Electrophoretic analysis of human respiratory syncytial virus (HRSV) messenger RNAs on agarose gel viewed under UV light, amplified by reverse transcriptase–polymerase chain reaction (bp indicates base pairs). Lane 1, Molecular weight marker from Hae III digestion of ϕX174. Lane 2, Positive control (cells transfected with HRSV complementary DNA). Lane 3, Negative control (distilled water instead of RNA extract). Lanes 4 to 11, Several biopsy samples. Samples in lanes 4, 6, 7, and 10 were recorded as RSV positive.
Electrophoretic analysis of interleukin 8 (IL-8) messenger RNAs on agarose gel viewed under UV light, amplified by reverse transcriptase–polymerase chain reaction (bp indicates base pairs). Lane 1, Molecular weight marker from Hinf digestion of ϕX174. Lane 2, Positive sample (cells transfected with human IL-8 complementary DNA). Lanes 3 to 9, Several effusion cell samples. Samples in lanes 5 to 9 were recorded as IL-8 positive. Lane 10, Negative control (distilled water instead of RNA extract).
Moyse E, Lyon M, Cordier G, Mornex J, Collet L, Froehlich P. Viral RNA in Middle Ear Mucosa and Exudates in Patients With Chronic Otitis Media With Effusion. Arch Otolaryngol Head Neck Surg. 2000;126(9):1105-1110. doi:10.1001/archotol.126.9.1105
To evaluate viral and cytokine signaling correlates of the persistent inflammation associated with chronic otitis media with effusion (OME).
Reverse transcriptase–polymerase chain reaction targeting RNA viruses frequently associated with OME (respiratory syncytial virus and parainfluenza virus type 3, the proinflammatory cytokines interleukin 8 and interleukin 1β, and RANTES [regulated upon activation, normal T cell expressed and secreted]) was performed on mucosal biopsy samples and on samples of the liquid and cellular compartments of inflammatory exudates obtained from 26 children (49 ears) with infected middle ears. Ribonucleic acid extracted from rapidly frozen samples was reverse transcribed by Moloney murine leukemia virus reverse transcriptase and amplified for 35 cycles using previously validated primers. Amplicons were evaluated by molecular size after agarose gel electrophoresis with ethidium bromide.
Most children had evidence of the presence of an RNA virus in at least one specimen. Respiratory syncytial virus was present in 40% and parainfluenza virus type 3 in 8% of effusions. Interleukin 8 messenger RNA was present in 21% of inflammatory exudates but never in cells from the mucosal biopsy samples.
Our data support a viral contribution to the cause of OME and suggest that the inflammatory cytokines observed derive more from cells in the inflammatory exudate than from those in the middle ear mucosa.
CHRONIC OTITIS media with effusion (OME) is characterized by the persistence of an effusion for more than 3 months.1 It is believed to result from a combination of blockage of the eustachian tube and infection, resulting in a persistent inflammatory process in the middle ear, and it may appear after either bacterial or viral acute otitis media.2 Associated viruses are those commonly infecting the rhinopharynx, such as adenoviruses, human respiratory syncytial virus (HRSV), or human parainfluenza viruses (HPIV) 1, 2, or 3,3 which may be found at moderate to high frequencies in OME effusions.4,5 Persistent inflammation might also be sustained by epithelial secretion of cytokines,6,7 and airway epithelial cells have been shown to produce proinflammatory cytokines (interleukins) (IL-1, IL-6, IL-8, and tumor necrosis factor) after stimulation by noxious substances, such as atmospheric pollutants, pollen, allergens, or microbial agents,8 as well as after viral stimulation in vivo9 or in vitro.10- 12 Proinflammatory cytokines have been detected in effusions from patients with OME.13- 15 The presence of IL-8 was well correlated with the inflammatory index,16 and IL-8 can induce OME-like symptoms in mice.17 Previous studies have examined the presence of viruses and cytokines only in effusions from patients with OME, so it could not be determined whether they were produced by the local mucosa or by exogenous inflammatory cells. In the present study, using reverse transcriptase–polymerase chain reaction (RT-PCR), we tested for the presence of messenger RNA (mRNA) from IL-8, IL-1β, and RANTES (regulated upon activation, normal T cell expressed and secreted), and for viral RNA from HRSV and HPIV-3 in biopsied mucosal cells and effusions from a series of patients.
Patients requiring unilateral or bilateral placement of pressure-equalizing tubes for OME were recruited for the study between October 1997 and April 1998. The study was approved by the ethics committee of the Leon Berard Hospital, Lyons, France. The individual characteristics of the 26 children enrolled (16 male, 10 female; age range, 2-11 years; mean, 3.2 years) are presented in Table 1. After myringotomy and before tube placement, 50 to 200 µL of effusion was collected in a sterile Eppendorf tube with 700 µL of mucolytic solution (10× stock solution diluted 1:5 in phosphate-buffered saline, 0.1 mol/L; pH 7.4) (Digester; Eurobio, Les Ulis, France). A small mucosal biopsy sample was taken through the tympanic membrane opening and snap frozen in a sterile Eppendorf tube in liquid isopentane/dry ice at −80°C. After a 15-minute digestion, the effusion sample was centrifuged at 10,000g and decanted, and the supernatant and cellular debris were frozen separately.
Frozen samples were placed in guanidinium isothiocyanate lysis buffer (350 µL of buffer for 1 hour for biopsy samples; 350 µL of buffer for 30 minutes for effusion cells; and 700 µL of buffer for 15 minutes for effusion supernatants), then mixed (using a Vortex), pumped 5 times through a 20-gauge needle, and applied to a silica gel, mini spin column (RNeasy; Qiagen, Les Ulis, France) following the manufacturer's instructions. Ribonucleic acid was eluted in 30 µL of sterile diethylpyrocarbonate-treated water and stored at −80°C until needed.
Total RNA from each sample was reverse transcribed by adding 100 U of Moloney murine leukemia virus reverse transcriptase (Promega, Charbonnières, France) to a solution of 5-mmol/L magnesium chloride, 1-mmol/L dideoxynucleotide triphosphates (dNTPs), 100 ng of random primer, and 40 U of RNAsin (Promega) in reverse transcriptase buffer (total volume, 20 µL). After sequential incubation at 20°C (15 minutes), 42°C (15 minutes), 99°C (5 minutes), and 4°C (5 minutes), the complementary DNAs (cDNAs) were amplified with cytokine-specific or virus-specific primers. A mixture containing 5 µL of cDNA, 2 U of Taq polymerase diluted in 5 µL of 10× Taq buffer (Eurobio), 20-µmol/L dNTPs, and 20 pmol of specific primers with appropriate magnesium chloride concentrations as listed in Table 2 (total volume, 50 µL) was submitted to 35 cycles of amplification in a Perkin-Elmer (Les Ulis, France) thermocycler. After initial denaturation at 95°C for 10 minutes, samples were iteratively denatured at 95°C for 1 minute, hybridized at the temperature listed in Table 2 for 1 minute, elongated at 72°C for 1 minute, then submitted to a final elongation at 72°C for 10 minutes. Viral cDNAs were further amplified by a second series of 35 cycles using the inner nested primers18 listed in Table 2.
Polymerase chain reaction (PCR) products were evaluated using 1.5% agarose gel electrophoresis in the presence of ethidium bromide and then were examined under UV light. Molecular weight markers were made from Hae III and Hinf digests of ϕX174 DNA (Promega). Positive controls were provided by extracts from cultured cells expressing the relevant virus or IL-8 or commercial standards for IL-1β and RANTES (R&D Systems, Abingdon, England). Negative controls included template-free distilled water instead of sample RNA in the retrotranscription step and template-free distilled water instead of sample cDNA for PCR amplifications. Conservation of detectable cellular mRNA in the samples was confirmed by amplification of cDNA from a spliced transcript of the glyceraldehyde-3′ phosphate dehydrogenase housekeeping (G3PDH) gene.
Some clinical and demographic characteristics of the 26 patients are summarized in Table 1. Surgery was performed on 3 patients (11, 24, and 25) unilaterally, and biopsy samples were obtained from only 1 of 2 ears that were operated on in patients 1 and 3. In some cases, there was not enough effusion to collect an adequate sample. Other samples were unsuitable for RT-PCR amplification because of RNA degradation, as shown by the absence of a positive signal using the G3PDH primers. Overall, 49 ears were investigated from the 26 patients and furnished a total of 118 analyzable cDNA samples: 47 from mucosal biopsy samples, 36 from exudate cell samples, and 35 from exudate liquid samples.
Polymerase chain reaction amplification of the cDNAs from the different samples using viral primers regularly produced multiple nonspecific bands on agarose gel electrophoresis. Positive reactions were scored only when a major band migrated at precisely the same position as the control positive viral cDNA sample in the same gel. The interfering bands might be eliminated by optimization of PCR conditions for each individual sample, but the amounts available were too small to attempt this. Viral RNA was detectable in at least one sample from most children who underwent surgery for otitis media in this series. Human respiratory syncytial virus was present in 19 children (73%) and HPIV-3 in 5 (19%); both viruses were present in patients 13 and 16 (Table 3 and Figure 1). No viral RNA could be detected in any sample from patients 6, 7, 11, or 21, but for patients 11 and 21, only biopsy samples were available for analysis, and patients 6 and 7 only provided suitable exudate samples from one ear. Human respiratory syncytial virus RNA was detected in biopsy samples from 4 (15%) of the 26 patients (4 [9%] of 47 ears); HPIV-3 was never seen in the mucosal biopsy samples. Viral RNA, usually HRSV, was more often present in exudate cells (10 [45%] of 22 patients; 10 [28%] of 36 ears) and especially in the effusion liquid (16 [76%] of 21 patients; 21 [60%] of 35 ears) (Table 4). Combining the figures for both HRSV and HPIV-3 in effusion liquid and cells, 19 (86%) of 22 patients showed evidence for the presence of a virus in the inflammatory exudate of at least one ear.
Expression of the proinflammatory cytokine IL-8 was investigated similarly using RT-PCR with primers spanning an exon-intron junction to discriminate between amplification of cDNA from the mRNA and that from genomic DNA sequences (Figure 2). Only bands of the molecular size corresponding to the spliced mRNA were scored as positive. Detectable IL-8 mRNA was never found in biopsy cell samples, but 7 (32%) of 22 patients with analyzable exudate from at least one ear showed significant IL-8 expression. In 2 patients (2 and 17), amplification was achieved from the effusion liquid as well as from cells, probably representing incomplete separation or leakage. Similar amplifications from primers specific for RANTES or IL-1β never produced a positive signal from either biopsy or exudate samples from any patient, despite satisfactory positive controls (not shown).
The etiology of OME is certainly complex and appears to involve infectious, inflammatory, and genetic components. An association with viral infection has been suggested, with 30% to 65% of exudate samples testing positive for HRSV by a PCR-based assay.11,12 In the present study, we found traces of HRSV in 58% (41/71) of effusion samples obtained at surgery. However, effusion samples may contain virus derived from airborne contamination (all children were hospitalized before the surgical procedure was performed) or reflux from the rhinopharynx by way of the eustachian tube. To determine the local involvement for viral infection, we also tested biopsy samples of middle ear mucosa obtained during surgery for tympanic tube insertion by specific nested RT-PCR. Human respiratory syncytial virus but not HPIV-3 RNA was present in mucosal cells from 4 (15%) of the 26 patients, showing that these cells can harbor and also may replicate the virus. Mucosal persistence of HRSV RNA and protein in the bronchi correlates with development of chronic airway inflammation.19,20 If local HRSV infection is indeed causative of OME, the low positivity rate in mucosal biopsy samples might be because of an irregular expression cycle, perhaps corresponding to the fluctuating course of the disease. Alternatively, HRSV might be only one of several possible causative viruses, and testing for a wider range of agents would be necessary. However, the small amount of material obtainable at biopsy limits the number of tests possible, and since the chinchilla animal model is not universally accepted as representative of human OME, an experimental approach is not available.
A frequent consequence of active or persistent viral infection is the maintenance of a recurrent and fluctuating inflammatory response.19 Otitis media with effusion has a fluctuating course with alternating exudative and dry periods and eventually long-term regression, particularly after tympanostomy tube placement. In a number of our patients, we found that mRNA coding for the proinflammatory cytokine IL-8 was actively produced by cells in the effusion but not by those in the mucosa. Interleukin 8 protein has been found in OME effusions16 and might be thought to be a causative agent. Our present findings suggest that it is more likely produced by the exudate cells as a consequence of inflammation. Indeed, more IL-8 is present in acute than chronic otitis media, and it is correlated with the presence of bacteria.21 We found no IL-8 produced by mucosal biopsy cells, although cultures of bronchial8,10 or nasal9 mucosae can produce IL-8 in viral infections. Again, none of the samples contained mRNA for RANTES or IL-1β, although these agents have recently been described in inflamed mucosae.22,23 This may indicate that the cause of OME is in some way distinct. Indeed, the majority of cells in the effusion are monocytes or T and B lymphocytes specifically targeted by stromal cell–derived factors 1α and 1β and by macrophage inflammatory protein 1α.24 It would be interesting to determine the production of these factors and to follow the sequence of viral and cytokine gene expression longitudinally during the course of the evolution of OME, if such a sequential study can be ethically conducted.
We found that HRSV or HPIV-3 was present in at least one sample of tissue or effusion from a large proportion of patients who underwent surgery for OME. The expression was not always uniform between the 2 affected ears, and the sometimes-associated expression of IL-8 may also be discordant. Ribonucleic acid viruses of this type are usually thought to be efficiently cleared on resolution of infection of an unimpaired individual, so the frequency of isolation implies an unexpected persistence. Alternatively, such viruses may be present more often than usually believed in the aural cavities of children of this age. We could not justify taking control samples from healthy ears to confirm the absence of viruses. Because the most frequently positive samples were inflammatory cells, which are derived from the circulation, we would expect blood leukocytes from healthy individuals to show signs of positivity if the virus were generally more persistent than believed. We know of no published data suggesting that this is the case. A recent study25 has shown an important hereditary component in the predisposition of young children to OME. Susceptible subjects might be genetically more likely to allow the establishment of a persistent viral infection, which can under favorable circumstances initiate and maintain an inflammatory response in the middle ear. Relevant antiviral therapy might be attempted as a treatment for OME.
Accepted for publication February 22, 2000.
This manuscript was presented at the 14th annual meeting of the American Society of Pediatric Otolaryngologists, Palm Desert, Calif, April 28, 1999.
The authors thank Timothy Greenland for the English-language revision and fruitful comments on basic virology and Catherine Berthet for her expert clerical assistance.
Corresponding author: Patrick Froehlich, MD, PhD, Département d'ORL et de Chirurgie Cervico-Faciale, Hôpital E. Herriot, 69003 Lyon, France (e-mail: firstname.lastname@example.org).