Expert view: Genetics-based nutrition is risky business
Dr Ruth Loos, from the Icahn School of Medicine at Mount Sinai, New York, spoke during ANS' Nutrition Live Online 2020 conference, to discuss the value of genetic information in predicting obesity.
She explained that previous studies have shown that our genes contribute around 40-70% of obesity risk while 30-60% comes from lifestyle. And so far, studies have identified 1,000 gene variants associated with obesity.
She questioned whether we already have the science needed to provide diet advice based on a person’s genetics.
"Some companies believe we do," she said, adding that she decided to sign up to a gene-based personalised nutrition app in order to get deeper insight into the advice they give and the data behind that advice.
So, the case study became a 48-year-old healthy and active woman who does not need to lose weight but wants to live a healthy life.
“I received my results six weeks later and they told me someone with my genotype will be more likely [than the average person] to gain weight on a high saturated fat diet.
“The data they provide to back up this prediction showed that, on my diet - with more than 22g saturated fat a day - my BMI will be 6% higher than an individual with a different genotype.
“I looked at the referenced studies by Corella. D., et al that the company cited and found two very nice studies showing my genotype group compared with two other genotype groups and showed my predicted BMI goes up much quicker on a high saturated fat diet than for those in the other groups... They even replicated those results in another study."
Challenges of using genetics for nutrition
However, Dr Loos said she had a number of concerns with companies using this information to predict weight gain.
"This is not data on weight gain. This does not show weight will increase by eating a high saturated fat diet. It’s simply cross-sectional information. It’s also only one variant. Imagine I carry another gene which creates the opposite effect – will that cancel out the effect of this single variant?
“Also, OK, I’m in the unlucky group which will gain more weight. What about this other group of people? Can we tell them they won’t gain as much weight with a higher intake of saturated fat? I’m not sure this is a safe message to tell this group because maybe it won’t affect their weight but it can still affect their risk of cardiovascular disease and other diseases.
“Another issue is, the same researchers did a third study, which included the observations of the people in the study and it found that the observations of the three groups completely overlapped.
"If you want to use the data for prediction, it’s important that the genotype completely separates the groups.
“Genetic association studies show small effects but we need to see bigger effects in order to be able to predict who is at a high risk and who is not.
“Studying one SNP at a time will never give you a complete picture. We also need to look at other variants - include information on physical activity, diet and so on.
“We need longitudinal studies and interventional studies to see if genotypes make people respond differently to the same diets.”
She concluded that using genetics to provide precision diets does not seem possible at this time.
Can genetics predict obesity?
Last year there was a study that created a genetic score to predict who is at most risk of developing obesity (Khera. A. V. et al, Cell) by combining data from 2.1million variants on a polygenic risk score. The study received huge amounts of media attention and a few months later they had launched a $50 test for people to predict their risk of obesity.
Loos questioned the validity of this test.
She argued that in order for a predictive test to have clinical relevance, it must have an area under the curve of 0.80 or higher. She said if a test has an area under the curve around 0.50, then it’s about as good a predictive tool as ‘flipping a coin’.
Khera's study never published the area under the curve so Loos and her team calculated this and found it was just under 0.64, which was "far off clinical utility".
The argument has been made that the genetic score could be used to predict obesity from early in life (Torkamani and Topol, Cell, 2019) and advice could be given to the parents of those children in the top 5% highest risk of obesity.
Loos explained her concerns with this.
“Then you treat 10 individuals to try to avoid obesity in four individuals. What’s more dangerous, is what about the other 90%? Do you tell them to just live their lives freely?... Not saying anything could put that population at higher risk.
“So using genotypes to predict obesity doesn’t work.”
Genes for nutritional advice?
So maybe genetics can't yet predict obesity but might people be more conscious to live a healthier life if they are warned about the potential health issues associated with their genes?
Two studies by Bloss et al (2011, NEJM and JMG 2013) did genotype profiling for 23 diseases on nearly 3,700 participants. Participants were asked about their weight, diet, exercise and anxiety and they were followed up at three months and at one year.
The team found that those who had been warned that their genes made them more susceptible to obesity were more likely to increase their fat intake at the three-month follow-up.
“This may be understandable because they will say ‘why bother eating a healthy diet if it’s all in my genes’,” suggested Loos.
However, after one year, the results went back to normal.
Loos concluded: "Clearly you may estimate people at risk of obesity but this does not mean we can easily treat people.
“Studies need to be followed up, we need longitudinal studies to see how people respond to certain diets. Lifestyle behaviour is hard to change so just informing people of genetic risk will not necessarily have the outcome you wish for.”
