Combined clinical characteristics and multiparametric MRI parameters for prediction of cribriform morphology in intermediate-risk prostate cancer patients

Abstract

Purpose: To develop a risk model with combined clinical characteristics and multiparametric MRI parameters for prediction of cribriform morphology in intermediate-risk prostate cancer (CaP) patients.

Methods: We retrospectively included 215 CaP patients received multiparametric MRIexamination, targeted biopsy (TB) combined with systematic biopsy (SB), radical prostatectomy and final Gleason group 2 or 3. Cribriform status was confirmed on both biopsy slices and whole-mount sections. Characteristics were stratified by cribriform status. Mann Whitney U test was performed for continuous variables and the χ² test for categorical variables. Univariate and multivariate logistic regression analyses were performed for significant predictors, followed by cribriform-risk nomogram construction. Receiver operating characteristic analysis was used for internal discrimination validation with corresponding area under the curve. Calibration curves were plotted and decision curve analysis was performed for clinical benefit exploration.

Results: Cribriform morphology was identified in 51.2% (110/215) patients. Cribriform-positive CaP demonstrated significantly higher prostate-specific antigen level, higher prostate-specific antigen density , larger lesion dimension on MRI, higher Prostate Imaging Reporting and Data System score, larger tumor dimension, higher Gleason score, higher pT stage, higher pN stage and more positive surgical margin (all P < 0.01). Sensitivities of TB, SB, TB + SB for detecting cribriform morphology were 28.2% (31/110), 22.7% (25/110), and 36.4% (40/110), respectively. Further, prostate-specific antigen density (P = 0.003), Prostate Imaging Reporting and Data System score (P < 0.001), and maximal biopsy Gleason score (P = 0.004) were independent predictors for positive cribriform morphology. Cribriform-risk nomogram was constructed with the 3 parameters and demonstrated considerable discrimination (area under the curve = 0.887, sensitivity 79.2%, specificity 84.0%) and calibration (mean absolute error 0.021), harboring net benefits with threshold probabilities range from 0 to 0.88.

Conclusion: A cribriform-risk nomogram was developed and well predicted aggressive cribriform morphology in intermediate-risk CaP patients.

Keywords: Biopsy; Cribriform morphology; Multiparametric MRI; Nomogram; Prostate cancer.