DEFINE-T2D Projects
The DEFINE-T2D has six core projects that are currently underway. See a description of each of the six projects below. For more information on planned, ongoing, and completed projects, as well as the process to propose a project, visit the DEFINE-T2D Publications & Presentations site.
Project title: Phenotypic clustering of individuals with type 2 diabetes
Description:
- This project assesses T2D subtypes and shows the utility and challenges of using established subtypes across clinical scenarios.
Aims:
- Aim 1: To apply established T2D clustering methods using clinical phenotypic data to subgroup individuals with T2D (i.e., subtypes) and assess clustering performance and describe T2D subtypes.
- Aim 2: Assess T2D subtypes with risk for clinical outcomes including progression in glycated hemoglobin (HbA1c) level and incidence of cardiovascular disease (and subtypes coronary heart disease, cerebrovascular disease, and heart failure), chronic kidney disease, retinopathy, atrial fib, and mortality, where available.
- Aim 3: Compare use of established T2D clustering methods with (3a) alternative clustering methods and (3b) use of simple clinical data to risk-stratify or predict clinical outcomes for individuals with T2D.
Project co-leads:
- Miriam Udler
- Mike Bancks
Project title: Polygenic score clustering of type 2 diabetes and prediabetes
Description:
- This project uses T2D and cardiometabolic disease polygenic scores to subtype and cluster individuals with prediabetes and T2D with the goal of developing genetic tools for early T2D identification.
Aims:
- Aim 1:To develop genetic clustering models in individuals with T2D and prediabetes using T2D partitioned polygenic risk score (pPS) derived from GWAS of T2D and related traits and characterize the clinical features and outcomes across clusters.
- Aim 2:To develop genetic clustering models in individuals with T2D and prediabetes using PRS derived from GWAS of cardiometabolic traits and characterize the clinical features and outcomes across clusters.
- Aim 3:To evaluate T2D pPS and multi-trait T2D PRS in phenotype-based T2D subtypes, and to integrate genetic and phenotypic clustering in individuals with T2D and prediabetes.
Project co-leads:
- Maggie Ng
- Josep Mercader
Project title: Clustering of type 2 diabetes and prediabetes
Description:
- This project examines individuals prior to T2D onset and uses metabolomics and proteomics to determine when lifestyle prevention efforts to reduce T2D severity may be most effective.
Aims:
- Aim 1:To examine whether participants with prediabetes who go on to develop T2D during the course of longitudinal study can be distinguished from those participants with prediabetes who do not go on to develop T2D during the course of longitudinal study via the clustering of their ‘omic data.
- Aim 2: To examine whether patients with T2D can be clustered into discrete groups according to their ‘omic data, prior to meeting clinical criteria to T2D (i.e., prior to disease onset)
Project co-leads:
- Lekki Wood
- Jerry Rotter
Project title: Consortium visibility project
Description:
- This projects goal is to raise awareness and highlight the clinical/translational impact of the DEFINE-T2D consortium.
Project co-leads: The BRC PI's
- Katerina Kechris
- Leslie Lange
- Wei Perng
- Ivana Yang
Project title: Harmonizing T2D-Associated Metabolites Across Metabolomics Platforms
Aims:
Aim 1: Characterize cross-platform comparability of overlapping metabolites.
- Aim 2: Develop imputation models for unmeasured metabolites (including unnamed features).
Project co-leads:
- Tamar Sofer
- Iain Konigsberg
Project title: Clinical phenotype clusters in persons with prediabetes
Description:
- The goal of this project is to identify cardiometabolic subtypes (aka clusters) in people with prediabetes and link these subtypes to risk of diabetes and complications. We will use state-of-the-art machine-learning methods to classify individuals with prediabetes into novel clusters based on clinical risk factors and link these clusters to risks of diabetes, cardiovascular disease, chronic kidney disease, and all-cause mortality.
Aims:
Aim 1: Derive and validate ML prediabetes clusters based on clinical risk factors.
Aim 2: Evaluate the role of ML prediabetes clusters with progression to diabetes and regression to normoglycemia.
Aim 3: Quantify the associations of ML prediabetes clusters with the risks of cardiovascular disease, chronic kidney disease, and mortality.
Project lead:
- Mary Rooney
Opportunity Fund Supplement Projects
These are supplemental projects that are supported through an administrative supplement in order to support/enhance the Consortium goals by funding projects that would support existing or future projects. The purpose of these projects is for data curation/generation in extant data and samples.
Year 2 Opportunity Fund Projects
Project title: Abstraction of medication use from EHR
Description:
- This project will develop a broadly-applicable curation and analysis framework that leverages longitudinal medication history data from EHR data sources in DEFINE-T2D, generating in a suite of tools, protocols, and best practices for the use of medication history data in the context of subtyping analyses
Aims:
- Develop a “curation and analysis” framework that leverages longitudinal medication history data from EHR sources in DEFINE-T2D by:
- (a) generating a suite of tools, protocols, and best practices for the
use of medication history data in the context of subtyping analyses (e.g., genetic and phenotype clustering) - (b) preparation of "harmonized" medication data across most or all EHR data sources in DEFINE-T2D for various analyses that will contribute to the consortium goals. (c) performing preliminary classification of T2D subtypes in response to T2D treatments and cardiometabolic diseases in a subset of BioVU (n=25,000).
- (a) generating a suite of tools, protocols, and best practices for the
Project lead:
- Eric Gamazon
Project title: Metabolomics data in Look AHEAD Trial
Description:
- This project will generate untargeted metabolomics profiling in stored fasting plasma samples collected at baseline from participants of the Look AHEAD trial. These data will be used to refine classification of T2D subtypes and investigate associations of T2D subtypes with risk for cardiovascular disease (CVD), chronic kidney disease, retinopathy, and mortality.
Aims:
- Aim 1: Generate untargeted metabolomics profiling data in stored fasting plasma samples collected at baseline among participants in the Look AHEAD trial. These data will be used to:
- (a) develop and cross validate data type specific models for classification of T2D subtypes based on clinical, genetic, and metabolomics data as well as models that integrate multiple data types
- (b) investigate associations of T2D subtypes with risk for cardiovascular disease (CVD), chronic kidney disease, retinopathy, and mortality.
Project co-leads:
- Mike Bancks
- Katerina Kechris
- Anna Bellatorre
Project title: C-peptide data in Look AHEAD Trial
Description:
- This project will generate C-peptide data in stored fasting plasma samples collected at baseline from participants of the Look AHEAD trial. These data will be used to refine classification of T2D subtypes and investigate associations of T2D subtypes with risk for cardiovascular disease (CVD), chronic kidney disease, retinopathy, and mortality.
Aims:
- Aim 1: Assay C-peptide from stored fasting serum samples collected at baseline for all participants in Look AHEAD. These data will be used to:
- (a) apply clustering models to data for c-peptide homeostatic model assessment of insulin resistance (HOMA2) together with other phenotypic measures to subtype T2D
- (b) investigate associations of T2D subtypes with risk for CVD, chronic kidney disease, retinopathy, and mortality.
Project co-leads:
- Mike Bancks
- Katerina Kechris
- Anna Bellatorre