Generation and assessment of multi-omic informed subtypes of type 2 diabetes in diverse populations (gems-T2D) - type 2 diabetes (T2D) represents one of the biggest global health problems of the…

Generation and assessment of multi-omic informed subtypes of type 2 diabetes in diverse populations (gems-T2D) - type 2 diabetes (T2D) represents one of the biggest global health problems of the 21ST century. Defined by the presence of hyperglycemia without another attributable cause, T2D is essentially a diagnosis of exclusion and thus both clinically and etiologically heterogeneous. Several efforts have been made to identify T2D subtypes using a variety of clustering approaches based on either clinical or genetic factors. Current T2D classification approaches have limitations, preventing their implementation into patient care. T2D subtypes based on clinical variables currently rely on measuring these traits at the time of diagnosis, which is not practical for standard clinical care, as these measures may change over time and in relationship to medication use. Genetic-based subtyping approaches have the advantage that they do not change over time but also do not capture the present metabolic state of the individual. Therefore, there is an urgent need to develop a universal and standardized T2D subtype classification to improve understanding of disease and provide a path forward for applying personalized medicine to people with T2D. To develop new T2D subtypes and understand their clinical utility, we have assembled a large and diverse collection of cohorts with phenotypic and multi-omic data, including some of the largest T2D clinical trials in the united states. These studies include glycemia reduction approaches in diabetes: a comparative effectiveness study (grade), look ahead (action for health in diabetes), treatment options for type 2 diabetes in adolescents and youth (today), and the diabetes prevention program. Additionally, we have assembled multiple large, diverse cohorts and biobanks with longitudinal data, including search for diabetes in youth (search), all of US, mass general brigham, and kaiser permanente. In aim 1, we will use state of the art machine-learning clustering approaches to integrate multiple data types, including clinical variables, genetics and metabolomics, to define new T2D subtypes and uncover disease mechanisms. In aim 2, we will evaluate the clinical relevance of existing and newly developed T2D subtypes using data from clinical trials involving glucose-lowering medications and lifestyle interventions, while considering ancestry, environmental factors and social determinants of health. In aim 3, we will develop simplified versions of the diabetes subtypes to assess the clinical utility of these subtypes in a "real world" setting. This proposal will provide a unified and standard classification of T2D and identify subgroups of patients more likely to respond to particular T2D medications and/or develop disease complications, providing a path forward for precision medicine for all.