Whole-Tumor Histogram Analysis of Multiple Diffusion Metrics for Glioma Genotyping

Abstract

Background The isocitrate dehydrogenase <i>(IDH)</i> genotype and <i>1p/19q</i> codeletion status are key molecular markers included in glioma pathologic diagnosis. Advanced diffusion models provide additional microstructural information. Purpose To compare the diagnostic performance of histogram features of multiple diffusion metrics in predicting glioma <i>IDH</i> and <i>1p/19q</i> genotyping. Materials and Methods In this prospective study, participants were enrolled from December 2018 to December 2020. Diffusion-weighted imaging was performed by using a spin-echo echo-planar imaging sequence with five <i>b</i> values (500, 1000, 1500, 2000, and 2500 sec/mm²) in 30 directions for every <i>b</i> value and one <i>b</i> value of 0. Diffusion metrics of diffusion-tensor imaging (DTI), diffusion-kurtosis imaging (DKI), neurite orientation dispersion and density imaging (NODDI), and mean apparent propagator (MAP) were calculated, and their histogram features were analyzed in regions that included the entire tumor and peritumoral edema. Comparisons between groups were performed according to <i>IDH</i> genotype and <i>1p/19q</i> codeletion status. Logistic regression analysis was used to predict the <i>IDH</i> and <i>1p/19q</i> genotypes. Results A total of 215 participants (115 men, 100 women; mean age, 48 years ± 13 [standard deviation]) with grade II (<i>n</i> = 68), grade III (<i>n</i> = 35), and grade IV (<i>n</i> = 112) glioma were included. Among the DTI, DKI, NODDI, MAP, and total diffusion models, there were no significant differences in the areas under the receiver operating characteristic curve (AUCs) for predicting <i>IDH</i> mutations (AUC, 0.76, 0.82, 0.78, 0.81, and 0.82, respectively; <i>P</i> > .05) and <i>1p/19q</i> codeletion in gliomas with <i>IDH</i> mutations (AUC, 0.83, 0.81, 0.82, 0.83, and 0.88, respectively; <i>P</i> > .05). A regression model with an <i>R</i>² value of 0.84 was used for the Ki-67 labeling index and histogram features of the diffusion metrics. Conclusion Whole-tumor histogram analysis of multiple diffusion metrics is a promising approach for glioma isocitrate dehydrogenase and <i>1p/19q</i> genotyping, and the performance of diffusion-tensor imaging is similar to that of advanced diffusion models. Clinical trial registration no. ChiCTR2100048119 © RSNA, 2021 <i>Online supplemental material is available for this article.</i> <i>An earlier incorrect version appeared online. This article was corrected on December 14, 2021.</i>