key: cord-0010197-gl8uuqej authors: Del Borrello, Giovanni; Stocchero, Matteo; Giordano, Giuseppe; Pirillo, Paola; Zanconato, Stefania; Da Dalt, Liviana; Carraro, Silvia; Esposito, Susanna; Baraldi, Eugenio title: New insights into pediatric community‐acquired pneumonia gained from untargeted metabolomics: A preliminary study date: 2019-12-10 journal: Pediatr Pulmonol DOI: 10.1002/ppul.24602 sha: 2a4ad1bf5652891d9487cf86429d5254c954e32e doc_id: 10197 cord_uid: gl8uuqej BACKGROUND: Available diagnostics often fail to distinguish viral from bacterial causes of pediatric community‐acquired pneumonia (pCAP). Metabolomics, which aims at characterizing diseases based on their metabolic signatures, has been applied to expand pathophysiological understanding of many diseases. In this exploratory study, we used the untargeted metabolomic analysis to shed new light on the etiology of pCAP. METHODS: Liquid chromatography coupled with mass spectrometry was used to quantify the metabolite content of urine samples collected from children hospitalized for CAP of pneumococcal or viral etiology, ascertained using a conservative algorithm combining microbiological and biochemical data. RESULTS: Fifty‐nine children with CAP were enrolled over 16 months. Pneumococcal and viral cases were distinguished by means of a multivariate model based on 93 metabolites, 20 of which were identified and considered as putative biomarkers. Among these, six metabolites belonged to the adrenal steroid synthesis and degradation pathway. CONCLUSIONS: This preliminary study suggests that viral and pneumococcal pneumonia differently affect the systemic metabolome, with a stronger disruption of the adrenal steroid pathway in pneumococcal pneumonia. This finding may lead to the discovery of novel diagnostic biomarkers and bring us closer to personalized therapy for pCAP. Community-acquired pneumonia (CAP) is the single most common cause of death among children worldwide, and a leading cause of hospitalization. 1, 2 Although the body of knowledge of the epidemiology, etiology, microbiology, and pathophysiology of CAP has expanded over the last few decades, basic patient management questions remain largely unanswered. In fact, when confronted with a feverish child with respiratory symptoms, the single most important management question is whether or not to start antibiotics, which translates into how confident is the treating physician that the child in question does not have a bacterial lower respiratory tract infection. 3, 4 Although epidemiological research has repeatedly pointed out that the large majority of lower respiratory infection in pediatric patients are caused by viruses, 2 physicians often lack the tools to reliably discriminate between bacterial and viral etiology [5] [6] [7] and a large percentage of children presenting with respiratory symptoms and fever are ultimately administered antibiotics. 8 The absence of certainty regarding CAP etiology thus prompts an overtreatment with the consequence of increasing the emergence of bacterial resistance. Metabolomics is the newest branch of the systems biology approach to biomedical research. [9] [10] [11] It aims to provide an unabridged description, both qualitative and quantitative, of all the metabolites (ie, molecules with a molecular weight of less than 1200 Da) found in human biofluids and tissues at a given time. In other words, it aims to define particular metabolic signatures, or fingerprints, characterizing a disease state. Powerful analytical tools are used to amass large amounts of data, which are analyzed and applied to previously-unsolvable, complex questions of biology. In particular, metabolomics can provide a more comprehensive overview of a diseaseʼs pathophysiology, identify new biomarkers for use in diagnostics, and point to potential therapeutic targets to consider in the search for new drugs. In the present exploratory investigation, a hypothesis-free approach based on untargeted metabolomics was applied to pediatric CAP (pCAP) in an effort to improve the diagnosis and clinical management of this common childhood infection. More specifically, our study aimed to elucidate whether a specific metabolic signature differentiates pneumococcal from viral pCAP. 40 The study was approved by each centerʼs institutional ethics committee. Written informed consent, signed by both parents, was required for a child to be included in the study; for children aged 8 years or older, the childʼs written informed assent was also required. Cases eligible for this study were hospitalized children less than 14 years of age with a diagnosis of moderate to severe pneumonia. This diagnosis of CAP was based on clinical presentation (presence of fever, symptoms suggestive of an acute respiratory illness defined as new cough or sputum production, chest pain, dyspnea, tachypnea), abnormal lung examination and chest X-ray results. 12, 13 Chest X-rays were evaluated by a blinded radiologist to the study participants. The decision to admit a child to the hospital for CAP was made by the attending physician in the emergency department. To increase the specificity of our findings and reduce the role of confounding variables, three exclusion criteria were strictly applied, concerning: infants (ie, children under 1 year of age), to avoid any diagnostic overlap between pneumonia and bronchiolitis; children with a previous diagnosis of chronic disease (HIV, asthma, immunodeficiency, CHD), to reduce the pathophysiological heterogeneity between CAP cases; and children given any oral or injected antibiotic therapy in the 48 hours preceding enrollment, to avoid cases of partially treated pneumonia, as the related pathophysiological profile differs from that of a lung infection devoid of any treatment. The pharmacological treatment of the recruited subjects was monitored during hospitalization to evaluate potential confounding effects on our findings. A three-step algorithm combining microbiological information obtained from the PCR assays and PCT levels was used to ascertain pneumonia etiology for subsequent metabolomic analysis. 14 Three variables were considered sequentially: presence or absence of respiratory viruses revealed by the multiplex PCR assay; presence or absence of S. pneumoniae on the PCR assay; and PCT cutoffs of 0.25 and 2.00 ng/mL. Using this algorithm, a viral etiology was assumed in the presence of respiratory viruses, the absence of S. pneumoniae, and PCT < 0.25 ng/mL; a pneumococcal etiology was assumed in the absence of respiratory viruses, the presence of S. pneumoniae, and PCT ≥ 2.00 ng/mL. Cases not falling into either category were labeled as "undetermined" and not included in the subsequent analysis. This strategy was applied to obtain "pure" groups for metabolomic investigation, by limiting the number of false-positive cases (ie, S. cases with inconsistent results between PCT measurement and PCR analysis) and were excluded from any further subclass analysis. Table 1 contains the clinical and demographic characteristics of the 27 patients Searching the available online metabolite databases generated a putative identification for 20 of these variables ( Table 2) . pneumoniae pleural infection, 23 probably as a consequence of the aggressive metabolic activity and amino acid biosynthesis induced by S. pneumoniae replication. Cyclic guanosine monophosphate (cGMP), the next significant molecule emerging from our analysis, mediates many of the proinflammatory and anti-inflammatory functions of nitric oxide (NO). 36 Tolllike receptor ligands induce the expression of cGMP by increasing the activity of NO synthetase; in fact, endotoxin infusions in healthy volunteers have been shown to raise both exhaled NO and cGMP plasma concentrations. 37 Bacteria may also increase the synthesis of cGMP by producing NO directly. 38 Our study corroborates these previous reports and provides evidence for the plausibility of using this metabolite as a diagnostic marker of bacterial infection. Another two molecules of interest to this discussion are 3-methylglutaryl-carnitine (an acyl-carnitine), and 3′-N-acetyl-neuraminyl-N-acetyllactosamine (a sialyl oligosaccharide). Carnitine metabolism is tied to mitochondrial homeostasis and has long been studied in the context of severe infections. Septic patients show many abnormalities in lipid metabolism, including a depletion of the cellular levels of L-carnitine, a reduction in its plasma levels, and an increase in its urinary excretion. 39 Three recent metabolomic studies also found an increase in the urinary concentration of acyl-carnitines in patients with bacterial pneumonia. 20, 25, 40 Indeed, the pneumococcal cytotoxin pneumolysin causes mitochondrial damage, which may account for the observed disarray in lipid metabolism. 41, 42 Sialyl oligosaccharides are broken down and digested by lysosomes, and a relative increase in sialyl oligosaccharides in the course of bacterial infections may represent a transient lysosomal dysfunction. Indeed, the role of lysosomes in human health and disease is just starting to be unveiled, but evidence is already accumulating of lysosomes serving as regulators of macrophage function, 43 and as key effectors of S. pneumoniae intracellular killing. 44 A recent meta-analysis of transcriptomic studies also showed that messenger RNA pathway associated with lysosomal function were among the most profoundly disrupted in the course of human sepsis. 45 In short, our study points to organelle dysfunction as a crucial discriminator between bacterial and viral infection. Viral pneumonia is associated with an increase in urine concentrations of glycyl-L-hydroxy-proline (gly-pro), an end product of collagen metabolism and the substrate of the enzyme prolidase, which promotes the surface expression of the interferon-I receptor and is a target of viral antagonism. 46, 47 The higher urinary levels of gly-pro found in our study may, therefore, reflect an impaired prolidase activity, which could be expected in the course of viral infection. Although state-of-the-art, high-throughput analytical methods and chemometric data processing methods were used in our study, it has some weaknesses. The most important concern the fact that we did not collect samples in therapy-naive patients, and this constitutes a veritable source of bias. This issue could have been avoided by collecting urine samples from all therapy-naive children presenting to the emergency department with respiratory symptoms and a fever, and then analyzing only those obtained from children actually enrolled in the study. This will be worth bearing in mind for future metabolomic studies conducted in acute care settings. Another weakness of our study concerns the small number of patients in our two groups, which may limit the validity of our findings. On the other hand, our adoption of stringent enrollment criteria and a conservative diagnostic algorithm ensured that the two groups were etiologically pure, thus enhancing the specificity of our findings. A further potential limitation of this exploratory study lies in the lack of any external validation set against which to test our models, which could only be validated internally by means of a cross-validation and stability selection to ensure the reliability of the findings emerging from the data collected. Though ours is only a preliminary study, the present findings seem to be a promising starting point for the conduction of larger, validated studies. In Global, regional, and national causes of child mortality in 2000-13, with projections to inform post-2015 priorities: an updated systematic analysis Community-acquired pneumonia requiring hospitalization among U.S. children A systematic review on the diagnosis of pediatric bacterial pneumonia: when gold is bronze Antibiotic therapy for pediatric community-acquired pneumonia: do we know when, what and for how long to treat? Does this child have pneumonia?: the rational clinical examination systematic review Specimen collection for the diagnosis of pediatric pneumonia Laboratory methods for determining pneumonia etiology in children Antibiotic prescribing in ambulatory pediatrics in the United States Opinion: understanding "global" systems biology: metabonomics and the continuum of metabolism Metabolomics: a new frontier for research in pediatrics Urinary metabolomics in newborns and infants The management of community-acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America British Thoracic Society guidelines for the management of community acquired pneumonia in children: update A practical approach estimating etiologic agents using real-time PCR in pediatric inpatients with community-acquired pneumonia Metabolomic profile of children with recurrent respiratory infections Wheezing preschool children with early-onset asthma reveal a specific metabolomic profile Metabolomic profile of low-copy number carriers at the salivary α-amylase gene suggests a metabolic shift toward lipid-based energy production What was old is new again: using the host response to diagnose infectious disease Systematic review regarding metabolic profiling for improved pathophysiological understanding of disease and outcome prediction in respiratory infections Pneumococcal pneumonia: potential for diagnosis through a urinary metabolic profile Plasma metabolomics for the diagnosis and prognosis of H1N1 influenza pneumonia Metabolomic analysis in severe childhood pneumonia in the Gambia, West Africa: findings from a pilot study Metabolomic profiling of infectious parapneumonic effusions reveals biomarkers for guiding management of children with streptococcus pneumoniae pneumonia Emerging biomarkers of illness severity: urinary metabolites associated with sepsis and necrotizing methicillin-resistant Staphylococcus aureus pneumonia. Pharmacotherapy Urine metabolomic profiling of children with respiratory tract infections in the emergency department: a pilot study The effects of ACTH on steroid metabolomic profiles in human adrenal cells Liquid chromatography-tandem mass spectrometry analysis of human adrenal vein 19-carbon steroids before and after ACTH stimulation Regulation of 18-oxocortisol and 18-hydroxycortisol by the renin-angiotensin system and ACTH in man The role of adrenal gland microenvironment in the HPA axis function and dysfunction during sepsis Reduced cortisol metabolism during critical illness Cortisol levels and adrenal response in severe community-acquired pneumonia: a systematic review of the literature Association of adrenal function and disease severity in community-acquired pneumonia Adrenal function is related to prognosis in moderate community-acquired pneumonia Altered balance of the aminogram in patients with sepsis-the relation to mortality The uraemic toxin phenylacetic acid contributes to inflammation by priming polymorphonuclear leucocytes Nitric oxide and superoxide in inflammation and immune regulation Exhaled NO and plasma cGMP increase after endotoxin infusion in healthy volunteers Neisseria meningitidis induces platelet inhibition and increases vascular endothelial permeability via nitric oxide regulated pathways Plasma carnitine levels and urinary carnitine excretion during sepsis Lipid mediators of inflammation as novel plasma biomarkers to identify patients with bacteremia Virulence factors in pneumococcal respiratory pathogenesis Pneumolysin causes neuronal cell death through mitochondrial damage Activator of G-protein signaling 3-induced lysosomal biogenesis limits macrophage intracellular bacterial infection Streptococcus pneumoniae invades endothelial host cells via multiple pathways and is killed in a lysosome dependent manner Lysosome and cytoskeleton pathways are robustly enriched in the blood of septic patients: a meta-analysis of transcriptomic data Flavivirus antagonism of type I interferon signaling reveals prolidase as a regulator of IFNAR1 surface expression SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section. How to cite this article New insights into pediatric communityacquired pneumonia gained from untargeted metabolomics: A preliminary study The authors declare that there are no conflict of interests. http://orcid.org/0000-0002-1829-3652