Autoantibodies to specific neurologic proteins are associated with subacute onset encephalopathies, which often present with seizures that are poorly controlled by conventional antiepileptic drugs (AEDs). Previous cross-sectional studies have found specific neurologic antibodies in a small proportion of people with established epilepsy, but these investigations have seldom included patients with recent diagnosis.
We screened two large epilepsy cohorts to investigate the prevalence of multiple autoantibodies in adult patients with either established or newly diagnosed, untreated epilepsy.
Eleven percent of patients had antibodies to one or more antigen: voltage-gated potassium channel (VGKC) complex proteins (5%), glycine receptors (3%), and glutamic acid decarboxylase (GAD) and N-methyl-d-aspartate (NMDA) receptors (1.7% each). There was no difference in the prevalence of antibodies, individually or collectively, between patients with established and newly diagnosed epilepsy or with generalized or focal epilepsy. There was, however, a significantly higher prevalence of positive antibody titers in patients with focal epilepsy of unknown cause than in those with structural/metabolic focal epilepsy (14.8% vs. 6.3%; p < 0.02). Newly diagnosed antibody-positive patients were less likely to achieve adequate seizure control with initial treatment than antibody-negative patients, but this difference failed to reach statistical significance.
The presence of autoantibodies is equally common in newly diagnosed and established epilepsy, it is therefore unlikely to be an epiphenomenon of long-standing refractory seizures.
With the exception of rare Mendelian seizure disorders that are explained by point mutations in the genes encoding voltage- and ligand-gated ion channels (Rees, 2010), the molecular etiology of most epilepsies remains unclear. We and others have postulated that an autoimmune attack on protein components of the central nervous system (CNS) could underlie some epileptic disorders of currently unexplained causation (Palace & Lang, 2000; Kullmann, 2010; Irani et al., 2011a,b).
There are now several recognized neurologic disorders associated with serum antibodies to neurologic proteins, many of which present with seizures. Antibodies to the voltage-gated potassium channel (VGKC)-complex have been identified in patients with limbic encephalitis (LE; Thieben et al., 2004; Vincent et al., 2004; Irani et al., 2008, 2010a), and more recently in patients with faciobrachial dystonic seizures (FBDS; Irani et al., 2011a,b), antecedent to the onset of amnesia and disorientation. These encephalopathies often display a monophasic course with antibody titers that decline substantially over 1–2 years (Buckley et al., 2001). Patients typically respond poorly to treatment with conventional antiepileptic drugs (AEDs), often experiencing heightened adverse effects, but they respond well to immunotherapies (Irani et al., 2011a,b).
Serum antibodies to neurologic proteins are relatively common in patients presenting with acute seizures of suspected autoimmune origin, as determined by inflammatory changes in cerebrospinal fluid or on neuroimaging (Quek et al., 2012). Antibodies to VGKC complex proteins and to glutamic acid decarboxylase (GAD) have also been reported in a small proportion of patients presenting with seizures as the main or sole symptom and without overt autoimmune involvement (Kwan et al., 2000; McKnight et al., 2005; Majoie et al., 2006; Niehusmann et al., 2009). The majority of such reports have been cross-sectional in nature, often recruiting patients from tertiary referral centers and thereby biasing the study population in favor of those with an extended history of predominantly refractory epilepsy. Therefore, it remains unclear whether the elevated antibody titers observed in these studies are the underlying cause of the epilepsy or simply a consequence of uncontrolled seizures and any associated neurologic damage. Recently, antibodies to other brain-expressed proteins, including N-methyl-d-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPA), γ-aminobutyric acid receptor B (GABAB), and glycine receptors (GLY-Rs), have been additionally reported in patients with encephalopathies (Graus et al., 2010; Bien & Scheffer, 2011; Vincent & Crino, 2011), but the prevalence of these novel antibodies remains relatively unknown in larger cohorts of sporadic epilepsy. Whether they coexist with more widely reported antibodies, such as VGKC and GAD, in a seizure disorder with broad autoimmunity is equally unclear.
The aim of this study was to determine the prevalence of various established and novel autoantibodies in two large cohorts of people with epilepsy and to compare that prevalence in terms of demographics, clinical characteristics, and treatment outcome.
The study population comprised two distinct cohorts of adults (aged 16 years and over) with either established epilepsy (n = 235; designated the clinic cohort [CC]) or new-onset epilepsy (n = 181; designated the newly diagnosed cohort [NDC]). Subjects with a history of alcohol or recreational drug abuse, suspected nonepileptic seizures, or a progressive neurologic disorder were specifically excluded. Appropriate ethics approval was obtained for the collection of serum samples and related clinical information for both cohorts, and written, informed consent was obtained from all patients.
The CC consisted of 235 consecutive patients with established epilepsy identified on admission to or attending specialist epilepsy clinics at two large teaching hospitals (John Radcliffe Hospital, Oxford, United Kingdom, and Royal Hallamshire Hospital, Sheffield, United Kingdom). Serum samples for the purposes of antibody analysis were obtained at routine outpatient appointments and relevant clinical information was extracted from hospital notes.
Newly diagnosed cohort
The NDC comprised a consecutive series of 181 patients with newly diagnosed, untreated epilepsy who had enrolled in a randomized, open-label trial of levetiracetam, lamotrigine, and topiramate monotherapy at the Epilepsy Unit, Western Infirmary, Glasgow, United Kingdom (Sills et al., 2009). Participants provided a venous blood sample at randomization (prior to initiation of first ever AED) and remained in the trial until failure of their assigned medication. All relevant clinical information was collected routinely in the conduct of the trial. A provisional assessment of treatment response was undertaken in NDC patients as follows: patients who had remained free from seizures for the first 6 months of the trial were considered responders, patients who continued to experience seizures during the first 6 months of the trial despite continuous exposure to their assigned medication at minimal therapeutic dose were considered nonresponders, and patients who failed to meet either of these criteria (as a result of drug change, inadequate dose, insufficient duration of follow-up, and so on) were considered unclassifiable and excluded from the analysis of treatment outcome. Any seizures occurring in the first 6 weeks of the trial were discounted to allow for differences in titration rate between the three trial medications.
A series of control patients were also included in the analysis for the purposes of comparison; these included healthy volunteers (n = 30), patients with other neurologic conditions such as stroke (n = 98), and patients with known autoimmune diseases (n = 20). Some controls have been reported previously in McKnight et al. (2005) but were reanalyzed in all of the assays for this study.
Serum samples were assayed as previously described (McKnight et al., 2005; Watson et al., 2005). VGKC-complex antibodies were determined by immunoprecipitation of 125I-α-dendrotoxin (DTX)–labeled rabbit whole brain extract, antibodies to voltage-gated calcium channels (VGCCs) were assayed using 125I-ω-conotoxin MVIIC-labeled rabbit cerebellar extracts, and α7-acetylcholine receptor (AChR) antibodies were determined using 125I-α-bungarotoxin–labeled extract derived from a clonal neuroblastoma cell line SH-EP1 hα7 that expresses the human α7-AChR (Watson et al., 2005). Samples were considered positive for VGKC, VGCC, or α7-AChR antibodies if the individual titer exceeded a concentration of 100 pm, recognized as three standard deviations above the mean titer observed in healthy controls (n = 30; McKnight et al., 2005; Watson et al., 2005). GAD antibodies were measured using a commercial radioimmunoassay (RSR Ltd., Cardiff, United Kingdom). Serum samples with GAD antibodies exceeding 10 units/ml were considered positive and tested at serial dilutions to obtain accurate titers.
The constructs used in cell based assays (CBAs) were as follows: for the NMDA-R assay we transfected cells with DNA encoding both NR1 and NR2B subunits as follows, enhanced green fluorecent protein (eGFP)-tagged NR1 (GRIN-1a) in pMo2-receiver, untagged NR2B (GRINB) in pcDNA3.1 hygro. For other CBAs the following vectors were used: eGFP-tagged glycine receptor α-subunit (GLRA1) in pcDNA3.1, both M1/M23 isoforms of AQP4 in pEGFP-C3 (Clontech, Oxford, United Kingdom) pEGFP-tagged contactin-associated protein-like 2 (CASPR2) (CNTNAP2) in pcDNA3.1 and leucine-rich, glioma inactivated 1 (LGI1) fused to the transmembrane region of CASPR2 in pcDNA3.1 as described previously (Hutchinson et al., 2008; Waters et al., 2008; Irani et al., 2010a,b, 2011a,b).
Human embryonic kidney (HEK293T) cells were grown on glass cover slips in six-well plates and maintained at 37°C in standard growth medium (Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin, streptomycin, and amphotericin, all GIBCO, Life Technologies, Paisley, United Kingdom). The cells were transiently transfected with constructs encoding the genes of interest using Lipofectamine 2000 (Invitrogen, Life Technologies, Paisley). For NMDA-R transfected cells, ketamine (500 μm) was added to the culture medium to prevent cytotoxicity due to calcium influx. On day 2 posttransfection, the transfected HEK cells were incubated with patient sera (1:20 dilution) for 1 h at room temperature, followed by fixation in 3% formaldehyde in phosphate-buffered saline (PBS) for 10 min and incubation with Alexa Fluor 568 goat anti-human IgG (Invitrogen) for 45 min. The cells were washed with PBS and the cover slips mounted on microscope slides using fluorescence mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI). The cells were analyzed on a fluorescence microscope (Leica DM2500, Leica Microsystems GmbH, Wetzlar, Germany) by two independent observers employing a semiquantitative scoring system ranging from 0 to 4 (Waters et al., 2008; Irani et al., 2010a,b); results were considered positive if scoring >1 in NMDA-R, AQP4, LGI1, and CASPR2 assays and ≥2 in the GLY-R assay.
The prevalence of neurologic antibodies (individually and collectively) was compared between all patients with epilepsy and nonepilepsy controls and also between CC and NDC patients. Further analyses were undertaken in both CC and NDC patients, with cohorts combined as necessary, to investigate the relative prevalence of antibodies in relation to demographics, clinical characteristics, and treatment outcome. Antibody results were expressed as relative prevalence throughout and analyzed statistically by chi-square test or Fisher's exact test, as appropriate. Statistical comparisons of demographics, clinical characteristics, and treatment outcome were undertaken by Mann-Whitney test (continuous variables) or chi-square test (categorical variables), as appropriate. All analyses were performed using Minitab for Windows (version 15, Minitab Software, State College, PA, U.S.A.) and/or GraphPad Prism (version 5, GraphPad Software, LaJolle, CA, U.S.A.).
A total of 416 people with epilepsy (CC = 235, NDC = 181) plus 148 controls (30 healthy volunteers, 98 neurologic controls, 20 autoimmune controls) were each tested for serum antibodies to VGKC, VGCC, GAD, NMDA-R, and GLY-R. A comparison of the CC and NDC patients is illustrated in Table 1. There were significantly more male patients in the NDC compared to the CC (p = 0.013). Age at enrollment into the study was similar between the two cohorts, but age at first-ever seizure was significantly higher in the NDC (p < 0.001), and significantly more patients in the NDC had unclassified epilepsy compared to focal epilepsy (p < 0.001). This is to be expected in a population of patients given a very recent diagnosis of epilepsy in an adult epileptology service.
Statistical comparisons were performed by Mann-Whitney test or chi-square test, as appropriate.
Median age at first seizure, years (range)
Median age at study enrollment, years (range)
Epilepsy type at enrollment
A total of 46 epilepsy patients (CC and NDC combined) had serum antibodies to one or more of VGKC, NMDA-R, GAD, or GLY-R, a prevalence that was significantly (p < 0.001) higher than in the combined control cohort (Table 2). Full details of the antibody-positive patients are given in supplementary Table S1 (Table S1a NDC, Table S1b CC). The most common antibody was to the VGKC-complex, found in 20 epilepsy patients (4.8% of total; titer range 110–500 pm) and only one of the controls (0.7% of total), a stroke patient with a titer of 108 pm (Fig. 1; Table 2). Antibodies were also detected against GLY-R (2.6% of epilepsy patients), NMDA-R (1.7% of epilepsy patients), and GAD (1.7% of epilepsy patients), none of which were observed in the control cohort (Fig. 1; Table 2). One individual from CC had elevated titers to two different antibodies (VGKC and GLY-R) but without remarkable clinical features. A significant difference (p = 0.0271, Fisher's exact test) was observed in the prevalence of GLY-R antibodies between the two cohorts, this was lost after correction for multiple testing. There were no further significant differences in the relative prevalence of any of the other positive antibody titers, either collectively or individually, between CC and NDC patients (Table 2).
Table 2. Presence of neurologic autoantibodies in cases and controls
None of the epilepsy patients (n = 416) or controls (n = 148) expressed antibodies to VGCC (Table 2). A random selection of CC and NDC patients were additionally investigated for the presence of antibodies to AQP4 (n = 90) and/or α7-AChR (n = 50), principally for the purposes of assay development and validation. None of these analyses was positive (data not shown).
All CC and NDC patients who had positive titers for VGKC (n = 20) plus 74 VGKC-negative patients (selected at random from CC and NDC) and 30 healthy controls were additionally tested for antibodies to the VGKC-complex associated proteins LGI1 and CASPR2 by CBA. Only one patient, with a positive VGKC-complex titer of 111 pm, had LGI1 antibodies. All other patients and controls were negative for LGI1 and all tests for CASPR2 were similarly negative. The LGI1 positive patient was a 61-year-old man with a 2-year history of focal dyscognitive seizures with oral automatisms (lip smacking) lasting a small number of minutes. He was subsequently retested and found to be negative for VGKC-complex and LGI1 antibodies following 4 months of treatment with controlled-release carbamazepine 400 mg twice daily.
There was no association between the presence of neurologic autoantibodies and age at enrollment and no difference in the relative prevalence of positive titers between males and females, when the CC and NDC cohorts were combined (Tables 3 and 4). Likewise, there was no difference in the presence of antibodies between patients with generalized, focal, or unclassified epilepsy, although all seven patients with elevated GAD antibodies (>100 units/ml) had focal epilepsy (Table 3). Further analysis of patients with focal epilepsy revealed a significantly higher prevalence of positive antibody titers in patients with focal epilepsies of unknown cause compared to patients with known structural/metabolic focal epilepsy (p < 0.02; Table 3). The duration of epilepsy at enrollment in the CC was not associated with the prevalence of elevated antibody titers, nor was the time since most recent seizure at enrollment in the NDC (p = ns, Table 4).
Table 3. Prevalence of neurologic autoantibodies by categorical demographic and clinical characteristics
Results are expressed as the number of samples showing a positive titer in clinic and newly diagnosed cohorts combined. Statistical comparisons were performed by chi-square test or Fisher's exact test, as appropriate.
Etiology was explored in patients with focal epilepsy alone.
Table 4. Prevalence of neurologic autoantibodies by continuous demographic and clinical characteristics
Median age at enrollment in years (range; CC + NDC)
Median duration of epilepsy in years (range; CC only)
Median time since most recent seizure in days (range; NDC only)
Results are expressed as the median value (range) in relation to date of enrollment for all antibodies combined. Age was investigated in the clinic cohort (CC) and newly diagnosed cohorts (NDCs) combined. Duration of epilepsy was investigated in the CC only. Time since most recent seizure was investigated in the NDC only. Statistical comparisons were performed by Mann-Whitney test.
The final analysis explored antibody status at diagnosis and its potential relationship with subsequent response to AED therapy. It was performed solely in the NDC who were participating in a prospective, randomized trial of AED monotherapy, with patients designated as responders or nonresponders to treatment as described above. Individuals with an elevated antibody titer at baseline (irrespective of antibody) were more than twice as likely (16.4% vs. 7.9%) to be unresponsive to initial drug treatment, when measured at 6 months after AED initiation, than those who were antibody negative (Table 5). This observation, however, failed to reach statistical significance (p = 0.18). There were no significant differences in seizure-free rates at 6 months amongst the three AEDs under investigation in the monotherapy trial (levetiracetam, lamotrigine, topiramate; data not shown), ruling out any potentially confounding influence of treatment arm.
Table 5. Presence of neurological autoantibodies and response to first ever antiepileptic drug in newly-diagnosed cohort
Statistical comparison of antibody positives in responder group compared to the non-responders was performed by Fisher's exact test.
Results are expressed as the number of samples showing a positive titre.
Previous studies have shown the presence of autoantibodies to VGKC complexes and GAD in cross-sectional cohorts of patients with epilepsy (Kwan et al., 2000; McKnight et al., 2005; Errichiello et al., 2009; Liimatainen et al., 2010; Irani et al., 2011a,b), in well-defined epilepsy syndromes (Irani et al., 2011a,b), and in patients with known or suspected preexisting neurologic autoimmunity (Quek et al., 2012). Antibodies to VGCC and GLY-R have never, to our knowledge, been systematically screened in the epilepsy population, and those to NMDA-R have only been investigated in two small pediatric studies, focused on LE and status epilepticus (Haberlandt et al., 2011; Suleiman et al., 2011). We looked for VGKC, VGCC, GAD, NMDA-R, and GLY-R antibodies in cohorts of consecutive, unselected patients with established epilepsy attending specialist clinics in Oxford and Sheffield and with newly diagnosed epilepsy taking part in a monotherapy trial in Glasgow.
We found positive antibodies to VGKC, GAD, NMDA-R, and GLY-R at a prevalence that was significantly higher than in either healthy or disease controls. In total, 46 (11%) of 416 patients with epilepsy in this analysis were antibody positive, with one patient showing an elevated titer to multiple antigens (VGKC and GLY-R). Prevalence of positive titers was comparable in patients with established and newly diagnosed epilepsy. There was no association with age, sex, epilepsy type, duration of epilepsy, or time since most recent seizure. There was a modest relationship with focal epilepsies of unknown cause and a trend toward AED unresponsiveness in patients with elevated antibody titers at diagnosis.
None of the patients in this study had clinical evidence of LE, NMDA-R–associated encephalitis, or any other antibody-mediated neuroinflammatory disease. As such, our findings suggest that the various neurologic autoantibodies associated with subacute onset encephalopathies may also give rise to sporadic epilepsy, similar to previous reports in a cohort of young female patients with unexplained new-onset epilepsy (Niehusmann et al., 2009). All seven patients with GAD antibodies had focal epilepsy, whereas six were female, which is consistent with a number of other studies (Kwan et al., 2000; Irani et al., 2011a,b) and more recent reports of a relationship between GAD antibodies and epilepsies arising in the temporal lobe (Peltola et al., 2000; Errichiello et al., 2009; Liimatainen et al., 2010; Irani et al., 2011a,b). Similarly, GAD antibodies have also recently been reported in a form of LE in young adult females with temporal lobe epilepsy and mild cognitive involvement (Malter et al., 2010). Five (8.6%) of the 58 patients with a diagnosis of generalized epilepsy had autoantibodies (three VGKC, two GLY-R), which is consistent with the relative prevalence of antibodies across our study as a whole. Patients with generalized epilepsies are usually presumed to have a genetic basis for their seizures, although in most cases this has not been formally confirmed. This issue is addressed in the recent revision to the terminology and concepts for organization of seizures and epilepsies (Berg et al., 2010). Autoantibodies directed against the same ion channels and receptors often implicated in genetic epilepsies could theoretically give rise to a clinically identical phenotype. Further studies on larger cohorts of generalized epilepsy patients to explore this possibility are thus warranted, and future revisions to the classification of epilepsies should consider autoimmunity as a significant etiologic contributor.
Uniquely, we tested 181 patients with a new diagnosis of epilepsy prior to the onset of treatment with AEDs and found a trend between antibody positivity and subsequent lack of response to the first-ever AED treatment. This is a preliminary finding and one that merits longer term follow-up in terms of drug responsiveness and further investigation to rule out potentially confounding influences on outcome. For example, it is possible that antibody-negative patients may have had a less acute and severe presentation, with lower pretreatment seizure number or density and an accordingly better prognosis. Nevertheless, this apparent lack of response to AEDs has been noted previously in VGKC-complex antibody-positive patients (Errichiello et al., 2009; Irani et al., 2011a,b) and there are increasing reports of VGKC antibody–positive patients who respond better to immunomodulatory therapy than to conventional AEDs (Lancaster et al., 2011; Quek et al., 2012). In a recent report of patients presenting with focal seizures and in whom an autoimmune etiology was suspected, 80% responded favorably to immunomodulation (Quek et al., 2012). Likewise in FBDS, a clinically distinctive syndrome associated with autoantibodies against LGI1 (Irani et al., 2010a) in which patients present with frequent brief, multiple episodes of facial and ipsilateral arm dystonia, electroencephalography (EEG) evidence of seizures, and progression to LE (Irani et al., 2008; Errichiello et al., 2009), few patients respond to AEDs, whereas immunotherapies have shown benefit (Irani et al., 2008; Errichiello et al., 2009).
Although the pathogenicity of the antibodies reported here has not been formally demonstrated, VGKC, GAD, NMDA-R and GLY-R antibody positive sera or immunoglobulin G (IgG) preparations thereof have all been shown to exert functional effects on neuronal tissues (Irani et al., 2010a,b; Lalic et al., 2011). As such, there are good reasons to believe that the antibodies detected in this study have the potential to be clinically significant, particularly when present early in the course of the disorder. Future immunologic screening studies should seek to preferentially include prospective cohorts of patients with new or recent-onset epilepsy to further understand the contribution of autoantibodies to the pathophysiology of epilepsy, to promote prompt and accurate diagnosis, and to encourage the consideration of alternative treatment options, including the possible use of immunotherapies.
The authors are most grateful to Professor Peter Rothwell for providing control sera and to Professor David Beeson for providing the construct GRIN-1a/pMo2-receiver. The study was supported by research awards from Epilepsy Research UK, NHS Greater Glasgow (West Research Endowment Funds), and by NIHR Oxford Biomedical Research Centre.
AV, PW, BL, and the Nuffield Department of Clinical Neurosciences in Oxford receive royalties and payments for antibody assays. TB, GJS, SH, MJB, and YH do not report any conflict of interest in respect of this study. We confirm that the authors have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.