OR20-4 NMRI-metabolomics Variables Associated With The Metabolically-healthy Obese Phenotype And The Transitions Between Metabolically-healthy And Unhealthy Obesity Phenotypes.

Abstract

BACKGROUND: The “Metabolically healthy obese (MHO)” phenotype is a research model to assess the pathophysiology of the obesity related complications. Some metabolites are associated with the MHO phenotype (i.e. branched-chain aminoacids and metabolites related with β-oxidation pathways). OBJECTIVE: To identify new metabolomics markers associated with MHO and its progression to “Metabolically unhealthy obesity (MUHO)” using targeted metabolomics. METHODS: Our data source was the Metabolic Syndrome Cohort (n=6114) that included healthy subjects from central Mexico. The cohort was developed to evaluate the risk of MS traits in incident T2D, hypertension and another outcomes. Subjects with metabolomics profile and BMI ≥30 Kg/ m2 (n=1124) were involved. MHO cases (n=66) were matched (1:2) with MUHO subjects (n=132) for sex, age and BMI. Metabolomics profile was measured using quantitative high-throughput Nuclear Magnetic Resonance (NMR). We performed principal component analysis (PCA) for variable dimension reduction. Logistic regression models were fitted to test the effect of the principal components to identify MHO phenotype. Cox Proportional-Hazards regression models were applied to evaluate the risk of progression or regression of MHO or MUHO phenotype during the follow-up. RESULTS: The glutamine concentration was higher in MHO subjects (p=0.009). Glucose, glycerol, isoleucine and glycoprotein acetyls were lower in MHO (p<0.05) compared with MUHO subjects. Using logistic regression models with PCA components as covariates, we identified four components significantly associated with MHO phenotype: PC14 (leucine and alanine) [OR=0.279, p=0.46], PC18 (phenylalanine and glycoprotein acetyl) [OR=0.055, p=0.13], PC19 (albumin) [OR=0.024, 0.17] and PC10 (lactate and histidine) [OR=1.790, p=0.048]. After a 3 years follow-up period, 30 MHO subjects progressed to MUHO (45.45%) and 20 MUHO regressed to MHO (15.15%). The PC that was associated to progression from MHO to MUHO was PC5 (lactate and glucose) [HR=1.16, p=0.021]. PC1 (pyruvate and glutamine), PC2 (β-hydroxybutyrate and glycerol) and PC19 were associated with regression of MUHO to MHO [HR 0.906, p=0.02], [HR1.19, p=0.02] and [HR 0.01, p=0.01] respectively. The clinical-metabolomics progression model that predict the transition MHO to MUHO was integrated by the VAI > 2.20 [HR 16.4, p<0.001], WHtI > 0.63 [HR=4.24, p=0.002] and PC5 [HR= 1.178, p=0.015], X2=8.426 (p=0.004). The model for regression from MUHO to MHO was integrated by age <47 years old [HR= 8.67, p=0.004], weight loss > 2% [HR 4.31, p=0.002] and PC1 [HR= 0.89, p=0.01], X2=7.55 (p=0.006). CONCLUSION: NMRI targeted metabolomics allows to identify the MHO phenotype and models composited by clinical and metabolomics variables are useful to predict the transition between MHO and MUHO phenotypes.