Introduction
Otitis media with effusion (OME) is one of the most common infectious diseases in the paediatric age group. Several predisposing factors have been suggested, including eustachian tube dysfunction, allergy, passive smoking, viral upper respiratory tract infections, and adenoidal hypertrophy or adenoiditis.Reference Ebmeyer, Furukawa, Pak, Ebmeyer, Sudhoff and Broide1,Reference Rovers, Schilder, Zielhuis and Rosenfeld2 Recent studies have investigated the possibility of a causal relationship between OME and the presence of Helicobacter pylori in the middle ear.Reference Agirdir, Bozova, Derin and Turhan3–Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8
Helicobacter pylori is a microaerophilic, Gram-negative, spiral organism which inhabits the mucus layer overlying the gastric epithelium.Reference Schreiber, Konradt, Groll, Scheid, Hanauer and Werling9 Worldwide, it is one of the most frequent infections. The prevalence rate of the disease is 20–50 per cent in developed countries and 80–90 per cent in underdeveloped countries. It has been shown to be associated with chronic gastritis, gastric and duodenal ulcer, and gastric cancer.Reference Suerbaum and Michetti10 These bacteria are most probably transmitted by the gastro–oral route, and have been identifiable in saliva and dental plaque.Reference Nguyen, Engstrand, Genta, Graham and el Zaatari11,Reference Li, Musich, Ha, Ferguson, Patel and Chi12 Helicobacter pylori has also been discovered in tonsil and adenoidal tissue, nasal polyps, nasal mucosa, and maxillary sinus mucosa.Reference Yilmaz, Aktepe, Cetinkol and Altuntas7,Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8
Direct transmission of H pylori from the stomach to the middle ear by reflux may occur.Reference Sudhoff, Bücker, Groll, Shagdarsuren, Dazert and Schreiber13 High pepsin and pepsinogen levels have recently been demonstrated in persistent middle-ear effusions in children, suggesting a role for gastroesophageal reflux in middle-ear disease.Reference Tasker, Dettmar, Panetti, Koufman, Birchall and Pearson14 A recent series of papers linking H pylori to OME contains contradictory data, and the detection methods employed, and, indeed, the study groups, are inconsistent.Reference Agirdir, Bozova, Derin and Turhan3–Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8
In this study, we aimed to clarify the role of H pylori in the pathogenesis of OME, based on the current evidence base.
Material and methods
Literature search
We undertook a literature search on 21st September 2006, using the Medline database, using search terms including: ‘Helicobacter pylori’, ‘otitis media with effusion’, ‘glue ear’ and ‘middle ear’. Six original research papers were identified.Reference Agirdir, Bozova, Derin and Turhan3–Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8 Two of these were randomised, controlled studiesReference Agirdir, Bozova, Derin and Turhan3,Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8 and four were prospective, cohort studies.Reference Bitar, Mahfouz, Soweid, Racoubian, Ghasham and Zaatari4–Reference Yilmaz, Aktepe, Cetinkol and Altuntas7
Randomised, controlled studies
In the first of the two randomised, controlled studies, Agirdir and co-workers investigated 45 patients between the ages of three and 13 years with a diagnosis of chronic otitis media with effusion (OME) and adenoidal hypertrophy.Reference Agirdir, Bozova, Derin and Turhan3 The study group comprised 30 patients (18 males (60 per cent) and 12 females (40 per cent)). Adenoidectomy with myringotomy, with bilateral ventilation tube insertion, was performed in this group. Middle-ear effusion samples and fragments of adenoid tissue were collected and the presence of H pylori was investigated by rapid urease testing (termed the campylobacter-like organism test) (Table I). The control group, comprising 15 age-matched patients (nine males (66.7 per cent) and six females (33.3 per cent)) also underwent adenoidectomy with myringotomy. These patients had a dry tap on myringotomy. Middle-ear washings and adenoid tissue samples were collected from the control group. Using the campylobacter-like organism test, these control samples were also tested for the presence of H pylori. The campylobacter-like organism test was positive for the middle-ear effusion samples of 20 (66.6 per cent) patients in the study group. None of the control group patient middle-ear washings had a positive campylobacter-like organism test result. There was no significant difference between the campylobacter-like organism test positivity for the adenoid tissue samples of the two groups. Agirdir et al. concluded that their findings demonstrated the presence of H pylori in middle-ear effusion fluid, and surmised that this may be responsible for the pathogenesis of OME.
Table I Summary of studies addressing relationship between helicobacter pylori and otitis media with effusion
CLO = campylobacter-like organism test; Hp = Helicobacter pylori; PCR = polymerase chain reaction; OME = otitis media with effusion
The second randomised, controlled trial, by Yılmaz and co-workers, studied aspirated middle-ear fluid and biopsies from the middle-ear promontory of 22 children with OME (nine girls and 13 boys, age range two to 13 years).Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8 All Children diagnosed with OME, adenoid hypertrophy and chronic tonsillitis underwent bilateral ventilation tube insertion and adenoidectomy and tonsillectomy. The control group comprised 20 children undergoing adenotosillectomy for chronic adenotonsillitis, without any signs of middle-ear disease. In this control group, a small biopsy was taken from the promontory after myringotomy. In both groups, 5-mm deep tissue specimens were obtained from the tonsils and adenoids, together with gastric juice samples, for H pylori analysis (Table I). Polymerase chain reaction analysis and microbiological culture indicated a statistically significant (p < 0.05) increase in the presence of H pylori in gastric juice and in tonsil, adenoid and middle-ear promontory samples in the study group, compared with the control group.
Prospective, cohort studies
Bitar and colleagues studied 18 consecutive paediatric patients (mean age 4.4 years; age range three to eight years; equal gender distribution) undergoing myringotomy and adenoidectomy for chronic OME or recurrent otitis media.Reference Bitar, Mahfouz, Soweid, Racoubian, Ghasham and Zaatari4 Aspirated middle-ear fluid samples were cultured on three types of agar (Brucella + laked horse blood, Brucella + sheep blood, and ‘chocolate’). A double polymerase chain reaction test was used to detect urease-C and adhesion subunit genes. Adenoid specimens underwent rapid urease enzyme and polymerase chain reaction testing (Table I). Middle-ear cultures were negative in all three media, for all 28 middle-ear fluid samples. Twenty-one of the 28 samples contained deoxyribonucleic acid (DNA), but polymerase chain reaction analysis revealed that none of this DNA was from H pylori. Ten of the 13 adenoid specimens obtained were positive for H pylori on rapid urease testing, but none were positive on polymerase chain reaction.
Morinaka and co-workers performed another prospective, nonrandomised study with nonpaired, nonmatched controls.Reference Morinaka, Tominaga and Nakamura6 They collected 15 effusion fluid samples from 15 patients with OME (male:female = 4:13; age range three to 85 years; mean age 54 years). Twelve gastric tissue biopsy samples from seven H pylori infected patients with gastric or duodenal ulcers were used as positive controls. Microscopy smears were prepared from the samples and immunostained with anti H pylori antibody, as well as being subjected to Gram and Giemsa staining. In addition, the middle-ear effusion and gastric tissue biopsy samples underwent campylobacter-like organism testing (Table I). The immunostained smears revealed spiral or curved, immunoreactive organisms in 12 (80 per cent) of the 15 middle-ear effusion specimens. The Giemsa-stained smears showed blue, spiral or curved organisms in 14 patients' samples. Three patients with positive immunohistochemistry also had a positive campylobacter-like organism test for OME samples. All 12 control gastric biopsy specimens from the seven positive control patients tested positive for H pylori using immuno- and Giemsa staining. Gram-staining showed Gram-negative bacteria.
Yılmaz and colleagues divided 38 consecutive children (average age 7.85 years) with adenoid hypertrophy and/or OME into two groups.Reference Yilmaz, Aktepe, Cetinkol and Altuntas7 The first group comprised 18 subjects with OME and adenoid hypertrophy, and the second group comprised 20 subjects having only adenoid hypertrophy. Each patient underwent the appropriate surgical procedure: myringotomy, placement of tympanostomy tubes and/or adenoidectomy. At myringotomy, middle-ear effusion samples were collected using a suction and collection device, and a core biopsy specimen was taken from each piece of adenoidal tissue following adenoidectomy. Deoxyribonucleic acid extracted from these samples was used to amplify the 23S ribosomal ribonucleic acid gene of H pylori by real-time polymerase chain reaction. In the first group, 34 effusion samples were obtained from the ears of 18 patients (two had unilateral OME). Real-time polymerase chain reaction analysis for H pylori was positive for 12 of the 18 children (67 per cent) and for 16 of the 34 ears (47 per cent). No positive reaction was seen in these patients' adenoid tissue samples. In the second group, a positive reaction was seen in the adenoid tissue of only one patient.
Karlidag and co-workers studied 38 patients with OME.Reference Karlidag, Bulut, Keles, Kaygusuz, Yalcin and Ozdarendeli5 In all cases, myringotomies were carried out (age range two to 12 years; male:female = 23:15). Aspirated middle-ear effusion samples were analysed by polymerase chain reaction. A total of 55 aspiration samples was collected. Seventeen patients had bilateral effusions, whereas 21 had unilateral effusions. Nine (16.3 per cent) of the 55 middle-ear effusion samples were shown to be H pylori positive by polymerase chain reaction.
Discussion
The behaviour of pathogenic bacteria within the middle ear is poorly understood. In recent years, advances in animal model research have led to major improvements in our knowledge of host–pathogen interactions in otitis media with effusion (OME).Reference Sudhoff, Bücker, Groll, Shagdarsuren, Dazert and Schreiber13,Reference White, Heavner, Hardy and Prazma15 However, OME, characterised by the persistence of a middle-ear effusion for three months or more, remains an unsolved problem. Recently, the relationship between H pylori and OME has been evaluated.
Bacteriological studies of OME using highly sensitive molecular biology techniques, such as polymerase chain reaction, have demonstrated that traditional culturing methods are inadequate to detect many viable bacteria present in OME. The presence of pathogens attached to the middle-ear mucosa as a bacterial biofilm, rather than as free-floating organisms in a middle-ear effusion, has previously been suggested to explain these observations.Reference Post16
The majority of the authors of papers identified in our systematic review concluded that they found interesting indications for a possible role of H pylori in OME. All of these authors proposed that further studies are needed to investigate the role of H pylori in the aetiology of OME.
Other authors who detected the presence of H pylori in the upper aerodigestive tract and middle ear have speculated that the presence of H pylori in the middle ear, adenoid and tonsillar tissue is most probably related to reflux of gastric contents into the pharynx.Reference Axon17 Acid reflux leads to mucosal inflammation and oedema in the nasopharynx, disturbing eustachian tube clearance and thus enabling nasopharyngeal bacteria to enter the middle ear. Helicobacter pylori may be one such bacterial species. Adenoid tissue may also act as a nidus of infection, disseminating H pylori to the middle ear via the eustachian tube in patients with OME. This theory is supported by evidence of significantly increased H pylori colonisation in the adenoids of patients with OME, compared with those with adenoid hypertrophy without OME.Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8
Is there a suitable microenvironment to sustain H pylori growth?
In order to reproduce itself, H pylori needs a microaerobic environment (i.e. 10 per cent CO2). Such a microaerobic environment is present in the middle ears of OME patients. The pH value of the luminal side of the mucus layer is 1.0–2.0, but on the mucosal side, which H pylori inhabits, it is approximately 5–7.Reference Schreiber, Konradt, Groll, Scheid, Hanauer and Werling9 The pH value of middle-ear effusions in chronic OME patients was found to be between 7.0 and 9.0. Such an alkaline environment will allow H pylori to survive; however, its growth will be impeded. Helicobacter pylori requires an acidic environment in order to survive in the presence of urea.Reference Clyne, Labigne and Drumm18,Reference Sjöström and Larsson19 None of the reports of its presence in extragastric sites are completely proven, since H pylori is highly localised to its ecological niche in the gastric mucus and needs dedicated conditions for growth.Reference Van and van der20 The presence of H pylori in other tissues might be explained by its existence as a transient pathogen. The reported finding of H pylori in the upper aerodigestive tract may give a false impression of its ability to survive in a ‘hostile’, non-acidic environment, as demonstrated by Berloco et al.,Reference Berloco, Cavallini, Di and Russo21 who showed daily variation in the amount of H pylori genomic material detected in the saliva of H pylori infected patients. More studies are needed to clarify this issue before a clear and definitive conclusion is reached regarding the significance and behaviour of H pylori detected in the upper aerodigestive tract.
How is the infection transmitted?
The exact route of H pylori transmission is still not clear. Most infections seem to be caused by human-to-human contacts within the familyReference Drumm, Perez-Perez, Blaser and Sherman22,Reference Rothenbacher, Bode, Berg, Knayer, Gonser and Adler23 through a ‘gastro–oral’ route.Reference Luzza, Mancuso, Imeneo, Contaldo, Giancotti and Pensabene24 Faecal and oral shedding of H pylori from healthy but infected children has been demonstrated.Reference Parsonnet, Shmuely and Haggerty25 A faecal–oral route,Reference Laporte, Pernes, Pronnier, Gottrand and Vincent26,Reference Thomas, Gibson, Darboe, Dale and Weaver27 possibly involving contaminated water,Reference Klein, Graham, Gaillour, Opekun and Smith28 has been discussed. Data on intrafamilial transmission of H pylori among children continue to be produced. Due to improved living standards, the prevalence of H pylori infection in industrialised countries has fallen to 5–50 per cent, depending on age.Reference Blecker, Lanciers, Hauser and Vandenplas29,Reference Grimm and Fischbach30 As it is not present in blood, H pylori may travel into the middle ear via gastro-esophageal reflux.
Tasker et al. investigated the presence of gastric juice in the middle ear in 65 children with chronic OME; they found that 91 per cent of their effusion samples contained pepsin and pepsinogen at levels that made gastric juice the most likely source.Reference Tasker, Dettmar, Panetti, Koufman, Birchall and Pearson14 However, the antigens used in this study were not specific and showed massive cross-reactivity in middle-ear samples.
A study by Rowland et al. also examined prospectively the age-specific incidence of H pylori in children aged younger than 24 months.Reference Rowland, Daly, Vaughan, Higgins, Bourke and Drumm31 They found that children in a developed country such as Ireland become infected at a very young age, usually before five years of age. According to these authors, because of the limitations of the Reference Sudhoff, Bücker, Groll, Shagdarsuren, Dazert and Schreiber13C urea breath test in very young children, the incidence of infection could not be evaluated in children younger than two years.
Which test to apply?
Many diagnostic methods have been developed to detect H pylori infection, both invasive (rapid urease test, histological analysis, microbiological culture and polymerase chain reaction) and non-invasive (serological analysis, urea breath test and, more recently, analysis for H pylori antigen in faeces). At present, no single test is absolutely reliable in detecting H pylori infection. If feasible, a combination of two tests is recommended.
The rapid urease test is a simple, inexpensive, quick and reliable method for detecting urease activity in a specimen, and it is considered by clinicians to be the initial test of choice for the diagnosis of H pylori.Reference McNulty and Wise32 The campylobacter-like organism test, an agar gel test, is the most widely used and best-studied rapid urease test.Reference Tseng, Wang and Wu33 The rapid urease test is based on the ability of H pylori to produce urease, which is not usually present in the stomach. Urease produced by the organism converts urea in the medium to ammonia, resulting in a rise of pH to values above 8 in the weakly buffered medium, causing a colour change in a pH indicator. Rapid urease tests are inexpensive, quick, reliable and can easily be performed in a local laboratory. This is in contrast to culture and histology, which usually take several days to obtain a result. The campylobacter-like organism test had a specificity and positive predictive value of 100 per cent, although its sensitivity is in the range of 90–99 per cent.
The rapid urease test has a high sensitivity in adults but a low sensitivity in children, possibly because the test depends on the concentration of H pylori for adequate urease activity.Reference Madani, Rabah and Tolia34 A positive rapid urease test in non-gastric tissue does not necessarily indicate the presence of H pylori, as the test is designed to detect a pH rise in an acidic sample, as in H pylori samples in the stomach. Use of the rapid urease test in a non-gastric, non-acidic tissue such as the adenoids and middle ear may lead to a high rate of false positive results, due a non-urease-dependent pH rise and to a high number of other bacteria, some of which may produce urease. Therefore, the campylobacter-like organism test will become positive in testing samples with a high buffer value at pH values above 8 (such as middle-ear effusion samples, in the case of OME) and will remain negative in neutral solutions with low buffer values (such as washings from a dry tap). Similar false positive results have been reported in achlorhydric patients, who are believed to have stomachs colonised by commensal organisms capable of producing urease. False positive results were investigated by Bitar et al.;Reference Bitar, Mahfouz, Soweid, Racoubian, Ghasham and Zaatari4 they found a high level of evidence for the presence of H pylori in adenoid specimens, using the campylobacter-like organism test. However, polymerase chain reaction testing of the same adenoid specimens failed to detect H pylori.
Although several diagnostic tests are available for the detection of H pylori infection, all of them have both advantages and disadvantages, and none can be considered as a single ‘gold standard’. A combination of biopsy-based methods (histological examination and culture) usually gives the most reliable diagnosis.
Assessing the six studies analysed, Agirdir et al. detected H pylori, using the campylobacter-like organism test, in 66.6 per cent of OME cases.Reference Agirdir, Bozova, Derin and Turhan3 Yılmaz et al., using a combination of polymerase chain reaction and culture, found H pylori in 45 per cent of their OME patients.Reference Yilmaz, Ceylan, Akyon, Ozcakyr and Gursel8 Morinaka et al. found that 80 per cent of OME fluid samples analysed were positive for H pylori by immunostaining; however, only 20 per cent had viable bacteria which reacted positively in the campylobacter-like organism test.Reference Morinaka, Tominaga and Nakamura6 Yılmaz et al. detected H pylori, using polymerase chain reaction, in 47 per cent of OME samples.Reference Yilmaz, Aktepe, Cetinkol and Altuntas7 Using the same method, Karlidag et al. detected H pylori in 16.3 per cent of OME samples.Reference Karlidag, Bulut, Keles, Kaygusuz, Yalcin and Ozdarendeli5 Bitar et al. failed to detect H pylori by either culture or polymerase chain reaction, and concluded that H pylori plays no part in OME in children and does not colonise adenoid tissue.Reference Bitar, Mahfouz, Soweid, Racoubian, Ghasham and Zaatari4
In the studies assessed, bacterial culture of H pylori had the highest rate of false negative results, which decreased its sensitivity (80 per cent), in comparison with other tests (86–100 per cent). However, its specificity was high (100 per cent), as it constituted definite proof of the presence of viable bacteria.Reference Kisa, Albay, Mas, Celasun and Doganci35 Despite this, difficulties in the isolation and culture of H pylori, and the demanding technical requirements involved, do not allow culture to be used as a single gold standard.
• There is growing interest in the presence of Helicobacter pylori in the upper aerodigestive tract and in the middle ears of patients with otitis media with effusion (OME)
• Direct transmission of H pylori from the stomach to the middle ear by reflux may occur
• This study critically evaluated the evidence for a possible role of H pylori in the middle ear with OME
• At present, there is poor evidence for the existence of H pylori-associated OME, and further research in the field is needed
Bacterial culture indicates the presence of living bacteria; however, polymerase chain reaction indicates the presence of bacterial DNA rather than the bacteria itself. Each method complements the other. When both are applied together, the sensitivity and specificity of results increase.
The anti H pylori antibody test has an accuraty between 80–90 per cent.Reference Malfertheiner, Megraud, O'Morain, Bazzoli, el-Omar and Graham36 Giemsa staining has been reported to have a sensitivity of 91 per cent and a specifity of 100 per cent.Reference Tokunaga, Shirahase, Yamamoto, Inao, Hamaguchi and Kanaji37
In the study by Morinaka et al., even though the majority of OME samples were positive for the presence of H pylori, only three of the 15 samples showed viable organisms.Reference Morinaka, Tominaga and Nakamura6 This shows that, although there may be reflux of gastric juice into the nasopharynx or middle ear, this in itself fails to prove that H pylori can survive in the middle ear and influence its physiology.
Can we distinguish between H pylori detection and infection?
The reason why some individuals remain infected with H pylori for life without displaying any symptoms, while others develop severe disease, has only been partially clarified. Presumably, it depends on multifactorial interactions between host immunological and physiological factors, bacterial virulence determinants, and environmental influences modulating the host response. The ability of H pylori to overcome defence mechanisms on mucosal surfaces, as well as to modulate the immune response by interfering with the gastric epithelial and immune cells seem to be the crucial factors for the infectionReference Nyström, Raghavan and Svennerholm38, however, a lot of questions are still unsolved. In contrast to the widely accepted “oral-oral” route of Helicobacter pylori transmission, it is possible to detect molecular traces of H. pylori in the environment.Reference Queralt, Bartolome and Araujo39 Since it is not possible to culture H. pylori from these samples, molecular traces do not indicate a colonization. In the studies assessed, efforts to isolate H pylori from OME effusion samples by culture also often failed. In some cases this may indicate a correct detection of Helicobacter traces without colonisation or infection.
Is the control group reliable?
Rees et al. showed that, in 13 per cent of patients with proven middle-ear effusion prior to anaesthesia, there was some displacement of the effusion during the early stages of nitrous oxide anaesthesia.Reference Rees and Freeland40 In the study by Bitar et al., general anaesthesia was induced via thiopental and continued using a combination of nitrogen protoxide, oxygen and sevoflorane, in each patient. Therefore, the selection of controls in the studies assessed seems inappropriate and implies a methodological bias.
Conclusion
From the studies assessed, we found no conclusive evidence for the existence of H pylori-associated otitis media with effusion (OME) in children. Reported data may suggest that acid reflux is the underlying aetiology explaining the presence of H pylori as a contaminant in the middle ear. All reviewed papers used indirect methods for the detection of H pylori, such as the campylobacter-like organism test, polymerase chain reaction without sequence confirmation, or immunohistochemistry. All currently published findings require additional confirmation. Further research in this field is strongly needed in order to delineate the presence of H pylori and to establish its role in the pathogenesis of OME.
