BMI Strongly Predicts Individual Risk for Diabetes

Brian Ference, MD, MPhil, MSc

As the prevalence of diabetes has increased over the years around the world, there is a great need to better identify at-risk individuals and to effectively intervene and prevent its development in that population.

According to new findings presented at the European Society of Cardiology (ESC 2020) Congress, body mass index (BMI) was found to be a much stronger risk factor than polygenic predisposition. Furthermore, these findings also showed that BMI has a threshold effect on one’s risk for diabetes.

A team, led by Brian Ference, MD, MPhil, MSc, professor and Director of Research in Translational Therapeutics at the University of Cambridge, assessed 445,000 participants enrolled in the UK Biobank. The primary goal of their analysis was to determine how polygenic score for diabetes can most effectively be integrated with BMI in order to predict lifetime risk of developing diabetes.

Of the entire assessed population, 54% were women, and the mean age at the last follow-up was 65 years.

They constructed the polygenic score for diabetes using 6.9cm variants.

BMI was determined from observational analyses using each participant’s measured BMI at the time they were enrolled into the UK Biobank and from Mendelian randomization analyses using a BMI genetic score.

Ference noted that there was no difference in BMI across the quintiles of the polygenic score, which he considered an interesting finding.

Additionally, participants in the highest polygenic score quintile had a 2.9-fold increased risk of diabetes when compared with those who were in the lowest quintile.

When they plotted the risk of diabetes over time, they found that participants with increasingly higher quintiles of polygenic score were at higher risk for diabetes regardless of age. As for participants with increasing quintiles of BMI measured in middle life, the data showed there were similar or steeper trajectories of risk.

Furthermore, they found that BMI had a threshold effect on the risk of diabetes as opposed to a cumulative effect—which is the case with long-term exposure to low-density lipoprotein, systolic blood pressure and risk of atherosclerosis.

This conclusion was supported by the observation that the risk of diabetes was similarly affected by lifelong exposure to increased BMI in Mandelian randomization analyses and short-term exposure to BMI in observational studies.

Thus, they inferred that insulin resistance and subsequent dysglycemia can occur once one has exceeded the BMI threshold, which may vary according to individual.

They also noticed that although the risk of diabetes increased with increasing quintiles of polygenic score from 4.2% to 10.6%, there was nonetheless a 10-fold variance in risk for each quintile depending on BMI differences.

For example, some participants who were in the lowest polygenic score quintile but exhibited high BMI had a 5-fold greater risk of diabetes than the participants with reversed characteristics.

And since there was an 11-fold stepwise gradient of increasing diabetes risk across BMI quintiles—versus 3-fold across polygenic score quintiles—Ference and team suggested that BMI should be used as the primary metric for estimating diabetes risk.

“As demonstrated for both cardiovascular disease and now diabetes, polygenic scores should not be used alone to estimate the risk of common diseases,” Ference said. “Instead, they should be integrated with the known modifiable causes of disease to more accurately predict risk and more accurately identify persons who may benefit from interventions to reduce risk."

He then addressed their findings on the non-cumulative effects of BMI exposure:

“The finding that BMI has a threshold effect rather than a cumulative effect on risk of diabetes implies that most cases of diabetes can be prevented by keeping BMI below the threshold at which dysglycemia develop or reversed after diabetes occurs by lowering BMI below the threshold of dysglycemia.”