Epilepsy susceptibility was consistently heightened in patients with autoimmune diseases (P < .001). Collectively, patients with any of the autoimmune diseases under study constituted 17.5% of the total epilepsy population. OR indicates odds ratio; SLE, systemic lupus erythematosus. Data markers indicate ORs and limit lines, 95% CIs.
Overall, children with an autoimmune disease had a 5-fold increased risk of epilepsy (P < .001 in all cases except otherwise indicated). OR indicates odds ratio; SLE, systemic lupus erythematosus. Data markers indicate ORs and limit lines, 95% CIs.aP = .006.bP = .008.
Overall, adults with an autoimmune disease had a 4-fold increased risk of epilepsy (P < .001 in all cases). OR indicates odds ratio; SLE, systemic lupus erythematosus. Data markers indicate ORs and limit lines, 95% CIs.
eTable 1. Medication list
eTable 2.ICD-9 codes used for identifying diagnoses
Ong M, Kohane IS, Cai T, Gorman MP, Mandl KD. Population-Level Evidence for an Autoimmune Etiology of Epilepsy. JAMA Neurol. 2014;71(5):569-574. doi:10.1001/jamaneurol.2014.188
Epilepsy is a debilitating condition, often with neither a known etiology nor an effective treatment. Autoimmune mechanisms have been increasingly identified.
To conduct a population-level study investigating the relationship between epilepsy and several common autoimmune diseases.
Design, Setting, and Participants
A retrospective population-based study using claims from a nationwide employer-provided health insurance plan in the United States. Participants were beneficiaries enrolled between 1999 and 2006 (N = 2 518 034).
Main Outcomes and Measures
We examined the relationship between epilepsy and 12 autoimmune diseases: type 1 diabetes mellitus, psoriasis, rheumatoid arthritis, Graves disease, Hashimoto thyroiditis, Crohn disease, ulcerative colitis, systemic lupus erythematosus, antiphospholipid syndrome, Sjögren syndrome, myasthenia gravis, and celiac disease.
The risk of epilepsy was significantly heightened among patients with autoimmune diseases (odds ratio, 3.8; 95% CI, 3.6-4.0; P < .001) and was especially pronounced in children (5.2; 4.1-6.5; P < .001). Elevated risk was consistently observed across all 12 autoimmune diseases.
Conclusions and Relevance
Epilepsy and autoimmune disease frequently co-occur; patients with either condition should undergo surveillance for the other. The potential role of autoimmunity must be given due consideration in epilepsy so that we are not overlooking a treatable cause.
Epilepsy is a debilitating condition affecting 0.5% to 1.0% of the world’s population. Therapies address manifestations rather than the underlying etiology, which remains unknown in most patients, one-third of whom have a condition that is refractory to antiepileptic therapy.1 Surgical interventions in epilepsy are often ineffective, with seizures recurring in 50% of patients within 5 years of surgery,2 and the number of patients who remain seizure free decreases further over the years to 38%.3 A deeper understanding of the underlying etiologies is necessary to develop new therapeutic approaches.
Specific autoimmune causes, typically associated with autoantibodies, have been increasingly identified in a subset of previously idiopathic seizure disorders.4- 10 In some of these situations, seizures are associated with other neurologic manifestations; in others, they are the only sign of neurologic autoimmunity. Small case studies and disease-specific investigations also report a high incidence of seizures in autoimmune diseases (ADs) such as systemic lupus erythematosus (SLE)11,12 and Hashimoto thyroiditis.13,14 Furthermore, published reports15 document success with immunotherapy in a substantial proportion of patients with presumed autoimmune mechanisms for their seizures.
Establishing an autoimmune basis in patients with idiopathic epilepsy is important because it highlights opportunities for developing new strategies for the treatment of medically refractory epilepsy. To date, evidence on the role of the autoimmune process in epileptogenesis is based mainly on animal studies and small-sample, disease-specific clinical observations. We conducted what we believe to be the largest population study to investigate the relationship between epilepsy and several common autoimmune diseases. Because clinical presentation of seizures, their etiology, and the presence of comorbidities in the elderly population differ considerably from those in younger patients, the present study focused on epilepsy in children (<18 years) and nonelderly adults (aged ≤65 years).
We conducted a population-based retrospective cohort study using claims data from a major nationwide employer-provided health insurance plan in the United States. The Boston Children’s Hospital Institutional Review Board approved the study and granted a waiver of consent. Data included dates of enrollment in the insurance program, outpatient and inpatient visits, and prescription drugs dispensed. Demographic data included sex and age. All encounters were coded with 4 or fewer International Classification of Diseases, Ninth Edition (ICD-9), codes. Prescription drugs were reported using the National Drug Code.
Participants were beneficiaries between January 1, 1999, and December 31, 2006, excluding adults older than 65 years. To ensure adequate follow-up, we included only individuals continuously enrolled for 4 years or more, and we considered only those with epilepsy diagnosed 2 or more years after entry into our study and with at least 2 years’ follow-up after the first recorded epilepsy diagnosis.
We assessed the relationship between epilepsy and 12 ADs selected a priori: type 1 diabetes mellitus, psoriasis, rheumatoid arthritis, Graves disease, Hashimoto thyroiditis, Crohn disease, ulcerative colitis, SLE, antiphospholipid syndrome, Sjögren syndrome, myasthenia gravis, and celiac disease. Outcomes were stratified based on age groups: (1) children (<18 years) and (2) nonelderly adults (≤65 years).
Furthermore, we examined the potential effects of common therapies used for treating AD including aminosalicylates, disease-modifying antirheumatic drugs, systemic glucocorticoids, nonsteroidal anti-inflammatory drugs (NSAIDs), anti–tumor necrosis factor agents, and other biologics (Supplement [eTable 1]). Only exposures that occurred before the first epileptic seizure were considered. Sex and age were included as covariates. Exposure to medications was expressed as a dichotomous variable.
Epilepsy was defined as 2 or more seizures occurring at least 24 hours apart within 2 years. Individuals with diagnoses of epilepsy and AD were identified using ICD-9 diagnostic codes (Supplement [eTable 2]) according to previously validated criteria16: (1) at least 1 acute inpatient encounter with the relevant ICD-9 code as the primary diagnosis or (2) at least 2 health care encounters with the relevant ICD-9 code within 2 years. These criteria have been demonstrated16 to achieve high accuracy in identifying patients from administrative data (sensitivity, 92.9%; specificity, 91.2%). To further strengthen the specificity of epilepsy case identification, we considered only individuals prescribed at least 1 course of an antiepileptic medication.
The risk of epilepsy in patients with AD was compared against the risk of epilepsy in individuals without AD using logistic regression, expressed as odds ratios (ORs) with 95% CIs. All analyses were performed using SPSS software, version 21 (IBM). All statistical tests were 2-sided.
A total of 2 518 034 individuals were included in our study; 0.4% of the study population developed epilepsy (Table 1). The risk of epilepsy was significantly heightened among patients with AD (OR, 3.8; 95% CI, 3.6-4.0; P < .001) (Figure 1). Collectively, individuals with AD accounted for 17.5% of patients with epilepsy in the study population.
Elevated risk was consistent across different ADs. Patients with antiphospholipid syndrome and SLE had the highest risk, with a 9-fold and 7-fold increased risk of epilepsy, respectively, followed by patients with type 1 diabetes and myasthenia gravis, eliciting a 5-fold increased risk of epilepsy. The risk of epilepsy was especially pronounced in children. Overall, children with AD had a 5-fold increased risk of epilepsy (Figure 2). In comparison, nonelderly adults with AD had a 4-fold increased risk of epilepsy (Figure 3).
The onset of epilepsy preceded the diagnosis of AD in 30% of cases. In 30% of the cases, the first epileptic seizure occurred within the first year after the AD diagnosis.
More than 70% of patients with AD and epilepsy were not exposed to antiepileptics for at least 2 years before the diagnosis of AD. The risk of epilepsy in patients with AD persisted after adjusting for medication use, including aminosalicylates, disease-modifying antirheumatic drugs, systemic glucocorticoids, NSAIDs, anti–tumor necrosis factor agents, and other biologics (OR, 4.1; 95 CI%, 3.9-4.3; P < .001) (Table 2). There was no evidence that any of the medications led to an increased risk of epilepsy. Patients exposed to aminosalicylates, NSAIDs, anti–tumor necrosis factor agents, and other biologics appeared to have a reduced risk of epilepsy.
Clinicians caring for patients with either AD or epilepsy should be aware of the strong association between them. Indeed, nearly 1 in 5 patients with epilepsy has a coexisting AD. Elevated epilepsy prevalence has been previously reported11,12,17- 19 in ADs in which the disease directly involves the brain. Systemic lupus erythematosus is associated with a range of inflammatory mechanisms in the brain, and cerebral ischemia is a common manifestation of antiphospholipid syndrome. Rates of epilepsy in SLE vary between 4% and 51%11,12,17 and in antiphospholipid syndrome from 3% to 8%.18,19 Our analysis is consistent with these published data. In addition, we established the association across a wide range of ADs, including those for which the primary biological mechanism is not known to directly affect the brain. Patients with myasthenia gravis had a 5-fold increased risk of epilepsy. Although our data do not elucidate the underlying causes of this relationship, they strongly support further effort to explore the potential role of autoimmunity in epileptogenesis. A focus on autoimmune mechanisms can guide translational approaches to new therapeutic options.
Seizures tend to occur within the first 1 to 2 years after the AD diagnosis. The risk of epilepsy is consistently higher in children with AD compared with adults with the same AD. Both clinical and biological features of AD may be influenced by the patient’s age at disease onset,20,21 with childhood-onset AD often more severe than adult-onset AD.22,23 Consistent with a previous study,11 our data showed that female sex is associated with a higher risk of epilepsy.
Prior studies on the effects of antiepileptics have produced inconsistent results. Antiepileptics have been reported to exert anti-inflammatory effects,24 but there also are reported cases of SLE-like symptoms caused by carbamazepine.25 However, evidence for causation remains largely anecdotal and inconsistent, with other studies26 failing to show a relationship between antiepileptic use and SLE. Our main findings are clearly not explained solely by immunologic adverse effects of antiepileptics, with 70% of patients with AD and epilepsy not exposed to antiepileptic medications for at least 2 years prior to the diagnosis of AD.
We also explored whether specific treatments for AD may cause or prevent seizures. Corticosteroids, certain immunologic agents, and NSAIDs have been found to induce seizures in some studies27- 29 and to reduce the risk of seizures in others.30- 32 We found the risk of epilepsy in patients with AD persisted, even after adjustment for these medications in regression models. Exposure to aminosalicylates, NSAIDs, and biologics appears to reduce the risk for epilepsy. However, because we did not study the duration and dose of exposure, whether these medications confer a protective effect against developing epilepsy is unknown and beyond the scope of the present study.
Among the growing list of neuronal autoantibodies identified in a subgroup of patients with epilepsy, some appear to play a pathogenic role while others may merely be markers of disease. For example, compelling clinical and laboratory evidence33- 36 support the pathogenicity of antibodies against the NR1 subunit of the N-methyl-d-aspartate receptor (NMDAR). Antibodies from patients with anti-NMDAR encephalitis cause a decrease in the density of NMDAR through antibody-mediated cross-linking and internalization, resulting in the impairment of NMDAR-mediated synaptic function.33,34 Clinical outcome of these patients has been found35,36 to correlate with antibody titers in cerebrospinal fluid, and in most cases, symptoms are reversible by immunotherapy. The pathogenic role of other autoantibodies, particularly those directed against intracellular antigens such as glutamic acid decarboxylase, is less clear.
The immune response in AD involves the adaptive immune system, for which the presence of autoimmune antibodies is just one manifestation, as well as the innate arm, as characterized by the increased synthesis and release of proinflammatory cytokines and chemokines. Thus, the occurrence of epilepsy in patients with AD might be attributable to the inflammatory component of AD, as evidenced by the expression of sustained inflammatory response in the resected brain tissue from patients undergoing surgery for refractory epilepsy, including activation of microglia and astrocytes and production of proinflammatory molecules.37- 39 Additionally, there is evidence of the anticonvulsant effects of selected anti-inflammatory drugs40,41 and the anti-inflammatory properties of some antiepileptic medications, such as valproate42 and carbamazepine.43
Although autoimmunity and neuroinflammation are likely to play a role in a subset of epilepsy patients with AD, seizures may also be a result of the cerebrovascular complications that are commonly associated with many ADs, including SLE, rheumatoid arthritis, Sjögren syndrome, type 1 diabetes, and inflammatory bowel disease. Thus, the risk of seizures in these patients is heightened independent of immunologic causes. However, given that susceptibility to cerebrovascular conditions correlates strongly with advancing age, our findings showing children with AD at a much higher risk of epilepsy compared with adults with the same autoimmune condition suggest that the relationship cannot be entirely attributable to the cerebrovascular complications secondary to AD. Furthermore, the association between epilepsy and ADs that do not directly affect the brain, including myasthenia gravis and psoriasis, strongly implicates the hypothesis that other mechanisms are involved.
In reality, epilepsy is not a single disease entity but a variety of disorders reflecting the underlying brain dysfunction that may result from many different causes. Similarly, it is likely that multiple factors contribute to the risk of epilepsy in patients with AD. First, some patients with AD may have coincidental epilepsy that is unrelated to autoimmune mechanisms; we have accounted for this via comparison with the non-AD population. Second, some patients with AD may have noninflammatory brain abnormalities that give rise to epilepsy; although this is likely to be true, we do not think it fully explains the heightened risk as described above. Third, some patients with AD and epilepsy may have autoantibodies, such as NMDAR antibodies. Finally, we speculate that the largest subgroup consists of patients with as-yet undefined autoimmune and inflammatory mechanisms that lead to epilepsy. Additional studies are needed to elucidate the pathogenesis of epilepsy in patients with ADs, especially in the latter subgroup. Identifying these mechanisms may also yield insights and novel treatment approaches for patients with epilepsy but without AD, particularly those without a known etiology and/or refractory epilepsy.
There are several limitations to our study. First, claims data have limited resolution and do not permit fine-grained classification of epilepsy type. However, the validity of using ICD codes in administrative data for identifying epilepsy cases has been demonstrated in several studies.16,44,45 To maximize the specificity of our case identification, we chose a stringent case definition for diagnosis of epilepsy that has been validated.16 In addition, we selected only patients with epilepsy who were given prescriptions for antiepileptic medications, thus minimizing the likelihood of misclassification of nonepileptic seizures, such as those induced by hypoglycemia, alcohol, or drugs. The consistency between our results and published data on the relationships between epilepsy and several ADs, including SLE and antiphospholipid syndrome, further validates our approach. Second, because our study participants were privately insured, findings from these data may not generalize to other populations. Third, because our data spanned 7 years and the set of ADs included in our study is nonexhaustive, the incidence of epilepsy in patients with AD may be underestimated.
Although there are limitations to claims data, the availability of a large number of patients makes it possible to study the relationships between rare diseases, which may not have been observable in traditional studies involving medical record reviews or surveys. Furthermore, claims data are systematically collected and provide longitudinal information that crosses facilities, geographic locations, and population demographics, thereby enhancing the generalizability of the research and limiting selection biases. Although our study does not prove that epilepsy and AD share a common pathophysiology, which remains to be elucidated, our data provide important epidemiology evidence that lends support to the hypothesis and highlights the need for in-depth investigation.
Epilepsy and AD frequently co-occur, and patients with either condition should undergo surveillance for the other. Because ADs affect 8% of the population and for reasons unknown their prevalence is rising, the link between autoimmunity and epilepsy will underpin a rise in the global burden of neurologic disease. The potential role of autoimmunity must be given due consideration in refractory epilepsy so that we are not overlooking a treatable etiology.
Accepted for Publication: January 31, 2014.
Corresponding Author: Kenneth D. Mandl, MD, MPH, Intelligent Health Laboratory, Harvard Medical School, 300 Longwood Ave, Boston, MA 02115 (firstname.lastname@example.org).
Published Online: March 31, 2014. doi:10.1001/jamaneurol.2014.188.
Author Contributions: Dr Ong had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Ong, Mandl.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Ong.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Ong, Kohane, Cai.
Obtained funding: Mandl.
Administrative, technical, or material support: Mandl.
Study supervision: Mandl.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Ong is supported by a fellowship from the National Health and Medical Research Council, Australia (grant APP1052871). Dr Kohane was funded in part by National Institutes of Health (NIH)/National Library of Medicine grant U54 LM008748 from the i2b2 National Center for Biomedical Computing and by grant NIH P50MH94267 from the Conte Center for Computational System Genomics of Neuropsychiatric Phenotypes. Dr Mandl was funded in part by NIH National Institute of General Medical Sciences grant R01GM104303.
Role of the Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Marc Natter, MD (Boston Children’s Hospital), provided intellectual input. No financial compensation was given.