Polygenic Risk, Rapid Childhood Growth, and the Development of Obesity

This post is about my recent paper on genetic risk for obesity and children’s growth and development. By way of background, genetic discovery research in obesity had mostly focused on adults. In the most successful discovery study, researchers from an international consortium assembled a quarter of a million adults and scanned their genomes for common variants that predicted who had a higher body mass index. They discovered 32 variants–called single nucleotide polymorphisms or “SNPs”–that were associated with higher BMI and thus risk for obesity. The question I set out to answer in my research was how these new genetic discoveries related to early life precursors to adult disease.

The developmental origins of obesity stretch back as far as growth in utero: higher weight at birth, faster weight gain in infancy, and an accelerated growth trajectory through childhood all predispose to obesity later in life. I wanted to know whether genetic risks discovered in samples of adults were related to these developmental processes.




Through my mentors I had access to data from the extraordinary Dunedin Multidisciplinary Health and Development Study, a 4-decade longitudinal study of a population-representative  birth cohort of about 1,000 individuals. For these 1,000 people I had data on their weight at birth, and on their height, weight, and a variety of other measurements at ages 3, 5, 7, 9, 11, 13, 15,  18, 21, 26, 32, and 38 years. I also had their DNA.

To measure genetic risk, I compiled information from GWAS to derive a “genetic risk score,” a kind of summary of genetic risk from all the different variants in the genome that GWAS had identified as predisposing to obesity.

I then asked if this genetic risk score predicted the developmental precursors of obesity. Did it predict faster growth in utero? in infancy? in childhood?

In fact, we found that genetic risk did not predict growth in utero. Babies weighed about the same, regardless of their genetic risk. But babies at higher genetic risk did gain more weight during  infancy and went on to grow faster through childhood. In turn, this rapid early growth led to obesity in adolescence and adulthood. Subsequently, cohort members with higher genetic risk were more likely to be obese as teenagers, as 20-somethings, in their 30s, and also chronically across those periods.


Strikingly, our analysis indicated that genetic risk translated to obesity only in those cohort members who grew rapidly earlier in life .

So the pattern was, genetic risk–>rapid early growth–>obesity. If the middle link was removed from that chain, genetic risk had no effect. In other words, for children who  grew at a normal rate during childhood, genetic risk was uncoupled from adult obesity.

My study was not designed to ask what protected this subset of children from the growth accelerating consequences of genetic risk. (though this is something I hope to pursue in the future) In fact, we understand relatively little about what leads some children to grow slowly and others rapidly, at least in populations where most of the kids have enough to eat. But understanding how nutrition, exercise, and other features of children’s lives affect how fast they grow now looks to be critical to understanding (and ultimately preventing) obesity.

Another puzzle the study left me with relates to family history. Children with heavier parents are likely to be heavier themselves. This is likely due to both genetic and environmental factors. I wanted to know if my genetic risk score added information over and above family history–was it really new or was this just another way of finding out what we could get from putting mom and dad on a scale? It turned out children’s genetic risk had very little to do with how big their parents were. Even more surprising, no matter whether mom and dad were thin or fat, higher genetic risk kids grew faster and were more likely to become obese, and lower genetic risk kids grew slower and were more likely to stay thin.










So, why do genetic risk and family history have so little to do with one another? The answer is that we just don’t know. We do have some ideas, and I will get into these in a later post.


This entry was posted on Wednesday, October 31st, 2012 at 9:23 pm and is filed under Papers. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

2 thoughts on “Polygenic Risk, Rapid Childhood Growth, and the Development of Obesity

  1. I read your paper. It is great work. Thanks also for putting it in simple words in this blog.

    I have a question for you. You mention that

    “In other words, for children who grew at a normal rate during childhood, genetic risk was uncoupled from adult obesity.”

    Has this been shown in the paper? If yes, where (which table/ figure/ part of text)?

    I tried to look at the paper and see if adjustment for childhood BMI had a NULL effect of polygenic risk score on adult BMI, but I don’t see that part. May be it is in the mediation analysis – but there you mention mediation is in part, not complete. So, I am confused.

    • Thanks AJ! Glad you found the paper and summary helpful. Regarding the sentence in question, it does refer to the mediation analysis. You are correct that we find only partial mediation in the structural equation analysis. The word “uncoupled” references the adjusted regression model, in which we see no statistically significant association between genetic risk and obesity net of early childhood growth.

Leave a Reply to Dan Belsky Cancel reply

Your email address will not be published. Required fields are marked *