Association of aggressive prostate cancer features on whole-mount histopathology and quantitative measures on PSMA PET.

Abstract

289 Background: The intraductal carcinoma (IDC) and the large cribriform growth pattern (LCP) are two architectural patterns of prostate cancer (PC) associated to more aggressive disease, high Gleason scores, advanced tumor stage, biochemical relapse, and distant metastasis. Predicting the presence of these pathology features could be of significant help in case whole-mount pathology (WMHP) is not available. The goal of this analysis was to assess the distribution of quantitative parameters from 68Ga-PSMA-11 PET/CT (PSMA PET) in lesions with and without LCP/IDC patterns on WMHP. We assessed the ability of pre-surgical PSMA PET imaging to predict the presence of aggressive pathology features. Methods: With IRB approval and HIPAA compliance, we derived a study cohort of PC patients who underwent PSMA PET prior to robotic radical prostatectomy (RALP). WMHP analysis described presence/absence of LCP and IDC patterns in each identified PC lesion. Board-certified physicians contoured all PC lesions on PSMA PET. Only lesions matching with WMHP were included in the analysis. PSMA PET metrics (SUVmax, SUVmean and PSMA-volume) were extracted. All lesions were categorized in sub-cohorts as LCP+/IDC+, LCP+/IDC-, LCP-/IDC-, and LCP-/IDC+. One way-ANOVA assessed significant differences among the imaging parameters in the sub-cohorts. The area under the curve (AUC) from ROC analysis was used to assess the ability of imaging metrics to predict the presence of aggressive PC features on WMHP. Results: The final cohort comprised 155 patients and 177 lesions. The median (IQR) PSA at time of RALP was 9.2 (5.55-15.05). 73/177 lesions (41.2%) were categorized as LCP+/IDC+, 63/177 (35.6%) as LCP-/IDC-, 37/177 (20.9%) as LCP+/IDC-, and 4 (2.3%) as LCP-/IDC-. SUVmax (p=0.007), SUVmean (p=0.048) and lesion size on PSMA PET (p=0.003) were significantly higher in lesions showing LCP, whereas only lesion size was significantly higher in lesions showing IDC (p=0.003). Lesions showing LCP+/IDC- had significantly higher SUVmax (p=0.002) and SUVmean (p=0.001) than lesions showing LCP-/IDC-. SUVmean was significantly higher in LCP+/IDC- than in LCP+/IDC+ (0.043). SUVmax and SUVmean were not significantly different among other sub-cohorts. Lesion size on PSMA PET was significantly higher in LCP+/IDC+ than in LCP-/IDC- (0.009). A logistic regression model analysis showed that PSMA PET quantitative measures predict the presence of LCP and/or IDC with moderate accuracy (AUC= 0.69; p=0.001). Conclusions: Higher SUVmax and SUVmean are found in lesions showing LCP, but not in those showing IDC. A linear regression model including SUVmax, SUVmean and PSMA PET lesion volume was able to predict with moderate accuracy the presence of pathology features of aggressive prostate cancer (AUC: 0.69).