BMI-anchored transcriptomics uncovers oxidative and inflammatory poles of obesity heterogeneity.

BACKGROUND

BMI remains the dominant clinical index of obesity, but does not resolve adipose molecular programs that distinguish metabolic resilience from vulnerability.

METHODS

Among 1,479 available human visceral adipose RNA-seq profiles, 1,298 samples from individuals with BMI 30-60 were included in the primary modeling cohort. Using BMI as a phenotypic anchor, we fit a LASSO regression model to derive transcriptome-predicted BMI and applied the resulting BMI-anchored transcriptomic model to mouse selection lines with divergent obesity phenotypes. In humans, directional discordance between measured and predicted BMI was used to define transcriptome-derived metabolically healthy obesity-like (MHO-like) and metabolically unhealthy obesity-like (MUO-like) states and to characterize pathophysiologic variation beyond measured BMI.

RESULTS

Model-selected genes converged on immune, metabolic, mitochondrial, growth, structural-remodeling, and proteostasis programs. Cross-species application showed that BMI-anchored transcriptomic scores more closely tracked adipose-tissue content than body-size condition factor, distinguishing lipid-anabolic and inflammatory features of obesity-prone DU6 mice from the oxidative-catabolic phenotype of DU6P mice characterized by increased muscle mass. Gene-level associations with the transcriptomic score mirrored DU6-associated expression changes, linking the cross-species axis to an obesity-vulnerable adipose state. In humans, MHO-like individuals showed enhanced mitochondrial energy-production programs and more favorable glycemic and lipid profiles, whereas MUO-like individuals showed stress-adaptive metabolic remodeling and greater cardiometabolic medication use. In a separate, non-overlapping validation cohort, projection onto the primary-cohort PCA space retained a signal associated with insulin resistance that remained essentially unchanged after adjustment for sex.

CONCLUSIONS

These findings support a conserved adipose transcriptomic axis linking oxidative metabolic resilience to healthier obesity and lipid-inflammatory remodeling to metabolic vulnerability.