Rapid etiological classification of meningitis by NMR spectroscopy based on metabolite profiles and host response

Preclinical Models for Cryptococcosis of the CNS and Their Characterization Using In Vivo Imaging Techniques

Infections caused by Cryptococcus neoformans and Cryptococcus gattii remain a challenge to our healthcare systems as they are still difficult to treat. In order to improve treatment success, in particular for infections that have disseminated to the central nervous system, a better understanding of the disease is needed, addressing questions like how it evolves from a pulmonary to a brain disease and how novel treatment approaches can be developed and validated. This requires not only clinical research and research on the microorganisms in a laboratory environment but also preclinical models in order to study cryptococci in the host. We provide an overview of available preclinical models, with particular emphasis on models of cryptococcosis in rodents. In order to further improve the characterization of rodent models, in particular the dynamic aspects of disease manifestation, development, and ultimate treatment, preclinical in vivo imaging methods are increasingly used, mainly in research for oncological, neurological, and cardiac diseases. In vivo imaging applications for fungal infections are rather sparse. A second aspect of this review is how research on models of cryptococcosis can benefit from in vivo imaging methods that not only provide information on morphology and tissue structure but also on function, metabolism, and cellular properties in a non-invasive way.

Methodological Developments for Metabolic NMR Spectroscopy from Cultured Cells to Tissue Extracts: Achievements, Progress and Pitfalls

This is a broad overview and critical review of a particular group of closely related ex vivo and in vitro metabolic NMR spectroscopic methods. The scope of interest comprises studies of cultured cells and excised tissue, either intact or after physicochemical extraction of metabolites. Our detailed discussion includes pitfalls that have led to erroneous statements in the published literature, some of which may cause serious problems in metabolic and biological interpretation of results. To cover a wide range of work from relevant research areas, we consider not only the most recent achievements in the field, but also techniques that proved to be valid and successful in the past, although they may not have generated a very significant number of papers more recently. Thus, this comparative review also aims at providing background information useful for judiciously choosing between the metabolic ex vivo/in vitro NMR methods presented. Finally, the methods of interest are discussed in the context of, and in relation to, other metabolic analysis protocols such as HR-MAS and cell perfusion NMR, as well as the mass spectrometry approach.

Trehalose as quantitative biomarker for in vivo diagnosis and treatment follow-up in cryptococcomas

Brain lesions caused by Cryptococcus neoformans or C. gattii (cryptococcomas) are typically difficult to diagnose correctly and treat effectively, but rapid differential diagnosis and treatment initiation are crucial for good outcomes. In previous studies, cultured cryptococcal isolates and ex vivo lesion material contained high concentrations of the virulence factor and fungal metabolite trehalose. Here, we studied the in vivo metabolic profile of cryptococcomas in the brain using magnetic resonance spectroscopy (MRS) and assessed the relationship between trehalose concentration, fungal burden, and treatment response in order to validate its suitability as marker for early and noninvasive diagnosis and its potential to monitor treatment in vivo. We investigated the metabolites present in early and late stage cryptococcomas using in vivo ¹H MRS in a murine model and evaluated changes in trehalose concentrations induced by disease progression and antifungal treatment. Animal data were compared to ¹H and ¹³C MR spectra of Cryptococcus cultures and in vivo data from 2 patients with cryptococcomas in the brain. In vivo MRS allowed the noninvasive detection of high concentrations of trehalose in cryptococcomas and showed a comparable metabolic profile of cryptococcomas in the murine model and human cases. Trehalose concentrations correlated strongly with the fungal burden. Treatment studies in cultures and animal models showed that trehalose concentrations decrease following exposure to effective antifungal therapy. Although further cases need to be studied for clinical validation, this translational study indicates that the noninvasive MRS-based detection of trehalose is a promising marker for diagnosis and therapeutic follow-up of cryptococcomas.

A validated 1H-NMR method for quantitative analysis of DOTAP lipid in nanoliposomes containing soluble Leishmania antigen

Leishmaniasis is a serious health problem that needs a suitable vaccine delivery system to control the disease. Cationic lipids such as 1, 2-dioleoyl-3-trimethylammonium-propane (DOTAP) have been widely used in nanoliposomes' formulation to deliver antigen and adjuvant at the same time to induce protection against Leishmaniasis. Therefore, it is necessary to accurately quantify DOTAP concentration in the formulation and biological materials. Due to the poor UV absorbance of DOTAP, the use of the conventional HPLC-UV method was impossible. Currently, an evaporative light scattering detector (ELSD) or MS/MS detector in conjunction with HPLC is used to quantify DOTAP. These methods have several disadvantages, including time- consuming during extraction procedure and decrease or/and even remove some components of the formulation. According to the advantages of the quantitative ¹H Nuclear Magnetic Resonance (¹H-NMR) spectroscopic method, a free extraction approach was developed to the assay of DOTAP in nanoliposomes containing Leishmania antigens. This method was carried out based on the relative ratio of signal integration of DOTAP [CH₂ (CH₂-CH = CH-CH₂)] in δ 2 ppm to a definite amount of an internal standard called dimethyl sulfone (DMSO₂). The q¹H-NMR method showed good precision (intra-day RSD = 1.8 % and inter-day RSD = 2.5 %), linearity (in the ranges of 1.3-7.8 mg. mL^-1 with correlation coefficients at 1), repeatability (RSD ≤ 2.39 %), and stability (RSD ≤ 2.32 %) for the quantification of the DOTAP without any extraction method. Considering all the experiments conducted in this study, NMR can be a feasible alternative to other traditional techniques for the simultaneous quantification of lipids in liposome formulations.

Metabolomic profiling of aqueous humor from glaucoma patients - The metabolomics in surgical ophthalmological patients (MISO) study

Glaucoma is still a poorly understood disease with a clear need for new biomarkers to help in diagnosis and potentially offer new therapeutic targets. We aimed to determine if the metabolic profile of aqueous humor (AH) as determined by nuclear magnetic resonance (NMR) spectroscopy allows the distinction between primary open-angle glaucoma patients and control subjects, and to distinguish between high-tension (POAG) and normal-tension glaucoma (NTG). We analysed the AH of patients with POAG, NTG and control subjects (n = 30/group). ¹H NMR spectra were acquired using a 400 MHz spectrometer. Principle component analysis (PCA), machine learning algorithms and descriptive statistics were applied to analyse the metabolic variance between groups, identify the spectral regions, and hereby potential metabolites that can act as biomarkers for glaucoma. According to PCA, fourteen regions of the NMR spectra were significant in explaining the metabolic variance between the glaucoma and control groups, with no differences found between POAG and NTG groups. These regions were further used in building a classifier for separating glaucoma from control patients, which achieved an AUC of 0.93. Peak integration was performed on these regions and a statistical analysis, after false discovery rate correction and adjustment for the different perioperative topical drug regimen, revealed that five of them were significantly different between groups. The glaucoma group showed a higher content in regions typical for betaine and taurine, possibly linked to neuroprotective mechanisms, and also a higher content in regions that are typical for glutamate, which can indicate damaged neurons and oxidative stress. These results show how aqueous humor metabolomics based on NMR spectroscopy can distinguish glaucoma patients from controls with a high accuracy. Further studies are needed to validate these results in order to incorporate them in clinical practice.

Metabolic characterization of tuberculous meningitis in a South African paediatric population using 1H NMR metabolomics

OBJECTIVE

To better characterize the cerebrospinal fluid (CSF) metabolic profile of tuberculous meningitis (TBM) cases using a South African paediatric cohort.

METHODS

¹H NMR metabolomics was used to analyse the CSF of a South African paediatric cohort. Univariate and multivariate statistical analyses were performed to compare a homogeneous control group with a well-defined TBM group.

RESULTS

Twenty metabolites were identified to discriminate TBM cases from controls. As expected, reduced glucose and elevated lactate were the dominating discriminators. A closer investigation of the CSF metabolic profile yielded 18 metabolites of statistical significance. Ten metabolites (acetate, alanine, choline, citrate, creatinine, isoleucine, lysine, myo-inositol, pyruvate and valine) overlapped with two other prior investigations. Eight metabolites (2-hydroxybutyrate, carnitine, creatine, creatine phosphate, glutamate, glutamine, guanidinoacetate and proline) were unique to our paediatric TBM cohort.

CONCLUSIONS

Through strict exclusion criteria, quality control checks and data filtering, eight unique CSF metabolites associated with TBM were identified for the first time and linked to: uncontrolled glucose metabolism, upregulated proline and creatine metabolism, detoxification and disrupted glutamate-glutamine cycle in the TBM samples. Associated with oxidative stress and chronic neuroinflammation, our findings collectively imply destabilization, and hence increased permeability, of the blood-brain barrier in the TBM cases.

Transcriptional Profiling of Patient Isolates Identifies a Novel TOR/Starvation Regulatory Pathway in Cryptococcal Virulence

Human infection with Cryptococcus causes up to a quarter of a million AIDS-related deaths annually and is the most common cause of nonviral meningitis in the United States. As an opportunistic fungal pathogen, Cryptococcus neoformans is distinguished by its ability to adapt to diverse host environments, including plants, amoebae, and mammals. In the present study, comparative transcriptomics of the fungus within human cerebrospinal fluid identified expression profiles representative of low-nutrient adaptive responses. Transcriptomics of fungal isolates from a cohort of HIV/AIDS patients identified high expression levels of an alternative carbon nutrient transporter gene, STL1, to be associated with poor early fungicidal activity, an important clinical prognostic marker. Mouse modeling and pathway analysis demonstrated a role for STL1 in mammalian pathogenesis and revealed that STL1 expression is regulated by a novel multigene regulatory mechanism involving the CAC2 subunit of the chromatin assembly complex 1, CAF-1. In this pathway, the global regulator of virulence gene VAD1 was found to transcriptionally regulate a cryptococcal homolog of a cytosolic protein, Ecm15, in turn required for nuclear transport of the Cac2 protein. Derepression of STL1 by the CAC2-containing CAF-1 complex was mediated by Cac2 and modulated binding and suppression of the STL1 enhancer element. Derepression of STL1 resulted in enhanced survival and growth of the fungus in the presence of low-nutrient, alternative carbon sources, facilitating virulence in mice. This study underscores the utility of ex vivo expression profiling of fungal clinical isolates and provides fundamental genetic understanding of saprophyte adaption to the human host.IMPORTANCE Cryptococcus is a fungal pathogen that kills an estimated quarter of a million individuals yearly and is the most common cause of nonviral meningitis in the United States. The fungus is carried in about 10% of the adult population and, after reactivation, causes disease in a wide variety of immunosuppressed individuals, including the HIV infected and patients receiving transplant conditioning, cancer therapy, or corticosteroid therapy for autoimmune diseases. The fungus is widely carried in the soil but can also cause infections in plants and mammals. However, the mechanisms for this widespread ability to infect a variety of hosts are poorly understood. The present study identified adaptation to low nutrients as a key property that allows the fungus to inhabit these diverse environments. Further studies identified a nutrient transporter gene, STL1, to be upregulated under low nutrients and to be associated with early fungicidal activity, a marker of poor clinical outcome in a cohort of HIV/AIDS patients. Understanding molecular mechanisms involved in adaptation to the human host may help to design better methods of control and treatment of widely dispersed fungal pathogens such as Cryptococcus.

Fungal Diseases in the 21st Century: The Near and Far Horizons

Fungal diseases became a major medical problem in the second half of the 20th century when advances in modern medicine together with the HIV epidemic resulted in large numbers of individuals with impaired immunity. Fungal diseases are difficult to manage because they tend to be chronic, hard to diagnose, and difficult to eradicate with antifungal drugs. This essay considers the future of medical mycology in the 21st century, extrapolating from current trends. In the near horizon, the prevalence of fungal diseases is likely to increase, as there will be more hosts with impaired immunity and drug resistance will inevitably increase after selection by antifungal drug use. We can expect progress in the development of new drugs, diagnostics, vaccines, and immunotherapies. In the far horizon, humanity may face new fungal diseases in association with climate change. Some current associations between chronic diseases and fungal infections could lead to the establishment of fungi as causative agents, which will greatly enhance their medical importance. All trends suggest that the importance of fungal diseases will increase in the 21st century, and enhanced human preparedness for this scourge will require more research investment in this group of infectious diseases.

Nuclear Magnetic Resonance Spectroscopy-Based Identification of Yeast

Rapid and robust high-throughput identification of environmental, industrial, or clinical yeast isolates is important whenever relatively large numbers of samples need to be processed in a cost-efficient way. Nuclear magnetic resonance (NMR) spectroscopy generates complex data based on metabolite profiles, chemical composition and possibly on medium consumption, which can not only be used for the assessment of metabolic pathways but also for accurate identification of yeast down to the subspecies level. Initial results on NMR based yeast identification where comparable with conventional and DNA-based identification. Potential advantages of NMR spectroscopy in mycological laboratories include not only accurate identification but also the potential of automated sample delivery, automated analysis using computer-based methods, rapid turnaround time, high throughput, and low running costs.We describe here the sample preparation, data acquisition and analysis for NMR-based yeast identification. In addition, a roadmap for the development of classification strategies is given that will result in the acquisition of a database and analysis algorithms for yeast identification in different environments.

High-Resolution (1)H NMR Spectroscopy Discriminates Amniotic Fluid of Fetuses with Congenital Diaphragmatic Hernia from Healthy Controls

Lung hypoplasia in congenital diaphragmatic hernia (CDH) is a life-threatening birth defect. Severe cases can be offered tracheal occlusion to boost prenatal lung development, although defining those to benefit remains challenging. Metabonomics of (1)H NMR spectra collected from amniotic fluid (AF) can identify general changes in diseased versus healthy fetuses. AF embodies lung secretions and hence might contain pulmonary next to general markers of disease in CDH fetuses. AF from 81 healthy and 22 CDH fetuses was collected. NMR spectroscopy was performed at 400 MHz to compare AF from fetuses with CDH against controls. Several advanced feature extraction methods based on statistical tests that explore spectral variability, similarity, and dissimilarity were applied and compared. This resulted in the identification of 30 spectral regions, which accounted for 80% variability between CDH and controls. Combination with automated classification discriminates AF from CDH versus healthy fetuses with up to 92% accuracy. Within the identified spectral regions, isoleucine, leucine, valine, pyruvate, GABA, glutamate, glutamine, citrate, creatine, creatinine, taurine, and glucose were the most concentrated metabolites. As the metabolite pattern of AF changes with fetal development, we have excluded metabolites with a high age-related variability and repeated the analysis with 12 spectral regions, which has resulted in similar classification accuracy. From this analysis, it was possible to distinguish between AF from CDH fetuses versus healthy controls independent of gestational age.

A hypothetical astrocyte-microglia lactate shuttle derived from a 1H NMR metabolomics analysis of cerebrospinal fluid from a cohort of South African children with tuberculous meningitis

Tuberculosis meningitis (TBM) is the most severe form of extra-pulmonary tuberculosis and is particularly intense in small children; there is no universally accepted algorithm for the diagnosis and substantiation of TB infection, which can lead to delayed intervention, a high risk factor for morbidity and mortality. In this study a proton magnetic resonance (¹H NMR)-based metabolomics analysis and several chemometric methods were applied to data generated from lumber cerebrospinal fluid (CSF) samples from three experimental groups: (1) South African infants and children with confirmed TBM, (2) non-meningitis South African infants and children as controls, and (3) neurological controls from the Netherlands. A total of 16 NMR-derived CSF metabolites were identified, which clearly differentiated between the controls and TBM cases under investigation. The defining metabolites were the combination of perturbed glucose and highly elevated lactate, common to some other neurological disorders. The remaining 14 metabolites of the host's response to TBM were likewise mainly energy-associated indicators. We subsequently generated a hypothesis expressed as an "astrocyte-microglia lactate shuttle" (AMLS) based on the host's response, which emerged from the NMR-metabolomics information. Activation of microglia, as implied by the AMLS hypothesis, does not, however, present a uniform process and involves intricate interactions and feedback loops between the microglia, astrocytes and neurons that hamper attempts to construct basic and linear cascades of cause and effect; TBM involves a complex integration of the responses from the various cell types present within the CNS, with microglia and the astrocytes as main players.

Viral pathogens in children hospitalized with features of central nervous system infection in a malaria-endemic region of Papua New Guinea

BACKGROUND

Viral central nervous system (CNS) infections are common in countries where malaria is endemic but, due to limited laboratory facilities, few studies have systematically examined the prevalence and clinical consequences of the presence of viruses in cerebrospinal fluid (CSF) from children with suspected CNS infection.

METHODS

We performed a prospective study of Papua New Guinean children hospitalized with signs and symptoms of CNS infection. CSF samples from 300 children without proven bacterial/fungal meningitis were analyzed for human herpes viruses (HHV), picornaviruses, influenza, adenoviruses, flaviviruses and bacteria.

RESULTS

Fifty-five children (18%) had viral (42), bacterial (20) or both viral and bacterial (7) nucleic acids (NA) identified in their CSF. Human herpes viruses accounted for 91% of all viruses found. The identification of viral or bacterial NA was not associated with any characteristic clinical features. By contrast, malaria was associated with increased identification of viral and bacterial NA and with impaired consciousness, multiple convulsions and age. Malaria was also inversely associated with an adverse outcome. Amongst children with HHV infection, those with HHV-6 and -7 were younger, were more likely have impaired consciousness and had a higher proportion of adverse outcomes than children with CMV. Dengue and enteroviral infections were infrequent. Adenoviral and influenza infections were not identified.

CONCLUSION

Infections with HHV-6, HHV-7, dengue and enterovirus have the potential to cause serious CNS disease in young PNG children. However most HHVs in this malaria-endemic setting should be considered to be the result of reactivation from a latent reservoir without clinical sequelae.

Cerebrospinal fluid lactate in post-neurosurgical bacterial meningitis diagnosis

BACKGROUND

Differential diagnosis between post-neurosurgical bacterial meningitis (PNBM) and aseptic meningitis is difficult. Inflammatory and biochemical cerebrospinal fluid (CSF) changes mimic those classically observed after CNS surgery. CSF lactate assay has therefore been proposed as a useful PNBM marker.

OBJECTIVE

To evaluate the diagnostic accuracy of CSF lactate as a PNBM marker in patients hospitalized after a neurosurgical procedure.

METHODS

Between July 2005 and June 2009, a prospective clinical study, in which all patients with clinical suspicion of PNBM were enrolled, was conducted at our neurosurgical Intensive Care Unit. PNBM diagnosis was categorized as proven, probable or negative before the analysis.

RESULTS

Seventy-nine patients, 51 males with a mean age of 50 years (range 32-68 years) were included. Surgery was elective in 76% patients, mostly for brain tumors (57%); thirty PNBM episodes were identified. CSF parameters were significantly different in glucose concentration (27 mg% vs. 73 mg%, p<0.001), lactate (8 mmol/L vs. 2.8 mmol/L, p<0.001), CSF neutrophil pleocytosis (850 mm(-3) vs. 10mm(-3), p<0.001), and protein levels (449 mg% vs. 98 mg%) between the PNBM and non-PNBM groups. The ROC curve that best fits PNBM diagnosis is lactate.

CONCLUSION

Increased CSF lactate is a useful PNBM marker, with better predictive value than CSF hypoglycorrhachia or pleocytosis. Lactate levels ≥ 4 mmol/L showed 97% sensitivity and 78% specificity, with a 97% negative predictive value.

An NMR metabolomics approach for the diagnosis of leptomeningeal carcinomatosis

Leptomeningeal carcinomatosis (LC) is the third most common metastatic complication of the central nervous system. However, the current modalities to reliably diagnose this condition are not satisfactory. Here, we report a preclinical proof of concept for a metabolomics-based diagnostic strategy using a rat LC model incorporating glioma cells that stably express green fluorescent protein. Cytologic diagnoses gave 66.7% sensitivity for the 7-day LC group and 0% for the 3-day LC group. MR imaging could not diagnose LC at these stages. In contrast, nuclear magnetic resonance-based metabolomics on cerebrospinal fluid detected marked differences between the normal and LC groups. Predictions based on the multivariate model provided sensitivity, specificity, and overall accuracy of 88% to 89% in both groups for LC diagnosis. Further statistical analyses identified lactate, acetate, and creatine as specific for the 7-day LC group, with glucose a specific marker of the normal group. Overall, we showed that the metabolomics approach provided both earlier and more accurate diagnostic results than cytology and MR imaging in current use.

Technologies and approaches to elucidate and model the virulence program of salmonella

Salmonella is a primary cause of enteric diseases in a variety of animals. During its evolution into a pathogenic bacterium, Salmonella acquired an elaborate regulatory network that responds to multiple environmental stimuli within host animals and integrates them resulting in fine regulation of the virulence program. The coordinated action by this regulatory network involves numerous virulence regulators, necessitating genome-wide profiling analysis to assess and combine efforts from multiple regulons. In this review we discuss recent high-throughput analytic approaches used to understand the regulatory network of Salmonella that controls virulence processes. Application of high-throughput analyses have generated large amounts of data and necessitated the development of computational approaches for data integration. Therefore, we also cover computer-aided network analyses to infer regulatory networks, and demonstrate how genome-scale data can be used to construct regulatory and metabolic systems models of Salmonella pathogenesis. Genes that are coordinately controlled by multiple virulence regulators under infectious conditions are more likely to be important for pathogenesis. Thus, reconstructing the global regulatory network during infection or, at the very least, under conditions that mimic the host cellular environment not only provides a bird's eye view of Salmonella survival strategy in response to hostile host environments but also serves as an efficient means to identify novel virulence factors that are essential for Salmonella to accomplish systemic infection in the host.

Qualitative metabolome analysis of human cerebrospinal fluid by 13C-/12C-isotope dansylation labeling combined with liquid chromatography Fourier transform ion cyclotron resonance mass spectrometry

Metabolome analysis of human cerebrospinal fluid (CSF) is challenging because of low abundance of metabolites present in a small volume of sample. We describe and apply a sensitive isotope labeling LC-MS technique for qualitative analysis of the CSF metabolome. After a CSF sample is divided into two aliquots, they are labeled by (13)C-dansyl and (12)C-dansyl chloride, respectively. The differentially labeled aliquots are then mixed and subjected to LC-MS using Fourier-transform ion cyclotron resonance mass spectrometry (FTICR MS). Dansylation offers significant improvement in the performance of chromatography separation and detection sensitivity. Moreover, peaks detected in the mass spectra can be readily analyzed for ion pair recognition and database search based on accurate mass and/or retention time information. It is shown that about 14,000 features can be detected in a 25-min LC-FTICR MS run of a dansyl-labeled CSF sample, from which about 500 metabolites can be profiled. Results from four CSF samples are compared to gauge the detectability of metabolites by this method. About 261 metabolites are commonly detected in replicate runs of four samples. In total, 1132 unique metabolite ion pairs are detected and 347 pairs (31%) matched with at least one metabolite in the Human Metabolome Database. We also report a dansylation library of 220 standard compounds and, using this library, about 85 metabolites can be positively identified. Among them, 21 metabolites have never been reported to be associated with CSF. These results illustrate that the dansylation LC-FTICR MS method can be used to analyze the CSF metabolome in a more comprehensive manner.

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<0.05) between groups were seen with the following metabolites: uric acid, hypoxanthine and glutamic acid were higher in plasma from cases, while L-tryptophan and adenosine-5'-diphosphate (ADP) were lower; uric acid and L-histidine were lower in urine from cases. The key limitation of this study is its small size.

CONCLUSIONS/SIGNIFICANCE

Metabolomic analysis clearly distinguished severe pneumonia patients from community controls. The metabolites identified are important for the host response to infection through antioxidant, inflammatory and antimicrobial pathways, and energy metabolism. Larger studies are needed to determine whether these findings are pneumonia-specific and to distinguish organism-specific responses. Metabolomics has considerable potential to improve diagnostics for childhood pneumonia.