Introduction
The use of anti-seizure medications (ASMs) has increased among pregnant mothers, with an almost three-fold increase seen in the worldwide during the past 12 years (1, 2). ASMs are used mainly for epilepsy during pregnancy; however, the greater number of prescriptions during pregnancy is driven largely by women with indications other than epilepsy, such as mood disorders or pain (1). Exposure to ASMs may lead to unwanted adverse maternal and neonatal outcomes (3); however, discontinuing the drug before or during pregnancy may lead to worsening or recurrence of the disease and thereby increase maternal mortality (4). Therefore, the decision to use an ASM during pregnancy is complex, and their reproductive safety should be understood.
ASMs theoretically pass through the human placenta and disrupt fetal neurodevelopment by changing the dopaminergic system (5). As dopamine receptors play a role in neurodevelopment (6), there are increasing concerns about adverse neurodevelopmental effects in offspring after prenatal exposure to ASMs. Animal studies have demonstrated that in utero exposure to these agents results in an increased risk of neurodevelopmental disorders, such as autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD), in offspring (7). Accordingly, several epidemiological studies (8–14) have investigated the effect of ASMs on ASD or ADHD risk, but the results were inconsistent. In the earliest cohort study, Christensen et al. (8) observed an increased risk of ASD among offspring exposed to valproate but not among those exposed to other ASMs; however, ASD was reportedly associated with exposure to other ASMs during pregnancy in several other studies (10, 12, 13). Prenatal exposure to ASMs increases the risk of ADHD in offspring; however, the results are inconsistent.
Due to the increased use of ASMs during pregnancy, determining the long-term effects of maternal use of ASMs on the neurodevelopment of offspring, including possible effects on the incidence of ASD or ADHD, is important. To date, no systematic review or meta-analysis has been conducted on this topic, so we conducted the present study to assess the association between maternal ASM exposure during pregnancy and the subsequent risk of neurodevelopmental disorders in offspring.
Methods
This systematic review and meta-analysis was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines.
Search strategy
A literature search was conducted of the PubMed and Embase databases on 1 March 2024, combining the following free-text words: (antiepileptic AND anti-seizure AND anticonvulsant) AND (autism OR autistic OR ASD OR attention-deficit/hyperactivity disorder OR ADHD OR attention-deficit disorder OR hyperkinetic disorder) AND (pregnancy OR mothers OR pregnant OR gestational OR prenatal OR perinatal OR gestation).
Inclusion criteria
We analyzed peer-reviewed epidemiological studies that examined the relationship between maternal exposure to ASMs during pregnancy and the subsequent risk of ASD or ADHD in offspring. A study was eligible if it used a cohort or case–control design, included a non-ASM exposure reference group, reported either the relative risk or hazard ratio or odds ratio (OR), and contained data allowing us to estimate risk. We excluded case reports, case series, animal studies, correspondence, editorials, and reviews.
Data extraction and quality assessment
Two reviewers independently extracted the data from the studies using an Excel spreadsheet, and any discrepancies were resolved by a third reviewer. The following data were collected from each study: first author, publication year, study location, study period, data source, information source for ASM exposure, exposure time, outcome measures, and statistical adjustments. Risk of bias and quality were assessed using the Newcastle-Ottawa scale (NOS), which was developed to assess the quality of epidemiological studies (15). This scale scores studies on three dimensions with nine questions relevant to research quality: the selection and comparability of the study groups and the relevance to the exposure (case–control) or outcome (cohort) of interest; all questions were scored as 0 or 1, and studies with scores ≥ 7 were considered to be of high quality.
Outcome measurements
The primary outcome was the risk of ASD or ADHD with ASM use during pregnancy compared with no treatment in the general population. To rule out the potential effect of confounding by indication (bipolar disorder or epilepsy), the risk was assessed by comparing mothers with a clinical indication for taking an ASM to mothers with a clinical indication for not taking an ASM. To identify potential sources of heterogeneity, subgroup analyses were stratified according to the individual ASM.
Statistical analysis
STATA 12.0 software (StataCorp LP, College Station, TX, United States) was utilized for all statistical analyses. Between-study heterogeneity was assessed using the χ2 test and I2 statistic; we considered I2 > 50% or p < 0.05 for the Q-statistic to indicate high heterogeneity (16). Pooled ORs with 95% confidence intervals (CIs) for dichotomous data were calculated using random or fixed-effects models according to the heterogeneity of the studies. We used a random effects model because its assumptions account for high heterogeneity among studies; otherwise, a fixed effects model was used (17). Publication bias was evaluated by Begg’s funnel plot (18). The funnel plot was not used to assess publication bias in analyses of <10 studies (19). We considered p < 0.05 to indicate a significant difference.
Discussion
There has been increasing interest in the association between the use of ASMs during pregnancy and neurodevelopmental disorders in offspring. The findings of our study suggest that maternal exposure to ASMs during pregnancy was associated with a modestly increased risk of ASD or ADHD in offspring, even after controlling for potential confounders by indication. However, subgroup analyses based on individual ASMs found that only valproate was associated with an increased risk of ASD or ADHD, suggesting that the overall association may be driven by valproate use alone.
In our primary analyses, these associations were explored in the general population, and an almost two-fold increased risk of ASD was observed. ASMs are widely used to treat epilepsy and psychiatric disorders and have been associated with neurodevelopmental disorders in offspring (20, 21). Thus, enrolling healthy pregnant women unexposed to ASMs as a reference group may have overestimated this association. The ideal control for confounding by indication is a reference group of women with epilepsy or a psychiatric disorder who did not take an ASM during pregnancy. In our secondary analysis, these associations were explored among mothers with a clinical indication, and the risk of ASD was reduced 1.38-fold. When we further limited our analysis to pregnant mothers with epilepsy, the combined OR of ASD was 1.37. Although we reported inconsistent results for ADHD risk, these findings may have been limited by the small number of studies included in our analysis.
Animal studies have demonstrated different neurodevelopmental effects of prenatal individual ASM exposure (22). Several systematic reviews have also summarized the evidence regarding the neurobehavioral teratogenicity of intrauterine ASM exposure (23–25). Most clinical data are derived from case reports and small case series studies (26), and findings for different types of ASM have been inconsistent. Drawing conclusions from epidemiological studies has been hampered by differences in methodology and limited sample sizes. In primary analyses, three types of ASM were associated with an increased risk of ASD; however, these findings could not be replicated in a secondary analysis, in which only exposure to valproate during pregnancy was associated with an increased risk of ASD or ADHD in offspring. The precise biological mechanisms underlying the relationship are not entirely clear. Valproate is a short-chain low-molecular-weight fatty acid that quickly crosses the placental and blood–brain barriers (27). Valproate disrupts the metabolism of glutamate and neuronal plasticity in the fetal brain, thus triggering a neurodevelopmental disorder (28). In addition, valproate is a histone deacetylation histone deacetylase inhibitor, which affects gene replication and transcription (29). Results from a recent animal model study suggest that maternal exposure to valproate promotes replication and transcription of autism-associated genes, leading to autism-like behaviors (30). Finally, exposure to maternal valproate has been linked to congenital malformations (31), which are associated with an increased risk of neurodevelopmental disorders (32).
This meta-analysis is the first to provide an overall estimate of the effect of maternal ASM use during pregnancy on ASD and ADHD risk in offspring. The strength of our meta-analysis lies in the exclusive use of cohort studies, which are less prone to bias when assessing drug exposure during pregnancy. Second, the overall sample of all studies for the meta-analysis comprised >10 million pregnancy episodes across nine different countries, which makes the findings more generalizable and robust. Third, all of the studies were considered of high quality, which may minimize the bias of the review. Fourth, the associations were investigated in subgroup analyses based on ASM indication and type.
However, our review also had several limitations. The most important limitation of our meta-analysis was the significant heterogeneity among the studies, although in the subgroup analysis based on individual ASM, heterogeneity was reduced to 0%. Second, due to the inherent nature of observational studies, those included in this meta-analysis were prone to various types of confounding. We conducted analyses by pooling the adjusted data. Third, the etiology of epilepsy contributing to different ASM choice may affect the contributors toward ASD or ADHD, which may be an important confounding factor. Fourth, there were limited data in the studies on the dose or use of combination ASM agents; therefore, we could confirm that any exposure parameter was associated with neurodevelopmental outcome. Fifth, due to the small number of included studies, we could not evaluated the interactions between different ASM contribute to the incidence of autism or ADHD; however, the results of subgroup analyses based on individual ASM were consistent with the findings of two included studies (9, 12), which evaluated the risk of ADHD or ASD in the offspring of women who used antiepileptic drugs in monotherapy during pregnancy.
In conclusion, our results suggest that the link between prenatal ASM exposure and ASD or ADHD risk in children was attributable to the indication, rather than there being a causal relation. However, maternal use of valproate during pregnancy was associated with a significantly increased risk of ASD or ADHD, even compared to mothers with bipolar disorder or epilepsy who were not exposed to ASMs in utero.
Author contributions
S-CX: Writing – review & editing, Methodology, Conceptualization. YZ: Writing – review & editing, Resources, Project administration, Investigation. H-YJ: Writing – review & editing, Visualization, Investigation, Formal analysis. JT: Writing – review & editing, Writing – original draft, Validation, Supervision, Funding acquisition, Formal analysis.