Scientists who recently announced an experimental genetic test that can help predict obesity got immediate pushback from other researchers, who wonder whether it is really useful.
The story behind this back-and-forth is, at its core, a question of when it’s worth diving deep into DNA databanks when there’s no obvious way to put that information into use.
The basic facts are not in dispute. Human behavior and our obesity-promoting environment have led to a surge in this condition over the past few decades. Today about 40% of American adults are obese and even more are overweight.
But genetics also plays an important role. People inherit genes that make them more or less likely to become seriously overweight.
While some diseases (like Huntington’s and Tay-Sachs) are caused by a single gene gone awry, that’s certainly not the case for common conditions, including obesity. Instead, thousands of genes apparently play a role in increasing obesity risk.
Many of those gene variants contribute a minuscule risk. Sekar Kathiresan, a cardiologist at Harvard and a geneticist at the Broad Institute, set out to see whether he and his team could find a bunch of these genetic variants and add up their effects. The goal was to identify genetic patterns that put people at the highest risk.
This genetic information “could explain why somebody’s so big, why they have so much trouble keeping their weight down,” Kathiresan says.
His team identified more than 2 million DNA variants of potential interest. He figures most of those variants are irrelevant, but his hunch is, hidden somewhere in there are a few thousand changes that each contribute at least a tiny bit to a person’s risk of developing obesity.
No single gene can do much to move the needle. But he says the composite result, called a polygenic risk score, is still potentially useful. Those with the highest risk scores were more likely to be severely obese (with a body mass index over 40). In fact, 43% of the people with the highest genetic scores were obese.
But the score is far from perfect. For instance, 17% of the people with the highest scores had normal body weights.
The team, with lead author Amit Khera, published its results in the journal Cell.
“The impact of the genetics — and this was a huge surprise to us as well — starts very early in life, in the preschool years, around the age of 3,” Kathiresan says.
That finding suggests prevention efforts are more likely to succeed if they also start in childhood. Kathiresan has a more philosophical takeaway from his work as well.
“I hope this work will hopefully destigmatize obesity and make it very similar to every other disease, which is a combination of both lifestyle and genetics.”
A lot of elaborate genetic analysis is behind the study, which involved more than 300,000 individuals. But the broad conclusions aren’t new.
Scientists already knew genetic risk factors can contribute significantly to obesity. And other studies show that obese children are at high risk for becoming obese adults.
Epidemiologist Cecile Janssens, a professor at Emory University, doesn’t think much of this strategy of adding up the tiny risks from millions of genetic variants to come up with a cumulative risk score.
“In all fairness, we don’t know whether all of these variants really matter,” she says. When asked about the value of doing a study like this, she says, “I have no clue.”
“It is not really answering a very relevant question from the biological perspective, and not really answering a very relevant question from a clinical perspective,” she says.
This type of analysis doesn’t reveal anything about the individual genes that are contributing to obesity, which means you can’t use it to understand the underlying biology. If obesity were a rare disease, a test like this could be useful to identify people at elevated risk. But since it affects 40% of Americans, Janssens says prevention efforts should include everybody.
She is among a group of scientists informally rebelling against the gene-centric way of looking at disease. It’s frustrating for them to see so much money poured into this kind of genetics work, rather than into efforts to change the environment and the behaviors that contribute to diseases like obesity.
Janssens also says that, despite the daunting effort involved in studying 2 million genetic variants, the resulting score still doesn’t explain even 10% of the variation the scientists observed in body mass index. (Kathiresan, who couches his conclusions differently, says the score explains about a quarter of the genetic risk.)
Scientists doing this kind of work hope that data like these, when presented to individuals, will prompt them to change their behavior.
Alas, that’s not supported by scientific reviews.
“This kind of personalized risk information has little [or] no impact on people’s actual behavior,” says Theresa Marteau, who directs the Behaviour and Health Research Unit at the University of Cambridge.
In fact, researchers have worried that when people learn that they are at high genetic risk for diseases like obesity, people would become fatalistic and stop trying to change their behaviors. Fortunately, Marteau says “in a review, we didn’t find any evidence for that.” It seems they just ignore the information.
Ewan Birney, who heads the European Bioinformatics Institute, has been watching this debate play out over the years. Birney agrees with the critics who say obesity isn’t the ideal disease for this kind of analysis.
“One needs to do more than just be able to show a strong statistical association,” he says. “One really needs to show that you can then use that to do an intervention.”
Birney also is wary of making too much of this information because it’s based primarily on data from the UK Biobank, as well as U.S. samples, in which racial minorities aren’t well represented.
There are other instances where these polygenic risk scores can be useful, he says. For example, a score that identifies people at high risk for heart disease identifies people who get the most benefit from cholesterol-lowering drugs called statins. (But it’s unclear whether it would be beneficial to give statins to people who score high on this test but wouldn’t ordinarily be identified as candidates for this medication).
Using a different method of analysis, called a genome-wide association study, scientists have identified more than 140 genes that can slightly increase the risk of obesity. Though their individual influence is small, they do provide clues about the biology of the disease.
For example, one of the relatively potent variants “actually relates to calorie-seeking behaviors,” says Ali Torkamani, who is director of genome informatics at the Scripps Research Translational Institute. Other variants are, unsurprisingly, related to the function of fat cells.
It’s also possible that a careful probe of the genes – rather than the abstract risk score – could identify genetic variants that actually reduce a person’s risk of obesity. A paper in the same issue of Cell as the one that featured work from Kathiresan’s group points in that direction.
While genes influence a person’s risk of obesity, the epidemic in this country is obviously far more extensive than simply people at high risk. And Torkamani notes that the risk score isn’t destiny. “It’s just a probability,” he says. “And you know, when you flip a coin sometimes it comes up heads and sometimes it comes up tails.”
You can contact NPR Science Correspondent Richard Harris at email@example.com.