Grading and proliferation assessment of diffuse astrocytic tumors with monoexponential, biexponential, and stretched-exponential diffusion-weighted imaging and diffusion kurtosis imaging

Purpose: To compare the main parameters derived from monoexponential, biexponential and stretched-exponential diffusion-weighted imaging (DWI) and diffusion kurtosis imaging (DKI) with respect to diagnostic performance for tumor grading and proliferation assessment in diffuse astrocytic tumors (DATs). Materials and methods: Fifty-eight pathologically confirmed DAT patients who underwent DWI and DKI on a 3-T scanner were prospectively collected and retrospectively reviewed. Measurements including the apparent diffusion coefficient (ADC), true diffusion coefficient (D), pseudodiffusion coefficient (D*), perfusion fraction (f), distributed diffusion coefficient (DDC), heterogeneity index (alpha), mean diffusivity (MD), fractional anisotropy (FA), and mean kurtosis (MK) were compared between tumor grades (II, III, and IV) by using a Jonckheere-Terpstra test. Receiver operating characteristic (ROC) curves were used to assess the diagnostic efficacy of these parameters. Spearman's rho with the Ki-67 labeling index (LI) was calculated for each parameter. Results: MK values differed significantly between all DAT subtypes and increased with grade. The ADC, D, f, DDC, alpha and MD values were significantly higher in grade II tumors than in grade III/IV tumors. D* values were significantly lower in grade II tumors than in grade IV tumors (all P < 0.05). In discriminating between grade II and III tumors, alpha, MK, MD, D and f had significantly greater area under the ROC curve (AUC) values than D* and FA (0.927, 0.901, 0.896, 0.895, and 0.889, respectively vs 0.659 and 0.598, respectively, P < 0.05). In discriminating between grade III and IV tumors, only MK demonstrated acceptable discrimination (AUC = 0.711). MK and D showed a strong correlation with the Ki-67 LI (rho = 0.791 and -0.789, respectively, P < 0.001). D*, f, MD, ADC, DDC and alpha showed a moderate correlation (vertical bar rho vertical bar ranged from 0.415 to 0.698, P < 0.05). Conclusion: MK and D have considerable potential to predict the degree of proliferation of DATs. MK could effectively characterize microstructural changes throughout the malignant transformation of DATs and provided useful complementary information for grading.