Anika Vaarhorst's Paper has been accepted by Circulation Cardiovascular Genetics
Anika's paper has been accepted by Circulation Cardiovascular Genetics. Please follow this link to read the prepublication version of the paper or a short summary below...
The traditional risk factors (TRF) sex, age, total and HDL cholesterol, systolic blood pressure, current smoking, diabetes, body mass index and parental history of myocardial infarction, are currently used to predict coronary heart disease (CHD) in an individual. During the last years investigators found many common genetic variants associated with CHD or CHD risk factors. In this study we investigated if genetic risk scores (GRS) based on these genetic risk variants can be used to predict CHD in addition to the currently used TRF.
We constructed 3 GRS by simply counting the number of genetic risk variants present in every individual. The first GRS was based on 179 genetic variants associated with CHD or CHD risk factors. The second was based on 153 genetic variants associated with CHD risk factors and the third on 29 genetic variants associated with CHD. We found that these 3 counted GRS were not associated with future CHD, nor did these GRS improve risk prediction. However, some genetic variants have a larger effect than other variants. Therefore, we adjusted for the effect sizes of genetic variants associated with CHD and found that this weighted CHD GRS was associated with future CHD. In addition, we saw some evidence for a minor improvement in risk prediction.
As an exploratory approach to achieve weighting and to reduce the number of genetic variants needed in a GRS, we used a statistical technique called LASSO regression. When this technique was applied on the 29 genetic variants associated with CHD, we found a set of 3 genetic variants. A GRS based on these 3 variants performed equal to the weighted CHD GRS. We also applied LASSO regression on all 179 genetic variants and found a set of 14 genetic variants. A GRS composed of these 14 genetic variants was associated with future CHD and we saw some evidence for improvement in risk prediction. These results need to be confirmed in an independent population.