Figure. Prevalence of major malformations in antiepileptic drug (AED) polytherapies with and without valproic acid (VPA). Results shown are from the North American AED Pregnancy Registry and 2 previously published studies, the UK Epilepsy Pregnancy Register6 and the International Lamotrigine Pregnancy Registry.7 CBZ indicates carbamazepine; LTG, lamotrigine.
Holmes LB, Mittendorf R, Shen A, Smith CR, Hernandez-Diaz S. Fetal Effects of Anticonvulsant PolytherapiesDifferent Risks From Different Drug Combinations. Arch Neurol. 2011;68(10):1275-1281. doi:10.1001/archneurol.2011.133
Author Affiliations: North American AED (Antiepileptic Drug) Pregnancy Registry (Drs Holmes, Hernandez-Diaz, and Mittendorf and Mss Shen and Smith), Genetics Unit, MassGeneral Hospital for Children (Dr Holmes), and Harvard School of Public Health (Dr Hernandez-Diaz), Boston, Massachusetts; and Departments of Obstetrics and Gynecology and Pediatrics, Loyola University Health System, Maywood, Illinois (Dr Mittendorf).
Objective To determine the frequency of malformations among infants born to women who had taken lamotrigine or carbamazepine as part of polytherapy during the first trimester of pregnancy.
Design A cohort of women enrolled during pregnancy in the North American AED (Antiepileptic Drug) Pregnancy Registry between February 1, 1997, and June 1, 2010. Information on AED use and demographic characteristics was collected in 3 telephone interviews.
Setting United States and Canada.
Patients A total of 6857 pregnant women taking an AED for any reason.
Main Outcome Measures Major congenital malformations were identified at birth and through the first 12 weeks after delivery. Diagnoses were based on the mother's report and confirmed by medical records. The risks of malformations were compared between polytherapy and monotherapy groups, using exact odds ratios (ORs) and 95% confidence intervals (CIs).
Results The risk of malformations was 1.9% among infants exposed to lamotrigine as monotherapy (n = 1441). Among the infants exposed to lamotrigine as polytherapy (n = 505), the risks were 9.1% for lamotrigine plus valproate sodium (OR, 5.0; 95% CI, 1.5-14.0) and 2.9% for lamotrigine plus any other AEDs (1.5; 0.7-3.0). The risk of malformations was 2.9% for the infants exposed to carbamazepine monotherapy (n = 1012). For the infants exposed to carbamazepine as polytherapy (n = 365), the risks were 15.4% for carbamazepine plus valproate (OR, 6.2; 95% CI, 2.0-16.5) and 2.5% for carbamazepine plus any other AEDs (0.8; 0.3-1.9). Confounding by factors such as periconceptional vitamin use, cigarette smoking, alcohol use, and chronic maternal diseases did not explain the results.
Conclusions The risk of malformations among infants exposed to lamotrigine and carbamazepine as polytherapy was higher than the corresponding monotherapies only when the polytherapy includes valproate. These findings suggest that counseling for fetal risks from AED polytherapy should be based on the specific drugs included.
For many years, it has been taught that the fetal risks from prenatal exposure to 2 or more anticonvulsant drugs are significantly greater than the risks from exposure to 1 anticonvulsant drug. In some studies,1,2 the risk of malformations increased with the addition of each anticonvulsant drug. The risk of malformations is widely considered to differ among anticonvulsant drugs when used as monotherapy. For example, in the North American AED (Antiepileptic Drug) Pregnancy Registry, the rate of major malformations for valproate sodium–exposed infants was 10.7%,3 for phenobarbital-exposed infants was 6.5%4 and for lamotrigine-exposed infants was 2.3%5 when used as monotherapies.
Recent analyses6,7 of specific AED polytherapies have suggested that the rate of major malformations was much higher when one of the anticonvulsants was valproate than when the second drug was carbamazepine or lamotrigine.
We present herein findings from the North American AED Pregnancy Registry (referred to as the Registry) on the risk of major malformations in infants exposed during pregnancy to the anticonvulsant drugs carbamazepine and lamotrigine in polytherapies with and without valproate.
To enroll in the Registry, women called its toll-free telephone number (1-888-233-2334). After informed consent was obtained, eligible women were interviewed 3 times: at enrollment, at 7 months' gestation, and 8 to 12 weeks after the expected date of delivery. During a computer-assisted telephone interview, the enrollees were asked about the dose, frequency, and medical indication of each anticonvulsant drug taken; the signs and symptoms of epilepsy (or a mood disorder); the apparent causes of the epilepsy; demographic characteristics; habits (eg, alcohol use, cigarette smoking, and use of illicit drugs); other potential teratogenic exposures (eg, maternal diabetes or taking the acne medication isotretinoin [Accutane]); other prescribed and over-the-counter medications taken; and family history of epilepsy (or mood disorders) and of birth defects. Enrollees were asked whether they were taking a multivitamin supplement—and a folic acid supplement in particular—at conception.
A woman was classified as being a “pure” prospective enrollee if she had not had any prenatal screening that could have identified malformations at the time of enrollment and, therefore, could not have known whether her fetus had a malformation. The “traditional” prospective enrollees could have some knowledge of the health status of the fetus, typically after having prenatal screening by ultrasonography. The term undetermined was used if the information recorded in the interviews, primarily in the early years of the Registry, was not sufficient to determine whether the study participant was a pure prospective enrollee. Women were not enrolled after the pregnancy had ended. Participants were eligible for analyses if they completed the follow-up and had a live-born or stillborn (>20 weeks' gestational age) infant, an infant who died in the first 28 days of life (ie, neonatal death), or an elective termination of pregnancy for which the findings in postmortem examinations were available.
Monotherapy was defined as exposure to only 1 anticonvulsant drug during the first 16 weeks of gestation. Polytherapy was defined as exposure to 2 or more anticonvulsant drugs simultaneously or in sequence at any time during this same period of pregnancy.
The informed consent document had been reviewed and approved annually since 1996 by the Human Studies Committee of the Massachusetts General Hospital and Partners HealthCare in Boston. Each enrolled mother was identified only by a study number in the analytic data to protect the confidentiality of the information compiled. Medical record release forms were mailed to the mothers to be signed and returned. These signed requests were sent to the woman's neurologist (or other treating physicians) and her infant's physicians. When mothers did not sign medical record release forms, they were recontacted and encouraged to provide this written permission. The parents of all infants with major malformations were interviewed whenever possible to review the family history for malformations and to ask whether a specific syndromic diagnosis had been established.
A major malformation was defined as a structural abnormality with surgical, medical, or cosmetic importance. The written descriptions in the pediatricians' examinations were reviewed separately by the clinical teratologist (L.B.H.), who was blinded to exposure status, to determine inclusion or exclusion. The physical features excluded were as follows: (1) minor anomalies (eg, simian crease); (2) deformations or positional deformities (eg, torticollis); (3) features due to prematurity (eg, undescended testes in an infant born at <37 weeks' gestational age); (4) birthmarks (eg, hemangiomas); (5) genetic disorders (eg, albinism); (6) chromosome abnormality (eg, Down syndrome); (7) functional deficits (eg, a failed hearing test during newborn screening); (8) any finding by prenatal sonography or at surgery or autopsy (eg, absence of 1 kidney) that was not identified by an examining pediatrician; (9) a physiological finding, such as a muscular ventricular septal defect, that caused no symptoms in the newborn and closed spontaneously; and (10) anomalies detected only by prenatal ultrasonography (and not during the physical examination by the pediatrician), such as unilateral renal agenesis. These physical features were excluded because not all infants had such examinations. In addition, among those who did have prenatal screening by ultrasonography, there was no uniformity of the timing of the studies, the equipment used, the expertise of the sonologists, or the body mass index of all women screened. The examination by a physician at birth was used as the gold standard for the detection of all malformations.
Two unexposed comparisons were used: one internal and one external. The internal comparison group consisted of pregnant women recruited among the friends and family members of each enrolled woman. These women were asked to call the Registry and were interviewed by the same personnel using the same questionnaires. The interviewer confirmed that women enrolled in the unexposed internal comparison group were not taking an anticonvulsant drug. The frequency of malformations identified before the postpartum interview, when the infant was 8 to 12 weeks of age, was compared between the AED-exposed infants and the internal unexposed comparison group.
The second unexposed comparison group, the external comparison group, included the live-born and stillborn infants, as well as elective terminations because of fetal anomalies, surveyed by the Active Malformations Surveillance Program at Brigham and Women's Hospital.8,9 This surveillance program, directed since 1972 by one of the authors (L.B.H.), uses the same inclusion and exclusion criteria as the Registry. Because the period for identification of major malformations in this comparison group was prenatally and between birth and 5 days of age, for the external comparison the period for identifying major malformations in the anticonvulsant-exposed group was restricted to prenatally or between birth and 5 days of age.
Each of the Registry's sponsoring companies has 1 or 2 representatives on its Steering Committee. However, the members of the Scientific Advisory Committee (listed in the “Additional Contributions” section of the “Acknowledgments”) and the staff of the Registry meet separately every 6 months, at which time decisions are made about the release of any findings. The findings for infants exposed to specific drugs are discussed blindly, that is, without identifying the drug. After a decision is made by the Scientific Advisory Committee, the members of the Steering Committee are informed of the decision. However, only the company that manufactures the drug being evaluated is informed of the identity of the drug. That company has a 30-day option of obtaining additional information from the Registry. After this period, the members of the Scientific Advisory and Steering Committees are informed about the identity of the drug for which the findings are being released.
We present herein the risks of major malformations in the infants born to enrolled women who took the most commonly reported AEDs—carbamazepine or lamotrigine—as polytherapy in the first 16 weeks of pregnancy. This risk was compared with that among women who used these drugs as monotherapy, as well as with the risk in the 2 unexposed reference groups. Exact odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated through multivariate logistic regression using SAS statistical software (version 9.1; SAS Institute, Inc, Cary, North Carolina).
The study included 6857 women taking an AED during some portion of pregnancy and 441 unexposed women who had enrolled in the Registry and had an estimated date of delivery between February 1, 1997, and June 1, 2010. The total number of pregnancies with analyzable data were as follows: 1441 for lamotrigine monotherapy, 505 for lamotrigine polytherapy, 1012 for carbamazepine monotherapy, 365 for carbamazepine polytherapy, and 408 for the unexposed control group. Altogether, 62.1% of these participants were pure prospective enrollees (Table 1).
The demographic characteristics of the enrolled women showed that, compared with monotherapy users, the polytherapy users were less educated and more often single mothers, smokers, and users of recreational drugs (Table 2). In addition, their seizures had begun earlier in life on average, and more of these women had seizures during pregnancy.
Among the 505 women using lamotrigine in polytherapy, the most common second AEDs used concomitantly or sequentially during the first 16 weeks of gestation were valproate (n = 55), phenobarbital (n = 35), and carbamazepine (n = 99) (Table 3). Among the infants exposed to lamotrigine polytherapies during their first trimester, 18 (3.6%) had major malformations (Table 4). Fifteen of the 18 malformations were detected between birth and 5 days of age and 3 were detected between 5 days and 12 weeks of age. The risks of malformations among the infants exposed to lamotrigine polytherapies were greater than those observed in infants exposed to lamotrigine monotherapy (1.9%) and the risks among unexposed controls in either of the comparison groups: internal (1.2%) and external (1.6%). Although based on small numbers, the risk of major malformations was higher for infants exposed to lamotrigine polytherapies that included valproate (9.1%) or phenobarbital (4.0%). Compared with lamotrigine monotherapy, the crude OR was 1.9 (95% CI, 1.0-3.5) for any lamotrigine polytherapy; it was 5.0 (1.5-14.0) when the polytherapy included valproate and 1.5 (0.7-3.0) when the polytherapy did not include valproate. For any lamotrigine polytherapy, adjustment for potential confounders moved the OR to 1.7 (95% CI, 0.9-3.1).
Among the 365 infants exposed to carbamazepine polytherapies, the most common AED combinations included valproate (n = 39), phenobarbital (n = 36), and lamotrigine (n = 99) (Table 5). The number of infants with major malformations after exposure to carbamazepine during their first trimester was 29 (2.9%) in the monotherapy group and 14 (3.8%) in the polytherapy group. Thirteen of the 14 malformations were detected between birth and 5 days of age; 1 was detected between 5 days and 12 weeks of age (Table 6). The risk of major malformations was particularly elevated for infants exposed to carbamazepine polytherapies that included valproate (15.4%). Compared with carbamazepine monotherapy, the crude OR was 1.4 (95% CI, 0.6-2.7) for any carbamazepine polytherapy; it was 6.2 (2.0-16.5) when the polytherapy included valproate and 0.8 (0.3-1.9) when the polytherapy did not include valproate. For any carbamazepine polytherapy, adjustment for potential confounders did not change the OR (1.4; 95% CI, 0.7-2.8).
Results did not change when these comparisons were restricted to pure prospective study participants, and the results were similar among nonsmokers and nonusers of recreational drugs. Comparisons with the external control group led to similar findings (Tables 3 and 5). We excluded from this analysis infants whose malformations were identified after 5 days of age because the malformed infants in the external comparison group were identified in the first 5 days of life.7,8
The anticonvulsant drugs carbamazepine and lamotrigine were the most commonly used AEDs as both monotherapy and polytherapy in the treatment of pregnant women enrolled in the Registry between 1997 and 2010. The enrollment of 365 and 505 pregnant women, respectively, who had received each of these drugs in polytherapies made it possible to evaluate the risk of major malformations overall.
Historically, the risk of malformations in polytherapy-exposed pregnancies was significantly higher than that among monotherapy-exposed pregnancies.1 The present analysis showed that, compared with unexposed pregnancies, a significantly increased risk of malformations occurred in infants exposed to polytherapy of carbamazepine or lamotrigine that included valproate, but not when those drugs were combined with AEDs other than valproate.
These findings are similar to the results published by the UK Epilepsy and Pregnancy Register6 and in the International Lamotrigine Pregnancy Registry.7 In the UK Epilepsy and Pregnancy Register,6 the rate of malformations was 8.8% among 62 pregnancies exposed to carbamazepine and valproic acid and 9.6% in 141 pregnancies exposed to lamotrigine and valproic acid (Figure). By comparison, none of the 118 infants exposed to carbamazepine and lamotrigine as polytherapy was malformed. In the International Lamotrigine Pregnancy Registry,7 the rate of malformations was 12.5% in 88 pregnancies exposed to lamotrigine plus valproate and 2.7% in 182 first-trimester exposures to lamotrigine polytherapy that did not include valproate. The comparable findings in these 2 separate pregnancy registries suggest that the fetal risks for malformations vary for the specific drugs used in polytherapy.
Although some studies1,2 had suggested that the risk of malformations increased with the addition of each anticonvulsant drug, no recent data on a higher risk among infants exposed to 2, 3, or 4 anticonvulsant drugs with or without valproic acid has been reported to our knowledge. However, a dose-response relationship for valproate has been suggested in at least 2 studies, which reported a higher risk for spina bifida11 and “dysmorphic features and structural anomalies”12(pp514-515) among the infants exposed to valproic acid at high doses.
Two unexposed comparison groups are used by the Registry, one external8,9 and the other internal.10 The time window for the detection of major malformations was birth to 5 days of age in the external comparison group8,9 and was up to 12 weeks of age in the internal comparison group.10 It was reassuring to see that there was no significant difference in the main conclusions from each of these comparisons.
Larger sample sizes will make it possible to determine the frequency of common malformations, such as myelomeningocele, heart defects, and oral clefts, in each of the polytherapy treatment groups. The study of larger numbers of monotherapy-exposed infants has made it possible to identify statistically significant risks of specific malformations. These have included (1) for valproic acid–exposed infants, a higher risk of spina bifida, cleft palate, hypospadias, and craniosynostosis13; (2) for carbamazepine-exposed infants, a higher risk of spina bifida14; and (3) for lamotrigine-exposed infants, a higher risk of cleft palate.5 More experience in a larger number of monotherapy-exposed pregnancies is needed to confirm and extend these correlations.
In conclusion, counseling for fetal risks from exposure to AED polytherapy should be based on the specific drugs included in the combinations. Polytherapies that include valproate pose a higher risk to the fetus than those without this drug.
Correspondence: Lewis B. Holmes, MD, Genetics Unit, MassGeneral Hospital for Children, 175 Cambridge St, Charles River Plaza, South Building, Fifth Floor, Room 504, Boston, MA 02114 (firstname.lastname@example.org).
Accepted for Publication: April 5, 2011.
Published Online: June 13, 2011. doi:10.1001/archneurol.2011.133
Author Contributions: Dr Mittendorf had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Holmes, Mittendorf, Smith, and Hernandez-Diaz. Acquisition of data: Shen and Smith. Analysis and interpretation of data: Holmes, Mittendorf, and Hernandez-Diaz. Drafting of the manuscript: Holmes, Mittendorf, and Hernandez-Diaz. Critical revision of the manuscript for important intellectual content: Holmes, Mittendorf, Shen, Smith, and Hernandez-Diaz. Statistical analysis: Mittendorf and Hernandez-Diaz. Obtained funding: Holmes. Administrative, technical, and material support: Shen and Smith. Study supervision: Holmes, Mittendorf, Shen, and Smith.
Financial Disclosure: All authors, except for Dr Mittendorf, received salary support from funds provided since 1997 by 6 sponsors of the Registry. At the time this manuscript was written, the sponsors were Abbott, Eisai, Novartis, Ortho-McNeil, Pfizer, and Sepracor.
Funding/Support: The North American AED Pregnancy Registry is supported by funds provided by 6 sponsors: Abbott Laboratories, Eisai Inc, Novartis, Ortho-McNeil, Pfizer Pharmaceuticals, and Sepracor Inc.
Additional Contributions: The authors thank the pregnant women who enrolled and assisted in obtaining medical information on them and their infants. The following members of the Scientific Advisory Committee contributed to the development of this study and this analysis: Janet Cragan, MD, Atlanta, Georgia; Allen Hauser, MD, New York, New York; Brandy Fureman, PhD, Bethesda, Maryland; and Mark Yerby, MD (chair), Portland, Oregon.