One of the strands of ‘dark matter’ co-influencing human behavior and health is ethnicity. We know there are differences between racial and ethnic groups in many behavioral and biological factors. Much of the previous research into why has focussed on cultural factors to explain differences in health, for example our previous work has examined whether differences in diet can partially explain differences in CVD-risk between US-based ethnic groups, or whether differences in family size can explain why UK minority families may be more accurate in rater their children’s ADHD-related behaviors than the Caucasians. However, despite what people may assume about differences in lifestyles (dietary and exercise patterns) between ethnic groups in the US, still we understand very little as to why Hispanic- , African- , and Native American- ancestry individuals carry a higher risk of CVD than European- and Asian- ancestry individuals.
In a project lead by Zhe Wang, we decided to look in the genome. It is well known that lipoprotein levels – a marker of CVD risk and metabolic (dys)function are heritable, and studies in Caucasian-ancestry individuals have identified and validated many genetic variants as associated with lipoprotein levels. In the first stage of analyses, Zhe examined whether these variants also associated with lipoprotein levels in Hispanic-, Asian- and African- ancestry Americans. Zhe found that the genetic associations were very similar in Hispanic- ancestry individuals as those with Caucasian ancestry. African- ancestry showed less similar patterns of associations, and those with Asian- ancestry hardly showed any of the same gene-liporotein associations.
In the next stage of analyses, Zhe examined whether these differences arose from differing genetic backgrounds between the groups. That is, there are patterns of correlations among our genetic variation – to put it at its simplest level: if you have genotype X at locus 1, you will have genotype Y at locus 2. At a more complex level we deal with genotypes across several loci (known as haplotypes) and get a probability (not a certainty) of a genotype at another locus. When we run a genome-wide analysis, we therefore do not need to look at all the variation in the genome – we just need to sequence enough loci to be able to infer what is likely to be going on in the rest of someone’s genome. So, when we say variant X is associated with a condition, in Phew. The problem is, this correlation pattern is different by ancestry (and so different by race or ethnicity). So, if variants don’t associate with the same outcomes in different groups we don’t know if it is because we haven’t “tagged” the causal variant, or because the genetic variants have different health and disease associations in different ethnic groups.
Zhe conducted a very systematic (and time-consuming!) examination of the differences in gene-outcome associations and found that the differing pattern of correlations between genetic variants for the ethnic groups did not explain the different gene-lipoprotein associations.
Zhe’s manuscript is available here: http://link.springer.com/article/10.1007%2Fs00439-017-1782-y
What does explain these differences? No one knows yet… More Science is needed!