The Distribution of Congenital Anomalies Within the VACTERL Association Among Tumor Necrosis Factor Antagonist-exposed Pregnancies Is Similar to the General Population

MATERIALS AND METHODS

The EUROCAT central database, covering a large European population, contains individual standardized anonymous records of congenital anomalies since 1980 including live births, stillbirths, or fetal death from 20 weeks of gestational age, and terminations of pregnancy following prenatal diagnosis of congenital anomalies of any gestational age⁸. The EUROCAT antiepileptic study database is a validated part of the EUROCAT central database and can be regarded as representative with respect to the distribution of congenital anomalies in the general population based on several publications. For this reason this dataset was used. This dataset comprises 98,075 major congenital anomalies covering 3.8 million births. Medication use in this database is representative for the use of medication in pregnancy in the general population delivering a pregnancy outcome with congenital anomalies. We expect a low prevalence of use of TNF-α antagonists in pregnancy (less than 1/1000). The use of antiepileptic drugs noted in this database was 5.7/1000. That is very low and therefore it is expected that there is no influence on the general distribution of specific birth defects⁹.

We classified the anomalies listed in Table 1 of the Carter publication³ according to EUROCAT anomaly subgroups. Only 31 of the 41 described cases could be classified to the EUROCAT classification. Ten cases were excluded because they contained unspecified congenital anomalies and minor anomalies according to the EUROCAT classification and/or were infants and fetuses with a chromosomal anomaly or single gene disorder (n = 3).

A working definition for VACTERL association was set up, as there are several definitions for VACTERL association⁴. For correct comparison we used the method used by Carter, et al³ (one or more congenital anomalies that are part of VACTERL) as well as the agreed definition of at least 3 of the individual anomalies present⁴, as listed in Table 1.

Analysis was performed comparing the proportion of different anomaly subgroups as published by Carter, et al with proportions found in the EUROCAT database⁹, using the chi-square test to determine significance (p < 0.05).

RESULTS

The EUROCAT database contained 110 cases coded as VACTERL association, which might be an underestimation. Using the agreed VACTERL definition⁴, including at least 3 anomalies present, in the EUROCAT database, 619 (531 + 73 + 15, respectively) cases fulfilled these criteria (Table 2).

Counting all individual anomalies in the VACTERL association solely, according to Carter’s definition, in the EUROCAT dataset, more than 50% of all cases of congenital anomaly would have a VACTERL association.

The congenital anomalies identified by Carter, et al included 9 minor anomalies that are excluded in the EUROCAT classification of malformations. These excluded anomalies belonged to the EUROCAT subgroups eye, genital, musculoskeletal, congenital heart disease, and respiratory.

Most cases from the Carter study had only one or 2 major anomalies, so comparison of proportions is imprecise (Table 3). In this context, one significant observation was seen for limb anomalies (p < 0.05).

DISCUSSION

This study shows that the cases with congenital anomalies exposed to TNF-α antagonists as reported by Carter, et al have a distribution of specific congenital anomalies similar to those in a representation of the general population-based congenital anomaly database EUROCAT.

Based on currently known in-utero exposures to TNF-α antagonists, no specific embryopathy has been identified^10,11,12.

A limitation for comparison of results between the Carter study and the EUROCAT data is the lack specificity of the congenital anomalies by Carter, et al, and because of this several anomalies could not be reclassified in detail according to the EUROCAT anomaly subgroups. Because cases with only minor anomalies are not included in the EUROCAT database, these cases in the publication of Carter, et al were excluded in our analysis.

Another possible limitation of our study is the difference between the FDA database and our database. The FDA AERS database is a spontaneous adverse reactions system and the EUROCAT database is a population-based congenital anomaly database.

In studies on VACTERL associations, a definition of the association should be included; for instance, the classic VATER association described by Quan, et al¹³ does not include the cardiac and limb associations described by Carter, et al. Källén, et al⁴ verified the original VATER association but not the extensions such as VACTERL.

Concluding protection by TNF-α antagonists for the congenital anomaly limb would be premature, because of the low number of cases (n = 2) in the population described by Carter, et al.

Our study was not designed to assess whether there is a generalized increased risk of congenital anomalies with use of TNF-α antagonists. In a postmarketing setting, 2 main approaches have been used to identify teratogenic effects of medicines, followup studies and case-control surveillance¹⁴. Carter, et al suggested a large followup study of pregnancies exposed to TNF-α antagonists, especially etanercept, to further assess a possible increase in congenital anomalies. Indeed, for this purpose, large pregnancy registries can be useful; however, a case-control study might be more appropriate to identify such a very rare association¹⁴.

The seemingly high number of congenital anomalies belonging to the VACTERL spectrum within pregnancies exposed to TNF-α antagonists described by Carter, et al could not be confirmed. This neither reinforces nor excludes the possibility of an increased absolute risk associated with TNF-α antagonists. Evaluation of risks of specific major congenital malformations associated with use of TNF-α antagonists can be performed best in a case-control study database designed to test signals of congenital anomalies¹⁴.