A Nomogram of MRI Features to Assess Muscle Invasion in VI‐RADS 2 Tumors With Stalk

Abstract

BackgroundVesical Imaging–Reporting and Data System (VI‐RADS) is widely used to assess the muscle‐invasive status of bladder cancer. However, the current classification efficacy of VI‐RASD 2 tumors of stalk is unsatisfactory.PurposeTo develop a nomogram to assess muscle‐invasive bladder cancer (MIBC) in VI‐RADS 2 tumors with stalk.Study TypeRetrospective.PopulationA total of 186 patients (age: 67.8 ± 12.7 years) with 15.1% females, divided randomly into a training cohort (N = 130) and validation cohort (N = 56).Field Strength/Sequence3‐T, T2‐weighted imaging (turbo spin‐echo), diffusion‐weighted imaging (breathing‐free spin‐echo), and dynamic contrast‐enhanced imaging (gradient‐echo).AssessmentTwenty‐one MRI features of tumors and stalks were developed from training cohort. The mean apparent diffusion coefficient (ADC) values of the tumor, stalk, and psoas muscles were calculated from the three circular regions of interest. The normalized T . The normalized ST . Three readers assessed the morphology of tumors and stalks.Statistical TestsThe final features of nomogram were selected by univariable logistic and the least absolute shrinkage and selection operator (LASSO) regression. The performance of the nomogram was assessed by the receiver operating characteristic (ROC) curve, calibration, and decision curve analysis.ResultsIn VI‐RADS 2 tumors with stalk, tumor size over 3 cm, increased stalk width, stalk morphology, decreased normalized T value, and increased normalized ST value were selected as the risk factors for MIBC. The AUC, accuracy, sensitivity, and specificity of the nomogram to assess MIBC were 0.969 (95% CI: 0.941–0.997), 92.3%, 94.1%, and 92.0% in training cohort and 0.940 (95% CI: 0.859–1.000), 89.3%, 75.0%, and 91.7% in validation cohort.Data ConclusionThis study constructed a nomogram for preoperative assessment of MIBC and modifying the current VI‐RADS.Level of Evidence3Technical EfficacyStage 2