Epidemiologist Nick Wareham on leptin, type 2 diabetes, and genome-wide association studies
The video is a part of the project British Scientists produced in collaboration between Serious Science and the British Council.
Type 2 diabetes and obesity are two very closely linked conditions. The question arises – what links them? One of the hypothesis is that they might be linked by common genetic predisposition. It’s been long known that both obesity which we characterize as excess body fat and is usually measured by having a high body mass index – the ratio of weight divided by height squared, being excessive with 30 – and also Type 2 Diabetes which is a disorder of chronic hyperglycemia. It’s well known that both of those conditions run in families, and from twin studies we know that the heritability of both conditions is around 50%. The big question is what is the molecular basis for that?
Now, early progress in trying to unravel that molecular basis relied on extreme phenotypes – monogenic conditions that occur and appear to have a more autosomal dominant pattern of inheritance. In the case of diabetes, the initial focus was on forms of diabetes that looked like Type 2 diabetes but which came on early on in adulthood and tended to have an autosomal dominant pattern of inheritance and run in families. And this gave rise to the notion of MODY, or maturity onset diabetes of the young. And modern genetics has allowed that condition to be unraveled into several different subtypes. And the importance of that is not only that it enables people to have a diagnosis, also that it enables a clear pathway of pathogenesis towards the condition to be unraveled. But, perhaps most importantly of all, it allows for personalization of treatment. And in some forms of MODY, patients have benefited from being switched to particular types of glucose-lowering treatment.
So, the key question is whether one can take those rare monogenic insights into human metabolism and disease and transfer them to the majority of the population in whom such rare forms of diabetes and obesity do not occur. Now, that is a much bigger scientific challenge and it only became tractable as a scientific problem with the development of technology. Our ability to characterize the range of variation in the whole genome has largely been driven by a combination of developments in genotyping and global collaboration in consortia around the world with the accumulation of very large population studies of people with obesity and diabetes. And it’s that combination of technology coupled together with a global collaboration that’s really given rise to insights into the disease.
In the case of diabetes, there are now a hundred or so diabetes genes that have been identified. They started with a gene called TCF7L2 which impacts on insulin secretion. That gene was actually identified by a different approach of positional cloning and not by this form of mass genotyping which we call a ‘genotype association study’. But it’s been followed by a whole range of genes that have been identified.
Progress in obesity has been somewhat different and actually the first gene that was identified was FTO, which was identified in the UK by a genome-wide association study. And was followed by other obesity genes the second of which was the melanocortin 4 receptor, which was identified to be associated with obesity by colleagues here in Cambridge. As in diabetes, there are now a hundred or so genes that have been associated with obesity.
In both conditions, obesity and diabetes, the question is – so what? How is this development in understanding of the genetics helped us? And, I think it’s helped us in a number of ways. It’s fueled other research that’s led from those genes to start thinking about what are the pathways that link from that genetic association to the pathogenesis and to the disease. In the case of FTO and obesity, that’s been a somewhat prolonged and difficult scientific journey. And we have not yet unraveled what it is that gives rise to the underlying biology association between FTO and obesity.
In the case of diabetes and hyperglycemia we’ve also identified novel biology from studying genetic association studies. For example, the melatonin receptor gene is associated with fasting hypoglycemia and this has given rise to new investigations into the links between melatonin, maybe circadian rhythms, and control of blood glucose levels. But, beyond the simple transference from genetic associations, epidemiology into biology there are other ways in which this understanding has helped science. Firstly, it has helped in terms of target validation. So, in the case of diabetes we know that there are a set of drugs that impact on the GLP1R receptor. This is a therapy for diabetes that also affects obesity and diabetes, but it wasn’t known whether it had any long-term complications, particularly on heart disease. And genetic variance in the GLP1R receptor are associated with lowering of blood glucose level and it’s also possible to show that they are also associated with lowering of heart disease risk. And this form of use of genetic epidemiology can help the pharmaceutical industry in identifying credible targets for intervention and also can pinpoint likely complications in side effects in the long term. This is important because the cost of developing a drug and then finding out late that it’s associated with side effects is very large.
The second area where genetics may help is in unraveling causal inference. In epidemiology, many factors are associated with outcomes, but it is impossible to tell whether they are causal or whether they are associated with the outcome via a different pathway, which we call confounding. This is a problem that’s inherent to the study of human populations, and it is impossible to get around by any form of analysis, however, genetic variance can be used to mimic a randomized control trial in an approach called Mendelian randomization. So, if you can find a genetic variant that’s associated with an intermediate trait, it’s unlikely to be confounded because that genetic variant has been present since birth and it reflects a lifetime’s exposure to the intermediate biomarker in question. And, it’s possible then to use genetics to try to unravel causal pathways to diabetes. Now, this may sound very technical but it’s extremely important because it could help us focus our attention on pathways to diabetes that are most likely to be causal and could help us avoid wasting time developing interventions for non-causal pathways.
I mean the scale of the studies that have been necessary to identify a common genetic basis of obesity and diabetes are truly huge. One of the first papers we wrote on this topic in around 2003 had a case-control study of five hundred people with diabetes and five hundred people without and involved the study of seventy-three single nucleotide polymorphisms. The studies we are doing now involve hundreds of thousands of people, and rather than studying seventy or eighty single nucleotide polymorphisms we are directly measuring about half a million and then imputing ten million, and that’s over the course of the last ten, fifteen years. So, the scale of the endeavor, the size of the study being undertaken, and the depth of the genetic information has really increased exponentially.
So, type 2 diabetes varies markedly around the world in its prevalence. So, one of the key questions is whether our understanding about genetic variation is a tool explaining the global variation in the prevalence of the condition. And the short answer to that is no. And that’s probably because common variance, which are the ones we are studying most commonly when we use genome-wide association studies are common to all populations and they tend to be traditional, old genes that probably pre-dated the mass human exodus from Africa. What is true is that more of the between population difference in genetics and probably in diabetes risk is explained by more recent genetic variation. And that tends to be rarer and tends to be particular to particular populations. And that’s a very difficult challenge. Now, there’s been some progress, certain groups studying particular population isolates have been able to identify rare genetic variance that cause disease in that population. For example, in a Greenland population there is a gene that particularly affects the two-hour glucose and not the fasting glucose, but that gene variant is almost entirely only found in that Greenland population. So, it has big insights into biology, but in terms of explaining variation in disease between populations we’re really only at the beginning of our understanding.