New Subependymal Enhancement After Radiation Therapy in High-Grade Glioma: Utilizing Morphological Features and DSC Perfusion MRI in Differentiate Progression and Post-Radiation Changes
Kyu Sung Choi,Inpyeong Hwang,Chul Kee Park,Sung Hye Park,Seung Hong Choi
DOI: https://doi.org/10.1002/jmri.29586
2024-09-05
Abstract:Background: The specific patterns of subependymal enhancement (SE) that frequently occur as radiation-induced changes in high-grade gliomas following radiotherapy are often overlooked. Perfusion MRI may offer a diagnostic clue. Purpose: To distinguish between radiation-induced SE and progression in adult high-grade diffuse gliomas after standard treatment. Study type: Retrospective. Population: Ninety-four consecutive high-grade diffuse glioma patients (mean age, 55 ± 14 years; 54 [57.4%] males) with new SE identified in follow-up MRI after completion of surgery plus chemoradiation: progression (N = 74) vs. regression (N = 20). Field strength/sequence: 3 T, gradient-echo dynamic susceptibility contrast-enhanced MRI, 3D gradient-echo contrast-enhanced T1-weighted imaging. Assessment: To differentiate between radiation changes and progression in SE evaluation, multivariable logistic regression was performed using significant variables among SE appearance interval, IDH mutation, morphological features, and rCBV. Cox regression was performed to predict the tumor progression. For the added value of the rCBV, a log-rank test was conducted between the multivariable logistic regression models with and without the rCBV. Statistical tests: Logistic regression, Cox regression, receiver operating characteristic analysis, log-rank test. Results: 38.3% (36/94) patients had first specific SE (9.2 ± 9.5 months after surgery), which disappeared in 21.3% (20/94) after 5.8 ± 5.8 months after initial appearance on post-radiation MRI. IDH mutation, elongated, small lesions with lower rCBV tended to regress: IDH mutation, elongation, diameter, and rCBV_p95; odds ratio, 0.32, 1.92, 1.70, and 2.47, respectively. Qualitative evaluation of shape revealed that thin and curvilinear-shaped SE tended to regress, indicating a significant correlation with quantitative shape features (r = 0.31). In Cox regression, rCBV and lesion shape were significant (hazard ratio = 1.09 and 0.54, respectively). For sub-centimeter lesions, the rCBV showed added value in predicting outcomes (area under the curve, 0.873 vs. 0.836; log-rank test). Data conclusion: Smaller, elongated lesions with lower rCBV and IDH mutation are associated with regression when differentiating radiation changes from progression in high-grade glioma with post-radiotherapy SE. Evidence level: 3 TECHNICAL EFFICACY: Stage 2.