Ototoxicity of Topical Azithromycin Solutions in the Guinea Pig FREE

Acute otitis media is one of the most common otologic disorders requiring outpatient antibiotic therapy. Infections in acute otitis media have been associated with the formation of bacterial biofilms, which make such infections difficult to treat.¹ Research continues into the development of new antibiotic formulations for otologic use that can effectively treat ear infections and stem the rise in bacterial resistance. There is long-standing interest in the development of formulations that would be simpler to administer and would limit systemic exposure to antibiotics.

Long-term, systemic administration of macrolide antibiotics has been an effective alternative to the more commonly used antibiotics in the treatment of otitis media,^2- 4 with few ototoxic effects in otherwise healthy patients. Macrolide antibiotic treatment is also effective against mycoplasma, mycobacteria, and other nonbacterial agents.⁵ Macrolide antibiotics can penetrate deeply into tissues and have been shown to interfere with quorum sensing of bacterial biofilms similar to those found in some types of otitis media.⁶ Azithromycin (AZ) has been shown to decrease Haemophilus influenzae biofilms in vitro.⁷ Therefore, ototopical formulations of AZ may potentially be used to treat external otitis or otitis media.

To our knowledge, the safety of ototopical formulations of AZ has not been systematically studied. Because AZ has been shown to be capable of rapid penetration into tissues, ototopical administration could increase inner ear concentration relative to what could be achieved by systemic dosing, thereby increasing the potential for ototoxicity. The focus of the present study was to determine whether a topical application of a therapeutic dose of AZ is likely to be safe to middle and inner ear tissues.

At The University of Texas Southwestern Medical Center at Dallas (UT Southwestern), 23 pigmented male guinea pigs, each weighing 330 to 540 g, were given different concentrations of AZ in a proprietary solution ototopically to determine the solution's effect on the middle and inner ear. To reduce variability in drug penetration, all animals were hydrated with 1 mL of saline, subcutaneously, on the day of surgery. For drug application, the animals were anesthetized with isoflurane inhalation; a small opening was made in the left tympanic bulla in an area ventrocaudal to the pinna, and 250 μL of test solution was carefully and slowly instilled into the left middle ear through the opening. The right ear served as the untreated control. An AZ solution of 3% (n = 3), 2% (n = 5), 1% (n = 5), or 0.5% (n = 5) or saline (n = 5) was given in the left ear. The skin was then closed, and the animals were allowed to recover from the anesthesia. The animals were monitored daily for signs of distress, vestibular upset, or wound infection. Two weeks after surgery, the animals were humanely killed and both temporal bones were harvested. All animal procedures were approved by the UT Southwestern Institutional Animal Care and Use Committee.

The temporal bones were divided into medial and lateral portions and photographed using a surgical microscope and digital camera. The inner ear of each temporal bone was then perfused with 2.5% glutaraldehyde in phosphate-buffered saline and fixed by immersion in glutaraldehyde for microdissection. Gross, unstained images of the middle ears were analyzed for the presence of hemorrhage, hyperemia, mucosal tissue adhesions, and mucosal thickening. The cochlea was stained with 1% osmium tetroxide in 0.1M sodium phosphate buffer, pH 7.4, and microdissected. The stria vascularis and organ of Corti were mounted on microscope slides in glycerol and studied as whole-mount preparations by light microscopy. In addition, selected segments of the organ of Corti were further processed by rinsing in phosphate buffer, dehydrating in a graded series of ethanol, critical point drying, and coating in gold for observation (Philips FEI XL30 environmental scanning electron microscope; FEI Company, Hillsboro, Oregon). The anatomy of the middle ear and organ of Corti was analyzed, scored, and statistically quantitated using a t test and linear regression.

To determine the extent of damage, micrographs of the middle ear were analyzed and quantitated by 2 examiners (K.S.P. and C.G.W.) in a blinded manner. Each examiner used a table with values representing the specific degree of damage observed (Table 1). Scores were reported for 4 indices: mucosal thickening, hyperemia, hematomas, and mucosal adhesions. Because adhesions tend to span the middle ear space, only 1 score for adhesions was given per ear. To account for variations in response in different areas of the middle ear mucosa, each ear was divided into 4 quadrants (dorsal lateral, dorsal medial, ventral lateral, and ventral medial), and each quadrant was given a score for the other 3 indices. The mean (SD) scores for mucosal thickening, hyperemia, and hematomas were added to the score for mucosal adhesions to give each ear a cumulative score for middle ear changes. Mean (SD) cumulative scores were compared between groups to determine if there was a statistically significant difference between the conditions of the middle ears in AZ-treated ears vs controls. In addition, the extent of change with increasing azithromycin concentration was analyzed using linear regression to determine if the damage correlated with dose.

The indices used in this study have been used previously^8- 10 to describe middle ear changes, but, to our knowledge, this is the first time they have been defined for use in a quantitative fashion. To determine the usefulness of the scoring indices used in this study, each index was analyzed independently to determine whether the individual scores reflect the overall outcome.

Initial qualitative examination of the inner ear damage, performed by 2 examiners (K.S.P. and C.G.W.), suggested that there was a relationship between the extent of damage and the concentration of AZ given. Therefore, a means was devised to estimate the damage quantitatively for inner hair cells (IHCs) and outer hair cells (OHCs) from base to apex. The severity of damage to the organs of Corti was quantified in all the AZ- and saline-treated ears, plus 5 randomly selected untreated controls. For quantification of the inner ear damage, the length of the organ of Corti was measured using a camera lucida, then the organ of Corti was separated into 10 segments of equal length from base to apex, and the tissue was analyzed in a blinded manner by 1 examiner (K.S.P.) for severity of damage of IHCs and OHCs using a scoring system with values representing specific degrees of damage (Table 2). All 3 rows of OHCs were scored together, with a score of 1 indicating similar to normal (≥ 90% of the hair cells present, with a near-normal appearance) and a score of 5 indicating severely damaged (no hair cells remaining in most [≥ 75%] of the segment). IHCs were scored separately from the OHCs, with a score of 1 indicating similar to normal (≥ 90% normal hair cell population) and 5 indicating severely damaged (only a very small portion [< 25%] of the hair cells appearing normal).

Statistical analysis was performed using a t test on the cumulative scores for middle ear damage and on the scores for IHC and OHC damage, summed base to apex. Linear regression was used to determine the correlation between the degree of damage in the middle ear relative to the AZ concentration plus the degree of damage in the inner ear (sum of IHC and OHC damage, base to apex) relative to AZ concentration.

All but 1 animal survived the duration of the study. That animal had an unrecognized preexisting gastrointestinal condition and was euthanized the day after surgery. All other animals (n = 22) recovered well from surgical anesthesia. Postsurgical monitoring of all animals revealed periodic circling and mild nystagmus the day after surgery in 2 of 3 animals receiving the 3% AZ solution. The symptoms were only evident when the animals were startled, and they resolved by the next day. None of the other animals (n = 20) exhibited signs of vestibular upset at any time.

Pathological analysis performed 2 weeks after AZ application revealed extensive middle ear changes for all formulations tested (Table 3). Changes were seen in the ears treated with AZ compared with control ears (Figure 1A). The changes seen included the development of thick, mucosal adhesions that occupied large portions of the middle ear space and were often attached to the otic capsule (Figure 1B). Hemorrhagic and hyperemic mucosal tissue was commonly seen, and the mucosa often looked thickened. The tympanic membrane typically showed minimal changes. The most common change seen in the saline-treated ears was hyperemia (Figure 2), but this was also seen in some untreated control ears. Adhesions were never seen in saline-treated or untreated controls. Mucosal thickening and hematomas were also rare in the control ears.

Changes seen were quantitated from photomicrographs of the middle ear. The mean (SD) cumulative scores for middle ear damage were 5.8 (1.4) for untreated control, 5.8 (2.4) for saline, 9.5 (0.0) for 0.5% AZ, 9.3 (1.9) for 1% AZ, 10.0 (1.4) for 2% AZ, and 9.8 (0.8) for 3% AZ. The scores for saline-treated ears were not significantly different from those of untreated controls (P = .77, df = 4). The scores for all AZ doses were significantly different from those of controls (0.05%: P < .001, df = 4; 1%: P = .01, df = 4; 2%: P = .005, df = 3; and 3%: P = .001, df = 2).

Cochlear dissection revealed changes within the organs of Corti after AZ administration. Degree of damage varied along the organ of Corti from the cochlear apex to base and between IHCs and OHCs (Figures 3, 4, 5, 6, and 7). Little nerve fiber loss was detected: it occurred in only 1 ear that was exposed to 3% AZ and was limited to areas of complete IHC and OHC loss. OHC changes ranged from scattered hair cell loss (Figure 3A) to areas where no normal OHCs remain (Figure 4B) to total loss of hair cells and supporting cells with only thin, squamouslike cells covering the basilar membrane. IHC changes ranged from scattered hair cell loss to areas where remnants of hair cells are intermixed with normal cells (Figure 5A) to areas of complete loss of hair cells and supporting cells (Figure 3A). There was a tendency for significant IHC damage to be present in areas of minimal damage to OHCs (Figures 3A and 5A).

The IHC damage seen in this study is uncommon compared with reports of ototoxic changes seen after administration of other antibiotics.^11- 13 After AZ administration, some IHCs were partially or completely extruded from the organ of Corti (Figures 3-5), with only a small portion of the cell remaining tethered at the lateral side of its cuticular plate. Free-floating or dislodged IHCs were not obvious, although the reticular lamina had depressions seen by the scanning electron microscope in positions where the IHCs had been lost (Figure 5B). It is possible that the hair cells completely dislodged from the surface of the reticular lamina and removal of the tectorial membrane allowed loose hair cells to float free.

In the ears in which the organ of Corti was microdissected, the stria vascularis was also microdissected from the cochlea and analyzed by light microscopy for any changes associated with AZ administration. Changes seen in the stria vascularis were inconsistent and similar to changes seen in controls.

Statistically significant inner ear changes were seen for all AZ formulations tested compared with controls. Cumulative IHC and OHC scores were as follows: untreated controls, 20.5 (0.6); saline, 21.4 (1.9); 0.5% AZ, 32.4 (4.8) (P = .005, df = 4); 1% AZ, 32.2 (6.6) (P = .02, df = 4); 2% AZ, 43.0 (6.1) (P < .001, df = 3); and 3% AZ, 53.0 (12.1) (P = .04, df = 2). The scores for saline-treated ears were not significantly different from those for control ears.

To determine whether the changes seen in the middle and inner ears were owing to AZ itself or were caused by other constituents of the AZ formulation, we examined the correlation of the extent of tissue damage with the concentration of AZ present within the formulations. The extent of middle ear damage correlated moderately with AZ concentration (r² = 0.59). Inner ear disease was highly correlated with the concentration of AZ administered (r² = 0.94).

Individual indices of middle ear damage were examined for their contribution to the overall score. Mean (SD) scores for AZ-treated ears were significantly different from those for control ears for all concentrations of AZ for mucosal thickening, hematoma, and mucosal adhesions (Table 3). Although hyperemia was slightly elevated in the 2% AZ and 3% AZ ears, it was not significantly different from the untreated or saline controls.

Administration of AZ solutions to the guinea pig middle ear caused damage to the inner and middle ear. The semiquantitative methods developed yielded results that could be statistically analyzed. The moderate correlation found between AZ concentration and the extent of middle ear damage suggests that some of the damage seen in the middle ear after application was caused by AZ, but other factors appear to be involved, such as the other constituents of the test solutions. Middle ear changes caused by the nonmedicinal ingredients of otic agents have been reported.¹⁰ The strong correlation of AZ concentration and inner ear damage suggests that AZ applied to the guinea pig's middle ear can cross into the inner ear and is the primary agent responsible for the ototoxicity.

The pattern of damage seen in this study, in which IHC damage is worse than OHC damage after antibiotic administration, is an unusual finding. The ototoxicity seen after administration of other antibiotics is typically greater for the OHCs or the effects for IHCs and OHCs are similar.^11- 12 In addition, the observation that dying IHCs tend to be extruded onto the endolymphatic surface has only rarely been reported in the literature on disease of the organ of Corti. Also, in this study, the supporting cells that surround the IHCs were often found to have disappeared in areas of extensive IHC loss. This type of damage is unexpected in view of findings from previous reports on systemically administered AZ,¹³ which suggest the target tissues for AZ ototoxicity are either the stria vascularis or the OHCs. Strial changes were not obvious in our study, although the methods used were not optimal for assessment of strial disease. It is also possible that the route of administration affects the drug target. Further study is needed to better determine the nature of cell death in the organ of Corti and possible morphologic effects on the stria vascularis after ototopical AZ administration.

Because of the development of antibiotic resistance in pathogens, there is a continuing need for discovery of effective, new antibiotics for treatment of infections. Macrolide antibiotics have been shown to be effective for treatment of otitis media. Their ability to interfere with quorum sensing in biofilms suggests that they may be effective in treating the more difficult cases of otitis media where biofilms are present. Although the present study was performed using single applications on the guinea pig model, the results suggest that AZ could be ototoxic in humans if topically applied to the ear. Therefore, AZ may not be appropriate for inclusion in formulations intended for ototopical use. In addition, there have been occasional cases of temporary and permanent hearing losses after systemic AZ treatment.¹⁴ That fact, along with the results of the present study showing ototoxicity of AZ occurring in a concentration-dependent manner, suggest that AZ can be ototoxic if it enters the inner ear in sufficient concentrations. Therefore, caution should be taken when administering AZ systemically.

Macrolides are an accepted and often preferred therapy for chronic rhinosinusitis.¹⁵ Their beneficial effect is a consequence not only of their bacteriologic and bacteriostatic effect but also because they downregulate proinflammatory mediators.¹⁶ They are often used in lower dosages (250 mg/d) for longer periods of time because of the immunomodulation effect. To date, no ototoxic effects have been reported in association with this treatment regimen. However, given that ototoxicity has occasionally occurred in patients receiving systemic AZ and the potential for some antibiotics to accumulate in perilymph over time, it may be advisable to audiometrically monitor patients who are receiving long-term therapy.

Correspondence: Karen S. Pawlowski, PhD, Department of Otolaryngology–Head and Neck Surgery, The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390-9035 (Karen.Pawlowski@utsouthwestern.edu).

Submitted for Publication: April 17, 2009; final revision received December 22, 2009; accepted January 12, 2010.

Author Contributions: Dr Pawlowski had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pawlowski, Si, Wright, Hosseini, and Roland. Acquisition of data: Pawlowski and Koulich. Analysis and interpretation of data: Pawlowski and Koulich. Drafting of the manuscript: Pawlowski. Critical revision of the manuscript for important intellectual content: Pawlowski, Si, Wright, Koulich, Hosseini, and Roland. Obtained funding: Si, Wright, Hosseini, and Roland. Administrative, technical, and material support: Si, Wright, Koulich, and Hosseini. Study supervision: Pawlowski.

Financial Disclosure: Dr Pawlowski has applied for a patent entitled “Micro- and non-patterned surface features to reduce implant fouling and regulate wound healing” (patent application 20090093879). Although Drs Hosseini and Si are employees of the sponsor of the study, InSite Vision Inc, they do not have any financial interest to disclose because InSite Vision Inc has decided not to pursue further development of an ototopically applied preparation of azithromycin, as mentioned in this article. Dr Roland has worked as a consultant or received honoraria from Alcon Laboratories, MED-EL Corporation, Advanced Bionics LLC, and Cochlear Corporation and has worked as a speaker and received honoraria from GlaxoSmithKline and Alcon Laboratories.

Funding/Support: This work was funded in part by InSite Vision Inc.

Previous Presentation: This study was presented at the 32nd Annual Midwinter Meeting of the Association for Research in Otolaryngology; February 18, 2009; Baltimore, Maryland.

Additional Contributions: Kipton Sheek, AAAS, and Paula Timmons, HT, LAT, provided assistance in performing the animal experiments. Christopher Gilpin, PhD, Laurie Mueller, BS, and Tom Januszewski, MS, in The University of Texas Southwestern Molecular and Cellular Imaging Facilities, provided assistance with scanning electron microscopy work.