White matter abnormalities and multivariate pattern analysis in anti-NMDA receptor encephalitis.

Objective: This study aimed to investigate white matter (WM) microstructural alterations and their relationship correlation with disease severity in anti-NMDA receptor (NMDAR) encephalitis. Multivariate pattern analysis (MVPA) was applied to discriminate between patients and healthy controls and explore potential imaging biomarkers. Methods: Thirty-two patients with anti-NMDAR encephalitis and 26 matched healthy controls underwent diffusion tensor imaging. Tract-based spatial statistics and atlas-based analysis were used to determine WM microstructural alterations between the two groups. MVPA, based on a machine-learning algorithm, was applied to classify patients and healthy controls. Results: Patients exhibited significantly reduced fractional anisotropy in the corpus callosum, fornix, cingulum, anterior limb of the internal capsule, and corona radiata. Moreover, mean diffusivity was increased in the anterior corona radiata and body of the corpus callosum. On the other hand, radial diffusivity was increased in the anterior limb of the internal capsule, cingulum, corpus callosum, corona radiata, and fornix. WM changes in the cingulum, fornix, and retrolenticular part of the internal capsule were correlated with disease severity. The accuracy, sensitivity, and specificity of fractional anisotropy-based classification were each 78.33%, while they were 67.71, 65.83, and 70% for radial diffusivity. Conclusion: Widespread WM lesions were detected in anti-NMDAR encephalitis. The correlation between WM abnormalities and disease severity suggests that these alterations may serve a key role in the pathophysiological mechanisms of anti-NMDAR encephalitis. The combination of tract-based spatial statistics and MVPA may provide more specific and complementary information at the group and individual levels.