FormalPara Key Points

Information was found on specific congenital anomalies in fetuses exposed to lamotrigine, topiramate, levetiracetam, gabapentin, and oxcarbazepine monotherapy in the first trimester.

The possible association between cleft lip and hypospadias and the use of topiramate in pregnancy should be investigated further.

1 Introduction

Users of anti-epileptic drugs (AEDs) regularly include women of childbearing age. During pregnancy, the treatment of epilepsy often needs to be continued, and—if possible—a pregnant woman with epilepsy should be free of seizures, because a seizure could harm both mother and fetus [1]. On the other hand, harm to the fetus from the use of medication in pregnancy should also be minimized. A balance must be found between benefits and risks for mother and child through agreement between the pregnant woman and the physician [2, 3]. Objective scientific information on the risks of AEDs to the fetus is essential for these discussions and decisions.

AEDs are classified in two groups: older AEDs and newer AEDs, the latter being introduced during the last two decades. Since the late 1990s, the use of newer AEDs has increased, especially for indications other than epilepsy such as neuropathic pain, mood disorders, migraine, and depression [4–6]. Lamotrigine, topiramate, gabapentin, and pregabalin are the most commonly used newer AEDs [4]. Women in particular seem to use more newer AEDs than do men, probably in an effort to avoid the use of valproate during childbearing years [7, 8]. About 0.5 % of pregnant women in Europe use AEDs, most often carbamazepine, valproic acid, or lamotrigine; the most frequently used AEDs differ between countries [9].

More information is needed on the risk of congenital anomalies in exposed pregnancies for the newer AEDs. These newer AEDs might have an increased risk for specific congenital malformations. We performed an extensive literature search to find signals of higher risks of specific congenital anomalies in relation to the use of newer AEDs in pregnancy.

2 Methods

2.1 Search Strategy

The newer AEDs included in this review were lamotrigine, topiramate, levetiracetam, gabapentin, oxcarbazepine, eslicarbazepine, felbamate, lacosamide, pregabalin, retigabine, rufinamide, stiripentol, tiagabine, vigabatrin, and zonisamide.

We searched PubMed for original articles using the following search strategy:

  1. 1.

    congenital abnormalities [medical subject heading; MeSH] OR pregnancy complications/drug therapy OR pregnancy complications/drug effects OR pregnancy outcome [MeSH]

    AND

  2. 2.

    (felbamate OR gabapentin OR lacosamide OR lamotrigine OR levetiracetam OR pregabalin OR topiramate OR vigabatrin OR eslicarbazepine OR oxcarbazepine OR rufinamide OR stiripentol OR zonisamide OR tiagabine OR retigabine OR pheneturide) NOT (models, animal [MeSH] OR animal experimentation [MeSH]).

We searched EMBASE for all newer AEDs separately using the following search strategy:

‘lamotrigine (or other drug) AND congenital malformation AND pregnancy NOT review’.

The search was conducted on 12 November 2014.

2.2 Selection

The articles were selected using the following inclusion criteria:

  • original randomized controlled trials or observational studies.

  • exposure to a newer AED as monotherapy in the first trimester (≤12 weeks of gestation).

  • information on congenital anomalies.

  • the most recent update of studies based on the same long-term databases or pregnancy registries.

  • enrolment of the pregnant women before the outcome of the pregnancy was known in the cohort and observational studies.

The articles from PubMed were selected based on the title and abstract. They were classified into two groups (studies including one AED and studies including more than one AED) and were read carefully, including the appendices. The articles were selected based on the inclusion criteria and categorised per AED; thus, the articles with more than one AED were categorised more than once. Additional articles found in EMBASE or the references that met the inclusion criteria were added (Fig. 1).

Fig. 1
figure 1

Selection process for articles included in the review

Full size image

2.3 Data Extraction

We reclassified the selected studies into five types based on the study design, which could differ from the design indicated by the authors:

  1. 1a.

    prospective cohort studies (with reference group).

  2. 1b.

    retrospective cohort studies (with reference group).

  3. 2a.

    prospective exposed groups (without reference group).

  4. 2b.

    retrospective exposed groups (without reference group).

  5. 3.

    case–control studies.

We analysed studies that included stillbirths, fetal deaths, and pregnancy terminations (‘all births’) and studies describing only live births both separately and in combination.

2.4 Data Analysis

Two authors (JJ and EG) reclassified all specific congenital anomalies into the congenital anomaly subgroups of the European Surveillance of Congenital Anomalies (EUROCAT) [10]. Only major anomalies were included in the analysis. When no information was provided on specific congenital anomalies, we followed the definition of major congenital anomalies in the article.

We calculated the overall major anomaly rate for every AED, for which cohort studies or studies with exposed groups (1a, 1b, 2a, and 2b) were selected, using the number of fetuses as the denominator. If the analysis in an article was based on pregnancies, and the exact number of fetuses was unknown, we used the number of pregnancies, assuming that multiple pregnancies are rare and that this difference would not greatly influence the results.

We calculated the prevalence of any specific congenital anomaly subgroup from studies with sufficiently detailed information.

The prevalence of specific anomalies was compared with the prevalence of the anomaly subgroup in the EUROCAT AED database, which covers 10,061,059 births from 21 regions in Europe (1995–2011) and has been described previously [11, 12]. We excluded registrations from this database with maternal AED exposure, maternal epilepsy, or chromosomal anomalies. A significantly higher prevalence of a specific congenital anomaly, based on three or more fetuses with anomalies among fetuses exposed to newer AEDs compared with the prevalence in the reference database, was considered a signal. This procedure is based on the method used by Jentink et al. [12] (see the Electronic Supplementary Material [ESM], Sect. 2), who searched for signals for investigation in a case–control study and noted that it is preferable to detect only the strongest signals. If one or two fetuses with a specific congenital anomaly are found within all the literature searched, it could be that coincidence may have played a major role. We decided to draw the limit at three or more fetuses with a specific anomaly. The case-control studies found were not included in the analysis but were summarized separately.

2.5 Statistical Analysis

We calculated the anomaly rate and confidence interval using the method described by Newcombe [13]: the Wilson ‘score’ method. To compare the prevalence of specific congenital anomalies in the fetuses exposed to newer AEDs with those in the reference database, we used Pearson’s Chi-squared test and Yates’ continuity correction. We used Microsoft® Office Excel (Microsoft® Corp., Redmond, WA, USA) and R version 3.1.0 (Free Software, Free Software Foundation, Boston, MA, USA) software for the data analysis. A p value <0.05 was considered statistically significant.

3 Results

3.1 General Results

We selected 341 articles from PubMed. In addition, six articles from EMBASE and references were added (Fig. 1; grey background). The 30 selected studies were categorized per AED and study design, as shown in Table 1. We found no randomized controlled trials.

Table 1 Number of included studies, stratified by anti-epileptic drug and study design
Full size table

Table 2 shows an overview of the cohort- and exposed group studies found, with the numbers of fetuses or pregnancies exposed to lamotrigine, topiramate, levetiracetam, gabapentin, or oxcarbazepine and the percentage of major congenital anomalies. Vajda et al. [14] provided no information on whether the congenital anomalies were major or minor, so we counted the total number of congenital anomalies. One study [15] included only live births.

Table 2 Overview of the studies including fetuses/pregnancies exposed to monotherapy with antiepileptic drugs in the first trimester
Full size table

Two studies were based on international pregnancy registries and were analysed separately: Cunnington et al. [16] used the International Lamotrigine Pregnancy Registry, with information on specific congenital anomalies of fetuses exposed to lamotrigine; Tomson et al. [17] based their study on the European Registry of Antiepileptic Drugs and Pregnancy (EURAP), a registry of pregnancies exposed to all types of AEDs in 42 countries. The latter provided only information about some congenital anomaly subgroups from pregnancies exposed to newer AEDs. Whether the congenital anomalies were major or minor was not defined, so we used the total number of congenital anomalies to calculate the anomaly rate.

No congenital anomaly was found in fetuses exposed to felbamate, lacosamide, pregabalin, tiagabine, vigabatrin, or zonisamide, but the number of exposed fetuses was very low (1 to 10), except for zonisamide (n = 97). No studies describing exposed pregnancies were found for eslicarbazepine, retigabine, rufinamide, or stiripentol.

The studies based on the International Lamotrigine Pregnancy Registry [16] and EURAP registry [17], both international pregnancy registries, were analysed separately because they involve data from, respectively, 43 and 42 countries and might overlap with the other studies.

Table 3 shows the prevalence of specific congenital anomaly subgroups according to the EUROCAT classifications based on the studies with information on specific congenital anomalies compared with that in the reference (EUROCAT AED) database. Two studies [18, 19] with information on congenital anomaly subgroups on a higher level (e.g., ‘nervous system’ or ‘cardiovascular heart defects’) among lamotrigine-exposed pregnancies were included in the calculation of the prevalence of these higher-level subgroups, using another denominator.

Table 3 Prevalence of specific anomaly subgroups of fetuses/pregnancies exposed to lamotrigine, topiramate, levetiracetam, gabapentin, or oxcarbazepine compared with the reference database
Full size table

Table 4 is a summary of the odds ratios (ORs) found in the three selected case–control studies. These case–control studies used databases of pregnancies with congenital anomalies and the (adjusted) ORs of specific congenital anomalies was determined comparing use and no use of specific AEDs.

Table 4 Summary of the selected case–control studies
Full size table

3.2 Lamotrigine

A total of 5197 fetuses exposed to lamotrigine monotherapy in the first trimester were described, 123 of which had major congenital anomalies. The anomaly rate was 2.3 % in the studies with all births; inclusion of the study with only live births [15] changed the rate to 2.4 % (Table 2). In the International Lamotrigine Pregnancy Registry [16], the anomaly rate was 1.7 %. The study based on the EURAP registry [17] reported an anomaly rate of 2.9 %.

The analysis of specific congenital anomalies is based on seven studies [18–24]. The data from the International Lamotrigine Pregnancy Registry [16] were also compared with the reference database (Table 3). For most specific congenital anomalies, the prevalence among fetuses exposed to lamotrigine monotherapy in the first trimester is comparable with the prevalence of the reference database. In the analysis based on the International Lamotrigine Pregnancy Registry [16], the specific congenital anomalies anencephaly and transposition of great vessels were significantly more prevalent than in the reference database, based on three or more fetuses with anomalies (1.77 vs. 0.20/1000, p < 0.001; 1.77 vs. 0.30/1000, p = 0.005, respectively).

In two case–control studies by Dolk et al. [25] (case-malformed–control study) and Werler et al. [26], no association between the use of lamotrigine in pregnancy and congenital anomalies was found (Table 4).

3.3 Topiramate

In seven studies, 553 fetuses were exposed to topiramate, 20 of whom had congenital anomalies (Table 2). The anomaly rate was 3.6 % without and with the study including only live births [15], which was higher than for the other newer AEDs. The study based on the EURAP registry [17] reported an anomaly rate of 6.8 %. The prevalence of specific congenital anomalies was calculated from six studies [14, 15, 20, 21, 27, 28]. Two specific congenital anomalies were significantly more prevalent, based on three or more fetuses with anomalies, in the fetuses exposed to topiramate than in the reference database: cleft lip with or without cleft palate (13.86 vs. 0.84/1000, p < 0.001) and hypospadias (7.92 vs. 1.53/1000, p = 0.002) (Table 3).

The case–control study by Margulis et al. [29] found an OR >1 for cleft lip with/without palate for pregnancies exposed to topiramate.

3.4 Levetiracetam

The major congenital anomaly rate of fetuses exposed to levetiracetam monotherapy in the first trimester was 1.1 % (11 of 957) and 1.6 % in the EURAP study [17]. Three studies provided specific information on congenital anomalies [20, 30, 31]. Based on three or more cases among fetuses exposed to levetiracetam, no congenital anomaly was more prevalent than in the reference database (Table 3).

3.5 Gabapentin

Four of 266 fetuses exposed to gabapentin in the first trimester had major congenital anomalies (Table 2). Congenital anomalies were not found in the study that included only live births [15] or in the EURAP study [17]. The overall anomaly rate was 1.5 %. We used seven studies with detailed information [15, 21, 28, 32–35] for the analysis of specific major congenital anomalies. No significant differences were found between the prevalence of the specific congenital anomalies and the prevalence in the reference database (Table 3).

3.6 Oxcarbazepine

Eight of 414 fetuses exposed to oxcarbazepine in the first trimester of pregnancy had major congenital anomalies. The anomaly rate was 2.0 % of the studies with all births and 1.9 % including the study with live births [15]. The EURAP study [17] had an anomaly rate of 3.3 %. The prevalence of specific congenital anomalies was calculated by including eight studies [15, 20, 36–41] (Table 3). Based on three or more cases, no higher prevalence of specific congenital anomalies than in the reference database was found.

4 Discussion

In this literature review, we found 30 articles that met our inclusion criteria. The total major anomaly rate of fetuses exposed to topiramate was slightly higher than of those exposed to the other newer AEDs. The anomaly rate for the EURAP study [17] was higher, probably because minor anomalies were included.

No congenital anomalies were found among pregnancies exposed to felbamate, lacosamide, pregabalin, tiagabine, vigabatrin, or zonisamide. Because the number of observations is very low, no conclusions could be drawn about the risk of congenital anomalies for pregnancies exposed to these AEDs. No articles on congenital anomalies were found in relation to the use of eslicarbazepine, retigabine, rufinamide, or stiripentol.

Four signals of possible associations of specific congenital anomalies and use of newer AED monotherapy in the first trimester of pregnancy were detected: two related to lamotrigine (anencephaly and transposition of great vessels) and two related to topiramate (cleft lip with or without cleft palate and hypospadias).

4.1 Lamotrigine

Two signals were found in the analysis of the study based on the International Lamotrigine Pregnancy Registry [16]. The first signal was a significantly higher prevalence of anencephaly (three cases) compared with the reference database. Anencephaly was also found in the analysis of the other combined studies but not significantly more frequently than in the reference database. The high prevalence of anencephaly was mentioned in the discussion of the study [16], but an association was not concluded because this finding was not supported by other studies.

The prevalence of transposition of great vessels is higher in the International Lamotrigine Pregnancy Registry [16] than in the reference database. In the combined analysis, transposition of great vessels was found; however, the prevalence was similar to that in the reference database. Cunnington et al. [16] stated that several cases of severe cardiac defects were found but that other studies fail to support an increased frequency of cardiac heart defects among fetuses exposed to lamotrigine in utero.

In this review, we did not find an association between lamotrigine and orofacial clefts, which was one of the conclusions in a previous review [42]. This conclusion was based on studies from the North American AED Pregnancy Registry, which found an increased frequency of isolated cleft palate in infants exposed to lamotrigine during pregnancy compared with the reference population [20, 43]. The case-malformed–control study by Dolk et al. [25] did not find an association between lamotrigine and orofacial clefts. A letter responding to the review stated that the comparison group in the North American AED Pregnancy Registry had a low prevalence of orofacial clefts compared with other databases [44]. In the case–control study by Werler et al. [26] in the USA, based on the National Birth Defect Prevention Study, no significant association was found between orofacial clefts and lamotrigine. That we did not find signals of an association between orofacial clefts and lamotrigine exposure in pregnancy, combining all known cohort studies, supports that there is no association between lamotrigine and orofacial clefts.

4.2 Topiramate

We found a signal for the congenital anomaly subgroup cleft lip with or without palate in fetuses exposed to topiramate in the first trimester of pregnancy. The prevalence is significantly higher than in the reference database and is based on seven fetuses. Five of these were from the study by Hernández-Díaz et al. (North American AED Pregnancy Registry) [20], and two were from the study by Hunt et al. (UK Epilepsy and Pregnancy Register) [27]. The conclusion of the study by Hernández-Díaz et al. [20] mentioned an association between topiramate monotherapy use in pregnancy and cleft lip. The conclusion of the study by Hunt et al. [27] also mentioned an association between topiramate and orofacial clefts, but their analysis was based on two fetuses with exposure to topiramate as monotherapy and two fetuses with topiramate in combination with other AEDs. The case–control study by Margulis et al. [29], using two North-American databases, found an association between topiramate and cleft lip with or without palate. Mines et al. [45] used four different data sources from the USA and found a higher prevalence of cleft lip in infants exposed to topiramate monotherapy than in infants of mothers formerly exposed to topiramate or other AEDs and mothers with similar medical profiles not exposed to topiramate. We conclude there is a strong signal indicating an association between cleft lip and topiramate monotherapy in the first trimester of pregnancy. This signal should be investigated further.

The second signal among fetuses exposed to topiramate monotherapy in the first trimester of pregnancy was a higher risk of hypospadias. Four fetuses with hypospadias were found in three different studies: one from the study by Vajda et al. (Australian Pregnancy Registry) [14], one from the study by Hunt et al. (UK Epilepsy and Pregnancy Register) [27], and two from the study by Hernández-Díaz et al. (North American AED Pregnancy Registry) [20]. Prevalence was significantly higher than in the reference database. Neither Hernández-Díaz et al. [20] nor Hunt et al. [27] mentioned a possible association between hypospadias and the use of topiramate monotherapy in their conclusions. Vajda et al. [28] also did not mention this association in their conclusion. Although they did indicate an association in a previous study [46], it was not for monotherapy.

4.3 Strengths and Limitations

A limitation of this review was the small sample size of fetuses exposed to newer AEDs in the observational studies included in the analysis of specific congenital anomalies. The number of exposed fetuses with information for specific congenital anomalies was 505 for topiramate, 756 for levetiracetam, 125 for gabapentin, and 340 for oxcarbazepine. The specific congenital anomaly subgroups with a prevalence of 0.1–5/1000 are most likely missed in these small samples. For lamotrigine, the number of pregnancies with information on specific congenital anomalies was higher (n = 1766 and 1699 from the International Lamotrigine Pregnancy Registry [16]), which increased the chance of finding signals.

We found few or no exposed fetuses for six other newer AEDs, except for zonisamide (97 exposed fetuses) and no congenital anomalies. We found no studies on exposure in pregnancy for five newer AEDs and therefore could draw no conclusions on possible associations with specific congenital anomalies.

Another limitation is that enrolment in most of the studies was voluntary. Selection bias may be present because well-educated, motivated, or interested women are more likely to register. It was unknown for most pregnancies whether the women took medication (prescribed or over the counter) other than the AEDs. This possible co-medication could also be teratogenic, as could other exposures such as smoking and alcohol use or other unknown circumstances affecting the fetus.

Not all the studies excluded chromosomal anomalies. A few major congenital anomalies from fetuses with chromosomal anomalies may have been included. Because we reclassified the anomalies according to EUROCAT, we consider this will not affect the results.

Differences also exist between the cohorts used in the literature and those in the EUROCAT AED database. The latter contains data from 1995 to 2011, whereas the literature includes studies with data from earlier years, starting from 1972 [41]. Because the data before 1995 are sparse, we consider this unlikely to have affected the conclusions. However, the data from the EUROCAT AED database are mainly from Europe, whereas the literature includes studies from the USA. In particular, one important study with considerable data is from the USA [20]. However, we think the differences between the populations are not large enough to affect the results.

5 Conclusion

In this literature review, we sourced information on specific congenital anomalies in fetuses exposed to lamotrigine, topiramate, levetiracetam, gabapentin, or oxcarbazepine monotherapy in the first trimester. Few or no exposed pregnancies and no congenital anomalies were reported for the remaining AEDs. Four signals of possible associations between the use of newer AEDs and specific congenital anomalies were detected: lamotrigine and anencephaly, lamotrigine and transposition of great vessels, topiramate and cleft lip, and topiramate and hypospadias. Of these, only the signal for the association between topiramate and cleft lip with or without cleft palate was considered strong. There might be an association between hypospadias and the use of topiramate in pregnancy, but possibly only for polytherapy. These signals should be investigated further.