Nuclear magnetic resonance -based metabolomics analysis of patients exposed to sulfur mustard in different stages using random forest method

authors:

avatar Bibi Fatemeh Nobakht Motlagh Ghoochani 1 , 2 , avatar Rasoul Aliannejad 3 , avatar Afsaneh Arefi Oskouie 4 , avatar fariba fattahi 5 , avatar Hossein Ali Sahakhah 6 , avatar Mostafa Rezaei-Tavirani ORCID 7 , *

– Proteomics Research Center, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran
Dept. of Basic Sciences, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran
Pulmonary Department, Shariati Hospital, Tehran University of Medical Sciences, Tehran, Iran
– Dept. of Basic Sciences, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran
Dept. of Chemistry, Sharif University of Technology, Tehran, Iran
Phhysiology Research Center and Dept. of Physiology, Semnan University of Medical Sciences, Semnan, Iran
Proteomics Research Center, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran

how to cite: Nobakht Motlagh Ghoochani B F, Aliannejad R, Arefi Oskouie A, fattahi F, Ali Sahakhah H, et al. Nuclear magnetic resonance -based metabolomics analysis of patients exposed to sulfur mustard in different stages using random forest method. koomesh. 2016;17(3):e153909. https://doi.org/10.5812/koomesh-153909.

Abstract

Introduction: Metabolomics is a powerful technique for determination of biomarkers. Here, we aimed to determine discriminatory metabolomic profiles in different stages of sulfur mustard-exposed patients (SMEPs). Materials and methods: Nuclear magnetic resonance spectroscopy was used to analyze serum samples from 17 SMEPs (normal group patients) and 17 SMEPs (severe group patients). Multivariate statistical analysis using random forest (RF) was performed on a ‘training set’ (70% of the total sample) in order to produce a discriminatory model classifying two groups of patients, and the model tested in the remaining subjects. Results: A classification model was derived using data from the training set with an area under the receiver operating curve (AUC) of 0.87. In the test set (the remaining 30% of subjects), the AUC was 0.8, thus RF model had good predictive power. We observed significant changes in lipid, amino acids and energy metabolism between two groups of patients. Conclusion: Nuclear magnetic resonance spectroscopy of serum successfully differentiates two groups of patients exposed to sulfur mustard. This technique has the potential to provide novel diagnostics and identify novel pathophysiological mechanisms, biomarkers and therapeutic targets