Background
More than six million children have been born worldwide since 1978 using assisted reproductive technologies (ART) [
1]. While the majority of these appear to be healthy, a higher incidence of some congenital abnormalities, pre-term birth and low birth weight (LBW) for gestational age has been noted among ART singletons [
2‐
5]. The latter is of increasing concern, since birth weight is a surrogate for fetal growth and a strong predictor of cardiometabolic disease (CMD) risk across the life-course [
6] (the Developmental Origins of Health and Disease (DOHaD);
https://dohadsoc.org/)).
Studies in both human populations [
7] and animal models [
8,
9] have clearly identified the periconceptional period immediately before and after fertilisation as a unique developmental window during which the long-term health of the individual may be programmed. This has raised significant concerns about ART children as it coincides with the period during which gamete and embryo manipulations take place in the non-physiological ART environment, with specific embryo manipulations associated with altered fetal growth and birth weight [
9].
Postnatal infant and child growth is arguably an even more important indicator of long-term disease risk; in particular, catch-up growth from LBW increases the risk of obesity and CMD [
10,
11]. Owing to the lack of routinely collected data, only a limited number of small (≤ 500 children) studies have been carried out on ART child growth and these generally show conflicting results. In three small European cohort studies, children born from intra-cytoplasmic sperm injection (ICSI) were found to be lighter than their target weight (for age and height) at age 3 years, but not at age 5 years [
12]. A longitudinal study [
13] noted that ART children had lower weight at birth and 3 months, with similar growth rates to age 3, compared with naturally conceived (NC) subfertile controls. Given the implications of altered early life growth for health risks later in life, it is clearly important to collect definitive data from larger groups of ART children. It is also far from clear which specific ART factors might be causing changes in birth weight, although extending the length of time embryos are kept in culture before (blastocyst) transfer, the type of culture medium and the use of frozen embryo transfer (FET) have been implicated [
5,
14‐
17]. Without this essential information on modifiable risk factors, it will be impossible to improve the ART process and reduce long-term risk to children.
The UK pioneered ART 40 years ago with the first successful birth in Oldham in July 1978 and thus is home to the first cohort of children now approaching middle age. However, the UK has been among the slowest countries to conduct long-term ART follow-up studies, because the original Human Fertilisation and Embryology Act (1990) did not envisage this. In line with shifts in public opinion, the Act was changed in 2010 and under the principle of presumed consent, the UK now possesses the world’s oldest and largest database of ART available for research, with approximately 110,000 children from 1991 to 2009. However, so far, this has only been exploited for a single cancer follow-up study [
18]. There have been no large studies of ART child growth or causal factors in the UK or anywhere in the world. Therefore, the primary aim of this work was to assess the role of ART treatment factors on growth from birth to age of school entry, in a large UK national study. A secondary aim was to provide comparative data for NC babies.
Discussion
Birth weight and early child growth are important predictors of long-term CMD risk [
6]. Although it is well established that ART children are at increased risk of low birth weight, there is little understanding of the causal factors in the ART process and there have been no large studies of postnatal growth in these children. Using the UK ART data register held by the HFEA, linked to maternity and child health data, we conducted a large-scale analysis of more than 5000 singleton ART children. Our key findings are that not only do ART babies born from fresh embryo transfers have lower weights, head circumference and length at birth than NC children, they grow more quickly and catch up by school age. Babies from FET cycles, by contrast, have greater weight, head circumference and length at birth and show similar growth to NC children.
The major strength of this study is the registry-based national cohort design. The loss to follow-up is a concern, although we believe most of this loss is administrative and we could detect no strong biases in those with and without data. The data on NC infants did not include parity and, therefore, we were not able to include this in the birth weight adjustment. It is likely that more NC babies were born to parous women than ART-conceived babies, which would lead to a small, but unquantifiable, over-estimation of birth weight in this group, in turn reducing the effect in ART-Fresh cycles while increasing the effect in FET cycles. Within-ART comparisons were parity-adjusted. Postnatal growth estimates for both ART-conceived and NC children are based on changes and are therefore largely unaffected by parity.
Newborn infants conceived by fresh embryo transfer were smaller than NC infants, with significant differences in birth weight, length and OFC. Some infants were small because they were born early, at gestational ages of < 37 or < 32 weeks, and there were more admissions to the NICU, probably due to conditions associated with prematurity. Other infants were small for gestational age (SGA), which may be due to a slowing down of intrauterine growth, implicating a reduction in placental transfer of nutrients. FET infants by comparison had a higher birth weight, length and OFC than NC infants. The birth weight findings are consistent with previous studies [
2,
4,
22], and the weight differences are remarkably consistent across studies, e.g. in a recent prospective RCT in China, FET babies were 161 g larger than fresh ART babies [
16] and in a large Japanese registry study, FET babies were 91 g larger than fresh ART, and 40 g larger than NC babies [
23], very similar findings to ours.
At 6 to 8 weeks of age, the weight, length and OFC of the FET infants and NC infants were similar, while weight, length and OFC in the fresh ART-conceived infants remained smaller. At school entry, weight, length and BMI in the three study groups were similar, showing that low birth weight fresh ART infants, when free from the constraints on intrauterine growth, resume their growth trajectory postnatally and attempt to achieve their genetic potential by catch-up growth. OFC measurements were not available; however, brain growth and somatic growth are likely to be in harmony. The BMI values at school entry were consistent with those reported previously at a similar age.
Previous studies of ART child growth have been small (≤ 500 children) and unable to adequately control for confounding factors, and show conflicting results [
12]. In three small European cohort studies, children born from intra-cytoplasmic sperm injection (ICSI) were found to be lighter than their target weight (for age and height) at age 3 years, but not at age 5 years [
12]. In a longitudinal study of approximately 200 ART children, they had lower weight at birth and 3 months, with no difference in early growth rate (velocity) compared to a subfertile NC control group; however, ART children then showed greater growth velocity from 3 months to 12 months, followed by similar childhood growth rates to age 3 [
13]. Small studies of IVF and ICSI children measured at birth and up to age 18 months in Taiwan [
24], 3 years in the USA [
25], 4 years in the Netherlands [
26] and 10–12 years in the UK [
27] showed few differences in growth [
12]. Data from a UK child screening programme [
28] showed that ART children were less likely to be overweight at 5 years old than controls, with significantly lower BMIs at 7 years of age than controls consisting of births after normal conception or to subfertile couples. Green et al. [
29] controlled for embryo freezing as a confounding factor, and showed that fresh transfer IVF children, while lower birthweight than FET or NC children, did not show catch up weight growth in childhood; however, ART conceived girls were taller than matched controls after fresh IVF and also FET [
29,
30]. Our data extend this current literature significantly by virtue of the large sample size and careful adjustment for confounding factors, by use of accurate growth data recorded by maternity units including length and OFC in addition to birth weight and, most importantly, by following growth of children to age of school entry.
It is currently unknown what alters fetal and child growth trajectories for children conceived through ART; however, it is noteworthy that the magnitude of impact of ART on birth weight is nearly as great as that of maternal smoking in pregnancy. Catch-up or catch-down growth (centile crossing) after birth are both significant predictors of long-term disease risk. Most LBW children show catch-up growth in the first few months, with accelerated growth associated with increased risk of obesity and CMD risk [
11,
31,
32]. Our study shows that fresh ART babies grow faster, from birth, over the first few weeks, than their NC counterparts with further catch-up growth then seen in fresh ART children by school entry age. This is associated, in NC children, with increased risk of obesity and CMD in adulthood [
10‐
12,
33,
34] and, in ART children, with greater weight gain during childhood, associated with increased cardiovascular risks [
13]. Reassuringly, however, catch-up was not seen in the smallest fresh ART birth weight centiles in our study, whereas the increased rate of catch-down growth observed was strongest for the largest birth weight centile babies.
Our major finding that children born from FET cycles differ significantly in birth weight and child growth from their fresh ART counterparts, strongly implicates some aspect of the ART treatment process itself rather than parental subfertility, as does the animal literature showing that adverse outcomes can be phenocopied in fertile mice subjected to ART [
9]. However, the mechanisms involved are unknown. One plausible hypothesis is that the uterine environment is hormonally dysregulated following ovarian hyper-stimulation, such that in fresh cycles fetal growth is restricted due to impaired placental function [
5,
35]. Alternatively, the ART environment may act via a direct epigenetic effect on the embryo; however, this would predict that the impaired fetal growth trajectory would be continued into postnatal life, and so is not supported by our data. Fetal growth in FET cycles may also be affected by some aspect of the embryo cryopreservation process, consistent with our previous finding that frozen/thawed embryos show altered gene expression [
5] and our finding (and from others [
23]) that FET babies have a consistently slightly higher birth weight compared to NC babies.
Other than the effects of embryo cryopreservation/FET, we failed to detect any significant associations between newborn or child weight, length and OFC and other ART patient or treatment factors available. However, the HFEA database lacks information on factors such as the type of culture medium embryos were grown in, an important omission in light of recent evidence that culture medium is associated with altered embryo gene expression [
15], epigenetics [
9] and fetal and postnatal growth following ART [
15], resulting in calls for action from professional societies and leading scientific journals [
5].
Acknowledgements
The authors gratefully acknowledge the considerable and vital roles of both the Human Fertilisation and Embryology Authority and NHS National Services Scotland in the conduct of this research. In particular, David Moysen, Nick Jones, Howard Ryan, Suzanne Hodgson and Deborah Bloor from the HFEA, who maintain the Assisted Reproductive Technology Register and facilitated its secure transfer to the National Safe Haven for Scotland and Jim Chalmers, Dave Clark, David Bailey, John Nolan, Carole Morris, Andy Duffy and Doug Kidd from NHS NSS who provided the Safe Haven as a platform for data storage and analysis, conducted the data linkage, monitored study outputs and replied to numerous and varied queries relating to all aspects of study conduct and logistics.