Subgroup comparison according to clinical phenotype and serostatus in autoimmune encephalitis: a multicenter retrospective study

Abstract

Background and purpose Autoimmune encephalitides (AE) include a spectrum of neurological disorders whose diagnosis revolves around the detection of neuronal antibodies (Abs). Consensus‐based diagnostic criteria (AE‐DC) allow clinic‐serological subgrouping of AE, with unclear prognostic implications. The impact of AE‐DC on patients’ management was studied, focusing on the subgroup of Ab‐negative‐AE. Methods This was a retrospective multicenter study on patients fulfilling AE‐DC. All patients underwent Ab testing with commercial cell‐based assays (CBAs) and, when available, in‐house assays (immunohistochemistry, live/fixed CBAs, neuronal cultures) that contributed to defining final categories. Patients were classified as Ab‐positive‐AE [N‐methyl‐d‐aspartate‐receptor encephalitis (NMDAR‐E), Ab‐positive limbic encephalitis (LE), definite‐AE] or Ab‐negative‐AE (Ab‐negative‐LE, probable‐AE, possible‐AE). Results Commercial CBAs detected neuronal Abs in 70/118 (59.3%) patients. Testing 37/48 Ab‐negative cases, in‐house assays identified Abs in 11 patients (29.7%). A hundred and eighteen patients fulfilled the AE‐DC, 81 (68.6%) with Ab‐positive‐AE (Ab‐positive‐LE, 40; NMDAR‐E, 32; definite‐AE, nine) and 37 (31.4%) with Ab‐negative‐AE (Ab‐negative‐LE, 17; probable/possible‐AE, 20). Clinical phenotypes were similar in Ab‐positive‐LE versus Ab‐negative‐LE. Twenty‐four/118 (20.3%) patients had tumors, and 19/118 (16.1%) relapsed, regardless of being Ab‐positive or Ab‐negative. Ab‐positive‐AE patients were treated earlier than Ab‐negative‐AE patients (P = 0.045), responded more frequently to treatments (92.3% vs. 65.6%, P < 0.001) and received second‐line therapies more often (33.3% vs. 10.8%, P = 0.01). Delays in first‐line therapy initiation were associated with poor response (P = 0.022; odds ratio 1.02; confidence interval 1.00–1.04). Conclusions In‐house diagnostics improved Ab detection allowing better patient management but was available in a patient subgroup only, implying possible Ab‐positive‐AE underestimation. Notwithstanding this limitation, our findings suggest that Ab‐negative‐AE and Ab‐positive‐AE patients share similar oncological profiles, warranting appropriate tumor screening. Ab‐negative‐AE patients risk worse responses due to delayed and less aggressive treatments.