomogram and MC-nomogram were based on discrimination, calibration, and decision curve analysis (DCA). A 3-fold cross-validation method was used to assess the stability of the nomogram. Results: There was no statistical difference in AUC between the C-nomogram (sensitivity=64%, specificity=65% and AUC=0.683), the M-nomogram (sensitivity=57%, specificity=88% and AUC=0.735) and the MC-nomogram (sensitivity= 64%, specificity=82% and AUC=0.756). The calibration curves of the three nomograms used to predict the risk of PSMs in patients with PCa showed good agreement. The net benefit of the MC-nomogram was higher than the others (range, 0.2-0.7). Conclusions: The mpMRI-based nomogram can predict PSMs in PCa patients. Although its AUC (0.735) is not statistically different from that of the clinical-based nomogram AUC (0.683). However, mpMRI-based nomogram has higher specificity (88% VS. 63%), model stability, and clinical benefit than clinical-based nomogram. And the predictive ability of mpMRI plus clinical parameters for PSMs is further improved.