Genetics of obesity and diabetes

Just looking at the families around us, it’s obvious that at least some aspect of our BMI - that’s body mass index, a handy if imperfect measure of weight - is encoded in our genes. This is borne out by genetic research, as well as the fairly obvious finding that a good chunk of our weight is down to our lifestyle. But what do we really know about how our genes influence our weight and metabolism, and - more importantly - our risk of metabolic diseases including type 2 diabetes. I spoke to Dr Robert Scott at the MRC Epidemiology Unit , based at the Institute of Metabolic Science in Cambridge, to find out how he’s hunting for genes involved in weight and disease, and that the answer may lie within the brain rather than the belly. I started by asking him how new genetic technology is helping the search...

Robert - So, it used to be the case in the past when we were doing these early genetic studies we would look at biological candidates and we would extract DNA from individuals, and we would genotype individual at known variants. So in the past, what we would’ve done was said, “Okay, this gene seems to be important in obesity or diabetes. Let’s look and compare individuals who are obese, compared to individuals who are not obese, and then look and see do the obese group carry more of this genetic variation than the other?” But what we can do now is we use this genome wide arrays where we can look at perhaps 1 or 2 million variants all at one time on this tiny microchip. So we extract DNA from individuals, we put the DNA onto this microchip, and it gives us information on what an individual’s genotype is at perhaps 2 million of these different genetic variants, all from the one experiment. So, we can then compare individuals who are obese versus non-obese, or individuals with and without diabetes, and look and see do the individuals with diabetes carry more particular genetic variants in individuals without diabetes.

The real approach that we’re taking now to identifying these genes for obesity, for metabolic disease, for diabetes is we take thes hypothesis free approaches where we measure genetic variance of millions of sites across the genome in lots and thousands of individuals, and then we performed a statistical test to see what are the genes that come out. We’re not driven by any particular hypothesis to say, “Okay, we’re only going to focus on genes which are involved in laying down fat or fat metabolism for example. We’ll just look at all the genes across the genome and see what comes out.” And what's really interesting is that most of the genes I think it’s fair to say, are not things which we would previously have said, “Okay, biologically, that looks interesting.”

Kat - The usual suspects.

Robert - Yeah, so they're not coming out with as the usual suspects as you see. They're coming out with things and giving us real insight into new biology. So perhaps then you might say, “Well, maybe we don’t know as much about the biology of obesity or diabetes as actually we thought we did” because we’re getting many new candidates, the function of which was previously unknown actually. So using these genetic approaches is giving us novel insight into biology of obesity, diabetes that we didn’t have before.

Kat - So, what sorts of genes are we talking about here that are involved in controlling things like weight and risk of diabetes, and other metabolic disease?

Robert - So, what's quite interesting is, when we do the experiments that we’re doing now, many of the genes which we’re coming up with are genes which have biological functions which we don’t really know actually, so we’re getting lots of novel insights. So, actually, what we found for BMI is that many of the genes that we’re identifying which are variants predisposing you to higher BMI are actually genes which seem to regulate your feeding behaviour, so they maybe make you have a preference for high fat foods or for high calorie foods and make you predisposed to want to eat more for example.

Kat - So, this is literally a gene almost for sweet tooth?

Robert - Well, yeah. One of the main genes for BMI or for risk of obesity seems to in the biological functions are being unravelled at the minute, but seems to give you a preference for high fat foods and what makes you want to eat these high fat foods.

Kat - It’s more of a fatty tooth rather than a sweet one.

Robert - Well perhaps, yeah.

Kat - And are you looking across a very wide section of society, because obviously, there's lots and lots of different people living in the UK with the different backgrounds, different genetic makeups. Do you find any interesting kinds of correlations between things like, say the risk of diabetes and other factors as well?

Robert - Well, that’s something we have to control for actually. So, for example, you're alluding to ethnic differences then one of the things you have to control for is this – what we term population stratification. So, exactly what you say. So, individuals for example, particular ethnic groups are predisposed to diabetes, but what you have to really be careful of is how you design your experiment because if of course you take individuals with diabetes, compare them to a control population without diabetes, but all of your individuals with diabetes are, for example, of south Asian descent (a group predisposed to developing diabetes) and all of your controls are of European descent, then what you're likely to find is not necessarily genes which are associated with diabetes, but genes or gene variants which are associated with your ethnic background.

What's ongoing at the moment is lots of multi-ethnic studies. We’re really looking in European individuals and individuals of south Asian descent, of African ancestry. What's really nice is that if you see a signal, a gene variant which is associated with diabetes in Europeans, if you then take that and see the same signal in individuals of African ancestry or South Asian ancestry, then it gives you real insight that these signals caused by something which is real and causal, rather than something which is a chance association. So, replicating things across multiple populations is really important actually.

Kat - How do you just marshal all this information and actually extract meaningful patterns from it? It’s a huge amount of data.

Robert - It really is a huge amount of data. So for example, in the standard experiments, we have for every individual, genotype information at 2 and a half million different mutations or polymorphisms as we term them (changes in the DNA sequence across individuals). Some of the more recent studies we’ve been doing, we’ve had up to quarter of a million individuals, so you could imagine 250,000 people with two and a half million different genetic variants, that's a whole lot of data that we have to handle. And that means when we're mining through all this information and trying to find out what's really important, we have to be statistically very rigorous.

So in a regular experiment what you say is, you perform your statistical test, you find your result, and then you compare that result to the chance of finding that result simply by chance. And normally, what one would see is, you say, “Okay, we’ll accept a 1 in 20 probability that this result is due to chance.” And we say, “Okay, there's a 95% likelihood it's not due to chance.” If that’s the case then we think this might be a real result. But if you're performing two and half million statistical tests then a whole lot of them, 100,000 or more, are going to meet that threshold based on chance alone, so these are not real. So of course, we can't then say, “Okay, 100,000 things look real.” What we have to do is be very rigorous and say, “Okay, we’re not going to accept this 1 in 20 threshold any more. We’re going to go to let’s say, 1 in a million chance.” So, things have to have a very defined, not necessarily a large effect, but a very precise and defined association with the disease, a very highly, statistically significant threshold for us to consider it real.

Kat - You're talking about a lot of people here. Where do you find them from? Where do you recruit all these people?

Robert - We’re analysing together lots of individual studies, so this is not one particular study of obesity in quarter of a million individuals, but what we do is we go across the world and we say, “Okay…” So for example, here and Cambridge, we have a number of studies, looking at the risk factors for obesity and diabetes, and we have studied perhaps 10,000 individuals. But what we can do is we collaborate with centres across the UK, across Europe, across the US. We bring together all these different studies, so each individual study will perform their analysis, and then we perform this method of data amalgamation to bring everything together, and tease out what's real and what's interesting.

Kat - You've looked through all these information, you found correlations with variations in genes that are important for the risk of obesity, for the risk of metabolic diseases. What are you going to do with this information now? How do we make this useful?

Robert - So, one of the things when the human genome project was kicking off was the belief that you could take DNA from everyone, measure your risk of disease and then you could say, “Okay, you're likely to develop obesity, diabetes and suchlike.” Actually, things have really turned around and really – and that was probably never likely to happen I think. What really we’re getting and the main benefit for doing this genetic studies is not necessarily to predict to become obese or diabetic, but like I've just mentioned to you, it gives us real insight into the biology of these diseases where we maybe had a prior belief that what’s important in obesity is how your body metabolises the food that you eat.

These new genes that we’re finding are telling us, “Well, okay that may well be important.” We have hypotheses that that’s important, but actually, people’s preference for particular foods are important, so it gives you novel insight into the biology of obesity. Similarly for diabetes, many of the genes we’re finding for diabetes that are involved in particular processes in the aetiology of diabetes and that perhaps not now maybe, but in 10 years and 20 years gives you novel insight into therapeutic targets.

Kat - Have there been any findings that have been very intriguing or things that you found particularly fascinating?

Robert - I think it’s quite easy for individuals like myself over Christmas will indulge and eat too many mince pies and we say “Well, I have genes which are making me eat too much.” Then one has to be careful not to ascribe everything to genes and I have responsibility to look after myself. So, one of the interesting things which has been found in studies previously is that while carrying particular genetic variants increases your BMI, what's been found previously is that if you compare the effect of these genetic variants amongst individuals who are physically active, who get a lot of exercise, who walk a lot, to individuals who are inactive, who don’t get as much exercise or physical activity as perhaps they should, then the effect of these genetic variants actually appears to be bigger in individuals who are physically inactive.

So while having, carrying particular genetic variants might increase your level of BMI, the really important public health message is that, if you're physically active you can perhaps offset some of that genetic risk. So being physically active and being careful about your diet is actually very important, despite perhaps you carrying an increased genetic risk.

Kat - We’re just moving so far away from the idea that, well, if you're overweight, it’s because you’ve eaten too many pies and you don’t do enough sport to a much more sophisticated and subtle idea of how our bodies work.

Robert - At it's very simplest level, if one is to eat less then exercise more, then of course, you can reduce your risk of obesity and your BMI will be lower. But I think it’s very fair to say that it’s very complex. Actually, it’s quite easy for people to ascribe being overweight to simply not doing enough exercise and eating way too much, but what are the causes of eating too much? Why are some individuals driven to eat more than others? And some of the differences, the inter-individual differences in how we do these things are likely to be done to genetic variation. So I think it’s a very complex story.

Kat - That was Robert Scott, from the MRC Epidemiology Unit in Cambridge.