Objective The aim of this study was to describe the frequency, attribution, outcome and predictors of seizures in systemic lupus erythematosus (SLE).
Methods The Systemic Lupus International Collaborating Clinics, or SLICC, performed a prospective inception cohort study. Demographic variables, global SLE disease activity (SLE Disease Activity Index 2000), cumulative organ damage (SLICC/American College of Rheumatology Damage Index (SDI)) and neuropsychiatric events were recorded at enrolment and annually. Lupus anticoagulant, anticardiolipin, anti-β2 glycoprotein-I, antiribosomal P and anti-NR2 glutamate receptor antibodies were measured at enrolment. Physician outcomes of seizures were recorded. Patient outcomes were derived from the SF-36 (36-Item Short Form Health Survey) mental component summary and physical component summary scores. Statistical analyses included Cox and linear regressions.
Results The cohort was 89.4% female with a mean follow-up of 3.5±2.9 years. Of 1631 patients, 75 (4.6%) had ≥1 seizure, the majority around the time of SLE diagnosis. Multivariate analysis indicated a higher risk of seizures with African race/ethnicity (HR (CI): 1.97 (1.07 to 3.63); p=0.03) and lower education status (1.97 (1.21 to 3.19); p<0.01). Higher damage scores (without neuropsychiatric variables) were associated with an increased risk of subsequent seizures (SDI=1:3.93 (1.46 to 10.55); SDI=2 or 3:1.57 (0.32 to 7.65); SDI≥4:7.86 (0.89 to 69.06); p=0.03). There was an association with disease activity but not with autoantibodies. Seizures attributed to SLE frequently resolved (59/78 (76%)) in the absence of antiseizure drugs. There was no significant impact on the mental component summary or physical component summary scores. Antimalarial drugs in the absence of immunosuppressive agents were associated with reduced seizure risk (0.07 (0.01 to 0.66); p=0.03).
Conclusion Seizures occurred close to SLE diagnosis, in patients with African race/ethnicity, lower educational status and cumulative organ damage. Most seizures resolved without a negative impact on health-related quality of life. Antimalarial drugs were associated with a protective effect.
Statistics from Altmetric.com
Nervous system disease in systemic lupus erythematosus (SLE) includes neurological and psychiatric events1,–,5 of which 19–38% are attributable to lupus.6 Previous studies clustered individual neuropsychiatric (NP) events into groups to study outcomes and impact on health-related quality of life (HRQoL). Although meritorious, assessment of individual NP manifestations is preferable but requires a large cohort of patients studied over a prolonged period.
Seizures are a manifestation of NPSLE and occur in 6–51%1,–,5 of adult and pediatric patients. Most studies have been cross-sectional and have not assessed patients from the time of diagnosis of SLE, an important issue in view of reports that seizures are most frequent early in the disease course. Here, we report our findings from a large, prospective, international, inception cohort of patients with SLE. Our objectives were to describe the frequency, attribution and outcome of seizure disorders in SLE and to identify clinical and laboratory predictors of their occurrence.
Patients and methods
Research study network
The study was conducted by the Systemic Lupus International Collaborating Clinics (SLICC)7 consisting of 37 investigators in 30 international academic medical centres in 11 countries. Data were collected per protocol at enrolment and annually thereafter, submitted to the coordinating centre in Halifax, Nova Scotia, Canada, and entered into a centralised Access database. Appropriate procedures ensured data quality, management and security. Capital Health Research Ethics Board, Halifax, and each of the participating centre's institutional research ethics review boards approved the study.
Patients fulfilled the American College of Rheumatology (ACR) SLE classification criteria,8 taken as the date of diagnosis, and provided written informed consent. Enrolment was permitted up to 15 months following the diagnosis. Demographic variables included age, gender, ethnicity, education and medication history. Lupus-related variables included the SLE Disease Activity Index 2000 (SLEDAI-2K)9 and SLICC/ACR Damage Index (SDI).10 Laboratory testing included haematology, chemistry and immunology required for SLEDAI-2K and SDI scores.
An enrolment window extended from 6 months prior to the diagnosis of SLE up to the enrolment date. NP events were characterised within this window using the ACR case definitions for NP syndromes,11 diagnosed by clinical evaluation and supported by investigations if clinically warranted as per the guidelines. Patients were reviewed annually with a 6-month window around the anticipated assessment date. New NP events since the previous study visit were determined at each assessment.
Seizures were classified into either primary generalised seizures with four subtypes (tonic clonic, atonic, absence and myoclonic) or focal seizures with two subtypes (simple and complex).11 The diagnosis was confirmed by independent description from a reliable witness, and a normal EEG did not exclude the diagnosis.
Other potential causes (‘exclusions’) or contributing factors (‘associations’) for each NP event11 were sought. These ‘non-SLE factors’ in part determined the attribution of events. For patients with seizures, several exclusions (vasovagal and cardiac syncope, hysteria, hyperventilation, tics, narcolepsy and cataplexy, labyrinthitis, alcohol and drug withdrawal, medications quinolones and imipenem, subarachnoid haemorrhage, hypoglycaemia, panic attacks, conversion disorders, malingering) and associations (thrombotic thrombocytopenic purpura/microangiopathy, stroke, transient ischemic attack, migraine, hypoglycaemia (mild), hypoxemia, uraemia, intracranial tumour and infection) were considered.
Recurrent episodes of the same NP event within the enrolment window or follow-up assessment period were recorded once. The date of the first episode was taken as the onset of the event.
Attribution of NP events
Decision rules were used to determine the attribution of all NP events including seizures. To optimise consistency, this was performed at the coordinating centre using data provided in the case record form by individual SLICC sites. Factors considered in the decision rules included the following: (1) onset of NP event(s) prior to the diagnosis of SLE, (2) concurrent non-SLE factor(s) identified from the ACR glossary for each NP syndrome and (3) ‘common’ NP events that are frequent in normal population controls as described by Ainiala et al.12 These include all headaches, anxiety, mild depression (mood disorders failing to meet criteria for ‘major depressive-like episodes’), mild cognitive impairment (deficits in less than three of the eight specified cognitive domains) and polyneuropathy without electrophysiological confirmation.
Decision rules of different stringency (models A and B) determined attribution of NP events, described in detail elsewhere.6 ,13 NP events that fulfilled criteria for model A (most stringent) or for model B (least stringent) were attributed to SLE. By definition, all NP events attributed to SLE using model A were included in the group of NP events using model B. Events not fulfilling these criteria were attributed to non-SLE causes.
Outcome of seizures
A physician-generated 7-point Likert scale compared the change in seizure activity between onset and study assessment (1=patient demise, 2=much worse, 3=worse, 4=no change, 5=improved, 6=much improved, 7=resolved).14 A patient-generated SF-36 (36-Item Short Form Health Survey) questionnaire provided mental component summary and physical component summary scores.14 ,15 These were not available to the physicians at the time of assessment.
Lupus anticoagulant, IgG anticardiolipin, anti-β2 glycoprotein-I, antiribosomal P and anti-NR2 glutamate receptor antibodies were measured at the enrolment visit at the Oklahoma Medical Research Foundation, USA, using previously described methodology.16,–,19
The associations of explanatory variables with the time to the first occurrence of all seizures and those attributed to SLE (model B) were examined using Cox proportional hazards regression. SF-36 analyses used linear regression and generalised estimating equations with a first-order autoregressive correlation structure to allow for correlation between multiple SF-36 measurements for the same patient (see online supplementary text S1).
A total of 1631 patients were recruited between October 1999 and January 2011. The median (range) number of patients enrolled in SLICC centres was 36 (8–202). Patients were most frequently women (89.4%), with a mean (±SD) age of 35.0±13.4 years and a wide ethnic distribution although predominantly Caucasian (table 1).
At enrolment, the mean disease duration was 5.6±4.8 months and 937 (57.5%) patients had a disease duration of <6 months. The prevalence of ACR classification criteria reflected an unselected patient population, receiving a typical range of lupus medications. The mean SLEDAI-2K and SDI scores indicated moderate global disease activity and minimal cumulative organ damage, respectively. The number of annual assessments varied from 1 to 12 with a mean follow-up of 3.5±2.9 years.
NP (≥1) occurred in 747 of 1631 (45.8%) patients, and 336 of 1631 (20.6%) patients had ≥2 events over the study period. The events and their attribution are summarised in table 2.
There were 1358 NP events, involving all 19 NP syndromes.11 The proportion of NP events attributed to SLE varied from 16.9% (model A) to 29.9% (model B) and occurred in 10.3% (model A) to 17.4% (model B) of patients. Of the 1358 NP events, 1256 (92.5%) involved the central nervous system and 102 (7.5%) involved the peripheral nervous system. The classification of events into diffuse and focal was 1071 (78.9%) and 287 (21.1%), respectively.
Over the study, 75 of 1631 (4.6%) patients had a seizure. In 59 of these 75 (78.7%) patients, seizures occurred during one observation period, and in 16 of these 75 (21.3%) patients, seizures occurred during two observation periods. There were a total of 91 seizures (66% generalised, 34% focal) in the 75 patients, and 9 patients had generalised and focal seizures (table 3).
Of the 91 seizures, 78 (86%) were attributed to SLE. The majority of seizures presented early in the disease course with a median (range) interval from the time of SLE diagnosis to onset of first seizure of 0.14 (−0.50 to 7.57) years.
Seizures and other NP events
Of the 91 seizures, 36 (39%) occurred in the absence of other NP events within the same period of observation, 28 (31%) presented concurrently with other NP events attributed to SLE (model B) and 27 (30%) presented with other NP events not attributed to SLE (model B). Seizures attributed to SLE (model B) were more frequent in the absence of other NP events compared to those not attributed to SLE (34/78 (44%) vs 2/13 (15%)) although statistical significance was not achieved (p=0.07, Fisher's exact test).
Predictors of seizures
Single-factor analyses revealed that female gender, race/ethnicity, lack of postsecondary education, younger age at SLE diagnosis, SDI (total) score, SDI (without NP) score and SDI (without seizures) score were individually associated with a higher risk of first occurrence of seizures (table 4).
Concurrent medications were also jointly associated with the risk of first seizure occurrence. In particular, the current use of corticosteroids and antimalarial drugs was associated with increased and reduced risks, respectively. There was no association with autoantibodies individually (table 4) or after combining lupus anticoagulant, IgG anticardiolipin or anti-β2 glycoprotein-I into a single antiphospholipid (aPL) panel either for all seizures (HR (CI) 1.42 (0.82 to 2.47)) or for seizures attributed to SLE (model B) (1.37 (0.78 to 2.47)). Among 1070 patients with available data for all three antibodies, 101 (9.4%) had >1 aPL and 279 (26.1%) had only 1 aPL. Likewise, there was no association with individual SLEDAI-2K scores. Using the adjusted mean SLEDAI-2K20 (AMS), 9 of 25 events, which occurred prior to the first follow-up assessment, were excluded as computing AMS scores requires a minimum of two SLEDAI-2K scores. Nevertheless, there was evidence of a relationship with seizure occurrence based on these analyses (p=0.07 for AMS (w/o NP variables) and p=0.04 for AMS (w/o seizures)). Similar results were obtained for seizures attributed to SLE (model B) with the exception that the association with SDI scores (without seizures) and current use of antimalarial drugs did not reach statistical significance at the 0.05 level.
Multivariate analysis (table 5, analysis I) indicated a significant relationship with race/ethnicity (global test p value=0.03) with a higher risk of seizures in African race/ethnicity (HR (CI): 1.97 (1.07 to 3.63)). Higher risk was also associated with lack of postsecondary education (1.96 (1.21 to 3.19); p=0.006) after adjustment for gender, race/ethnicity and age at diagnosis, and there was no interaction between race/ethnicity and education (p=0.964). Inclusion of SDI scores in the analysis (table 5, analysis II) reduced the available events for analysis from 70 to 20 as SDI scores cannot be computed in patients with <6 months disease duration. Nevertheless, when SDI scores were included, postsecondary education remained an important predictor and the association with race/ethnicity was reduced. Adjusting for gender, race/ethnicity, age at diagnosis and postsecondary education, patients with higher SDI scores (without NP) were associated with an increased risk of subsequent seizures, reaching statistical significance when SDI scores were treated both as a continuous variable (1.50 (1.04 to 2.17); p=0.03) and as a categorical variable (SDI=1:3.93 (1.46 to 10.55); SDI=2 or 3:1.57 (0.32 to 7.65); SDI≥4:7.86 (0.89 to 69.06); p=0.03). After adjustment for prior medications (table 5, analysis III), this risk was less significant (p=0.07), demonstrating confounding between disease severity and medications. An overall association with the four specified medication variables was found (global test p value=0.04). Corticosteroids were associated with an increased risk of seizures but with a wide CI. Prior use of antimalarial drugs in the absence of immunosuppressive agents was the most notable treatment effect on seizures, conferring a reduced risk (0.07 (0.01 to 0.66)). A formal test for an interaction between antimalarial drugs and immunosuppressive agents was significant (p=0.03).
Without adjusting for SDI scores (see table 5, analysis IV), the global medication test generated a p value of 0.01 with again a significant interaction (p=0.02). This result derives primarily from a comparison of 7 events in 715 observation intervals for patients not taking immunosuppresants or antimalarials versus 2 events in 2210 observation intervals for patients taking only antimalarials. Because of missing values, adjustment for AMS (w/o NP variables) alters this finding because 5 events and 225 intervals in the first group and 1 event and 655 intervals in the second group are excluded; hence, the interaction, and the global medication effect, is not detected. If the missing AMS values are replaced by the current SLEDAI-2K values, the significant findings are again seen. Thus, the paucity of data makes a definitive statement on the possible confounding of medication and disease activity effects difficult. Similar results were obtained for seizures attributed to SLE (model B).
Comparable results were also obtained in a multivariate analysis of medication effects based on the inclusion of propensity scores for the use of steroids and antimalarials alone. Factors included in the propensity scores included all those in table 5, analysis III, as well as SLEDAI-2K. More severe disease was linked to steroid use while less severe disease was linked to antimalarials alone. In this analysis, steroid use had an estimated HR of 9.9 (with a CI of 1.26 to 78.57) while antimalarials alone had an estimated HR of 0.08 (0.01 to 0.70).
Outcome of seizures
Using physician assessment (figure 1), seizures attributed to SLE were more likely to resolve (59/78 (76%)) compared to seizures attributed to non-SLE causes (7/13 (54%)). While more patients with seizures attributed to SLE stopped taking antiseizure drugs at the first follow-up assessment compared to patients with seizures attributed to non-SLE causes (19/66=29% vs 2/12=17%) and similarly by the second follow-up assessment (16/50=32% vs 1/9=11%), the small number of seizures (especially non-SLE seizures) during the follow-up period precluded formal statistical testing. Using patient self-report, the mean (±SD) mental component summary score in patients with seizures and no other NP events was comparable to that in patients without NP events (45±13.1 vs 48.5±10.9; p=0.21). A similar outcome was seen in the mean (±SD) physical component summary scores (40.0±10.8 vs 42.8±11.2; p=0.34).
Seizures are a manifestation of NPSLE but may follow a different course to idiopathic seizures. This has implications for diagnosis and treatment. Previous studies have been limited by cross-sectional design, small sample size and short duration of follow-up. We studied a large, prospective, international, multi-ethnic disease inception cohort of adult patients with SLE. The risk of seizures was higher in patients of African race/ethnicity and lower educational status and in patients with organ damage outside of the nervous system. Most seizures were attributable to lupus and occurred in close proximity to the diagnosis of SLE. Most seizures resolved, did not require long-term antiseizure medication and were not associated with a negative impact on mental or physical HRQoL. The association with lupus-related treatments was complex, but antimalarial drugs were associated with a protective effect.
Prevalence estimates of seizures in SLE have varied between 6%3 and 51%.5 The SLICC cohort is well positioned to determine the frequency of seizures due to the size of the cohort, wide global geographic distribution, enrolment of patients close to diagnosis, standardised data collection and attribution rules. The cumulative frequency of 4.6% over a mean of 3.5 years in our study is comparable to other large studies. For example, 600 patients in the LUMINA cohort21 had a frequency of 6.7%, usually early in the disease course. Gonzalez-Duarte et al22 studied 1200 patients with SLE of variable disease duration and reported long-term outcomes for a subset of the cohort over a mean of 5 years. Seizures occurred in 12.5% of patients, 54% of them within 1 year of SLE diagnosis and recurred in 53%.
Although single-factor analyses identified several variables at enrolment with subsequent seizures, African race/ethnicity and lack of postsecondary education remained significant in multivariate analysis. The same association with race/ethnicity was reported in the LUMINA cohort.21 Lack of education is a surrogate for socioeconomic status23 and a predictor of functional status24 and mortality25 in SLE. In the present study, the association with education may be due to impaired access or adherence to antiseizure or lupus-related treatments. The association with cumulative organ damage has been reported in the LUMINA cohort21 and also in the current study albeit with some confounding with concurrent medication use. Overall, these data indicate that patients with lupus who have seizures have more severe disease both within and outside the nervous system.
Previous studies have noted that seizures occur in patients with higher global SLE disease activity.21 ,26 ,27 We did not find an association with higher SLEDAI-2K scores, even in patients whose seizures were attributed to SLE. As SLEDAI-2K scores were computed at the enrolment and follow-up visits, rather than at the time of seizure occurrence, they do not reflect global disease activity at the precise time of the NP event. However, the association with AMS and with corticosteroids supports the likelihood that patients with SLE who have seizures likely had more active SLE.
The autoantibodies selected for study reflected those most frequently associated with NPSLE. Antiphospholipid antibodies have previously been reported in association with seizures in some,26 ,27 but not all,22 studies and no such association was found in the current study. This discrepancy may be due to differences in patient selection and study design. Alternatively, autoantibody status at the enrolment visit only may be insufficient to determine the true risk of these autoantibodies. Rather, an assessment of autoantibodies over time may be required to identify a potential clinical–serological linkage.
Of considerable interest is the inverse association between seizures and antimalarials, which suggests a possible protective effect, an observation also reported in the LUMINA cohort.21 The potential for confounding by indication was addressed in the current study by propensity scores and regression modelling in the analysis. Both approaches produced similar results in keeping with the findings of Shah et al.28 Although the precise mechanism remains to be determined, our findings further support the short-term29 and long-term30,–,33 benefit associated with antimalarial use in SLE. This may also explain the ability to discontinue anticonvulsant treatment, which was more frequent in patients with seizures attributed to SLE. Overall, seizure disorders in patients with SLE have a favourable outcome as indicated by a lower recurrence rate, more frequent discontinuation of antiseizure medications and no detectable impact on patient self-report HRQoL.
There are limitations to the current study. First, a measure of global SLE disease activity was not universally available at close proximity to the time of seizure occurrence, which limited the ability to examine the association with generalised disease activity. Second, the association with persistent autoantibodies needs to be examined in subsequent studies. Third, the number of patients with seizures attributed to non-SLE causes was relatively small, which limited the power of some statistical analyses. Finally, although the duration of follow-up extended to 12 years, the mean follow-up was 3.5 years, which is insufficient to capture the lifetime experience of patients with SLE who have seizures. Despite these limitations, the size and characteristics of the SLICC NPSLE inception cohort and the standardised approach to data collection and analyses have provided information that will help to guide management of seizure disorders in patients with SLE worldwide. Our findings support the recent EULAR (European League Against Rheumatism) recommendations on seizures in patients with SLE.34
The corresponding author has the right to grant on behalf of all authors and does grant on behalf of all authors an exclusive license (or non-exclusive for government employees) on a worldwide basis to the BMJ Publishing Group and its Licensees to permit this article (if accepted) to be published in Annals of the Rheumatic Diseases and any other BMJPGL products to exploit all subsidiary rights, as set out in our license (http://group.bmj.com/products/journals/instructions-for-authors/licence-forms).
Ethics approval This study was conducted with the approval of the Capital Health Research Ethics Board, Halifax, NS, Canada.
Provenance and peer review Not commissioned; externally peer reviewed.
Funding JG Hanly was supported by Canadian Institutes of Health Research grant MOP-57752 (Capital Health Research Fund). Li Su was supported by MRC (UK) U105261167. V Farewell was supported by MRC (UK) U105261167. Dr Sang-Cheol Bae's work was supported by the Korea Healthcare Technology R&D Project, Ministry for Health and Welfare, Republic of Korea (A080588). The Montreal General Hospital Lupus Clinic was partially supported by the Singer Family Fund for Lupus Research. Dr Clarke is a National Scholar of the Fonds de la recherché en santé de Quebec. Dr Paul R. Fortin is a Distinguished Senior Investigator of The Arthritis Society with additional support from the Arthritis Centre of Excellence, University of Toronto. Dr Ramsey-Goldman's work was supported by the National Institutes of Health (grants UL-1RR-025741, K24-AR-02318 and P60-AR-48098). Dr Ruiz-Irastorza was supported by the Department of Education, Universities and Research of the Basque Government.
Competing interests None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.