Characterization of the nasopharyngeal and middle ear microbiota in gastroesophageal reflux-prone versus gastroesophageal reflux non-prone children

Thirty children up to 12 years of age who suffered from recurrent AOM (5), chronic OME (23), or both (2), and who were listed for tympanostomy tube placement, were included in the study (Table 1). Children were diagnosed based on the surgeon’s report.

In total, 11/30 MEF specimens and 11/30 NPS specimens (not, by definition, paired specimens) were excluded from 16S rRNA gene sequencing due to low DNA content (< 1000 16S rRNA gene copies/μL). Antibiotic use ≤ 6 months prior to inclusion in the study did not affect the exclusion of samples due to low DNA content (Table 2; MEF p = 0.13, NPS p = 0.45). The microbial composition of the remaining 38 specimens were sequenced and the data were processed using mothur. To reduce the effects of uneven sequencing depths, sequences were rarefied to 1000 sequences per specimen, which was sufficient to obtain a high degree of sequence coverage with an average Good’s coverage value of 0.98 (± 0.01).

Sequencing of the MEF microbiota resulted in the identification of 29 bacterial operational taxonomic units (OTUs) in 19 MEF specimens with an abundance > 1%, with an average of 4 OTUs (± 3) per specimen. Sequencing the nasopharyngeal microbiota identified 39 bacterial OTUs in 19 NPS specimens with > 1% abundance, with an average of 9 OTUs (± 4) per specimen.

The most commonly detected OTUs were Alloiococcus spp. and Turicella spp., present in 12 and 11 MEF specimens, respectively. There has been some debate as to whether members of the bacterial genera Alloiococcus and Turicella actually play a role in the pathogenesis of OM in children or if they are part of the commensal microbiota (and invading the middle ear following perforation) [17‐23]. Though A. otitidis and T. otitidis are frequently detected in OM children, it is questionable whether these organisms have enough pathogenic potential to induce OM, although studies have shown that A. otitidis may have enough immunogenic potential to modulate a host immune response [24‐26]. In addition, A. otitidis is able to form both single- and multispecies biofilms with Haemophilus influenzae. When present in polymicrobial biofilms, A. otitidis can promote H. influenzae growth and survival by increasing biofilm production in adverse growth conditions and by altering antimicrobial resistance [27]. In this study, the use of antibiotics ≤ 6 months prior to inclusion in the study did not have an effect on the detection of Alloiococcus spp., Haemophilus spp., Streptococcus spp., or Moraxella spp. (Table 3). These findings are clinically important, as antibiotics overuse or misuse can promote the development of bacterial resistance to antibiotics. Alloiococcus spp. and Turicella spp. were not detected by 16S rRNA gene sequencing in any nasopharyngeal swab. Previous PCR- and culture-based studies from other populations have reported A. otitidis in 7–12% of nasopharyngeal swabs, in contrast to the study of Marsh et al., who found no A. otitidis in paired nasopharyngeal swabs [17, 18, 28, 29]. Turicella otitidis is isolated almost exclusively from middle ear exudates [20, 23]. However, our study was limited to 30 children and it is possible that Alloiococcus spp. and Turicella spp. would have been detected in a larger cohort of children. Failure to detect Alloiococcus spp. and Turicella spp. in nasopharyngeal swabs from any of the children with positive ear MEF samples suggests that they are unlikely to be primary otopathogens in this population, since nasopharyngeal colonization is considered to be the antecedent of OM [30]. To better understand the potential role of Alloiococcus spp. and Turicella spp. in OM, further studies into host interactions and colonization of adjacent sites in healthy and diseased children will be necessary. Ultimately, the findings of this study will need to be further validated in the future with larger cohort sizes.

OTUs consistent with classical otopathogens such as Haemophilus spp., Streptococcus spp., and Moraxella spp. were detected in 9, 6, and 1 MEF specimens, respectively (Fig. 1). In addition, staphylococci were also found in the MEF of seven children, with an abundance up to 100%. Our findings are consistent with several other studies; for example, Haemophilus (influenzae) has been found to be the most common pathogen isolated in the context of widespread conjugate pneumococcal vaccination [31, 32]. Further, the presence of Haemophilus (influenzae) has been associated with an increased risk of OM, whereas the presence of Corynebacterium spp. and Dolosigranulum spp. have been associated with a decreased risk of OM [33, 34].

No upper gastrointestinal tract microbiota previously associated with OM (e.g., H. pylori) was detected in the middle ear. Further, we did not find any microbial difference in the MEF of children with or without GER (Table 4), suggesting that GER was not associated with the translocation of stomach microbiota to the middle ear in our cohort.

To study the co-occurrence of OM-related pathogens in the nasopharynx and middle ear, we compared the microbiota obtained from 12 paired NPS and MEF specimens (Fig. 2). Nine OTUs were present in both specimen types, of which four OTUs, including Haemophilus spp., Streptococcus spp., Parvimonas spp., and Staphylococcus spp. were present in both paired specimen types from 10/12 children. Importantly, Haemophilus spp. and Streptococcus spp. were only detected in MEF when the nasopharynx of the child was concurrently colonized with these bacterial genera. This finding is in agreement with previous studies, and suggests that the more easily accessible nasopharyngeal microbiota could be used to predict the presence of pediatric otopathogens in the middle ears of children suffering from OM. However, it should be noted that the majority of OTUs, including the dominant genera Alloiococcus spp. and Turicella spp., were unique to the sampling site and, therefore, caution is necessary when using the nasopharyngeal microbiota as a proxy for the MEF microbiota, as discussed previously by van Dongen et al. [35].

The strength of this publication is that it is the first study to investigate the possible association between the microbial composition of the nasopharynx and middle ear in GER-prone and GER non-prone children using culture-free 16S rRNA gene sequencing. However, several limitations in the study protocol should be recognized.

Firstly, an actual diagnosis of GER is difficult to prove, as most cases of pediatric GER are diagnosed based solely on the clinical presentation and parental observation (and not, for example, on the presence of pepsin and bile acids in the middle ear). There are no recognized classic physical signs of GER in the pediatric population, and, in general, a doctor diagnoses GER by reviewing an infant’s symptoms and medical history. In this study, only parental-reported clinical information was collected. Essentially, we asked parents retrospectively if their child was diagnosed with reflux in the previous three months and if their child had been diagnosed with reflux at any time in the past. To date, the most thoroughly evaluated questionnaire for infant symptoms is the Infant Gastroesophageal Reflux Questionnaire Revised (I-GERQ-R) [36]. We designed our own questionnaire, as even an established questionnaire such as the I-GERQ-R was not validated to be adequately specific to differentiate GER infants from other symptomatic, but non-GER, infants, particularly if the treatments being tested are directed at acid reflux. For that purpose, additional inclusion criteria, such as esophageal pH monitoring or histology, are currently necessary to ensure adequate diagnostic specificity [36]. In the current study, we did not confirm the parental-reported diagnosis of GER by the identification of bile acids/pepsin in MEF or measurement of the MEF pH levels. These tools have proven to be useful for determining the presence of pathological GER [4‐6]. The absence of these tools in our study may have influenced the fact that we found no statistically significant association between the microbiota profiles of GER-prone and GER non-prone children.

Secondly, our study included only children primarily diagnosed with OME, as these children are mainly listed for tympanostomy tube placement. Most children with AOM experience a self-limiting illness and lack the presence of fluid in the middle ear. Many will not present to a doctor. This means that an association between GER and AOM disease is much more difficult to prove than between GER and OME disease, but does not affect the hypothesis that GER may be associated with increased OM disease in GER-prone children.

Thirdly, as previously discussed, using the more easily accessible nasopharyngeal microbiota to predict the presence of pediatric otopathogens in the middle ears of children suffering from OME and AOM is questionable. Because of the size of the population studied and lack of data from other studies, data interpretation should be made with caution and regarded as preliminary. Further, we were unable to analyze the microbiota to the species level (a common problem with microbiota analysis [13]) and were, therefore, limited to bacterial taxonomic identification at the genus level.