The teratogenic effect of antiepileptic drugs was suggested by Meadow in 1968, who associated minor facial anomalies in 14–33% of the children exposed. Some malformative syndromes are associated with specific drugs, such as phenytoin (PHT), trimethadione, valproate (VPA), primidone, carbamazepine (CBZ), and phenobarbital. Most of these syndromes have been related to common metabolic pathways, high levels of the drug and/or polytherapy; however, some authors have reported the association of facial anomalies with the genetic etiology of the epilepsy.

Facial dysmorphism has been described in a non-specific, non-objective manner and without comparison to a control group; the clinical approach could be improved with the addition of anthropometric measurement.

The purpose of this article is to describe the facial characteristics of the offspring of epileptic mothers with and without pharmacological treatment by means of 22 anthropometric measurements, in order to compare them with a control group composed of the offspring of non-epileptic women, previously described in the literature, and to correlate the facial anomalies with the specific drug.

Full-term eutrophic newborns (37–42 weeks gestation and weight according to gestational age) of epileptic mothers who attended the Epilepsy Clinic of the National Institute of Perinatology, a third-level gyneco-obstetric center in Mexico City from September 1, 1993 to November 1, 1996 were included. These mothers were enrolled in a follow-up program from 18 weeks of gestation. All mothers have Mexican parents and grandparents.

All newborns in the intensive care unit, and those with congenital malformations with a different specific recognizable etiology were excluded.

The neonates were evaluated by the principal author, Díaz-Romero, between the first 12 and 48 h of life. The physician used a metallic calliper (vernier) calibrated in millimeters (mm), and a glazed fiberglass tape for the head circumference, with a variation coefficient of less than 0.5 mm.

With the newborns asleep, the following three facial points were marked with a dermic pencil: nasion (at the base of the nose in the frontonasal suture); gnathion (at the mandibular symphysis), and tragus (the medial point in front of the auditive conduct). The measurements were performed twice and collected by a second observer; the final value was the average of both measurements. All measurements were correlated with sex and with gestational age, estimated by the last menstrual period date and corroborated by the Capurro scale.

The values obtained were compared with those of a control group of offspring of mothers without epilepsy; for this purpose, an interval where 95% of the central values were considered as common values (± 2 standard deviations of the average) was defined and the remaining 5% of the uncommon values were considered to be in an alert zone. The uncommon measurements were correlated in each patient with the drug used during pregnancy. In those measurements where results were affected by more than 0.25, the odds ratio with the Wolf modification was used. These were then compared with the measurements of the newborns of non-epileptic mothers by means of Fisher’s test.

Seventy-two full-term eutrophic newborns were studied, classified according to the number of drugs exposed in the following: the polytherapy group in four cases (2 exposed to CBZ-VPA, and 2 to CBZ-PHT); the monotherapy group in 60 babies exposed to only one antiepileptic drugs (AED) (CBZ-26, PHT-21, VPA-10, Pb-two, and clonazepam-one), and a group of eight offspring of epileptic women without seizures during pregnancy and without exposure to any drug.

During the individual analyses, 70 cases (proportion = 0.97) were found with at least one measurement in the alert zone (nasion-tragus in 47 cases, bizygomatic in 32, and nasion-internalcanthal distance in 33); the mode and average of the uncommon values were 3, with a predominance of the mid-line area. All measurements were found in the common values range in only two cases.

No differences were found between groups of neonates who received AED in the form of monotherapy (VPA vs. PHT, VPA vs. CBZ, PHT vs. CBZ, etc.), and had one or a combination of two to four uncommon anthropometric measurements (Fisher’s test, p = non-significant). Eighteen of the 70 cases had uncommon values in 5 or more anthropometric measurements; 12 were exposed to CBZ, 3 to PHT, and 3 to VPA (p = non-significant). Additionally, no case was found with five or more measurements affected in the group without exposure to drugs.

Of the 22 measurements studied, 20 were located at the ± third SD of the distribution curve of frequencies at least once. In the head circumference and filtrum length measurements, there were no values at the second or third SD.

The five anthropometric measurements affected in the larger number of patients were analyzed separately.

1. Nasion-tragus distance (measurement 18): Number of neonates with values at the alert zone = 45 (proportion = 0.65). The more affected were those exposed to CBZ: odds ratio = 4.52 (CI 95% = 0.84–24) in 19 out of 26, and PHT: odds ratio = 4.16 (CI 95% = 0.74–23) in 15 out of 26 exposed.

2. Bizygomatic distance (measurement 13): Number of affected neonates = 34 (proportion = 0.44). This measurement was the only value to the left of the curve in the alert zone in 6 out of 7 exposed to phenobarbital or clonazepam, odds ratio = 18 CI 95% (1.26–25.5), and in 4 out of 6 with monotherapy of AVP, odds ratio = 4.50 (CI 95% = 0.58–34.6).

3. Nasion internalcanthal distance (measurement 5): Number of affected neonates = 33 (proportion = 0.45). In this area, a higher measurement was observed in the neonates exposed to CBZ, 13 out of 26, with an odds ratio of 3.00 (CI 95% = 0.50–17.70), and PHT in 13 out of 21 with an odds ratio of 4.87 (CI 95% 0.78–30.28).

4. Internal intercanthal distance (measurement 2): Number of affected neonates = 23 (proportion = 0.32). The drug with a higher number of cases in the alert zone up to the right side was CBZ in 13 out of 26 exposed neonates, with an odds ratio of 1.66 (CI 95% = 0.33–8.46). We considered this feature to be telecanthum. The same patterns were observed in two of the seven neonates exposed to phenobarbital, and in the baby exposed to clonazepam with an odds ratio of 0.066 (CI 95% = 0.075–5.87). The newborns with anomalies in the inner canthal distance also had higher nasion-canthalinternal and nasion-tragus distances.

5. Filtrum width (measurement 10): Number of affected neonates = 23 (proportion = 0.31). These measurements were higher in the 4 out of 7 exposed to phenobarbital and those exposed to clonazepam, with an odds ratio of 4.00 (CI 95% = 0.44–35.78).