New Insights Into the Role of Antibiotic Use in Infancy and the Upper Airway Microbiome in Childhood Asthma Development

(See the Major Article by Toivonen et al on pages 1546–54.)

The current body of research suggests a detrimental effect of antibiotic use in infancy on childhood asthma [1, 2], a disease that affects approximately 8% of the pediatric population in the United States and is one of the leading causes of healthcare utilization and work and school absenteeism in this country [3]. However, little is known about the potential mechanisms underlying this association. Furthermore, the reported associations of antibiotic use in infancy with childhood asthma may not be truly causal but instead explained by reverse causation (as infants at high risk of or with childhood asthma could receive more antibiotics than those without or at low risk of childhood asthma), recall bias (particularly for retrospective studies), and/or confounding by indication (ie, as increased antibiotic use could be just a marker of recurrent acute respiratory infections [ARIs], which could, in turn, be the true risk factor for childhood asthma).

In this issue of Clinical Infectious Diseases, Toivonen et al [4] use data from a population-based birth cohort study in Finland (N = 697) to (1) examine the association of antibiotic use in infancy with childhood asthma and (2) assess whether the upper airway microbiome is a mediator of this association. The findings of this well-conducted study confirm the findings of previous studies in this field that have shown a positive relationship between antibiotic use in infancy and the onset of childhood asthma [1, 2]. What is novel and important is the inclusion of nasal samples at ages 2, 13, and 24 months to characterize the upper airway microbiome throughout early life and the application of causal inference methodology to determine whether longitudinal changes in the upper airway microbial patterns mediate the association of antibiotic use in infancy and childhood asthma. The authors conclude that they do (particularly through a reduction in the abundance of taxa from the genus Moraxella), but the effect is modest, accounting for approximately 16% of the total effect. They also found similar results in several sensitivity analyses.

The study by Toivonen et al addresses a very important question, has a clear and novel hypothesis, a prospective design that minimizes the risk of reverse causation and recall bias, and a statistical approach that decreases the risk of confounding by indication through adjusting for the number of ARIs in infancy, thus addressing many of the limitations of prior observational studies in this field. Furthermore, the authors use a causal mediation framework to explore potential underlying pathways, as well as next-generation sequencing and new bioinformatic techniques to take into consideration the longitudinal development of the upper airway microbiome in infancy, which—unlike the gut microbiome—has been relatively understudied.

The immediate question we could ask is, if the association between antibiotic use in infancy and childhood asthma is only partly mediated by changes in the development of the upper airway microbiome, what else explains this association? To answer this question, we should consider several points. First, is it still possible that the relationship is still not truly causal, due to residual confounding by measured or unmeasured variables. For example, although the authors adjusted for the number of ARIs in infancy, they did not include data on the precise etiology or severity of these ARIs, and it has been shown that severe ARIs by specific respiratory viruses (such as human rhinovirus or respiratory syncytial viruses) could lead to a higher risk of childhood asthma [5]. Furthermore, there remains the question of whether there is a shared genetic influence on the upper airway microbiome and susceptibility to ARIs, as individuals with asthma have been shown to be at increased risk of certain types of bacterial infections. Thus, there could be a shared genetic susceptibility between the risk of ARIs and the use of antibiotics in infancy, early-life microbiome profiles, and childhood asthma, which confounds the relationships identified in this and other observational studies. The authors did not have nasal samples at birth, which would have been helpful to assess if the upper airway microbiome patterns were similar between groups. Second, the true proportion of the total effect explained by changes in the development of the upper airway microbiome could be higher. This could be due to the fact the authors only examined the nasal microbiome but no other upper airway communities (such as the nasopharyngeal or oropharyngeal microbiome) that have also been shown to be relevant for other childhood respiratory outcomes (including childhood asthma) [6–8]. Likewise, the use of antibiotics in infancy could impact other microbial ecosystems not assessed in this study (such as the lower airway or gut microbiomes). In addition, although the authors found 6 different upper airway microbiome profiles, they had to focus on one of these due to the complexity of human microbiome data and to simplify mediation analyses, which may have led to biased estimates. In the same context, their statistical analyses only focused on the β-diversity (ie, overall structure) of the upper airway microbiome, but other important metrics of this community were not considered (such as the α-diversity [richness ± evenness] or individual taxa abundances). Of note, other studies using next-generation sequencing have shown that upper airway microbiome profiles characterized by a low abundance of Moraxella in infancy are actually associated with a higher risk of some pediatric wheezing illnesses, so how exactly this micro-organism impacts childhood respiratory health remains unclear [8]. Last, it is possible that the use of antibiotics in infancy leads to childhood asthma through microbiome-independent anti-inflammatory or immune-modulatory effects [9].

The randomization of infants to antibiotics cannot be done, so it is likely that future evidence of an association between antibiotic use in infancy and childhood asthma will continue to be obtained from observational studies. Establishing whether this association is truly causal or not is important, as the misuse of antibiotics in early life is widespread, the prevalence of childhood asthma and its impact on the healthcare system are high, and it has been postulated that the use of antibiotics in infancy could be responsible for approximately 15% of all cases of childhood asthma [10]. The adoption of a causal mediation framework, such as the one used by Toivonen et al, in future observational studies could be extremely helpful to generate strong epidemiological evidence to support a causal link. If a causal association is established and a specific window of opportunity to act is identified, then healthcare policies targeting the use of antibiotics in early life could have an enormous public health impact. For health policy makers and clinicians, these findings lend additional support to current guidelines on antibiotic stewardship and prevention of antibiotic-related adverse effects.

Notes

Financial support. C. R. -S. reports grants from the National Institutes of Health (NIH) (K23HL148638) and The Francis Family Foundation Parker B. Francis Fellowship, outside the submitted work. T. V. H. reports grants from the NIH (U19AI095227 and R01AI136526) and World Health Organization, and an advisory board payment from Pfizer, outside the submitted work.

Potential conflicts of interest. The authors: No reported conflicts of interest. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.