To assess the association between ART and anorectal malformations, we compared data from the multicenter population-based CURE-Net with nationwide data from the German IVF register and the Federal Statistical Office. We observed a strongly and significantly increased risk of ARM after both IVF and ICSI. Further, separate analyses of patients with isolated ARM and those with associated anomalies as well as with a VATER/VACTERL association showed strongly increased risks in each group. Furthermore, a strongly increased risk of ARM was seen among both singletons and multiple births.
Our finding of an increased risk for ARM is in agreement with three previous epidemiological studies from Italy and Sweden [13–15]. All these studies observed a significantly increased risk for ARM in general (Ericson and Källén : relative risk [RR], 3.1; 95% CI, 1.3–6.1; Källén et al. : OR, 4.7; 95% CI, 3.2–6.9; Midrio et al. : OR, 13.31; 95% CI, 4.0–39.6). However, Källén et al.  could not confirm their results with other cases in a second study two years later (OR, 0.90; 95% CI, 0.37–2.18), but analysis of the first and second data set together showed again a highly significant association between ART and ARM (OR, 3.11; 95% CI, 2.21–4.52; P < 0.001). Heterogeneity across both data sets was significant (χ2 = 43.5). One American study by Reefhuis et al.  stratified ARM infants by plurality of birth and reported an increased risk of ARM after ART for singletons (OR, 3.4; 95% CI, 1.2–8.3), but not for multiple births (OR, 1.3; 95% CI, 0.5–3.3). Three of the five studies mentioned above have identified their cases through the Swedish Medical Birth Register [13, 14, 28]. The remaining two studies obtained their information from the pediatric surgery unit of the University of Padua  and existing birth defects surveillance systems in the United States of America . Studies were heterogeneous with respect to case numbers and control types. Case numbers ranged from 28 ARM cases  to 533 ARM cases . Control groups were register-based [13–15, 28] or from the same source populations as the case infants . Control groups consisting of nationwide data were only available in the Swedish studies [13, 14, 28]. Period of data acquisition ranged from six years [10, 15, 28] to 19 years . Our risk estimates were mostly higher than those of previous studies, but in the light of the overlapping confidence intervals it is unclear to what extent these differences might reflect chance variation.
The large sample size of the present study allowed differentiation between isolated ARM and ARM with associated defects which is common in approximately 64% of all ARM patients . This mainly concerns anomalies of the kidney (31%), heart (24%) and vertebra (20%). To our knowledge, this study is the first study specifically addressing risk of isolated ARM, ARM with multiple anomalies and the VATER/VACTERL association. Similarly strong and significant associations were found for all three subgroups.
The previously published studies assessed the association between ARM and ART use (IVF and ICSI) in general , or between ARM and IVF [13–15, 28]. However, to show possible differences between IVF and ICSI procedures we analysed data separately. There was a highly increased risk of ARM after both ART procedures. Normally, ICSI is used when previous IVF procedures were unsuccessful or in case of severe male infertility . Although the studies by Bonduelle et al.  and Lie et al.  did not indicate an increased risk for malformed children born after ICSI compared to IVF, this procedure is often considered to be risky because it is more invasive than routine IVF .
Compared to other congenital malformations, ARM is not likely to be specifically associated with ART. For example, a higher risk is also reported for esophageal atresia, hypospadias (second or third degree), cleft palate, septal heart defects and the exstrophy-epispadias complex [10–12]. Our finding of an increased risk for ARM associated with ART may be due to ART per se and/or the infertility problems of couples who conceive following ART [34–37]. Regarding the inheritance in ARM families, the study by Falcone et al.  reported a positive family history in 1.4% of patients with ARM. In our data set, only one father and three siblings were affected with ARM, also resulting in a proportion of 1.4%. The respective four cases were all naturally conceived. In our cohort, there was no indication that families with major defects conceived via ART more often than families without major defects. However, due to missing data for our control group, it was not possible to examine clustering in more detail to confirm potential genetic pathways. Blastogenesis defects, arising in the first four weeks of pregnancy, are also conceivable as reason for the higher risk of ARM . In addition, it is known that maternal overweight, obesity and diabetes, identified as risk factors for ARM in the systematic review and meta-analysis by Zwink et al. , occur more often in women with polycystic ovary syndrome [40–42]. This hormonal disorder is also the most common cause of infertility due to menstrual disorders.
Plurality of birth is both more common after ART and strongly related to ARM. However, our data clearly indicate that the strong relation between ART and ARM was not explained by plurality of birth. Strong associations between ART and ARM were seen among both singletons and multiple births. Compared to previous studies, our twinning rate of all ARM cases is slightly higher [43, 44]. Due to lack of pertinent information in the control group, it was not possible to analyse the risk of monozygotic and dizygotic twins separately. The study by Harris et al.  assumed that the increased risk for twins seems to be mainly associated with monozygotic twinning.
Our study has specific strengths and limitations. Strengths include the large sample size of both ARM cases and controls. With 295 ARM cases it was possible to differentiate between ART types as well as between isolated ARM and ARM with multiple anomalies. The nationwide control group included more than 10 million births between 1997 and 2010. As data of 2011 will be only available at the end of 2012 it was not possible to include them in the analysis. Since reporting of ART procedures in Germany is mandatory by law, the registration of ART in Germany can be considered complete. To avoid any delay in reporting, we chose the updated numbers of ART published in the respective subsequent annual report. In contrast to the worldwide increase [1, 2], proportions remained nearly the same over the years in Germany. However, there was a decrease in the performed procedures in 2004 because of changes in the German health system. These comprised less insurance eligibility of treatment of ART so that couples must pay half of the arising costs by themselves. In addition, services of the German health insurance company are dependent on the age of couples and available for only three treatment cycles. Different than in other countries, the embryo transfer in Germany is limited to a maximum of three embryos per treatment cycle.
Despite major efforts of nationwide recruitment of ARM cases, recruitment cannot be considered complete. In particular, active nationwide registration started in only 2009 and earlier cases could only be identified and recruited retrospectively in cooperation with hospitals and the self-help organisation SoMA. However, major selection bias from incomplete recruitment of cases could have occurred only if recruitment would have varied by application of ART for which we do not see obvious reasons. A selection bias might also have occurred in case of differences in prenatal care with potential impact on pregnancy termination between ART and natural conceptions. As all parents reported on regular prenatal control examinations an impact on the results seems unlikely. In addition, the average number of ultrasound examinations during pregnancy did not differ relevantly by history of ART in our data set (ARM patients born with ART: 15 examinations; ARM patients born without ART: 12 examinations). Nationwide data on termination of pregnancy and death of ARM patients during the first days of life are not yet available. Therefore, analyses were limited to live-births only. Possible differences in the registration of ART due to different collecting tools used in the CURE-Net register and the IVF register are also unlikely.
Retrospective collection of data could also have compromised the validity of ART information among cases. However, IVF and ICSI are very special procedures that are likely to be recalled correctly. Furthermore, a subgroup analysis with prospectively collected ARM cases born between 2009 and 2011 confirmed our findings. As we have no information about the nationwide proportion of ARM patients born with and without ART an overlap in ARM patients can not be excluded. However, since ARM is a rare malformation with an incidence of 1 in 2,500 to 1 in 3,500 live-births [17–19] an influence on the results can be neglected. Exclusion of the ARM cases from national data did not change the result.
Finally, use of controls from a nationwide registry hindered comprehensive control for potential confounding factors. We could control, however, for plurality of birth, a key risk factor of ARM, and found a strong risk increase to persist in both singletons and multiple births. It seems unlikely that this persistent strong risk could be explained by confounding. Nevertheless, our data do not allow differentiating the potential impact of ART itself or of their possible indication, such as factors associated with maternal or paternal infertility, or characteristics of the parents, such as maternal age [37, 45], or of preceding treatments, such as hormonal therapies.
Despite its limitations, our study provides evidence for a strong association between ART and ARM. Further research should aim to elucidate the underlying mechanisms and to identify possible preventive actions to limit the frequent occurrence of ARM after application of these techniques.