Introduction & Objective: Obesity affects ~890 million adults and causes an estimated 5 million deaths each year. However, other than BMI and several clinical severity scales, there is no accepted framework for stratifying obesity into clinically actionable or etiologically distinct subtypes. Here, we identify and characterize two major subtypes of human obesity using a data-driven approach.
Methods: We performed multi-dimensional discordance analysis on adipose tissue gene expression profiles and morphological data from monozyogtic twins and validated the resulting two data-driven obesity subtypes against independent cohorts containing a wide range of clinical covariates.
Results: We identified two novel data-driven subtypes of human obesity, Type-A and -B, comprising ~90% of all patients with obesity. Type-A and -B are molecularly, epigenetically, morphologically, and physiologically distinct. At any given BMI, Type-B individuals exhibit relative hyperinsulinemia, relative lean mass increase, higher adipose tissue inflammation, and deregulated lipid homeostasis/adipogenesis gene expression. Mouse models and twin studies suggest that Type-B resembles a developmentally programmed obesity. These subtypes validated in the Leipzig Child Adipose Tissue cohort and in silico analysis reveal that Type-A/-B signatures can be found across the spectrum of obesity presentation. Our latest unpublished findings, integrating data available across several major obesity cohorts, indicate that Type-B obesity is reduced in insulin-responsive adipocytes (AdipoPLIN), and enriched for cardiometabolic comorbidities, irrespective of where the individuals are on the landscape of obesity.
Conclusion: Type-A and -B obesity define a major axis of human obesity heterogeneity, which might direct specific clinical intervention on patients. Type-B represents a higher risk form of obesity linked to incidence of comorbidities and suggestive of a potentially insulin-driven etiology.
L. Fagnocchi: None. J. Pospisilik: None.
