Obesity is a complex and heterogeneous condition with diverse presentations and highly variable responses to treatment. While interventions such as lifestyle modification, dietary changes, bariatric surgery, and pharmacotherapy can lead to meaningful and sustained weight loss, their effectiveness is influenced by numerous individual factors.(1) Currently, there is no reliable system to match patients to the most appropriate treatment options. Precision nutrition aims to personalize interventions by considering biological, behavioral, and environmental factors. Improved understanding of individual molecular responses is essential to guide targeted treatment strategies.
We investigated shared molecular features in adipose tissue from individuals (n= 185, 71% female) who successfully lost weight through bariatric surgery or lifestyle interventions by analyzing publicly available transcriptomic datasets from the Gene Expression Omnibus (GEO). Three studies per intervention type were included. Paired linear mixed models using the limma package in R were used to identify differentially expressed genes pre- and post-intervention.(2)
In a complementary approach, genes differentially expressed in Peripheral Blood Mononuclear Cells (PBMCs) between high and low responders (n=55, 38% female) were used to build a logistic regression model to predict weight loss, across four separate dietary interventions. In adipose tissue, both intervention types modulated the Adipogenesis pathway (WP236). Leptin expression was positively associated with weight loss, but only up to 10%, suggesting that dysregulation beyond this threshold may contribute to weight regain. A novel association between RBL2 expression and weight loss was also identified. For the PBMC-based model, one trial served as the discovery cohort. Four genes—OLFM4, DEFA3, ELANE, and MS4A3—were significantly downregulated (p < 0.05) in high responders. Using these genes, a prediction model was fitted. The final fitted model combined OLFM4 expression with baseline weight to predict treatment response with 76.4% accuracy, 41.7% sensitivity, and 100% specificity.
The transcriptome appears to be sensitive to weight loss and can predict weight loss response. The exact genes and pathways that are sensitive appear to be tissue-specific and suggest that weight loss affects multiple tissues across the body. This study highlights the potential of tissue-specific transcriptomic profiling to inform personalized obesity treatment. Shared molecular responses in adipose tissue suggest common biological pathways underlying successful interventions. Additionally, gene expression in PBMCs can be used to develop predictive models for treatment response, with potential for clinical translation using digital droplet PCR as a scalable alternative to sequencing. Further validation in larger, more diverse cohorts is warranted