Lipids as key biomarkers in unravelling the pathophysiology of obesity-related metabolic dysregulation

• Published in: Heliyon
• Main Author: Osman A.A.; Chin S.-F.; Teh L.-K.; Abdullah N.; Abdul Murad N.A.; Jamal R.
• Format: Article
• Language: English
• Published: Elsevier Ltd 2025
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85216761625&doi=10.1016%2fj.heliyon.2025.e42197&partnerID=40&md5=6ef8f6711baf03c72555e537694ac68e

Background and objective: Obesity is intricately linked with metabolic disturbances. The comprehensive exploration of metabolomes is important in unravelling the complexities of obesity development. This study was aimed to discern unique metabolite signatures in obese and lean individuals using liquid chromatography-mass spectrometry quadruple time-of-flight (LC-MS/Q-TOF), with the goal of elucidating their roles in obesity. Methods: A total of 160 serum samples (Discovery, n = 60 and Validation, n = 100) of obese and lean individuals with stable Body Mass Index (BMI) values were retrieved from The Malaysian Cohort biobank. Metabolic profiles were obtained using LC-MS/Q-TOF in dual-polarity mode. Metabolites were identified using a molecular feature and chemical formula algorithm, followed by a differential analysis using MetaboAnalyst 5.0. Validation of potential metabolites was conducted by assessing their presence through collision-induced dissociation (CID) using a targeted tandem MS approach. Results: A total of 85 significantly differentially expressed metabolites (p-value <0.05; −1.5 < FC > 1.5) were identified between the lean and the obese individuals, with the lipid class being the most prominent. A stepwise logistic regression revealed three metabolites associated with increased risk of obesity (14-methylheptadecanoic acid, 4′-apo-beta,psi-caroten-4'al and 6E,9E-octadecadienoic acid), and three with lower risk of obesity (19:0(11Me), 7,8-Dihydro-3b,6a-dihydroxy-alpha-ionol 9-[apiosyl-(1->6)-glucoside] and 4Z-Decenyl acetate). The model exhibited outstanding performance with an AUC value of 0.95. The predictive model underwent evaluation across four machine learning algorithms consistently demonstrated the highest predictive accuracy of 0.821, aligning with the findings from the classical logistic regression statistical model. Notably, the presence of 4′-apo-beta,psi-caroten-4′-al showed a statistically significant difference between the lean and obese individuals among the metabolites included in the model. Conclusions: Our findings highlight the significance of lipids in obesity-related metabolic alterations, providing insights into the pathophysiological mechanisms contributing to obesity. This underscores their potential as biomarkers for metabolic dysregulation associated with obesity. © 2025