Detecting the muscle invasiveness of bladder cancer: An application of diffusion kurtosis imaging and tumor contact length.

PURPOSE

To evaluate the diagnostic efficacy of diffusion kurtosis imaging (DKI) parameters and tumor contact length (TCL) among clinical and radiological factors for preoperative prediction of muscle-invasive bladder cancer (MIBC).

METHOD

A total of ninety-seven patients underwent 3.0 T MRI scan with propeller fast spin-echo T2WI, echo planar imaging diffusion-weighted imaging (DWI), and dynamic contrast-enhanced imaging (DCE). Two radiologists independently viewed multiparametric MRI (mpMRI) of each patient, graded the VI-RADS, drew the region of interest (ROI) and measured TCL. Interclass correlation coefficients (ICCs), Kappa statistics, Kolmogorov-Smirnov test, Mann-Whitney U tests, chi-square tests, logistic regression analyses, Hosmer-Lemeshow tests, receiver operating characteristic curve (ROC) analysis, and area under the curve (AUC) were applied.

RESULTS

The mean K of NMIBC group (0.62 ± 0.01) was significantly lower than that of MIBC group (0.79 ± 0.08). The mean TCL of MIBC group (4.66 ± 1.89) was significantly larger than TCL of NMIBC group (1.88 ± 1.50) (all p < 0.01). At the corresponding cut-off, AUC of TCL, K, VI-RADS and the combination of K and TCL were 0.87, 0.92, 0.90, and 0.95, respectively. TCL and K were risk factors of BC muscle invasion at both univariate and multivariate analysis.

CONCLUSIONS

K performed better than conventional DWI in predicting MIBC. K and TCL were independent risk factors of MIBC and could complement VI-RADS for predicting muscle invasion. The combination of K and TCL had the largest AUC and highest accuracy among all parameters.