Study shows genetically-matched diets don’t work. Except it doesn’t.
A recent study, published earlier this week, has been widely reported in the media and lay press as proof that genetically matched diets don’t work. The study contains some important findings, but the genetic part of the study (the bit which has garnered the most headlines) leaves quite a lot to be desired, so did they actually prove that genetically-matched diet advice doesn’t work? Let’s take a closer look and find out.
What did the researchers do?
The researchers randomly assigned just over 600 subjects to follow either a low-fat or low-carb diet for 12 months. The subjects attended 22 total sessions with a dietitian over the 12 month period. The subjects were told to consume “high-quality whole foods” – essentially to increase their intake of vegetables, and reduce their intakes of sugar, refined foods, and trans fats.
What did they find?
The main findings of the study were:
1. Going on a diet causes weight loss, even though these subjects didn’t deliberately count calories – they just made better food choices (which led them to consume less calories.)
2. There was no difference (in terms of statistical significance) in terms of weight loss between the low carbohydrate and low fat diets used in this study.
3. Based on a 3 SNP model, there was no modifying effect of genotype on the amount of weight lost.
What does this mean?
Given recent media reports, you’d be forgiven for thinking that this study proves that genotype-matched diets are not effective.
This study does not show this.
Rather, it shows that a retrofitted combination of three SNPs (PPARG, ADRB2, and FABP2) to predict optimal diet type is ineffective, when dietary adherence is high. These two things are not the same; it’s possible that other genes, such as FTO, TCF7L2, and IRS1, would have successfully predicted optimal diet type; we don’t know, because they weren’t tested for.
FTO, for example, has been shown through a number of studies, with larger sample sizes than this one, to explain variation in weight loss following a diet. It’s not entirely clear why these three SNPs were selected; the researchers cite an abstract from 2010, which itself has not been peer-reviewed and the choice to omit different SNPs with better studied diet interaction is a puzzling one. Given that the abstract the researchers are basing their choice on was presented 8 years ago, it’s a strange choice, and further it’s not clear why the original abstract hasn’t been published in a journal – perhaps it failed to pass the peer-review process. The study itself is used by other genetic testing companies in the formulation of their reports; perhaps now is the time for those companies to remove their claims. So effectively, to reach their assumption on the role of genetics in the effectiveness of dietary changes the researchers appear to be basing their genetic panel on an unpublished abstract from 8 years ago. This is a flawed approach, and begs the question – why not include more SNPs in the study?
It’s also possible that the role of genetically-matched advice is something less binary, that is to say that what matters is how many calories your consume, it’s how you get there where other factors come in to play. In this study, the subjects on both types of diet consumed the same amount of calories, and so it should be expected that weight loss would be similar. Given that the purported mechanism of some other SNPs on diet adherence is due to appetite control and satiety, this could be a potential confounding variable. For example, if those with a risk allele for FTO tend to find saturated fat less filling – and therefore eat more food as a result – they are more likely to gain weight on a diet higher in saturated fat. However, in studies such as this, where calorie intake was the same (because the subjects knew they were in a study), then this doesn’t happen – those with the risk allele of FTO don’t consume more food, perhaps because they are more motivated, and as a result both diet types are effective.
The subjects in this study had 22 sessions with a dietician over a 12 month period, who educated them on the best dietary choices for them to make. They were also blinded to their genetic results. This doesn’t necessarily reflect real-life; here, people are often working things out by themselves. In this situation, there is the greater potential for a genetically matched diet to be effective, when compared to people who aren’t following a genetically matched diet. This was shown in a 2007 study, although this clearly had a much smaller sample size than the more recent study. Here, the subjects knew whether they had undertaken a genetic test or not, and those that had lost significantly more weight over a 12 month period.
Bringing this all together, the most important aspect when it comes to losing weight is consuming fewer calories than you require. This is not exciting, new advice, we’ve known this for a great while. Eat less, lose weight. The main ways to do this are to either consume fewer calories through food, burn more through exercise, or a combination of the two – so how do we get there?
Consuming fewer calories is fairly miserable, it often leads to people under-consuming foods they enjoy eating (like ice cream), and replacing them with foods they don’t necessarily enjoy (like celery). Being able to stick to a diet – and by that, I mean being able to continue to under-consume calories for a significant portion of time – is therefore a matter of adherence; can the person stay on their diet long enough to lose fat?
There are numerous ways to increase adherence; regular sessions with a dietician will certainly help, as would group support – two things used in this present study, which is why the subjects lost weight; (they consumed an average of 1600kcal per day for the duration of the study, quite a reduction.) No matter your genotype, if you reduce your calorie intake – you will likely lose weight. However, this support structure isn’t always available to people outside of studies like these, which is where genetically matched diets may help, as understanding how a particular risk allele interacts with a particular food helps the decision making process.
Stating that the results of this study show that genetically matched diets don’t work is plain incorrect; the study shows that, when dietary adherence is high, a certain selection of 3 SNPs (with lesser effect sizes than other SNPs) don’t predict weight loss.
But what happens if we add a greater number of SNPs, particularly those that have greater effect sizes? What about if, instead of rigorously ensuring diet adherence occurs, we test it under real-world scenarios, with people following the diet themselves at home? Here, the tentative research suggests that genetically-guided information may enhance adherence, and also weight loss. We do need more research, as always, that’s why we are actively involved in a number of studies currently to help progress this understanding.
What this study shows is that retrospective analysis of just 3 SNPs doesn’t predict weight loss in this cohort of people. That’s it. Let’s not misinterpret the findings of a single study to suit our cognitive bias. We take great steps to not do this in choosing how we curate our evidence base to show effectiveness of genetic-diet interactions (using only SNP’s that have a multiple published studies behind them, for example) so I recommend the same approach be applied to those wishing to show the other side of the debate.
