Plasma metabolomics profiling for the prediction of cytomegalovirus DNAemia and analysis of virus–host interaction in allogeneic stem cell transplant recipients

Metabolomics analysis of biofluids is increasingly being recognized as a useful tool for the diagnosis and management of a number of infectious diseases. Here we showed that plasma metabolomics profiling by untargeted 1H nuclear magnetic resonance may allow the anticipation of the occurrence of cytomegalovirus (CMV) DNAemia in allogeneic stem cell transplant. For this purpose, key discriminatory metabolites were total glutathione, taurine, methylamine, trimethylamine N-oxide and lactate, all of which were upregulated in patients eventually developing CMV DNAemia. The overall classification accuracy (predictability) of the projection to latent structure discriminant analysis (PLS-DA) model in cross-validation technical replicates was 73 %. Increased levels of alanine, lactate and total fatty acids, and a shift in the fatty acid profile towards unsaturated species, were observed in patients with detectable CMV DNA in plasma. The classification accuracy of this PLS-DA model in cross-validation technical replicates was 81 %. Plasma metabolomics profiling may prove useful for identifying patients at highest risk for CMV DNAemia thus allowing early inception of antiviral therapy.