Untargeted metabolomics profiling of childhood asthma: linking metabolic pattern to disease severity

Abstract

Background

Asthma is a heterogeneous inflammatory airway disorder associated with complex metabolic alterations. Despite significant advances in diagnosis and treatment, its prevalence and disease burden continue to rise, particularly among children. Recent research reveals that metabolic alterations play a crucial role in the disease development, severity, and treatment response. This study aimed to investigate dysregulation of serum metabolites in children with moderate-to-severe asthma compared with healthy controls using an untargeted metabolomic profiling approach.

Results

Hierarchical clustering and stratified analyses identified distinct metabolic signatures differentiating asthmatic patients from controls, with progressive changes reflecting increasing disease severity. Thirty-nine metabolites were found to be commonly dysregulated in moderate and severe asthma. Notably, adenosine and orotic acid were significantly upregulated, while 5-methoxytryptophol (5-MTX) was markedly downregulated, showing its lowest levels in severe asthma. Pathway enrichment analysis revealed significant disruptions in pyrimidine, amino acid, and urea cycle pathways.

Conclusions

The identified metabolites and disrupted metabolic pathways may offer potential biomarkers of disease severity and provide insights to guide future therapeutic strategies. However, their utility in clinical practice as diagnostic or prognostic biomarkers across the full spectrum of asthma requires further validation in independent cohorts that include children with mild disease.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40348-026-00219-2.

Background

Asthma is a heterogeneous chronic inflammatory disease of the airways and the most common chronic respiratory condition in childhood worldwide, affecting approximately 14% of children and young individuals [1]. It contributes substantially to morbidity, healthcare burden, and reduced quality of life [2]. Childhood asthma involves complex pathophysiological and molecular processes, encompassing multiple cell types, mediators, and immune pathways [1, 3]. Clinically, it presents with variable symptoms, including cough, wheezing, shortness of breath, and chest tightness, alongside differing degrees of severity and airway hyper-responsiveness assessed by lung function tests and treatment response [1, 3]. Spirometry remains central to diagnosis, with decreased peak expiratory flow (PEF) and forced expiratory volume in one second (FEV1) indicating impaired lung function; however, the natural fluctuation of PEF and FEV1 in children, combined with episodic symptoms, makes early diagnosis challenging [4]. Timely diagnosis and treatment are crucial for effective disease control, reduction of exacerbations, and prevention of irreversible airway remodeling [5]. Therefore, identifying novel diagnostic biomarkers is essential for early prediction, monitoring airway dysfunction and inflammation, and assessing therapeutic responses at the molecular level.

Metabolomics, an advanced and sensitive analytical approach for profiling small molecules produced by cellular activity, employs mass spectrometry (MS) and informatics tools to provide insights into pathogenic pathways of complex diseases such as asthma [5]. It has become central to biomarker discovery, disease characterization, and evaluation of treatment responses. While metabolomics studies have explored the heterogeneity and treatment response patterns of childhood asthma [5], only limited work has examined serum metabolomic alterations associated with disease severity.

In this study, we investigate the serum metabolomic profiles of children with varying asthma severities using an untargeted liquid chromatography–mass spectrometry (LC–MS) approach. This analysis aims to deepen our understanding of the molecular mechanisms underlying childhood asthma and its severity, and to identify dysregulated metabolites and pathways potentially linked to disease progression.

Methods

Study design and selection criteria

This cross-sectional study included 150 participants (100 asthmatic children and 50 age-matched healthy controls). Asthmatic children were diagnosed with persistent asthma and treated with either inhaled corticosteroid (ICS) or ICS/long-acting beta-agonist (controller) combination therapy for at least three consecutive months. Children were recruited from the pediatric outpatient asthma clinic at Jordan University Hospital (JUH). Children with comorbid disorders, including chronic lung disease of prematurity, severe gastroesophageal reflux disease, aspiration disorders, or vocal cord dysfunction, were excluded. Healthy controls were recruited from general pediatric outpatient visits (non-respiratory) at the same institution. Eligibility required no personal or family history of asthma or chronic respiratory disease, no recurrent wheezing or allergy requiring treatment, no use of inhaled or systemic steroids or other asthma medications, and no acute illness or respiratory symptoms in the preceding 4 weeks.

Demographics and clinical data collection

Demographics and clinical data of the enrolled children were collected by the research assistant. Parents were asked to complete medical history questionnaires, including the Asthma Control Questionnaire. Asthma control was assessed using the self-reported Asthma Control Test (ACT). This validated 5-question questionnaire evaluates asthma control over the past 4 weeks by assessing activity limitations, symptom frequency, nocturnal awakenings, rescue medication use, and overall control perception [6]. Each question has a point rating: 5–15 indicates poorly controlled asthma, 16–19 shows partially controlled asthma, and 20 or more indicates controlled asthma [6]. For the control group, given their asymptomatic and disease-free status, ACT was not administered, as it is validated only for individuals with asthma.

The asthmatic children (n = 100) were classified into two groups(n = 54 with moderate and n = 46 with severe asthma) based on their clinical symptoms and the number and type of medications used. Asthma severity classification was determined by board-certified pediatric pulmonologists based on a comprehensive clinical evaluation, including symptom frequency, exacerbation history, medication requirements, and level of asthma control, in line with routine clinical practice for pediatric asthma management. Specifically, children classified as “moderate asthma” had persistent daily symptoms or frequent exacerbations requiring ICS (± long-acting beta-agonist as controller) but retained relatively preserved lung function and required no or only moderate controller therapy escalation. Those classified as “severe asthma” had frequent or continuous symptoms, more frequent exacerbations, or hospitalisations (or more intensive therapy), and clinician judgment that the disease was more difficult to control despite ICS. Although spirometry is commonly used in asthma assessment, lung function measurements were not consistently available for all enrolled children due to age-related cooperation limitations; therefore, spirometric parameters were not included in the present analysis. Most patients who visited the respiratory clinic had moderate or severe asthma, with very few cases of mild asthma. All of the asthmatics were receiving ICS. Some asthmatic patients received azithromycin, which was administered at the discretion of the pediatric pulmonologist in children with severe or poorly controlled asthma, recurrent exacerbations, as an adjunct to standard ICS-based therapy.

Moreover, participants were classified based on the Centers for Disease Control and Prevention (CDC) categorization of body mass index (BMI)-age and sex-specific percentiles on a growth chart [7]. Underweight is below the 5th percentile, healthy weight is the 5th to less than the 85th percentile, overweight is the 85th to less than the 95th percentile, and obesity is the 95th percentile or greater.

Sample collection, storage, and dye extraction protocol

Blood samples were collected from participating children by a well-trained nurse in plain tubes and directly centrifuged at 1500 × g for 10 min at 4◦C. The obtained serum was transferred to microtubes and stored at −20 ◦C until extraction. Briefly, A total of 345 µL of cold LC–MS grade methanol (Fisher Scientific, Hampton, NH, USA) was added to 35 µL of serum, followed by 172.5 µL of HPLC-grade chloroform (Fisher Scientific, USA). The mixture was shaken for 30 s, then 88 µL of LC–MS grade water (Fisher Scientific, USA) was added and shaken again for 30 s. Subsequently, equal volumes of chloroform and water (172.5 µL each) were added, and the mixture was centrifuged at 10,000 rpm for 5 min. From the upper separated phase (aqueous metabolites), 400 μL was transferred to a new Eppendorf tube, dried using a vacuum centrifugal evaporator, and stored at −80 ◦C until further analysis.

Metabolomic analysis

An ultra-high performance liquid chromatography coupled to electrospray ionization and quadrupole time of flight mass spectrometry (UHPLC-ESI-QTOF-MS/MS) was operated to perform untargeted metabolomics analysis. The system used an electrospray ionization (ESI) source, a solvent delivery system pump (HPG 1300), an autosampler, and a temperature-controlled column compartment. The mobile phases consisted of Phase A (water containing 0.1% formic acid) and Phase B (acetonitrile (ACN) with 0.1% formic acid). A 10 µL aliquot of the sample was injected, and chromatographic separation was carried out using a Hamilton® IntensitySolo 2 C18 column (100 mm × 2.1 mm, 1.8 µm particle size) maintained at 35 °C. The separation followed a gradient elution starting with 99% water. [8, 9].

Most polar metabolites were eluted earlier in the chromatographic run, while less polar and non-polar compounds exhibited longer retention times. A microflow technique was applied, using a moderate flow rate of 0.25 mL/min and a higher rate of 0.35 mL/min for column washing and method conditioning. The ESI source parameters for each injection were as follows: drying gas was delivered at 10.0 L/min and maintained at 220 °C; the capillary voltage was 4500 V; and the nebulizer pressure was 2.2 bar. For MS2 data acquisition, collision energy was stepped from 100 to 250%, with a base energy of 20 eV and an End Plate offset of 500 V. External calibration was performed using sodium format. The data acquisition was divided into two phases: an initial Auto MS scan (0–0.3 min) for the sodium format calibration, followed by an Auto MS/MS scan (0.3–30 min) involving fragmentation. The automatic in-run mass scan range was 20–1300 m/z. The precursor ion window was ± 0.5 m/z, with up to three precursor ions selected per cycle. Each cycle lasted 0.5 s, and the intensity threshold for precursor selection was 400 counts(cts).

Quality Control (QC) procedure

A pooled QC sample was prepared by combining 10 µL aliquots from all study samples and injected at regular intervals, approximately every 6–8 injections, throughout the analytical run to monitor system stability and analytical reproducibility. Retention time stability and signal intensity reproducibility were assessed across all QC injections and features with poor reproducibility (QC RSD > 25%) were excluded to ensure high data quality. Evaluation of QC signal patterns throughout the batch confirmed that no signal drift occurred, and therefore, no drift-correction algorithms were required. To enhance technical reliability, each sample was injected in duplicate, and the mean of the duplicate measurements was used for all subsequent analyses.

All samples, including pooled QC samples, were analyzed in a single analytical batch on the same UHPLC-ESI-QTOF-MS instrument. A pooled QC sample was injected at the beginning and end, and approximately every 6–8 injections throughout the sequence, to monitor technical variability and ensure analytical reproducibility. The median relative standard deviation (RSD) of all detected features in the pooled QC samples was below 25%, indicating excellent technical consistency and reproducibility. To prevent confounding of biological effects with injection order, samples were randomized prior to injection using a stratified randomization scheme. In contrast, QC injections were regularly distributed throughout the run to continuously assess system performance.

Data processing and statistical analysis

Data analysis was conducted using MetaboScape® 4.0 software (Bruker Daltonics) [10]. In the T-ReX 2D/3D processing workflow, the bucketing parameters were set to an intensity threshold of 1,000, a peak length of 7 spectra, and a peak area for feature quantification. The scans were carried out within a retention time of 0.3 to 25 min and a mass-to-charge (m/z) range of 50 to 1000. Each sample underwent duplicate analysis by LC-QTOF, resulting in a total of 300 injections.

Metabolite identification was achieved by aligning MS/MS spectra and retention times with entries in the Human Metabolome Database (HMDB), an annotated resource tailored for metabolomics research. The compounds identified by MS/MS were annotated using library matching. Following annotation, metabolites were filtered by selecting those with the highest annotation quality (AQ) scores, representing the optimal retention time values, MS/MS scores, m/z values, mSigma values, and analyte list spectral library.

The metabolite data sets were exported in CSV format and analyzed further using MetaboAnalyst 5.0, a robust tool for metabolomics data interpretation [11]. Data filtering thresholds included a p-value of < 0.05 for significance and a fold change cutoff of 1.5. In this untargeted metabolomics workflow, raw data were processed with pooled QC samples used for normalization to correct for technical variation, ensuring that observed differences reflect actual biological variation. All samples were processed and analyzed under consistent extraction and injection conditions. Data transformation was not applied, as the distribution of peak intensities showed no significant skewness after feature filtering. To ensure equal contributions from all variables in multivariate analyses, auto-scaling (mean-centring followed by standardisation) was performed in MetaboAnalyst 5.0. Analytical methods, including Principal Component Analysis (PCA), Sparse Partial Least Squares Discriminant Analysis (sPLS-DA), volcano plots, and pathway enrichment analysis, were applied to assess differences among the experimental groups. To control for multiple comparisons and minimise false positives, the false discovery rate (FDR) correction method was used. Before statistical testing, data distributions and variance homogeneity were visually examined using histograms. One-way analysis of variance (ANOVA) was performed to compare the three experimental groups: group 1 (healthy, as a control), group 2 (moderate), and group 3 (severe asthma). For pairwise comparisons, independent Student’s t-tests were employed between the following groups: Asthma vs. Healthy, Moderate vs. Healthy, Severe vs. Healthy, and Severe vs. Moderate. The dataset was examined for missing values; none were present for any of the measured variables, so no imputation was required.

Results

Clinical characteristics and demographics of the study population

A total of 150 participants were included in the study, comprising 100 asthmatic patients (54 with moderate asthma and 46 with severe asthma) and 50 healthy controls. The clinical characteristics and demographic data of the study groups are presented in Table 1. The mean age of the study participants was comparable across the three groups. Although controls were slightly older than asthmatic patients, the age difference between groups was not significant and of limited clinical relevance. Gender distribution was comparable across groups, with no statistically significant difference observed (p = 0.25). Most participants were male (65% of asthmatics and 64% of controls). Asthma control, assessed using ACT, varied markedly between the moderate and severe asthma groups, with a significantly higher proportion of controlled asthma in the moderate group and a greater prevalence of partially controlled and uncontrolled asthma among patients with severe asthma (p < 0.001). Asthma control status, as assessed by ACT scores, showed clinical heterogeneity within the asthma cohort, suggesting distinct phenotypes despite a shared diagnosis. Based on the CDC’s BMI-for-age classification, BMI category distribution differed significantly across the three groups (p = 0.021). Compared with controls, patients with moderate and severe asthma exhibited higher proportions of overweight and obesity, while healthy weight status was more prevalent among controls. Allergic comorbidities were common among asthmatic patients, particularly allergic rhinitis, which affected most moderate and severe asthma cases. The prevalence of atopic dermatitis, allergic conjunctivitis, and food allergy was higher in the severe asthma group compared with the moderate group; however, these differences were not statistically significant.

Table 1.

Clinical characteristics and demographics of the study groups

Characteristic	Controls (n = 50)	Moderate asthma (n = 54)	Severe asthma (n = 46)	P-value
Age (years), mean ± SD	12.8 ± 2.78	11.55 ± 2.45	11.65 ± 3.23	0.050¥
Gender, n (%)				0.25¥
Female	18 (36.0%)	16 (29.6%)	19 (41.3%)
Male	32 (64.0%)	38 (70.4%)	27 (58.7%)
Asthma control based on ACT score, n (%)				< 0.001#
Controlled (ACT ≥ 20)	—	33 (61.1%)	10 (21.7%)
Partially controlled (ACT 16–19)	—	5 (9.3%)	14 (30.4%)
Not controlled (ACT ≤ 15)	—	16 (29.6%)	22 (47.8%)
BMI category, n (%)				0.021¥
Underweight	2 (4.0%)	5 (9.3%)	3 (6.5%)
Healthy weight	38 (76.0%)	27 (50.0%)	26 (56.5%)
Overweight	6 (12.0%)	11 (20.4%)	11 (23.9%)
Obese	4 (8.0%)	11 (20.4%)	6 (13.0%)
Allergies and comorbidities, n (%)
Allergic rhinitis	—	41 (75.9%)	38 (82.6%)	0.39#
Atopic dermatitis	—	7 (13.0%)	7 (15.2%)	0.75#
Allergic conjunctivitis	—	12 (22.2%)	15 (32.6%)	0.23#
Food allergy	—	1 (1.9%)	5 (10.9%)	0.09#

Data are presented as mean ± standard deviation or number (percentage)

ACT Asthma Control Test, BMI Body mass index

^¥Overall p-values compare the three groups (controls, moderate asthma, severe asthma) using one-way ANOVA for continuous variables (age) and the Chi-square test of independence for categorical variables (gender and BMI category)

^#Pairwise p-values compare moderate versus severe asthma using the independent-samples t test for age and the Chi-square test (or Fisher’s exact test when expected cell counts were < 5) for categorical variables (ACT-based asthma control, BMI category, and allergic comorbidities)

Regarding pharmacological management, all included asthmatic patients (n = 100) were receiving inhaled corticosteroids. Those taking fluticasone in different strengths (250 mcg, 500 mcg, and 1000 mcg) twice or three times a day accounted for 85%, while those taking budesonide and beclomethasone accounted for 10% and 5%, respectively. Moreover, the majority of asthmatic patients (99%, n = 99) were receiving a short-acting beta-agonist (SABA), and 61% (n = 61) were receiving a long-acting beta-agonist (LABA). Only 13% (n = 13) received azithromycin antibiotics.

Overview of the metabolomics profiling among the study groups

Following untargeted metabolomics analysis using UHPLC-ESI-QTOF-MS, 8,037 metabolic features were identified. After QC, 161 features were removed as replicates, leaving 7,876 features. Metabolite identities were annotated and filtered against the Human Metabolome Database (HMDB), yielding a refined set of 120 metabolites. After excluding the exogenous metabolites, 90 endogenous metabolites were retained for further statistical analysis.

The metabolomics profiles of the study groups were analyzed using sPLS-DA (Fig. 1). The scores plot illustrates distinct clustering patterns among asthmatic patients and healthy controls, highlighting significant metabolic differences associated with the disease (Fig. 1). However, the metabolic profiles of moderate and severe asthmatic patients largely overlapped, indicating a high degree of similarity in their biochemical alterations (Fig. 1). These findings demonstrate that while asthma is metabolically distinct from the healthy state, the metabolic signatures of moderate and severe asthma are closely associated. Moreover, one-way ANOVA analysis annotated 59 metabolites with statistically significant variation across the three groups (control, moderate, and severe) (p < 0.05). The identities of the 59 shared metabolites among the three groups were presented in Supplementary Table 1.

Figure 1.

Metabolomics profiling of the study groups. The sparse partial least squares-discriminant analysis (sPLS-DA) was performed for the study groups, including healthy controls and asthmatic patients(moderate and severe asthmatics). Component 1 (sPLS-DA latent variable 1, 9.8% variance) and Component 2 (sPLS-DA latent variable 2, 7.5% variance)

Metabolomics profiling of asthmatic patients and control groups

The sPLS-DA analysis demonstrated complete separation between asthmatics and healthy controls, indicating robust metabolic differences in a binary comparison (Fig. 2A). In addition, volcano plot analysis, considering a t-test moderated by FDR p < 0.05 and a Fold Change (FC) cut-off of 1.5, revealed that 26 metabolites were significantly dysregulated between the two groups (Fig. 2B). Among them, 23 and 3 were up- and down-regulated in asthmatic patients compared to controls. The significantly upregulated metabolites include adenosine, orotic acid, and 1,3-dimethyluracil. In contrast, isocitric acid, L-fucose, and 5-MTX were significantly downregulated. The heat map of the significantly dysregulated metabolites is presented in Fig. 2C. Hierarchical clustering revealed distinct metabolic profiles between asthmatics and control groups, with controls clustering more closely together. In contrast, asthmatics displayed metabolic alterations associated with disease status. These patterns are consistent with disease-related metabolic signatures and may have potential utility for asthma classification, although no causal or prognostic inference can be drawn from the present data. (Fig. 2C). Furthermore, pathway analysis comparing asthmatic and control groups revealed significant alterations in several metabolic pathways, notably pyrimidine, purine, and histidine and arginine metabolism (Fig. 2D). In line with this, functional enrichment analysis also highlighted substantial effects of asthma on several biochemical pathways, including phenylacetate, pyrimidine, and glycine and serine metabolism (Fig. 2D).

Figure 2.

Dysregulated metabolites between asthmatic patients and controls and pathway analysis

Metabolomics profiling of moderately asthmatic patients and controls

To investigate metabolomic alterations associated with asthma progression, we first conducted a binary comparison between moderately asthmatic subjects (n = 54) and controls (n = 50). The sPLS-DA analysis revealed a clear separation between the two groups, indicating differences in their metabolic profiles (Fig. 3A). The volcano plot analysis, using a t-test with FDR correction (p < 0.05) and a fold change (FC) threshold of 1.5, identified 26 metabolites that were significantly dysregulated between moderately asthmatic individuals and controls (Fig. 3B). Of these, 25 metabolites were upregulated, with adenosine and orotic acid showing the highest fold changes. On the other hand, only one metabolite, isocitric acid, was significantly downregulated in the moderate asthma group compared to controls (Fig. 3B). Pathway analysis identified significant alterations in histidine, pyrimidine, and arginine metabolism. Additionally, functional enrichment analysis highlighted notable effects of asthma on pathways such as glycine and serine metabolism, phenylacetate metabolism, the urea cycle, and ammonia recycling (Fig. 3D).

Figure 3.

Metabolomic profiling of moderate asthmatic patients and controls. (A) The sparse partial least squares-discriminant analysis (sPLS-DA) for moderate asthma and control. Component 1 (sPLS-DA latent variable 1, 11.9% variance) and Component 2 (sPLS-DA latent variable 2, 8.2% variance) (B) Volcano plot showing metabolites that were significantly altered in the moderate asthma group compared to controls (FC = 1.5), where 25 metabolites were upregulated, and only one metabolite was downregulated. (C) Pathway analysis and (D) Enrichment analysis of significantly dysregulated metabolites (n = 26) between moderate asthmatic patients and controls

Metabolomics profiling of severe asthmatic patients and controls

Another binary comparison was conducted to investigate metabolomic profiling between severe asthmatics (n = 46) and controls (n = 50) and to annotate dysregulated metabolites associated with severe asthma. The sPLS-DA analysis shows pronounced clustering and separation between the two groups (Fig. 4A), underscoring a marked distinction in metabolic profiles between the severe asthma group and healthy controls. Furthermore, volcano plot analysis annotated 28 significantly altered metabolites, FC (1.5) (p < 0.05). Among these, 20 metabolites were upregulated, including adenosine, orotic acid, and 1,3-dimethyluracil, which exhibited high fold changes. Conversely, isocitric acid remained consistently reduced in this comparison, alongside seven additional downregulated metabolites, including 5-MTX, deoxycytidine, uric acid, and 6-(methylamino) purine (Fig. 4B). Pathway and functional enrichment analysis of the significantly dysregulated metabolites revealed significant disruptions in several metabolic pathways, notably pyrimidine metabolism, urea cycle, and amino acid metabolism, including arginine, glycine, proline, histidine, and serine metabolism (Figs. 4C and 4D).

Figure 4.

Metabolomics profiling of severe asthmatic patients and controls. (A) A sparse partial least squares-discriminant analysis (sPLS-DA) of the metabolomics profile of severe asthmatics versus controls showing the separation and clustering. Component 1 (sPLS-DA latent variable 1, 13.1% variance) and Component 2 (sPLS-DA latent variable 2, 9.5% variance). (B) Volcano plot showing the statistically significant dysregulated metabolites (n = 28) (FC = 1.5) and cutoff p-value < 0.05. The levels of 20 metabolites were upregulated, whereas 8 were downregulated in severe asthmatics compared with the control group. (C) and (D) A Pathway and enrichment analysis, demonstrating the main pathways included in the metabolic changes based on the 28 dysregulated metabolites in the severe asthmatic group versus the healthy

Metabolomics alteration associated with asthma severity: a profiling between moderate and severe asthmatic patients

To examine metabolomic changes associated with asthma progression and to explore dysregulated metabolites that may serve as potential prognostic biomarkers for asthma, a final binary comparison was carried out between asthma patient subgroups: moderate (n = 54) vs severe (n = 46). The sPLS-DA analysis showed substantial overlap between the compared groups (Fig. 5A), reflecting the shared metabolic background typical of individuals with asthma, despite differences in disease severity. To further investigate the shared and unique metabolites between the two groups, a Venn diagram was used to compare metabolites annotated in moderate asthma and severe asthma with those in healthy controls (Fig. 5B). The diagram shows that 18 dysregulated metabolites were unique to severe asthma (blue circle), 11 to moderate asthma (red circle), and 39 were shared in the overlapping region (Fig. 5B). The identity of the detected shared and unique metabolites is presented in Supplementary Table 2. These findings highlight numerous common metabolic alterations associated with asthma and identify subgroup-specific metabolites that may be linked to disease severity. However, given the substantial overlap between moderate and severe asthma and the absence of a mild asthma group, these metabolites should be interpreted as exploratory markers rather than definitive prognostic biomarkers. Out of the 39 shared metabolites, 5 metabolites, namely, 1,3-Dimethyluracil, 5-Methoxytryptophan, 5-MTX, Adenosine, and orotic acid, were selected based on statistical significance P < 0.05, fold change (1.5), and consistent trend in disease severity. The average intensity of these selected common dysregulated metabolites was assessed across healthy control, moderate, and severe asthmatic patients (Fig. 5C). 1,3-Dimethyluracil showed a steady increase from healthy controls to patients with moderate asthma and reached its highest intensity in severe asthma, indicating a positive association with disease progression. 5-Methoxytryptophan showed a modest increase between healthy and moderate asthma patients and remained relatively stable in severe asthma, suggesting a minor role in distinguishing disease severity. On the other hand, 5-MTX exhibited an opposite trend, indicating a negative correlation with asthma severity. Adenosine showed one of the most pronounced patterns, starting at low levels in healthy controls, increasing moderately in patients with moderate asthma, and then rising sharply in severe asthma, making it a strong candidate biomarker of disease progression. Finally, orotic acid followed a similar upward path to 1,3-dimethyluracil, with gradual increases across the severity spectrum, further supporting its potential role as a metabolite that tracks asthma severity (Fig. 5C).

Figure 5.

Metabolomics Alteration Associated with Asthma Severity. (A) The sparse partial least squares-discriminant analysis (sPLS-DA) of moderate and severe asthma groups. Component 1 (sPLS-DA latent variable 1, 7.8% variance) and Component 2 (sPLS-DA latent variable 2, 3.9% variance). (B) A Venn diagram illustrates the overlap between severe and moderate asthmatic patients, annotating 39 metabolites as significantly dysregulated shared metabolites between the two groups. (C) The line graph shows the mean intensities of representative metabolites that are significantly dysregulated in healthy controls, moderate asthma, and severe asthma. Adenosine (yellow line), Orotic acid (green line), 5-Methoxytryptophan (orange line), 1,3-Dimethyluracil (blue line), and 5-MTX (gray line).

Discussion

In this study, the metabolomic alterations associated with asthma severity were investigated, providing valuable insights into biochemical changes in pediatric asthma and identifying potential new diagnostic and prognostic metabolites. The results revealed distinct clustering of metabolite expression between asthmatics (moderate and severe) and the control group, as analyzed using the sPLS-DA model. Specifically, moderate asthma was associated with dysregulation of 26 metabolites (25 upregulated and one downregulated), whereas severe asthma was associated with dysregulation of 28 metabolites (20 upregulated and 8 downregulated) compared with controls. Notably, the sPLS-DA model clearly showed overlap between the moderate and severe asthma groups. Further analysis using a Venn diagram annotated 39 shared dysregulated metabolites between the moderate and severe asthmatics. These include 5-MTX, 1,3-Dimethyluracil, 5-Methoxytryptophan, adenosine, and orotic acid, which exhibited pronounced changes in their levels across the disease state.

Levels of 5-MTX, an endogenous derivative of L-tryptophan and a serotonin metabolite, were downregulated in asthmatic patients, with the lowest levels in severe cases. This metabolite is reported to have anti-inflammatory, anti-fibrotic, and antioxidant activities, which could potentially contribute to tissue protection in various organ systems [12]. The detected decline in 5-MTX levels suggests an association with disease severity, but the data do not support a direct mechanistic link. Our pathway analysis indicated a potential alteration in tryptophan metabolism [13]. It has been revealed that when melatonin synthesis is elevated, more tryptophan may be directed toward its production, potentially leaving less substrate available for 5-MTX formation [13]. In support of this, our pathway analysis demonstrated a dysregulation in tryptophan metabolism. Asthma in children, particularly the nocturnal type, has been linked to disrupted melatonin. Previous research has shown that patients with nocturnal asthma have markedly elevated serum melatonin levels compared with healthy controls [14]. Melatonin has been implicated in bronchoconstriction in asthma by activating MT2 receptors on airway smooth muscle [14]. This may influence airway tone and responsiveness to β2-agonist medications [14]. Therefore, the observed decrease in 5-MTX levels may reflect alterations in tryptophan-related metabolic pathways in severe asthma; however, a direct mechanistic link with melatonin production cannot be established from this study. Additionally, changes in 5-MTX may be associated with modulation of cytokines such as IL-2 and IL-6, although causality cannot be established from our data [15].

Isocitric acid, a metabolite within the citric acid cycle, was found to be significantly downregulated in all conducted binary comparisons. Recent studies suggest that lower levels of isocitric acid and related metabolites, such as citric acid, are associated with persistent cough, a common feature of asthma [16]. Acidic airway conditions influenced by these metabolites could activate capsaicin-sensitive sensory neurons, potentially contributing to airway responses such as cough and bronchoconstriction [17]. Microaspiration of acidic substances may further influence the release of neuropeptides, such as tachykinins and bradykinin [17]. Notably, bronchodilators have been shown to reduce the cough response to citric acid inhalation in asthmatic patients, supporting the role of acid-sensitive mechanisms in symptom development [18]. Collectively, these findings align with our results, which showed a significant decrease in isocitric acid levels in asthmatic patients compared to healthy controls (3.4-fold), suggesting a potential role for this metabolite in the pathophysiology of asthma.

Adenosine is a signaling molecule that has been associated with airway inflammation and hyperresponsiveness in asthma [19]. Elevated concentrations of adenosine have been identified in various biological samples from asthmatic patients, including bronchoalveolar fluid, blood, and exhaled breath [19]. Inhalation of adenosine can induce airway narrowing in individuals with asthma [20]. This response may involve A1 receptor activation and mast cell stimulation [20]. Intravenous administration of adenosine has been reported to worsen breathing difficulties and may provoke bronchospasm in some asthmatic patients [21]. Although blocking the A3 adenosine receptor has shown promise in reducing asthma symptoms, research findings remain mixed, and more studies are required to clarify its therapeutic potential [22]. All these findings support our results, which demonstrated elevated adenosine levels in asthmatic patients compared to healthy controls (5.2-fold), further reinforcing its potential role in airway inflammation and hyperresponsiveness. Additionally, adenosine was markedly elevated in the moderate asthma group and reached its highest levels in the severe asthma group compared with healthy controls. The high fold change observed reflects elevated adenosine levels in severe asthma, which may be associated with disease severity; however, its prognostic value cannot be inferred from this cross-sectional analysis.

Our results also revealed that orotic acid, a key intermediate in pyrimidine metabolism, was upregulated in asthmatic patients compared to healthy controls (4.9-fold). Moreover, when stratifying the analysis by disease severity, orotic acid levels remained significantly elevated in moderate (25.9-fold) and severe (34.8-fold) asthmatic patients compared to healthy controls. In addition, the pathway and functional enrichment analyses conducted in our study showed dysregulation of pyrimidine metabolism across all binary comparisons. This highlights a potential disturbance in the de novo pyrimidine synthesis pathway, which produces essential nucleotides such as uridine and cytidine [23, 24]. The accumulation of orotic acid in asthmatic patients could reflect increased demand for nucleotide production or alterations in enzymatic activity, such as uridine monophosphate (UMP) synthase. However, our data do not establish causality [25]. Asthma is a chronic inflammatory condition. These metabolic shifts may be consistent with increased nucleotide turnover, but direct mechanistic links cannot be confirmed from this study [26]. The consistent increase in orotic acid levels across severity groups is associated with asthma severity and supports its consideration as a candidate biomarker for further validation studies. In addition to orotic acid, other metabolites linked to pyrimidine metabolism, such as thymidine, uridine, DUMP, and glutamine, were also significantly altered, strengthening the idea that this pathway may serve as a valuable source of biomarkers for asthma severity and progression. Nevertheless, further studies are needed to confirm the specificity and diagnostic utility of these findings in broader patient populations.

1,3-Dimethyluracil is a methylated derivative of uracil formed during the metabolism of methylxanthines such as caffeine, theophylline, and theobromine, and its presence in biological samples could denote exposure to or metabolism of methylxanthine-containing substances .

Our results showed upregulation in 1,3-dimethyluracil levels in asthmatic patients compared to healthy controls (3.7-fold). Additionally, 1,3-dimethyluracil levels were increased in both moderate and severe asthmatic patients compared to controls. However, the increase in severe asthma was not pronounced, indicating that its elevation does not increase with disease severity. Chronic inflammation and oxidative stress in asthma could influence DNA and RNA repair requirements, potentially affecting methylation patterns and metabolite levels, including 1,3-dimethyluracil. These observations are supported by murine models showing elevated oxidative DNA damage markers and DNA repair proteins under inflammatory conditions [27]. The potential diagnostic role of 1,3-dimethyluracil in asthma warrants further investigation to clarify its association with disease progression.

5-Methoxytryptophan (5-MTP), a tryptophan metabolite, was upregulated in the disease groups. Although the relationship between 5-MTP and asthma is still being explored, current evidence suggests that elevated 5-MTP levels represent a compensatory anti-inflammatory response rather than a causal factor [28]. 5-MTP has been proposed as an endogenous mediator that may help counteract inflammation and tissue remodeling, although its role in asthma remains to be further investigated. Previous studies have suggested that 5-MTP belongs to a novel group of endothelium-derived protective metabolites that safeguard against endothelial barrier disruption and excessive systemic inflammatory responses [29]. Our results indicate higher 5-MTP levels in asthmatic patients, which may reflect an association with inflammatory burden and tissue remodeling in asthma. However, a compensatory role cannot be determined from the present data. When comparing severe asthmatic patients with healthy controls, 5-MTP levels were found to be significantly elevated (P-value < 5). This marked increase demonstrates that in the context of severe disease, the body may upregulate 5-MTP as a compensatory mechanism to counteract heightened airway inflammation and structural remodeling. The 5-MTP levels were also elevated in patients with moderate asthma compared with healthy controls; however, the magnitude of the increase was smaller than that observed in patients with severe asthma. This graded response may represent a compensatory mechanism in response to increasing airway inflammation and remodeling, though causality cannot be confirmed.

Moreover, the results showed dysregulation in several amino acid metabolism pathways, including glycine and serine metabolism, arginine and proline metabolism, histidine metabolism, and arginine biosynthesis. In addition, enrichment analysis identified dysregulation of the urea cycle in asthmatic children. This biochemical cycle plays a vital role in detoxifying ammonia generated from amino acid catabolism by converting it into urea [30]. Arginine is an intermediate in the urea cycle, and its dysregulation is associated with alterations in L-arginine metabolism. It has been reported that severe asthma is frequently associated with increased arginase activity, which depletes L-arginine levels and elevates the risk of exacerbations [31]. As the primary substrate for nitric oxide (NO) synthesis, L-arginine is essential for maintaining airway tone, regulating immune responses, and controlling inflammation [32]. In asthmatic airways, excessive arginase activity limits L-arginine availability, leading to reduced protective NO production, increased oxidative stress, and structural airway changes. These metabolic disturbances have been reported to be associated with airway hyperresponsiveness and inflammation; however, in the present study, their contribution may also be influenced by medication use and other clinical factors [33]. Consequently, measuring L-arginine levels may serve as a valuable biomarker of disease severity and as a potential therapeutic target.

Similarly, disruptions in glutamine metabolism were observed in asthmatic patients, with downstream effects on phenylacetate metabolism. A previous study found that children with severe asthma exhibit markedly reduced systemic glutamine levels compared to those with mild to moderate asthma [34]. Reduced glutamine levels are associated with elevated exhaled nitric oxide, a marker of airway inflammation, as well as a higher proportion of eosinophils in the blood [34].

The findings of this study provide a comprehensive characterization of the metabolomic alterations associated with pediatric asthma. By examining metabolomic profiles in both moderate and severe asthma, the study identifies potential biomarkers associated with disease progression and severity.

Several limitations should be considered when interpreting the results of this study. The modest sample size and cross-sectional design limit generalizability and preclude assessment of temporal or causal relationships. In addition, serum metabolomic profiles may not fully capture local inflammatory and metabolic processes occurring within the airways. Although systemic metabolic alterations often accompany airway inflammation in asthma, the identified serum signatures likely reflect a composite of disease-related, systemic inflammatory, and treatment-related effects rather than direct measures of airway-specific pathology.

Another important limitation is the absence of children with mild asthma. As most participants were recruited from a tertiary care setting, the findings primarily characterize metabolomic alterations associated with moderate-to-severe pediatric asthma and do not allow evaluation of whether metabolic changes occur gradually from health to mild disease or represent a more dichotomous transition.

Furthermore, asthma severity classification relied on clinical assessment and ACT scores, and the lack of systematically collected lung function parameters (e.g., FEV₁ or FEV₁/FVC) limits direct correlation between metabolomic alterations and objective measures of airway obstruction or reversibility. Asthma severity remains a heterogeneous construct influenced by multiple factors, including treatment intensity, symptom perception, allergic comorbidities, diet, BMI, and variability in airway inflammation. The limited availability of objective clinical and inflammatory biomarkers in our cohort (e.g., eosinophil counts, IgE, or longitudinal lung function parameters) restricts the ability to fully characterise underlying asthma phenotypes and may therefore constrain the interpretation of metabolomic differences observed between moderate and severe asthma groups. In addition, differences in asthma control (as measured by ACT scores) may reflect varying acute inflammatory states, which could independently influence systemic metabolomic profiles. Taken together, these limitations indicate that the present findings are exploratory and warrant validation in larger longitudinal studies that incorporate detailed clinical phenotyping and airway-derived samples.

Conclusions

The study identified distinct metabolomic profiles in asthmatic children compared to healthy controls, and an overlapping profile between moderate and severe asthmatics, highlighting the complexity and specificity of the metabolic pathways involved in asthma. A total of 26 metabolites were dysregulated, with 23 upregulated and 3 downregulated in asthmatic children compared to controls. These findings suggest that several circulating metabolites are associated with asthma and its severity in moderate-to-severe pediatric cases; however, their utility in clinical practice as diagnostic or prognostic biomarkers across the full spectrum of asthma requires further validation in independent cohorts that include children with mild disease.

Abbreviations

AUC

Area Under the Curve

ANOVA

Analysis of Variance

BMI

Body Mass Index

CI

Confidence Interval

CV

Coefficient of Variation

ESI

Electrospray Ionization

FDR

False Discovery Rate

FEV₁

Forced Expiratory Volume in One Second

FVC

Forced Vital Capacity

HILIC

Hydrophilic Interaction Liquid Chromatography

HMDB

Human Metabolome Database

ICS

Inhaled Corticosteroids

LABA

Long-Acting Beta-Agonist

LC–MS

Liquid Chromatography–Mass Spectrometry

MS/MS

Tandem Mass Spectrometry

OCS

Oral Corticosteroids

PCA

Principal Component Analysis

PLS-DA

Partial Least Squares Discriminant Analysis

QC

Quality Control

QC-RLSC

Quality Control–Based Robust LOESS Signal Correction

ROC

Receiver Operating Characteristic

SABA

Short-Acting Beta-Agonist

SD

Standard Deviation

SE

Standard Error

UHPLC

Ultra-High Performance Liquid Chromatography

VIP

Variable Importance in Projection

Authors’ contributions

S.M.A. conceived the study idea, developed the concept, and led the manuscript writing. M.A.-I. and E.A.Z. contributed to patient recruitment, data collection, and revising the final version of the manuscript. Y.A.H. and B.M.S. performed metabolomics and data analysis and revised the manuscript. F.A.A. was responsible for sample collection. A.Y.A., W.E.-H., Z.M.A., E.A.-G., K.H.A., and Y.B. critically reviewed and approved the final manuscript. M.H.S. supervised the metabolomics analysis, conceived the study idea, critically reviewed, and approved the final manuscript.

Funding

This work has been funded by the Deanship of Scientific Research, University of Jordan. Grant number: 1011/2023/19.

Data availability

All data generated or analyzed during this study are included in this published article [and its supplementary information files manuscript.

Declarations

Ethics approval and consent to participate

All procedures in this study adhered to the ethical standards outlined in the Declaration of Helsinki and the guidelines set forth by the International Conference on Harmonization Clinical Practice (ICH-GCP). Parents of all participants were required to provide written informed consent, indicating their approval of their children's involvement in the study. The Institutional Review Board (IRB) at the hospital and the University of Jordan reviewed and approved the study protocol (approval no. 10/2023/26529).

Consent for publication

All authors have read and agreed to the published version of the manuscript.

Competing interests

The authors declare no competing interests.