Investigation of associations between the neonatal gut microbiota and severe viral lower respiratory tract infections in the first 2 years of life: a birth cohort study with metagenomics

Abstract

Background

Early-life gut microbiota affects immune system development, including the lung immune response (gut–lung axis). We aimed to investigate whether gut microbiota composition in neonates in the first week of life is associated with hospital admissions for viral lower respiratory tract infections (vLRTIs).

Methods

The Baby Biome Study (BBS) is a prospective birth cohort, which enrolled mother–baby pairs between Jan 1, 2016, and Dec 31, 2017, at three UK hospitals. In the present study, we only included BBS babies with a sequenced first-week stool sample and successful data linkage. Stool was collected in the first week of life for shotgun-metagenomic sequencing. We examined the following microbiota features: alpha diversity (Chao1, Shannon, and Simpson indices) and community structures (cluster-partitioning against medoids method). The participants were followed up through linkage to the Hospital Episode Statistics-Admitted Patient Care (HES-APC) database to determine vLRTI hospital admission incidence in the first 2 years of life. We used Poisson mixed-effects models for univariable and multivariable analyses to evaluate the association between microbiota features and vLRTI hospital admission incidence, adjusting for confounders identified through direct acyclic graphs.

Findings

3305 (95%) of the 3476 BBS-enrolled babies for whom consent to data linkage was obtained were included in the present study. 1111 (34%) babies had a first-week sequenced stool sample, of whom 1082 (97%; 564 born vaginally and 518 born by caesarean section) were successfully linked to HES-APC, and had median follow-up of 2·0 years (IQR 1·4–2·9). Most babies were born at term (996 [92%] ≥37 weeks gestational age and 1070 [99%] >35 weeks gestational age) and healthy (1050 [97%] had no comorbidities), and 520 (48%) were female and 562 (52%) were male. Higher first-week gut microbiota alpha diversity was associated with reduced rates of vLRTI hospital admission (Chao1 Index adjusted hazard ratio [HR] 0·92 [95% CI 0·85–0·99]; Shannon Index adjusted HR 0·57 [0·33–0·98]; and Simpson Index adjusted HR 0·36 [0·11–1·20]). Three microbiota clusters were identified. Cluster 1 had a mixed composition and cluster 2 was dominated by Bifidobacterium breve, with both clusters observed in babies born vaginally and by caesarean section. Cluster 3 was found only in vaginally born babies and was dominated by Bifidobacterium longum. Having cluster 1 (mixed) or cluster 2 (B breve dominated) was independently associated with increased rates of vLRTI hospital admission compared with cluster 3 (B longum dominated; cluster 1 [mixed] 3·05 [1·25–7·41] and cluster 2 [B breve dominated] 2·80 [1·06–7·44]).

Interpretation

We report observational evidence that first-week gut microbiota differences are associated with clinically severe vLRTI in young children. This study identified bacterial species that could be of interest for vLRTI prevention. This finding has important implications for the design of future research and intervention strategies.

Funding

The Wellcome Trust and Wellcome Sanger Institute core funding.