Autoimmune Epilepsy:  Clinical Characteristics and Response to Immunotherapy FREE

Quiz Ref IDAntiepileptic drugs (AEDs) are the mainstay of treatment for epilepsy, but seizures continue in one-third of patients despite appropriate AED therapeutic trials.¹ Even in the current era, the etiology of epilepsy often remains unclear.² Seizures are a common symptom in autoimmune neurologic disorders, particularly in limbic encephalitis or multifocal paraneoplastic disorders.^3- 16 Autoantibody specificities recognized in the setting of paraneoplastic limbic encephalitis include antineuronal nuclear antibody type 1, collapsin response-mediator protein 5 (CRMP-5), and Ma2. Voltage-gated potassium channel (VGKC) complex and glutamic acid decarboxylase 65 (GAD65) antibodies, often nonparaneoplastic in etiology, have been reported in patients with limbic encephalitis^7,13,15- 16 and idiopathic epilepsy with AED-resistant seizures.^17- 22 Newly identified autoantibody specificities that strongly correlate with clinical seizures include N -methyl-D-aspartate (NMDA),²³ γ-aminobutyric acid B,²⁴ and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors.²⁵

Accumulating data support an autoimmune basis in patients with AED-resistant seizures,^17- 22 including those lacking a typical “limbic encephalitis” phenotype.^26- 28 Identification of an immune basis is important because adjunctive immunotherapy may slow, halt, or even reverse the epileptogenic process in these patients. In a cohort study, autoimmune antibodies were detected in 14% of patients with epilepsy.²⁷ This study, along with several case reports and series,^17,19,22 suggested a potential benefit of immunotherapy in improving seizure control. Herein, we review the clinical characteristics and responses to immunotherapy for patients with suspected autoimmune epilepsy, evaluated jointly in an Autoimmune Neurology Clinic and Epilepsy Clinic, whose sole or predominant presenting symptom was recurrent, uncontrolled seizures.

With approval of the Mayo Clinic institutional review board, we searched the Mayo Clinic computerized diagnostic index to identify patients who were evaluated in both the Autoimmune Neurology Clinic and Epilepsy Clinic between January 1, 2005, and December 31, 2010, whose evaluations led to a diagnosis of autoimmune epilepsy. Autoimmune epilepsy was defined as (1) epilepsy as the exclusive (n = 11) or predominant (n = 21) presenting concern and (2) autoimmune pathogenesis suspected by the treating physicians based on detection of a neural autoantibody, inflammatory cerebrospinal fluid (CSF) (leukocytosis or CSF-exclusive oligoclonal immunoglobulin bands), or magnetic resonance imaging (MRI) characteristics suggesting inflammation (T2 hyperintensities, contrast enhancement on gadolinium studies, and/or restricted diffusion).

Demographic and clinical characteristics (seizure semiology, clinical course, and associated symptoms) were recorded. Head MRIs and whole-body radiolabeled fluorodeoxyglucose positron emission tomography (FDG-PET) scans were reviewed by at least 2 investigators (A.M.L.Q., A.L.K., and R.E.W.) blinded to the clinical data (one, a neuroradiologist). The electroencephalogram (EEG) studies were scalp recordings acquired via electrodes applied using the international 10-20 system for electrode placement. All routine EEGs comprised two 1-channel and all prolonged 30-channel digital EEG recordings. Longitudinal and transverse bipolar, Cz and ear/mastoid referential, and Laplacian montages were used as indicated to optimize seizure detection and localization. Results of neural autoantibody screening were recorded.^3,13- 14 A composite substrate of mouse cerebellum, midbrain, stomach, and kidney was used in a standardized indirect immunofluorescence assay to detect the following neuronal and glial nuclear and cytoplasmic IgG autoantibodies: ANNA types 1 (anti-Hu), 2 (anti-Ri), and 3; Purkinje cell cytoplasmic autoantibodies types 1 (anti-Yo), 2, and Tr; CRMP-5; amphiphysin; and antiglial/neuronal nuclear antibody type 1. In-house assays included radioimmunoprecipitation to detect antibodies reactive with cation channel complexes (neuronal voltage-gated calcium channels [P/Q type and N type], VGKC complex, nicotinic acetylcholine receptors of skeletal muscle and autonomic ganglionic types) and GAD65, enzyme-linked immunosorbent (skeletal muscle striational antibodies) and recombinant Western blot (CRMP-5–IgG). Frequencies of these neural autoantibodies in healthy controls (Table 1A, and Table 1 B, footnote) were previously reported.²⁹ Ma/Ta antibodies were identified via recombinant Western blot (Athena Diagnostics).

Supplementary immunofluorescence assays were performed on sections of mouse cerebral cortex, hippocampus, and thalamus to detect synapse-reactive IgG autoantibodies specific for NMDA, AMPA, and γ-aminobutyric acid B receptors. N -methyl-D-aspartate receptor seropositivity was confirmed molecularly by immunofluorescence on HEK293 cells transfected with NMDA receptor complementary DNA (product of EUROIMMUN). Sera positive for VGKC complex antibodies by radioimmunoprecipitation were analyzed further for IgGs specific for leucine-rich, glioma-inactivated 1 (Lgi1) or contactin-associated proteinlike 2 (Caspr2) using HEK293 cells transfected with Lgi1 or Caspr2 complementary DNA (product of EUROIMMUN). These proteins coprecipitate with Kv1 VGKC complexes solubilized from mammalian brain membranes and ligated with iodine 125–labeled α-dendrotoxin.^8- 9 All sera tested in 126 healthy controls were negative for VGKC complex, Lgi1, or Caspr2 autoantibodies.

Response to immunotherapy was categorized on the basis of physician and patient reports of seizure freedom, seizure improvement (reduction in seizure frequency and severity), or no change.

Data were expressed as median (range and interquartile range) for continuous variables and counts (percentages) for categorical variables. Differences between responders (seizure freedom or improvement) and nonresponders were compared using an unpaired t test, analysis of variance, and Wilcoxon rank sum tests for continuous measures and χ² and Fisher exact tests for categorical variables.

Clinical, radiological, EEG, autoimmune serologic values, and immunotherapeutic outcomes for 32 patients are presented in Table 1A, Table 1 B and Table 2. All presented with recurrent seizures. Fifty-nine percent were female. Median seizure onset age was 56.0 years (range, 5-79 years). Median history of seizure activity prior to Mayo Clinic presentation was 5 months (range, 3 weeks to 12 years). Quiz Ref IDAn autoimmune basis was suspected based on detection of a neural autoantibody (91%), inflammatory CSF (leukocytosis or CSF-exclusive oligoclonal immunoglobulin bands) (31%), or MRI characteristics suggesting inflammation (63%).

Partial seizures were the predominant clinical presentation: simple partial and/or auras, 27 of 32 (84%); complex partial, 26 of 32 (81%); and secondary generalized tonic-clonic, 17 of 32 (53%). Seizure semiologies were variable or changed over time in 12 patients (38%). Most patients (81%) had received 2 or more AEDs at presentation (median, 3 AEDs), yet seizures were frequent: 26 (81%) had daily seizures; the remaining had at least 1 seizure per month.

Two patients had undergone epilepsy surgery without seizure benefit elsewhere (anterior temporal lobectomy plus amygdalohippocampectomy and frontal corticectomy, patients 5 and 14, respectively); none had a neoplasm. Perivascular chronic inflammatory cell infiltrates (mainly T lymphocytes) were noted on histopathology review at our institution in patient 5; details for the other patient are unavailable. Continuation of poorly controlled seizures prompted postoperative referral to Mayo Clinic.

All 32 patients had EEGs recorded in our institution (eTable 1), with a median of 2 per patient (range, 1-14). Prolonged video EEG monitoring was performed in 13 (41%). The following abnormalities were recorded: interictal epileptiform discharges, 20; electrographic seizures, 15; focal slowing, 13; and generalized slowing, 12. Three patients (patients 4, 9, and 13) had no EEG abnormalities detected, of whom only 1 (patient 13) had MRI inflammatory changes.

Additional manifestations included memory and cognitive difficulties, 20 (63%); personality changes, 8 (25%); and depression or anxiety, 6 (19%). Neurocognitive changes developed subsequently in 3 of 11 patients who did not have memory or affective changes at presentation (34%).

Magnetic resonance imaging brain scans were available for review in all patients (Figure and eTable 2). Fifteen (47%) had normal MRIs at the time of initial seizure evaluation. Abnormalities were observed in 22 (17 at initial evaluation, 5 on follow-up imaging): probable inflammatory changes were interpreted in 20 (63%); 2 showed postsurgical changes. Among the 5 patients whose inflammatory changes were only detected on subsequent imaging, the median interval between normal and subsequent abnormal scans was 4 months (range, 1-8 months). Abnormalities deemed inflammatory included swelling and T2 hyperintensity involving the amygdalohippocampal complex (17 patients [53%]) and extramedial temporal structures (6 patients [19%]). Six of 19 gadolinium studies demonstrated contrast enhancement (32%). Five of 19 diffusion-weighted sequence MRIs demonstrated restricted diffusion (26%). Prior to immunotherapy, 4 patients had radiographic features indistinguishable from medial temporal sclerosis.

Whole-body FDG-PET images, performed as a screen for occult malignancies in 20 patients, were reviewed. Brain sections of these studies showed medial temporal region hypermetabolism in 11 patients and left parietal cortex hypermetabolism in 1. No clinical seizures were reported to have occurred during PET acquisition in any patient. However, specific inquiry as to the presence or absence of seizure activity during acquisition is not a part of the routine procedure during PET, and none were performed with concurrent EEG monitoring. Medial temporal and extratemporal hypometabolism was detected in 1 patient.

Neural autoantibodies were identified in 29 patients (91%). Specificities were VGKC complex, 18; GAD65, 7; CRMP-5, 2; Ma (PNMA1 and PNMA2), 1; NMDA receptor, 1; and neuronal nicotinic acetylcholine receptor, ganglionic type, 1. Among the 18 patients who had VGKC complex IgG, 14 (78%) bound to Lgi1, 1 bound to Caspr2, and 3 were of unknown specificity (eFigure 1). The 3 patients who lacked detectable neural autoantibodies (patients 7, 11, and 21) had other features that supported the likelihood of autoimmune epilepsy: 2 had inflammatory CSF, all 3 had inflammatory MRI abnormalities, 2 had a personal history of cancer (1 prostate and 1 breast), and 1 had coexistent autoimmune disease (thyroid disease and celiac sprue). None had laboratory findings to indicate an infectious etiology.

The identification of a neural autoantibody led in 3 patients (patients 2, 4, and 16) to prospective detection of cancer: 2 with VGKC complex antibodies had thyroid or prostate carcinoma and 1 patient with CRMP-5 antibody had recurrent bladder cancer. Cerebrospinal fluid abnormalities were found in 19 of 30 patients (63%) evaluated: elevated leukocyte count (>5/μL), 5 patients; CSF-exclusive oligoclonal bands, 5 patients; and elevated protein level (>35 mg/dL), 17 patients.

Immunotherapy was instituted in 27 of 32 patients for the treatment of persistent seizures despite AED therapy (Table 3). Initial immunotherapy comprised intravenous methylprednisolone alone (IVMP) (n = 12); intravenous immune globulin alone (IVIg) (n = 3); and combinations of IVMP, IVIg, cyclophosphamide, or plasmapheresis (n = 12). The median follow-up period was 17 months (range, 3-72 months). At last follow-up, 22 of 27 patients (81%) had improved clinically after initiation of immunotherapy. The median time from seizure onset to initiating immunotherapy was 4 months for responders and 22 months for nonresponders (P < .05). All 15 VGKC complex antibody–positive patients and 3 of 5 GAD65-seropositive patients (60%) reported benefit (P < .05 and P = .17, respectively) (eTable 3). Five responders had relapses during follow-up. With further immunotherapy and/or AED treatment, 2 eventually achieved seizure control. Their autoantibody specificities were CRMP-5, 1; GAD65, 1; and VGKC complex (Lgi1), 3. Five patients did not respond to immunotherapy. However, 2 of the 5 demonstrated subsequent improvement after AEDs were changed (patients 19 and 29).

Eighteen patients (67%) achieved seizure freedom over a median period of 10 months (range, 2-48 months). Eight of those patients (44%) were seizure free within 12 weeks of immunotherapy initiation. Eight patients (44%) had no residual deficits, but others experienced residual neurologic deficits, despite achieving seizure freedom. Cognitive and memory concerns were improved but persisted in 8 (44%). Four patients had behavioral or mood changes. One patient (patient 32) had residual aphasia having presented with intractable aphasic seizures and left cortical inflammatory changes. For long-term maintenance, immunotherapy comprised azathioprine only, 2; mycophenolate mofetil only, 11; or combinations of azathioprine, mycophenolate mofetil, prednisolone, rituximab, or methotrexate, 5.

Postimmunotherapy imaging was available for review in 15 of the patients whose scans had revealed evidence of inflammation (eTable 2). Four patients had no evidence of radiological changes. Five showed reduction in hyperintensity size, and 5 patients developed hippocampal atrophy and sclerosis. One patient with initial T2 hyperintensity in the right posterolateral temporal lobe before immunotherapy had complete resolution of this abnormality. Postimmunotherapy antibody values were available for 10 patients (Table 1A, and Table 1 B, last column) who had a favorable immunotherapy response. Of 7 VGKC complex–seropositive patients, posttherapy values were lower in 3 and undetectable in another 3. CRMP-5 IgG did not persist in 2 patients and GAD65 antibody level decreased in 1.

One patient (patient 30) who had VGKC complex antibodies was seizure free for a 2-week period after receiving a third AED. An immunotherapy trial was initiated because of significant residual memory impairment but not for seizures; cognition improved within 3 months, and seizures did not recur. Four patients did not receive immunotherapy. Two patients declined, and the need for immunotherapy in the third patient (patient 2) was obviated because he became seizure free following removal of a thyroid papillary carcinoma found in the malignancy screening that was prompted by VGKC complex antibody detection. Seizure resolution followed. A fourth patient (patient 6), whose seizures also were associated with VGKC complex antibodies (3.5 nmol/L; normal range, ≤0.02 nmol/L), was refractory to the first AED (levetiracetam), but seizures were controlled after a second AED was started (lamotrigine). The AED therapy was discontinued after 3 years, and the patient remained seizure free 12 months later.

All 32 patients for whom we describe clinical, serologic, and imaging findings had refractory epilepsy of presumed autoimmune basis. The intractability, high seizure frequency, and striking improvement in seizure control achieved following immunotherapy in many warrant emphasis: 81% had significant improvement in seizure status and 67% achieved seizure freedom, a majority of whom were AED resistant.

Our study supports previously noted links between neurologic autoimmunity and epilepsy.^{18- 21,26- 28} Recurrent seizures were the early and predominant clinical manifestation in the patients of our report. An autoimmune etiology is identified most readily in patients who present with the full syndrome of limbic encephalitis, characterized by subacute memory impairment with affective changes and temporal lobe seizures. The diagnosis of autoimmune limbic encephalitis is aided by detection of neural autoantibodies with radiological or pathological evidence of temporomedial inflammation and in some cases a history of neoplasia in the preceding 5 years.³² Limbic encephalitis has been suggested as a precedent of hippocampal sclerosis and adult-onset temporal lobe epilepsy.^6,33 In our report, one-third of the patients had seizures as their exclusive presentation without other recognized clinical accompaniments of limbic encephalitis. Although the remaining two-thirds had additional neurologic problems, including cognitive and personality changes, they had presented with predominant concerns of high daily seizure burden. This prevented clear distinction of the contribution of inflammatory limbic lesions vs seizure activity to the evolving neurocognitive impairment. Furthermore, 15 patients had normal MRI brain scans at initial presentation, and among 12 patients who had subsequent MRIs, a median of 4 months elapsed before subsequent imaging showed development of inflammatory changes in 5.

The primary aim of this study was to report the clinical features and immunotherapy response in a cohort diagnosed with autoimmune epilepsy. The study was not designed to compare clinical features of this entity with those of epilepsy from other etiologies. The diagnosis of autoimmune epilepsy requires a high level of suspicion at initial evaluation. The clinical presentations in our patients were heterogeneous, but some general observations can be made. Quiz Ref IDData from the current cohort suggest that autoimmune investigation should be considered in the presence of 1 or more of the following: an unusually high seizure frequency, intraindividual seizure variability or multifocality, AED resistance, personal or family history of autoimmunity (either organ specific [eg, thyroid disease, diabetes mellitus, pernicious anemia, or celiac disease] or non–organ specific [rheumatoid arthritis or systemic lupus erythematosus]), or recent or past neoplasia. Serological testing is increasingly valuable as an aid to establishing the diagnosis of an autoimmune etiology. As illustrated in the patients we presented, other laboratory and radiological findings may be normal. Serial MRI findings were consistent with inflammation in several patients. When detected, these radiological findings (sometimes indistinguishable from medial temporal sclerosis) supported the diagnosis of autoimmune epilepsy. Quiz Ref IDHowever, MRIs were normal in about half of patients. Cerebrospinal fluid was also normal in nearly half the patients despite the presence of an autoimmune neurologic disorder. Hence, the presence of normal CSF or MRI does not exclude an immune-mediated process. The role of brain FDG-PET in these patients warrants further study. While focal hypometabolism is more typically seen in the epilepsy population, focal hypermetabolism was fortuitously noted in this cohort who underwent whole-body FDG-PET primarily for malignancy purposes. It is our continuing observation that autoimmune epilepsies are underrecognized.

A majority of the patients in this study had neuronal VGKC complex autoantibodies. This serological marker aids the diagnosis of idiopathic and less commonly paraneoplastic autoimmune neurologic disorders. It is impressive that the seizure disorder was immunotherapy responsive in all seropositive patients. Voltage-gated potassium channel complex autoimmunity was first reported in patients with neuromyotonia³⁴ (Isaacs syndrome), Morvan syndrome,³⁵ and limbic encephalitis.³⁶ A broader spectrum of neurologic phenotypes affecting all levels of the nervous system has been described.^30,37 Two independent groups recently reported that the target autoantigens in these disorders are generally not VGKC complex channel proteins per se but neuronal proteins (Lgi1 and Caspr2) that respectively associate with a subset of Kv1 VGKC complexes at synapses and at juxtaparanodes of myelinated axons.^8- 9 Lgi1 was the target antigen in 78% of our VGKC complex antibody–positive patients. One had antibodies targeting Caspr2. Previous reports have implicated Lgi1 as the principal target antigen in limbic encephalitis, while Caspr2 is more commonly, but not exclusively, associated with peripheral nervous system manifestations.^8- 9 Lgi1 is recognized as a causative gene in autosomal-dominant partial epilepsy with auditory features.³⁸ It encodes a secreted protein that links 2 epilepsy-related receptors, ADAM22 and ADAM23, creating a complex that incorporates presynaptic potassium channels and postsynaptic AMPA receptor scaffolds. Fukata and colleagues³⁹ demonstrated that disruption of the Lgi1-linked synaptic complex causes abnormal synaptic transmission and epilepsy. Recently, faciobrachial dystonic seizures were reported to precede Lgi1 antibody–associated encephalitis, suggesting that early immunotherapy could prevent the evolution to limbic encephalitis.²²We identified similar seizures in 6 of 14 (43%) Lgi1-seropositive patients in this cohort, often accompanied by other seizure semiologies. We also noted piloerection as a semiological feature in 4 of 14 (29%).

One patient in our study had NMDA receptor autoantibodies.²³N -methyl-D-aspartate receptor autoimmune encephalitis is often accompanied by ovarian teratoma and a stereotypic clinical evolution starting with a viral-like prodrome, psychiatric symptoms, memory impairment, dyskinesias, seizures, and progressing coma and hypoventilation.⁴⁰ Most reported cases have had seizures at presentation,⁴¹ but these were overshadowed or accompanied by neurocognitive disturbances. Our patient presented with AED-intractable aphasic seizures and evolving left cortical inflammatory changes.

When autoimmune epilepsy is suspected on clinical grounds, CSF evaluation and comprehensive screening for neural autoantibodies are indicated. Selective autoantibody testing is not advised because no single neural antibody is definitively associated with seizures. Failure to detect a neural antibody does not exclude the diagnosis of autoimmune epilepsy when other clinical clues exist. If autoimmune epilepsy is suspected, a trial of 6 to 12 weeks of immunotherapy (IVMP or IVIg daily for 3 days and then weekly) is justifiable in the absence of other treatment options and may serve as additional evidence for an autoimmune etiology when a favorable seizure response is observed.⁴² In 22 of 27 patients (81%), this therapeutic trial was positive, and early treatment was associated with a favorable outcome (P < .05). Long-term immunosuppressive treatment, overlapping with gradual taper of IVMP or IVIg, should be considered for patients whose seizures respond favorably to the initial trial of immunotherapy. Despite this, relapses may still occur.

Our study is limited by its retrospective design and the fact that AED changes were not restricted during the period of immunotherapy. The patients' poorly controlled seizures necessitated continuing AED changes during immunotherapy initiation, complicating interpretation of the contribution of immunotherapy to seizure control. However, the likelihood that such changes accounted for improved clinical response in these patients is well below the proportion of patients responding to immunotherapy trial.¹Quiz Ref IDClinical experience suggests that immunotherapy should not be used alone to control seizures but should be used in combination with AEDs to optimize seizure control. The clinical spectrum of autoimmune epilepsy is still unknown. In a series of patients with epilepsy, VGKC complex antibodies were detected in 10%; NMDA receptor antibodies, in 7% of newly diagnosed patients; and GAD65 antibodies, in 1.6% to 1.7%.⁴³ It is conceivable that we are only identifying patients with the most severe presentations in this heterogeneous group, and the burden of this entity remains underappreciated in patients with milder epilepsies. Questions remaining unanswered include the natural history of autoimmune epilepsy, the selection criteria for patients with epilepsy most likely to benefit from an autoimmune evaluation, the timing for immunotherapy trial, and optimal duration of long-term immunotherapy maintenance.