Radiologic and Pathologic Concordance of Multiparametric MRI for Detecting Unilateral Prostate Cancer: Implications for Focal Therapy Selection.

BACKGROUND

Accurate selection for focal therapy depends on reliable assessment of tumor laterality. Although multiparametric MRI improves local staging, unilateral findings often underestimate pathological disease extent. We hypothesized that mpMRI-defined unilateral disease does not consistently reflect true organ-confined laterality and that integrating imaging with clinical and biopsy variables enhances patient selection.

METHODS

A cross-sectional analysis of a consecutive cohort of patients was conducted undergoing radical prostatectomy between 2016 and 2025. All patients had preoperative multiparametric MRI (mpMRI) and combined targeted plus systematic biopsy. Radiological staging was assigned according to PI-RADS v2.1 and TNM criteria. Whole-mount prostatectomy specimens served as the reference standard. Unilateral organ-confined disease (pT2a-pT2b) was grouped as pT2ab and defined as the pathological endpoint of interest. Diagnostic accuracy, radiological-pathological concordance, and predictors of bilateral or locally advanced disease were analyzed using logistic regression models.

RESULTS

Among 232 patients, 66 (28.4%) had unilateral organ-confined disease (pT2ab) and 166 (71.6%) had bilateral or locally advanced disease. Baseline clinical variables were comparable between groups. mpMRI showed moderate radiological-pathological agreement (51.7%; κ = 0.30) and systematically overestimated organ-confined disease. Although rT2ab staging was highly sensitive for pT2ab (84.8%), specificity was limited (51.8%). A parsimonious multivariable model combining radiological T stage, biopsy laterality, and percentage of positive cores improved discrimination (AUC 0.78-0.79), showed excellent calibration, and enabled clinically meaningful rule-out and rule-in strategies across predefined probability thresholds.

CONCLUSIONS

mpMRI alone may overestimate unilateral disease; integrating biopsy-derived tumor burden and laterality improves pathological prediction and supports safer candidate selection for focal therapy.