Diagnostic performance of diffusion-weighted MR imaging at 3.0 T in predicting muscle invasion in urinary bladder cancer: utility of evaluating the morphology of the reactive tumor stalk.

Purpose: To evaluate the diagnostic performance of stalk morphology on diffusion-weighted imaging (DWI) in comparison with conventional MRI in predicting muscle invasion in urinary bladder cancer.Methods: The study was prospective and approved by the institutional ethics committee. A written informed consent was obtained from all the patients. The study included 56 patients who presented with bladder mass between January 2014 and November 2015. After excluding 16 patients, 40 patients with 92 tumors were assessed. All the 40 patients underwent MRI at 3.0 Tesla (Achieva, Philips) inclusive of DWI (b0, 500, 1000 and 1500). Two radiologists evaluated the images independently, and disparities were resolved through consensus. For predicting muscle invasion on T2-weighted images, tumor morphology (papillary versus non-papillary), distensibility of the underlying bladder wall, and perivesical fat infiltration were evaluated. On DWI, the criterion used in a previous study (Takeuchi et al.) was used along with tumor stalk morphology. Findings were compared with histopathology using Pearson’s χ2 test, and diagnostic performance indices were calculated.Results: All the evaluated features were present with significantly higher frequency in muscle-invasive tumors (p < 0.001). The finding of absent or distorted stalk on DWI had the highest sensitivity (87.5%) and specificity (97.6%). Conventional imaging features of non-papillary stalk morphology, restricted distension of underlying bladder wall, perivesical fat infiltration, as well as the previous DWI criterion were less sensitive (56.3%, 68.8%, 56.3% and 56.3%, respectively) in predicting muscle invasion.Conclusions: Assessment of the morphology of the reactive tumor stalk on DWI has better diagnostic performance in predicting muscle invasion than conventional MRI. [ABSTRACT FROM AUTHOR]