|Home | About | Journals | Submit | Contact Us | Français|
Chronic suppurative otitis media (CSOM) presents with purulent otorrhea (ear discharge), is characterized by chronic inflammation of the middle ear and mastoid cavity, and contributes to a significant disease burden worldwide. Current antibiotic therapy is guided by swab culture results. In the absence of detailed molecular microbiology studies of CSOM patients, our current understanding of the microbiota of CSOM (and indeed of the healthy ear) remains incomplete. In this prospective study, 24 patients with CSOM were recruited, along with 22 healthy controls. Culture-based techniques and 16S rRNA gene amplicon sequencing were used to profile the bacterial community for each patient. Comparisons between patients with and without cholesteatoma in the middle ear and mastoid cavity were also made. A major finding was that the middle ear of many healthy controls was not sterile, which is contradictory to the results of previous studies. However, sequencing data showed that Staphylococcus aureus, along with a range of other Gram-positive and Gram-negative organisms, were present in all subgroups of CSOM and healthy controls. Large interpatient variability in the microbiota was observed within each subgroup of CSOM and controls, and there was no bacterial community “signature” which was characteristic of either health or disease. Comparisons of the culture results with the molecular data show that culture-based techniques underestimate the diversity of bacteria found within the ear. This study reports the first detailed examination of bacterial profiles of the ear in healthy controls and patients with CSOM.
Chronic suppurative otitis media (CSOM) is one of the most common childhood diseases worldwide (1). It carries a significant disease burden, estimated at 2 million disability-adjusted life years (2, 3). CSOM is characterized by chronic ear discharge through a perforated tympanic membrane for more than 6 weeks to 3 months (2, 4, 5). Its prevalence is related to poor socioeconomic conditions, and it is relatively uncommon in developed countries. Estimates suggest that between 65 and 300 million cases occur worldwide, with 60% of these cases suffering significant hearing impairment (3, 6, 7). Globally, 28,000 deaths per year due to complications of CSOM have been reported (2). There is an enormous financial burden (around $5 billion per annum in the United States) associated with otitis media and its sequelae, including chronic suppurative otitis media (8, 9).
CSOM often begins as an acute infection of the middle ear, acute otitis media (AOM), which occurs in up to 80% of children by the age of 3 (8, 10). While most cases resolve spontaneously, a small minority of patients progress to a chronic phase characterized by chronic purulent ear discharge through a perforated tympanic membrane with associated inflammation of the mastoid and middle ear mucosa and hearing loss. CSOM can occur with or without cholesteatoma (epithelial inclusion cyst), but the presence of cholesteatoma does not necessarily alter the clinical symptoms. Intracranial complications such as brain abscess and meningitis contribute to the morbidity and occasional mortality of this condition (2, 3).
The pathogenesis of CSOM remains poorly understood. Complex interactions between the environment, microbes, and host are thought to lead to the development of this multifactorial disease (3, 10, 11). Topical and oral antibiotics are prescribed to patients based on bacterial culture results when available, but the clinical benefit of antibiotic therapy is not always clear (3). Surgery may prevent local, regional, or systemic complications, but some patients may continue to have ear discharge postoperatively (12). Research into the putative microbial causes of CSOM has so far been reliant on culture-based techniques. In these studies, Staphylococcus aureus and Pseudomonas aeruginosa were the most commonly isolated bacteria, with methicillin-resistant S. aureus (MRSA) isolated in some cases (10, 13,–21). However, there are treatment failures even when these specific organisms are targeted.
It has previously been assumed that a healthy individual's ear is sterile (22, 23). Infections in the middle ear are thought to occur when pathogens enter the middle ear through the external ear canal or Eustachian tube. Other theories to explain the persistent nature of this disease and repeated infection include toxin production by P. aeruginosa (24), microbes embedding within the dead/damaged tissue (cholesteatoma) (16, 18), formation of biofilms (25, 26), recurrent bacterial infection from the nasopharynx not covered by the antibiotics prescribed (21), or development of antibiotic resistance (20, 27). Fluorescence in situ hybridization (FISH) has previously shown S. aureus biofilms to be present within the tissue of CSOM patients (28). Biofilm has also been demonstrated on the middle ear mucosa of children with chronic otitis media (23) and in middle ear effusions from children with recurrent AOM (29, 30), which is a likely risk factor for CSOM. These evade the host immune response and antibiotic therapy by growing in a protective biofilm matrix.
The aim of this study was to characterize the bacterial communities within the middle ears and mastoids of patients with and without CSOM. Bacterial community profiles obtained using culture and molecular-based techniques were compared, with the latter in particular providing novel insights into the microbiota of CSOM patients. While previous studies have described the microbiota of otitis media with effusion (31, 32), there is limited information on the microbiota associated with CSOM (33).
This was a prospective study of 24 patients undergoing mastoid surgery for CSOM and 22 patients with healthy middle ears undergoing either cochlear implantation (CI) or benign brain tumor (vestibular schwannoma) (T) removal via the mastoid and middle ear. None of the study patients received antibiotics within the 2 months preceding surgery, but all patients (subjects and controls) received intravenous cefazolin on induction as part of the institutional perioperative protocol. The CSOM patients were subdivided based on the presence or absence of cholesteatoma. Patients undergoing surgery for CSOM were also categorized according to the extent of surgery (canal wall up [CWU] or canal wall down [CWD] procedure). Patients with cholesteatoma were graded by disease severity (34), but there is no such disease severity scale currently available for CSOM without cholesteatoma. Patients with anatomical temporal bone abnormalities or immune deficiencies were excluded.
The study was approved by the New Zealand Health and Disability Ethics Committee (NTX/12/03/024).
Mastoid surgery was performed under general anesthesia and sterile conditions. The ear canal was left intact in more limited disease (CWU) but taken down and the mastoid air cell system externalized (canal wall down,) in more extensive disease. Assessment of disease severity was based on clinical and imaging findings that were evaluated by an otolaryngologist (M.N.). Intraoperatively, tissue samples and two sterile rayon-tipped swab (Copan 170KS01) samples were taken from the mastoid and middle ear. Collected swab samples were placed immediately on ice and then frozen (−20°C) within 2 h of surgery.
Conventional microbiology swab samples were also collected during surgery and sent to a hospital laboratory for bacterial culture analysis. Swab samples were inoculated on the following culture media (Fort Richard Laboratories Ltd., New Zealand): Columbia sheep blood, supplemented chocolate with bacitracin, MacConkey, colistin-nalidixic acid, and Sabouraud dextrose agars. The former two were incubated at 37°C in ambient air supplemented with 5% CO2, while the latter three were incubated at 37°C in ambient air. Swab samples collected from patients with chronic infections were also plated on brain heart infusion medium and incubated anaerobically. Significant bacterial species were identified using the Vitek MS system (bioMérieux) and antibiotic sensitivity testing was performed on the Vitek 2 system (bioMérieux) or by using disc diffusion criteria as appropriate and interpreted using the Clinical and Laboratory Standards Institute guidelines (35). Samples that led to cultivation of a single colony or more were considered to be positive.
Swab samples were thawed on ice and placed into a sterile Lysing Matrix E tube (MP Biomedicals, Australia). Genomic DNA was extracted from the paired swabs using the AllPrep DNA/RNA isolation kit (Qiagen), as described previously (36). A negative extraction control to test for contamination originating from the DNA isolation kit was carried out in triplicate using 200 μl of sterile water. In no case did these extraction controls yield measurable DNA, and subsequent PCRs including this DNA were invariably negative.
A nested PCR approach was used to amplify the bacterial 16S rRNA genes from genomic DNA due to difficulties in obtaining PCR products from most samples after 35 cycles of amplification. Primers 616V (37) and 1492R (38) were used to amplify nearly the full length (targeting Escherichia coli positions 8 to 1513) of the bacterial 16S rRNA gene, as described previously (39). After 12 initial PCR cycles, 1 μl from the first PCR was used as the template for the second PCR amplification. In this second reaction, the V3 to V4 region of the bacterial 16S rRNA gene was amplified (35 cycles) using primers 341F and 806R (read length of 465 bp) (40). The PCR mixture included the following: genomic DNA template (~100 ng), equimolar concentrations (0.2 μM) of each primer, deoxynucleoside triphosphates (dNTPs) (0.2 mM), HotStar PCR buffer (1×), MgCl2 (2 mM), 0.5 U HotStar DNA polymerase (Qiagen), and PCR-grade water to a final volume of 25 μl. PCRs were conducted under the following conditions: 95°C for 15 min, followed by 35 cycles of 95°C for 30 s, 55°C for 30 s, and 70°C for 40 s, with a final step of 3 min at 70°C. Duplicate PCRs were conducted for each sample. Negative and positive controls were run for all PCRs. In addition, 1 μl of the initial PCR negative control was used as a template in the second PCR to ensure that the final PCR products obtained were not due to contamination. Study samples that produced no amplification product after the nested PCR amplification (47 cycles in total) were considered to be negative. Amplicons were purified using Agencourt AMPure beads (Beckman Coulter Inc.) and then subjected to both quality (Agilent Bioanalyzer DNA high-sensitivity [HS] assays) and quantity (Qubit double-stranded DNA (dsDNA) HS assay kit) checks.
Sequencing on an Illumina MiSeq (2 × 300 bp, paired-end reads) was performed at a commercial sequencing provider (Macrogen, South Korea). Raw sequences were analyzed using a combination of the UPARSE and QIIME software packages (41, 42). In brief, raw sequences were merged, trimmed, and quality checked using UPARSE. Unique sequences were clustered in UPARSE into operational taxonomic units (OTUs) at a threshold of 97% 16S rRNA gene sequence similarity. Chimera checks were performed by checking the sequences against a chimera database (SILVA gold chimera reference database). Taxonomic assignment of the sequences was subsequently performed in QIIME using RDP Classifier 2.2 and the SILVA 16S rRNA gene database (version 119) as the reference (43). Sequences were aligned using PyNAST (44), and a phylogenetic tree was generated via FastTree 2.1.3 (45). Samples were rarefied to an even sequencing depth of 1,168 sequences per sample. Alpha and beta diversity analyses (including UniFrac ) were conducted in QIIME.
The software programs Prism 6 and PRIMER6 were used to statistically analyze the data sets. Permutational analysis of variance (PERMANOVA) was performed in PRIMER6 (version 6.1.13) using Jaccard, Bray-Curtis, and weighted and unweighted UniFrac distances. Multidimensional scaling plots were constructed from the weighted UniFrac distance matrix in PRIMER6. Kruskal-Wallis tests were performed with Dunn's multiple comparison corrections to measure significant differences (P < 0.05) between variables.
Raw sequences were uploaded to the NCBI Sequence Read Archive, and the accession numbers can be found via BioProject record number PRJNA320336.
A total of 24 patients with CSOM were recruited, along with 22 healthy controls. Most patients with cholesteatoma had extensive middle ear and mastoid disease (grades 3 to 5). The patients' demographic information is presented in Table 1.
Swab samples collected from the control (CI and T) and noncholesteatoma patients with the CWU procedure led to no microbial growth on culture plates. In contrast, patients with cholesteatoma (CWU and CWD procedures) had positive growth for both middle ear (40 to 60%) and mastoid (20 to 50%) samples (Fig. 1A). None of the negative controls produced any amplification products after the nested PCR amplification process and were considered to be negative. A range of organisms were identified as Gram-positive or anaerobic and included Corynebacterium jeikeium and Shewanella algae. In addition, noncholesteatoma patients undergoing a CWD procedure had higher rates of positive cultures from the middle ear (80%) than from the mastoid (20%) swab samples. Microbes identified in these patients included Klebsiella oxytoca, P. aeruginosa, C. jeikeium, Aspergillus flavus, and methicillin-resistant S. aureus.
16S rRNA gene sequences could be obtained from middle ear and mastoid samples of 50 to 80% of cholesteatoma patients, 35 to 80% of noncholesteatoma patients, 43% of control CI patients, and 14% of control T patients (Fig. 1B). The success in obtaining amplifiable DNA was not consistent between sample sites (middle ear and mastoid) for each patient. The number of sequences per sample varied between 1,168 and 86,952. The total numbers of OTUs obtained after quality trimming of the sequences across all mastoid (n = 21) and middle ear (n = 25) samples were 156 and 163, respectively. Noncholesteatoma patients had the lowest bacterial diversity compared with the cholesteatoma and control groups, as measured by the number of observed species and Shannon and Simpson indices (see Fig. S1 in the supplemental material). However, these observed differences were not significant (Kruskal-Wallis test).
Interpatient differences explained the majority of the variation in bacterial composition observed in the data set of CSOM patients (R2 = 34.4%, P = 0.002). Interestingly, such high variation was not observed in the control group. In addition, no significant differences were measured between the sampling site (mastoid versus middle ear) or disease status or subgroupings or age.
The microbiota of the healthy ear was composed of a range of bacteria, including members of the genera Novosphingobium, Staphylococcus, Streptococcus, Escherichia-Shigella, and Burkholderia (Fig. 2). The bacterial compositions of CSOM were also variable between patients and included members of the genera Haemophilus, Staphylococcus, Alloiococcus, and Streptococcus. The genus Pseudomonas was found in a minor proportion (4/46) of samples and with variable relative abundance ranging from 0.01 to 99%. Staphylococcus, another commonly cultured organism from ear discharge, was detected in 39% (18/46) of the samples, including healthy controls, with relative abundance ranging from 0.01 to 99%. Propionibacterium was found in the middle ears and mastoids of both control and CSOM patients (<11% relative sequence abundance). In addition, Alloiococcus and Haemophilus were detected largely in cholesteatoma patients (Fig. 3).
All samples from this study with sequence information were plotted on a multidimensional scaling (MDS) plot using weighted UniFrac distances to visualize clustering of sample types (Fig. 4). However, no clustering was observed based on disease status (with and without CSOM) or sampling site (mastoid versus middle ear) or the CWU or CWD procedure.
Our current understanding of the microbiology of the middle ear and mastoid mucosa in healthy and diseased states has been largely derived from bacterial cultivation studies. These results are inevitably skewed due to the failure of some bacteria to grow in standard culture media (47). In contrast, cultivation-independent molecular techniques are able to provide a more accurate assessment of the microbial communities growing on the mucosa. Here, we applied molecular techniques to determine the microbiota in both patients with CSOM and those with healthy middle ear mucosa, and our findings are very different from those of previous culture-based studies.
It has been a long-held belief that the middle ear of a healthy individual is sterile (22, 23). In contrast, the results from this study show a diverse bacterial community being detected in 45% of healthy mastoids and middle ears. Notably, our cultivation-based data, which are targeted toward suspected pathogens, detected no bacteria in either the middle ears or mastoids of healthy individuals. Only with the application of cultivation-independent, molecular biology techniques were bacteria detected in these patients. Even then, a highly permissive PCR protocol (comprising >45 PCR cycles) was required, suggesting that bacterial loads on the mucosa of healthy individuals are low. In addition, the swab samples contained traces of blood, which could have potentially inhibited PCRs in some samples. Future studies should focus on confirming the findings from this study by using alternative approaches such as FISH.
CSOM is thought to result from bacteria ascending from the nasopharynx via the Eustachian tube (48) or invading from the external auditory meatus (ear canal) through a perforation of the tympanic membrane. This ultimately causes irreversible mucosal changes that produce chronic otorrhea (ear discharge) and lead to hearing loss. However, our demonstration that many healthy middle ears do in fact contain bacteria demands consideration of a potential new concept for initiation of CSOM. Infections within the middle ear and mastoid may need to be viewed as a consequence of a disturbance to the normal balance of commensal bacteria (i.e., dysbiosis), at least in some individuals. Under such a model, pathogenesis would likely still involve invasion of bacteria from the nasopharynx or ear canal, but after a perturbation of the resident bacterial community rather than invasion of a previously sterile space.
Alternatively, or in addition to the above, CSOM may arise from middle ear or mastoid commensal bacteria alone, without contributions from adjacent anatomical sites. Indeed, potential pathogens, such as Staphylococcus, Pseudomonas, Streptococcus, and Moraxella, which have been implicated in ear disease using cultivation techniques, were also detected among healthy controls, supporting the notion of the resident microbiota as a potential source of infection from within. Perturbations to the commensal microbiota could conceivably occur due to, for example, excessive antibiotic usage and/or compromised immunity.
At other mucosal sites, commensal organisms play important roles in defense against opportunistic bacterial invasion. In the gut, for example, communication and regulation between bacteria can have positive or negative effects on bacterial growth. Such interactions appear to be mediated through signaling molecules released by bacteria and reabsorbed by the host cell (49, 50). It is possible, but remains to be demonstrated, that similar interactions occur in the middle ear.
The majority of the variation in the bacterial communities of CSOM patients was accounted for by interpersonal differences. As this observation was not seen for controls, it further supports the suggestion that the bacterial communities of CSOM patients are disrupted, leading to an imbalance in community structure. Such observations have been reported previously in the gut (51) and paranasal sinuses (52, 53). Interestingly, sample site (mastoid versus middle ear) within an individual had no influence on bacterial community variation, suggesting that similar communities were found at both sites. Substantial interpatient variation has been observed in a number of human sinus, gut, and skin microbiomes (35, 54, 55). Ethnicity has been shown to contribute to some of this variation (56). Due to the small patient numbers in this study, no such associations were observed in this study (Table 1). It is possible that with a larger study (more patients), ethnicity signatures might become more apparent.
The diversity of pathogens/bacteria identified in swab samples by hospital cultivation techniques is typically an underestimate. Routine diagnostic hospital laboratories will normally selectively culture for known pathogens; therefore, other cultivable organisms may be overlooked. Notwithstanding this caveat, given that much of the human microbiota is uncultivable or extremely hard to grow (47), molecular tools are able to offer better insights into the bacterial community composition of a given sample. Based on the results of this study, we did not observe a specific microbial profile typical of CSOM. Putative pathogens such as Staphylococcus, Pseudomonas, and Haemophilus were also observed in healthy control microbiota. The next step for this research is to elucidate the influence of the microbiota on the host.
We demonstrated the presence of Propionibacterium in normal mastoid and middle ear mucosa. The relative sequence abundance of Propionibacterium was significantly higher in controls than in patients with CSOM. Members of the genus Propionibacterium are part of the normal microbiota of the skin, mouth, and gut (57,–60). They are believed to play a role in stabilizing the normal microbiota by occupying niches which then cannot be invaded by pathogens (58, 59). Moraxella is a known pathogen of the human respiratory tract, typically causing pneumonias in adults and acute otitis media in children (48). It is not a classic pathogen reported in CSOM. The occurrence of Moraxella was variable between groups, but it was mostly found in control and noncholesteatoma patients at relatively low abundance. Our findings thus support it being part of the commensal microbiota in normal middle ears and mastoids but not a significant player in CSOM.
There has been debate about whether members of the bacterial genus Alloiococcus play a role in the pathogenesis of otitis media in children or if they are part of the commensal microbiota (61, 62). Using molecular techniques, we found higher relative abundance in CSOM patients with cholesteatoma than in controls or other subject groups. To better understand the potential role of Alloiococcus in CSOM, further study into host interaction and colonization of adjacent sites in normal and diseased patients will be necessary. Of note, Staphylococcus was identified in the microbiota of many patients in our study, but was Pseudomonas detected in relatively few patients. These results contrast with those of many culture-based studies in which P. aeruginosa and S. aureus were the dominant bacteria (10, 13−16, 18). Differences in the capabilities of certain species to grow in culture media may account for much of the disparity between culture-based and molecular techniques (47). The preliminary findings of this small study will need to be further validated in the future with larger cohort sizes.
In light of our findings, the importance of Pseudomonas as a pathogen in CSOM requires further investigation. Patients in this study would not have benefited from antibiotics targeting Pseudomonas in CSOM. This is clinically important, as overuse of antipseudomonal antibiotics can promote bacterial resistance (10, 17, 27). More generally, the use of broad-spectrum antibiotics may have an unfavorable effect on commensal bacterial communities, potentially affording pathogens a competitive advantage by eliminating competing commensal bacteria. Antibiotics administered early in life may alter the biomass and composition of the normal microbiota and reduce the exposure of potential pathogens to the immune system. In studies on gut microbiota, these mechanisms have been reported to result in dysbiosis (63). It is possible that antibiotic use for CSOM early in life may cause dysbiosis and may have a detrimental effect on a patient's long-term ear health. Further studies are needed to examine the effect of antibiotics on the middle ear microbiota in health and disease.
In conclusion, the healthy middle ear and mastoid mucosa are dependent on normal Eustachian tube function and mucosal immunity and an absence of allergy and infection (3). In human beings, commensal bacteria have been shown to coexist to maintain a healthy environment in a variety of body sites, and this study suggests that a bacterial community is often present in the normal human middle ear and mastoid. There is a disparity between culture and molecular assessment of the middle ear and mastoid, which may lead to inappropriate prescribing of antibiotics. The use of molecular techniques to describe the microbiota should further refine our understanding of the role of the middle ear and mastoid microbiota in the pathogenesis of chronic inflammatory ear diseases.
We thank all of the participants of this study.
The research reported here was funded by an A+ trust research grant (Auckland District Health Board).
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JCM.01068-16.