Introduction:

Multiparametric MRI (mpMRI) is widely used in prostate cancer (PCa) diagnosis for its high sensitivity and negative predictive value (NPV). However, we previously showed the NPV of the PIRADS <3 threshold is only 77.1% in biopsy-naïve Black men—substantially lower than in predominantly White populations. While some MRI-invisible cancers are indolent, many are clinically significant (csPCa). This study aimed to identify demographic, pathologic and radiologic predictors of MRI-invisible csPCa in a diverse cohort undergoing mpMRI and radical prostatectomy (RP).

Methods:

We analyzed 61 patients (30 Black, 31 White/Hispanic) with pre-biopsy mpMRI followed by RP and whole-mount pathology. csPCa was defined as Gleason Grade Group (GG) 2–5 tumors ≥5 mm. Tumors were matched to PIRADS ≥3 lesions by a multidisciplinary panel. MRI-invisible tumors lacked radiologic-pathologic concordance. Logistic regression models (limited to 3 covariates) identified predictors of invisibility. We initially incorporated patient-level random intercepts, however the estimated random effects were negligible with minimal variance, and were therefore omitted from the models. 

Results:

Among 85 csPCa lesions, 20 (23.6%) were MRI-invisible csPCas ≥5mm. MRI-invisible tumors were smaller (0.12 vs. 2.17 cm³, p<0.001) and less likely to show extracapsular extension (5% vs. 39%, p=0.005). In multivariate analysis, Black race, extracapsular extension, horizontal width, apex-to-base length and the number of cs tumors were associated with MRI-invisibility. Black race was a consistent independent predictor across multivariable models with Model AUCs of 0.84–0.95 and the odds ratio for Black race ranging from 3.96–8.67 (p<0.05). Smaller tumor size and more csPCa tumors at RP also retained significance.

Conclusion:

MRI-invisible csPCa is associated with smaller tumor size including tumor length in the sagittal plane, multifocality, and Black race and negatively associated with extracapsular extension. These findings underscore limitations of mpMRI and highlight the need for enhanced diagnostic tools and race-sensitive risk stratification strategies.

Funding: 1. Using Prostate Index and MRI in Combination for Cost-Effectively Detecting High Grade Prostate Cancer, R01CA249973; 2. The Robert H. Lurie Comprehensive Cancer Center, P30CA060553-28