Identifying Four Obesity Axes Through Integrative Multiomics and Imaging Analysis

Abstract

We aimed to identify distinct axes of obesity using advanced magnetic resonance imaging (MRI)-derived phenotypes. We used 24 MRI-derived fat distribution and muscle volume measures (UK Biobank; N = 33,122) to construct obesity axes through principal component analysis. Genome-wide association studies were performed for each axis to uncover genetic factors, followed by pathway enrichment, genetic correlation, and Mendelian randomization analyses to investigate disease associations. Four primary obesity axes were identified: 1) general obesity, reflecting higher fat accumulation in all regions (visceral, subcutaneous, and ectopic fat); 2) muscle dominant, indicating greater muscle volume; 3) peripheral fat, associated with higher subcutaneous fat in abdominal and thigh regions; and 4) lower-body fat, characterized by increased lower-body subcutaneous fat and reduced ectopic fat. Each axis was associated with distinct genetic loci and pathways. For instance, the lower-body fat axis was associated with RSPO3 and COBLL1, which are emerging as promising candidates for therapeutic targeting. Disease risks varied across axes; the general obesity axis was correlated with higher risks of metabolic and cardiovascular diseases, whereas the lower-body fat axis seemed to protect against type 2 diabetes and cardiovascular disease. This study highlights the heterogeneity of obesity through the identification of obesity axes and emphasizes the potential to extend beyond BMI in defining and treating obesity for obesity-related disease management.

Article highlights: This study aimed to address potential limitations of BMI by exploring the heterogeneity of obesity using magnetic resonance imaging-derived fat distribution and muscle volume measures. We sought to identify distinct obesity axes and investigate their genetic, metabolic, and disease associations. Four obesity axes were identified (general obesity, muscle dominant, peripheral fat, and lower-body fat), each linked to unique genetic loci, metabolic traits, and disease risks. These findings emphasize the potential to extend beyond BMI in defining and managing obesity, offering a more nuanced framework for understanding and treating obesity-related diseases.

Figure 1

Characteristics of obesity axes. Radial plots display the magnitudes of PC loadings for the four obesity axes from men (A–D) and women (E–H). Points above the inner circle indicate positive loadings, reflecting traits that contribute positively to the respective obesity axis, whereas points below the inner circle represent negative loadings, indicating traits that contribute inversely to the axis. General obesity (A and E), muscle dominant (B and F), peripheral fat (C and G), and lower-body fat (D and H) axes are shown. ASAT, abdominal subcutaneous adipose tissue; TSAT, thigh subcutaneous adipose tissue; VAT, visceral adipose tissue.

Figure 1

Characteristics of obesity axes. Radial plots display the magnitudes of PC loadings for the four obesity axes from men (A–D) and women (E–H). Points above the inner circle indicate positive loadings, reflecting traits that contribute positively to the respective obesity axis, whereas points below the inner circle represent negative loadings, indicating traits that contribute inversely to the axis. General obesity (A and E), muscle dominant (B and F), peripheral fat (C and G), and lower-body fat (D and H) axes are shown. ASAT, abdominal subcutaneous adipose tissue; TSAT, thigh subcutaneous adipose tissue; VAT, visceral adipose tissue.

Figure 2

MRI scans and fat distribution patterns across obesity axes: general obesity (A and B), muscle dominant (C and D), peripheral fat (E and F), and lower-body fat (G and H). MRI scans illustrate the contrasting fat distribution patterns observed in individuals with the lowest (A, C, E, and G) and highest (B, D, F, and H) scores along each obesity axis. These visual comparisons highlight the distinctive anatomical fat accumulation and muscle distribution associated with each axis.

Figure 3

Relationship between obesity axes and age in men (A) and women (B). Scatter plots depict the variation in scores for each obesity axis across different ages. General obesity scores tended to increase with age, whereas scores for other axes, such as the lower-body fat axis, decreased in older individuals.

Figure 4

Ancestry-related variation in obesity axes: general obesity (A), muscle dominant (B), peripheral fat (C), and lower-body fat (D). Density plots show the distribution of scores for each obesity axis across different ancestry groups: African (AFR), Central/South Asian (CSA), East Asian (EAS), and European (EUR). pop, population.

Figure 5

Genetic correlations between obesity axes and selected biomarkers, lifestyle traits, and psychological disorders. Heat map of genetic correlations (rg) between obesity axes and anthropometric traits, insulin-related traits, and metabolic biomarkers (A); metabolites (B); lifestyle traits and psychological disorders (C); and various disease outcomes, including cardiovascular disease and type 2 diabetes (D). Colors and their intensities represent the correlation coefficients (rg), with asterisks indicating statistical significance after multiple testing correction (P < 0.00011). ADHD, attention-deficit/hyperactivity disorder; HOMA-B, HOMA for β-cell function; HOMA-IR, HOMA for insulin resistance; MASLD, metabolic dysfunction–associated steatotic liver disease; MUFA, monounsaturated fatty acid; OCD, obsessive-compulsive disorder; PTSD, posttraumatic stress disorder; PUFA, polyunsaturated fatty acid; SHBG, sex hormone–binding globulin.

Figure 6

MR. The heat map illustrates causal associations between obesity axes and selected disease outcomes. Colors and their intensities represent the direction and strength of associations determined by the IVW method. Asterisks indicate statistical significance after Benjamini-Hochberg correction (adjusted P < 0.05). MASLD, metabolic dysfunction–associated steatotic liver disease.