Magnetic resonance diffusion tensor imaging histogram analysis facilitates non-invasive prediction of IDH1 and ATRX mutation status in WHO grade 2/3 gliomas

Abstract ObjectivesTo investigate whether magnetic resonance diffusion tensor imaging (DTI) histogram analysis can provide significant value for the non-invasive prediction of IDH1 combined with ATRX gene mutation status in WHO grade 2/3 gliomas.Methods97 patients with pathologically confirmed WHO grade Ⅱ and III gliomas were classified into IDH1 gene wild type (IDH1wt) and IDH1 gene mutation type (IDH1mut). Patients with IDH1 mutation are further divided into patients with ATRX mutation(IDH1mut ATRX⁻) and those without ATRX mutation (IDH1mut ATRX⁺). The histogram characteristics of two indicators of DTI, apparent diffusion coefficient (ADC) and fractional anisotropy (FA), were calculated for the areas covering the whole tumour and the peritumour oedema for comparison between the groups. The accuracy of diagnosis was evaluated and compared with the ROC curve.ResultsThe whole-lesion DTI histogram can significantly differentiate IDH1 gene status in lower-grade gliomas, and distinguish whether IDH1 mutations are accompanied by ATRX gene mutations. The FA 5th percentile is the best predictor of IDH mutations, and the ADC mean is the best predictor of IDH mutations accompanied by ATRX mutations. The combined parameters had better predictive efficacy than the single parameters. The maximum combined parameter of IDH1wt and IDH1mut for the identification of WHO grade Il and III adult diffuse glioma had a 0.703 AUC, 70.2% sensitivity and 70.0% specificity. The predicted maximum combined AUC for IDH1mut ATRX⁻ and IDH1mut ATRX⁺ was 0.741, with a sensitivity of 78.8% and specificity of 62.5%.ConclusionDTI histogram analysis based on the tumour global domain can help to non-invasively predict the status of IDH and ATRX genes in lower-grade gliomas and assist in clinical decision-making and prognosis.