Introduction: Genome-wide association studies have identified numerous glycemic traits-associated SNPs with myocardial infarction (MI) and stroke. However, conventional regression methods combining these large numbers of highly significant SNPs in a weighted genetic risk score (wGRS) have only explained a small proportion of variance in MI and stroke. Therefore, we used a machine-learning algorithm (gradient boosting) to test the association of genetic risk for glycemic traits with MI and stroke in the UK Biobank population.

Methods: We identified 4626 SNPs associated with glycemic traits from the NGHRI catalogue for GWAS studies. wGRS were calculated using the effect estimates from the GWAS studies. We used a gradient-boosting machine-learning (GBM) model to identify the relative influence (RI) of baseline variables and wGRS of the glycemic traits on prevalent stroke and MI in the UK Biobank population.

Results: The study consisted of 409,633 individuals (53% females) with a median age of 58 (51-63) years and a median BMI of 26.7 (24.1-29.8) kg/m2. The prevalence rates of MI and stroke were 3.6% and 2.3%. In the GBM model, top wGRS associated with MI were type 2 diabetes (RI=14.6) , blood glucose levels adjusted for fasting time (RI=12.2) and beta-cell glucose sensitivity (RI=10.06) . In contrast, the top wGRS associated with stroke were HbA1c (RI=12.13) and fasting insulin levels (RI=10.31) . The addition of wGRS’s to the model comprising the baseline characteristics slightly improved the AUC for predicting stroke and MI.

Conclusion: We showed a differential effect of wGRS for various glycemic traits on the risk of MI and stroke in the UK Biobank population. However, even in machine learning models, these wGRS for various glycemic traits had a limited utility in predicting stroke and MI in the general population.


H.Deshmukh: None. M.Miles: None. A.Bhaiji: None. N.Shah: None. S.Akbar: None. M.Papageorgiou: None. T.Sathyapalan: n/a.


NIHR Clinical Lectureship

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at