Presentation Authors: Soichiro Yoshida*, Hiroshi Tanaka, Toshiki Kijima, Minato Yokoyama, Junichiro Ishioka, Yoh Matusoka, Kazutaka Saito, Yasuhisa Fujii, Tokyo, Japan
Introduction: Multiparametric MRI has been widely used for local assessment of bladder cancer. In 2018, the Vesical Imaging-Reporting And Data System (VI-RADS) was proposed to standardize diagnostic criteria and report the risk of muscle-invasion (Panebianco et al. Eur Uro 2018). A five-point VI-RADS score is generated using the individual structural category (SC), diffusion-weighted (DW) category, and contrast-enhanced (CE) category. SC, DW, and CE categories were assessed using a five-point scale for T2-, DW-, and dynamic CE- (T2WI, DWI, and DCE) imaging, respectively. However, the diagnostic ability of VI-RADS has not been reported.
Methods: Among the primary bladder cancer patients who were diagnosed with pT1 or higher urothelial carcinoma between 2009 and 2016, we retrospectively evaluated 115 (MIBC, n= 46 [40%]; NMIBC, n= 69 [60%]) who underwent MRI examination including DWI before TUR. The patients whose index bladder cancer was less than 10 mm on T2WI, and in whom muscle sampling was not performed were excluded from the analysis. MRI was performed using a 1.5-Tesla imager (Intera Achieva; Philips). One radiologist with 11 years of experience in reading abdominal DWI interpreted the imaging data. The index tumor was scored according to the VI-RADS in each patient. We compared VI-RADS scores with histological diagnosis for muscle invasion.
Results: T2WI and DWI were obtained in all 115 patients, and DCE was performed in 87 of these patients (76%). The mean SC, DW, and CE category of NMIBC/MIBC was 3.1/4.5, 2.6/4.2 and 3.1/4.5, respectively. With a category threshold of 2, the negative predictive value (NPV) of T2W, DWI, and DCE was 95%, 95%, and 94%, respectively, while the positive predictive value (PPV) was low for all modalities (51-59%). If a category threshold of 3 was used, the PPV/NPV/accuracy was 71/90/81% for T2W, 82/86/84% for DWI, and 81/86/79% for DCE, respectively. The area under the curve (AUC) for the diagnosis of muscle invasion was 0.83 for T2WI, 0.88 for DWI, and 0.84 for DCE. A VI-RADS score of T2WI + DWI and T2WI + DWI + DCE resulted in AUCs of 0.88 and 0.86, respectively.
Conclusions: VI-RADS scoring is useful for bladder cancer staging. DWI was the leading sequence for accurately diagnosing muscle invasion. Combined use of DCE to T2WI and DWI did not necessarily increase detection accuracy of muscle invasion.