Marlene Rietz

Background: Type 2 diabetes (T2D) may cause major diabetes-related complications including cardiovascular disease (CVD), nephropathy, retinopathy, and neuropathies. Evidence profiling the interactions of risk factors in these complications is scarce. Objectives and hypotheses: We aim to investigate the interactions between multidimensional risk factors, including genomics, molecular biomarkers, medication, comorbidities, and objectively measured physical activity (PA), in association with the development of and mortality from diabetes-related complications in the Danish Centre for Strategic Research in Type 2 Diabetes (DD2) cohort. This integration will support complex risk prediction models advancing personalised diabetology. Methods: Until 2024, approximately 12,700 newly diagnosed individuals with T2D were enrolled in the DD2 cohort. Core data collection included anthropometrics, accelerometry, plasma and urine samples, self-reported lifestyle and T2D heritability, and health registry data. Data describing complication outcomes will be obtained from Danish national registries. Polygenic risk scores for T2D, complications, comorbidities, and adverse lifestyle habits will be calculated from plasma-based genome-wide sequencing (GWS). Unsupervised machine learning will be used to group clinical biomarkers into patterns associated with increased risk of complications. For a subset of the DD2, prospective trajectories of PA and sedentary time (ST) will be computed. Registry data describing T2D pharmacological treatment and comorbidities will be integrated with individualised risk profiles. Cox-proportional hazards models will be created for each risk factor and interactions. Lastly, all risk factors extracted from the DD2 study will be combined using a gradient boosting model (GBM) to predict personalised risk scores for T2D complications independently of interactions. Potential Impact: Analysing risk factor dynamics in T2D complications will allow for effective personalised medicine.