Characterize Disease Progression Subphenotypes in Real World Populations with Overweight and Obesity using a...

Area of Science:

• Obesity research
• Precision medicine
• Real-world evidence

Background:

• Obesity is a complex condition with diverse individual risks and treatment responses.
• Limited research exists on long-term obesity progression heterogeneity and its impact on associated outcomes and treatment efficacy.

Purpose of the Study:

• To identify obesity progression subphenotypes in a real-world population over 10 years using electronic health records (EHRs).
• To evaluate the heterogeneity of treatment effects (HTE) of glucagon-like peptide-1 receptor agonists (GLP-1RAs) across identified subphenotypes for major obesity-related diseases.

Main Methods:

• Retrospective cohort study of adults with overweight or obesity using EHR data.
• Developed a graph neural network (GNN) model to identify obesity progression subphenotypes.
• Emulated a target trial comparing GLP-1RA users and non-users within each subphenotype to assess disease outcomes.

Main Results:

• Identified three subphenotypes: progressive, intermediate, and stable, with distinct BMI trajectories and baseline characteristics.
• Significant differences in obesity-associated comorbidity risks and all-cause mortality were observed across subphenotypes (p < 0.001).
• GLP-1RA use was linked to reduced heart failure risk in two subphenotypes and a potential increase in chronic kidney disease risk in the progressive group.

Conclusions:

• Obesity progression subphenotypes derived from EHRs demonstrate significant heterogeneity in clinical risks and treatment responses.
• Distinct BMI trajectory groups suggest potential for data-driven, phenotype-guided care in obesity management.
• Findings highlight meaningful variation in GLP-1RA treatment effects, supporting precision obesity medicine and the need for prospective validation of HTE.