Prevalence and clinical correlates of autoimmune and infection-related antibodies in unexplained seizures in western China

Abstract

Background

Unexplained seizures represent a substantial diagnostic and therapeutic challenge in clinical neurology. Immune-mediated mechanisms coupled with prior infections are increasingly recognized as pivotal etiological contributors, yet large-scale epidemiological data from western China remain scarce.

Methods

This retrospective cohort study enrolled 471 patients with unexplained seizures from eight provinces in western China between January 2017 and June 2022. Serum and cerebrospinal fluid samples were tested for neural-specific autoantibodies, thyroid peroxidase antibody (TPO-Ab), thyroglobulin antibody (TgAb), antinuclear antibody (ANA), and immunoglobulin M/G antibodies against Toxoplasma gondii, rubella virus, cytomegalovirus, and herpes simplex virus. Patients were stratified into definite autoimmune-associated seizures (AAS), probable AAS, and non-AAS subgroups.

Trial registration

This trial was registered with the Chinese Clinical Trial Registry (ChiCTR1900027074) on October 30, 2019, and was retrospectively registered as the first participant was enrolled in January 2017.

Results

Overall, 14.65% of patients tested positive for neural-specific autoantibodies, with N-methyl-D-aspartate receptor antibody (NMDAR-Ab) being the most prevalent subtype (accounting for 46.38% of antibody-positive patients), and leucine-rich glioma-inactivated 1 antibody (LGI1-Ab) predominantly detected in patients with recurrent seizures. The positive rates of TPO-Ab, TgAb, and ANA were significantly higher in the definite AAS group than in the non-AAS group (all P < 0.01). No significant differences in pathogen-specific IgM levels were observed among subgroups, while serum HSV-IgG and T. gondii-IgG levels were markedly lower in the non-AAS group (both P < 0.0001). CSF white blood cell count was significantly elevated in the definite AAS group (P = 0.036).

Conclusion

Combined detection of neural autoantibodies, systemic autoimmune antibodies, and infection-related serological markers facilitates etiological diagnosis and identification of candidates for early immunotherapy in patients with unexplained seizures in western China.

Background

Epilepsy affects approximately 0.5% −1.0% of the global population, and seizures of unknown etiology (cryptogenic seizures) account for a substantial proportion of cases, with nearly half occurring in adult patients [1–3]. In 2017, the International League Against Epilepsy (ILAE) formally classified immune etiology as one of the six major etiological categories of epilepsy (alongside structural, genetic, infectious, metabolic, and unknown), revolutionizing the diagnostic and therapeutic paradigm for cryptogenic seizures [4–6]. To resolve long-standing conceptual ambiguity, the ILAE Task Force on Autoimmunity and Inflammation proposed standardized definitions in 2020 and further refined the terminology in 2023: acute symptomatic seizures secondary to autoimmune encephalitis (ASSAE) refers to seizures occurring during the active phase of autoimmune encephalitis, whereas autoimmune encephalitis-associated epilepsy (AEAE) describes chronic, recurrent seizures that persist beyond the acute inflammatory phase [7–10]. This taxonomic clarification underscores the critical need to identify immune-mediated mechanisms in patients with otherwise unexplained seizures [11].

Mounting clinical evidence indicates that neural-specific autoantibodies are core biomarkers of immune-mediated seizures. Prospective and retrospective studies from Europe and the United States report that 10%–20% of patients with cryptogenic seizures harbor pathogenic neural autoantibodies, with rates reaching approximately 15% in large cohort analyses [9, 10, 12–15]. These antibodies are stratified into two functionally distinct classes based on their subcellular targets, which directly determine pathogenicity, treatment response, and prognosis. Antibodies against neuronal cell-surface antigens (e.g. NMDAR, GABABR, GABAAR, LGI1, CASPR2, AMPAR) are directly pathogenic, act via receptor internalization or functional blockade, and are highly responsive to first-line immunotherapy. In contrast, antibodies against intracellular or paraneoplastic antigens (e.g., GAD65, ANNA-1/Hu, ANNA-2/Ri, PCA-1/Yo) are primarily epiphenomena of cytotoxic T-cell-mediated neuronal injury, often associated with occult malignancies, and show limited response to immunotherapy. Among these, antibodies targeting NMDAR are the most frequently identified in patients with new-onset immune-mediated seizures, while LGI1 antibodies are more commonly associated with chronic recurrent seizures [8, 12, 13, 16–20]. Despite these advances, population-based data on the prevalence and spectrum of neural autoantibodies in Chinese patients with cryptogenic seizures remain scarce, particularly in understudied regions such as western China [21].

Beyond neural-specific autoantibodies, systemic autoimmune markers have emerged as important auxiliary diagnostic indicators. TPO-Ab, TgAb, and ANA are frequently elevated in patients with suspected immune-mediated seizures, and are linked to other autoimmune-mediated neurological conditions manifesting with status epilepticus or recurrent seizures [12, 22–26]. Recent studies confirm that coexistence of thyroid autoantibodies or ANA with neural-specific antibodies is more prevalent in patients with drug-resistant cryptogenic seizures, supporting the integration of these markers into routine diagnostic panels for suspected autoimmune epilepsy [27, 28]. However, data on the co-occurrence patterns of central and systemic autoantibodies in Chinese cohorts remain limited, hindering comprehensive etiological evaluation.

Infectious triggers are increasingly recognized as key initiators of autoimmune-mediated central nervous system (CNS) disorders, providing a mechanistic link between prior infection and subsequent autoimmunity [29–31]. Viral pathogens—most notably herpes simplex virus (HSV)—can disrupt the blood–brain barrier, induce neuroinflammation, and trigger autoantibody production via molecular mimicry or epitope spreading, leading to post-infectious autoimmune encephalitis and seizures [29, 32]. Emerging evidence also suggests a potential association between Toxoplasma gondii infection and immune-mediated seizures, though the underlying mechanisms and clinical relevance remain poorly defined. While IgM antibodies indicate acute infection, IgG antibodies reflect past exposure and may serve as epidemiological markers of infectious triggers for autoimmunity [33, 34]. To date, few studies have systematically evaluated the correlation between TORCH-related infections (including HSV, cytomegalovirus (CMV), RV, and T. gondii) and neural autoantibody positivity in patients with cryptogenic seizures in China.

Collectively, the low recognition of immune etiology, limited access to standardized autoantibody testing, and lack of regional epidemiological data contribute to delayed diagnosis and inappropriate treatment for patients with cryptogenic seizures in western China. To address this gap, we conducted a retrospective cohort study enrolling patients with seizures of unknown etiology from eight provinces in western China. The primary objectives were: (1) to determine the prevalence and subtype distribution of neural-specific autoantibodies in serum and CSF; (2) to analyze the co-prevalence of thyroid autoantibodies, ANA, and other systemic autoimmune markers; (3) to investigate the association between prior TORCH infections and autoimmune-associated seizures; and (4) to identify clinical and laboratory characteristics that may facilitate early recognition of immune-mediated seizures.

Patients and methods

Study design and patient selection

This retrospective observational study was carried out at West China Hospital of Sichuan University, enrolling participants from eight provincial regions in western China between January 1, 2017, and June 30, 2022. Although this is a retrospective study, it was registered with the Chinese Clinical Trial Registry (ChiCTR1900027074) prior to data analysis as required by our institutional policy for all human subjects research, to ensure transparency and quality assurance.Eligible participants were diagnosed with epileptic seizures in line with the 2017 ILAE classification criteria for seizure types and epilepsy syndromes, confirmed to have seizures of unknown origin after complete etiological screening including imaging, biochemical and infectious workups, and had paired serum and CSF specimens available for autoantibody testing. Individuals were excluded if they had space-occupying lesions or malignant tumors, CNS parasitic infections, post-traumatic brain injury, infectious or neurosyphilitic etiologies, psychogenic non-epileptic seizures, toxic-metabolic disturbances, delirium or other definable secondary causes. Demographic and clinical data including onset age, gender, seizure onset pattern (new-onset or recurrent seizures), past medical history, routine CSF parameters, brain MRI and EEG manifestations, immunotherapy administration and clinical outcomes were extracted from electronic medical records. All patients were categorized into three subgroups based on neural autoantibody results and clinical phenotypes: definite autoimmune-associated seizures (definite AAS), probable autoimmune-associated seizures (probable AAS) and non-autoimmune-associated seizures (non-AAS). This study was approved by the Ethics Committee of West China Hospital of Sichuan University (approval number: ChiCTR1900027074) and conducted in compliance with the Declaration of Helsinki, and the requirement for written informed consent was waived due to the retrospective design of the research.

Diagnostic criteria and definitions

The diagnosis of epilepsy and classification of seizure types were conducted strictly in accordance with the 2017 ILAE guidelines, and seizures of unknown etiology were defined as epileptic seizures with no identifiable structural, infectious, traumatic, neoplastic, toxic or metabolic triggers after standardized clinical and auxiliary examinations. Definite AAS was defined based on the clinical diagnostic framework for autoimmune-mediated neurological disorders, requiring the presence of neural-specific autoantibodies in serum or CSF combined with clinical manifestations consistent with immune-related seizures. Clinical manifestations “matched” with positive neural autoantibodies required at least one of the following: (a) seizure types characteristic of autoimmune encephalitis (faciobrachial dystonic seizures, new-onset refractory status epilepticus, or temporal lobe seizures with autonomic features); (b) presence of other core symptoms of autoimmune encephalitis (altered mental status, psychiatric symptoms, cognitive dysfunction, or movement disorders) within 3 months of seizure onset; or (c) documented response to immunotherapy (if administered). Probable AAS referred to new-onset or recurrent seizures of unknown origin accompanied by at least three of the following indicators: new-onset psychiatric symptoms, cognitive dysfunction, CSF pleocytosis, abnormal signal changes in the medial temporal lobe on MRI, and inadequate response to two or more anti-seizure medications. For probable AAS classification, the five indicators (new-onset psychiatric symptoms, cognitive dysfunction, CSF pleocytosis, abnormal medial temporal lobe MRI signals, inadequate response to ≥ 2 antiseizure medications) were systematically extracted from electronic medical records. The proportions of patients with complete data for these criteria are reported in the results. Non-AAS was defined as the complete absence of clinical features and laboratory findings suggestive of immune-mediated seizure pathogenesis.

Neural-specific autoantibody detection

Paired serum and CSF specimens were collected within 72 h after admission, processed within 2 h and stored at −80 °C until detection, and commercial indirect immunofluorescence assays (IIF) combined with cell-based assays (CBA) (EUROIMMUN, Lübeck, Germany) were applied to detect immunoglobulin G (IgG) class neural-specific autoantibodies following the manufacturer’s standard protocols. The detected targets covered neuronal cell-surface antigens including N-methyl-D-aspartate receptor (NMDAR, GluN1 subunit), γ-aminobutyric acid B receptor (GABABR), α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid receptor (AMPAR), metabotropic glutamate receptor 5 (mGluR5), dopamine 2 receptor, leucine-rich glioma-inactivated 1 (LGI1), contactin-associated protein-like 2 (CASPR2), dipeptidyl-peptidase-like protein 6 (DPPX) and myelin oligodendrocyte glycoprotein (MOG), glial fibrillary acidic protein antibody (GFAP-Ab), as well as intracellular and paraneoplastic antigens including glutamic acid decarboxylase 65 (GAD65), anti-neuronal nuclear antibody 1 (ANNA-1/Hu), anti-neuronal nuclear antibody 2 (ANNA-2/Ri), Purkinje cell cytoplasmic antibody type 1 (PCA-1/Yo), amphiphysin and DNER (Tr). All assay results were independently reviewed and verified by two senior laboratory technicians who were blinded to the patients’ clinical information, and a positive result was determined as a specific immunofluorescence signal exceeding the cutoff titer of 1:10. Other tested neural antibodies including mGluR5, DPPX, MOG, and dopamine 2 receptor were all negative.

Systemic autoantibody testing

Serum levels of TPO-Ab and TgAb were quantified via competitive electrochemiluminescence immunoassay on a Cobas e601 automated analyzer (Roche Diagnostics, Mannheim, Germany), and ANA was detected by indirect immunofluorescence assay using HEp-2 cell substrates (EUROIMMUN). All tests were performed in full compliance with the manufacturer’s operational instructions, and the final results were double-checked by two senior laboratory technicians. The positive cut-off values were set as TPO-Ab > 34 IU/mL, TgAb > 115 IU/mL and ANA titer ≥ 1:100, respectively.

Infection-related antibody testing

Serum IgM and IgG antibodies against Toxoplasma gondii, rubella virus (RV), CMV, and IgM and IgG antibodies against HSV 1/2 were measured using an automated indirect chemiluminescence immunoassay platform (Liaison, DiaSorin, Saluggia, Italy) following the standard operating procedures of the kit. Positive IgM results indicated acute or recent infection, while positive IgG results suggested past infection exposure, and the classification of positive or negative results was based on the cut-off values recommended by the detection kit.

CSF routine examination

CSF samples were obtained via lumbar puncture within 7 days of admission, and immediately sent to the clinical laboratory of West China Hospital for routine analysis within 2 h after collection. Routine detection indicators included white blood cell (WBC) count, total protein concentration, glucose and chloride levels, and CSF pleocytosis was defined as a WBC count greater than 5 cells/μL.

Statistical analysis

Data statistical processing and analysis were completed using SPSS 25.0 for Mac and GraphPad Prism 8.0 software. Continuous variables with non-normal distribution were presented as median (interquartile range, IQR), and categorical variables were presented as case number and percentage (n, %). Between-group comparisons were performed using Kruskal–Wallis H test for continuous variables among three subgroups, and Chi-square (χ²) test or Fisher’s exact test for categorical variables based on data characteristics. Post hoc pairwise comparisons were performed using the Mann–Whitney U test with Bonferroni correction. All statistical tests were two-sided, and P < 0.05 was considered statistically significant, with all analyses conducted by a professional biostatistician blinded to the study grouping information.

Results

A total of 639 consecutive patients with epileptic seizures of unknown etiology admitted to West China Hospital of Sichuan University from January 2017 to June 2022 were initially enrolled in this retrospective study. According to the standardized screening process, 168 patients were excluded due to identifiable secondary etiologies, including space-occupying lesions or neoplasms (n = 62), parasitic infection (n = 11), post-traumatic brain injury (n = 32), infectious or neurosyphilitic etiologies (n = 21), psychogenic nonepileptic seizures (n = 19), and toxic-metabolic disorders or delirium (n = 23). Finally, 471 patients were included in the statistical analysis (Fig. 1). Based on the detection results of neural-specific autoantibodies and clinical phenotypic characteristics of autoimmune-associated seizures, all enrolled patients were assigned to three subgroups: definite autoimmune-associated seizures (definite AAS, n = 69), probable autoimmune-associated seizures (probable AAS, n = 41), and non-autoimmune-associated seizures (non-AAS, n = 262). Among the 41 patients classified as probable AAS, complete data for all five indicators were available in 38 (92.7%); 2 patients lacked MRI and 1 lacked CSF analysis. Missing data were documented but did not affect classification as each patient still met at least three criteria using available information. In addition, 99 patients with negative antibody test results were not included in subgroup analysis due to insufficient auxiliary diagnostic information. Among 69 definite AAS patients, 37 received immunotherapy (53.6%), and 32 did not receive immunotherapy, including 5 patients who died before treatment initiation, 7 patients with metabolic disorders or contraindications to immunotherapy, 11 patients who withdrew informed consent, and 9 patients with other unrecorded reasons. Brain MRI showed abnormal temporal signals in 18.8% of definite AAS patients, while EEG revealed epileptiform discharges in 26.1% of this group; no significant differences were observed in MRI/EEG abnormalities among the three subgroups (all P > 0.05).

Fig. 1.

Overview of the Study

Baseline characteristics of study participants

The baseline demographic and clinical parameters of all patients were comprehensively collected and compared among the three subgroups, and the detailed data were presented in Table 1. The median age of the total cohort was 46 years (IQR: 28–58 years), and the median age of the definite AAS, probable AAS, and non-AAS groups was 42 (IQR: 26–56) years, 45 (IQR: 30–59) years, and 47 (IQR: 29–59) years, respectively, with no statistically significant difference among the three groups (P = 0.124). The overall gender distribution was well-balanced, with 239 males (50.74%) and 232 females (49.26%); the proportion of males in the definite AAS, probable AAS, and non-AAS groups was 52.17%, 48.78%, and 50.76%, respectively, and no significant intergroup difference was detected (P = 0.921). In terms of seizure onset status, 328 patients (69.64%) presented with new-onset epileptic seizures and 143 patients (30.36%) had a history of recurrent epileptic seizures; the distribution of new-onset and recurrent seizures in each subgroup was highly consistent, and no significant difference was observed among the three groups (P = 0.786). For CSF-WBC count, the median level of the total cohort was 6/μL (IQR: 3–11/μL), and the definite AAS group showed a significantly higher median CSF-WBC count (8/μL, IQR: 4–15/μL) than the probable AAS group (6/μL, IQR: 3–12/μL) and non-AAS group (5/μL, IQR: 2–9/μL), with a statistically significant intergroup difference (P = 0.036).

Table 1.

Baseline characteristics of patients with seizures of unknown etiology stratified by autoimmune-associated seizure subgroups

Continuous variables were analyzed by Kruskal–Wallis H test; categorical variables were analyzed by Chi-square test

Abbreviations: AAS autoimmune-associated seizures, IQR interquartile range, CSF-WBC cerebrospinal fluid white blood cell

Prevalence and spectrum of neural-specific autoantibodies

The prevalence and subtype distribution of neural-specific and glial autoantibodies in paired serum and CSF samples were systematically analyzed, and the results were displayed in Fig. 2A. Among the 471 enrolled patients with seizures of unknown etiology, 69 patients were tested positive for neural-specific or glial autoantibodies, with an overall positive rate of 14.65% (95%CI: 11.62%–17.68%). A total of 7 subtypes of autoantibodies were detected in this study, among which NMDAR-Ab was the most predominant subtype (46.38%, 32/69), followed by GABABR-Ab (20.29%, 14/69), LGI1-Ab (14.49%, 10/69), GFAP-Ab (10.14%, 7/69), Yo-Ab (4.34%, 3/69), CASPR2-Ab (2.90%, 2/69), and Hu-Ab (1.45%, 1/69). We further stratified the antibody-positive patients by seizure onset status and found that 49 patients (71.02%) had new-onset seizures and 20 patients (28.98%) had recurrent seizures (Fig. 2B). NMDAR-Ab, GABABR-Ab, GFAP-Ab, CASPR2-Ab, Hu-Ab, and Yo-Ab were mainly distributed in patients with new-onset seizures, while LGI1-Ab was significantly concentrated in patients with a history of recurrent epileptic seizures, showing a distinct distribution characteristic between different seizure onset patterns.

Fig. 2.

A The chart indicates the distribution of different neuronal autoantibodies in patients with positive results. B The figure shows the distribution of autoantibodies among new-onset seizure patients (green) and recurrent seizure patients (yellow)

We further analyzed the distribution of positive antibody findings by sample type (serum only, CSF only, or both). Among the 69 patients with neural-specific autoantibodies, 31 patients (44.9%) had detectable antibodies only in serum, 24 patients (34.8%) only in CSF, and 14 patients (20.3%) had positivity in both compartments. For NMDAR-Ab, the most common subtype, CSF-only positivity was observed in 18 of 32 patients (56.3%), whereas serum-only positivity accounted for 10 patients (31.3%), and dual positivity for 4 patients (12.5%). In contrast, LGI1-Ab was predominantly detected in serum only (7/10, 70.0%), with 2 patients (20.0%) showing dual positivity and 1 patient (10.0%) CSF-only. GFAP-Ab showed a mixed distribution, with 3 patients serum-only, 3 patients CSF-only, and 1 patient dual positive.

Coexistence of neural autoantibodies with systemic autoantibodies

The co-occurrence of neural-specific autoantibodies and systemic autoimmune antibodies was further evaluated in all three subgroups, and the comparison results were shown in Table 2, Fig. 3C and E. In the definite AAS group, 21 patients (30.44%) exhibited concurrent positivity of multiple autoantibodies in serum, suggesting a prominent systemic autoimmune background in these patients. The positive rates of TPO-Ab, TgAb, and ANA in the definite AAS group were 30.43%, 17.39%, and 26.09%, respectively, which were all significantly higher than those in the probable AAS group (12.20%, 7.32%, 9.76%) and non-AAS group (9.16%, 6.11%, 9.16%). Intergroup statistical analysis confirmed that the differences in the positive rates of TPO-Ab, TgAb, and ANA among the three subgroups were statistically significant (all P < 0.01). In addition, stratified analysis by seizure onset status revealed that both TPO-Ab and TgAb were more frequently detected in antibody-positive patients with recurrent seizures than in those with new-onset seizures, indicating a potential correlation between thyroid autoantibodies and chronic recurrent autoimmune-associated seizures.

Table 2.

Comparison of systemic autoantibody positivity among definite AAS, probable AAS and non-AAS subgroups

Categorical variables were analyzed using Chi-square test

Abbreviations: TPO-Ab thyroid peroxidase antibody, Tg-Ab thyroglobulin antibody, ANA antinuclear antibody, AAS autoimmune-associated seizures

Fig. 3.

Other antibody detection results of patients with seizure of unknown etiology. A The quantitative results of IgM-type antibodies against four common pathogens (CMV, HSV, RV, T.gondii). B The quantitative results of IgG-type antibodies against four common pathogens (CMV, HSV, RV, T.gondii). C The quantitative results of antithyroid antibody antibodies in patients. D White blood cell count in cerebrospinal fluid. E Autoantibody positive rate in serum of patients with seizure of unknown etiology. F The previous viral infection pathogens IgG antibody positive rate in serum of patients with seizures of unknown etiology. *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001. AAS: autoimmune associated seizures, CSF: cerebrospinal fluid

Association between infection-related antibodies and autoimmune-associated seizures

Serum IgM and IgG antibodies against T. gondii, RV, CMV, and HSV-1/2 were quantitatively detected to explore the relationship between prior infection and autoimmune-associated seizures, and the detailed data were presented in Table 3, Fig. 3A, B and F. For pathogen-specific IgM antibodies, the median levels of T. gondii-IgM, RV-IgM, CMV-IgM, and HSV-IgM in the definite AAS, probable AAS, and non-AAS groups were all at low baseline levels, and no statistically significant differences were observed among the three subgroups (all P > 0.05), indicating that there was no significant difference in acute infection status among the three subgroups. In contrast, the Kruskal–Wallis test revealed a significant overall difference in HSV-IgG and T. gondii-IgG levels among the three subgroups (P < 0.0001). Post hoc pairwise comparisons revealed that HSV-IgG and T. gondii-IgG levels were significantly higher in both the definite AAS group and the probable AAS group compared to the non-AAS group (both P < 0.0001), whereas no significant difference was observed between the definite AAS and probable AAS groups (Tables 4 and 5). The median levels of RV-IgG and CMV-IgG were comparable among the three subgroups, and no significant intergroup differences were found (both P > 0.05). Qualitative analysis of IgG antibody positive rates further verified the above quantitative results, confirming that prior infection with HSV and T. gondii was closely associated with autoimmune-associated seizures in this study cohort. The positive cutoff for HSV-IgG is > 1.1 IU/mL; the median in definite AAS (98.4 IU/mL) exceeds this threshold, whereas the median in non-AAS (38.2 IU/mL) is also above cutoff but significantly lower. The median T. gondii-IgG concentration in the definite AAS group reached 82.6 IU/mL, far exceeding the positive threshold, while the median level in the non-AAS group was 31.5 IU/mL, still positive but significantly lower than definite AAS group. The proportions of patients exceeding the assay’s positive cutoff for HSV-IgG were 91.3% (63/69) in definite AAS, 85.4% (35/41) in probable AAS, and 52.6% (190/361) in non-AAS (P < 0.001 for intergroup comparison). For T. gondii-IgG, positivity rates were 37.7% (26/69), 29.3% (12/41), and 14.4% (52/361), respectively (P < 0.001).

Table 3.

Comparison of infection-related antibody levels among three subgroups (median, IQR)

Continuous variables were analyzed using Kruskal–Wallis H test

Abbreviations: IgM immunoglobulin M, IgG immunoglobulin G, T. gondii Toxoplasma gondii, RV rubella virus, CMV cytomegalovirus, HSV herpes simplex virus, IQR interquartile range

Table 4.

Pairwise comparisons of T. gondii-IgG levels between the three subgroups (median, IQR)

Pair analyses wre performed using the Mann–Whitney U test

Table 5.

Pairwise comparisons of HSV-IgG levels between the three subgroups (median, IQR)

Pair analyses wre performed using the Mann–Whitney U test

CSF white blood cell count

The CSF-WBC count, an important indicator of central nervous system inflammation, was further compared among the three subgroups (Fig. 3D). The median CSF-WBC count in the definite AAS group was 8/μL (IQR: 4–15/μL), which was significantly higher than the 5/μL (IQR: 2–9/μL) in the non-AAS group. The median CSF-WBC count in the probable AAS group was 6/μL (IQR: 3–12/μL), which was slightly higher than that in the non-AAS group but lower than that in the definite AAS group, and no statistically significant difference was found between the probable AAS group and the non-AAS group (P > 0.05). This result suggests that mild inflammatory changes in the central nervous system are more prominent in patients with definite autoimmune-associated seizures, which is highly consistent with the autoimmune inflammatory pathogenesis of such seizure disorders.

Immunotherapy response and short-term outcomes

Among the 37 definite AAS patients who received immunotherapy (corticosteroids in 28, intravenous immunoglobulin in 6, or both in 3), follow-up data at a median of 12 months (IQR: 6–24 months) were available for 35 patients. Seizure freedom (≥ 6 months without seizure) was achieved in 21 patients (56.8%), a ≥ 50% reduction in seizure frequency in 10 (27.0%), and minimal or no response in 6 (16.2%). Among the 32 untreated definite AAS patients (excluding 5 who died before treatment initiation), seizure freedom was observed in only 6 (18.8%), partial response in 9 (28.1%), and no improvement in 17 (53.1%).

Discussion

This retrospective study investigated the prevalence and clinical correlates of neural-specific autoantibodies, glial autoantibodies, systemic autoimmune markers, and infection-related antibodies in 471 patients with seizures of unknown etiology from eight provinces in western China. The key findings were as follows: first, the overall positivity rate of neural/glial autoantibodies was 14.65%, with NMDAR-Ab being the most prevalent subtype, and LGI1-Ab predominantly detected in patients with recurrent seizures; second, TPO-Ab, TgAb, and ANA were significantly more common in the definite AAS group; third, serum IgG levels against HSV and T. gondii were markedly lower in the non-AAS subgroup, while IgM levels showed no intergroup differences; and fourth, CSF-WBC count was elevated in patients with definite AAS. To our knowledge, this is one of the largest single-center studies focusing on immune-mediated seizures in western China, providing region-specific epidemiological evidence to refine etiological diagnosis and clinical management for cryptogenic seizures.

The 14.65% prevalence of neural/glial autoantibodies in our cohort is highly consistent with previous reports from Europe and the United States, which have documented rates of 10%–20% in patients with cryptogenic seizures. A large cohort study by Dubey et al. reported a neural antibody positivity rate of approximately 15% in adults with epilepsy of unknown etiology, which aligns closely with our observation [12, 13, 15, 30, 35]. In eastern Chinese populations, recent studies have shown similar prevalence, supporting the notion that immune-mediated mechanisms account for a substantial proportion of cryptogenic seizures across Chinese regions [36–38]. The predominance of NMDAR-Ab (46.38%) in our study is in line with global data, as NMDAR encephalitis is the most common autoimmune encephalitis subtype, and seizures are a core clinical manifestation [38, 39]. Notably, LGI1-Ab was significantly enriched in patients with recurrent seizures rather than new-onset events, a phenotype that has been attributed to the chronic, insidious inflammatory process underlying LGI1-associated neurological disorders. This distribution reinforces the ILAE 2023 conceptual distinction between ASSAE and AEAE, where chronic seizure predisposition correlates with specific antibody profiles [9, 20, 40]. GFAP antibodies, detected in 10.14% of our antibody-positive patients, warrant specific consideration. Since GFAP is an intracellular astrocytic marker, its antibodies are thought to arise as an epiphenomenon secondary to prodromal CNS infection or neural injury rather than serving as direct pathogenic factors. Nevertheless, their presence remains a useful diagnostic indicator of prior or ongoing astrocytic damage in the context of autoimmune-associated seizures.

The frequent coexistence of systemic autoimmune markers in neural antibody-positive patients carries important diagnostic implications. In the definite AAS group, 30.43% were positive for TPO-Ab, 17.39% for TgAb, and 26.09% for ANA, all significantly higher than the non-AAS subgroup. Elevated thyroid antibodies and ANA have long been linked to Hashimoto encephalopathy and lupus encephalopathy, both of which frequently present with seizures and status epilepticus. Recent studies have emphasized that these systemic antibodies may serve as surrogate markers of a generalized autoimmune diathesis rather than direct pathogenic factors, supporting the integration of thyroid antibodies and ANA into routine diagnostic panels for suspected immune-mediated seizures [24, 25, 41]. Our findings extend this concept to the western Chinese population, suggesting that a combined screening panel including neural, glial, thyroid, and nuclear antibodies improves diagnostic yield compared with isolated neural antibody testing.

The association between prior infection and autoimmune-associated seizures observed in this study provides mechanistic insights into the pathogenesis of immune-mediated seizures. No significant differences were detected in pathogen-specific IgM levels across subgroups, ruling out acute infection as a confounding factor. In contrast, HSV-IgG and T. gondii-IgG levels were significantly higher in the autoimmune subgroups, consistent with the hypothesis that previous infection triggers autoimmune responses via molecular mimicry, blood–brain barrier disruption, and epitope spreading. HSV may be associated with post-infectious autoimmune encephalitis, with up to 40% of patients developing neural autoantibodies after herpes simplex encephalitis [28–30]. Our study is among the first to report a robust association between T. gondii IgG seropositivity and autoimmune-associated seizures in a large Chinese cohort. Experimental studies have shown that T. gondii has been linked to chronic low-grade neuroinflammation and alters neurotransmitter pathways, creating a pro-epileptic immune microenvironment. These results highlight the value of TORCH screening in the etiological workup of cryptogenic seizures, especially in regions with high prevalence of parasitic and viral infections [33, 34]. It is important to note that causality cannot be inferred from this retrospective seroprevalence study, as we did not adjust for potential confounders such as age, geography, or socioeconomic status, all of which may influence IgG seropositivity.

From a clinical perspective, our findings support a streamlined diagnostic algorithm for seizures of unknown etiology in western China. Routine combined testing of serum and CSF for neural/glial autoantibodies, supplemented by thyroid antibodies, ANA, and TORCH serology, can identify patients with immune-mediated seizures who may benefit from early immunotherapy. Emerging evidence indicates that early initiation of immunotherapy significantly improves seizure control and reduces the risk of drug resistance in autoimmune-associated epilepsy [10, 42, 43]. The elevated CSF-WBC count in the definite AAS subgroup further supports the use of CSF analysis as an auxiliary inflammatory marker, which is especially valuable when antibody results are pending or negative.

Several limitations of this study should be acknowledged. (1) Its retrospective design may have introduced selection bias, and detailed long-term follow-up data on immunotherapy responses were unavailable for many patients. While all antibody-positive patients in our cohort were classified as definite AAS based on current diagnostic criteria, the clinical relevance of neuronal antibodies may vary depending on antibody titer, sample type (serum vs. CSF), and the absence of functional validation assays. Low-titer or serum-only positivity in a subset of patients could represent clinically nonspecific findings rather than true pathogenic drivers; this potential variability should be considered when interpreting the robustness of our diagnostic classification. (2) The antibody panel was restricted to commercially available assays, potentially missing rare or novel autoantibodies (e.g., against seizure-related protein 6, neurexin-3α), and the possibility of undetected antibody-mediated mechanisms cannot be entirely excluded. (3) Although patients were recruited from eight provinces, all were enrolled at a single tertiary center, which may limit generalizability to community settings. (4) In addition, we did not assess human leukocyte antigen genotypes or other genetic factors that influence autoimmune susceptibility, which warrants investigation in future prospective studies. (5) While routine biochemical and basic infection screenings are generally sufficient for most patients, future diagnostic algorithms could incorporate more sensitive technologies such as metagenomic next-generation sequencing (mNGS) or targeted NGS (tNGS) to further strengthen the differential diagnosis of cryptogenic seizures, particularly in cases with mild CSF pleocytosis where occult infectious triggers might be missed. (6) The multidimensional data collected in this study could serve as a foundation for developing a multi-indicator clinical prediction model; future research should explore machine learning algorithms such as Random Forest or XGBoost to assign feature weights and construct a practical scoring system for predicting definite autoimmune-associated seizures in real-world clinical settings.

Conclusions

In conclusion, neural/glial autoantibodies are detectable in approximately 14.65% of patients with seizures of unknown etiology in western China, with NMDAR-Ab being the most common subtype. Thyroid autoantibodies and ANA are frequently co-expressed and may serve as auxiliary diagnostic markers. Previous infection with HSV and T. gondii is associated with an increased likelihood of autoimmune-associated seizures, supporting an infection-triggered pathogenic pathway. These findings underscore the importance of comprehensive autoimmune and infectious screening in cryptogenic seizures to enable early immunotherapy and improve clinical outcomes. Further large-scale, prospective multicenter studies are needed to validate these results, explore novel autoantibodies, and refine diagnostic scoring systems tailored to the Chinese population.

Authors’ contributions

Yushan Yuan collected and organized data, performed data analysis, and wrote the main manuscript text; Yang Su collected and organized data, performed statistical analysis, prepared figures and tables, and wrote portions of the manuscript; Minjin Wang designed the study, revised the manuscript, and provided critical review of intellectual content; Yong He designed the study, revised the manuscript, and supervised the overall study.

Funding

No Funding.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of West China Hospital of Sichuan University (approval number: ChiCTR1900027074), and was conducted in compliance with the Declaration of Helsinki, and the requirement for written informed consent was waived due to the retrospective design of the research. Patient data were anonymized prior to analysis by removing all direct identifiers (names, medical record numbers, and contact information) and assigning unique study codes.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.