Increased Nasopharyngeal Density and Concurrent Carriage of Streptococcus pneumoniae, Haemophilus influenzae, and Moraxella catarrhalis Are Associated with Pneumonia in Febrile Children

Background We assessed nasopharyngeal (NP) carriage of five pathogens in febrile children with and without acute respiratory infection (ARI) of the upper (URTI) or lower tract, attending health facilities in Tanzania. Methods NP swabs collected from children (N = 960) aged 2 months to 10 years, and with a temperature ≥38°C, were utilized to quantify bacterial density of S. pneumoniae (Sp), H. influenzae (Hi), M. catarrhalis (Mc), S. aureus (Sa), and N. meningitidis (Nm). We determined associations between presence of individual species, densities, or concurrent carriage of all species combination with respiratory diseases including clinical pneumonia, pneumonia with normal chest radiography (CXR) and endpoint pneumonia. Results Individual carriage, and NP density, of Sp, Hi, or Mc, but not Sa, or Nm, was significantly associated with febrile ARI and clinical pneumonia when compared to febrile non-ARI episodes. Density was also significantly increased in severe pneumonia when compared to mild URTI (Sp, p<0.002; Hi p<0.001; Mc, p = 0.014). Accordingly, concurrent carriage of Sp+, Hi+, and Mc+, in the absence of Sa- and Nm-, was significantly more prevalent in children with ARI (p = 0.03), or clinical pneumonia (p<0.001) than non-ARI, and in children with clinical pneumonia (p = 0.0007) than URTI. Furthermore, Sp+, Hi+, and Mc+ differentiated children with pneumonia with normal CXR, or endpoint pneumonia, from those with URTI, and non-ARI cases. Conclusions Concurrent NP carriage of Sp, Hi, and Mc was a predictor of clinical pneumonia and identified children with pneumonia with normal CXR and endpoint pneumonia from those with febrile URTI, or non-ARI episodes.


Introduction
The nasopharynx is an ecologic reservoir for human bacterial pathogens such as Streptococcus pneumoniae (Sp), Moraxella catarrhalis (Mc), Haemophilus influenzae (Hi), Staphylococcus aureus (Sa), and Neisseria meningitidis (Nm) [1]. Whereas these species form part of the nasopharyngeal microbiome [1,2], they are also the source of several of the most prevalent causes of morbidity and mortality to human kind, which include diseases such as acute otitis media, pneumonia, bacteremia and meningitis [3].
Carriage of these nasopharyngeal species in healthy children varies amongst different studies and geographic regions [4]. In general, carriage prevalence of these bacteria is lower in industrialized countries than in resource-limited nations. The tendency, however, is that carriage of Sp, Hi or Mc increases during childhood, peaking at the age of 3 years, and then decreases [1]. Conversely, Nm carriage is low during childhood, but peaks in prevalence in young adults [5], whereas nasopharyngeal carriage of Sa decreases during childhood and remains relatively low thereafter [6,7].
There are limited studies focused on investigating nasopharyngeal carriage during disease episodes. Studies conducted in Vietnamese children (<2 years old) showed a similar prevalence of nasopharyngeal carriage of Sp, Hi or Mc in children with pneumonia compared to healthy controls but an increased Sp nasopharyngeal density was observed in pneumonia patients, compared to controls [8]. Carriage of Sp, Hi or Mc has also been associated with the development of otitis media and sinusitis [9,10]. An increased nasopharyngeal carriage of Sp has also been associated with infection with influenza virus, rhinovirus, and adenovirus in admitted South African children with pneumonia [11] and to influenza virus and parainfluenza virus in Peruvian children with acute respiratory infection (ARI) [12].
The complex milieu of these nasopharyngeal (NP) pathogens can also be modified by factors such as the use of antimicrobial medicines or vaccines, or the innate immune response, which include the development of an acute infection [13,14]. To the best of our knowledge, carriage dynamics by all these five species (i.e., Sp, Mc, Hi, Sa and Nm) have not been previously investigated at the same time in the nasopharynx of ill children. The present study investigated carriage of these five major human pathogens in a cohort of urban and rural Tanzanian children presenting with an acute febrile illness [15]. The associations between presence of these bacteria, concurrent carriage or nasopharyngeal densities, and disease conditions were assessed. More precisely, we investigated differences in bacterial carriage according to: 1) respiratory disease versus other type of infections causing fever, 2) clinical pneumonia versus upper respiratory tract infections, 3) type of radiological findings in children with clinical pneumonia, and 4) disease severity.

Study area and population
Nasopharyngeal swabs (NP) were collected from 1005 febrile children, aged 2 months to 10 years, presenting at the outpatient clinic of Amana District Hospital in the economical capital of Tanzania, Dar es Salaam, and at the outpatient clinic of St. Francis Designated District Hospital in Ifakara, Kilombero District, a small rural town in South central Tanzania, as described elsewhere [15]. Briefly, the enrollment period was from April to August 2008 for patients of Dar es Salaam and from June to December 2008 for those of Ifakara. Children with an axillary temperature of ! 38˚C and requiring no immediate lifesaving procedures were assessed for inclusion criteria: 1) first visit for the present illness, 2) fever duration 1 week, 3) chief reason for visit not injury/trauma, 4) no antimalarial or antibiotic received during the preceding week, and 5) no severe malnutrition. A questionnaire and clinical examination were administered. At the time of study, enrollment in the Expanded Program on Immunization in Tanza

Definition of febrile diseases
Final diagnosis(ses) for acute febrile illness was established based on criteria from the World Health Organization (WHO), Infectious Diseases Society of America guidelines and systematic reviews [15]. Acute respiratory infection (ARI) was defined as any acute ( 1 week) infection manifested by at least one respiratory sign or symptom localized to the upper or lower respiratory tract and divided in two categories: clinical pneumonia or upper respiratory tract infection (URTI). Children with clinical pneumonia were further divided in three categories based on chest radiography (CXR) findings, according to the WHO Pneumococcal Trials Ad Hoc Committee recommendations [16,17]: alveolar consolidation and/or pleural effusion was categorized as endpoint pneumonia; other infiltrates (that in all these children corresponded to peribronchial thickening +/-atelectasis, compatible with the clinical entity of bronchiolitis), were categorized as pneumonia with other infiltrates; normal radiography was categorized as pneumonia with normal CXR. Severe febrile disease (whenever due to ARI or another type of infection) was defined as the presence of at least one of the following features: respiratory distress, impaired consciousness, seizures, meningismus, cardiovascular failure, renal failure, severe anemia (hemoglobin <5 g/dl), severe dehydration, jaundice, and severe malnutrition.

Specimen collection and storage
NP swabs were collected according to recommendations from the WHO [18] and immediately stored in 1 ml of STGG (skim-milk, tryptone, glucose and glycerol) transport medium [19], vortexed for 20 s with the swabs inside to release bacteria into the STGG and frozen at -80˚C until further analysis.
1 h in a 37˚C in water bath. The subsequent steps were carried out according to the Qiagen DNA mini kit protocol, as detailed elsewhere [20,21]. DNAs were eluted in 100 μl of elution buffer and stored at -80˚C. DNA from reference strains Sp (TIGR4), Sa (American Type Culture Collection (ATCC) 25923), Hi (Centers for Disease Control and Prevention (CDC) reference strain M5216), Mc (CDC reference strain M15757), and Nm (CDC8201085) [22] were also extracted from overnight cultures using the QIAamp kit. DNA concentration was measured by the Nanodrop method (Nanodrop Technologies, Wilmington, DE).

Quantitative PCR
Quantitative PCR (qPCR) assays targeting the following genes lytA, nuc, hpd, copB, and sodC, carried by all Sp, Sa, Hi, Mc, and Nm respectively, were performed. The qPCR assays utilized published primers and probes at concentrations optimized in this study and shown in Table 1 [23 -28].

Demographic and clinical characteristics of the febrile children
A total of 1005 febrile children were consecutively recruited. Nine hundred and sixty febrile children, for whom enough NP material was available, were included in this study. Demographic and clinical characteristics of these children are described in Table 2. Among the children included in the analysis, 62% presented at the clinic with an ARI (44% with URTI, 12% with pneumonia with normal CXR, 3% with pneumonia with infiltrates and 3% with endpoint pneumonia), whereas 23% were diagnosed with other diseases, and 15% had unknown disease. About 6% of children had malaria, 6% gastrointestinal disease, 2% typhoid, 4% urinary tract infection, 5% systemic disease and less than 1% presented with occult bacteremia, or skin disease. Overall, 13% of the children had a severe febrile illness, and among children with ARI, 10.6% had a severe presentation ( Table 2).

Prevalence of individual nasopharyngeal carriage of bacterial pathogens and association with respiratory infections
The overall prevalence of nasopharyngeal carriage of Sp was 81%, Hi 75%, Sa 23%, Mc 91%, and Nm 51% (

Concurrent carriage of bacterial pathogens and association with respiratory infections
Further analyses revealed that 94.2% of febrile children enrolled were carrying more than one species of the five bacterial species tested in this study (Table 4). Concurrent carriage of Sp, Hi, and Mc was associated with respiratory infections (S1 Table). Carriage of all three Sp, Hi, and Mc [in the absence of Sa and Nm (Saand Nm -)] was significantly more prevalent in the nasopharynx of children with febrile ARI than in children with a febrile non-ARI episode (p = 0.035) or in children with clinical pneumonia vs URTI (p = 0.009) (S1 Table).
Furthermore, concurrent carriage of these three species, in the absence of Saand Nm -, was also significantly more prevalent in the nasopharynx of children with pneumonia with normal CXR than those children with URTI (p = 0.018) (S1 Table). The sensitivity and specificity to differentiate pneumonia with normal CXR and endpoint pneumonia from non-ARI cases was 75% and 80% or 81% and 80%, respectively, whereas the sensitivity to differentiate pneumonia with normal CXR, or endpoint pneumonia, from URTI cases was 75% and 77%, or 81.5%, and 77.6%, respectively.  The odds of carrying Sp, Hi, Sa, and Mc in a child enrolled in the urban clinic was significantly less than that of a child enrolled in the rural clinic. This association was found regardless of the local prevalence of ARI among all febrile episodes (that was lower in the urban than in the rural site). All other possible covariates were assessed but found not to be significantly associated with bacterial carriage.

Association between bacterial density and respiratory infections
Density was categorized according to increasing bacterial load. Table 6 shows that 64.0%, 75.1%, or 81.3% of children carried in the nasopharynx >1x10 6 cfu/ml of Sp, Hi or Mc, respectively, whereas only 9.9%, or 13.3%, of children carried >1x10 6 cfu/ml of Nm or Sa, respectively. We next investigated differences in nasopharyngeal density in children with respiratory infection in comparison to children with a non-ARI episode. As shown in Fig 1 and S2 Table, NP density of Sp, Hi, or Mc, was significantly higher in febrile ARI cases (clinical pneumonia or children with pneumonia with normal CXR) than in febrile non-ARI episodes [p<0.001 for all cases (S2 Table)]. Similarly, nasopharyngeal density of Hi or Mc was found significantly increased in clinical pneumonia when compared to URTI, (not shown). Nasopharyngeal density of Sp, Hi or Mc, was significantly higher in children with severe clinical pneumonia when compared to mild URTI (p<0.002, p<0.001 and p = 0.014 respectively) (Fig 2). Nasopharyngeal density of Sa and Nm were similarly detected in children with respiratory and non-respiratory infection.

Discussion
We have studied, for the first time, nasopharyngeal carriage of five major human bacterial pathogens, Sp, Hi, Mc, Sa, and Nm in children presenting the most common childhood syndrome: acute fever. Our analyses demonstrate that Tanzanian febrile children experiencing an episode of any type of ARI (including URTI, clinical pneumonia or pneumonia with normal CXR) carried significantly more Sp, Hi, and Mc, at the same time in the nasopharynx, in the absence of Sa and Nm than children with non-ARI infections. Accordingly, bacterial densities of these three species were found to be significantly increased in the nasopharynx of children with respiratory infection, when compared to those with non-ARI infections. Such a significantly higher prevalence, or increased density, was not observed in children carrying Sa or Nm.
Carriage studies including bacterial species investigated here, in children with ARI, have increased the last few years. The high prevalence of nasopharyngeal carriage of Sp, Hi, and Mc (87.5%, 82.6%, and 96.5%, respectively) in children with pneumonia, whether clinical or radiological, that we observed, differs from that obtained in a recent study of Vietnamese children with radiological pneumonia, whose carriage prevalence was 38.7, 50 and 28.1% for Sp, Hi and Mc, respectively [8]. A study by Wolter et al (2014) analyzed South African children experiencing invasive pneumococcal pneumonia and showed a 53% prevalence of nasopharyngeal carriage of the pneumococcus [11].
An increased nasopharyngeal pneumococcal density has been associated with pneumococcal pneumonia in HIV-infected adults [30] and children with pneumococcal or radiological pneumonia [8,11,30]. In line with this evidence, our study demonstrates a statistically higher pneumococcal density in the nasopharynx of children with ARI (median, 2.25x10 6 cfu/ml), URTI (median, 1.73x10 6 cfu/ml), clinical pneumonia (median, 2.05x10 6 cfu/ml), and pneumonia with normal CXR (median, 1.90x10 6 cfu/ml), when compared to those suffering from a non-ARI episode (median, 9.17x10 5 cfu/ml), which represents a *2.5, *1.8, *2.2, or *2 fold-increase in pneumococcal density, respectively. Furthermore, we found a 4.2-fold increase of pneumococcal nasopharyngeal density when children with severe clinical pneumonia were compared to those with mild URTI, and a statistically significant high density in four children who died of pneumonia (median, 2.27x10 7 cfu/ml), compared to children who survived (p = 0.003). These findings suggest that there might be a correlation between high nasopharyngeal density and disease severity, although further studies would be required to confirm these observations. Nasopharyngeal bacterial density in children with respiratory infection (ARI, URTI, clinical pneumonia and pneumonia with normal CXR) was also found to be increased when evaluated for Hi and Mc, but not for Sa and Nm. While nasopharyngeal pneumococcal density has not been found useful to assist in the diagnosis of radiological pneumonia in children [8], our findings of a 2.5-fold, or 2-fold, increase in pneumococcal density in the nasopharynx of children with clinical pneumonia, or ARI, respectively, may facilitate secondary bacterial infection.
In healthy children the most common positive association seen in the nasopharynx is between Sp and Hi, whereas a negative association between Sp and Sa has been observed [31]. When we modeled our prevalence data, the strongest positive association was detected between Sp and Mc, followed by Hi and Mc, suggesting that Mc may drive the carriage of the other two species. Acquiring evidence to support whether Mc may play a central role in driving carriage of Sp and Hi will require further efforts. Carriage of Mc does not affect carriage of Sa or Nm as this bacterium, Mc, was neither negatively, nor positively associated with Nm or Sa, in contrast to both Sp and Hi which were both positively associated with Nm and negatively associated with carriage of Sa.
There are some limitations in this manuscript that need to be mentioned. For example, bacterial cultures were not obtained, and pneumococcal types were not investigated. The latter information may have been relevant in view that potential association between pneumococcal serotypes and individual species (i.e., Mc, Hi, Sa, or Nm) could not be explored. Another important limitation relates to the fact that viral infections have not been included in this paper, since the presentation of results would have been too complex.
In summary, our study demonstrated a significant association in febrile children between concurrent nasopharyngeal carriage of Sp, Hi, and Mc and respiratory infection. Firstly, when individually analyzed, carriage prevalence of Sp, Hi, or Mc was significantly increased in febrile ARI cases, cases of URTI, and children with clinical pneumonia. Secondly, when we considered all five species, our analyses showed that in the absence of Sa and Nm, concurrent carriage of Sp, Hi, and Mc was significantly more prevalent in the nasopharynx of children with clinical pneumonia and pneumonia with abnormal CXR, in comparison to non-ARI cases and URTI. Thirdly, when we assessed nasopharyngeal density, our study demonstrated a significantly increased density of Sp, Hi, and Mc in cases of febrile respiratory infection vs non-ARI. These findings call for the development of quantitative multiplex point-of-care tests, or at least semiquantitative tests that would ideally allow better prediction of a LRTI.
Supporting Information S1