Generating genetic prediction scores of circulating biomarker levels that can be leveraged as instrument variables in Mendelian Randomization analysis of complex traits.
Principal Investigator: Dr Sara Lindstroem
Approved Research ID: 55120
Approval date: March 11th 2020
A drawback with scientific studies trying to determine the link between risk factors and disease is that they can be subject to bias. For example, it is difficult to determine if hormone levels are associated with breast cancer, since such study would traditionally require that you as a researcher collect hormone level data from breast cancer cases before their cancer diagnosis. As an alternative approach, it is possible to replace hormone levels by genetic predictors of hormone levels, an approach called Mendelian Randomization. Studying DNA variants that are linked to hormone levels is attractive since it is easy to measure at relatively low cost and they do not change by a cancer diagnosis, meaning that you can collect the data after cancer diagnosis. UK Biobank has collected biomarker and genome-wide genetic data on all 500,000 participants. We propose to use these data to identify genetic predictors of a range of biomarkers, allowing us to use those genetic predictors as proxies for biomarkers in our other studies. This will allow us to assess if these biomarkers are associated with our outcomes of interest (e.g. breast cancer, breast tissue composition), for which we have collected genetic data but not biomarker data on. The biomarkers we propose to include are oestradiol, testosterone, SHBG, C-reactive protein, IGF-1, vitamin D, glucose, HDL, LDL, cholesterol, triglycerides, apolipoprotein A, apolipoprotein B, lipoprotein (a), and calcium. These analyses will help elucidate potential associations between circulating biomarkers and breast tissue composition as previous observational studies have been small and suffered from complex confounding relationships with chiefly adiposity measures. We expect this project to take two years.