ART and offspring anomalies: How concerned should we be?

Lack of a proper control group 

In determining whether ART places offspring at increased risk of anomalies, our pivotal difficulty is a lack of proper control groups for comparisons.1, 2, 3 Couples undergoing ART are not representative of the general population, yet we typically compare the two. This apples and oranges comparison is not appropriate. ART couples differ at the outset by virtue of their infertility and the underlying problem or problems responsible for it. Either or both individuals may have systemic abnormalities that place their offspring at increased risk for complications, and this might have little to do with ART technology per se. The true control group for ART couples would consist of infertile couples who do not undergo ART, but this is obviously not practical.

ART couples differ from the general population in other ways as well, and these factors may or may not be amenable to adjustment. Women undergoing ART are generally older than women conceiving naturally (27–28 years). Advanced maternal age is associated with increased rates of pregnancy loss and numerical chromosomal abnormalities (aneuploidy). Advanced paternal age is associated with de novo Mendelian mutations. Independent of age, some ART couples have experienced higher than expected rates of pregnancy loss prior to undergoing ART.

Other confounding variables include maternal disease and maternal exposure to toxins. Direction of bias is not always clear, however, and may even favor ART patients, in that women undergoing ART may be exposed to fewer toxins, especially alcohol and cigarettes.

Differences in evaluating and reporting 

When ART is used, surveillance begins during the cycle of conception and is more rigorous throughout the gestation than in naturally conceived pregnancies. This may partially explain the elevated rates of pregnancy loss among ART couples compared with those in the population at large. The ideal way to assess these rates, and thus attain accuracy, is through life-table analysis, taking into account the differing number of weeks of observation.

An analogous problem occurs when birth defect data for ART offspring are accumulated over a protracted time interval after birth. The cumulative rate thus obtained cannot validly be compared to that in birth defect registries, which are typically based on ascertainment only during hospitalization and the 1 to 3 days following birth. These registries form the basis of the traditional 2% to 3% anomaly rate cited for the general population. However, perhaps one third of anomalies evidenced at one year of life are not recognized at birth, and are therefore not represented in the 2% to 3% figure.

Also playing a role are clinical biases and inconsistent criteria for classifying anomalies. It is likely that the examination of neonates for malformations is more rigorous in ART offspring than in the offspring of naturally conceived pregnancies. If so, anomaly rates associated with ART would show a spurious increase. Compounding this further is the fact that even minor anomalies have been recorded in some ART registries, while minor anomalies are generally not recorded in the birth defect registries. In addition, ART registries have included internal anomalies detected solely on the basis of ultrasound; again, the general surveillance birth defect registries do not consider or factor in this information.

Pregnancies not requiring ICSI 

Notwithstanding the difficulty of comparing ART couples to others, and the fact that anomaly rates are subject to overestimation, there are plausible reasons why ART offspring could be at increased risk for major anomalies. Chief among them are the following:

Because of these recognized risks, surveillance has long been pursued as a means of monitoring for ART anomalies, although it has not always been a priority. So, what do the descriptive studies, registries, and population databases tell us about the safety of ART?

In the 1980s, the National Institute of Child Health and Human Development (NICHD) conducted a case-control study of 83 IVF children and 93 matched non-IVF children whose parents sought treatment at the Jones Institute for Reproductive Medicine.4 The rate of major malformations was low in both groups (2 IVF, 1 non-IVF) and not significantly different. Throughout the 1990s, registries of the American Society for Reproductive Medicine (ASRM), the Society for Assisted Reproductive Technology (SART), and the Centers for Disease Control and Prevention (CDC) also reported low rates of anomalies.1, 2, 3 However, claims to this effect should be tempered, because the US registry is now recognized as never having been intended for assessing birth defects.

Registry data from other countries, which generally have been more systematic, also tend to show no increase in anomalies. In the United Kingdom, Beral and Doyle evaluated the outcomes of 1,267 pregnancies resulting from IVF or gamete intrafallopian tube (GIFT) transfer.5 This 1978–1987 data set encompassed all UK clinics registering with a voluntary licensing authority. The finding: One or more congenital anomalies in the first week of life in 35 of 1,581 infants (2.2%), an incidence of malformations comparable to that in the general UK population.

A more granular analysis was conducted at the UK's Bourne-Hallam clinics. Among 961 IVF infants born at the clinics between 1978 and 1987, the malformation rate was 2.7% in multiple births and 2.4% in singleton births, again within the range expected for the general UK population.6

In France, statistics have been compiled by the French National IVF registry. Their 1993 report of data from 1986 to1990 revealed anomalies in 2% of IVF infants.7 A follow-up study of 375 children aged 6 to 13 years revealed height, weight, and scholastic performance levels that were no different than expected for the population at large.8 This trend has been consistently observed once correction is made for birth weight and multiple gestation.

Data from Australia have tended to raise more concerns than those from the United States and Europe. In one early report, the frequency of major congenital anomalies was 2.2% (cohort size, 371,697); however, 6 of the 37 infants with malformations had spina bifida.9 Later data from Australia continued to show an increase in neural tube defects, even when there was no overall increase in anomalies in the same cohort.10 More recently, Hansen and colleagues11 reported a 2-fold higher risk of major malformations in 837 IVF offspring, compared with 4,000 comparison children.

A rigorous cohort study may provide the most solid answer to our questions about ART's risks, and designing such a study is not theoretically difficult.12, 13 However, the logistics of doing so are daunting and the costs nearly prohibitive. Until then, the best data sets to consider are probably population-based registries in Scandinavia. In several such countries, independent and well-established registries are used to record traditional vital statistics and data on birth defects, cancer, and now ART. Importantly, these data sets can be linked.

One study using population-based data sets from Sweden compared anomaly rates in the general population of 1,690,577 births to that of 9,111 IVF cases not requiring ICSI.14 The overall anomaly rate was no different in the groups once adjustments were made (odds ratio 0.89).

However, the same study reported a 3-fold increase in neural tube defects among the IVF infants compared with the control population; alimentary atresia and omphalocele were also observed at 3 times the frequency. In a similar study conducted in Finland, Koivurova et al15 reported a 4-fold increase in cardiac malformations among IVF offspring, although no overall increase was seen in the rates of other anoma1ies.

Pregnancies requiring ICSI 

When ICSI is used to address oligospermia and azoospermia, there are additional reasons for concern about anomalies in offspring.3 Foremost is that oligospermia and azoospermia may reflect sex chromosome aneuploidy or balanced translocations. Among couples undergoing ICSI, 2.0% of males and 1.4% of females had a translocation.16 If the male partner was not azoospermic but oligospermic, 4.2% of the female partners had a translocation. Such findings suggest that subfertility in both partners will result in clinical infertility, whereas this may not be manifested if only a single partner is subfertile.

In the Swedish IVF Registry, 47 of 1,139 ICSI infants (4%) showed a major anomaly.14 Hypospadias in particular was markedly higher among the infants conceived through ICSI than those conceived naturally, with the 7 detected cases producing a relative risk of 2.9. Other ICSI data sets have also reported increased cases of hypospadias, albeit at a low absolute rate.17 Increased hypospadias among ICSI offspring could reflect the transmission of mutant paternal genes, whose malfunction was responsible for the spermatogenic abnormalities that necessitated ICSI in the first place.

The most thorough surveillance of ICSI pregnancies has been conducted in Brussels by Bonduelle and coworkers,18, 19 who have been compiling data for more than a decade. In 2002, they reported a malformation rate of 3.4% following ICSI (n = 2,840) and 3.8% following standard IVF (n = 2,955). These rates are higher than those from traditional birth defect registries, but this probably reflects more rigorous ascertainment of anomalies by the researchers.

A recent multi-center cohort study by Bonduelle and colleagues compared the physical health of 5-year-old children conceived after ICSI (n = 540), standard IVF (n = 437), and through natural means (n = 538). This study included the Brussels unit and several other centers throughout Europe.20 Of the ICSI children, 4.2% had a major malformation. Compared with the control group (represented by the naturally conceived children), the odds ratio for a major malformation was 2.77 in the ICSI group and 1.80 in the group receiving IVF alone. The higher rate among ICSI children was accounted for in part by hypospadias, findings similar to those reported by Hansen et al11 and Wennerholm et al.17

Another concern in ICSI offspring is increased rate of chromosomal abnormalities. In the Brussels cohort, prenatal cytogenetic studies revealed an incidence of sex chromosome abnormalities among ICSI offspring of 0.8%, compared to the population incidence of 0.2%.19 The frequency of de novo autosomal structural aberrations (also 0.8%) was likewise found to be higher than expected. In addition, one observes increased balanced translocations in ICSI offspring, but this reflects transmission from a similarly heterozygous parent.

The ostensibly increased rate of de novo chromosomal abnormalities in ICSI offspring could have several possible explanations:

Beckwith-Wiedemann syndrome 

Two case control studies have shown an association between ART and BWS, an overgrowth disorder characterized by macrosomia, macroglossia, omphalocele, and embryonal cancer. DeBaun et al reviewed BWS registry cases, finding that 3 of 65 children with BWS (4.6%) had been conceived through ART.26 Molecular studies in 6 BWS children (all of them ART offspring) showed 5 of them to have unexpected and atypical maternal expression of the imprinted BWS gene. That overexpression occurs in an overgrowth disorder like BWS is intriguing because in sheep, cloned animals are larger than normal (LOS, large offspring syndrome). An association between ART and BWS was also reported in the UK by Maher et al.27 Of 149 BWS children, 6 (4%) had been the product of ART—3 of IVF alone and 3 of ICSI in addition to IVF.

More recent studies have eased concern over ART per se. Chang et al studied 12 BWS children who were the product of ART; no common implicating factor was identified with respect to ovulation stimulation or the type of culture medium used.28 Ludwig et al studied the imprinting disorder Angelman syndrome, and found that among 16 cases 4 were born not to ART but to subfertile couples.29 The 4 couples did not require ART; however, either their time to conception exceeded 2 years, or they required hormonal stimulation. This may well be another example of the perturbation responsible for the infertility predisposing offspring to disorders of imprinting.

Other possible associations 

The Danish IVF cohort study has also provided relevant data on imprinting diseases and IVF. Lidegaard et al cross-linked several hospital registries to search for disorders known or thought to be related to imprinting, including BWS, Prader-Willi syndrome, Angelman syndrome, and, more speculatively, cancer, mental disease, and cerebral palsy.30 The researchers compared the frequency of these conditions in 442,349 children who were the result of non-IVF pregnancies and in 6,052 children conceived through IVF; the cohort included all singleton births reported between 1995 and 2002. Only for cerebral palsy was a group difference found. Compared with non-IVF children, IVF children were nearly twice as likely to have cerebral palsy (relative risk, 1.8). IVF was not associated with any of the other imprinting-related diseases.

A systematic review from Australia 

An Australian group reviewed publications through March 2003 that offered data on the prevalence of birth defects among IVF, ICSI, and naturally conceived infants.31 Twenty-five studies were suitable for inclusion, and of these, two thirds showed at least a 25% increased risk of birth defects in ART infants (odds ratio >1.25). The same group of researchers had earlier shown a 2-fold increase in anomalies among 837 IVF offspring.11

One may or may not feel that an odds ratio of this magnitude is disturbing. Once again, the results could be explained by the fact that the infertile population is biologically different from the general population. Moreover, the absolute rate of anomalies following ART would be low. Finally, the authors noted that their conclusions, which were based on aggregate data, sometimes differed from those of the individual studies. The discrepancy could reflect inadequate power in these individual investigations.

The Children's Health Panel report 

A US study likely to prove authoritative is that conducted by the ART Children's Health Panel, sponsored by the Genetics and Public Policy Center in Washington, DC (Kathy Hudson, PhD, Director). Consisting of experts in genetics, reproductive medicine, epidemiology, pediatrics, and public policy, this panel presented preliminary conclusions at the 2004 ASRM and the 2004 American Society of Human Genetics meetings.32 The group systematically reviewed studies in which ART offspring were compared to naturally conceived offspring, as well as studies in which IVF alone was compared to ICSI with IVF. The conclusions were collapsed into 3 standard categories: evidence sufficient, evidence insufficient, and evidence inadequate.

The panel's most reassuring statement was that “the evidence is suggestive of no association between ART and overall serious malformations.” In addition, they concluded that the evidence was “suggestive of no association between early childhood cancer and ICSI or IVF.”

The panel expressed its greatest concerns with respect to hypospadias and the imprinting disorders. For hypospadias, 9 cohorts had comparison groups sufficient for analysis; however, in only 3 of these cohorts were there more than 3 ART-associated hypospadias cases. Thus, the evidence was considered “inadequate to determine whether children conceived with ART have a slightly elevated risk of hypospadias.” The same conclusion was reached after the data were stratified for ICSI alone. Although these conclusions are less reassuring than ones of “no association,” there is some solace in the absence of a conclusion supporting a clear association.

For imprinting disorders also, decisive statements could not be made. The evidence was “suggestive but not sufficient to conclude that ART techniques may increase the frequency of rare genetic syndromes characterized by loss of imprinting.” Lastly, the panel suggested that even if ART slightly increased the rate of offspring with a rare disorder, there would be “low absolute rates and hence very few affected children.”