Metabolomic Analysis in Severe Childhood Pneumonia in The Gambia, West Africa: Findings from a Pilot Study

Background: Pneumonia remains the leading cause of death in young children globally and improved diagnostics are needed to better identify cases and reduce case fatality. Metabolomics, a rapidly evolving field aimed at characterizing metabolites in biofluids, has the potential to improve diagnostics in a range of diseases. The objective of this pilot study is to apply metabolomic analysis to childhood pneumonia to explore its potential to improve pneumonia diagnosis in a high-burden setting. Methodology/Principal Findings: Eleven children with World Health Organization (WHO)-defined severe pneumonia of non-homogeneous aetiology were selected in The Gambia, West Africa, along with community controls. Metabolomic analysis of matched plasma and urine samples was undertaken using Ultra Performance Liquid Chromatography (UPLC) coupled to Time-of-Flight Mass Spectrometry (TOFMS). Biomarker extraction was done using SIMCA-P+ and Random Forests (RF). ‘Unsupervised’ (blinded) data were analyzed by Principal Component Analysis (PCA), while ‘supervised’ (unblinded) analysis was by Partial Least Squares-Discriminant Analysis (PLS-DA) and Orthogonal Projection to Latent Structures (OPLS). Potential markers were extracted from S-plots constructed following analysis with OPLS, and markers were chosen based on their contribution to the variation and correlation within the data set. The dataset was additionally analyzed with the machine-learning algorithm RF in order to address issues of model overfitting and markers were selected based on their variable importance ranking. Unsupervised PCA analysis revealed good separation of pneumonia and control groups, with even clearer separation of the groups with PLS-DA and OPLS analysis. Statistically significant differences (p