Temporal interactions of microbiota in longitudinal nasopharyngeal samples and association with lower respiratory tract infection

Abstract

During aetiology of respiratory illnesses, it is widely accepted that infection is preceded by nasopharyngeal (NP) colonisation with bacteria and that NP flora develop early in childhood (during the first year of life). The presence of multiple NP bacteria results in competitive and synergistic associations, however temporal organism interactions have rarely been explored due to limited availability of longitudinal data sets, and the complex statistical methods needed. This study aimed to identify, describe and quantify the temporal interactions existing between selected key bacteria colonizing the nasopharynx in young children (up to 1 year old), and to further compare these patterns in children who go on to develop pneumonia compared to those who do not. The significance of the study, as well as the objectives of the study, methods and data analysis plan are outlined in the study protocol (Part A). A summary of what is currently known about NP bacterial species interactions is presented as part of the literature review (Part B). The primary aim of the literature review was to describe the prevalence of NP carriage of four NP colonizing bacteria of interest: S. pneumoniae, S. aureus, H. influenzae and M. catarrhalis in children, as well as identify any risk factors or confounding associations. The literature review furthermore aimed to identify previously described NP bacterial species interaction patterns, as well as providing a summary of statistical approaches previously employed in the studying bacterial interactions. A manuscript presenting the subsequent analysis of these data is included as Part C. This study was a secondary data analysis of 760 infants enrolled in a birth cohort with NP swabs collected every two weeks for the first year of life and additionally at episodes of lower respiratory tract infections (LRTI). Kaplan-Meier estimates were used to visualize time to first carriage. Generalised estimating equations with a logit link and adjusted for repeated measures were used to estimate the time varying association of NP bacteria carriage with development of pneumonia, while enabling adjustment by key confounders. Markov multi state models (MSMs) were used to describe NP bacterial acquisition with age and estimation of clearance probabilities, new acquisition or persistent acquisition. There were 760 individuals included in the analysis, with a total of 16,346 NP samples available and a median 364 person-days (IQR 346 – 365 person-days). S. pneumoniae was predominant, found in >55% of all samples and demonstrating carriage in >95% of individuals at least once by 12 months of age. S. aureus was both less common (25% of samples and 88% of individuals) but also had a strikingly different pattern of first acquisition compared to the other three organisms, demonstrating a rapid increase in carriage prevalence until approximately four weeks and subsequently decreasing. S. pneumoniae had the highest co-carriage prevalence overall with H. influenzae and M. catarrhalis (both 25%) but this varied by age category. In contrast, co-carriage with S. aureus was less prevalent with either S. pneumoniae (12%), H. influenzae (5%) or M. catarrhalis (6%). Co-carriage frequencies differed considerably by age category, at least partially reflecting the relative prevalence of carriage by age. Carriage and co-carriage rates were similar among those children that experienced LRTI compared to those that did not. Seasonal carriage varied, but to a small extent compared to variance by age. Models adjusting for sex, site, season of birth and age found temporally sustained positive associations between the co-carriages of S. pneumoniae with H. influenzae, and M. catarrhalis, but no association with S. aureus. Clear differences occur in the co-carriage patterns of S. pneumoniae with other organisms. The probability of acquisition of S. pneumoniae is modified by earlier carriage of H. influenzae or M. catarrhalis. Positive H. influenzae carriage increases the probability of acquisition of S. pneumoniae with transition probabilities from 0.15 (95% CI 0.14-0.17) to 0.36 (95% CI 0.17, 0.54) after 28 days of age, compared to the same period probability of acquisition of S. pneumoniae alone at 0.015 (95% CI 0.043-0.076) to 0.088 (95% CI 0.075- 0.10). There is no difference in the clearance of S. pneumoniae related to H. influenzae carriage, but clearance of H. influenzae before 6 months of age is far less likely if coming from a state of co-carriage (probability between 0.04 - 0.07) compared to sole carriage (probability 0.23 - 0.12). The only evidence of differences in clearance probability in the models investigating S. pneumoniae and M. catarrhalis are in the probability of M. catarrhalis clearance before 28d which is 0.24 (95% CI 0.15 - 0.38) if carried alone and only 0.058 (55%CI 0.01 - 0.30) if carried with S. pneumoniae, though these confidence intervals overlap.Through this modelling we found positive sustained interactions between S. pneumoniae and both H. influenzae, and M. catarrhalis, where models indicated that preceding carriage or colonisation with either H. influenzae, and M. catarrhalis may increase the risk of colonisation with S. pneumonia. Timing of carriage and overall prevalence of carriage are in line with other findings in similar populations with overall high exposure to S. pneumoniae, H. influenzae, M. catarrhalis during the first year of life and rapid and early exposure to S. aureus. Carriage, co-carriage and transition frequency did not vary appreciably when comparing children who experienced LRTI in the first year of life compared to those who did not, suggesting that overall exposures are similar, but that further modelling is required to understand the specific timing of associations in relations to LRTI.