ACCELERATED BRAIN AGING IN AUTOIMMUNE ENCEPHALITIS BASED ON BRAIN-PREDICTED AGE DIFFERENCE

Autoimmune encephalitis (AE) is a severe neuroimmunological disorder frequently associated with persistent cognitive deficits, especially memory impairment, which are closely linked to functional and structural brain network alterations. However, understanding the relation between cognitive changes and the often complex metrics derived from connectome analyses remains challenging. To address this, brain-predicted age difference (brain-PAD) has emerged as a single, interpretable biomarker that may capture the relationship between cognitive dysfunction and pathological network alterations. We applied DeepBrainNet (Bashyam et al., 2020), a validated CNN trained on a diversified set of MRI scans (N=11,729), to estimate brain age from T1-weighted MRI images. The model was applied to imaging data from 113 AE patients and 305 healthy participants. For each scan, SHAP values were calculated to identify younger and older brain regions on generated voxel maps. This approach provides insights into regional brain aging patterns that may reflect maladaptive or compensatory mechanisms in AE and can be related to neurocognitive performance. Using OLS regression (R²=0.869), we found significantly increased brain-PAD in the AE cohort (BETA=4.0467, p<0.0001) compared to the HC subjects. Higher brain-PAD may indicate more maladaptive and fewer compensatory mechanisms in AE, supporting its potential value as a clinically meaningful and scalable biomarker for both research and clinical applications in autoimmune encephalitis.
Reference. Bashyam, V. M., Erus, G., … Davatzikos, C. (2020). MRI signatures of brain age and disease over the lifespan based on a deep brain network and 14468 individuals worldwide. Brain, 143(7), 2312–2324.