2.1 Patients

We have collected detailed information on 37 patients diagnosed with GAD65 neurological autoimmunity, admitted and treated at the Neurology Department of Xuanwu Hospital, Capital Medical University, from June of 2017 to December of 2023; these patients were subsequently followed up. All patients exhibited clinical manifestations indicative of nervous system involvement, tested positive for anti-GAD65 antibodies in blood and/or CSF, and underwent thorough evaluations to exclude other potential etiologies by neurologists. Informed written consent was acquired from patients or their representatives.

We provide comprehensive descriptions of the patients' demographic characteristics, clinical phenotypes, laboratory findings, imaging results, time from symptom onset to initial immunotherapy, and disease outcomes. Disease severity at the nadir after onset and neurological functional status at the last follow-up were assessed using the modified Rankin Scale (mRS), Mini-Mental State Examination (MMSE), and Montreal Cognitive Assessment (MoCA) scales depending on different phenotypes.

2.2 Outcome Measures

Response to immunotherapy (steroids, intravenous immunoglobulin (IVIg), plasma exchange (PLEX), mycophenolate mofetil (MMF), and rituximab) was classified as no response, mild response, partial response (i.e., minimal residual clinical signs/symptoms), or good response (i.e., no residual clinical signs/symptoms). The assessment of treatment response was carried out by two clinicians, F Xu and My Zhang. If there is a disagreement, it will be further evaluated by Zd. Qiu until a consensus is reached.

2.3 Laboratory Investigations

Upon admission, all patients underwent lumbar puncture, with CSF analyzed for white cell count, protein levels, and specific oligoclonal bands (OCB). Routine diagnostic workup for suspected neuroimmune disorders and encephalitis included assessing antibodies associated with autoimmune encephalitis, paraneoplastic neurological syndromes, and CNS demyelinating diseases via CBA in both serum and CSF. Additionally, some patients were concurrently tested using ELISA and immunoblotting methods. Serum antinuclear antibodies (ANA), antineutrophil cytoplasmic antibodies, anticardiolipin antibodies, and rheumatic factor (RA) were tested in all patients to exclude systemic autoimmunity and to seek concomitant autoantibodies. All patients were screened for diabetes and thyroid disease. CSF samples from all patients were also screened for viruses and bacteria to exclude infections in the CNS.

2.4 Imaging

All patients underwent cerebral MRI examinations, encompassing T1-weighted imaging, T2-weighted imaging, FLAIR sequences, diffusion-weighted imaging, and gadolinium-enhanced T1-weighted imaging. Positron emission tomography/computed tomography (PET/CT) scans were conducted for 12 patients. Our analysis focused on lesion localization, signal intensity, metabolic profiles, and relevant imaging parameters.

2.5 Statistical Analysis

This study constituted a follow-up investigation. For variables that disobeyed Gaussian distribution, the median (interquartile, IQR) was present. Numbers of patients in different subgroups were described as values and ratios. The Spearman rank correlation was utilized to identify correlations between time to treatment and immunological treatment response types, with Spearman r > 0.60 considered as strong correlations. All analyses were implemented by using R software, version 4.2.2 (R Foundation for Statistical Computing).

3.1 Demographic and Clinical Characteristics

This study included 37 patients from 16 provinces and 31 cities around China, comprising 14 males (37.8%) and 23 females (62.8%), including 2 minors, with a male-to-female ratio of approximately 1: 2. Epilepsy was the most common phenotype, affecting 15 patients (40.5%), followed by limbic encephalitis (7 cases, 18.9%), overlap syndromes (7 cases, 18.9%), peripheral neuropathy (3 cases, 8.1%), brainstem dysfunction (2 cases, 5.4%), parkinsonism (2 cases, 5.4%), and cerebellar ataxia (1 case, 2.7%). Notably, Epilepsy and limbic encephalitis were predominant among the phenotypes observed in this study, whereas cerebellar ataxia and stiff person syndrome were more common in patients with overlap syndromes. There were isolated cases of brainstem dysfunction, parkinsonism, or peripheral neuropathy, highlighting distinct phenotype distributions compared to previous studies. Median age of symptom onset for all patients was 41.5 (25–59) years, with a median interval of 9 (1.5–36) months from onset to definitive diagnosis. Table 1 summarizes these patient characteristics.

3.2 Laboratory Characteristics

Five patients showed elevated white cell counts in CSF, with a median of 15 (13–33) × 10⁶/L (reference range: 0–8 × 10⁶/L) and mild protein elevation was observed in 5 patients with a median of 53 (49–54) mg/dL (reference range: 15–45 mg/dL). CSF-specific OCB were present in 23 patients (62.2%). Eight patients exhibited elevated 24-h intrathecal IgG synthesis rates with a median of 16.9 (11.2–23.9) mg/24 h (reference range: 0–9 mg/24 h).

Using the CBA method, all 37 patients tested positive for serum anti-GAD65 antibodies with a median titer of 1:320 (1:100–1:320). In CSF, the positivity rate was 91.9%, with a median titer of 1:320 (1:100–1:640). Among them, 6 patients had serum and CSF anti-GAD65 antibody concentrations exceeding 1000 U/mL (reference value: < 5 U/L). Immunoblotting assays confirmed positive results in both serum and CSF for 8 patients. Five patients also tested positive for other neuronal antibodies, including anti-GABABR CV2, SOX1, Hu, Ri, Titin, and amphiphysin antibodies, with 3 cases showing combined positivity for anti-SOX1 antibodies and 3 patients displaying dual or more antibody positivity. Nine patients underwent re-evaluation for anti-GAD65 antibodies at a median interval of 10 (6–19) months after immunotherapy; only 1 patient showed a negative conversion of serum antibody, while the remaining 8 patients retained positivity in serum (median 1:100, 1:100–1:320) and CSF (median 1:100, 1:100–1:320) with stable levels.

Systemic antibody testing revealed positive anti-thyroid peroxidase (TPO) antibodies/anti-thyroglobulin antibodies of 16 patients (43.2%), with both antibodies positive in 10 cases. Based on thyroid function tests, ultrasound findings, and endocrinology consultations, 11 patients (29.7%) were diagnosed with Hashimoto's thyroiditis, including 1 patient diagnosed with Sjögren's syndrome due to longstanding xerostomia, xerophthalmia, and positive anti-nuclear antibodies, anti-SSA antibodies, and anti-Ro-52 antibodies. Seven patients (18.9%) tested positive for serum anti-nuclear antibodies, and 3 patients were positive for anti-Ro-52 antibodies. Three patients (8.1%) were diagnosed with T1DM and received regular subcutaneous insulin therapy.

3.3 Diverse Clinical Presentations of GAD65 Neurological Autoimmunity

In our cohort, epilepsy patients were primarily young and middle-aged, with diverse seizure types. There were significant differences among patients in the median time from onset to definitive diagnosis and initiation of immunotherapy, as well as in their responses to immunotherapy. Some patients experienced significant improvement with steroids or IVIg alone, while a few patients who received multiple treatments only achieved slight improvement. One 62-year-old female patient with limbic encephalitis was found to have anti-GAD65, CV2, and GABABR antibodies in both serum and CSF. This patient was transferred to the Neuro Intensive Care Unit (NICU) due to the severity of symptoms. PET/CT indicated a possible lung tumor, and paraneoplastic limbic encephalitis could not be ruled out. Unfortunately, the patient was lost to follow-up.

In this cohort, two female patients presented exclusively with brainstem dysfunction. One patient experienced dizziness, diplopia, right-sided ptosis, intermittent nausea, and vomiting. Despite initial treatment with intravenous methylprednisolone (IVMP), IVIg, and MMF, which was discontinued due to inadequate symptom relief and intolerable side effects, she eventually became bedridden and lost independence (see Table 2. Patient 1). The other patient started with vertigo and developed persistent diplopia, with physical examination revealing left eye abduction and limited downward gaze. Treatment included IVMP and sequential rituximab (RTX), leading to gradual symptom improvement (see Table 2. Patient 2).

Two male patients in this cohort displayed symptoms of parkinsonism. A 63-year-old male presented with bradykinesia, personality changes over 2 years, dysphagia, and declining memory over 6 months. Brain MRI indicated midbrain atrophy and the “hummingbird sign”, consistent with imaging features of progressive supranuclear palsy (PSP) (see Figure 1C). Treatment involved IVMP for 5 days followed by tapering oral steroid (OS) and concurrent MMF, resulting in significant symptom relief and restored independence (see Table 2. Patient 3). Another patient, a 77-year-old male, experienced 7 years of left limb bradykinesia and clumsiness unresponsive to levodopa. Initial treatment with IVIg provided slight improvement, but subsequent treatment with steroids and MMF was discontinued due to intolerable side effects (hallucinations, increased unsteady gait, and elevated transaminases). Symptomatic stability was maintained with intermittent IVIg without further deterioration (see Table 2. Patient 4 and Figure 1D). Notably, this patient repeated serum GAD65 antibody testing six times over 17 months, consistently showing titers at 1:100.

Four patients in this cohort suffered from peripheral nerve damage: one with subacute sensory neuropathy, one with subacute sensory neuropathy and myelopathy (see Figure 1F), one with Lambert-Eaton myasthenic syndrome (LEMS), and one with peripheral neuropathy combined with HIV infection. PET-CT or SPECT scans in the first three patients suggested the possibility of malignant lesions, accompanied by one or more additional neuronal antibodies. Two were lost to follow-up, one did not receive immunotherapy and died approximately 3 months after discharge, and one patient with HIV infection experienced symptom disappearance after IVIg, supported by effective HIV treatment at a specialist hospital (see Table 3). It should be noted that the titers of GAD65 antibody in serum or CSF of these patients are relatively low compared to patients with other clinical phenotypes, with median titers of 1:64 (1:56–1:73) and 1:64 (1:48–1:64). Unfortunately, pathology data were unavailable for these patients.

Seven patients were diagnosed with overlap syndromes, comprising 18.9% of the cohort and encompassing six overlap patterns. These included combinations such as cerebellar ataxia with limbic encephalitis, stiff person syndrome, and brainstem dysfunction, peripheral neuropathy with myelopathy, as well as epilepsy with stiff person syndrome or brainstem dysfunction.

3.4 Imaging Findings

All patients underwent comprehensive brain MRI evaluations, revealing abnormalities in 21 cases, none of which were gadolinium-enhanced. The hippocampus and/or hippocampal gyrus (71.43%) and temporal lobes (33.33%) were predominantly affected, with involvement also noted in the brainstem, cerebellum, amygdala, insula, basal ganglia, and corona radiata, predominantly showing high signal intensity on T2-weighted imaging. Hippocampal atrophy was most common, some accompanied by brainstem or cerebellar atrophy, with isolated cases exhibiting hippocampal swelling and thickening of the temporal lobe cortex (see Figure 1A). One patient with overlap syndromes of cerebellar ataxia, stiff person syndrome, and brainstem dysfunction had brainstem T2-hyperintensity indicative of brainstem involvement (see Figure 1E).

Twelve patients underwent PET/CT scans, revealing metabolic abnormalities in various brain regions for 10 patients: temporal lobe (80%), frontal lobe (70%), parietal lobe (70%), basal ganglia (40%), occipital lobe (30%), hippocampus (30%), and cerebellum (20%). Among these, 7 patients showed hypometabolism in ≥ 1 brain region, including 2 cases of epilepsy, 2 cases of overlap syndromes, 2 cases of limbic encephalitis, and 1 case of parkinsonism (e.g., Figure 2B,C), while 3 patients exhibited both hypo- and hypermetabolic regions, including 2 cases of limbic encephalitis (see Figure 2A,E) and 1 case of parkinsonism (see Figure 2D). Among the 4 patients with peripheral neuropathy, two with subacute sensory neuropathy showed hypermetabolism regions in the lungs (see Figure 2F), with one also showing hypermetabolism lesions in the gastrointestinal tract, suggesting potential malignant lesions. One patient diagnosed with Lambert-Eaton myasthenic syndrome showed abnormal bone metabolic activity in the right acetabulum and right ischium on single photon emission computed tomography (SPECT), with metastasis not ruled out (see Figure 2G).

3.5 Treatment and Prognosis

Thirty-six patients (97.3%) received immunotherapy within a median time of 8.5 (1.5–37.5) months from symptom onset (see Table 1). Four cases were lost to follow-up, leaving a median follow-up period of 20.5 (16–37.3) months among the remaining 32 patients (see Table 4). Acute-phase treatments involved IVMP, IVIg, and PLEX, with 15 patients (46.9%) receiving IVMP alone, 5 patients (15.6%) receiving IVIg alone, 12 patients (37.5%) receiving combination or sequential treatments of IVMP, IVIg, or PLEX, but only one patient underwent all three treatments sequentially. Maintenance-phase treatments involved MMF or RTX, with 9 patients (28.1%) receiving MMF alone or with OS, 5 patients (15.6%) receiving RTX infusion (periodic single dose of 375 mg/m2 of body surface area; reinfusion will be given when CD19+ B-cell counts of total lymphocytes reach 1%) alone or with OS, and 5 patients (15.6%) initially receiving MMF without marked improvement, converting to RTX infusion, then 3 patients with epilepsy experiencing a reduction in seizure frequency and less severity in seizure types, and one patient with limbic encephalitis achieving complete remission.

Most patients responded favorably to immunotherapy, including good response (n = 8) and partial response (n = 18). Notably, among the 8 patients with good response, including 4 with epilepsy and 3 with limbic encephalitis accompanied by seizures, all received concurrent antiepileptic drug (see Tables 1 and 4). However, 2 patients with epilepsy and parkinsonism, respectively showed only mild response, 4 patients with limbic encephalitis accompanied by seizures, cerebellar ataxia, brainstem dysfunction, and epilepsy respectively, did not respond to immunotherapy, experiencing ongoing symptom deterioration (see Table 1). One patient with epilepsy and another with limbic encephalitis in the aforementioned 6 patients underwent various treatments, including IVMP, MMF, and RTX, without marked improvement, and were seizure-free following epileptogenic focus resection. Spearman's rank correlation was used to assess the relationship between median time from symptom onset to first immunotherapy and response types, but only weak correlation was found, as shown in Figure 3.