Abstract

It is unclear whether clinical models including the Partin tables (PT), the Memorial Sloan Kettering Cancer Center nomogram (MSKCCn), and the cancer of the prostate risk assessment (CAPRA) can benefit from incorporating multiparametric magnetic resonance imaging (mpMRI) when staging prostate cancer (PCa). To compare the accuracy of clinical models, mpMRI, and mpMRI plus clinical models in predicting stage ≥pT3 of PCa. Prospective monocentric cohort study. Seventy-three patients who underwent radical prostatectomy between 2016-2018. 3.0T using turbo spin echo (TSE) imaging, single-shot echoplanar diffusion-weighted imaging, and T1 -weighted high-resolution-isotropic-volume-examination (THRIVE) contrast-enhanced imaging. We calculated the probability of extraprostatic extension (EPE) using the PT and MSKCC, as well as the CAPRA score. Three readers with 2-8 years of experience in mpMRI independently staged PCa on imaging. Receiver operating characteristics analysis and logistic regression analysis to investigate the per-patient accuracy of mpMRI vs. clinical models vs. mpMRI plus clinical models in predicting stage ≥pT3. The alpha level was 0.05. Median probability for EPE and MSKCCn was 27.3% and 47.0%, respectively. Median CAPRA score was 3. Stage ≥pT3 occurred in 32.9% of patients. Areas under the curve (AUCs) were 0.62 for PT, 0.62 for MSKCCn, 0.64 for CAPRA, and 0.73-0.75 for mpMRI (readers 1-3) (P > 0.05 for all comparisons). Compared with mpMRI, the combination of mpMRI with PT or MSKCCn provided lower AUCs (P > 0.05 for all the readers), while the combination with CAPRA provided significantly higher (P < 0.05) AUCs in the case of readers 1 and 3. On multivariable analysis, mpMRI by reader 1 was the only independent predictor of stage ≥pT3 (odds ratio 7.40). DATA CONCLUSION: mpMRI was more accurate than clinical models and mpMRI plus clinical models in predicting stage ≥pT3, except for the combination of mpMRI and CAPRA in two out of three readers. 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;50:1604-1613.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call