Distinguishing IDH-mutant astrocytomas from IDH-wildtype glioblastomas using qualitative and quantitative MRI features: a WHO CNS5 study

Using qualitative and quantitative magnetic resonance imaging (MRI) features, this study aimed to distinguish between isocitrate dehydrogenase (IDH)-mutant astrocytomas (IDH-mA) and IDH-wildtype glioblastomas (IDH-wG) based on the fifth edition of the World Health Organization’s (WHO’s) classification of central nervous system (CNS) tumors (WHO CNS5), published in 2021. We enrolled 87 IDH-mA and 102 IDH-wG patients with pathologically confirmed disease according to the WHO CNS5 standard. Pretreatment brain MRI images and genetic information were obtained for each patient. Qualitative imaging features were assessed, including the side of lesion center, multifocality/multicentricity, hemorrhage, pial invasion, ependymal invasion, cortical involvement, midline location invasion, and enhancement mode. The quantitative imaging features assessed included tumor volume-related metrics and the relative apparent diffusion coefficient (rADC)-related metrics based on tumor segmentation. Contrast-enhanced and non-enhanced areas of the tumors were analyzed separately. univariable analysis and logistic regression were used to select the candidate predictors. The discrimination performance of the logistic regression model was evaluated using the area under the receiver operating characteristic curve (AUC). Internal validation was performed using the bootstrap approach. In terms of the qualitative features, IDH-mA exhibited less multifocality/ multicentricity (p = 0.032), more hemorrhage (p = 0.009), and more cortical involvement (p = 0.009) than IDH-wG. Regarding the quantitative imaging features, IDH-mA demonstrated higher values in Vall (p = 0.001), Vne (p < 0.001), rmaxADCce (p < 0.001), rminADCne (p = 0.015), rmaxADCne (p = 0.004), and rmeanADCall (p = 0.001) than IDH-wG. In the multivariable analysis of all patients, multifocality/multicentricity (odds ratio [OR] = 2.87, p = 0.033), Vne (OR = 1.02, p = 0.049), and rmaxADCce (OR = 2.82, p < 0.001) were independent predictive factors for distinguishing IDH-mA from IDH-wG. A combination of multifocality/multicentricity, Vne, and rmaxADCce (model) had a superior performance in distinguishing IDH-mA from IDH-wG, with an AUC, accuracy, sensitivity, and specificity of 0.849 (95% confidence interval [CI], 0.791–0.908), 76.5%, 66.3%, and 92.3%, respectively. We found that the combination of multifocality/multicentricity, Vne, and rmaxADCce, can help distinguish between IDH-mA and IDH-wG effectively.