To characterize MRI features in anti-NMDA receptor encephalitis (anti-NMDA-Re) and associations with clinical severity and outcomes.

Imaging features and utility of MRI as a biomarker of clinical severity and outcomes in anti-NMDA-Re is incompletely understood.

This is an observational cohort study of adult patients with anti-NMDA-Re (CSF-positive) and early brain MRI scans (≤6 months from symptom onset). Clinical characteristics, including clinical assessment scale in autoimmune encephalitis (CASE) and modified Rankin Scale (mRS), were obtained at first hospital admission and last follow-up. T-tests compared clinical and neuroimaging features.

Forty-nine patients with anti-NMDA-Re were identified. Thirty-two adult patients with available early MRI were included (mean age=34 years [standard deviation (SD)=15 years]; 70% female), including 5 with para/post-herpes simplex virus encephalitis (HSVe). MRI was performed at a median of 0 days (interquartile range [IQR]= -31 to 10 days) from hospitalization. Median follow-up was 1.9 years (IQR=0.5-2.8 years).

MRI was abnormal in 14 patients (44%). Eleven (34%) demonstrated fluid-attenuated inversion recovery (FLAIR) hyperintensities- 6 (19%) involved meninges, 6 (19%) hippocampi, and 5 (16%) other temporal lobe structures (including 4 with para/post-HSVe). Six (19%) had contrast-enhancement (parenchymal: 5 [16%], meningeal: 1 [3%]). Five (16%) demonstrated diffusion restriction.

Cross-sectionally, abnormal MRI did not associate with disease severity/disability (p>0.05).  At last follow-up, patients with initial abnormal MRI had less improvement in their mRS (mean [SD]= -0.45 [1.92] vs -2.23 [1.42], p=0.02) and CASE scores (0.0 [7.50] vs -7.85 [7.09], p=0.02). When patients with HSVe were removed, this association was no longer significant.

This study demonstrates 1) Initial MRI in anti-NMDA-Re is often normal; few demonstrate meningeal FLAIR hyperintensity, while temporal T2-hyperintensity was mainly seen in patients with para/post-HSVe; 2) Initial MRI was not associated with disease severity/disability after excluding patients with para/post-HSVe, potentially due to reversible neuronal dysfunction that is not necessarily destructive, distinct from para/post-HSVe.