Abstract: PD11-10
Sources of Funding: none

Introduction

The ability to accurately identify prostate cancer on multiparametric MRI (mpMRI) has been shown to be directly related to the size and grade of the tumor. No studies to date have evaluated how different architectural patterns affect the performance of mpMRI. Our objective was to determine whether Gleason pattern 4 architecture impacts tumor visibility on mpMRI and correlate with final histopathology.

Methods

A total of 83 tumors were identified within 22 radical prostatectomy specimens from patients who underwent MR/US fusion biopsy followed by radical prostatectomy at our institution from January 2015-July 2016. Each tumor focus was characterized by: % Gleason pattern 4, overall Gleason score, predominant architectural pattern (poorly formed, cribriform, fused), tumor size, axis (anterior/posterior), region (apex/mid/base), laterality (left/right), and presence of extraprostatic extension. Each region of interest (ROI) on mpMRI, ranging from PIRADS 3-, was re-reviewed and paired with its corresponding pathological tumor focus. Tumors not paired with an ROI were classified as "not visible." Multiple logistic regression was performed to determine predictors of tumor visibility._x000D_

Results

Out of 83 tumors identified, 33/83 (40%) were Gleason 3+3 (36%) and 50/83 (36%) were Gleason 3+4 or above. Among Gleason pattern 4 tumors, 14/50 (28%), 18/50 (36%), and 18/50 (36%) had predominantly poorly formed, cribriform, and fused gland architecture, respectively. MpMRI detected 17/30 (57%) of Gleason 3+4, 6/14 (43%) of Gleason 4+3, and 2/5 (40%) of Gleason 4+4 and above. Among tumors containing Gleason pattern 4, increasing tumor size and non-cribriform architecture predicted individual tumor detection on multivariate analysis (p = 0.002 and p = 0.011, respectively). Cribriform predominant tumors were detected by MRI in 5/17 (21%) compared to 20/33 (61%) when poorly formed or fused glands were the predominant architecture (p = 0.01). The size threshold for detection of cribriform tumors was higher than that of other architectural patterns (Figure)._x000D_

Conclusions

Independent of lesion size, cribriform predominant tumors are less visible on mpMRI. Clinically this could limit the utility of mpMRI in men with known cribriform tumors, since these tumors behave more aggressively than other Gleason pattern 4 tumors._x000D_

Funding

none