Glycoproteins have an important role in

Glycoproteins have an important role in inflammation, and their synthesis and secretion increases during inflammatory processes (Connelly et al., 2016). We observed that the levels of circulating glycoproteins were higher in patients with pneumonia than in those without. Patients who developed VAP had both elevated levels of N-acetyl glycoproteins and also O-acetyl glycoproteins which are normally intracellular and within the extracellular matrix. Glycoproteins are potential biomarkers for inflammation (Otvos et al., 2015) and our results suggest that signals belonging to N-acetyl and O-acetyl groups from glycoproteins could be use as markers for pneumonia and in particular VAP. Other metabolites that discriminated patients with pneumonia and VAP were phospholipids, especially choline, and amino acids, in particular phenylalanine, alanine, and glutamine. Phenylalanine is elevated in sepsis (Mickiewicz et al., 2013, 2014), possibly representing a reduction in its conversion to tyrosine, an increase in oxidative stress (Neurauter et al., 2008), as a result of immune activation (Ploder et al., 2008), or due to alterations in the catabolism of skeletal muscle. Alanine and glutamine were reduced in those with pneumonia as in previous sepsis studies (Mickiewicz et al., 2013, 2014). These changes may reflect alterations in nutritional status in these conditions and the alteration in the release of amino acids from muscle proteins. Our data also suggested phospholipids may be reduced in those with pneumonia compared with those with brain injury, a consistent finding with work differentiating pneumonia from other causes of sepsis (Neugebauer et al., 2016). Comparison of these data to previous studies demonstrates a number of differences. Analysis of plasma from children with pneumonia (Laiakis et al., 2010) found elevated uric acid, hypoxanthine and glutamic nmda receptor levels with decreases in adenosine diphosphate and tryptophan. Adults in ICU (Seymour et al., 2013) with community acquired pneumonia had higher levels of bile acids, metabolites of steroid metabolism and metabolites related to oxidative stress in non-survivors. Differences in these findings may be because the previous studies used different control groups to the current study and because mass spectrometry was used for analysis. Mass spectrometry is able to detect metabolites at lower concentrations than NMR. However, some metabolites are unsuitable for analysis with mass spectrometry due to their non-volatility (GC-MS), their ability to be ionized, or they can be influenced by ion suppression. Moreover, NMR allows direct quantification of metabolites whereas MS requires targeted assays to achieve quantification. Future work would ideally use parallel sample profiling using both techniques. Other differences in the metabolic markers of pneumonia may arise from the different patient populations, or differences due to comparing survivors to non-survivors as opposed to infected versus non-infected. Differences in causative organisms between studies may result in different metabolites being important, as organisms can be differentiated based on metabolic profiling (Slupsky et al., 2009a). This is especially relevant when considering the ability of metabolic profiling to differentiate community acquired and ventilator acquired pneumonias. Community acquired pneumonia is mostly caused by Pneumococcus and Haemophilus whereas VAP is often caused by gram negative organisms, and resistant organisms such as Staphylococcus aureus. Microbiological results from this study reflected this heterogeneity. The metabolic phenotype of pneumonia may vary depending on the microorganism causing the infection. A larger study will be necessary to stratify the patients with pneumonia according to the organisms causing the disease and define metabolic phenotypes associated with each. Limitations of this study must be considered. The small number of patients in each group, especially those with VAP, where there were just seven patinets, could have led to biased model estimates, this may be represented by the differences seen between the R2 and Q2 values in these models. Because of the limited number of patients in each group we were unable to provide a validation cohort which would have been the ideal method to test the models. However, the current data are a promising starting point from which to perform a larger study. In this study NMR was the analytical platform of choice. Although NMR is both reliable and reproducible it lacks the sensitivity of mass spectrometry (MS) and is not an ideal platform to identify specific lipids, apart from the lipoprotein subclasses. Further work would employ parallel analysis with both NMR and MS and the application of Liquid Chromatography – Mass Spectrometry lipidomic analysis to specifically investigate the role of lipid species in more detail. The study used a group of brain injured patients as a control group and it must be recognized that the findings may represent changes seen in brain injury as opposed to changes in pneumonia. However, the fact that similar changes were seen in those with VAP, all of whom also had a brain injury, to those with pneumonia lends some support to the findings being related to the pneumonic process. Further work looking at different cohorts of intensive care patients needs to be done to clarify the specificity of our findings to pneumonia. Finally as no true “gold standard” tests exist by which to diagnose VAP or pneumonia, the comparison groups in this study may not be entirely discrete. However, every effort was made to ensure correct classification of patients using a combination of clinical “bedside” opinion, objective CPIS and independent assessment. This work looked at VAP in a cohort of brain injured patients, further work will be needed to establish if the current findings apply to other groups of critically ill patients and to all populations. Some clustering was seen in our data, potentially accounted for by ethnicity, which may mean a larger study is needed to accurately stratify patients based on such demographic details.