Recently, I completed my PhD studies in the EDIT lab, wrote my thesis, and finished reading all the papers in that pile on my desk (or two of those three). Here I describe my work on the effect of the genome on body mass index and its relationship with our environment.
Body mass index (BMI) is a crude measure of your body’s mass, defined as the mass in kilograms of each square metre of your body, assuming you were a giant square with each side as long as you are tall. Typically, a BMI of less than 18.5 is deemed “underweight”, more than 25 is deemed “overweight”, and more than 30 “obese” (World Health Organization). It is imperfect (not least because most of us are not squares). For example, we might be more concerned about the amount of excess fat an individual has than their excess muscle, because excess fat is associated with an increased risk of high blood pressure (Janssen et al, 2004). BMI does not discriminate between different contributors to mass, so people with unusually high muscle mass (such as athletes) might have an “obese” BMI, but not be at risk for future ill health. This means that the real effects of excess fat on ill health might be difficult to spot by measuring BMI across the population because some of the “obese” individuals do not have an excess of fat (Centre for Disease Control).
“ People who have had depression tend to weigh more, and people who are heavier are more likely to develop depression.”
However, just because BMI is crude does not mean it is not useful. BMI is simple to measure (compared to a complex measure like fat-free mass, which requires specialist equipment), which means it is relatively easy to obtain on many people. BMI is also correlated with negative outcomes in later life, including heart disease and diabetes (National Institutes of Health). Particularly, for my interest, it is also associated with depression – people who have had depression tend to weigh more, and people who are heavier are more likely to develop depression (Luppino et al, 2010).
Even more interestingly for me as a psychiatric geneticist, recent cutting-edge work in genetics has given new life to the idea that BMI may be, in part, a behavioural trait. A number of methods that map genetic variation to regions of biological interest, and to sets of interacting proteins, suggest that variants associated with BMI tend to map to proteins found in the brain and nerves (Locke et al, 2015; Finucane et al, 2015). In comparison, variants associated with fat independent of BMI map close to genes associated with laying down fat (Shungin et al, 2015).
“ Recent cutting-edge work in genetics has given new life to the idea that BMI may be in part a behavioural trait.”
The idea that BMI might be a behavioural trait, and that the genetic influences on BMI might also act through changing people’s behaviour, led me to the studies that form the second half of my thesis (which, for those who missed the first exciting instalment is concerned with using the whole genome to study how genes and the environment influence our behaviour).
First, I was interested in how the observed relationship between depression and BMI altered when you considered the genetic influences on each trait. That is, do you see a relationship between BMI and depression after removing the effect of genetics? To examine this, I looked in a group of twenty thousand individuals from the UK, seven thousand of whom reported suffering from depression at some point in their lives. As expected, we saw a small but significant relationship between depression and higher BMI.
I then used a method called polygenic risk scoring (PRS) to assess the effect of the genome on each trait. PRS uses previous studies of a trait to assign two scores to each variant in the genome, ranking them by how strongly they are associated with the trait (the beta value), and how likely it is that this occurred by chance (the infamous p-value). By selecting all variants with p values smaller than a certain threshold, and looking at those variants in your group of interest, you can create scores designed to estimate that trait from your participant’s genes. Doing this in my group, I found that my gene-estimates of BMI and depression were significantly associated with the real thing. However, my gene-estimates of BMI did not predict real depression, nor the other way around.
Finally, I put all three scores into models exploring the association of BMI with depression when considering the gene-estimates of depression and BMI. The association between depression and BMI was not altered when either or both of the gene-estimates were in the model – this suggests that genetic effects don’t contribute a lot to the relationship between depression and BMI. However, that conclusion comes with a heavy pinch of salt – the gene-estimates I used did not capture the full effect of the genome, because the common variant influences we can capture are each individually very small, and there are rarer variant influences (and multi-variant influences) that are difficult to capture with current technology.
Our behaviours are influenced by our genetics and by the environments to which we are exposed. As such, I was interested in how the immediate social environment influences BMI in childhood, and across adolescence. Parenting has previously been shown to influence children’s BMI, both directly through providing food and exercise opportunities, but also indirectly through inspiring children’s behaviours. I was interested to know how the impact of parental warmth and discipline affects BMI in the context of genetic effects. Using a smaller cohort of three thousand children followed through adolescence, I assessed the effects of parenting and the gene-estimate of BMI (again using PRS) on real BMI, and on BMI change from age 11 to age 16. As my analyses progressed, it became clear that a more general measure of the social environment, the socio-economic status (SES) of the parents, was also interesting as an environment to test.
Harsher parenting, considered alone, was associated with higher BMI – but when SES was also included in the model, the effect of parenting reduced considerably. This suggests there may be a general effect of the social environment on BMI in childhood, but that it is not just the effects of parenting that matter. The gene-estimate was again associated with the real BMI. When the effects of genetics and each environment were considered together, parenting and the gene-estimate appeared to be largely separate effects. However, there were very weak indications that the more general measure, SES, may alter the effects of genetics, such that the effect of genetics was weaker in individuals from richer families. Again, however, caveats apply – the effects seen were small and would need to be investigated in larger, independent studies to be convincing, especially given the limitations of gene-estimates previously noted.
– World Health Organization: BMI Classification
– Janssen I, Katzmarzyk PT, Ross R. Waist circumference and not body mass index explains obesity-related health risk. The American journal of clinical nutrition. 2004 Mar 1;79(3):379-84.
– Centre for Disease Control: Body Mass Index: Considerations for Practitioners
– National Institutes of Health: Health Risks of Being Overweight
– Luppino FS, de Wit LM, Bouvy PF, Stijnen T, Cuijpers P, Penninx BW, Zitman FG. Overweight, obesity, and depression: a systematic review and meta-analysis of longitudinal studies. Archives of general psychiatry. 2010 Mar 1;67(3):220-9.
– Locke AE, Kahali B, Berndt SI, Justice AE, Pers TH, Day FR, Powell C, Vedantam S, Buchkovich ML, Yang J, Croteau-Chonka DC. Genetic studies of body mass index yield new insights for obesity biology. Nature. 2015 Feb 12;518(7538):197-206.
– Finucane HK, Bulik-Sullivan B, Gusev A, Trynka G, Reshef Y, Loh PR, Anttila V, Xu H, Zang C, Farh K, Ripke S. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nature genetics. 2015 Nov 1;47(11):1228-35.
– Shungin D, Winkler TW, Croteau-Chonka DC, Ferreira T, Locke AE, Mägi R, Strawbridge RJ, Pers TH, Fischer K, Justice AE, Workalemahu T. New genetic loci link adipose and insulin biology to body fat distribution. Nature. 2015 Feb 12;518(7538):187-96.