Abstract
Despite antiretroviral therapy, pneumonias from pathogens such as pneumococcus continue to cause significant morbidity and mortality in HIV-1-infected individuals. Respiratory infections occur despite high CD4 counts and low viral loads; therefore, better understanding of lung immunity and infection predictors is necessary. We tested whether metabolomics, an integrated biosystems approach to molecular fingerprinting, could differentiate such individual characteristics. Bronchoalveolar lavage fluid (BALf ) was collected from otherwise healthy HIV-1-infected individuals and healthy controls. A liquid chromatography-high-resolution mass spectrometry method was used to detect metabolites in BALf. Statistical and bioinformatic analyses used false discovery rate (FDR) and orthogonally corrected partial least-squares discriminant analysis (OPLS-DA) to identify groupwise discriminatory factors as the top 5% of metabolites contributing to 95% separation of HIV-1 and control. We enrolled 24 subjects with HIV-1 (median CD4=432) and 24 controls. A total of 115 accurate mass m/z features from C18 and AE analysis were significantly different between HIV-1 subjects and controls (FDR=0.05). Hierarchical cluster analysis revealed clusters of metabolites, which discriminated the samples according to HIV-1 status (FDR=0.05). Several of these did not match any metabolites in metabolomics databases; mass-to-charge 325.065 ([M+H](+)) was significantly higher (FDR=0.05) in the BAL of HIV-1-infected subjects and matched pyochelin, a siderophore-produced Pseudomonas aeruginosa. Metabolic profiles in BALf differentiated healthy HIV-1-infected subjects and controls. The lack of association with known human metabolites and inclusion of a match to a bacterial metabolite suggest that the differences could reflect the host's lung microbiome and/or be related to subclinical infection in HIV-1-infected patients.