Temporal Lobe Epilepsy With GAD Antibodies—Time to Give Up,or Time to Double Down?

Abstract

Temporal Lobe Epilepsy Associated With Glutamic Acid Decarboxylase Antibodies: Defining a Distinct Epilepsy Syndrome

Rada A, Reisch LM, Hagemann A, Woermann FG, Bien CG. Neurol Neuroimmunol Neuroinflamm. 2025;12(4): e200422(1)-e200422(13). doi:10.1212/NXI.0000000000200422

Background and Objectives: Epilepsy associated with glutamic acid decarboxylase antibodies (GAD-abs) often presents as temporal lobe epilepsy (TLE). However, detailed data on this condition are lacking. The aim of this study was to demonstrate that TLE with GAD-abs (GAD-TLE) is distinct from other forms of TLE and to describe the subgroup of patients with epilepsy and GAD-abs but no evidence of TLE. Methods: Inpatients from 2011 to 2022 with high serum GAD-abs (≥1:80, confirmed by a cell-based assay and typical staining pattern on a tissue-based assay) were included. The control group comprised patients with mesial TLE (EMU-mTLE) with hippocampal sclerosis or nonlesional epilepsy diagnosed on the epilepsy monitoring unit (EMU) and having negative results on a broad panel of neural antibodies. Data were retrospectively collected from first admission (visit 1 [V1]) and most recent follow-up (visit 2 [V2]). Parameters included demographic data, comorbidities, seizure characteristics, electroencephalography findings, MRI results, verbal memory performance, GAD-ab titers (V1), therapies, seizure and memory outcomes, and occupational status (V2). Results: Of 81 GAD-ab–positive patients, 71 (88%) had TLE. Compared with 40 EMU-mTLE controls, patients with GAD-TLE were more often female (84% vs 33%, p < 0.001), more frequently had autoimmune comorbidities (46% vs 2.5%, p < 0.001), musicogenic seizures (18% vs 0%, p = 0.004), and seizure clusters (25% vs 5%, p = 0.002) but less frequently had hippocampal sclerosis (25% vs 54%, p = 0.007). At follow-up (GAD-TLE: median 5 years; EMU-mTLE: 2 years), only 14% of patients with GAD-TLE were terminally seizure-free for ≥1 year, compared with 39% of patients with EMU-mTLE (p = 0.04). Verbal memory remained stable, and occupational status was unchanged. Ten GAD-ab–positive cases without evidence of TLE were identified, including 3 with generalized features. Discussion: Epilepsy associated with GAD-abs predominantly manifests as TLE and as such is an epilepsy syndrome (GAD-TLE). In comparison with EMU-confirmed patients with mTLE without GAD-abs, the seizure outcome is inferior but the neuropsychological performance is usually normal and remains stable. A subset of patients with epilepsy with high-titer GAD-abs do not have TLE, some presenting with generalized features.

From the addition of immune to the ILAE epilepsy etiology classification, ¹ to the conceptual definition of autoimmune associated epilepsy as a distinct construct from acute symptomatic seizures due to autoimmune encephalitis, ² the last decade has seen a rise in an interest in diagnosing and treating autoimmune causes of chronic epilepsy. Translating this motivation to practically diagnosing these patients and addressing their etiology in a tailored therapeutic approach has been more challenging. ³ In their latest publication, Rada et al ⁴ refine the clinical syndrome of glutamic acid decarboxylase (GAD) antibody associated temporal lobe epilepsy (TLE), and raise new questions on where to find GAD autoimmune encephalitis associated epilepsy (AEAE), and whether the search is worthwhile.

In a period of over a decade, this epilepsy center in Germany tested about one fifth of referred patients for neural autoantibodies, including to GAD, and found 81 GAD-antibody positive patients, of which 71 had TLE, roughly equal to 3% of the overall new epilepsy referrals. This figure mirrors other studies indicating that about one in twenty patients with focal epilepsy of unknown cause carry neural antibodies,^5,6 of which GAD is likely the most common. The clinical parameters followed by the authors are in line with a framework largely accepted in the epilepsy community, although not yet operationalized in a guideline or clinical scoring tool: patients with adult onset focal epilepsy of unknown cause, with suspected temporal lobe seizures and associated cognitive or psychiatric symptoms. Of note, not all features were necessary for antibody testing. To be included in the GAD-TLE cohort, high-titer serum GAD antibodies were necessary—of the 71 subjects, about half also had intrathecal synthesis identified on cerebrospinal fluid testing. The authors then selected a control cohort defined as TLE (without additional features suggestive of “temporal-plus” epilepsy), either nonlesional or with hippocampal sclerosis (HS), and without neural antibodies. A small proportion (3%) of this control, non-GAD, TLE cohort received immunotherapy for suspected antibody-negative AEAE, and no comment is made on whether cerebrospinal fluid testing was performed, so this control cohort may not have been entirely non-autoimmune. It is also worth noting that 10 patients with high titer GAD antibodies did not have TLE but harbored a different epilepsy syndrome—three with idiopathic generalized epilepsy, the remainder with extratemporal epilepsy, unclassified or psychogenic nonepileptic seizures. The clinical significance of GAD antibodies in these patients is unclear.

When comparing GAD-TLE to this antibody-negative nonlesional or HS-TLE cohort, several clinical features emerge. Female sex, comorbid type 1 diabetes mellitus were highly prevalent in the GAD-TLE cohort. Electroclinically, musicogenic seizures were present in one fifth of the GAD-TLE cohort and never seen in the control epilepsy group, while seizure clusters, deja-vu and jamais vu also were more common with GAD-TLE. Bitemporal or no interictal epileptiform discharges were more common in the GAD-TLE group, which was also less likely to harbor HS. Notably, no cognitive profile differences were noted between GAD and control group, going against a general (unproven) notion that autoimmune associated epilepsy must have significant cognitive dysfunction (as indicated in the profile described by the authors to prompt antibody testing). About a third of both GAD-TLE and the control TLE groups had objective deficits on verbal memory tasks.

No deterioration in occupational status was noted in the GAD-TLE, again despite the impression that this is a severe and refractory epilepsy syndrome with a poor prognosis. ³ The authors posit that this stable functional status may be related to generally preserved cognition in the majority of the GAD-TLE group. Meanwhile, seizure outcomes reported were in line with the existing literature—only a minority became seizure free, two after temporal lobe resections, the remainder after a combination of antiseizure medication and immunotherapy changes. While the general autoimmune encephalitis literature is clear on the importance of timely immunotherapy, ⁷ the time to immunotherapy in this cohort did not impact the chances of seizure freedom. Therefore, seizure outcomes are poor, but when good seizure outcomes occur, they are not uniformly due to correct identification of the etiology and its treatment with immunotherapy. In this cohort, classical epilepsy management (in the form of epilepsy surgery and adjustments to antiseizure medications) could be effective.

These findings raise the question—is the quest for autoimmune associated epilepsy worth it? Since the epileptological approach to treatment can be successful, why bother testing broadly, if immunotherapy does not lead to seizure freedom like in acute symptomatic seizures from surface antibody mediated autoimmune encephalitis? Using retrospective data runs the risk of perpetuating a self-fulfilling prophecy. There is a sense of nihilism when it comes to GAD-TLE. Patients do not become seizure free with immunotherapy, so why look for it aggressively and early—indeed, the median time to diagnosis in this cohort of GAD-TLE was 6 years. But histopathological data shows a rapid reduction in the density of cytotoxic T cell infiltrates in GAD-TLE,^3,8 and a concurrent increase in MRI findings consistent with hippocampal sclerosis. Is the race to the diagnosis just as time-sensitive in GAD-TLE as it is with surface antibody mediated autoimmune encephalitis? Miraculous recovery after immunotherapy in GAD-TLE lives in the realm of case reports, ³ but when early treatment is rare, that may be the only reasonable expectation. Also, the extrapolation of autoimmune encephalitis immunotherapy principles, based on antibody depletion, to GAD-TLE may be misguided. The primary mechanism of immune-mediated injury in GAD-TLE is likely through cytotoxic T cells, which are not targeted with the majority of autoimmune encephalitis immunotherapies. Tailoring immunotherapies to the precise immunological mechanisms, when the disease is still active, is much more likely to lead to a change in their currently grim seizure outcome.

If we are to double down on early identification of those patients, a precise understanding of the phenotype becomes crucial. This publication by Rada and colleagues brings us closer to a precise syndrome with specific demographics, comorbidities, and electroclinical characteristics. It is however worth noting that one fifth of their referrals underwent testing for autoantibodies, and despite this, some surprises occurred with 10 patients with an unexpected non-TLE syndrome. Defining clinical characteristics through testing of a prospectively recruited unbiased cohort of focal epilepsy of unknown cause for GAD (and other neural) antibodies and comparing to a clearly non-autoimmune epilepsy cohort will likely lend the cleanest answers on what the phenotype is, so that we can start broadly and rationally testing epilepsy as early as possible.