Serum-integrated omics reveal the host response landscape for severe pediatric community-acquired pneumonia

Abstract

Objective: Community-acquired pneumonia (CAP) is the primary cause of death for children under five years of age globally. Hence, it is essential to investigate new early biomarkers and potential mechanisms involved in disease severity.

Methods: Proteomics combined with metabolomics was performed to identify biomarkers suitable for early diagnosis of severe CAP. In the training cohort, proteomics and metabolomics were performed on serum samples obtained from 20 severe CAPs (S-CAPs), 15 non-severe CAPs (NS-CAPs) and 15 healthy controls (CONs). In the verification cohort, selected biomarkers and their combinations were validated using ELISA and metabolomics in an independent cohort of 129 subjects. Finally, a combined proteomics and metabolomics analysis was performed to understand the major pathological features and reasons for severity of CAP.

Results: The proteomic and metabolic signature was markedly different between S-CAPs, NS-CAPs and CONs. A new serum biomarker panel including 2 proteins [C-reactive protein (CRP), lipopolysaccharide (LBP)] and 3 metabolites [Fasciculol C, PE (14:0/16:1(19Z)), PS (20:0/22:6(4Z, 7Z, 10Z, 13Z, 16Z, 19Z))] was developed to identify CAP and to distinguish severe pneumonia. Pathway analysis of changes revealed activation of the cell death pathway, a dysregulated complement system, coagulation cascade and platelet function, and the inflammatory responses as contributors to tissue damage in children with CAP. Additionally, activation of glycolysis and higher levels of nucleotides led to imbalanced deoxyribonucleotide pools contributing to the development of severe CAP. Finally, dysregulated lipid metabolism was also identified as a potential pathological mechanism for severe progression of CAP.

Conclusion: The integrated analysis of the proteome and metabolome might open up new ways in diagnosing and uncovering the complexity of severity of CAP.

Keywords: Community-acquired pneumonia; Diagnosis; Host response; Metabolomics; Proteomics.

Fig. 1

Study overview. A Study overview. 50 subjects including 20 S-CAPs, 15 NS-CAPs and 15 CONs from cohort 1 were recruited for proteomic and metabolomic analysis. Nine DAPs were verified with ELSIA in cohort 1. The DAPs and DAMs were then used to identify potential biomarker combinations for severe CAP diagnosis. B Selected biomarkers were verified using an independent cohort with 129 blinded subjects (cohort 2). C Protein-metabolite crosstalk was examined using integrated analysis. Proteomic and metabolomic signatures were analyzed to uncover the molecular profile for severe CAP

Fig. 2

Identification of differentially abundant proteins in S-CAPs from cohort 1. A PLS-DA score plots for S-CAPs, NS-CAPs and CONs. B Venn diagram of the number of DAPs. C GO enrichment analysis for all DAPs with the top 20 GO terms shown. Yellow cycle highlights acute-phase response; Red cycle highlights platelet dysfunction; Orange cycle highlights immune response; Purple cycle highlights metabolic processes. D KEGG analysis for all DAPs with the top 13 KEGG terms shown. Red cycle highlights platelet dysfunction; Green cycle highlights cell death. E The interaction network for proteins involved in the cell death, inflammatory response, immune response, platelet dysfunction and metabolic pathways. Green squares represent pathways; purple circles represent the altered proteins; solid lines represent association between the pathways and proteins

Fig. 3

Identification of Differentially Abundant Metabolites in S-CAP from Cohort 1. A Venn diagram of the number of DAPs. B SMPDB analysis of the DAMs from cluster 1. Top 25 KEGG terms are shown. C PLS-DA score plots for NS-CAPs and CONs. (D) PLS-DA score plots for S-CAPs and CONs. E PLS-DA score plots for S-CAPs and NS-CAPs. F Parameters for assessing the quality of the PLS-DA model for NS-CAPs and CONs. G Parameters for assessing the quality of the PLS-DA model for S-CAPs and CONs. H Parameters for assessing the quality of the PLS-DA model for S-CAPs and CONs

Fig. 4

Identification and verification of potential biomarkers for classification of S-CAPs. A Classification and regression tree analysis using 2 DAPs and 3 DAMs with 6 terminal nodes. The selected splitting variables are shown in the nodes. B AUC values for 5 biomarkers and the combined panel were calculated to differentiate NS-CAPs from CONs in cohort 2. C AUC values for 5 biomarkers and the combined panel were calculated to differentiate S-CAPs from CONs in cohort 2. D AUC values for 5 biomarkers and the combined panel were calculated to differentiate S-CAPs from NS-CAPs in cohort 2

Fig. 5

Activated death system, dysregulated complement system and platelet function in S-CAP cases. A Hierarchical clustering illustrating four DAP patterns across three groups. The red line is the center line of the trend for each gene cluster. B KEGG terms enriched in cluster 1 and cluster 4. Top 20 KEGG terms are shown. Red lines highlight cell death-related pathways. C KEGG terms enriched in cluster 2. Top 20 KEGG terms are shown. Red lines highlight complement and coagulation cascade pathways. D KEGG terms enriched in cluster 3. Top 20 KEGG terms are shown. Red lines highlight platelet activation pathway

Fig. 6

Cross-talk between glucose metabolism and nucleotide metabolism implicated in progression to severe disease in CAP. A Glycolysis and B Pentose phosphate pathway (PPP) were activated during the initial onset of CAP and progression toward severe disease. Most enzymes involved in glycolysis were significantly upregulated in CAPs. C Circulating levels of TCA metabolites in serum. Increased proteins and metabolites are labeled in red. Decreased proteins and metabolites were labeled in blue. Statistical significance was determined using the FDR-adjusted p-value. *p < 0.05; **p < 0.01; ***p < 0.00

Fig. 7

Dysregulated lipid metabolism in CAP. A Heatmap showing expression levels of apolipoproteins in CONs, S-CAPs, and NS-CAPs. B Representative apolipoprotein expression changes across 3 groups. Square and bars represent the mean and standard deviation, respectively. Statistical significance was determined using the FDR-adjusted p-value. *p < 0.05; **p < 0.01; ***p < 0.001. C The interaction network for apolipoproteins and proteins associated with macrophage function. D Correlation analysis of inflammatory-associated proteins and apolipoproteins. Red and blue numbers represent positive and negative correlation, respectively. * means correlation p value < 0.05. ** means correlation p value < 0.01. E Heatmap of DAMs that are associated with fatty acyls, glycerolipids, glycerophospholipids, prenol lipids, sphingolipids, steroid and steroid derivatives