Background Anti-N-methyl-d-aspartate (NMDA) receptor encephalitis is a severe but treatable autoimmune disorder which diagnosis depends on sensitive and specific antibody testing. We aimed to assess the sensitivity and specificity of serum and CSF antibody testing in patients with anti-NMDA receptor encephalitis, and the relation between titres, relapses, outcome, and epitope repertoire. Methods In this observational study, we used rat brain immunohistochemistry and cell-based assays (CBA) with fixed or live NMDA receptor-expressing cells to determine the sensitivity and specificity of antibody testing in paired serum and CSF samples. Samples were obtained at diagnosis from patients with anti-NMDA receptor encephalitis and from control participants worldwide. We deemed a patient to be antibody positive if their serum, their CSF, or both tested positive with both immunohistochemistry and CBA techniques; we determined titres with serial sample dilution using brain immunohistochemistry. We examined samples from 45 patients (25 with good outcome [modified Rankin Scale, mRS 0–2], ten with poor outcome [mRS 3–6], and ten with relapses) at three or more timepoints. We determined the epitope repertoire in the samples of 23 patients with CBA expressing GluN1-NMDA receptor mutants. Findings We analysed samples from 250 patients with anti-NMDA receptor encephalitis and 100 control participants. All 250 patients had NMDA receptor antibodies in CSF but only 214 had antibodies in serum (sensitivity 100·0% [98·5–100·0%] vs 85·6% [80·7–89·4%], p<0·0001). Serum immunohistochemistry testing was more often in agreement with CBA with fixed cells (77 [71%] of 108) than with CBA with live cells (63 [58%] of 108, p=0·0056). In multivariable analysis, CSF and serum titres were higher in patients with poor outcome than in those with good outcome (CSF dilution 340 vs 129, difference 211, [95% CI 1–421], p=0·049; serum dilution 7370 vs 1243, difference 6127 [2369–9885], p=0·0025), and in patients with teratoma than in those without teratoma (CSF 395 vs 110, difference 285 [134–437], p=0·0079; serum 5515 vs 1644, difference 3870 [548–7193], p=0·024). Over time there was a decrease of antibody titres in the 35 patients with good or poor outcome and samples followed at three timepoints regardless of outcome (from diagnosis to last follow-up: CSF 614 to 76, difference 538 [288–788]; serum 5460 to 1564, difference 3896 [2428–5362]; both p<0·0001). Relapses were associated with a change in titre more often in CSF than in serum (14 of 19 vs seven of 16, p=0·037). After recovery, 24 of 28 CSF samples and 17 of 23 serum samples from patients remained antibody positive. Patients' antibodies targeted a main epitope region at GluN1 aminoacid 369; the epitope repertoire did not differ between patients with different outcomes, and did not change during relapses. Interpretation The sensitivity of NMDA receptor antibody testing is higher in CSF than in serum. Antibody titres in CSF and serum were higher in patients with poor outcome or teratoma than in patients with good outcome or no tumour. The titre change in CSF was more closely related with relapses than was that in serum. These findings emphasise the importance of including CSF in antibody studies, and that antibody titres can complement clinical assessments. Funding Dutch Cancer Society, National Institutes of Health, McKnight Neuroscience of Brain Disorders award, the Fondo de Investigaciones Sanitarias, ErasmusMC fellowship, and Fundació la Marató de TV3.

Antibody titres at diagnosis and during follow-up of anti-NMDA receptor encephalitis: a retrospective study