Nasopharyngeal Bacterial Interactions in Children

TOC summary: Pathogen prevalence differs during periods of health and at onset of acute otitis media.

Moraxella catarrhalis (2). The human nasopharynx is an ecologic reservoir of these and other bacteria. A broad variety of commensal bacteria and potential bacterial pathogens colonize the nasopharynx (3,4). Colonization of the nasopharynx is a fi rst, and essential, step toward development of respiratory bacterial infections (3). Viruses can join the microbial mix as a prelude to secondary bacterial infections of the respiratory tract (5)(6)(7).
More information about microbial interactions in the nasopharynx is needed (8). These interactions can be altered by therapeutic (e.g, antimicrobial drug) and vaccine (e.g.; pneumococcal conjugate vaccination) interventions, resulting in synergistic or competitive outcomes. Information about interactions of the major bacterial respiratory pathogens in the nasopharynx and the conditions conducive to progression to infection (e.g., concurrent viral upper respiratory infections) is limited.
Microbial species can interact synergistically to promote persistence of colonization (positive, or synergistic, association) or they can compete (negative, or competitive, association) (4,9). Interactions between bacteria can alter the composition of a microbial community and affect incidence of disease (4). Several studies have reported competitive associations between colonized S. pneumoniae and S. aureus in the nasopharynx of children, raising concerns that eradication of S. pneumoniae from the nasopharynx by the heptavalent pneumococcal conjugate vaccine (PCV7) might lead to increased S. aureus colonization and subsequent infections (10)(11)(12)(13). The introduction of the 13-valent pneumococcal conjugate vaccine will probably exacerbate this effect. Other variables that alter nasopharynx colonization patterns in children include age, gender, daycare attendance, history of having been breast-fed, environmental exposure to tobacco smoke, and otitis-prone condition (14,15). Several recent reports have described interactions among the 3 major pathogens-S. pneumoniae, H. infl uenzae, and M. catarrhalis-in young children (8,10,11,16), but the results were contradictory (9). We investigated the interactions of these 3 pathogens in the nasopharynx of young children while healthy (healthy visits) and at onset of AOM (AOM visits). Our aims were to understand differences in nasopharynx colonization rates and bacterial interactions according to the child's health status.

Study Design and Participants
We analyzed data collected during June 2006-May 2011 from children enrolled in a 5-year prospective study supported by the National Institute of Deafness and Other Communication Disorders. In that study, healthy children with no previous episodes of pneumonia, sinusitis, or AOM were enrolled at 6 months of age from 5 middle-class, suburban pediatric practices in Rochester, New York, USA. Nasopharyngeal and oropharyngeal samples were obtained from healthy children at 6,9,12,15,18, and 24 months of age and examined for S. pneumoniae, H. infl uenzae, M. catarrhalis, and S. aureus. If symptoms compatible with an AOM infection developed, a tympanocentesis was performed, as described, to confi rm the diagnosis (17). At the time of diagnosis, nasopharyngeal and oropharyngeal samples were obtained for bacterial pathogen cultures. All children received age-appropriate standard vaccinations, including pneumococcal conjugate vaccine (PCV7) (Prevnar; Wyeth Pharmaceuticals, Collegeville, PA, USA).
We analyzed culture data from nasopharyngeal samples collected during 1,183 healthy visits and 334 AOM visits among 320 children 6-24 months of age. All samples included in this study were from children who had not received antimicrobial therapy for at least 3 weeks. Nasopharynx colonization at healthy versus AOM visits was compared among children at 6,9,12,15,18, and 24 months of age. This time frame includes peak incidence of AOM infection caused by S. pneumoniae, H. infl uenzae, and M. catarrhalis.
Nasopharyngeal and oropharyngeal samples were obtained for culture as described (18). The pathogens S. pneumoniae, H. infl uenzae, M. catarrhalis and S. aureus were isolated and identifi ed according to the Manual of Clinical Microbiology (19).
The study was approved by the Institutional Review Board of the University of Rochester and the Rochester General Hospital. Written informed consent was obtained from parents or guardians before the children were enrolled.

Statistical Analyses
The rates of nasopharynx colonization among children of the same age at healthy and AOM visits were compared by using the Fisher exact test and GraphPad Prism software (www.graphpad.com). Bacterial interactions were analyzed by using repeated measures logistic regression models. Predicted outcomes of colonization with S. pneumoniae, H. infl uenzae, and M. catarrhalis were examined by using multivariate logistic regression. Generalized estimating equations were used to model exchangeable correlation within participants (20). Two logistic regression models (1 for healthy visits and 1 for AOM visits) were calculated by using R version 2.13.2 (www.r-project.org/). To examine the effects of covariates on each of the 3 pathogens, we modeled colonization of each pathogen separately by using the remaining 2 pathogens as predictors and including the interaction term (8). Because few S. aureus were isolated, we did not separately model colonization outcome for S. aureus (8). For each model, we estimated odds ratios (ORs) for the response pathogen given the presence of each predictor pathogen alone, then jointly; synergistic associations between bacteria are indicated by OR>1; competitive associations, by OR<1. The absence of both predictor pathogens was used as the reference condition. p<0.05 was considered signifi cant; p<0.01, strongly signifi cant; p<0.1, weakly signifi cant. The model was also used to estimate the OR for each pathogen relative to the risk factors of sex, age, daycare attendance, history of having been breast-fed, environmental exposure to tobacco smoke, and otitis-prone condition.

Polymicrobial Nasopharynx Colonization
Differences in nasopharynx colonization rates between healthy and AOM visits remained generally similar across the longitudinal samplings ( Table 1). The rate of nasopharynx colonization by S. pneumoniae for children of all ages at healthy visits, when neither H. infl uenzae nor M. catarrhalis was present, was 14.2%; this rate did not differ statistically from that at AOM visits (14.4%; p = 0.93) ( Table 1). In contrast, the rate of nasopharynx colonization by H. infl uenzae for children of all ages at AOM visits, when neither S. pneumoniae nor M. catarrhalis was present, was 19.5%; this rate was >4-fold higher than the rate at healthy visits (4.5%; p<0.0001) ( Table 1). The rate of nasopharynx colonization by M. catarrhalis for children of all ages at healthy visits, when neither H. infl uenzae or S. pneumoniae was present, was 20.6%; this rate was 2-fold higher than that for AOM visits (10.5%; p<0.0001) ( Table 1).
Polymicrobial colonization of the nasopharynx was signifi cantly less common at healthy visits than at AOM visits. At the onset of an infection, polymicrobial colonization increased by 1.5-2.8-fold overall for children of all ages (p<0.05 for all). At healthy visits, the proportion of children with polymicrobial nasopharynx colonization was 18.1% (214/1,183); whereas at AOM visits, the proportion was 45.5% (152/334) (p<0.0001). A comparison of singlepathogen colonization and polymicrobial colonization at ages 6, 9, 12, 15, 18, and 24 months and differences between healthy visits and AOM visits are shown in the Figure. The overall colonization rates for any of the 3 pathogens were 57.4% (679/1,183) at healthy visits and 89.8% (300/334) at AOM visits (p<0.0001). The culturepositive rates at healthy visits were 30 The higher prevalence of a potential otopathogen in the nasopharynx need not imply greater association. For example, the marginal rates for the pathogens in the healthy group (all ages) were 0.303, 0.117, and 0.363, the product of which is 0.0129 (Table 1). The actual rate for the occurrence of all 3 is 0.028, indicating an increase in association compared with chance. In contrast, the corresponding rates for the AOM group are 0.527, 0.479, 0.434, the product of which is 0.1096, compared with an actual rate of occurrence of 0.087 for all 3, indicating a decrease in association compared with chance. These totals are therefore compatible with the ORs reported in Table 2.  S. pneumoniae (OR 2.34, p = 0.016) and H. infl uenzae colonization (OR 3.1, p = 0.0017) at healthy visits but not at AOM visits. We found no signifi cant association between smoking exposures and colonization or between a history of breastfeeding and colonization at either type of visit (Table 2). Being male was positively associated with M. catarrhalis colonization (OR 1.94, p = 0.005) at AOM visits but not at healthy visits.

Discussion
We found that patterns of nasopharynx colonization associations in children differed at onset of AOM and when healthy. H. infl uenzae colonization was competitively associated with S. pneumoniae and M. catarrhalis colonization at AOM visits but not at healthy visits; the rates of nasopharynx colonization by H. infl uenzae were 4-fold higher during AOM visits than during healthy visits.
Our fi ndings suggest that during the PCV era, H. infl uenzae might increase as a bacterial pathogen of AOM. Our data among young children show that nasopharynx colonization studies of healthy children might not refl ect the polymicrobial mix at the time of onset of AOM; the nasopharyngeal environment during onset of AOM favors H. infl uenzae colonization. It is the mix of bacteria at time of infection that determines which organisms are most likely to cause infection (12). We have previously shown that among children, the changes in nasopharynx colonization by S. pneumoniae and H. infl uenzae caused by PCV7 resulted in a remarkable proportionate decrease in AOM infection caused by S. pneumoniae and a proportionate increase in  (17,21,22). At onset of AOM, when H. infl uenzae co-colonizes with S. pneumoniae or M. catarrhalis, H. infl uenzae predominates over these 2 bacteria to cause AOM (18). The elimination of S. pneumoniae strains expressing PCV7 serotypes has resulted in the remaining S. pneumoniae strains, except serotype 19A, competing less effectively with H. infl uenzae in the nasopharynx (18). Therefore, with a further reduction in nasopharynx colonization by S. pneumoniae, including strains expressing serotype 19A (as will probably result from the recent introduction of PCV13 in some countries), our data suggest that H. infl uenzae might fi ll the nasopharyngeal niche at the onset of AOM, and consequently, H. infl uenzae might become a more prominent cause of AOM.

AOM infection caused by H. infl uenzae
S. aureus also can cause respiratory infections; several reports have suggested that S. aureus might be replacing S. pneumoniae as a dominant nasopharynx colonizer as a consequence of the introduction of PCV. Concern has been expressed that S. aureus might emerge as a more prominent respiratory pathogen (10)(11)(12)(13)23). However, our results do not support that S. aureus has emerged as a frequent pathogen of AOM after introduction PCV7.
Knowledge Although we found a synergistic association between S. pneumoniae and M. catarrhalis at healthy visits and a signifi cant increase of polymicrobial colonization at AOM visits, the tendency toward polymicrobial colonization did not result in synergistic associations at AOM visits. On the contrary, competitive associations between H. infl uenzae and S. pneumoniae and between H. infl uenzae and M.
catarrhalis were found at AOM visits. The increases in colonization of individual bacterial pathogens during AOM might randomly result in an increase of polymicrobial colonization and might not result from synergistic associations among the potential pathogens. Further study on the mechanism is needed.
The mechanisms to explain competitive and synergistic interactions among S. pneumoniae, H. infl uenzae, M. catarrhalis, and S. aureus have been explored. In a mouse model, Lysenkoet al. found that when H. infl uenzae colonized with S. pneumoniae in the nasopharynx, S. pneumoniae was rapidly cleared (24). The competitive interaction was dependent on cellular components of H. infl uenzae activating the host innate immune response involving complement and neutrophils; the end result was the killing of S. pneumoniae (24). The mouse model results are consistent with our observations among children but differ from results observed during in vitro experiments that predict that S. pneumoniae should inhibit the growth of H. infl uenzae (25,26). A competitive association between S. pneumoniae and S. aureus might be mediated by S. pneumoniae production of hydrogen peroxide (25,27). Our results and those of others suggest that this mechanism might be active during times of health but not, on the basis of our observation, at onset of AOM (10,11,13,28).
We did evaluate the effects of age, sex, daycare attendance, breast-feeding history, exposure to tobacco smoke, and an immunologically driven increased susceptibility to AOM (15,29) in our study population. The results differed at healthy versus AOM visits except for the effect of daycare attendance on H. infl uenzae colonization; daycare attendance had a positive association with H. infl uenzae colonization at both healthy and AOM visits. Others have shown daycare attendance to be consistently associated with increased colonization by S. pneumoniae, H. infl uenzae, and M. catarrhalis in healthy children and in children with viral upper respiratory infections (8).
Our previous studies have shown that children who are prone to otitis have much weaker immune responses to S. pneumoniae surface antigens than those who are not prone (15,30). We therefore assessed the otitis-prone condition as a predictor in the models in this study. We found that the otitis-prone condition was positively associated with S. pneumoniae and H. infl uenzae colonization at healthy visits but not at AOM visits.
The rates of nasopharynx colonization with S. pneumoniae, H. infl uenzae, and M. catarrhalis and the mix at onset of AOM that we observed occurred mostly in the context of a concurrent viral upper respiratory tract infection. A limitation of our study is that we did not evaluate bacterial-viral interactions. Revai et al. (31) and Chonmaitree et al. (32) found that among young children, viral upper respiratory tract infections preceded >90% of AOM infections. At onset of AOM, 93% of the children in our study had clinical signs of a viral upper respiratory tract infection; the predominant viruses detected were infl uenza, parainfl uenza, respiratory syncytial, and adenovirus (A. Chang, unpub. data). The culture and PCR methods used in that study did not identify rhinovirus, metapneumovirus, or bocavirus. Previous studies have shown that viral upper respiratory tract infections affect bacterial nasopharynx colonization (5,7). Respiratory viruses up-regulate epithelial cell receptors for some species of respiratory bacteria (7); they cause infl ammation and down-regulate innate and adaptive host defenses (7,33). Different viruses have varying effects on different bacteria (34)(35)(36). The comparison of healthy visits with AOM visits suggests that viral infections alter the host nasopharynx environment by facilitating a shift in the polymicrobial mix and enhancing polymicrobial colonization.
We did not study the additional effect of antimicrobial drugs on nasopharynx colonization. Such treatment modifi es nasopharynx colonization patterns. Pettigrew et al. (8) showed that antimicrobial drug therapy was associated with a lower prevalence of colonization with S. pneumoniae and M. catarrhalis but not with H. infl uenzae. Varon et al. (37) showed that colonization by S. pneumoniae, H. infl uenzae, and M. catarrhalis decreased after antimicrobial drug thereapy; the reduction in colonization was less for H. infl uenzae than for S. pneumoniae or M. catarrhalis. Current national guidelines endorse the use of amoxicillin for fi rstline treatment of AOM, sinusitis, and community-acquired pneumonia in children (38)(39)(40). Amoxicillin is ineffective for eradicating β-lactamase-producing H. infl uenzae and M. catarrhalis. We have previously shown that ≈65% of H. infl uenzae and 100% of M. catarrhalis colonizing the nasopharynx of children in our study population elaborate β-lactamase (17,21,22). Therefore, empiric treatment with amoxicillin would probably increase H. infl uenzae and M. catarrhalis nasopharynx colonization in children.
Our results led us to 2 conclusions. First, nasopharyngeal bacterial interactions among S. pneumoniae, H. infl uenzae, and M. catarrhalis differ during health and at onset of AOM in young children. Second, at the onset of AOM, the nasopharynx environment among children vaccinated with PCV7 is favorable for H. infl uenzae colonization. Consequently, our results predict that H. infl uenzae might be become a more prominent bacterial pathogen of AOM in the era of PCV. Further studies of virus-bacterium-host interactions in the nasopharynx and additional studies of the mechanisms driving the observed shifts in bacterial species and polymicrobial makeup are needed.