Background
Assisted reproductive techniques (ART) are methods used to achieve pregnancy in case of female and/or male infertility. ART are known to be associated with a higher risk of multiple pregnancies [
1,
2], mostly due to superovulation or transfer of two or more eggs or embryos. An overall higher risk of congenital malformations in fetuses conceived following ART has been reported in several studies [
3‐
10] using different study populations and designs and methods of analysis and at times inconsistent results, particularly for specific anomalies [
6].
Few studies have assessed specifically the risk of congenital heart defects (CHD) associated with ART [
6,
11‐
15]. These studies have found an increased risk of 30-50% (Odds ratios ~1.3-1.5) of the overall risk of CHD, which varied across categories of CHD and methods of ART [
12‐
15]. In particular, a recent study assessing the risk for four individual CHD, (Transposition of Great Arteries, Hypoplastic Left Heart, Coarctation of Aorta and Tetralogy of Fallot (TOF)) [
15] found a specifically higher risk of TOF associated with ART. In contrast, the authors found no statistically significant associations with the other three defects examined and the effect sizes (odds ratios) were close to the null value (i.e., close to one).
Even if multiple pregnancies may be independently associated with a higher risk of congenital anomalies [
16,
17] and this may be specifically the case for CHD, relatively little information exist on the specific association between multiple pregnancies and CHD [
18‐
20], particularly for individual defects [
21].
Given: i) the previous finding of the specific association between ART and TOF [
15], ii) the known association between ART and multiple pregnancies, and iii) the possible association of the latter with the risk of congenital anomalies, the objective of the present study was to assess the impact of multiple pregnancies in the association between TOF and ART using a path-analysis model.
Discussion
Using population-based data on approximately 400 cases of tetralogy of Fallot (TOF) and 4000 malformed controls, we assessed the previously reported higher risk of TOF [
15] that may be due (mediated by) multiple pregnancies in fetuses conceived following assisted reproductive technologies (ART). We used a path-analysis model based on a counterfactual approach in order to decompose the total effect associated with ART into a direct and an indirect effect (i.e. that due to the association between ART and multiple pregnancies).
Our results showed that of the overall 2.6-fold higher odds of TOF in fetuses conceived following ART, 20% was due to the higher likelihood of multiple pregnancies in fetuses conceived following ART (the “indirect effect”). This suggests in turn that multiple pregnancies themselves may be associated with a higher risk of TOF, albeit much less so than ART. The contribution of multiple pregnancies to the higher odds of TOF associated with ART (i.e. the indirect effect) varied according to the method of ART. Our estimates suggested that 30% of the total effect of IVF only was due to multiple pregnancies vs. 11% in the case of IVF + ICSI. The results were similar for analyses restricted to isolated cases of TOF (excluding those associated with chromosomal or other anomalies).
In order to estimate the risk of congenital malformations in multiple pregnancies conceived following ART, previous studies have generally included separate (stratified) analyses for singletons and multiple pregnancies [
8,
9,
12,
13,
15,
32]. This approach can be informative in examining the association between ART and the risk of congenital malformations, in particular regarding the possible existence of an interaction effect between ART and multiple pregnancies. However, these studies did not formally test for statistical significance of any such interaction and had limited power to detect them if they indeed existed. In any case, this approach does not allow a formal estimation of the proportion of the effect of ART on the risk of congenital malformations that may be due to multiple pregnancies.
We found that the “direct” effect associated with ART (i.e. that not mediated by multiple pregnancies) was substantially greater than the indirect effect. This was particularly so in the case of IVF + ICSI. This finding may provide clues as to the underlying mechanisms of the association between ART, particularly for IVF + ICSI, and TOF. The developmental basis of TOF is complex, incompletely elucidated and probably multifactorial [
33,
34]. The implication of genomic imprinting [
35‐
38] and a role for cardiac neural crest cells in the development of TOF have been suggested [
34,
39] but any implication of ART in these pathways remains to be shown.
Regarding any effects (the indirect effect) due to twinning, our study was not designed to elucidate what may be the underlying mechanism of any association between twinning and risk of TOF (independently of exposure to ART). In particular, we did not take into account zygosity (or chorionicity), whereas the risk of congenital anomalies in monozygotic twins appears to be higher than that of dizygotic twins [
16,
17,
20]. Our results suggest however, that any effects due to twinning
per se, whatever its underlying mechanism may be, are likely to be small. Hence, it is possible that twin pregnancies, even in the case of monozygotic twins, are not (or only to a small extent are) associated with a higher risk of TOF, whereas for other CHD twinning
per se may be a more important risk factor [
16,
17,
20].
We also did not account for any potential effect of the “vanishing” twin syndrome on our estimates. Several studies have found that ART multiple pregnancies in which a twin vanished had poorer obstetrical and perinatal outcomes [
40‐
43]. However, to our knowledge, there is no documented association between the vanishing twin syndrome and risk of birth defects in general or of CHD in particular. If indeed such an association exists in the case of TOF, misclassification of vanishing twin pregnancies as singletons would result in an underestimation of the mediating effect of multiple pregnancies in the association between ART and risk of TOF.
Our study has certain limitations. Even if the total number of cases of TOF and controls was fairly large, the numbers of cases exposed to the different methods of ART were small and hence confidence interval were fairly wide reflecting the relative imprecision of our estimates of ORs particularly for specific techniques of ART.
Although we chose frequent and heterogeneous malformations for which no association with ART exist in the literature, bias due to the association between ART and malformations in the control group cannot be excluded. In particular, if ART were associated with a higher risk of one or more of the malformations in the control group, the consequence would be an underestimation of the true association between ART and TOF.
Due to a lack of available information in our data, we did not adjust for certain maternal conditions including pregestational or gestational diabetes and obesity, tobacco, alcohol and drug consumption. Although these factors may be associated with risk of birth defects in general, or of CHD in particular, their associations, if any, with the risk of specific CHD, including TOF are generally not known.
We also could not adjust for maternal folic acid/multivitamins use that is associated with a lower risk of CHD [
44] and in addition is more frequent in women who conceive following ART [
13,
45]. Hence, we cannot rule out residual confounding due to risk (or protective) factors associated with ART and/or twinning that were not taken into account.
In addition, our study was not designed to and cannot disentangle the effects that may be due to infertility of couples vs. ART or twinning
per se[
4,
9,
46]. Some authors have found that the risk of birth defects is increased in infertile women who conceive spontaneously, which suggests that the underlying infertility problems may explain part or perhaps all of the association between ART and risk of birth defects. However, the strategy of “adjusting for infertility” (time to pregnancy or duration of infertility) has been criticized [
5] as it could be “synonymous” with exposure to ART. In any case, in our data, essentially all women who had a duration of infertility of more than two years had conceived following ART; adjustment for infertility duration was therefore not feasible in our study.
The frequency of missing data was low for exposure to ART and maternal characteristics. However, paternal age was missing for 25% of the study population. We adjusted for paternal age as it can be related to ART exposure and particularly ICSI. Results were nevertheless similar when models did not include paternal age and in analyses that included multiple imputation for paternal age (results not shown, available from authors).
Acknowledgements
Inserm Unit 953 has received a grant from the Bettencourt Foundation (Coups d’élan pour la Recherche française) in support of its research activities.
Funding
This work was supported by grants from the Agence de Biomédecine (Saint-Denis La Plaine, France) (to B.K.). The Paris Registry of Congenital Malformations received financial support from INSERM (Paris, France) and the Institut de Veille Sanitaire (Saint-Maurice, France). The EPICARD study was supported by three grants from the Ministry of Health (PHRC 2004, 2008 and 2011). Additional funding for the EPICARD study was provided by the AREMCAR Association (Association pour la Recherche et l’Etude des Maladies Cardiovasculaires).
EPICARD study group
Principal Investigators: François Goffinet, Babak Khoshnood
Steering Committee: Damien Bonnet (Hôpital Necker Enfants Malades, AP-HP, Centre de référence M3C, Université Paris Descartes, Paris)
Johanna Calderon (INSERM U953)
Drina Candilis (Université Paris-Diderot, Paris)
Anne-Lise Delezoide (Hôpital Robert Debré, AP-HP, Service de biologie du Développement, Université Paris-Diderot, Paris)
Catherine De Vigan (INSERM U953, Paris)
François Goffinet (Groupe Hospitalier Cochin-Hôtel Dieu, AP-HP, Maternité Port-Royal et INSERM U953, Université Paris Descartes, Paris)
Lucile Houyel (Hôpital Marie Lannelongue, Service de chirurgie des cardiopathies congénitales, Le Plessis-Robinson)
Jean-Marie Jouannic (Hôpital Trousseau, AP-HP, Centre pluridisciplinaire de diagnosticprénatal, UPMC, Paris)
Babak Khoshnood (INSERM U953, Paris)
Nathalie Lelong (INSERM U953, Paris)
Suzel Magnier (Hôpital Robert Debré, AP-HP, Service de cardiologie, Paris)
Jean-François Magny (Institut de Puériculture et de périnatologie, Service de néonatologie, Paris)
Caroline Rambaud (Hôpital Raymond Poincarré, AP-HP, Service d’anatomie et cytologie pathologiques – Médecine légale, UVSQ, Garches)
Dominique Salomon (INSERM U953, Paris)
Véronique Vodovar (INSERM U953, Paris)
Project Coordination and Data Analysis Committee: François Goffinet, Babak Khoshnood, Nathalie Lelong, Anne-Claire Thieulin, Thibaut Andrieu, Véronique Vodovar
Independent Data Monitoring Committee (URC Paris Centre et CIC Cochin Necker Mère Enfant): Maggy Chausson, Anissa Brinis, Laure Faure, Maryline Delattre, Jean-Marc Treluyer (Groupe Hospitalier Cochin-Hôtel Dieu, AP-HP, Université Paris Descartes, Paris)
External Scientific Committee: Gérard Bréart, Dominique Cabrol, Alain Sérraf, Daniel Sidi, Marcel Voyer
Participating Centers: The Greater Paris Area (Paris and its surrounding suburbs) public (AP-HP) and private maternity units, Departments of Pediatric Cardiology and Pediatric Cardiac Surgery, pediatric cardiologists in private practice, Neonatal Intensive Care Units, Pediatric Intensive Care Units, Emergency Transfer Services (SMUR), Departments of Pathology, Sudden Death Centers, Departments of Family and Infant Protection (DFPE)
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
BK conceived the study. KT conducted the main statistical analyses and wrote the first draft of the manuscript with BK. NL assisted with statistical analysis. LH, DB and FG contributed to the conceptualization of ideas and made suggestions about the required analyses. LH and DB provided expertise as pediatric cardiologists. All of the authors contributed to the interpretation of findings and revisions of the article. All authors read and approved the final manuscript.