This retrospective, multicenter cohort study enrolled consecutive patients newly diagnosed with AE between January 2018 and October 2022 at four tertiary medical centers in China: the First Affiliated Hospital of Henan University of Science and Technology, the People’s Hospital of Gaozhou, Beijing Tiantan Hospital, and Tianjin Medical University General Hospital. The study was approved by the Ethics Committee of The First Affiliated Hospital, and College of Clinical Medicine of Henan University of Science and Technology, and conducted in accordance with the ethical principles outlined in the Declaration of Helsinki, and all patients provided informed consent and approved participation (No. 2022-09-B067).

Inclusion criteria were: 1) over 18 years old; 2) meeting the diagnostic criteria for AE based on clinical signs and symptoms (15); 3) positive AE-related antibodies in serum and/or cerebrospinal fluid (CSF) tests; 4) complete baseline and clinical information; 5) informed consent for data collection and analysis. The exclusion criteria included: 1) pregnancy or lactation; 2) complicated with other acute neurological diseases; 3) with significant disability (modified Rankin scale[mRS] score ≥ 1) before; 4) active systemic infection, malignancy, or significant hematological disorders; 5) prior immunotherapy (corticosteroids, IVIG, plasmapheresis, rituximab, cyclophosphamide, etc.) within 4 weeks before admission.

A standardized data extraction form was developed to collect data from electronic medical records, including demographics, clinical manifestations, laboratory results (serum/CSF), mRS scores, clinical assessment scale in autoimmune encephalitis (CASE) scores, treatment, and prognosis. Certified neurologists conducted the diagnosis and treatment of all AE patients in accordance with the Chinese Expert Consensus on the Diagnosis and Management of Autoimmune Encephalitis (15).

Autoantibodies associated with AE were detected in both serum and CSF samples using a cell-based assay (CBA) with fixed transfected cells. All assays were performed at the KingMed Center for Clinical Laboratory (Guangzhou, China). Semi-quantitative titers were categorized as negative (<1:10), weakly positive (1:10-1:32), moderately positive (1:100-1:320), or strongly positive (≥1:1000) (16). ANA were detected by indirect immunofluorescence on HEp-2 cells (Euroimmun AG, Lübeck, Germany), with a positive cutoff set at >1:80. Anti-double-stranded DNA (anti-dsDNA) antibodies were also determined using indirect immunofluorescence (Euroimmun AG, Lübeck, Germany), with a positive cutoff set at >1:10. ENA, including anti-Smith antibody (Sm), Anti-Sjogren’s-Syndrome-related antigen A (SSA), Anti-Sjogren’s-Syndrome-related antigen B (SSB), Anti-Ro52 antibodies (Ro52), anti-ribonucleoprotein antibody (RNP), anti-histidyl-tRNA synthetase antibody (Jo-1), and anti-scleroderma 70 antibody (Scl-70), were detected by an immunoblotting assay (Euroimmun AG, Lübeck, Germany). The same platform was used to detect anti-mitochondrial antibody (AMA), Anti-polymyositis and scleroderma antibodies (PM-Scl), anti-histone, and anti-CENP-B antibodies. Anti-thyroglobulin (TG-Ab) and anti-thyroid peroxidase (TPO-Ab) antibodies were quantified using chemiluminescent immunoassays (Beckman Coulter, USA), with positive thresholds defined as > 12 IU/mL and > 60 IU/mL, respectively. Anticardiolipin antibodies (aCL) of IgG and IgM isotypes were measured by enzyme-linked immunosorbent assay (ELISA) (Euroimmun AG, Lübeck, Germany). Anti-neutrophil cytoplasmic antibodies (ANCA) were initially screened using indirect immunofluorescence, and specific reactivities to myeloperoxidase (MPO) and proteinase 3 (PR3) were subsequently confirmed with ELISA kits from Euroimmun AG. All laboratory procedures were rigorously conducted in accordance with the manufacturers’ standardized protocols. SAOS-Ab positivity was defined as the presence of at least one of the above autoantibodies meeting its respective clinical significance threshold. Based on this definition, patients were stratified into SAOS-Ab-positive and SAOS-Ab-negative groups.

The mRS and the CASE were pre-specified as primary clinical outcome measures owing to their established relevance in evaluating disease severity and predicting patient outcomes. The CASE score was used to assess the disease severity on admission. The total score of the CASE is 27 points. It consists of nine items: seizures, memory dysfunction, psychiatric symptoms, consciousness impairment, language problems, dyskinesia, gait instability and ataxia, brainstem dysfunction, and weakness (17). The mRS, which ranges from 0 (asymptomatic) to 6 (death), was employed to assess functional disability and to evaluate the outcome at discharge. Patient outcomes were assessed at hospital discharge using the mRS, with a score of ≥ 3 defined as a poor outcome and a score of < 3 as a good outcome (18). Admission to the intensive care unit (ICU) was deemed necessary if patients exhibited one or more of the following conditions: status epilepticus, delirium, coma, respiratory failure, severe sepsis, single or multiple organ failure, requirement for mechanical ventilation or vasopressor support, elevated APACHE II score, or other critical complications such as post-resuscitation status or increased intracranial pressure. The treatment for the enrolled patients encompassed four principal intervention strategies: anti-inflammatory (corticosteroids), IVIG, B cell targeted therapy (rituximab), plasmapheresis, and broad-spectrum immunosuppressants (including cyclophosphamide, cyclosporine, tacrolimus, and mycophenolate mofetil). All patients were followed up via telephone or interview for 24 months by independently blinded neurologists. A relapse was defined as the onset of new or recurrent neurological symptoms consistent with AE, not attributable to other causes, which led to a medical consultation resulting in a formal diagnosis in an outpatient or inpatient setting.

Continuous variables’ differences between groups were analyzed using the Mann-Whitney U test for nonparametric parameters. Pearson’s chi-square test and Fisher’s exact test were employed to examine the differences in categorical variables’ proportions between groups. The Wilcoxon Rank Sum Test was utilized to investigate ranked data. Candidate predictors of recurrence were first evaluated using univariate Cox regression analysis, and factors significant in the univariate analysis were entered into a multivariable Cox regression model. The forward LR (forward stepwise regression based on maximum likelihood estimate) method was used to perform multivariable Cox regression analysis. Statistical analyses were performed using SPSS Statistics for Windows version 22, and figures were generated using GraphPad Prism 9.4. p-values less than 0.05 were considered indicative of statistical significance.

A total of 218 patients were enrolled in the study and 25 were excluded (Figure 1). The prevalence of SAOS-Ab seropositivity was 57.3% (125/218). The demographic and basic clinical characteristics of the two groups are summarized in Table 1. There was no significant difference in the median age at onset between the SAOS-Ab-positive (39, IQR: 21-56) and negative (40, IQR: 24-57) groups (p = 0.578). However, a significant gender disparity was observed (p = 0.004), with a higher proportion of females in the SAOS-Ab-positive group (55.2% vs. 35.5%). The distribution of specific neuronal autoantibody subtypes (e.g., anti-NMDAR, anti-LGI1) did not differ significantly between the two groups (p > 0.05 for all) (Table 2), indicating that the SAOS-Ab phenomenon is not restricted to a specific AE subtype. The spectrum of detected SAOS-Ab is detailed in Table 1. Among the 125 positive patients, the most frequently detected autoantibodies were TPO-Ab (56.0%), ANA (47.2%), TG-Ab (46.4%), and Ro52 (43.2%). A substantial proportion of patients (77.6%) had non-organ-specific antibodies, while 66.4% had organ-specific antibodies, with a significant overlap between the two groups. No significant differences were observed in the prevalence of comorbidities between the two groups, including coronary heart disease (CHD), hypertension, diabetes, hyperlipidemia, and chronic obstructive pulmonary disease (COPD) (Table 1).

Comprehensive immunological profiling revealed significant differences between the groups. Patients in the SAOS-Ab-positive group exhibited markedly higher levels of the proportion of CD19⁺ B lymphocytes (17.66% vs. 13.79%, p = 0.0027), IgG (median 14.96 g/L vs. 8.11 g/L, p < 0.0001) and IgM (median 1.25 g/L vs. 0.87 g/L, p = 0.0334) in peripheral blood (Figures 2A–C). In contrast, no significant differences were observed in CSF IgG, serum IgA and serum IgE levels (p > 0.05 for all) (Figures 2D–F). Levels of systemic inflammatory markers, including tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-1beta (IL-1β), and C-reactive protein (CRP), were also comparable between groups (Table 1). Notably, while serum titers of the primary neuronal autoantibodies were similar, CSF titers were significantly higher in the SAOS-Ab-positive cohort (p = 0.034) (Table 2), suggesting a possible link between systemic immune activation and intrathecal antibody production.

Disease severity at onset was significantly greater in SAOS-Ab-positive patients, as reflected by higher CASE scores (p < 0.0001) (Figure 3A). A greater proportion of these patients were classified as having severe disease (24.0% vs. 8.6%), and a smaller proportion had mild disease (21.6% vs. 57.0%) compared to the SAOS-Ab-negative group (Figure 3A). This increased severity was reflected in the clinical symptoms. SAOS-Ab-positive patients had a significantly higher CASE scores in weakness (28.0% vs 10.8%; p = 0.005), brainstem dysfunction (47.2% vs 20.4%; p = 0.017), gait instability and ataxia (40.0% vs 24.7%; p = 0.019), dyskinesia (71.2% vs 57.0%; p = 0.025), language problems (50.4% vs 32.3%; p = 0.004), cognitive impairment (32.9% vs 16.1%; p = 0.004), psychiatric symptoms (73.6% vs 38.7%; p < 0.001), memory dysfunction (76.8% vs 25.6%; p < 0.001), seizures (79.2% vs 35.5%; p = 0.004) (Figure 3B). Furthermore, a significantly higher rate of ICU admission was observed among SAOS-Ab-positive patients at the time of hospital admission(21.6% vs 7.5%; p = 0.005) (Table 1).

Treatment patterns differed between the groups. SAOS-Ab-positive patients received first-line corticosteroids (89.6% vs. 72.0%, p = 0.002) and second-line B cell targeted therapy (17.6% vs. 5.3%, p = 0.006) more frequently (Table 1). At discharge, the functional outcome was significantly worse in SAOS-Ab-positive patients, as reflected by higher mRS scores (p < 0.0001) (Figure 3C). A poor outcome (mRS 3-6) was observed in 47.2% (59/125) of SAOS-Ab-positive patients compared to 26.9% (25/93) of SAOS-Ab-negative patients (p = 0.002) (Table 1). Univariable binary logistic regression analysis identified a significant association between SAOS-Ab positivity and poor outcome (OR: 2.427, 95% CI: 1.36-4.331, p = 0.003), with specific antibodies including ANA (OR: 2.998, 95% CI: 1.617-5.555, p < 0.001), Ro52 (OR: 1.893, 95% CI: 1.014-3.533, p = 0.045), and AMA (OR: 2.637, 95% CI: 1.291-5.388, p = 0.008) also demonstrating significant associations. Among organ-specific autoantibodies, both antithyroid antibodies (OR: 2.061, 95% CI: 1.167-3.638, p = 0.013) and TPO antibodies (OR: 2.042, 95% CI: 1.141-3.654, p = 0.016) were significantly associated with poor outcomes (Figure 4).

Kaplan-Meier analysis for relapse risk was performed in AE patients starting from the time of discharge. During a median follow-up of 18.2 months (IQR: 5.7-24.0), 4 patients (3.20%) were lost to follow-up in the SAOS-Ab-positive group, compared to 2 patients (2.15%) in the SAOS-Ab-negative group (Table 1). This attrition rate of below 10% in both cohorts is unlikely to have introduced significant bias into the comparative analysis. Among those patients, 48 had experienced relapse, while 164 had achieved remission (77.4%). Compared to the SAOS-Ab-negative group, the relapse rate was significantly higher in the SAOS-Ab-positive group (29.8% vs. 13.2%, p = 0.0043) (Table 1).

Candidate predictors of recurrence were initially evaluated using univariate Cox proportional hazards regression analysis, including age at onset, gender, CASE scores, mRS scores, ICU admission, and treatment regimens. A significantly elevated relapse risk was observed in the SAOS-Ab-positive group (HR: 2.439, 95% CI: 1.269-4.690, p = 0.008) and among patients who did not receive immunosuppressant therapy (HR: 0.336, 95% CI: 0.157-0.7195, p = 0.005). In contrast, no significant associations were found for corticosteroids (p = 0.057), IVIG (p = 0.302), B cell targeted therapy (p = 0.748), or plasmapheresis (p = 0.642). Variables significant in the univariate analysis were entered into a multivariable Cox regression model (forward likelihood ratio method). In the multivariable Cox regression model, after adjusting for established risk factors, positivity for SAOS-Ab remained an independent risk factor for relapse (adjusted HR: 2.270, 95% CI: 1.174-4.387, p = 0.015) (Figure 5).