Mark McCarthy and Rachel Freathy

Birth Weight

With colleagues in the EGG (Early Growth Genetics) Consortium, we have published a paper in Nature describing genetic analyses of birth weight in over 150,000 samples. In this blog, we detail the rationale behind the study, summarise the main findings, and place the research in a wider context.

The rising prevalence of diabetes, obesity and related conditions represents a major challenge for human health. For any given individual, the risk of developing these conditions is partially dependent on the profile of the genetic variants they inherited from their parents. It is also heavily influenced by a wide range of environmental factors, most obviously those related to dietary intake and physical activity during adulthood.

The impact of the environment on chronic disease risk may however extend to exposures far earlier in life. Around 25 years ago, it was noted that, in some historical cohorts, those who had been recorded as having low birth weight were at substantially increased risk of diabetes and coronary disease many decades later. This observation has been widely reproduced in a variety of data sets around the world. More recently, it has been shown that, in populations with particularly high prevalence of type 2 diabetes, the same appears to be true of individuals with birth weights at the higher end of the scale. In other words, the shape of the distribution between birth weight and later diabetes risk follows a “U” or “reverse J” shape: from the perspective of future disease risk, it is best to be near the average.

The dominant explanation for these observations has been provided by the Developmental Origins of Health and Disease (DOHaD) hypothesis which suggests that these relationships between early growth (for which birth weight is the most readily available measure) and later disease result from the long-term effects of exposing an individual to either too little (or too much) nutrition during early life. For babies exposed to poor intrauterine nutrition, for example, as a result of maternal deprivation or placental insufficiency, the consequences extend beyond poor fetal growth and low birth weight, to include a series of hardwired changes in the “set up” of the metabolic profile of the body that increase disease risk in adulthood. Studies in rodents provide confirmation of many aspects of this hypothesis, as have studies of the offspring of women exposed to extreme deprivation during pregnancy (typically as a result of conflict). Teleologically, this metabolic programming can be seen as an adaptation on behalf of the child for the expectation of an equally harsh postnatal environment, one which has adverse consequences when that postnatal environment turns out to be more luxurious than anticipated.

However, several of us felt that this might not be the only explanation. One other way to connect early growth and later disease could be via DNA variants that influence BOTH birth weight and later disease risk. Perhaps some individuals inherit DNA variant profiles that simultaneously put them at greater risk of reduced growth in early life and of diabetes and other cardiometabolic diseases in adulthood. This is also biologically plausible: the hormone insulin plays a critical role in both early growth and normal adult metabolism, and genetic differences that result in a reduction in insulin secretion would tend to push individuals towards both low birth weight and later diabetes. Thus, two temporally distinct phenotypes could be connected by the same genotype. There are examples of rare genetic conditions (most notably in the glucokinase gene) where such mechanisms are clearly in play.

In the present study, we set out to ask whether or not there was evidence that common, shared genetic differences that influence birth weight showed overlap with the genetic differences already known to be involved in influencing individual risk of diabetes, and other cardiovascular and metabolic conditions.

This study involved 164 scientists from 117 institutions, based in 17 countries on four continents, who have been working as part of the Early Growth Genetics (EGG) consortium. Together, we assembled data from over 150,000 individuals who had the combination of a recorded birth weight and detailed information on genetic variation across their genomes. Nearly half of those samples came from the UK Biobank, a massive study of middle-aged participants from the UK for which data has recently become available (see our other blog post).

We found around 60 regions of the genome that harboured genetic differences that were significantly associated with differences in birth weight. Together, variation at these 60 sites accounts for about 2% of variation in birth weight, similar to the impact of maternal smoking or obesity. In fact, these 60 regions are just the most visible “tip of the iceberg” in terms of the overall contribution of genetic variation to birth weight. Analysis of the full data set allows us to estimate that, in total, around one sixth of the variation in individual birth weight is attributable to genetics (approximately the same impact as an additional week of gestation at term).

Armed with these data, we next compared the genome-wide patterns of birth weight association, with those for late onset diseases such as diabetes, high blood pressure and coronary heart disease. We found substantial overlap between them. In other words, much of the same genetic variation that contributes to birth weight differences between people also contributes to differences in susceptibility to disease in later life. This is evidence that genetic variation does indeed contribute to the relationship between early growth and later disease risk. Indeed, in some more detailed (but still somewhat preliminary) analyses, we found that most of the relationship between abnormal early growth and later disease could be explained by shared genetic factors. We were able to show that these genetic factors were acting through a series of shared processes to influence early growth and later disease including those connected to metabolism, growth and development.

Crucially, we were able to start to look beyond the simple relationships between the child’s genetic profile, their early growth potential, and their risk of later disease. When considering events occurring before birth, one also needs to consider the essential contribution made by the mother to the growth of the fetus. Not only is the baby’s growth dependent on the baby’s own genetic profile, it is also heavily influenced by the gestational environment that the mother provides: that maternal environment will in turn be influenced by the mother’s genetic profile. And, since each baby inherits half of its genes from its mother, the maternal and offspring genetic profiles will partly overlap with one another.

This sets up a complex web of interacting influences, which we need to disentangle (see our other post). Could the overlap we detected between the baby’s genetic profile and their future risk of diabetes and heart disease be driven, not by the impact of those genetic variants within the child, but instead through their shared presence in the mother and mediated through their influence on the maternal environment? Rather than a direct effect of those genetic differences on both early growth and later disease, could those genetic differences have their dominant effect at the level of the mother, with only indirect effects on fetal growth and disease risk?

Those are relatively simple questions to pose, but difficult ones to answer. To do so requires large collections of samples, most obviously from offspring and their mothers (and ideally their fathers), that are only now starting to emerge. On the basis of the analyses that we have been able to do so far, the headline answer we can offer is that both direct (fetal) and indirect (maternal) mechanisms are involved, but that the former seems to be more important than the latter. In other words, the more important mechanism linking fetal genotype with adult disease risk involves the direct effects of the child’s genotype on both early growth and later disease.

But this is most definitely not the entire story. When we looked at the links between early growth and diabetes, a complex picture emerged. Some of the genetic differences that we know are responsible for an increased risk of diabetes in adulthood were associated with higher birth weight in earlier life. However, others were clearly associated with lower birth weight.

We believe (and have evidence!) that this reflects competition between two processes. In the baby, diabetes risk alleles have a direct effect that reduces growth. This is because those variants reduce insulin levels (or the sensitivity of tissues to the actions of insulin), which in turn reduce insulin-mediated early growth: many decades later, those same alleles increasing the risk of later diabetes. At the same time, exactly the same alleles, when present in the mother, put that mother at increased risk of gestational diabetes, leading to increased transfer of sugars across the placenta, stimulating the fetus to produce more insulin, and promoting fetal growth. These two oppositional processes play out in different ways for different genetic variants depending on the variants’ relative impact in early and later life.

Birth Weight
Summary of key relationships: Black arrows reflect purely genetic mechanisms, connecting parental to offspring genes, with shared impacts on fetal growth and later disease risk. Red arrows reflect the DoHAD programming model whereby intrauterine environment influences early growth and leads to long term programming effects on future health. In the joint model that emerges from our data, we add (green arrows) the impact of maternal genotype on intrauterine environment, and emphasise (blue arrow) that the baby’s birthweight is a readout of growth, but not of itself on the causal pathway to future disease risk

Much work remains to be done to further tease apart these processes, but our data provide the strongest evidence yet that direct genetic effects and indirect environmental influences (some of them also secondary to maternal genetic effects) interplay to regulate early growth and influence risk of disease in later life. As a result of this study, we have a clearer idea, a “road map”, for how we should frame and focus future studies to tackle this complexity. In particular, we are planning more genetic analyses of family data (children and their mothers and fathers).

How does this research impact on clinical care? At the heart of this research is a desire to better understand the mechanisms underlying the development of diabetes, obesity and related conditions. This is basic research, but fundamental to developing improved strategies for the prevention and treatment of these conditions.

More immediately, this research matters because, according to the DOHaD hypothesis, improvements in antenatal care and the alleviation of intrauterine deprivation will eventually help to forestall the rising prevalence of diabetes and related conditions. Ongoing research that allows us to quantify the contributions of the various mechanisms that influence the relationship between early growth and subsequent diabetes should provide a better sense as to how far we should expect advances in pregnancy care to achieve those goals.


Dr. Rachel Freathy, of the University of Exeter Medical School, and Prof. Mark McCarthy, of the University of Oxford, are part of the writing group who co-led a major paper on the genetics of birth weight, published in Nature on 28 September 2016