Nasopharyngeal Dysbiosis Precedes the Development of Lower Respiratory Tract Infections in Young Infants, a Longitudinal Infant Cohort Study

Abstract

Background: Infants suffering from lower respiratory tract infections (LRTIs) have distinct nasopharyngeal (NP) microbiome profiles that correlate with severity of disease. Whether these profiles precede the infection or are a consequence of it, is unknown. In order to answer this question, longitudinal studies are needed.

Methods: We conducted a retrospective analysis of NP samples collected in a longitudinal birth cohort study of Zambian mother-infant pairs. Samples were collected every two weeks from 1-week through 14-weeks of age. Ten of the infants in the cohort who developed LRTI were matched 1:3 with healthy comparators. We completed 16S rRNA gene sequencing on the samples each of these infants contributed and compared the NP microbiome of the healthy infants to infants who developed LRTI.

Results: The infant NP microbiome maturation was characterized by transitioning from Staphylococcus dominant to respiratory-genera dominant profiles during the first three months of life, similar to what is described in the literature. Interestingly, infants who developed LRTI had distinct NP microbiome characteristics before infection, in most cases as early as the first week of life. Their NP microbiome was characterized by the presence of Novosphingobium, Delftia, high relative abundance of Anaerobacillus, Bacillus, and low relative abundance of Dolosigranulum, compared to the healthy controls. Mothers of infants with LRTI also had low relative abundance of Dolosigranulum in their baseline samples compared to mothers of infants that did not develop an LRTI.

Conclusions: Our results suggest that specific characteristics of the NP microbiome precede LRTI in young infants and may be present in their mothers as well. Early dysbiosis may play a role in the causal pathway leading to LRTI or could be a marker of underlying immunological, environmental, or genetic characteristics that predispose to LRTI.

Keywords: Dysbiosis; Longitudinal Cohort study; Lower Respiratory Tract Infection; Nasopharyngeal Microbiome.

Figure 1.

The maturation of the NP microbiomes of A) healthy, asymptomatic infants (n=30), and B) LRTI infants (n=10) over three months of observation. These stacked bar plots show the average relative abundance of the most common genera found in infant NPs, with samples binned by age.

Figure 2.. Nonmetric multidimensional scaling (NMDS) ordination plots of all infants’ (n=40) nasopharyngeal (NP) samples.

We applied 3-dimensional NMDS ordination to the Bray-Curtis dissimilarity matrix between all infants’ NP swabs, and projected vectors into that ordination space representing the best fit correlations for the age at sampling (the black arrows) and LRTI status (the cyan arrows represent control infants). Age is highly correlated with the first NMDS axis, and samples on the young end of the age vector mostly belong to the Staphylococcus-dominated profile, whereas samples on the older end tend to belong more to the Haemophilus and Moraxella-dominated profiles. The Dolosigranulum-dominated profile is associated with the healthy end of the vector for LRTI status, while the Anaerbacillus-dominated profile is associated with disease.

Figure 3.. Relative abundance Z-scores of specific genera of LRTI infants compared to healthy controls (light blue), by weeks from infection.

Figure 4.. The taxonomic profiles of all infant NP samples (n=40), clustered by pairwise Bray-Curtis dissimilarity.

Clusters were defined by performing hierarchical clustering on the beta diversity matrix and then cutting the resulting dendrogram into an optimal number of clusters according to the A) Silhouette index (6 clusters) and B) Frey index (13 clusters). The color bars above the stacked bar plots indicate the infants’ ages at the time of each sample and their LRTI status – “healthy” indicates an infant which did not develop LRTI symptoms during our observation.

Figure 5.. Stacked bar plots showing the average relative abundance of the most common genera found in mothers NPs at first time point.