Comparative Adipose Tissue and Cardiometabolic System Remodeling by Tirzepatide and Semaglutide: an AI-integrated Multimodal Metabolomics and Translational Mechanistic Study

This study is a single-center, randomized, open-label, active-comparator trial designed to characterize the differential effects of tirzepatide and semaglutide on adipose tissue biology and cardiometabolic remodeling over 12 months in adults with obesity and metabolic syndrome. The trial integrates pharmacologic weight-loss therapy with multimodal phenotyping to quantify changes across body composition, hepatic steatosis, metabolic biomarkers, functional strength, and AI-derived electrophysiologic signatures.

The scientific premise is that incretin-based therapies may produce qualitatively distinct patterns of fat loss, ectopic fat mobilization, and cardiometabolic improvement. Tirzepatide, a dual GIP/GLP-1 receptor agonist, may induce greater reductions in visceral adiposity and hepatic fat compared with GLP-1 receptor agonism alone. This trial is designed to directly compare these mechanistic profiles using harmonized imaging, biochemical, and digital phenotyping platforms.

Participants are randomized 1:1 to tirzepatide or semaglutide using a computer-generated permuted block scheme stratified by sex and baseline BMI category. Both interventions follow standard titration schedules and are paired with structured lifestyle counseling to ensure comparable background care. Study visits occur at baseline, 3, 6, and 12 months, with deep phenotyping at baseline, 6, and 12 months.

The phenotyping framework emphasizes regional adiposity, lean mass preservation, hepatic steatosis, and cardiometabolic risk signatures. Whole-body DXA provides quantification of total and regional fat and lean mass, enabling derivation of fat-to-lean mass loss ratios and redistribution indices. Abdominal MRI with proton density fat fraction (PDFF) quantifies liver fat content and abdominal fat compartments. Functional and behavioral assessments-including grip strength and the Chinese version of the Weight-Related Eating Questionnaire-characterize changes in physical function and eating behavior patterns over time.

Fasting biochemical panels measure glycemic indices, lipid metabolism, inflammation, renal and hepatic function, and cardiometabolic biomarkers. Standard 12-lead ECGs are processed through a validated AI pipeline to generate electrophysiologic risk features (e.g., predicted ASCVD risk, ECG-derived heart age), which serve as exploratory digital biomarkers.