Presentation Authors: Fumitaka KOGA*, Kazumasa Sakamoto, Kosuke Takemura, Hiroaki Suzuki, Madoka Kataoka, Masaya Ito, Yasukazu Nakanishi, Ken-ichi Tobisu, Tokyo, Japan
Introduction: Accurate preoperative T staging is crucial for selecting candidates of neoadjuvant chemotherapy and regional lymph node dissection among UTUC patients. However, cT staging with standard CT/MRI is often inconsistent with pT stage. Moreover, it is difficult to discriminate T2 from Ta-T1 on standard imaging. Apparent diffusion coefficient values (ADC), measured on diffusion-weighted MRI, reflect invasive potential of bladder UC. We developed algorithms predicting â‰¥pT2 and LAD incorporating ADC in UTUC patients.
Methods: Subjects included 58 pathologically confirmed UTUC patients who took MRI/CT before nephroureterectomy (NU) at a single center between April 2013 and August 2018. ADC of tumor and the psoas muscle were measured on ADC maps and standardized tumor ADC (minimal tumor ADC/mean psoas ADC) were used to overcome the incompatibility between different MRI protocols. Best cutoffs of ADC were determined by partition analysis. Multivariate logistic regression analysis was used to develop models predicting â‰¥pT2 and locally advanced disease (LAD, â‰¥pT3 or pN+). Parameters examined included age, sex, tumor location, voided urine cytology (VUC), cT stage, hydronephrosis, and ADC.
Results: Of the 58 patients, 16 (28%) were diagnosed with â‰¥cT3 on CT/MRI. Pathological examination of NU revealed â‰¥pT2 and LAD in 35 (60%) and 31 (53%), respectively. When using cT stage (â‰¥cT3 vs. â‰¤cT2) on CT/MRI as a predictor, positive predictive values of â‰¥pT2 and LAD were 81% for both; however, â‰¥pT2 and LAD were understaged in 22 (63%) and 18 (58%), respectively. ADC were significantly lower in â‰¥pT2 (P=.04) and LAD (P=.03) than their less invasive counterparts. Multivariate analyses identified positive VUC (OR 8.0, P < .01) and ADC < 1.145 (the best cutoff: OR 20, P < .01) as predictors of â‰¥pT2, and hydronephrosis (OR 3.8, P=.03) and ADC < 1.145 (OR 33x106, P < .01) as predictors of LAD. When incorporating these predictors into cT stage, predictive accuracy improved from AUC 0.62 to 0.79 for â‰¥pT2 and from AUC 0.65 to 0.79 for LAD. When using algorithms predicting â‰¥pT2 and LAD proposed based on the above results, understaging was reduced by 41% (9/22) for â‰¥pT2 and by 72% (13/18) for LAD.
Conclusions: ADC could reflect invasive potential of UTUC. Algorithms incorporating ADC and VUC/hydronephrosis into cT stage on standard imaging may improve predictive accuracy of â‰¥pT2 and locally advanced UTUC.