MSACL 2019 EU Abstract

Biomarker Analysis by Mass Spectrometry and Artificial Intelligence Techniques of Obese Human Plasma

Flavia Luisa Dias Audibert (1); Luiz Cláudio Navarro (1), Diogo Noin de Oliveira (1), Jeany Delafiori (1), Tatiane Melina Guerreiro (1), Carlos Fernando Odir Rodrigues Melo (1), Flávia Troncon Rosa (2), Maria Angelica Ehara Watanabe (3), Anderson Rezende Rocha (1), Rodrigo Ramos Catharino (1)
(1) University of Campinas, Campinas, SP (2) University Center Philadelphia, Londrina, PR (3) State University of Londrina, Londrina, PR

Warning: Undefined variable $headshot in /var/www/html/view_abstract/view_abstract_in_program.php on line 704

Flávia Luísa Dias Audibert (Presenter) University of Campinas >> POSTER (PDF)

Presenter Bio: I am Flavia Luisa Dias Audibert, and I have a bachelor’s degree in Nutrition by the Filadélfia University Center (UNIFIL). After completing my bachelor’s degree, I did a specialization in to Health Applied Biology at the State University of Londrina, guided by Prof. Dr. Maria Angelica Ehara Watanabe, with emphasis in immunology, molecular biology and breast cancer. I completed my master's degree in the same laboratory, with research related to immunology and obesity. It was then that I decided to continue the studies of obesity, but working with mass spectrometry focusing on metabolomic area with emphasis on the elucidation of mechanisms of pathogenesis and the application of this technique for the discovery and development of biomarkers for diagnosis. Currently I continue studying Obesity in the Innovare Biomarkers Laboratory under supervision of Prof. Dr. Rodrigo Ramos Catharino where

Relevant Financial Disclosures (within past 24 months) No relevant financial relationship(s) to disclose. Conflict mitigation NOT REQUIRED.

Introduction
Obesity is characterized as a worldwide epidemic, accounting for more than 600 million diagnosed with obesity. It occurs due to an energy imbalance resulting in the accumulation of fat in the cells of adipose tissue. Clinical recognition of obesity already has pre-established parameters, but the involvement of metabolic variations makes functional diagnosis and prognosis difficult. Although it's possible to classify individuals from new diagnostic methods, these methods do not contemplate all stages of the metabolic process related to adipogenesis. The metabolomics complement the studies in the area of genomics and proteomics, analyzing the final products of the cellular metabolic pathways. Therefore, the aim of this study is the determination of biomarkers present in the plasma of obese patients by mass spectrometry.

Methods
To carry out the project, 90 volunteers of both sexes with diagnosis of overweight and obesity were selected. The participants were selected according to the Body Mass Index (BMI) with BMI individuals above 26 kg / m². For the control group, 90 volunteers of both sexes, eutrophic and free of obesity-associated comorbidities, were selected. From each individual of the groups, a 10mL blood tube with heparin was collected. From this material, 100μL of plasma was used for metabolomic analysis. Plasma samples were immersed in organic solvent (CH3OH) and subjected to chemical protonation or deprotonation. They were then analyzed in Mass Spectrometers MALDI-LTQ-MS (Thermo Scientific) and LTQ-Orbitrap Discovery (Thermo Scientific). Chemical markers will be determined from a machine learning algorithm (Random Forest). Random Forest is a robust and reliable classification method with high predictive performance and low generalization error that fits multiple decision trees and chooses a class that best aggregates the results of those trees.

Results
The mean BMIs for the non-obese and obese groups were 22.3 kg/m2 (SD: ± 2.8; range: 17.4-32.0 kg/m2) and 33.3 kg/m2 (SD: ± 6.5; range: 25.1-66.1 kg/m2), respectively (p<0.0001). The body fat percentage (BF%) average in non-obese and obese groups was 21.7% (SD: ± 3.5; range: 13.7-29.9 %) and 40.3% (SD: ± 9.2; range: 24.9-62.0 %), respectively. The waist circumference average in obese group was 107.1cm (SD: ± 14.6; range: 75-149 cm). The non-obese group had no comorbidities (high blood pressure or diabetes), whereas 15 (37.5%) had hypertension and 7 (17.5%) had diabetes in the obese group. The samples were analyzed and the data collected for the accomplishment of the statistical analysis. The chemical markers of the obese group were determined by random forest.

Conclusions & Discussion
From the algorithm, 8 specific markers were identified for the obese group. We are now in the stage of elucidation of the molecules and their metabolic pathways.