Integrating the Effect of BMI and Polygenic Scores to estimate Lifetime Risk and Identify Optimal Treatment Targets to Prevent or Reverse Diabetes

Presented at the ESC congress 2020 by: Prof. Brian Ference Cambridge, UK

Introduction and methods

Diabetes is a risk factor for CVD, observed in observational studies and RCTs. Diabetes doubles the risk of CV events. Prevention of diabetes can be a strategy for the reduction of CVD. Polygenic scores (PGS) have emerged as a risk factor for diabetes. As genetic variants are present from birth, calculating a PGS early in life may identify those with a high lifetime risk of diabetes.

This study was performed to evaluate the risk of diabetes at all levels of a PGS for diabetes depending on differences in BMI. Furthermore, the effect of lifelong exposure to increased BMI with short-term exposure to increased BMI was evaluated on risk of diabetes at all levels of polygenic predisposition.

This study used data of 445,765 participants in the UK Biobank. PGS for diabetes was calculated using 6.9 million genetic variants. In addition, a BMI genetic score was calculated including 255 variants to assess lifelong exposure to BMI on risk of diabetes. The score was used as an instrumental randomization to naturally randomize participants to a higher or lower lifelong exposure to BMI. The primary outcomes was T2DM. In observational studies middle-life BMI was measured.

Main results

• Quintiles based on PGS for diabetes showed only slightly different levels for BMI.
• Risk of diabetes was increased in those in the highest quintile of PGS compared to the lowest PGS (HR 2.90, 95%CI: 2.79-3.02).
• Those with higher PGS for diabetes had higher risk of diabetes for all ages, and trajectories of risk were increasingly steeper compared to those with lowest PGS for diabetes.
• Similar, participants with higher quintiles of PGS for BMI had increasingly steeper trajectories for risk of diabetes.
• Risk of diabetes due to each unit increase in lifelong exposure to BMI in Mendelian randomization analysis was similar to effect of each unit increase in BMI measured in mid-life (OR 1.26 [1.24-1.28] per 1 kg/m²Δ BMI and OR 1.22 [1.21-1.23] per 1 kg/m²Δ BMI, respectively. This indicates that BMI has a threshold and not a cumulative effect on risk of diabetes.
• Within each quintile of PGS for diabetes, risk of diabetes differed by at least 10-fold depending on BMI.
• There was a 2.5-fold gradient of increased risk across PGS quintiles and a 11-fold gradient of increased risk across BMI quintiles, suggesting that BMI is a more powerful risk factor for diabetes.
• Within each quintile of BMI, risk of diabetes differed by 2-fold depending on PGS quintile.

Conclusion

BMI and polygenetic predisposition for diabetes have independent and additive effect on risk of developing diabetes. In addition, BMI appears to have a greater impact on risk of diabetes than PGS. Lifelong exposure to increased BMI in Mendelian randomization appears to have same effect on risk of diabetes as short-term exposure to BMI in observational studies. The authors suggest that PGS for diabetes should not be used alone to estimate risk, but should be integrated with known modifiable causes of diseases to more accurately predict risk and more accurately identify those who may benefit from interventions to reduce risk.

- Our reporting is based on the information provided at the ESC congress -