Ζεκερίδου, Αναστασία

Gephyrin is a postsynaptic scaffold protein essential for inhibitory neurotransmission. Gephyrin‐immunoglobulin G (IgG) was reported, decades ago, in a single case of paraneoplastic stiff‐person‐like syndrome, but its broader clinical relevance remains unknown. While investigating unclassified synaptic IgGs mimicking amphiphysin‐IgG, we identified gephyrin‐IgG in 3 patients using human protein microarray. Gephyrin specificity was confirmed using confocal tissue‐based immunofluorescence assay, cell‐based assay, and Western blot. The 3 patients, all female, had: encephalitis (2) and myeloneuropathy and breast cancer (1). All had inflammatory cerebrospinal fluid and abnormal magnetic resonance imaging, and improved following immunotherapy. Gephyrin‐IgG is associated with a treatable autoimmune neurological disease, sometimes paraneoplastic. ANN NEUROL 2026

Abstract Severe neuropathies with predominant involvement of motor fibres can resemble lower motor neuron disease (LMND) phenotypes. Given the fatal prognosis of LMND, identifying underlying autoimmune syndromes is crucial to provide treatment options to patients. We investigated a novel autoantibody binding pattern observed on murine teased sciatic nerve fibres. Target antigens were identified using immunoprecipitation combined with mass spectrometry. Target specificity of these autoantibodies was validated in cell-based assays, neutralization assays and knock-out models. A retrospective study cohort consisting of different neuropathies (chronic inflammatory demyelinating polyradiculopathy n = 86, Guillain–Barré syndrome n = 37, multifocal motor neuropathy n = 18, diabetic neuropathy n = 30, other inflammatory neuropathies n = 10), amyotrophic lateral sclerosis (n = 50), multiple sclerosis (n = 50) and healthy controls (n = 50) was negative for septin multimer autoantibodies. Histopathological analysis of skin and the sural nerve including electron microscopy was performed in one seropositive patient, and autoantibody binding was characterized in vitro. Extensive immunotherapy was initiated in one patient, with clinical and serological follow-up over 4 years. Among 3543 total samples tested, three patients (two male, one female)—diagnosed with the LMND variant of amyotrophic lateral sclerosis (aged 65, 72 and 79 years, respectively)—showed a novel and distinct autoantibody binding pattern of indirect immunofluorescence staining on peripheral nerves, targeting Schmidt–Lanterman incisures (SLIs), paranodes and the abaxonal myelin. Target identification and validation revealed septin multimers as autoantibody epitopes. Despite the primarily intracellular location of septins, autoantibody binding was evident in living myelinated dorsal root ganglia, primarily at SLIs (‘incisuropathy’). Septin multimer autoantibodies further initiated complement deposition on fixed and permeabilized cell-based assays. Sural nerve and skin biopsies showed inflammation, myelin and axonal pathology. Extensive immunotherapy in one patient was followed by disease stabilization over 3 years. The other two patients died of rapid disease progression: one of them received no immunotherapy while the other had ineffective treatments with single administrations of intravenous immunoglobulin and rituximab. Our data suggest that septin multimer autoimmunity occurs in severe motor-predominant neuropathies which can clinically resemble a neurodegenerative LMND. Screening for septin multimer autoantibodies should be considered in patients presenting with this phenotype. Follow-up studies need to determine the direct pathogenicity of septin multimer autoantibodies, their potential as a biomarker of an autoimmune syndrome and responses to immunotherapy in larger cohorts.

ABSTRACT Background Long‐term outcome data for stiff person spectrum disorder (SPSD) remain limited and sustained response to treatment varies. Methods In this retrospective cohort study, we included patients with probable or definite SPSD (Mayo Clinic criteria), evaluated at Mayo Clinic (12/1995‐07/2024) with ≥ 5 years of follow‐up after diagnosis. Primary outcome measures included the modified Rankin scale (mRS) and use of gait aids. Results Fifty‐four patients were included, 41 (76%) were female; median age at onset was 45 years (16–78). Most were antibody‐positive (51 [94%]), with high‐titer (serum, ≥ 20 nmol/L) GAD65‐IgG in 45 (83%). Nine patients had a stiff‐limb presentation (17%). The median interval from onset to diagnosis was 2 years (0–34), and median follow‐up was 9 years (5–26). Most patients (46 [85%]) received immune therapy; 37 (82%) reported improvement. All patients had significant impairment (mRS ≥ 2) at nadir, with 31 (57%) losing functional independence (mRS ≥ 3) and 40 (75%) requiring bilateral gait assistance. The median time from onset to nadir was 6 years (0–39). At last follow‐up, most (37 [69%]) were functionally independent (mRS ≤ 2), including 8 (15%) with minimal restrictions (mRS ≤ 1). Bilateral gait assistance was needed in 22 (41%). Among 40 patients with ≥ 10 years of follow‐up, 22 (56%) had an mRS ≤ 2 and 25 (63%) required a gait aid at last follow‐up. Poor outcome in this population was associated with bilateral gait assistance at nadir and older age ( p < 0.05). Conclusions Although over half of SPSD patients lose functional independence at disease nadir, many regain independence and achieve well‐controlled symptoms.

ABSTRACT Objective To (1) validate GAD65‐ELISA detection and quantification for type 1 diabetes mellitus and autoimmune neurological diagnoses, (2) correlate ELISA results (reference range < 5 IU/mL) with established radioimmunoprecipitation assay (RIA; ≤ 0.02 nmol/L), and (3) define ELISA clinical utility and pitfalls. Methods Serum performance for diabetes (FDA‐cleared, undiluted) was verified, and neurological laboratory‐developed serum and CSF dilution protocols were validated to extend the reportable range beyond > 250 IU/mL. ELISA and RIA values were correlated, including established neurological RIA cut‐offs (serum ≥ 20 nmol/L; CSF any positive). Results ELISA met analytical criteria (precision, accuracy, sensitivity, specificity, reference range) in serum and CSF and was clinically equivalent to RIA for autoimmune diabetes. Neurological ELISA cut‐offs were established at 10,000 IU/mL (serum) and 100 IU/mL (CSF). Precision was better below 10,000 IU/mL (CVs < 20%) than above (CVs 30.8% serum; 26.5% CSF). ELISA–RIA correlation and agreement was stronger below the neurological cut‐off (R 2 = 0.89) than above (R 2 = 0.36). Positive agreement for RIA‐defined neurological disease was 100% in serum and CSF; all serums were > 10,000 IU/mL. Clinical specificity was 97.5% in serum and 100% in CSF, exceeding reported RIA specificity. Screen results > 250 IU/mL spanned a wide range of dilution values; many were below the neurological cut‐off. Most patients with paired serum/CSF positivity showed elevated GAD65 IgG indices. Interpretation GAD65 ELISA and RIA have equivalent sensitivity for autoimmune diabetes and neurological testing, with higher specificity for ELISA. A serum cut‐off of 10,000 IU/mL is informative but requires clinical context, and dilution of screen > 250 IU/mL samples is essential for neurological interpretation.

Objective To estimate the prevalence, incidence, and disease burden of autoimmune seizures and epilepsies in Olmsted County. Methods We conducted a retrospective, population‐based study using the Rochester Epidemiology Project (1996–2019). Age‐ and sex‐standardized rates (2019 United States standard population) per 100,000 were calculated with 95% confidence intervals (95% CIs). Temporal trends in incidence were assessed over 3 consecutive 8‐year intervals (1996–2003, 2004–2011, and 2012–2019) using Poisson regression to estimate rate ratios per period and trend. Disability‐adjusted life years (DALYs), years of life lost (YLL), and years lived with disability (YLD) and 95% uncertainty intervals (95% UIs) were estimated for 2019. Results Twenty‐three patients were identified between 1996 and 2019. In 2019, the standardized prevalence of all autoimmune seizures and epilepsies was 9.81 (95% CI, 5.59–15.21) per 100,000, including autoimmune‐associated epilepsy (AAE) 7.27 (3.74–11.95) and acute symptomatic seizures secondary to autoimmune encephalitis (ASSAE) 2.55 (0.69–5.60). From 1996 to 2019, the standardized incidence was 0.50 (0.29–0.77) per 100,000 person‐years overall, with AAE 0.36 (0.19–0.59) and ASSAE 0.15 (0.05–0.30). By period, the standardized incidence of all autoimmune seizures and epilepsies suggested a rising trend from 0.28 (95% CI, 0.06–0.68) in 1996 to 2003 to 0.60 (95% CI, 0.24–1.12) in 2004–2011, then plateaued at 0.53 (95% CI, 0.21–1.00) in 2012 to 2019 (rate ratio per 8‐year interval 1.30, 95% CI, 0.71–2.38), largely driven by the seropositive subgroup. The standardized DALY rate was 26.27 (95% Uncertainly intervals [UI], 5.30–67.09) per 100,000 (YLL 20.07 [0.00–60.21] and YLD 6.20 [5.06–7.39]). Interpretation Although autoimmune seizures and epilepsies remain rare, they result in a disproportionately high disease burden. ANN NEUROL 2026;100:74–83

Abstract Introduction Cila-cel is associated with a 10-15% risk of delayed neurotoxicity (DNT) with rates of cranial nerve palsies (IEC-NP) and parkinsonism (IEC-PKS) seen at a rate of 10% and 1-4% respectively and can occur as early as 3-4 weeks after infusion. We and others have shown that high post-infusion peak absolute lymphocyte count (ALCpeak) of &gt;3x10^9/L is associated with an increased risk of delayed neurotoxicity (Lim et al., ASCO 2025) whilst the risk was negligible (~1%) when ALCpeak fell below that threshold. Since December 2024, we implemented prophylactic administration of dexamethasone (PPX DEX) when ALC was &gt;3x10^9/L (defined as high ALCpeak). Outcome of these patients are described here. Methods Analysis of clinical outcome included 230 pts treated between Feb 2022 and Nov 2024 as a historical cohort and 102 patients treated between Dec 2024 and May 2025 after the introduction of dexamethasone prophylaxis using an oral dose of 10mg twice a day for at least three days in the high ALCpeak cohort. Results No significant differences were seen between the new and historical cohort in pre-lymphodepletion disease burden (bone marrow plasma cell %, involved light chain or M-spike level), and rates of all/high-grade CRS or ICANS, IEC-HS and rates of tocilizumub use. Median follow-up in the new cohort of 102 pts was 3.6 months (95% CI 3.3-4.2). Eleven (10.8%) developed DNT [3 (2.9%) with IEC-PKS, 8 (7.8%) with IEC-NP]. High ALCpeak of &gt;3x10^9/L was observed in 51 (50%) pts. Among pts with high ALCpeak, DNT was observed in 10 (20%) pts [3 (5.9%) with IEC-PKS, 7 (13.7%) with IEC-NP] whereas among 51 pts with ALCpeak &lt;3x10^9/L, DNT was observed in 1 (2%) pt [1 (2%) with IEC-NP]. In the high ALCpeak cohort, 38 (75%) received DEX [6 (12%) for toxicity management, 32 (63%) PPX DEX]. Of the 38 pts treated with DEX, DNT was observed in 6 (16%) pts (2 (5.3%) IEC-PKS, 4 (10.5%) IEC-CNP). We next compared DNT rates between this new cohort of high ALCpeak pts receiving DEX (n=38) against the historical cohort with high ALCpeak (n=98). Estimated 90-day DNT by Kaplan-Meier method was (19.2% vs 20.5%; p=0.53): IEC-PKS (5.8% vs 7.2%; p=0.60) and IEC-NP (11.7% vs 13.3%; p=0.58). In patients with high ALCpeak, the use of DEX did not appear to significantly reduce the risk of DNT compared to historical controls (HR 0.76, 0.33-1.78), IEC-PKS (HR 0.67, 0.14-3.10) or IEC-NP (0.73, 0.24-2.23). We next compared ALC expansion over the first month post CAR-T infusion (area under the curve, AUC). Among the new cohort of 102 patients, no difference was seen in the AUC of pts who received DEX but did not develop DNT vs those who developed DNT (31.0 vs 57.9, p=0.28); No difference in AUC was seen in the new high ALCpeak cohort receiving DEX vs the historical cohort with high ALCpeak (36.0 vs 43.8, p=0.15) Of the eight cases of IEC-NP, 4 received PPX DEX. No noticeable differences were observed in the severity of presentation, laterality of cranial nerve involvement and rates of multiple cranial nerve involvement between these 4 cases who received PPX DEX and the 16 cases from the historical cohort. At the time of database freeze, 3 of the 4 cases had resolved and 1 was improving. Median time to complete symptom resolution was 38 days vs 69 days in the historical cohort. Of the three cases of IEC-PKS, 2 received PPX DEX. Median time to symptom onset for IEC-PKS was 46 days (vs. 24 days in historical cohort). Both cases had severe symptoms non-responsive to first line high-dose methylprednisolone, second line intrathecal chemotherapy and third line high-dose cyclophosphamide. Both pts subsequently died from sepsis 28 and 67 days post-cyclophosphamide. Conclusion While ALCpeak &gt;3x10^9/L predicts risk of DNT in pts treated with cilta-cel, the use of PPX DEX does not appear to significantly reduce ALC expansion and does not appear to significantly reduce the risk of DNT, IEC-PKS or IEC-NP. Other strategies to reduce ALC or alter T cell functions may need to be considered.

ABSTRACT Objective We sought to characterize the sixth most common finding in our neuroimmunological laboratory practice (tissue assay‐observed unclassified neural antibodies [UNAs]), combining protein microarray and phage immunoprecipitation sequencing (PhIP‐Seq). Methods Patient specimens (258; 133 serums; 125 CSF) meeting UNA criteria were profiled; October 2022–September 2023. Top‐ranking candidate antigens were validated in silico, by dual‐staining confocal microscopy, and ≥ 1 protein‐specific assay. Clinical data were reviewed. Results Among 21 patients, 11 autoantibodies were characterized (serum, 19; CSF, all 9 available). Autoantigens were CACNA1I, 1; CAMK2B, 2; CLIP2, 1; FMN2, 2; MAP1A, 2; MAP2, 5; NECAB1, 1; SNAP91, 3; SRCIN1, 1; SYNJ1, 1; SYT3, 2. Analytical validation was by confocal TIIFA (all), western blot (10/10 available), and cell‐based assay (5/5 performed). Clinical accompaniments were: encephalitis, 6; brainstem encephalitis, 2; encephalomyelitis, 2; cerebellar ataxia, 2; longitudinally extensive transverse myelitis (LETM), 2; sensory neuronopathy, 1; peripheral neuropathy, 4, and movement disorders, 2. Inflammatory MRI abnormalities were identified in 5/16 patients (31%) with CNS disorders: T2 signal change (2), LETM (2), leptomeningeal enhancement (1). Seven of 8 (88%) had inflammatory CSF (pleocytosis, 5 [median 25.5 cells, range 7–294]; elevated IgG index/synthesis rate, 4; CSF‐exclusive oligoclonal bands, 4). Six had paraneoplastic causation (lung cancer, 2; other, 4); 3 were postinfectious (1 each of COVID‐19, HSV‐1, and post‐Group A streptococcal infection). Of 9 immunotherapy‐treated patients, 5 improved. Interpretation UNAs are partly accounted for by a repertoire of diverse mostly intracellular synaptic antigens. Their characterization is expedited by protein arrays and PhIPSeq. Further individual studies are needed to assess them as disease biomarkers.