Identification of the Bacterial Pathogens in Children with Otitis Media: A Study in the Northwestern Portuguese District of Braga

Understanding the bacterial etiology of otitis media (OM) is important when designing and evaluating the best course of treatment. This study analyzed middle ear fluid (MEF) and nasopharynx (NP) samples collected from 49 children with OM undergoing myringotomy in the northwestern Portuguese district of Braga. A correlation between species in the NP and MEF was observed following pathogen detection by culture and quantitative polymerase chain reaction (qPCR) methods. Bacterial identification using culturing methods showed that Moraxella catarrhalis was the most representative in NP and MEF, followed by Streptococcus pneumoniae. However, qPCR of MEF showed a higher prevalence (61%) of Haemophilus influenzae. S. pneumoniae was not the most frequently identified species, but it still remains one of the leading causes of OM in this region despite 93.9% of the children being vaccinated with the pneumococcal conjugate vaccine. Furthermore, 46% of the samples analyzed by qPCR identified more than two bacterial species. M. catarrhalis and S. pneumoniae were the most frequent combination identified in NP and MEF samples by culturing methods. Additionally, a few NP and MEF samples simultaneously presented the three main otopathogens. These results point out that polymicrobial infections play an important role in OM. Further studies characterizing the serotypes of the strains isolated, their resistance profile, and their biofilm forming ability would help in the development of more targeted strategies against otitis media.


Introduction
Otitis media (OM) is a significant public health problem in childhood. Middle ear inflammation is the most frequent reason children visit their doctor, take antibiotics, and undergo surgery [1]. Acute otitis media (AOM) and otitis media with effusion (OME) are the two major sub-classifications of OM. In AOM, the presence of middle ear fluid (MEF) is accompanied by otalgia, otorrhea, and fever, among other signs and symptoms of an acute infection. In OME, which can be developed as a sequel to AOM or as a new onset after viral infection, MEF is present in the absence of signs and symptoms of infection, with the most common alteration detected being hearing loss [2]. Recurrent AOM (RAOM) are very common, as well as persistent OME (chronic OME, COME), with myringotomy and the insertion of ventilation tubes often being performed for their treatment [1].
As well as its clinical relevance, OM also has a significant economic impact associated with medical and nonmedical costs. For instance, in Portugal, the mean costs per doctordiagnosed OM episode are estimated at 334 euros divided into three categories. The of malignancy, organ transplantation, systemic infectious diseases, and immune deficiency were excluded.
Samples from the nasopharynx (NP) and the middle ear fluid (MEF) were collected during the surgery. Patients' relevant data were registered in a form by the physician responsible for the surgery collecting different data including age, gender, type of otitis media (acute, with effusion, recurrent, chronic), laterality, myringotomy (first intervention or reintervention), antibiotics prescribed prior to surgery, PCV vaccination status. NP samples were collected by gently rotating a sterile cotton wool swab in the nasopharynx, then placed on Amies transport media (Deltalab 300284 or VWR 710-0438P). Before the surgical procedure, the external ear was disinfected with povidone-iodine for 2 min, then a washing step with saline (NaCl 0.9%) was performed to eliminate the antiseptic agent. After myringotomy, MEF was aspirated using a suction device coupled to a collector (Vygon 5534.06). The ethical committee of the hospitals approved the study, and written consent was obtained from all patients involved.

Bacterial Culture
Samples were stored at 4 • C temperature and transported to the laboratory. Once at the lab, NP swabs were immediately cultured (as described below). At the same time, MEF samples were divided into two fractions: one was immediately used for bacterial culture, and another was stored at −80 • C for subsequent bacterial DNA extraction.
All NP swabs and MEF specimens were examined for specific bacterial species using conventional culture methods. Briefly, samples were streaked on 5% (v/v) sheep blood agar and chocolate agar and incubated at 37 • C with 5% CO 2 for 24 to 48 h. They were also streaked on 5% (v/v) sheep blood agar and MacConkey agar and incubated at 37 • C, aerobically, for the same period.
Species identification was also based on specific tests, such as catalase and oxidase tests. S. pneumoniae was specifically identified by the optochin disk test and H. influenzae by the V and X factor growth requirements.

Specific Quantitative PCR of MEF Samples
Detection of specific individual bacterial pathogens in MEF samples was also evaluated by quantitative PCR.
Calibration curves were performed using DNA extracted from individual strains. Briefly, DNA was extracted from cultures of S. pneumoniae R6st, H. influenzae C894248, M. catarrhalis U225012, S. aureus ATCC 6358, and P. aeruginosa PAO1 (Table 1) using the E.Z.N.A. Bacterial DNA Kit (OMEGA bio-tek) following the manufacturer's instructions. The concentrations and purity ratios were measured (Nanodrop), and the DNA integrity was confirmed by agarose gel electrophoresis (2% agarose gel). Initially, primers described by Sillanpaa et al. (2016) were tested by conventional PCR. Amplification was performed using primers (Table 2) [11]. Three pairs of primers targeting the 16S rRNA genes of S. pneumoniae, H. influenzae, and M. catarrhalis, originally reported by Hendolin et al. (1997), were also used [17]. The multiplex-PCR reaction described by the authors was adapted for KAPA-Taq DNA Polymerase conditions. The other three pairs of primers target the lytA gene of S. pneumoniae, the nucA gene of S. aureus, and the oprL gene of P. aeruginosa. Genes were amplified by PCR using KAPA-Taq DNA Polymerase following the manufacturer's instructions, and amplification confirmed by agarose gel electrophoresis ( Figure S1). Then, calibration curves by quantitative PCR were performed using 1:10 DNA dilutions for each bacterial species and Xpert Fast SYBR Mastermix (Grisp, Portugal) following the manufacturer's instructions ( Figures S2 and S3).
MEF samples (approximately 250 µL) were defrosted and spiked with 1 µL of a plasmid (empty pET-28a(+)) as an exogenous internal control. The DNA was extracted using the E.Z.N.A. Universal Pathogen Kit (OMEGA bio-tek) following the manufacturer's instructions. The detection of the specific pathogens was performed by individual qPCR reactions using the primers described above and Xpert Fast SYBR Mastermix (Grisp, Portugal), in duplicate. The exogenous internal control was detected using the universal T7 primers (Forward (5 → 3 ): TAATACGACTCACTATAGGG; Reverse (5 → 3 ): GCTAGT-TATTGCTCAGCGG).

Study Population
From January 2018 to February 2020, 49 samples were collected from the nasopharynx (NP) and the middle ear fluid (MEF) of children with OM. The patients' sex was well-distributed, and the mean age of the individuals was 3.5 years (±1.7), ranging from 1 to 9 years old, and distributed as in Figure 1. Patients' history questionnaires were filled in by the responsible doctors and are compiled in Table S1. Seven of the cases were classified as RAOM, while the other 42 were due to OME, of which 15 were COME. Moreover, 29 children had bilateral OM, while the remaining 20 had unilateral OM on the right (15) or left (5) ear. The myringotomy was performed for the first time, except in two cases where these were reinterventions (Table  S1, patients 1 and 2). Furthermore, in only two cases, adenoidectomy was not performed at the same time as the myringotomy. Only three of these children had not been vaccinated with a pneumococcal conjugate vaccine (PCV). Moreover, 57% of the children had not taken antibiotics in the last six months. Most of the children taking antibiotics one week to six months before the myringotomy received amoxicillin/clavulanic acid, while a few received penicillin (2), ceftriaxone (1), or cefixime (1) six months prior to the surgery.

Identification of Bacterial Species
The presence of bacterial pathogens in NP and MEF of 49 children with OM was detected by culture, and MEF samples were further evaluated by qPCR.
Concerning the MEF samples, otopathogens were identified in 93.88% of them, either by culture or qPCR, with 57.14% or 89.80% of the samples being Culture+ or PCR+, respectively, for at least one pathogen. The most common bacteria isolated by culture from the MEF (Figure 2b) was again M. catarrhalis (40.82%), followed by S. pneumoniae (38.78%), with H. influenzae and S. aureus being isolated from the same number of samples (14.29%). M. catarrhalis was once more frequently found in co-occurrence with S. pneumoniae in the MEF ( Figure 3b  Patients' history questionnaires were filled in by the responsible doctors and are compiled in Table S1. Seven of the cases were classified as RAOM, while the other 42 were due to OME, of which 15 were COME. Moreover, 29 children had bilateral OM, while the remaining 20 had unilateral OM on the right (15) or left (5) ear. The myringotomy was performed for the first time, except in two cases where these were reinterventions (Table S1, patients 1 and 2). Furthermore, in only two cases, adenoidectomy was not performed at the same time as the myringotomy. Only three of these children had not been vaccinated with a pneumococcal conjugate vaccine (PCV). Moreover, 57% of the children had not taken antibiotics in the last six months. Most of the children taking antibiotics one week to six months before the myringotomy received amoxicillin/clavulanic acid, while a few received penicillin (2), ceftriaxone (1), or cefixime (1) six months prior to the surgery.

Identification of Bacterial Species
The presence of bacterial pathogens in NP and MEF of 49 children with OM was detected by culture, and MEF samples were further evaluated by qPCR.
Concerning the MEF samples, otopathogens were identified in 93.88% of them, either by culture or qPCR, with 57.14% or 89.80% of the samples being Culture+ or PCR+, respectively, for at least one pathogen. The most common bacteria isolated by culture from the MEF (Figure 2b) was again M. catarrhalis (40.82%), followed by S. pneumoniae (38.78%), with H. influenzae and S. aureus being isolated from the same number of samples (14.29%). M. catarrhalis was once more frequently found in co-occurrence with S. pneumoniae in the MEF (Figure 3b   H. influenzae was detected alone by qPCR in 28.57% of the total MEF samples ( Figure  3c). The presence of two or three different species was identified in 20.41% and 22.45% of samples, respectively, while in 4.08% of samples, four pathogens were detected.
More than half of the NP Culture+ isolates were also detected in the MEF sample of the same patient by culture or qPCR (Figure 4). For instance, 76.00% of the S. pneumoniae Culture+ NP samples were positive for S. pneumoniae in the MEF sample, of which 48.00%  H. influenzae was detected alone by qPCR in 28.57% of the total MEF samples ( Figure  3c). The presence of two or three different species was identified in 20.41% and 22.45% of samples, respectively, while in 4.08% of samples, four pathogens were detected.
More than half of the NP Culture+ isolates were also detected in the MEF sample of the same patient by culture or qPCR (Figure 4). For instance, 76.00% of the S. pneumoniae Culture+ NP samples were positive for S. pneumoniae in the MEF sample, of which 48.00% H. influenzae was detected alone by qPCR in 28.57% of the total MEF samples (Figure 3c). The presence of two or three different species was identified in 20.41% and 22.45% of samples, respectively, while in 4.08% of samples, four pathogens were detected.
More than half of the NP Culture+ isolates were also detected in the MEF sample of the same patient by culture or qPCR (Figure 4). For instance, 76.00% of the S. pneumoniae Culture+ NP samples were positive for S. pneumoniae in the MEF sample, of which 48.00% were culture positive (36.00% Culture+/PCR+ plus 12.00% Culture+/PCR+) and 40.00% qPCR positive (28.00% Culture−/PCR+ plus 12.00% Culture+/PCR+) (Figure 4b).

Discussion
The prevalence of bacterial pathogens in children with otitis media has been investigated in different European countries [21][22][23][24], but only one study (which analyzed samples from AOM cases with spontaneous otorrhea) has been performed in Portugal [20]. The implementation of the PCV has been reported to impact OM etiology. This vaccine was first licensed in Europe in 2001, and, since then, countries have been introducing it in their national immunization programs. It started as a 7-valent vaccine (PCV-7), protecting against seven pneumococcal serotypes, later substituted by the PCV-10, and then by the PCV-13 [25]. In Portugal, the PCV-13 was introduced in the routine childhood immunization program in 2015, with the different doses being administered at 2, 4, and 12 months of age [26], similarly to most European countries [27]. Before that, it was recommended and free to children at increased risk for pneumococcal disease [28]. The percentage of children immunized in Europe is high (80%) [29], as were those from which samples were collected (93.88%). After the implementation of PCV, studies reported a change in pathogen dominance from S. pneumoniae to H. influenzae (in France and Greece) [21,22], as well as the replacement of S. pneumoniae serotypes covered by the vaccine by others (in France and Spain) [21,24,30]. Moreover, the high use and misuse of antibiotics may have impacted the prevalence of some bacterial species and serotypes [22]. Amoxicillin-clavulanic acid, the gold standard of otitis media treatment, is the antibiotic most commonly used in Portugal, including in the pediatric population [31]. Importantly, 42.86% of the children included in this study had taken amoxicillin-clavulanic acid up to six months before the myringotomy. Despite this, the antibiotic consumption rate in Portugal is below (but near) the average of European countries [31].
In this study, M. catarrhalis was the most frequent pathogen isolated by culture from NP samples, being in agreement with studies that report this pathogen as the most common NP colonizer in both children with OM or who are healthy [16,32,33]. M. catarrhalis was frequently associated with S. pneumoniae. While colonization by M. catarrhalis alone is not reported to be associated with an increased risk of OM development, this is significantly enhanced when it co-colonizes the NP with S. pneumoniae [33]. The co-infection with M. catarrhalis affected the middle ear ascension of S. pneumoniae in mice and chinchilla models [34]. Moreover, the co-colonization with M.

Discussion
The prevalence of bacterial pathogens in children with otitis media has been investigated in different European countries [21][22][23][24], but only one study (which analyzed samples from AOM cases with spontaneous otorrhea) has been performed in Portugal [20]. The implementation of the PCV has been reported to impact OM etiology. This vaccine was first licensed in Europe in 2001, and, since then, countries have been introducing it in their national immunization programs. It started as a 7-valent vaccine (PCV-7), protecting against seven pneumococcal serotypes, later substituted by the PCV-10, and then by the PCV-13 [25]. In Portugal, the PCV-13 was introduced in the routine childhood immunization program in 2015, with the different doses being administered at 2, 4, and 12 months of age [26], similarly to most European countries [27]. Before that, it was recommended and free to children at increased risk for pneumococcal disease [28]. The percentage of children immunized in Europe is high (80%) [29], as were those from which samples were collected (93.88%). After the implementation of PCV, studies reported a change in pathogen dominance from S. pneumoniae to H. influenzae (in France and Greece) [21,22], as well as the replacement of S. pneumoniae serotypes covered by the vaccine by others (in France and Spain) [21,24,30]. Moreover, the high use and misuse of antibiotics may have impacted the prevalence of some bacterial species and serotypes [22]. Amoxicillin-clavulanic acid, the gold standard of otitis media treatment, is the antibiotic most commonly used in Portugal, including in the pediatric population [31]. Importantly, 42.86% of the children included in this study had taken amoxicillin-clavulanic acid up to six months before the myringotomy. Despite this, the antibiotic consumption rate in Portugal is below (but near) the average of European countries [31].
In this study, M. catarrhalis was the most frequent pathogen isolated by culture from NP samples, being in agreement with studies that report this pathogen as the most common NP colonizer in both children with OM or who are healthy [16,32,33]. M. catarrhalis was frequently associated with S. pneumoniae. While colonization by M. catarrhalis alone is not reported to be associated with an increased risk of OM development, this is significantly enhanced when it co-colonizes the NP with S. pneumoniae [33]. The co-infection with M. catarrhalis affected the middle ear ascension of S. pneumoniae in mice and chinchilla models [34]. Moreover, the co-colonization with M. catarrhalis may negatively impact the treatment since β-lactamase-producing M. catarrhalis can confer passive protection to S. pneumoniae, promoting resistance to the first-line antibiotic amoxicillin [34].
Most of the samples analyzed are from OME cases (85.71%). Therefore, the identification of H. influenzae as the most prevalent pathogen in the MEF agrees with other studies that report H. influenzae as the predominant pathogen in OME [8,14,35]. However, the reported prevalence rates are highly variable due to different health care programs, particularly vaccination, in various geographical locations [36]. The fact that most children were immunized with a pneumococcal conjugate vaccine possibly contributed to a lower incidence of S. pneumoniae. The identification of S. aureus in 42.86% of samples may look somewhat surprising, but a recent study also reports a high prevalence (27.78%) of S. aureus in MEF samples from OME cases [14]. Nevertheless, differences in pathogens' prevalence in studies conducted in the same country have been observed. For instance, two studies conducted in Iran reported a prevalence of H. influenzae of 95.2% in the city of Shiraz [37] and 11% in the city of Tehran [38] using PCR methods. The diversity of the results available may also result from variability in sample collection, the population studied, specific disease stage, and, of course, the detection techniques used (culture, PCR, next-generation sequencing) [36].
PCR methods usually improve the detection of bacteria in MEF samples to a magnitude that varies among the available reports [8,11,39]. Indeed, in the present study, the detection of H. influenzae was largely improved using qPCR compared to culturing (Figure 2b) by 4.3 times. This is in agreement with previous studies, where, for instance, increased detection of H. influenzae of five [40] or ten [19] times using PCR methods compared to culture was reported. H. influenzae is a fastidious organism that is difficult to culture, which could have affected its detection by this approach. Moreover, the fact that PCR methods do not depend on bacterial viability could also contribute to their increased sensitivity [19]. However, it has been shown that the DNA of nonviable bacteria does not persist in the MEF and cannot be detected by PCR 3 days after inoculation in the ME of chinchillas [41]. In contrast, antibiotic-treated bacteria persist in the MEF for weeks and can be detected by PCR but not by culture [41]. Furthermore, the presence of bacterial biofilms has been extensively pointed to being one of the main reasons for Culture−/PCR+ results [42,43].
However, qPCR did not increase the detection of S. pneumoniae and M. catarrhalis, which might be related to the DNA extraction method. It should be noted that, in some cases, only a few colonies (one or two) were isolated from the MEF samples, and the levels of bacteria in the sample might not have been enough for successful DNA extraction. While this hypothesis could be valid, we used an exogenous internal control, consisting of empty plasmid pET-28a(+), to safeguard against the loss of DNA during the extraction procedure or the presence of PCR inhibitors in the sample. This control was successfully detected in 91.84% of the samples. The nature of the sample, such as the viscosity, or the presence of inhibitors, such as blood, might have interfered in the DNA extraction process and may explain the lack of success in detecting the exogenous internal control in 8.16% of samples.
In These show that polymicrobial infections play an important role in OM. Since most of the MEF samples correspond to OME cases (42 of 49), of which at least 15 are chronic, this is in accordance with the disease continuum hypothesis. According to this theory, S. pneumoniae is often involved in an initial AOM episode, which causes mucosal damage and predisposes children to subsequent more complex episodes of RAOM and COME. These most frequently involve non-typeable H. influenzae alone or a mixed-species infection that can be present in the form of biofilms [44].
NP colonization is a usual first step in OM development [5]; therefore, a correlation between species present in the NP and MEF was already expected. However, recent studies suggest that the NP reservoir theory may not explain all etiologies of MEF production [36]. Next-generation sequencing studies have reported similarities [45] and dissimilarities [46] between the MEF and the adenoidal microbiome in children with OME. Therefore, more extensive investigations are needed to validate or disprove the existing theory.
This study presented some limitations. The bioburden could not be quantified with accuracy since the volume of MEF varied between samples, and some of them were diluted in saline during the collection procedure, while others were not. Therefore, qPCR was used only as a qualitative method instead of quantitative. Moreover, using the universal SYBR Green fluorescent dye instead of specific probes for each bacterial pathogen could have reduced the specificity of the qPCR reactions. Furthermore, the presence of other pathogens, such as A. otitis, which may play an important role in both AOM and OME [47,48], was not assessed. Indeed, novel species, particularly Alloiococcus and Turicella, are being identified at significant relative abundances in MEF samples using metagenomic DNA sequencing, which captures all available DNA in a non-discriminatory manner [36].

Conclusions
To our knowledge, this is the first study investigating the prevalence of bacterial pathogens in the middle ear fluid of children with otitis media undergoing myringotomy in Portugal. The majority of the cases had a known diagnosis corresponding to otitis media with effusion, with H. influenzae being the most prevalent causative agent. qPCR was substantially more sensitive than culturing to detect this pathogen, showing the importance of using molecular methods. However, this was not valid for S. pneumoniae and M. catarrhalis, which were more frequently detected by culture. The development of H. influenzae vaccines, which currently only target H. influenzae type B, will be important to reduce the burden of the disease. Moreover, a better understanding of the prevalence of these pathogens and others not included in this study could be achieved by using nextgeneration sequencing methods, which would give information about the whole middle ear microbiome.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms10010054/s1, Figure S1: Agarose gel electrophoresis of amplified genes, Figure S2: Standard curves and melting curves for qPCR quantification of the internal exogenous control (pET-28a(+)), Figure S3: Standard curves and melting curves for qPCR quantification of middle ear fluid bacterial pathogens, Table S1: Demographic information (age, sex) and clinical characteristics of the study subjects.
Informed Consent Statement: Patient consent was waived since they had consented the surgery and the study does not identify the participating patients.
Data Availability Statement: Not applicable.