The value of textural analysis based on dynamic contrast-enhanced-MRI in predicting IDH genetic phenotypes of high-grade gliomas

Objective: To investigate the value of dynamic contrast-enhanced (DCE) -MRI based textural analysis in differentiating IDH mutated high-grade gliomas from IDH gene wild types. Methods: Twenty-nine patients with high grade gliomas collected from April 2016 to December 2019 in First People's Hospital of Changzhou were assessed retrospectively, including 10 patients with IDH mutation and 19 patients with IDH gene wild type. All patients underwent DCE, conventional plain and enhanced MR scanning. Omni Kinetics software was used to perform DCE-MRI data processing, volume transfer constant (K^trans), ratio constant of tracer refluxing from tissue to plasma (Kep), extravascular extracellular space per unit volume of tissue (Ve), blood plasma volume (Vp) and area under the gadolinium concentration-time curve (AUC) were obtained. Five commonly used textural features, including Energy, Entropy, Inertia, Correlation, and Invert Difference Moment (IDM), were generated based on gray-level co-occurrence matrices. The independent samples t test (normal distribution and equal variance) or Mann-Whitney rank sum test (abnormal distribution or unequal variance) was used to compare the differences in textural features of DCE-MRI parameters between IDH mutated group and IDH gene wild type group. Receiver operating characteristic (ROC) curves were used to evaluate the efficiency of textural features of DCE-MRI parameters in differentiating IDH mutated high-grade gliomas from IDH gene wild types for statistically significant textural features. Results: Entropy of K^trans and Ve for IDH mutated group were 5.368 ± 1.458 and 6.698 ± 1.081, respectively; while the corresponding values were 7.334 ± 1.385 and 8.213 ± 1.320 for IDH gene wild type group, respectively. The difference between the two groups was statistically significant (t values were -3.570, -3.113, P values were 0.001, 0.004, respectively). Inverse difference moment of K^trans and Ve for IDH mutated group were 0.567 ± 0.147 and 0.417 ± 0.106, respectively; while for IDH gene wild type group, the values were 0.393 ± 0.119 and 0.296 ± 0.101, respectively. The difference between the two groups was statistically significant (t values were 3.452, 3.014, P values were 0.002, 0.006, respectively). In all textural features, the area under the ROC curve of entropy of K^trans was the largest (0.874), and the sensitivity was the highest (100%), and the specificity of IDM of Ve was the highest (94.7%). Conclusion: Textural analysis of DCE-MRI can help to differentiate IDH mutated high-grade gliomas from IDH gene wild types. (authors)