Metabolomics-guided insights on Bariatric Surgery: a longitudinal chemometrics approach over 1H NMR spectra from serum samples

MARTINEZ BILESIO, ANDRÉS R.; ARGUELLO, MARÍA A.; MATELLICANI, GUSTAVO; NASURDI, ALEJANDRO; BARRERA, MARIA M.; PAZ, MAXIMILIANO; ROCCA, LILIANA; SCIARA MARIELA; FAY, FABIAN; JAUMOT SOLER JOAQUIM; RASIA, RODOLFO M.; GARCÍA REIRIZ, ALEJANDRO G.; BURDISSO PAULA

Congreso; CAC2022: 18th Chemometrics in Analytical Chemistry Conference; 2022

Bariatric surgery is considered the most efficient treatment for diseases related tomorbid obesity [1]. This surgical procedure has proven successful for weight loss, but also forthe progression control of type 2 diabetes considering its metabolic impact [2]. However, theeffect of bariatric surgery on metabolism is still not well defined. In this sense, metabolomics hasemerged in the biomedical research as a field able to shed some light on obesity-relatedmetabolic diseases [3]. The metabolome analysis through high-throughput nuclear magneticresonance (NMR) coupled with chemometric processing provides a picture of the mainmetabolic patterns at a particular time. This approach allows defining metabolic phenotypes(metabotypes) of responses affected by a certain treatment.In the present study, we aimed to discriminate metabolic signatures linked to bariatricsurgery and determine potential adaptations according to the metabolic evolution and theclinical parameters of different patients. A chemometric assisted 1 H NMR metabolomicsapproach was used in order to analyze serum samples of subjects with morbid obesity (n = 15),before (2-3 weeks) and after (48 hours, 5 days, 1 month, 6 months and 12 months) bariatricsurgery.In the first step, a multivariate curve resolution (MCR)-based strategy was applied toobtain the peak integrals profiles of the previously identified metabolites (biomarkers) [4, 5]. Bythis means, a significant reduction in the dimensionality of the dataset was achieved, fromaround 30,000 variables (chemical shifts) in the spectral data matrix to 49 variables(metabolites) in the peak integrals data matrix, without losing the potential for biologicalinterpretation.In the second step, different multivariate analyses were performed over this reducedfeatures matrix. Chemometric methods, including exploratory analysis (principal componentanalysis, PCA), statistical assessment of the effects of the studied factors (ANOVAsimultaneous component analysis, ASCA) and sample discrimination (partial least squaresdiscriminant analysis, PLS-DA), allowed identifying the main metabolic responses associatedwith the bariatric surgery. We defined two metabotypes of response independently of gender,age or body mass index (BMI). In addition, it was possible to identify those metabolitesassociated with the metabolic changes induced by the bariatric surgery. Finally, the evaluationof the dataset by MCR applying the non-negative constraint elucidated the general metabolicprofiles over time, distinguishing three primary temporal trends throughout the bariatric surgeryevolution.Although further studies are needed, our results open new hypotheses in the study ofobesity-linked co-morbidities and provide a comprehensive view of the metabolic changes afterthe surgery.