Proteomic profiling of genetically defined endotypes reveals molecular heterogeneity of obesity

Background Obesity is a heterogeneous condition driven by multiple biological pathways, yet the molecular architecture underlying distinct obesity endotypes remains incompletely characterized. Recently, eleven obesity endotypes have been identified using genetic clustering, capturing distinct biological mechanisms underlying obesity risk. Whether these endotypes are characterized by distinct circulating proteomic signatures that map to different underlying biological pathways remains unclear. Purpose To characterize endotype-specific proteomic signatures of obesity using large-scale population proteomics and to identify biological pathways underlying obesity heterogeneity. Methods We analyzed data from 38,037 participants in the UK Biobank with plasma proteomics measured using the Olink platform (2,923 proteins). Their endotype scores was calculated using endotype-specific partitioned polygenic score (pPSs). Proteome-wide association analyses were conducted for each endotype using multivariable linear regression, adjusting for age, sex, ancestry principal components, lifestyle factors, and additional covariates. To capture overall obesity-related biological effects, primary analyses were conducted without body mass index (BMI) adjustment, with BMI-adjusted models evaluated as sensitivity analyses. Multiple testing in models was addressed using Bonferroni and false discovery rate (FDR) correction. Pathway enrichment analysis was conducted using Reactome with Benjamini–Hochberg FDR correction. Results Across endotypes, we identified 1,209 significant protein associations after Bonferroni correction, with substantial variation in the number, magnitude, and direction of effects across endotypes (Figure 1). Reactome pathway enrichment revealed endotype-specific biological programs, including immune regulation, lipid metabolism, and adipogenesis (Figure 2). The extent to which protein-level associations aggregated into coherent pathways differed markedly across endotypes, reflecting heterogeneity in the organization of proteomic signals. Following BMI adjustment, the number and magnitude of significant protein associations were attenuated, while the direction of effects for remaining associations was largely preserved, consistent with a combination of BMI-mediated and BMI-independent proteomic effects. Conclusion Genetically defined obesity endotypes exhibit distinct circulating proteomic signatures and pathway-level organization, revealing substantial molecular heterogeneity underlying obesity. These findings clarify endotype-specific obesity biology and provide a molecular framework for studying heterogeneity in obesity.Figure 1Figure 2