NEW YORK (AP) — Can a genetic test identify newborns at risk of becoming severely obese by middle age? Researchers say they have come up with one, and that it might allow interventions in childhood to avoid that fate.

The test examines more than 2 million spots in a person's genetic code, seeking variants that individually nudge a person's obesity risk up by a tiny amount. The researchers drew on previously published data about those variants to create a risk score.

A high score didn't guarantee obesity, nor a low score rule it out. But middle-aged people with scores in the top 10 percent were 25 times as likely to be severely obese as those in the bottom 10 percent, scientists reported in a paper released Thursday by the journal Cell.

Those two groups were separated by an average weight difference of about 29 pounds (13 kilograms), researchers said.

Analysis showed the genetic propensity to obesity began having an effect on weight around age 3. Up to about age 8, "you might be able to make a difference in the kids who are born susceptible to obesity," said one author of the study, Dr. Sekar Kathiresan of Massachusetts General Hospital in Boston and the Broad Institute of MIT and Harvard.

But it will take further research to see whether intervening would work, Kathiresan said.

The results for middle age came from a study of about 288,000 people. Overall, the risk-score research included data from more than 300,000 people at various ages. Severe obesity was defined as a body mass index of 40 or more.

Results show genetic inheritance "plays a large role in how heavy one gets," Kathiresan said. The risk score probably takes about half of a person's genetic propensity into account, he said, and it shows similar accuracy in predicting ordinary obesity, defined as a BMI of 30 or more.

Even if one inherits a propensity for obesity, he said, "you still have control over your fate. You're not fated to be obese, but it's very clear those individuals who've inherited susceptibility have to work that much harder to keep the weight off."

Among study participants with the highest scores, he noted, 17 percent were of normal weight. Other analyses show that people who remain lean despite an inherited propensity for obesity tend to eat better and have more physical activity than others with a high score who got fat.

"So you can do something about it," he said.

Ruth Loos, a professor of environmental medicine and public health who did not participate in the study, said the risk score explains more of people's genetic tendency than previous studies did. But she doubted it would be useful in a doctor's office.

It is "never going to be a good predictor," said Loos, of the Icahn School of Medicine at Mt. Sinai in New York.

She noted that in one test of the score's predictive power, only 58 of the 371 subjects scoring in the top 10 percent ended up severely obese. And many other severely obese people didn't score in the top 10 percent, she said.

Simply asking about family history of obesity would probably work better, she said. The large role of lifestyle in obesity means a purely gene-based predictor will never be perfect, she said.

Kathiresan, a cardiologist, said the risk score is best seen as a risk indicator, like high cholesterol. Most people with high cholesterol don't get heart attacks, he said, but they do run a higher risk than others.

Jason Boardman of the University of Colorado Boulder said genetic variants might affect body size indirectly, through an impact on lifestyles or other social and behavioral traits. Other research suggests the effect of genes on size depends on what kind of social and physical environment a person lives in, including such factors as access to unhealthy foods, he said in an email.

___

The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute's Department of Science Education. The AP is solely responsible for all content.

Copyright 2019 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.