The airway microbiota of neonates colonized with asthma-associated pathogenic bacteria

Asthma is a common childhood chronic disease influenced by prenatal and early-life factors, many of which relate to microbial exposures. Prior prospective cohort studies have associated early airway and gut microbiota with later asthma. In the COPSAC2000 cohort, hypopharyngeal colonization at 1 month by pathogenic bacteria (Streptococcus pneumoniae, Haemophilus influenzae, Moraxella catarrhalis) detected by culture was linked to asthma by age 5. In the later COPSAC2010 cohort, 16S rRNA gene sequencing of hypopharyngeal aspirates associated anaerobic taxa, notably Veillonella and Prevotella, with asthma by age 6. It has remained unclear whether discrepancies across studies reflect differences in sampling site (nasopharynx vs hypopharynx), detection method (culturing vs sequencing), or cohort characteristics. The present study analyzes archived 1-month nasopharyngeal swabs from COPSAC2000 using 16S rRNA gene sequencing to test whether sequencing-based measures from nasopharyngeal samples recapitulate the previously reported culture-based pathogenic bacteria association with asthma, and to assess whether other taxa, especially Veillonella and Prevotella, are associated with later asthma.

Background literature indicates that early-life airway microbiota is linked to subsequent wheeze/asthma across multiple cohorts. The original COPSAC2000 culture-based study associated neonatal hypopharyngeal colonization by S. pneumoniae, H. influenzae, and M. catarrhalis with asthma and exacerbations by age 5. In COPSAC2010, hypopharyngeal 16S sequencing associated Veillonella and Prevotella with asthma by age 6. The nasopharyngeal and hypopharyngeal niches exchange microbes yet may be colonized by different species. Culture-independent techniques, such as 16S sequencing, detect more taxa, including non-cultivable anaerobes. Other cohorts (CAS, MARC-35, COAST, DORMICe, STEPS) have variously associated early colonization dominated by Moraxella, Streptococcus, Haemophilus, or Staphylococcus with later wheeze/asthma, underscoring a general link between early airway microbiota and asthma risk, while leaving uncertainties about the specific causal taxa versus host or functional drivers.

Study design and population: The study used samples from the COPSAC2000 prospective birth cohort (n=411) of children born to mothers with doctor-diagnosed asthma in Copenhagen, Denmark. Exclusions included severe congenital abnormalities, preterm birth <36 weeks, lung symptoms, or need for mechanical ventilation before enrollment at 1 month of age (1998–2001). Children were followed at 1 and 6 months, then every 6 months to age 7, with acute visits for respiratory illness. Asthma-related endpoints were diagnosed solely by COPSAC physicians using structured, quantitative algorithms and parental daily symptom diaries. Sampling: Nasopharyngeal swabs were collected at ~4 weeks from asymptomatic neonates (1999–2001) by placing and rotating a cotton-tipped aluminum swab in the posterior nasopharynx, then immersed in SP4 mycoplasma transport medium. Samples were kept at +4 °C until lab delivery the next day for mycoplasma culture, and remaining material was immediately frozen at −80 °C and stored ~20 years until DNA extraction in 2021. Hypopharyngeal aspirates were collected the same day as swabs in the original protocol. DNA extraction and sequencing: Microbial DNA was extracted from 250 µl saline using the NucleoSpin 96 Soil kit on an epMotion robotic platform. Negative controls (PBS, molecular-grade water) and a mock community positive control were included. The 16S rRNA gene V3–V4 regions were amplified using primers 341F/Uni806R with Phusion High-Fidelity polymerase (30 cycles; second PCR 15 cycles to add adapters). Amplicons were purified (AMPure XP), normalized (SequalPrep), pooled, concentrated (Zymo Clean & Concentrator), quantified (NanoDrop), and sequenced on Illumina MiSeq (2×300 bp, v3 kit; 8% PhiX). Bioinformatics: Reads were demultiplexed, adapters removed with Cutadapt, and ASVs inferred using QIIME2/DADA2. Taxonomic annotation used the AnnotIEM pipeline combining EzBioCloud, NCBI RefSeq, RDP, and SILVA, with high-confidence species calls; performance validated on a mock community (19/20 species correctly annotated, one at genus level). ASVs mapping to human GRCh38 (Bowtie2 very-sensitive) were removed. Potential contaminants were identified with decontam (prevalence mode, probability >0.5), retaining typical airway/skin taxa (e.g., Cutibacterium acnes, Dolosigranulum pigrum, H. influenzae, M. catarrhalis, S. aureus, S. epidermidis, S. hominis, S. pneumoniae) to avoid over-filtering. After QC, 285 nasopharyngeal samples with ≥2000 reads (median depth 60,723; IQR 45,384–79,958) and 4023 ASVs remained; median richness 36 ASVs (IQR 23–48). Four samples were removed for low read depth. Clinical endpoints: Wheezy episodes (≥3 days of cough/wheeze/breathlessness), recurrent wheeze (≥5 episodes in 6 months or 4 weeks continuous symptoms), persistent wheeze/asthma (recurrent wheeze responding to 3-month ICS and relapsing upon cessation), and exacerbations requiring OCS/high-dose ICS or hospitalization. Statistical analysis: Baseline comparisons used Chi-squared, Fisher’s exact, or Wilcoxon rank-sum tests. Species-level relative abundance from nasopharyngeal sequencing was compared to hypopharyngeal culture positivity for S. pneumoniae, H. influenzae, M. catarrhalis, and S. aureus using Wilcoxon tests and ROC AUC. A pathogen score was computed as the sample-wise sum of relative abundances of ASVs annotated as S. pneumoniae, H. influenzae, and M. catarrhalis, log-transformed with a pseudocount; tertiles used for KM visualization. Associations with time-to-event endpoints (e.g., persistent wheeze/asthma to age 7) were tested via Cox proportional hazards regression (sex-stratified and interaction tested). Adjusted models included siblings, delivery mode, living environment, maternal smoking, season of birth, sex, maternal antibiotic use during pregnancy, breastfeeding at 1 month, and gestational age. Alpha diversity (Shannon, Faith’s PD) and beta diversity (unweighted and weighted UniFrac; PERMANOVA via adonis2) were analyzed adjusting for library size and run. Differential abundance at species level used Cox regression (log2 relative abundance) and DESeq2, adjusting for log(library size) and sequencing run; FDR controlled by Benjamini–Hochberg. Taxon co-occurrence/correlation used Spearman and SparCC with hierarchical clustering; p-values by 1000 bootstrap resamples. Resolution within Moraxella was assessed with RibDif. Comparative data: Sequencing outcomes were compared to hypopharyngeal cultures in COPSAC2000 and to hypopharyngeal V4 16S sequencing in COPSAC2010 (processed with the same DADA2 + AnnotIEM pipeline).

• Sequencing–culture concordance: Among 244 children with both datasets, nasopharyngeal relative abundances were strongly higher in culture-positive vs culture-negative children for the targeted species (Wilcoxon p<0.0001 for all): • S. pneumoniae median [IQR] 7.18% [0.53; 21.1] vs 0.03% [0.00; 0.62] • H. influenzae 0.00% [0.00; 42.4] vs 0.00% [0.00; 0.00] • M. catarrhalis 3.05% [0.42; 11.1] vs 0.00% [0.00; 0.00] • S. aureus 44.2% [0.45; 79.0] vs 0.15% [0.08; 0.32] High species-level specificity was observed when comparing culture results with top ASVs in the same genus.
• Pathogen score and asthma: The summed relative abundance (pathogen score) of S. pneumoniae, H. influenzae, and M. catarrhalis was associated with increased risk of persistent wheeze/asthma by age 7 (Cox HR per SD 1.50, 95% CI 1.12–2.01, p=0.0066, n=285). Results were similar by sex (male HR 1.52 [1.07–2.16], p=0.02, n=144; female HR 1.58 [0.94–2.65], p=0.08, n=141; interaction p=0.90). Adjustment for covariates yielded aHR 1.48 [1.06–2.05], p=0.02 (n=241).
• Independence from culture: The association disappeared in children without cultured pathogenic bacteria (HR 0.93 [0.60–1.46], p=0.44, n=188) and when adjusting for culture positivity (HR 1.06 [0.73–1.52], p=0.76). Conversely, the culture–asthma association remained significant when adjusting for the pathogen score (aHR 3.50 [1.69–7.25], p=0.0007, n=244), suggesting culture positivity was a stronger predictor.
• Other endpoints: The pathogen score associated with recurrent wheeze and early exacerbations; for hospitalization by age 5, culture-defined pathogen colonization showed HR 3.88 [1.92–7.86], p=0.00016 (n=319), whereas the pathogen score showed only a non-significant trend.
• Diversity and composition: No significant associations between alpha diversity (Shannon HR 0.72 [0.42–1.22], p=0.22; Faith’s PD HR 1.02 [0.88–1.18], p=0.78; n=285) or beta diversity (PERMANOVA unweighted UniFrac F=1.21, p=0.17; log-weighted UniFrac F=1.17, p=0.25; n=221) and asthma. Differential abundance analyses (Cox and DESeq2) identified S. pneumoniae and H. influenzae as significant after FDR correction; M. catarrhalis and Moraxella lincolnii were nominal but did not pass FDR.
• Veillonella/Prevotella: Detection in COPSAC2000 nasopharyngeal swabs was lower than in COPSAC2010 hypopharyngeal aspirates (Veillonella 40.4% vs 77.2%; Prevotella 17.9% vs 42.3%). Neither genus was associated with persistent wheeze/asthma or differentially abundant at genus level. However, both correlated positively with the pathogen score (Veillonella Spearman ρ=0.25, p<0.0001; Prevotella ρ=0.14, p=0.016; n=285) and clustered with Streptococcus and Haemophilus in correlation analyses, suggesting latent community structures.
• Sequencing performance and sample metrics: 285 samples successfully sequenced (median depth 60,723 reads; 4023 ASVs; median richness 36 ASVs).

This study shows that early-life nasopharyngeal 16S rRNA gene sequencing recapitulates the earlier hypopharyngeal culture-based association between pathogenic bacteria and childhood asthma in the high-risk COPSAC2000 cohort. A pathogen score derived from nasopharyngeal relative abundances of S. pneumoniae, H. influenzae, and M. catarrhalis associated with persistent wheeze/asthma and exacerbations. However, when jointly modeled with culture data or restricted to culture-negative children, the sequencing-based association attenuated, indicating that culture positivity captured the key predictive information, possibly reflecting differences between relative and absolute abundance metrics. No broad community diversity signals were associated with asthma, and beyond the cultured species, only M. lincolnii appeared nominally associated and did not survive FDR correction. The lack of replication for Veillonella and Prevotella as asthma-associated taxa in these nasopharyngeal samples likely reflects sampling-site differences and long-term storage/processing effects; nevertheless, their positive correlation and clustering with Streptococcus and Haemophilus suggest overlapping ecological or environmental determinants linking these taxa to pathogen colonization and asthma risk observed in other cohorts. Across multiple international cohorts, varied taxa and community types have been linked to wheeze/asthma, underscoring a robust association between early airway microbiota and later respiratory outcomes. Yet, mechanistic causality remains uncertain—whether microbes drive disease trajectories or reflect underlying host susceptibilities. These findings reinforce the association’s robustness across methods and sites while highlighting the need for studies dissecting host–microbe interactions and microbial functions to inform preventive strategies.

Nasopharyngeal colonization at 1 month, measured by 16S sequencing and summarized as a pathogen score from S. pneumoniae, H. influenzae, and M. catarrhalis, was associated with persistent wheeze/asthma by age 7, recapitulating prior culture-based hypopharyngeal findings in COPSAC2000. Veillonella and Prevotella were less frequently detected in nasopharyngeal swabs than in hypopharyngeal aspirates from COPSAC2010 and were not directly associated with asthma but correlated with the pathogen score, suggesting shared community structures. Future work should focus on temporospatial dynamics of upper airway colonization, absolute abundance measurements, host–microbe interactions, and mechanistic models to establish causality and identify targets for early-life asthma prevention.

Key limitations include inherent 16S rRNA gene sequencing constraints (copy number variation, amplification bias, read errors, contamination) and the inability to detect viruses and fungi. Species-level annotations via AnnotIEM are putative, though validated against controls. Cohort comparisons are limited by different primer sets and targeted 16S regions (V3–V4 vs V4). The nasopharyngeal swabs were biobanked for ~20 years and underwent mycoplasma culture before freezing, potentially causing selective degradation of certain taxa (e.g., anaerobes), which could skew composition and preclude detection of some associations, especially when comparing with COPSAC2010. Original COPSAC2000 hypopharyngeal samples were unavailable for parallel sequencing, limiting direct site/method comparisons. The study was not designed to compare sampling locations or methods. Residual confounding from differences between cohorts (maternal asthma selection, geography, smoking rates, vaccination era changes affecting S. pneumoniae epidemiology) cannot be excluded. Relative abundance analyses may miss effects linked to absolute bacterial load.