Integrated multiomics reveals a clinically relevant CXCL8-centric protein network in pediatric postinfectious bronchiolitis obliterans

Abstract

Background

Pediatric postinfectious bronchiolitis obliterans (PIBO) is a long-term sequela of severe lower respiratory tract infection. However, the underlying pathogenic mechanisms within the airways remain poorly understood.

Methods

Bronchoalveolar lavage fluid (BALF) from six children with PIBO and three control children was subjected to transcriptomic, proteomics and untargeted metabolomics analyses. Integrated analysis identified key differentially expressed candidates, among which protein candidates were subsequently validated by ELISAs.

Results

Transcriptomic analysis revealed altered pathways in the BALF of children with PIBO, including pathways related to epithelial differentiation, inflammatory responses, and metabolic processes involving amino acids, purines, lipids, and vitamins. Proteomic analysis further revealed epithelial injury, inflammatory activation, tissue remodeling, and metabolic dysregulation in PIBO patients. Protein–protein interaction (PPI) network analysis identified CXCL8 (interleukin-8, IL-8) as the central hub. Integrative transcriptomic-proteomic correlation analysis further revealed that CXCL8 transcript levels were correlated with the greatest number of differentially expressed proteins (DEPs). Subsequent ELISA validation confirmed elevated CXCL8 protein levels in children with PIBO, which were positively correlated with the predicted FEV1%. Furthermore, 54 proteins, including KRT19, OSM, ADAM8, and PADI2, were significantly correlated with CXCL8 expression. These proteins spanned epithelial, inflammatory, remodeling, and metabolic processes, indicating a clinically relevant network. Untargeted metabolomics analysis revealed alterations in purine, cofactors, and amino acid metabolism, with pyridoxine (vitamin B6) as the most significantly altered metabolite. Integrated multiomic analysis revealed widespread correlations between differentially expressed metabolites and proteins. Notably, putative pyridoxine expression was negatively correlated with the expression of CXCL8 and other key inflammatory proteins, which is consistent with the evidence of disrupted vitamin metabolism observed by transcriptomics. In vitro experiments further showed that pyridoxine treatment reduced CXCL8 secretion in small airway epithelial cells. Together, these findings suggest a potential link between vitamin B6 and inflammatory processes in PIBO.

Conclusions

Multiomic integration revealed molecular signatures characterized by a CXCL8-centric protein network, a potential disruption of vitamin B6 homeostasis, and altered pathways involved in epithelial differentiation, immunity, and metabolism. These signatures converge on core pathogenic mechanisms within the airway microenvironment, including epithelial injury, inflammatory activation, tissue remodeling, and metabolic dysregulation. Collectively, these findings establish an integrative framework for understanding the pathogenesis of PIBO and highlight potential molecular targets for biomarker discovery and therapeutic intervention.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12931-026-03882-1.

Background

Postinfectious bronchiolitis obliterans (PIBO) is a common chronic obstructive airway disease in children that typically arises secondary to severe lower respiratory tract infections, especially adenovirus pneumonia [1]. Although the overall incidence is low, it can reach 28% among children with severe adenovirus pneumonia and as high as 50% in critically ill children requiring mechanical ventilation [2, 3]. In contrast, PIBO is distinctly rare in adults, whose cases are more frequently associated with conditions such as transplantation, inhalational injuries, or autoimmune disorders rather than as a postinfectious sequelae. PIBO predominantly affects infants and young children during their rapid developmental stage, resulting in persistent airflow limitation, protracted cough, wheezing, and exercise intolerance [4]. Some patients require long-term home oxygen therapy, which severely impairs quality of life and growth, imposing a substantial disease burden on pediatric health care systems.

The pathological hallmark of PIBO is fibrotic obliteration of the bronchioles. Disease is initiated by severe airway epithelial injury caused by pathogens [5]. This injury disrupts epithelial barrier integrity and triggers a sustained imbalance between inflammation and repair [6]. In this chronic inflammatory microenvironment, fibroblasts become aberrantly activated, leading to excessive extracellular matrix (ECM) deposition and eventual fibrotic obliteration of the airways. However, the precise pathogenic mechanisms that link initial epithelial injury to chronic inflammation and ultimately to irreversible fibrosis remain poorly understood, hindering the discovery of clinically relevant biomarkers and the development of targeted therapies [7].

Bronchoalveolar lavage fluid (BALF) directly reflects the immune and metabolic state of the airway microenvironment, making it an ideal specimen for investigating PIBO pathogenesis [8, 9]. Given that PIBO predominantly affects infants and young children, lung biopsy is rarely performed due to its highly invasive nature and associated ethical and clinical risks. BALF represents a more feasible and clinically accessible alternative for molecular profiling in this vulnerable population. The cell pellet enables transcriptomic analysis of resident immune cells and sloughed structural cells (e.g., epithelial cells and fibroblasts), while the cell-free supernatant permits proteomic and metabolomic profiling of key protein effectors and metabolites. Conventional single-omics approaches suffer from low throughput and insufficient sensitivity, especially when they are applied to small-volume, low-abundance BALF samples [10]. In contrast, Olink ultrasensitive proteomics and untargeted metabolomics enable deep profiling from minimal input, facilitating integrated multiomic analysis [11, 12].

Here, we performed an integrated multiomic analysis of the same BALF samples from children with PIBO, including cellular transcriptomics, Olink proteomics, and untargeted metabolomics of the supernatant. By integrating gene expression, protein abundance, and metabolic profiles, we first identified a set of molecular features, including a CXCL8-centric protein interaction network, potential disrupted vitamin B6 homeostasis, and perturbed pathways involved in epithelial differentiation, immune responses, and metabolism. These features collectively underpin the core pathogenic mechanisms of PIBO, including epithelial injury, immune activation, tissue remodeling, and metabolic dysregulation. Together, our findings provide multilevel evidence to advance the understanding of the pathogenesis of PIBO, facilitate the identification of candidate biomarkers, and reveal potential therapeutic targets.

Methods

Participants and sample collection

Children diagnosed with PIBO at Beijing Children's Hospital, Capital Medical University, in July 2025 were enrolled in the PIBO group. The diagnostic criteria for PIBO followed the Expert Consensus on Diagnosis and Treatment of Bronchiolitis Obliterans in Children (2023) and included: (1) a history of severe lower respiratory tract infection; (2) persistent or recurrent airway obstructive symptoms and signs persisting for more than six weeks, with poor response to bronchodilator therapy; (3) pulmonary function tests showing obstructive or mixed ventilatory dysfunction; (4) high-resolution chest CT (HRCT) demonstrating mosaic perfusion, bronchial wall thickening, air trapping, or other characteristic findings; and (5) exclusion of other chronic lung diseases.

The control group comprised children who underwent bronchoscopy during the same period either for foreign body aspiration (without a history of severe respiratory tract infection) or for evaluation of congenital pulmonary airway malformation (CPAM) without a history of severe lower respiratory tract infection. CPAM was diagnosed according to the National Expert Consensus on Diagnosis and Treatment of Congenital Pulmonary Airway Malformations in China (2021).

Inclusion criteria for both groups were: (1) age ≤ 18 years; (2) meeting the respective diagnostic criteria as described above; and (3) undergoing bronchoscopy with BALF collection. Exclusion criteria for both groups were: (1) primary ciliary dyskinesia, bronchial asthma, or other chronic lung diseases; (2) congenital heart disease or immunodeficiency; and (3) incomplete clinical data or guardian refusal to participate.

All the subjects underwent bronchoscopy under local anesthesia, and BALF was collected using conventional methods. The samples were immediately centrifuged at 1500 rpm for 10 min at 4 °C. The cell pellet and supernatant were then collected separately and stored at − 80 °C for subsequent analysis.

Transcriptome sequencing

Total RNA was extracted from BALF cell pellets using TRIzol® reagent according to the manufacturer's instructions. RNA integrity was assessed using a 5300 Bioanalyzer (Agilent), and RNA concentration was measured with an ND-2000 spectrophotometer (NanoDrop Technologies). Only high-quality RNA samples (RQN > 4.5) were used for library construction. RNA purification, reverse transcription, library preparation, and sequencing were performed at Shanghai Majorbio Biopharm Biotechnology Co., Ltd. (Shanghai, China) following the manufacturer's protocols (Illumina, San Diego, CA, USA).

The raw paired-end reads were trimmed and quality-controlled by fastp with default parameters. Then clean reads were separately aligned to the reference genome with orientation mode using HISAT2 software. To quantify gene expression levels, RSEM was used to calculate transcript abundances based on the raw read counts (the expression level of each transcript was estimated according to the transcripts per million reads, TPM, method). Differential expression analysis between the two groups was performed using the DESeq2 package (version 1.53.1) in R. Genes with an absolute log₂fold change (|log₂FC|) ≥ 1 and adjust P < 0.05 were considered significantly differentially expressed. Gene Ontology (GO) enrichment analysis was conducted using the goatools Python package. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed using a custom Python script based on the SciPy library (hypergeometric test). Visualization of enrichment results was performed using the R package ggplot2.

Proteomic analysis of BALF supernatants

BALF supernatants from children with PIBO and healthy controls were collected for protein detection, with 50 µL of sample used per subject. Protein profiling was performed using the Olink Reveal platform (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. This platform is based on proximity extension assay (PEA) technology coupled with next-generation sequencing for signal readout, enabling the simultaneous detection of approximately 1,000 proteins.

Briefly, oligonucleotide-labeled antibody pairs specific to each target protein bind to the corresponding proteins in the sample, allowing the adjacent oligonucleotides to hybridize and extend, thereby generating a unique protein-specific “barcode”. After library preparation, sequencing was carried out on an Illumina® NovaSeq™ 6000 system. The raw data underwent quality control and normalization and were subsequently converted to Olink’s proprietary relative abundance unit (NPX value). Detailed assay performance validation data are available on the manufacturer’s website (www.olink.com).

Differentially expressed proteins between the PIBO and control groups were identified using Student’s t test, with a significance threshold of P < 0.05 and a fold change cutoff of FC ≥ 1.2 or FC ≤ 1/1.2. Subsequent visualization of GO enrichment, KEGG pathway, and Reactome pathway analyses was performed using the OmicStudio tools at https://www.omicstudio.cn/tool.

Metabolomic analysis of BALF supernatants

BALF supernatants from children with PIBO and healthy controls were collected for untargeted metabolomics analysis using liquid chromatography‒mass spectrometry (LC‒MS), with 150 µL of sample used per subject. After sample pretreatment, analysis was performed on a Thermo Vanquish Flex ultra-performance liquid chromatography (UPLC) system coupled with a Thermo Q Exactive Plus mass spectrometer. Peak detection, alignment, and quality control were carried out using XCMS software, followed by metabolite identification with metaX. The identified metabolites were then annotated by matching against public databases such as the KEGG. Differentially abundant metabolites were screened using a t test with the criteria FC ≥ 1.2 or FC ≤ 1/1.2 and P < 0.05. Subsequently, quantitative analysis of metabolites, sample correlation analysis, and differential analysis were performed. For the identified differentially abundant metabolites, further analyses included KEGG functional enrichment analysis, interaction network analysis, and metabolite‒metabolite correlation analysis.

Enzyme-linked immunosorbent assays (ELISAs)

To further validate the proteomics findings, the concentrations of CXCL8 and MMP9 in the BALF supernatants were measured using ELISAs. CXCL8 was detected using a Human IL-8 Uncoated ELISA kit (Invitrogen, cat. no. 88–8086), and MMP9 was detected using a Human Total MMP-9 ELISA kit (USCN, cat. no. SEA553HU). All procedures were performed strictly according to the manufacturers’ instructions. The absorbance was measured at 450 nm using a microplate reader, and sample concentrations were calculated on the basis of standard curves. Correlation analyses were conducted to assess the relationships between CXCL8 levels and lung function parameters (predicted FEV1%), as well as the correlations between CXCL8 levels and MMP9 levels.

Primary human small airway epithelial cells culture and treatment

Human small airway epithelial cells were obtained from ScienCell and cultured in PneumaCult™-Ex Plus Medium (STEMCELL Technologies) at 37 °C in a humidified 5% CO₂ incubator. The culture medium was refreshed every 2–3 days. Upon reaching approximately 80% confluence, cells were treated with pyridoxine (MCE, cat. no. HY-B1328) at a final concentration of 20 µM or an equal volume of DMSO for 24 h. After treatment, cell culture supernatants were collected and centrifuged at 3000 × g for 10 min to remove debris. CXCL8 levels in the supernatants were measured by ELISA.

Statistical analysis

To determine the appropriate correlation methods, we first assessed the distribution characteristics of the proteomics and metabolomics data using the Shapiro–Wilk normality test. The Olink proteomics data satisfied the normality assumption; hence, Pearson correlation analysis was used for protein–protein correlations. Conversely, the metabolomics data deviated from normality, so Spearman’s rank correlation was applied for metabolomics-internal and metabolomics-proteomics cross-omics correlations.

Based on these criteria, we calculated the Pearson correlation coefficients between CXCL8 and all quantified proteins, and proteins with |r|> 0.7 and P < 0.05 were considered significantly correlated with CXCL8. For integrative multi-omics analysis, Spearman’s rank correlation was used to construct correlation networks between differentially abundant metabolites and differentially expressed proteins, with |r|> 0.6 and P < 0.05 as the threshold for statistical significance.

All the statistical analyses were performed using R software or GraphPad Prism. Comparisons between two groups were conducted using Student’s t test. A P value < 0.05 was considered to indicate statistical significance.

Results

Transcriptomic landscape of the local airway microenvironment in PIBO patients

To explore the transcriptomic landscape of the local airway microenvironment in PIBO patients, we collected BALF cells from six children with PIBO and three control children who underwent bronchoscopy at the Respiratory Department of Beijing Children’s Hospital (July 2025). The inclusion criteria and clinical characteristics are detailed in the Methods section, and baseline information of the participants is summarized in Supplementary Table 1. Principal component analysis (PCA) revealed a clear separation in the transcriptomic landscape between the PIBO and control groups. Together, PC1 and PC2 explained 75.96% of the total variance, indicating a substantial difference in transcription between the two groups despite the limited sample size (Fig. 1A). Volcano plot analysis revealed differentially expressed genes (DEGs) in the PIBO group compared with those in the control group. A total of 6,465 significant DEGs were identified (|log₂FC|> 1, Padj < 0.05), of which 5,330 were upregulated and 1,135 were downregulated (Fig. 1B), suggesting a broadly activated cellular state in the local airway of PIBO patients. The acute-phase protein genes SAA1 and SAA2 were markedly upregulated in the PIBO group, indicating a robust local inflammatory response. In addition, inflammation-related genes (e.g., CX3CL1 and CXCL8) and epithelial-specific genes (e.g., FOXJ1 and KRT19) were significantly upregulated, suggesting that epithelial injury and repair processes were ongoing.

Fig. 1.

Transcriptomic profiling of BALF cells reveals epithelial injury, immune activation, and metabolic pathway alterations in PIBO patients. A PCA of BALF transcriptomes. The PCA plot shows a clear separation between the PIBO group and the control group at the transcriptome level. B Volcano plot of DEGs in the BALF transcriptome. DEGs were identified using thresholds of |log₂FC|> 1 and adjusted P < 0.05. Inflammation-associated genes such as SAA1 and SAA2 were significantly upregulated in the PIBO group. C Functional enrichment analysis of DEGs based on the GO annotation and KEGG/Reactome pathway databases. The DEGs were significantly enriched in pathways related to epithelial differentiation, NETs, IL-17 signaling, and the ECM–receptor interaction. D Heatmap showing the expression patterns of representative DEGs associated with epithelial injury, inflammation, and tissue remodeling in the two groups

Functional enrichment analysis of the DEGs revealed coordinated activation of multiple pathways in the PIBO airway microenvironment (Fig. 1C). The enriched biological processes included (1) cilium-related signaling pathways, suggesting epithelial denudation and ciliary dysfunction; (2) neutrophil extracellular trap (NET) formation, indicating neutrophil-mediated immunopathology; (3) IL-17 signaling pathway activation, indicating a potential role for Th17-type immune responses; (4) tissue remodeling and fibrosis-associated pathways, including ECM and ECM-receptor interactions, indicating aberrant repair; and (5) metabolic pathways, such as vitamin B6, lipid and amino acid metabolism, suggesting global alterations in cellular energy metabolism and biosynthesis under disease conditions. Such metabolic dysregulation may further exacerbate inflammatory signals, modulate immune cell function and promote aberrant tissue repair. A heatmap of representative DEGs further supported these findings (Fig. 1D), showing coordinated upregulation of epithelial markers (e.g., KRT19 and TP63) and chemokines (e.g., CX3CL1 and CXCL8) in the PIBO group, which suggests that epithelial injury and local inflammation may together drive disease progression within the BALF microenvironment. However, the transcriptomic signals demonstrated some degree of inter-sample heterogeneity, which may relate to variations in the proportion of shed structural cells, such as epithelial cells and fibroblasts, in the BALF pellets.

Proteomics reveals a CXCL8-centered pathological network in PIBO patients

Transcriptomic analysis of BALF cells provided key insights into cellular changes associated with the progression of PIBO. To characterize the functional protein network potentially involved in PIBO pathogenesis, we performed Olink targeted proteomics on BALF supernatants and profiled molecular alterations in the local airway microenvironment.

As shown in Fig. 2A, PCA of the proteomic data revealed that PC1 and PC2 explained 61.02% and 24.24% of the total variance, respectively, indicating distinct proteomic distribution profiles between the PIBO and control groups (Fig. 2A). Volcano plots revealed significantly differentially expressed proteins between the PIBO and control groups (Fig. 2B). A total of 76 upregulated and 7 downregulated proteins were identified (FC ≥ 1.2 or ≤ 1/1.2, P < 0.05). To further explore the functional associations among the significantly differentially expressed proteins, a protein‒protein interaction (PPI) network was constructed on the basis of the STRING database. CXCL8 was identified as a central hub in the protein‒protein interaction network; it exhibited the highest node degree (degree = 10) and directly interacted with 10 proteins, including MMP13, TLR2, NOS2, ARG1, and RIGI. Among these genes, CXCL8 had the highest interaction score with MMP13 and NOS2 (score ≥ 0.9), indicating a functionally closely linked regulatory module (Fig. 2C). To further explore the co-expression relationships among the Olink proteomics data, we calculated pairwise Spearman correlation coefficients across all detected proteins and constructed a co-expression network using the top 100 significant correlation pairs (Fig. 2D). Notably, CXCL8 again occupied a central hub position in this network. This result corroborates the STRING-based PPI network analysis, jointly supporting the central role of CXCL8 in the PIBO patients. Integrative analysis of differentially expressed proteins and transcriptomic data revealed four association patterns: concordant regulation (both transcript and protein upregulation or downregulation), as well as discordant patterns (transcript upregulation/protein downregulation, and transcript downregulation/protein upregulation) (Fig. 2E). Notably, the CXCL8 transcript correlated with the greatest number of differentially expressed proteins (Supplementary Fig. 1), indicating that its transcriptional change most consistently aligned with widespread protein abundance alterations. These findings further establish CXCL8 as a core hub gene, as evidenced by transcript‒protein covariation. Consistently, CXCL8 has previously been proposed as a potential target in PIBO [9]. On the basis of these lines of evidence, CXCL8 was selected as the primary candidate for further validation, along with the classic PIBO marker MMP9, and an ELISA was performed on BALF samples from an independent cohort at our center that included 13 children with PIBO and 8 children in the control group. The results revealed that the CXCL8 protein level was significantly elevated in the PIBO group and, in the 12 patients with available lung function data, was negatively correlated with lung function, confirming its association with the clinical phenotype of PIBO. A significant positive correlation was also observed between CXCL8 and MMP9 protein expression, suggesting their potential synergistic involvement in matrix remodeling in PIBO (Fig. 2F). To expand the CXCL8-centric molecular network, a full correlation analysis was conducted on proteins that were significantly associated with CXCL8 expression in the Olink proteomic data, yielding 54 proteins with significant correlations (/r/> 0.7, P < 0.05) (Fig. 2G). High expression of these key differentially expressed proteins collectively delineated a pathogenic cascade from epithelial injury to irreversible fibrosis in PIBO. Elevated levels of the epithelial injury marker KRT19 indicated airway epithelial barrier disruption. The upregulation of CXCL8 and OSM suggested innate immune involvement. The marked upregulation of TNFSF14 and TNFSF19 expression was associated with adaptive immune regulation. Furthermore, increased levels of MMP12, ADAM8, and ADAM15 were observed, indicating airway remodeling. In addition, elevated expression of ARG1 and PADI2 indicated metabolic dysregulation in patients with PIBO. GO and KEGG enrichment analyses revealed significant enrichment of genes related to amino acid metabolism, ECM remodeling, and immune signaling pathway dysregulation in the PIBO airway microenvironment (Fig. 2H). This Olink proteomics study revealed a CXCL8-centered pathological network in PIBO, encompassing epithelial injury, inflammatory activation, matrix remodeling and metabolic dysregulation. The significant negative correlation between the CXCL8 protein level and lung function indicates that this network may contribute to lung function decline in PIBO patients, underscoring its potential clinical relevance.

Fig. 2.

Proteomic profiling of BALF supernatant identifies CXCL8 as a central hub within the PIBO protein interaction network. A PCA plot showing distinct separation between the PIBO and control groups at the protein level. B Volcano plot of differentially expressed proteins (FC ≥ 1.2 or FC ≤ 1/1.2, P < 0.05). The upregulated proteins in the PIBO group included KRT19, MMP12, CXCL8, CEACAM8, ARG1, and others. C PPI network of the core differentially expressed proteins. Nodes represent proteins, with node size and color intensity reflecting degree. Edges represent interactions; thicker lines indicate higher interaction scores. D Protein co-expression network based on the top 100 significant Spearman correlation pairs from Olink proteomics data. E Integrative proteomic–transcriptomic four-quadrant plot. The scatter plot shows log₂ fold changes of RNA (x-axis) vs. protein (y-axis) for differentially expressed genes. Genes are stratified into four quadrants on the basis of the direction of change. F ELISA validation of CXCL8 expression. Up: CXCL8 levels are significantly elevated in PIBO patients, n = 21; Middle: CXCL8 levels correlate negatively with the predicted FEV1%, n = 12; Down: MMP9 expression is positively correlated with that of CXCL8,n = 21. G Proteomics correlation network of CXCL8. An interaction network was constructed from 54 proteins significantly correlated with CXCL8 (|r|> 0.7, P < 0.05). H Functional enrichment analysis of differentially expressed proteins. The significantly enriched pathways included response to bacteria, response to virus, neutrophil activation and degranulation, ECM organization, and amino acid metabolic pathways

Metabolomics reveals amino acid metabolic dysregulation and vitamin B6 downregulation in the PIBO airway microenvironment

To complement the transcriptomic and proteomic evidence of lipid and amino acid metabolic abnormalities, we performed an untargeted metabolomic analysis of the BALF supernatants. PCA revealed a clear separation between the PIBO and control groups, which was consistent with the proteomic data. PC1 and PC2 explained 53.98% of the total variance, with PC1 accounting for 36.45% and PC2 accounting for 17.53%, indicating a marked difference in metabolic profiles between the two groups (Fig. 3A). Volcano plots revealed multiple putatively annotated differentially abundant metabolites (FC ≥ 1.2 or ≤ 1/1.2, P < 0.05), including lipid peroxidation products (15-HETE, 9,12,13-TriHOME, and 13(S)-HODE), purine metabolites (uric acid, AMP) and amino acid derivatives (Leu-Asp, guanidinosuccinic acid) (Fig. 3B). Notably, pyridoxine (vitamin B6) was identified as the most significantly altered metabolite in this untargeted screening, with levels markedly decreased in the BALF of PIBO patients. These findings are consistent with the enrichment of the vitamin B6 metabolism pathway in the results of the transcriptomic KEGG analysis, suggesting a potential perturbation of vitamin B6 homeostasis in the airway microenvironment (Fig. 3C).

Fig. 3.

BALF metabolomics reveals metabolic reprogramming of the airway microenvironment in PIBO patients. A PCA plot of BALF metabolomics data showing clear separation between the PIBO and control groups, indicating systemic differences in metabolic profiles. B Volcano plot of differentially abundant metabolites (FC ≥ 1.2 or ≤ 1/1.2, P < 0.05). The metabolites significantly upregulated in PIBO include 15-HETE, uric acid, and 13(S)-HODE, whereas pyridoxine and stearic acid are significantly downregulated. C Boxplot showing pyridoxine levels. Pyridoxine levels were significantly lower in the PIBO group than in the control group. D KEGG pathway enrichment analysis of differentially abundant metabolites. The significantly enriched pathways included nucleotide metabolism, purine metabolism, biosynthesis of amino acids, and biosynthesis of cofactors. E Correlation network of the top 100 metabolite pairs based on Spearman’s correlation analysis

Pathway enrichment analysis of differentially abundant metabolites further revealed significant enrichment in core metabolic pathways, including biosynthesis of amino acids, nucleotide metabolism, purine metabolism, and biosynthesis of cofactors (Fig. 3D). These findings suggest metabolic disturbances in the PIBO airway microenvironment, potentially involving amino acid and energy metabolism. To investigate coordinated changes among key metabolites, we constructed a correlation network using the top 100 metabolite pairs based on Spearman’s correlation analysis (Fig. 3E). The network showed a subset of amino acids and their derivatives that exhibited strong correlations, with some pairs reaching r = 1.0. Several lipid-related metabolites, including stearic acid and its correlated partners, also showed strong correlations. Together, these findings suggest a high degree of coordinated metabolic alterations in the PIBO airway microenvironment, predominantly involving amino acid metabolism.

Integrated multiomic analysis reveals core molecular signatures of PIBO

To further investigate regulatory associations between proteins and metabolites in PIBO, we performed an integrated correlation analysis of proteomic and metabolomic data (Fig. 4A). Widespread correlations were observed between proteins and metabolites in PIBO BALF, with the strongest positive correlations predominantly involving amino acid metabolites and several key inflammatory proteins. Notably, the metalloproteinase ADAM8 exhibited strong positive correlations with several amino acid metabolites, including Leu–Asp and Pro-Ile, attaining a maximum of r = 0.97. These amino acid metabolites were strongly correlated with multiple proteins in the PIBO proteome, including OSM, CXCL8, VNN2, TNFSF14 and ARG1 (r = 0.75–0.98), suggesting a highly coordinated module linking amino acid metabolism and inflammatory signaling.

Fig. 4.

Multiomics integration reveals core molecular signatures of PIBO. A Correlation heatmap of differentially expressed proteins and differentially abundant metabolites in the BALF of patients. B Correlation between pyridoxine and CXCL8 protein levels in BALF. Spearman’s correlation analysis was performed (n = 9). C Effect of pyridoxine on CXCL8 secretion in human small airway epithelial cells. Cells were treated with or without pyridoxine, and CXCL8 levels in the supernatant were measured by ELISA (n = 3 independent experiments). D Integrative KEGG pathway enrichment analysis of PIBO BALF proteomics and metabolomics data. E Potential pathogenic mechanisms in the PIBO airway microenvironment revealed by integrative multiomics analysis

In contrast, pyridoxine (vitamin B6) showed consistent negative correlations with multiple airway inflammation-related proteins, including CXCL8 (r = − 0.7), MMP13 (r = − 0.72), and ADAM8 (r = − 0.77), indicating a potential inverse relationship between vitamin B6 metabolism and inflammatory protein expression in the PIBO airway microenvironment. To illustrate this relationship more directly, we further plotted the correlation between pyridoxine and CXCL8 protein levels (Fig. 4B), which confirmed a significant negative correlation. To complement this correlative observation, we performed an in vitro experiment in which human small airway epithelial cells were treated with pyridoxine, and CXCL8 levels in the supernatant were measured by ELISA. The results showed that pyridoxine treatment significantly reduced CXCL8 secretion, further supporting a potential regulatory link between vitamin B6 metabolism and CXCL8 mediated inflammation (Fig. 4C). Vitamin B6 is an essential water-soluble vitamin that exerts anti-inflammatory effects by modulating inflammatory cytokine metabolism and oxidative stress balance. The negative correlation between vitamin B6 and inflammatory proteins, together with the observation that vitamin B6 treatment reduced CXCL8 secretion in cell-based ELISA, may suggest a possible association between dysregulated vitamin B6 metabolism and airway inflammation. Thus, vitamin B6 supplementation might have a modulating effect on inflammatory responses, pointing to vitamin B6 metabolism as a potential target.

Integrative KEGG pathway enrichment analysis of the proteomic and metabolomic datasets revealed core pathways that were significantly enriched in both datasets (Fig. 4D). Shared enrichment of amino acid metabolism pathways (e.g., biosynthesis of amino acids; alanine, aspartate, and glutamate metabolism), cofactor biosynthesis pathways, and signaling pathways (PI3K-Akt, mTOR, and AMPK) was observed. Notably, these enriched pathways partially overlapped with those in which strong protein-metabolite correlations were detected. Integrated analysis of BALF cell transcriptomic, supernatant proteomic, and metabolomic data thus yielded a multi-dimensional molecular profile of the PIBO airway microenvironment. These multi-layer data may reflect a cascading pathological process involving epithelial injury, inflammatory activation, tissue remodeling, and metabolic dysregulation, as schematized in Fig. 4E. Transcriptomic evidence of upregulated epithelial markers (KRT19 and FOXJ1) and cilia-related pathways indicated extensive epithelial shedding and ciliary dysfunction, which subsequently triggered local inflammatory remodeling, as evidenced by the upregulation of chemokines, including CXCL8 and CX3CL1. Among these, CXCL8 was broadly associated with multiple inflammation- and remodeling-related proteins, driving a robust neutrophil-driven inflammatory response. As inflammation persists, tissue remodeling and metabolic dysregulation progressively amplify this process.

Discussion

Integration of multiomics insights into the pathogenesis of PIBO

By integrating multiomics data, this study systematically delineates a pathological network in the PIBO airway microenvironment characterized by epithelial injury, inflammatory activation, tissue remodeling, and metabolic dysregulation. Transcriptomic analysis revealed aberrant expression of epithelial marker genes and widespread activation of pathways involved in neutrophil degranulation and ECM remodeling. These findings were further supported by proteomic data. Specifically, Olink proteomics detected altered expression of KRT19 in the BALF of PIBO patients, suggesting potential epithelial injury. Upregulation of the chemokine CXCL8 and the matrix metalloproteinase MMP9 indicated activation of neutrophils and macrophages. Moreover, the expression of pro-fibrotic cytokines, including ADAM8, OSM, and TNFSF14, was observed, potentially reflecting fibroblast activation. Collectively, these findings suggest the coexistence of immune cell activation, epithelial injury, and aberrant airway remodeling in the PIBO microenvironment. Metabolomic analysis further revealed the downstream metabolic consequences of these pathological processes. Rather than serving as passive responses to persistent inflammation, these metabolic alterations may actively contribute to the maintenance of chronic disease progression by influencing cellular survival and function.

Methodological innovation and clinical translational value

This study employed a complementary multiomics integration strategy. Transcriptomic analysis revealed disease-associated gene expression and regulatory networks, proteomic analysis identified functional effector molecules involved in pathogenesis, and metabolomic analysis profiled global alterations in downstream metabolites. This integrated approach overcomes the limitations of single-omics analyses and provides a more systematic understanding of the pathophysiology of PIBO.

CXCL8 was identified as a central hub in the BALF, and multiple CXCL8-correlated proteins were identified, suggesting that they may play cooperative roles in the PIBO inflammatory network and disease progression. BALF CXCL8 levels were negatively correlated with lung function (predicted FEV1%), supporting its potential as a severity biomarker and suggesting further investigation into CXCL8-associated proteins and multimarker tools. Metabolomic analysis revealed the downregulation of pyridoxine and its negative correlation with CXCL8, suggesting a potential association between vitamin B6 metabolism and inflammation, although this finding was derived from untargeted metabolomics and requires targeted validation. Consistently, in vitro experiments showed that pyridoxine treatment reduced CXCL8 secretion in small airway epithelial cells, providing preliminary supportive evidence for this potential link.

Limitations and future perspectives

Several limitations should be acknowledged. This study was based on a small discovery cohort with a cross-sectional, single-center design, which may attenuate the detection of weak signals and preclude longitudinal assessment of disease progression. Consequently, the generalizability of our findings and their implications for disease trajectories require further validation in independent cohorts. In addition, PIBO can be triggered by diverse pathogens (e.g., adenovirus, influenza virus, RSV, and Mycoplasma pneumoniae), and different pathogens may induce distinct initial molecular responses. Although PIBO is fundamentally a chronic post-infectious sequelae in which the acute infection has resolved, pathogen diversity may still act as a potential confounding factor affecting the interpretation of molecular profiles. However, due to the retrospective nature of this study, we were unable to obtain complete pathogen identification data for all patients. Future prospective studies with systematic documentation of causative pathogens are needed to clarify their potential influence on the PIBO molecular phenotype.

Furthermore, the nature of BALF cell pellets imposes inherent constraints on transcriptomic analysis. The abundant presence of shed epithelial cells in PIBO BALF likely creates a dilution effect, whereby RNA contributions from epithelial cells dominate the sequencing data and compress signals from immune cells and fibroblasts. This explains why fibrosis-associated genes (e.g., FN1, VIM, and TGFB1) and immune cell markers (e.g., CD4, CD8A, CD8B, and CD68) were not significantly upregulated in the PIBO transcriptome. Moreover, many immune effector functions rely on protein secretion, and their transcripts may be unstable in BALF cells, further limiting the detection of immune-related signals by conventional differential analysis. Transcriptomic analysis alone cannot resolve the dynamics of specific cell subsets or effector molecule status. Future studies employing single cell RNA sequencing (scRNA-seq) could more precisely dissect the distinct contributions of different cell subsets (e.g., epithelial cells, macrophages, and neutrophils) in the BALF to the inflammatory response in PIBO. However, the integration of proteomic and metabolomic data partially compensates for these limitations at the protein and metabolite levels, offering a more comprehensive view of the airway molecular landscape in PIBO patients.

Moreover, the metabolomic findings in this study were derived from untargeted LC‑MS/MS analysis, with metabolite identification primarily based on database matching (Level 2/3 putative annotation) without validation using authentic standards. Notably, vitamin B6 (pyridoxine) is chemically unstable and may undergo degradation during long-term sample storage, potentially affecting its quantitative accuracy. Therefore, the pyridoxine-related findings in this study should be interpreted as exploratory and hypothesis-generating, and warrant further validation using freshly collected samples with targeted metabolomic approaches.

To overcome these limitations, future large-scale, multicenter prospective studies are needed to validate the clinical utility of the CXCL8-centered network (e.g., TNFSF14 and ADAM8) and its detectability in noninvasive samples such as blood or urine. Functional studies in animal models and cellular systems are also needed to establish causal roles for these key molecules. The downregulation of vitamin B6 and its negative correlation with CXCL8 particularly warrant in vitro and in vivo investigations into its role in airway inflammation.

Overall, this study offers a preliminary framework for further investigation into core proteins (e.g., CXCL8 and ADAM8) as candidate molecules for targeted therapy, and suggests a potential association between vitamin B6 metabolism and airway inflammation that warrants further mechanistic and functional exploration.

Conclusions

This study presents one of the first integrated multiomic analyses in children with PIBO, combining transcriptomics of BALF cell pellets with Olink proteomics and untargeted metabolomics of BALF supernatants, thereby systematically delineating the multidimensional molecular landscape of the local airway microenvironment. The integrated data collectively suggest a pathological cascade involving epithelial injury, inflammatory activation, tissue remodeling, and metabolic dysregulation. CXCL8 was identified as a central hub protein in this network, and its level in BALF was significantly negatively correlated with lung function, supporting its potential as a severity biomarker and therapeutic target. Metabolomic analysis further revealed marked downregulation of pyridoxine expression and its negative correlation with the expression of CXCL8 and other inflammatory proteins, suggesting a potential link between vitamin B6 metabolism and airway inflammation that warrants further investigation. Overall, this study provides a preliminary multiomics framework for understanding PIBO pathogenesis, developing multimarker assessment tools and identifying candidate molecules for targeted therapies that require further functional validation.

Abbreviations

PIBO

Postinfectious bronchiolitis obliterans

BALF

Bronchoalveolar lavage fluid

CPAM

Congenital pulmonary airway malformation

PCA

Principal component analysis

DEG

Differentially expressed gene

DEP

Differentially expressed protein

ECM

Extracellular matrix

NET

Neutrophil extracellular trap

PPI

Protein‒protein interaction

STRING

Search Tool for the Retrieval of Interacting Genes/Proteins

HRCT

High-resolution Chest CT

TPM

Transcripts per million

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

PEA

Proximity extension assay

ELISA

Enzyme-linked immunosorbent assay

Authors’ contributions

BX and LS conceived and designed the study, supervised the entire study, and revised the manuscript. HY designed and performed the in vitro cell experiments. MJ and LG performed the experiments, analyzed the multiomics data, and drafted the manuscript. JY and QQ contributed to patient recruitment and clinical evaluation. WC assisted with the statistical analysis. JL assisted with clinical data collection. All the authors read and approved the final manuscript.

Funding

This study was supported by Clinical and translational medicine research project of Chinese Academy of Medical Sciences (2023-I2M-C&T-A-015); Major Project of the Ministry of Science and Technology of China (2024ZD0522505); Capital Health Research and Development of Special (2026-1Q-1083);The National Natural Science Foundation of China (Grant No. 81870552); the National Key R&D Program of China (Grant No. 2018YFC2000400); Beijing Municipal Natural Science Foundation (No. L246011); Capital's Funds for Health Improvement and Research(2026-1Q-1072);Funding for Reform and Development of Beijing Municipal Health Commission (No. EYGF-HX-05).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Beijing Children's Hospital, Capital Medical University (2024-Y-321-D) and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from the guardians of all the participants.

Consent for publication

Not applicable.

Competing interests

All the authors declare that they have no competing interests.