Mechanisms of neurodevelopmental toxicity of topiramate

Abstract

Prescriptions for antiseizure medications (ASMs) have been rapidly growing over the last several decades due, in part, to an expanding list of clinical indications for which they are now prescribed. This trend has raised concern for potential adverse neurodevelopmental outcomes in ASM-exposed pregnancies. Recent large scale population studies have suggested that the use of topiramate (TOPAMAX, Janssen-Cilag), when prescribed for seizure control, migraines, and/or weight management, is associated with an increased risk for autism spectrum disorder (ASD), intellectual disability, and attention-deficit/hyperactivity disorder (ADHD) in exposed offspring. Here, we critically review epidemiologic evidence demonstrating the neurobehavioral teratogenicity of topiramate and speculate on the neuromolecular mechanisms by which prenatal exposure may perturb neurocognitive development. Specifically, we explore the potential role of topiramate’s pharmacological interactions with ligand- and voltage-gated ion channels, especially GABAergic signaling, its effects on DNA methylation and histone acetylation, whether topiramate induces oxidative stress, and its association with fetal growth restriction as possible mechanisms contributing to neurodevelopmental toxicity. Resolving this biology will be necessary to reduce the risk of adverse pregnancy outcomes caused by topiramate or other ASMs.

1. Introduction

Topiramate is an antiseizure medication (ASM) approved in 1996 for seizure control in individuals with epilepsy syndromes (Fariba and Saadabadi 2024). Like many ASMs, it can be prescribed as monotherapy or used as an adjunct with other ASMs. Aside from seizure control, it is also prescribed as a migraine prophylactic and a weight loss aid (Pearl et al. 2023; Wajid et al. 2023; Lupianez-Merly et al. 2024). In addition to these indications, for which it is officially approved by the Federal Food and Drug Administration, topiramate is also prescribed off-label for the treatment of neuropathic pain, binge eating disorder, alcohol dependence, non-migraine headache, Tourette syndrome, essential tremor, and several mood and personality disorders (bipolarism, depression, post-traumatic stress, obsessive-compulsive disorder, borderline personality disorder) (Pearl et al. 2023; Fariba and Saadabadi 2024). Topiramate is taken orally, with typical doses ranging from 25 to 400 mg per day based on tolerability and indication (higher doses are used for seizure prevention). Doses may often need to be adjusted in the endeavor to achieve seizure freedom, and topiramate clearance rates may vary when used adjunctively with other ASMs that have enzyme inducing or inhibiting effects. As with other ASMs, topiramate can result in adverse cognitive effects resulting in varying degrees of disability, such as memory impairment, reductions in cognitive speed, and inattention (Donegan et al. 2015; Brandt et al. 2015; Javed et al. 2015). Topiramate is also a carbonic anhydrase inhibitor (Dodgson et al. 2000) that can result in a non-anion gap metabolic acidosis. This may be asymptomatic, or lead to nephrolithiasis and osteoporosis (Vega et al. 2007). Other adverse effects include depression, suicidal thoughts and behavior, hyperammonemia, effects on growth in pediatric patients in terms of both reduced height and weight, and a wide range of ocular side effects, including myopic shift, choroidal effusion syndrome, and angle closure glaucoma (Mechrgui and Kanani 2022).

2. Congenital malformations associated with topiramate

Fetal exposure to topiramate has been associated with an increased incidence of structural birth defects. A 2017 systematic review that examined 96 studies found that topiramate exposure was associated with increased odds of congenital malformations and prenatal growth restriction (Veroniki et al. 2017). Additionally, studies using data from various pregnancy registries, insurance or Medicaid claims, or electronic medical records have reported increased risk of cleft lip and cleft palate as well as lower birth weights in infants exposed to topiramate in utero (Hunt et al. 2008; Margulis et al. 2012; Hernández-Díaz et al. 2012; Hernández-Díaz et al. 2014; Mines et al. 2014; Hernandez-Diaz et al. 2018; Blotière et al. 2019; Cohen et al. 2023). Among these studies, there was considerable heterogeneity in sample size, prescribing indications, analytic methodologies, and types of exposure (dosing, mono vs polytherapy). When these studies are taken together, the increased risk of oral clefts in topiramate-exposed pregnancies is generally reported between 1.5 and 6.8-fold. Studies that were able to perform a dose analysis generally reported a higher risk of clefts with higher doses (Hernandez-Diaz et al. 2018; Cohen et al. 2023). For example, using Medicaid Analytic eXtract data, Hernandez-Diaz and colleagues reported a risk ratio of 1.64 (CI: 0.53–5.07) for doses of 100 mg/d or less, while the risk ratio for doses over 100 mg/d was 5.16 (CI: 1.95–13.73) (Hernandez-Diaz et al. 2018). However, it is important to note that orofacial clefts have still been reported even for doses as low as 25 mg/d (Margulis et al. 2012). A Cochrane Library systematic review examining risk for monotherapy ASM exposures and major congenital malformations found that topiramate was significantly associated with birth defects prevalence in pooled data from cohort studies and, notably, that the specific risk for orofacial clefts was significantly higher for topiramate when compared to other ASMs (Bromley et al. 2023). Indeed, prescribing information for topiramate disclose that preclinical testing across multiple rodent species have identified craniofacial defects, reduced fetal weight, and other unspecified structural malformations at clinically relevant doses.

3. Pharmacological action of topiramate

The pharmacological effects of topiramate are elicited through several mechanisms of action, which serve as a starting point for proposing modes of neurodevelopmental toxicity. Topiramate is believed to control seizures through effects on multiple receptors or membrane channels that regulate the transmission activity of neurons (Meldrum and Rogawski 2007), including GABA_A receptors, kainate receptors, AMPA and NMDA-type glutamate receptors, as well as voltage-gated sodium and calcium channels. GABA_A receptors play a principal role in fast inhibitory neurotransmission, and topiramate has been shown to enhance the activity of these receptors (White et al. 1997; Petroff et al. 1999; Petroff et al. 2001; Kuzniecky et al. 2002). AMPA and NMDA receptors modulate excitatory synaptic transmission and play central roles in synaptic plasticity that underlies learning and memory (Chater and Goda 2014; Jewett and Thapa 2024). Kainate receptors regulate release of neurotransmitters from presynaptic neurons, and have been implicated in autism as well as other neurological conditions (Strutz-Seebohm et al. 2006; Aller et al. 2015; Epsztein et al. 2017; Koromina et al. 2019; Valbuena and Lerma 2021). There is strong evidence demonstrating that topiramate inhibits AMPA and kainate receptors (Gibbs et al. 2000; Skradski and White 2000; Gryder and Rogawski 2003; Qian and Noebels 2003; Poulsen et al. 2004; Braga et al. 2009), and there are also data to suggest that it similarly inhibits NMDA activity (Rawls et al. 2009; Yilmaz et al. 2011; Motaghinejad et al. 2017). The inhibitory actions of topiramate on these receptors are not necessarily direct. Fukushima and co-workers tested excitatory inhibition of topiramate in HEK cell lines expressing specific AMPA, NMDA, and kainate receptors, and did not observe any direct antagonism (Fukushima et al. 2020). The authors postulated this was due to the fact that topiramate antagonizes AMPA and kainate receptors through phosphorylation-dependent allosteric mechanisms (Ängehagen et al. 2004; Fukushima et al. 2020). Finally, topiramate has been demonstrated to block voltage-gated sodium channels (Zona et al. 1997) and antagonize voltage gated calcium channels, which are essential to neuronal excitability (Wang et al. 2017). This review will explore how these various actions of topiramate may perturb the normal course of neurodevelopment during fetal exposure and result in syndromes of pervasive neurodevelopment featuring varying degrees of intellectual disability and autism spectrum disorder (ASD).

4. Neurodevelopmental consequences of prenatal topiramate exposure: clinical evidence

Efforts to identify neurodevelopmental risks of in utero ASM exposures have been historically challenged by the inherent inability to conduct double-blind, placebo-controlled assessments for neuroteratogenic safety. In the absence of such, we are left with retrospective assessments or uncontrolled prospective observational studies, which can suffer from recall or selection bias as well as limited sample sizes (Meador and Loring 2016). For example, a retrospective study published in 2016 using data from the UK Epilepsy and Pregnancy Register (UKEPR) reported on 27 pregnancies exposed to topiramate, and failed to find reductions in child cognitive abilities (IQ, verbal, nonverbal, and expressive language) for children between the ages of five and nine, despite finding all of these effects in children of valproate-exposed pregnancies (Bromley et al. 2016). However, the valproate arm of the study had a considerably larger sample size (47 children). A larger observational cohort from France looking at pregnancies exposed to nine different ASMs, including 477 topiramate-exposed pregnancies also did not find any association between prenatal topiramate exposure and autism (Blotière et al. 2020); however, it is important to note that the study design, in this case, used lamotrigine-exposed pregnancies as the reference group. Additionally, the authors report a median follow-up age of only 3.7 years, which they admit is an additional limitation of their study since later diagnoses and less severe ASD cases were more likely to be omitted. In 2018, a population-based cohort study used Danish pregnancy register data to examine in utero exposure to various ASMs and prevalence of learning disabilities in children during their first year of compulsory education (Bech et al. 2018). The investigators used 434 unexposed pregnancies of mothers who previously used ASMs greater than 90 days before conception as controls. According to their adjusted regression model, the odds ratio for learning disabilities in topiramate-exposed pregnancies was 5.82 (95% CI of 1.21–27.97), which was statistically significant. However, this reported outcome appears to be based on fewer than four topiramate-exposed pregnancies, since samples sizes of less than four could not be reported due to Danish privacy restrictions.

Increasingly improved datasets and study designs in recent years have significantly enhanced our understanding of neurodevelopmental outcomes in topiramate-exposed pregnancies. The most convincing epidemiological evidence supporting increased risk for ASD or intellectual disability from topiramate exposure during pregnancy comes from a 2022 report published in JAMA Neurology. This population-based study used public registry data from five Nordic nations (SCAN-AED), which included 24,825 children prenatally exposed to ASMs, of which, 471 were exposed to topiramate monotherapy, and 148 were exposed to topiramate/lamotrigine polytherapy (Bjørk et al. 2022). Topiramate exposure (compared with no ASM) was associated with an ~2.7-fold increased risk of ASD and an ~3.4–3.9-fold risk of intellectual disability. This effect was similar when considering topiramate use for women with epilepsy or for the total population. This increased risk appeared to be dose-dependent, increasing from 1.71 for women taking less than 100 mg/d to 2.93 in women taking 100 mg/d or more. The same SCAN-AED data source was used in another population-based study which examined early onset psychiatric disorders which found an association between topiramate exposure and attention-deficit/hyperactivity disorder (ADHD) (Dreier et al. 2023). The study included 38,661 children of mothers with epilepsy, of which 290 were exposed to topiramate monotherapy during pregnancy. Of topiramate-exposed children, 16 were diagnosed with ADHD, amounting to an adjusted hazard ratio of 2.38 (95% CI of 1.40–4.06). The authors also noted non-significant elevated risks for ASD and intellectual disability in the topiramate exposed cohort which were slightly lower than reported by their 2022 study. The reason for the lower risk signature in the 2023 study was attributed to different inclusion criteria regarding the drug’s exposure window during pregnancy (Dreier et al. 2023).

More recently, another population-based study published in The New England Journal of Medicine (NEJM) examined ASD risk in pregnancies exposed to topiramate (Hernández-Díaz Sonia et al. 2024). This study utilized Medicaid Analytic eXtract data and MarketScan health insurance data to define a study population in the United States which included pregnancies of women with dispensed prescriptions for topiramate, valproate, or lamotrigine compared to an unexposed reference group of pregnancies. Like Bjørk et al 2022, the authors’ primary analysis focused on 28,952 pregnancies of women with epilepsy of which 1,030 were exposed to topiramate; however, they also examined a total cohort of over 4 million pregnancies regardless of epilepsy status of which 2,469 were exposed to topiramate. While they found an increased cumulative 8 year incidence of ASD and an elevated hazard ratio of 2.17 (95% CI of 1.54–1.94) in topiramate-exposed offspring compared to their total cohort, these results were similar for the other ASMs included in the study (Hernández-Díaz Sonia et al. 2024). When they restricted their cohort to only include epileptic pregnancies, cumulative incidence and adjusted hazard ratios for ASD in topiramate-exposed pregnancies were comparable to those of the reference group, which contradicted the SCAN-AED findings.

In 2023, an analysis of adaptive behaviors in topiramate-exposed children was conducted through the UKEPR. This study measured Vineland Adaptive Behavior scores in a cohort of 21 children from topiramate-exposed pregnancies (doses between 100–800mg/d) using telephone interviews (Knight et al. 2023). Compared to a historically obtained normative sample of 2,560 children, they found significantly reduced global adaptive behavior composite (ABC) scores in topiramate-exposed children, along with significantly reduced scores in daily living skills and socialization domains. They also examined the effect of topiramate dosage and found that ABC, socialization, and communication scores all correlated negatively with increased doses. Four of these 21 cases had formal diagnoses of ASD, which the authors noted was a higher rate than would be expected in the general UK population (Knight et al. 2023). Notably, three of these four ASD cases were associated with exposures greater than 200 mg/d. Knight and colleagues do note, however, that their study was plagued by recruitment challenges due to the COVID-19 pandemic which caused them to modify their study design and may have introduced selection bias. Other studies examining risks of autistic traits or intellectual disabilities in children of ASM-exposed pregnancies either did not have any topiramate exposures in their dataset, or had relatively small sample sizes of topiramate exposure, and thus, could not report any significant associations (Veiby et al. 2013; Bromley and Marson 2014; Bjørk et al. 2018; Husebye et al. 2018; Husebye et al. 2020).

Thus, the clinical evidence for neurodevelopmental toxicity of topiramate remains controversial. While several observational studies in Denmark and the United Kingdom have found elevated incidence of ASD, intellectual disability, ADHD, poor adaptive behavior, and learning impairment in children of topiramate-exposed pregnancies (Bech et al. 2018; Bjørk et al. 2022; Knight et al. 2023; Dreier et al. 2023), similar studies in France and the United States have suggested that topiramate does not carry any more risk for ASD outcomes than lamotrigine for women with epilepsy (Blotière et al. 2020; Hernández-Díaz Sonia et al. 2024). In all cases, the rates of ASD or other adverse neurodevelopmental outcomes are relatively low, and each of these studies differ in various aspects of study design, ranging from data sources to data classification and analysis. When comparing the Nordic and American studies, for example, Meador has raised the point that they used differing criteria to categorize topiramate-exposed pregnancies, with the American study restricting their examination to exposures during the second half of pregnancy arguing that this period of development coincides with significant synaptogenesis (Hernández-Díaz Sonia et al. 2024; Meador Kimford J. 2024). Regardless, the conflicting clinical evidence for neurodevelopmental risk highlights the necessity for further research and more direct approaches to clarify the potential effects of topiramate on prenatal neurological development.

5. Proposed mechanisms of neurodevelopmental toxicity of topiramate

How do the known pharmacological actions of topiramate overlap with causes and mechanisms of ASD or other neurodevelopmental disorders? ASD is a complex trait for which many genetic risk factors have been identified (Won et al. 2013), and for which several environmental precipitants have also been implicated (Rossignol et al. 2014; Yenkoyan et al. 2024). Multiple inherited or de novo genomic variants may act in concert with environmental factors and pharmaceutical exposures (Ornoy et al. 2015). Given the sheer number of genetic factors that have been linked to autism, it can be difficult to conceptualize how these various factors may function and interact on a larger mechanistic level. Krishnan and colleagues adopted a machine learning strategy to examine the genetic basis of ASD from a functional perspective (Krishnan et al. 2016). This methodology used known ASD-associated genes (which were weighted based on the level of evidence supporting an association) and integrated it with a human brain-specific functional interaction network to predict other genes that may be associated with ASD pathology and analyze how these risk genes function in the cellular and molecular context of the brain. Some of the key functional modules they identified, which had been previously implicated in ASD pathology, included synaptic transmission and plasticity, neuronal function (such as ion transport), effects on signaling pathways (including pathways involved in fetal development), epigenetic regulation (such as histone modification and chromatin remodeling), and regulation of cell division. Examples of well characterized genetic risk factors for ASD include genes required for neuronal development and function, such as RELN, DYRK1A, SHANK3, SYNGAP1, CDKL5, CNTNAP2, SYN1, NRXN1, RBFOX1, FMR1, and SCN1A/2A (Wang et al. 2023). Additionally, several chromatin remodelers and methyltransferases (important for epigenetic regulation of gene expression) have been implicated, including MECP2, ANDP, ARID1B, ASH1L, CHD2, CHD8, POGZ, and EHMT1 (Wang et al. 2023). These gene products all play critical roles in embryonic and postnatal neurogenesis, and regulate the development of axons, dendrites, and synapses (Zhou et al. 2024). These genetic risk factors ultimately translate to cell and tissue level alterations in brain development that impact neuronal connectivity. For example, autistic brains have been found to demonstrate changes in cell density in cerebellar Purkinje neurons (Bruchhage et al. 2018) and a host of other interneuron types (Lawrence et al. 2010). While, ASD certainly has a strong genetic etiological component, the risk for ASD can be influenced by a multitude of environmental factors as well, such as prenatal exposure to certain drugs (Ornoy et al. 2015). Other proposed causes of autism include oxidative stress, autoimmune diseases, pathologies of gastrointestinal permeability, and several metabolic disorders (Lyall et al. 2017). In addition, maternal age, nutrition, weight, and diabetes status are all risk factors. Finally, there is considerable overlap between risk factors for ASD and those for epilepsy, suggesting common mechanisms (Keller et al. 2017). The following sections will explore evidence supporting specific mechanisms through which topiramate may interact with these known ASD-associated pathways during gestation, resulting in an increased risk for adverse neurodevelopmental outcomes.

5.1. Effects on ligand- and voltage-gated ion channels

Topiramate interacts with receptors with established roles in synapse development and function (Südhof 2018). Not only do mutations in these receptors impart autism risk, but these receptors have functional interactions with other proteins coded by ASD risk factor genes. For example, the GABAergic system has been consistently implicated in ASD pathology through both human genetic and animal model studies (Zhao et al. 2021). Genetic risk variants in GABA receptor subunits have been identified in genomic association studies (Ma et al. 2005; Noroozi et al. 2018), and the chromosomal region that is duplicated in 15q11-q13 duplication syndrome (a syndrome for which symptoms include autism and epilepsy phenotypes) contains genes for multiple GABA_A receptor subunits (Hogart et al. 2007). Transgenic mice deficient in the ß₃ subunit of the GABA_A receptor develop seizures and autism-like behavioral phenotypes (DeLorey et al. 1998). Notably, they also develop cleft palate (Culiat et al. 1993; Culiat et al. 1995), a hallmark of topiramate teratogenicity. Classical experiments have shown that GABA exposure inhibits mouse palatal shelf reorientation in ex vivo embryo culture (Zimmerman and Wee 1984); while cleft palate has also been observed in mouse fetuses lacking Gad67, which codes for an enzyme that participates in GABA synthesis (Condie et al. 1997). Thus, too much or too little GABA signaling is problematic, and topiramate’s agonism of GABA_A may tie together effects on cognitive development and orofacial malformations. In addition to these direct genetic manipulations of GABA signaling, multiple genetic mouse models of autism demonstrate altered expression of GABA_A subunits, variability in the number of GABAergic neurons, and subsequent impairment of GABAergic neurotransmission (Zhao et al. 2021). Moreover, valproate-induced autism models in mice have also been experimentally demonstrated to have altered expression of GABA receptor subunits as well as other genes or proteins that participate in GABAergic activity in specific regions of the brain (Zhao et al. 2021).

Since topiramate is an agonist of GABA receptors, it is likely to interfere with normal GABAergic transmission required for prenatal brain development. GABA receptors are expressed in the developing prenatal brain, and importantly, GABA receptor activity seems to be primarily excitatory instead of inhibitory in the context of fetal development, due to lower extracellular chlorine concentrations observed in these early critical windows. Therefore, it is only after birth that GABA switches to becoming an inhibitory signal (Miles 1999). GABA receptor-modulated calcium signaling has been implicated in the migration of interneurons into the developing fetal neocortex. In this context, GABA functions as a chemoattractant, guiding migrating interneuron precursor cells to their appropriate location during cortical development (Behar et al. 2000; Li et al. 2004; Li et al. 2018; Warm et al. 2022). In support of this model, conditional transgenic impairment of GABA signaling in embryonic endothelial cells was shown to alter both migration distance and laminar positioning of cortical interneurons (Li et al. 2018); while pharmacological agonism of GABA_A receptors alters migration speeds of neurons and causes accumulation of migrating neurons in upper cortical layers (Heck et al. 2007). This embryonic migration of neurons is essentially linked to these cells functional specification and subsequent maturation extending into the postnatal period (Warm et al. 2022; Wamsley and Fishell 2017; Lim, Mi, et al. 2018; Lim, Pakan, et al. 2018). Therefore, in a vulnerable genetic background, prenatal topiramate exposure may hyperactivate GABAergic signaling to disrupt the migration, distribution, and specification of neurons, ultimately influencing their interconnectivity (Herlenius and Lagercrantz 2001).

Topiramate may similarly influence the development of neurons through its inhibition of NMDA, AMPA, and kainate receptors. Each of these receptors have salient functions in embryonic and fetal brain development, and their expression and functionality has been detected even prior to embryo implantation (Špirková et al. 2022). But their roles in neural development undoubtedly occur at later developmental stages; for example, NMDA receptor subunits are reportedly found throughout second trimester human brains (Bagasrawala et al. 2017). NMDA, AMPA, and kainate receptors are ionotropic glutamate receptors, and typically respond in an excitatory manner to glutamic acid and its various neurotransmitter analogs. Glutamatergic signaling through AMPA and kainate receptors have been shown to function in the embryonic neocortex of rats to inhibit DNA synthesis and may thereby regulate proliferation of developing neurons (LaMantia 1995; LoTurco et al. 1995); while NMDA receptors have been shown to mediate establishment of functional sensorimotor synapses in rat embryonic spinal cord (Ziskind-Conhaim 1990). Moreover, prenatal antagonism of NMDA receptors in rats seems to induce apoptosis in developing neurons due to excessive calcium influx (Lipton and Nakanishi 1999; Herlenius and Lagercrantz 2001). There is also evidence suggesting that topiramate’s pharmacological effects are tied to genetic variants in these receptors. For example, the rs2832407 polymorphism in GRIK1 (encoding a kainate receptor subunit) was found to be predictive for topiramate’s effectiveness in treating alcoholism (Kranzler et al. 2014). A follow up study showed that neurons with one or two copies of this genetic variant had differential impairment of excitatory responses after exposure to topiramate (Lieberman et al. 2020). Thus, topiramate may very well impact neurodevelopment through effects on both GABA and glutamate signaling (Nisar et al. 2022).

Finally, topiramate is an antagonist of voltage-gated sodium channels (Zona et al. 1997), which propagate action potentials during neuronal excitation. Various subunits of voltage gated sodium channels are expressed throughout the developing embryonic brain (Beckh et al. 1989; Felts et al. 1997; Whitaker et al. 2000; Cheah et al. 2013; Liang et al. 2021), and genetic variants in genes coding these subunits result in cortical defects, intellectual disability, ASD, developmental delay, and epilepsy (which in many cases is not treatment responsive) (Liang et al. 2021; Sanders et al. 2018; Zaman et al. 2020; Hanly et al. 2021; Barbieri et al. 2023). Topiramate appears to inhibit voltage-gated sodium channels by modifying their phosphorylation state, thus modifying their activity (Curia et al. 2007). As such, inhibition of these sodium channels is another plausible mechanism through which topiramate may contribute to ASD and intellectual disability. Importantly, genetic variants in sodium channel genes have been associated as contributing determinants to the efficacy of sodium channel-inhibiting ASMs in terms of their ability to control seizures (Lin et al. 2021). Therefore, any individual mechanism for adverse neurodevelopmental outcomes associated with exposure to these drugs is likely similarly a product of gene-drug interactions.

5.2. Transcriptomic and Epigenetic Effects

Many autism-associated genes code for enzymes or transcription factors that regulate or respond to multiple epigenetic factors (MECP2, CHD8, etc.), suggesting that they play salient roles in the etiology of ASD; while several studies have revealed altered epigenomes in ASD cases (Sun et al. 2016; Waye and Cheng 2018). Beyond topiramate’s acute influences on voltage- or ligand-gated channels, it may impart more long-term changes in gene expression through transcriptional and epigenetic effects. A cell culture study comparing the effects of multiple ASMs on histone hyperacetylation found that topiramate, like valproate, is a histone deacetylase (HDAC) inhibitor (Eyal et al. 2004); while evidence from the valproate-induced mouse model of autism suggests that valproate’s mechanism of action for ASD pathogenesis may be tied to HDAC inhibition (Nicolini and Fahnestock 2018). Kataoka and co-workers found that treating mice in utero with valpromide, a valproate analog that does not inhibit HDACs, did not result in similar behavioral phenotypes observed in valproate-exposed mice (Kataoka et al. 2013). Thus, topiramate has the potential to function through a similar mechanism of action. Another study examining cortical spreading depolarization in migraine cases with and without topiramate treatment found a reduction in the number of regions of hypermethylated DNA, suggesting topiramate may promote demethylation or inhibit DNA methylases (Vila-Pueyo et al. 2023). There are also several examples of topiramate having effects on gene expression in certain experimental models, such as one testing the efficacy of topical topiramate to promote wound healing (Jara et al. 2018) and another which found that topiramate increased expression of TGFß1 and SOX9, among other genes, in embryonic palatal mesenchyme, proposing this as a potential teratogenic mechanism for the induction of cleft palate (Rafi et al. 2021). Notably, in this case, the gene expression changes were proposed to be elicited through topiramate’s effect on GABA signaling, and not on direct epigenetic actions. Thus, topiramate does not have to directly interact with epigenome regulators, like HDACs, to modify the timing and regulation of gene expression.

5.3. Oxidative Stress

Increased oxidative stress has also been proposed as a causative factor for ASD (Bjørklund et al. 2020). Evidence regarding the effects of topiramate on oxidative stress is conflicting. In vitro studies using cultured astrocytes found that topiramate increased markers of oxidative stress, including malondialdehyde (MDA), nitric oxide, and lactate dehydrogenase, and also resulted in a dose-dependent increase in reactive oxygen species (Pavone and Cardile 2003). Similar effects have been observed in a mouse study examining the hepatic and renal toxicity of topiramate (El Makawy et al. 2022), where topiramate was found to inhibit two antioxidant enzymes (super oxide dismutase and catalase). This study reported similar dose-dependent increases in MDA and nitric oxide levels. Multiple other studies in mice or cell culture have reported observing markers of oxidative stress in models of topiramate exposure, usually in the form of increased MDA, increased levels of nitric oxide, or decreased levels of glutathione (Huang et al. 2007; Agarwal et al. 2011; El Makawy et al. 2019; Gündüz et al. 2021; Mabrouk et al. 2022). In most cases, these effects were observed in a dose-dependent manner or in experimental groups that received high doses of topiramate.

At the same time, many studies provide evidence for an antioxidant effect. For example, a study looking at the potential of topiramate to prevent skin flap necrosis after plastic surgery measured a decrease in MDA and an increase in glutathione levels (Ahmadzadeh et al. 2022); however, this study noted that topiramate did raise levels of super oxide dismutase at high doses. One study suggests that these conflicting results may be because topiramate has direct antioxidant activities (N. Cárdenas-Rodríguez et al. 2013), implying that acute or low dose exposures may have neuroprotective or antioxidant effects, while higher doses or chronic exposures may inhibit antioxidant enzymes leading to oxidative stress (Noemí Cárdenas-Rodríguez et al. 2013).

So, it may be more difficult to argue oxidative stress as a mode of action for topiramate induced neurotoxicity given this conflicting evidence. However, one notable study found that rats exposed to topiramate between postnatal days 16 and 28 led to vascular dysfunction in the adult mice, which the authors attributed to elevated superoxide anion concentrations (Vidigal et al. 2022). These data suggest unrealized toxic effects of topiramate on the cardiovascular system and cardiovascular development, which may be important during embryonic development when the maternal and fetal vascular system are integrated via the placenta. There is some evidence to suggest that defects in placental vasculature, which are often linked with diabetes and obesity, may be associated with ASD (Bronson et al. 2017; Chang et al. 2017). Insufficient placental development is also a risk factor for fetal growth restriction, which has an established association with gestational topiramate exposure (Sun et al. 2020).

5.4. Fetal Growth Restriction

Multiple epidemiological studies as well as animal studies have demonstrated that topiramate increases the risk for children to be born small for gestational age or with low birth weights (Hernández-Díaz et al. 2014; Veroniki et al. 2017). ASD is substantively more common in children born small for gestational age or with a low birth weight (Hultman et al. 2002; Gardener et al. 2011; Lampi et al. 2012; Talmi et al. 2020; Song et al. 2022), although it is unclear if this correlation with restricted growth is causative for ASD or merely coincident with shared risk factors (Lyall et al. 2017). Regardless, this connection could potentially explain some cases of ASD or intellectual disabilities tied to prenatal topiramate exposure, especially if the affected child is known to have been born small for gestational age or underweight. The recently published study of adaptive behavior of children from topiramate-exposed UKEPR pregnancies highlighted this potential association. Of the 21 children included in their study, six were born small for gestational age (defined as below the 10^th percentile), which was significantly higher than expected in the UK population (Knight et al. 2023). They also observed a significant negative correlation between birthweight centile and topiramate dose. Of the four children in their study with diagnosed ASD, all were born below the 25^th percentile, with one child born below the 10^th percentile. This study also determined that lower birthweights in children exposed to topiramate in utero correlated with reduced ABC scores, with specifically lower scores in the Socialization domain. Therefore, future epidemiological studies or meta-analyses exploring neurocognitive developmental risk of topiramate or other ASMs should consider birthweight outcomes as a critical covariate when possible.

6. Discussion

A growing body of clinical evidence ties fetal topiramate exposure to an increased risk of orofacial malformations (e.g., cleft lip or palate), low birth weights, and neurobehavioral syndromes that encompass autism spectrum disorder, intellectual disability, and ADHD. It is not yet understood whether these distinct exposure outcomes co-occur in individual children as a syndrome, especially since the epidemiological studies examining neurodevelopmental outcomes largely exclude children with major congenital malformations, precluding their joint interpretation. Long-term follow ups are also needed to more thoroughly explore developmental outcomes of children exposed to topiramate in utero. In the meantime, more targeted, mechanistic studies addressing neurodevelopmental toxicity of topiramate would improve decision making for prescribers and regulatory agencies. Data from multiple animal and cell culture models identify several pharmacodynamic mechanisms by which gestational topiramate exposure may pervasively impair neurocognition. Many of these mechanisms intersect with known pathways associated with ASD and other forms of neurodevelopmental disability. Unfortunately, animal studies specifically investigating neurobehavioral outcomes of in utero topiramate exposure are severely lacking, unlike with some ASMs, such as valproate (Roullet et al. 2013; Bass et al. 2020).

Topiramate prescriptions continue to increase, as the list of clinical indications for topiramate grows longer. According to the Clinicalc DrugStats database, which sources prescription data from the Medical Expenditure Panel Survey, topiramate prescriptions in the United States increased by 34% between 2013 and 2021; and the number of patients receiving topiramate prescriptions increased 57%. Given the recent data supporting risk for adverse neurodevelopmental outcomes, there is growing alarm and calls for regulatory agencies to (i) identify suitable alternatives to topiramate during preconceptional counseling for women of childbearing potential, and (ii) more closely monitor topiramate use (Anonymous 2022; Braillon 2023; Wells-Gatnik and Martelletti 2023). The uncertainty around the neurodevelopmental safety of topiramate has led the European Medicines Agency to recommend that topiramate be placed on a pregnancy prevention program (EMA 2023). Considering the ongoing debate, determining the molecular, cellular, and pharmacological factors contributing to neurodevelopmental risk of topiramate and other ASMs is essential knowledge for identifying next generation ASM therapies that provide wider margins of safety to the mother and developing fetus.

Abbreviations:

ASM

antiseizure medication

ASD

autism spectrum disorder

GABA

gamma-aminobutyric acid

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

NMDA

N-methyl-D-aspartate

HEK

human embryonic kidney

UKEPR

UK Epilepsy and Pregnancy Register

IQ

intelligence quotient

CI

confidence interval

SCAN-AED

Scandinavia multi-registry study of antiepileptic drug teratogenicity

ADHD

attention-deficit/hyperactivity disorder

ABC

adaptive behavior composite

UK

United Kingdom

HDAC

histone deacetylase

MDA

malondialdehyde