Project description
This project description is the application we submitted to UK Biobank.
A1. Project title
Identifying preventable risk factors and potential drug-targets in diabetes and associated diseases
A2. Research question(s) and aim(s)
The overarching aim of this project is to investigate the association between genetic and environmental factors and diabetes and associated diseases/conditions (e.g., cardiovascular, kidney and thyroid diseases, obesity and related diseases, e.g., cancers and mental health disorders, including cognitive function). To examine the nature of the investigated associations, we will integrate results from different statistical methods, where each of the methods has different and largely unrelated sources of potential bias, i.e., triangulation. Besides common regression analyses, the methods will include, but not be restricted to dyadic multilevel modeling, instrumental variable approach, mediation analyses, artificial intelligence (AI), machine learning and image analyses. This will enable us to estimate the relative contribution of heritability (genetic variants as instruments for exposure) and environmental exposures (early life, adult lifestyle, parental and spousal/household exposure) on the risk of developing diabetes and associated diseases/conditions. This could potentially help identify preventable risk factors and potential drug-targets in diabetes and associated diseases.
A3. The background and scientific rationale of the proposed research project in general
Diabetes is a major public health burden with an estimated global prevalence of 463 million people. An estimated half of all people with diabetes are unaware of their disease and consequently will delay the onset of treatment and increase the risk of developing complications and comorbidities. These complications and comorbidities include certain types of cancers and mental health disorders (Parkinson’s disease, dementia, mood disorders), cardiovascular disease (CVD), gastro-intestinal disorders, nephropathy, neuropathy, and retinopathy. These complications also often share the same risk factors as diabetes, such as obesity, smoking, sedentary lifestyle, hypertension, and dyslipidemia, making it difficult to disentangle the relative contribution that diabetes might have on the risk for complications. For instance, people with diabetes have a higher prevalence of thyroid disorders compared with the normal population (10.8% vs. 6.6%). Diabetes and thyroid disorders have been shown to mutually influence each other: thyroid hormones contribute to the regulation of carbohydrate metabolism and pancreatic function, and diabetes affects thyroid function tests to variable extents. The presence of thyroid dysfunction may affect diabetes control, such as for the relationship between hyperthyroidism and worsening glycemic control and increased insulin requirements.
A4. A brief description of the method(s) to be used
In addition to simple regression analyses in observational studies, we will employ several other statistical methods. While observational studies are prone to confounding and reverse causation, the genetic methods are largely free from both, and thus allow for inference on causality. For the genetic analysis we will use one-sample (e.g., Wald ratio) as well as two-sample Mendelian Randomization (MR) methods (inverse variance weighted, weighted median, MR-Egger, MR-PRESSO, MR Steiger), as the latter method performs well even within a single large sample such as the UK Biobank. Since the different two-sample MR methods make different assumptions about horizontal pleiotropy (when a genetic variant affects the outcome of interest through traits or pathways other than or in addition to the intermediate phenotype under investigation), we will apply them all in order to lower risk of false positive findings and bias.
To examine the pathways through which an exposure affects an outcome, we will employ mediation analyses: multivariable MR (MVMR) and the NetCoupler-algorithm. MVMR allows for equivalent analysis to mediation within the MR framework, and can therefore estimate mediation effects. NetCoupler is an algorithm that estimates causal pathways between a set of -omic (e.g., metabolomics, lipidomics) or other high-dimensional metabolic data and either a disease outcome, an exposure, or both.
Dyadic multilevel analysis will be used to identify hour-by-hour patterns of covariation in daily physical activity and sleep between couples over a seven day time window. Image analyses based on different established as well as novel modalities will be used to investigate the associations between metabolic phenotypes and diabetes with cardiovascular and cerebral/cognitive complications, i.e. target organ damage (e.g., cardiac function, vascular integrity, carotid remodeling, structural cerebral change) and clinical complications (cardiovascular event, cognitive function).
A5. The type and size of dataset required
Our request is for data for the whole cohort. It would include the following information: population characteristics, employment, demographics, comorbidities, hospital inpatient data, critical care data, primary care data, ECG, accelerometer, genomic and metabolomic data, biochemical markers, lifestyle and environmental exposures (including diet), health outcomes and imaging data.
A6. The expected value of the research
We expect to be able to estimate the relative contribution of heritability (genetic variants as instruments for exposure) and environmental exposures (early life, adult lifestyle and parental and spousal/household exposure) on the risk of developing diabetes and associated diseases/conditions. Identifying potentially preventable risk factors and potential drug-targets in diabetes and associated diseases/conditions has the potential to improve prevention and treatment of these diseases/conditions.
A7. Keywords summarising proposed research project
Diabetes, aetiology, comorbidities, complications, life course
A8. Please provide a lay summary of your research project
The number of people living with diabetes is increasing worldwide. People with diabetes are at a higher risk for getting other health complications like cardiovascular disease, kidney and thyroid diseases, cancer, and mental health disorders. These complications also often share the same risk factors as diabetes, such as obesity, smoking, sedentary lifestyle, hypertension, and dyslipidemia. We want to understand how diabetes contributes to these complications/conditions (and vice versa) so that we can use that information in trying to find potential treatments with drugs or other interventions. For instance, if diabetes impacts a certain disease because of genes, this might be a good place to consider how drugs might help. Or if we find that a specific lifestyle behaviour or feature in the environment, an intervention might be a better approach to use. The project is going to last at least three years.