Patients between 2 and 6 years of age with COME and planned myringotomy and/or insertion of ventilation tubes were included in the study. Surgical intervention was indicated in patients with otoscopic evidence of refractory MEE behind an intact tympanic membrane for at least three months. Prior to surgery, conservative therapeutic measures such as Eustachian tube ventilation systems (Otovent®, Otobar®) as well as nutritive measures to lower gastroesophageal reflux had failed in those patients. Tympanometry and audiometry were performed preoperatively in all patients. The ASA Score was used to evaluate physical status of patients. Patients were included if their ASA Scores were I or II (14). Patients were excluded from the study if they had otoscopic evidence or exhibited symptoms of acute otitis media, acute upper respiratory tract infections, or congenital conditions associated with COME (trisomy 21, mucopolysaccharidosis or cleft palate). Patient history was carefully checked for known immunodeficiency, chronic cardiac or respiratory diseases. Age, sex, weight and height were recorded upon patient admittance, and BMI was classified using WHO criteria (15).

All patients were under general anesthesia during sample acquisition. An indirect transoral nasopharyngoscopy and/or adenoidectomy, and a direct microscopic myringotomy and/or ventilation tube insertion were carried out. MEE was sampled for each ear separately using a special secretion trap (Tracheal-Saugset, Medizintechnik Andreas Fahl, Wiener Neudorf, Austria) and smeared for bacterial culturing. The adenoids were swabbed during nasopharyngoscopy. Blood samples (1.8 ml) were collected during surgery (Sarstedt S- Monovette red 1,8 ml with K3 EDTA, Sarstedt AG & Co.KG, Nümbrecht, Germany).

Samples were processed at the Institutes of Virology, Hygiene and Medical Microbiology and the Department of Dermatology at the Medical University Innsbruck within 2–24 h.

RNA/DNA viral extraction was performed with NucliSENS® easyMag® (bioMérieux, Vienna, Austria). Nucleic acids were extracted from 200 µl of sample to a final volume of 110 µl. In all cases, 2 µl of internal control (IC) was added directly to the lysis buffer of each extraction to check for successful extraction of the nucleic acid and monitor possible PCR inhibition. The respiratory pathogen PCR was conducted using FTD kit (Siemens Healthineers Company, Wien). This commercial kit permits the simultaneous detection of one bacterial and 19 viral pathogens, namely Mycoplasma pneumoniae (M.pneu), influenza virus A (INFA), influenza virus A H1N1 (INFA H1N1), influenza virus B (INFB), parainfluenzavirus 1 (PIV1), parainfluenzavirus 2 (PIV2), parainfluenzavirus 3 (PIV3), adenovirus (Ad), human metapneumovirus A and B (HMPV A/B), coronavirus NL63 (CoV-NL63), coronavirus OC43 (CoV-OC43), coronavirus 229E (CoV- 229E), coronavirus HKU1 (CoV-HKU1), RSV, RV, enterovirus (EnV),human bocavirus (hBoV), human parechovirus (hPeV) and picornavirus. A fluorescence signal from a dual-labeled probe was used to detect the presence of specific viral sequences in the reaction. Presence of viral sequences were recorded as Ct values. The limit of detection was set at >40 cycles, Ct values <30 were interpreted as active viral replication. Thermocycling was performed on a Rotor-Gene 6000/Q (QIAGEN). To validate the results, an extraction control, i.e., the IC—as well as positive and negative controls were carried out as part of the assay (16, 17).

Culture media (Columbia bouillon, Columbia blood agar and and boiled blood agar) for aerobe cultures were inoculated with the smears and incubated over night at 37 ± 2°C for two consecutive nights. Cultures were examined for growth of pathogenes, followed by identification using MALDI TOF (MALDI TOF MS, Bruker Daltonik GmbH, Bremen, Germany). During this process pathogen-specific molecules embedded in a matrix were loosened and ionized with a laser, identified via mass spectrometry and compared to an online data base (18).

MEEs (500 µl–1 ml) were filled up to 10 ml with PBS supplemented with 2% FCS and 2 mM EDTA. Cell suspension was filtered through a 100 µm cell strainer to remove debris. Cell suspension was then split into four 5 ml polystyrene round-bottom tubes (both Falcon® from Corning, New York, NY, USA). Cells were pelleted by centrifugation at 1.400 rpm for 8 min at 4°C. Cell pellets were resuspended in 100 µL of PBS supplemented with 2% FCS and 2 mM EDTA. Samples were stained with either BD Multitest™ CD3/CD16 + CD56/CD45/CD19 or BD Multitest™ CD3/CD8/CD45/CD4 four-color direct immunofluorescence reagent (BD Biosciences, San Jose, CA, USA). After 30 min of incubation at 4°C samples were washed and stained with LIVE/DEAD™ fixable violet dead cell stain kit from Invitrogen (Thermo Fischer Scientific, Waltham, MA, USA). Samples were stained with BD MultitestTM CD3/CD8/CD45/CD4 with a combination of monoclonal antibodies against either PD-1 (BV510-labelled, clone EH12.1), T cell immunoglobulin and mucin-domain containing-3 (Tim-3) (BV421-labelled, clone 7D3), or Lymphocyte-activation gene 3 (Lag-3) (BV421-labelled, clone 74–530), all purchased from BD Biosciences (San Jose, CA, USA). Parallel to MEE, 100 µl of EDTA blood of the study participants was stained as described above. In blood samples, erythrocytes were lysed after the staining using BD Pharm Lyse^TM solution (BD Biosciences, San Jose, CA, USA). Stained cells were fixed in PBS containing 1% formalin and measured using a FACS Canto^TM II cell analyzer (BD Biosciences, San Jose, CA, USA). Data were analyzed with FlowJo^TM software (v10, BD Biosciences, San Jose, CA, USA) (13, 19). Representative data for gating and expression of check-point molecules PD-1, Tim-3 and Lag-3 are shown in Supplementary Figures S1, S2.

Patient data, results viral genome detection and bacterial cultures were collected in an Excel sheet (Microsoft Office, 2016, Redmond, Washington, USA). Statistical analysis was performed with SPSS Statistics (IBM SPSS Statistics 25, Armonk, New York, USA). Fisher’s exact test was used to compare detection rates of viral genome. For nominal and ordinal scaled data, frequencies were recorded. Non-linear correlations of immune cell proportions and risk factors were analysed using Spearman’s Rho. A non-parametric Kruskal-Wallis test was used to determine immune cell proportions and expression of exhaustion markers PD-1, Tim-3, Lag-3 on T cells in blood and MEE.

A total of 69 children were included in the study of whom 48 (70%) were male and 21 (30%) were female. 47% of the children were below the age of 4 years. The weight of the children was classified as healthy weight (41, 59%), overweight (14, 20%) and underweight (14, 20%). Mostly symmetrical tympanograms type B and C were recorded in 67 (97%) and 2 (3%) cases, respectively. Hearing thresholds were mostly between 20 and 40 dB (34, 49%) and between 40 and 60 dB (28, 40%), a threshold between 0 and 20 dB was recorded six times (9%).

Viral genome was detected in MEE of 44/69 children (64%). RV had the highest prevalence (29/69, 43%), followed by PIV (18/69, 26%). RSV was not detected (Table 1). In 19/69 (27.5%) patients, the singular sample contained genome of more than one viral species genome, mostly RV and PIV (11/19 samples). If MEE existed in both ears (27 cases), separate samples of each ear were tested for viral genome. In 14/27 cases, no viral genome was detected in the second ear. Different genomes in were detected in 16/27 cases, when samples of both ears were collected. Coronaviruses were significantly more often detected in cases of bilateral MEE than in unilateral MEE (p = 0.048). The average Ct value was 33.6 (min 21.1, max 39.3). Ct values < 30 were measured in 10 cases (14.5%), RV (7/69, 10%), hPeV (3/69, 4%) and one simultaneous EnV. Association of virus detection in middle ear fluid with the weight was analyzed using the chi-square test. Being overweight was associated with viral presence in middle ear fluid (p = 0.006) (Table 2). Ct values as a measure of viral load were similar in overweight and healthy weight children (both 33.6 on avg.).

The nasopharynx was swabbed and assessed for pathogenic viruses in 37/69 children. Positive PCR results were obtained in 31/37 patients (84%). The highest proportion of viruses detected were PIV (16/37, 43%) and RV (15/37, 40.5%), followed by Ad (8/37,22%), CoV (7/37,19%) and hBoV (6/37,16%) (Table 1). More than one (up to 5) viral species in one nasopharyngeal swab were found in 21/37 (57%) children, with RV and PIV in one third (7/21). The average Ct value was 33.5 (min 21.4, max 38.21). Ct values <30 were observed in 13 cases (35%), mostly for CoV (5/37, 13.5%), RV (4/37, 11%), and hPeV (4/37, 11%) and two simultaneous BoV and PIV. Associations of BMI and virus detection in nasopharyngeal swabs was not significant upon chi-square analysis (data not shown).

Virus-PCR of middle ear fluids and nasopharyngeal swabs yielded concordant results for RVs (Cohen’s kappa = 0.4; p = 0.013), PIV (Cohen’s kappa = 0.55; p = 0.001), and hBoV (Cohen’s kappa = 0.375; p = 0.015). RV and PIV Results were not concordant for CoV (Cohen’s kappa =0.006; p = 0.97) and Ad (Cohen’s kappa = 0.1; p = 0.44). Ct values <30 were measured in both nasopharyngeal swabs and MEE 2 times each.

In 42/69 children bacterial smears were cultured. Positive bacterial cultures of the MEE were observed in 11 children. Haemophilus influenzae and Moraxella catarrhalis were positive in 5 children (12%) and Staphylococcus haemolyticus was positive in 1 child. Detection of bacteria in MEE was not associated with virus detection in MEE (Fisher’s exact p = 0.68) or virus detection in nasopharyngeal swabs (Fisher’s exact p = 0.61). Synchronization of numbers in a Venn diagram revealed bacterial presence in the middle ear 7 times, when viral genome was detected in MEE and the nasopharynx (Figure 1).

Nasopharyngeal swabs, bacterial smears and FACS analysis were not available in all cases due to failures of sample collection and transportation. Application of viral plaque assays, confocal scan laser microscopy and tissue FAXS may enhance aspects of viral replicative activity, bacterial biofilm formation, viral- bacterial interaction and cellular composition in future experiments. In this study, pharyngeal swabs were included for the sole purpose of correlating the results with the viruses detected in the middle ear. Inclusion of additional control groups, such as children with obstructive sleep apnea undergoing adenoidectomy might broaden the knowledge on viral and bacterial profiles that trigger MEE in the future (39) In addition, we could not compare the results of MEE to control group with healthy middle ears at the same age for ethical reasons. However, the feasibility, i.e. virus detection and FACS analysis of middle ear washes, was evaluated in an adult cohort undergoing stapes surgery or cochlear implantation (see Supplementary Table S1).