Vestibular and Cochlear Ototoxicity of Topical Antiseptics Assessed by Evoked Potentials †
Article first published online: 2 JAN 2009
Copyright © 2000 The Triological Society
Volume 110, Issue 9, pages 1522–1527, September 2000
How to Cite
Perez, R., Freeman, S., Sohmer, H. and Sichel, J.-Y. (2000), Vestibular and Cochlear Ototoxicity of Topical Antiseptics Assessed by Evoked Potentials . The Laryngoscope, 110: 1522–1527. doi: 10.1097/00005537-200009000-00021
Supported by the Israel Science Foundation, The Israel Academy of Sciences and Humanities–The Jack Adler Foundation.
- Issue published online: 2 JAN 2009
- Article first published online: 2 JAN 2009
- Manuscript Accepted: 5 MAY 2000
- vestibular ototoxicity;
- vestibular evoked potentials;
- auditory evoked response;
Objectives/Hypothesis To evaluate and compare the effect of chlorhexidine gluconate, povidone-iodine, and alcohol—three antiseptics used before ear surgery—on the function of the vestibular and cochlear parts of the sand rat's inner ear. The assessment of damage is based on the recording of vestibular evoked potentials (VsEPs) and auditory brainstem response (ABR).
Study Design Prospective controlled animal study.
Methods Fat sand rats were randomly assigned to five different groups, each receiving topical application of a different agent: saline (control), gentamicin (ototoxic control), chlorhexidine, povidone-iodine, and alcohol. Right-side total labyrinthectomy was performed, and a polyethylene tube was inserted into the left (contralateral) middle ear. After baseline recordings were taken of VsEPs and ABR, each animal received five consecutive daily applications of the specific agent into the left middle ear. Three days after the fifth application, evoked potential recordings (VsEPs and ABRs) were repeated and compared with baseline measurements.
Results Administration of saline affected neither VsEPs nor ABR. In contrast, as expected, neither of these responses could be recorded after gentamicin application. After application of chlorhexidine all waves disappeared in all sand rats. Alcohol caused the waves to disappear in some of the animals only. Povidone-iodine did not affect VsEP recordings and had only a small effect on ABR.
Conclusions Chlorhexidine and alcohol had a clear toxic effect on the vestibular and cochlear function of the inner ear of the sand rat, whereas povidone-iodine did not. Thus, taking into consideration that this is an animal study, it appears that povidone-iodine might be preferable to the other agents tested in disinfecting ears with a perforated tympanic membrane.
Ototoxicity of topical antiseptics was first suspected three decades ago. The suspicion arose as a result of reported sensorineural hearing loss following simple myringoplasty procedures when antiseptics were used before surgery. 1 During the past 30 years, relatively little study has been directed to the potential ototoxicity of these commonly used agents. 2–9 Most of these investigations referred to damage to the cochlear part of the inner ear, assessing it by pathological cochlear histology and abnormal responses in different electrophysiological tests. Limited study has been directed to vestibular ototoxicity, 4,6 and the evaluation of vestibular damage in these reports was mainly based on description of histological findings, not on objective measurement of vestibular end organ function.
In recent years, a new method for induction and recording of short-latency vestibular evoked potentials (VsEPs) in response to angular 10–12 and linear 13,14 acceleration has been developed in both laboratory animals 10–14 and human subjects. 15,16 It has been shown that the first peak of the VsEPs is the compound action potential of the vestibular nerve fibers synchronously activated by the stimulus 17 and therefore reliably reflects the function of the vestibular end-organ. 18 Thus VsEP recordings are an objective method for directly evaluating vestibular end-organ function, as opposed to other, less direct methods, such as the vestibular ocular reflex, which have been used previously.
Making use of VsEPs, supplemented by the extensively used auditory nerve responses and auditory brainstem responses (ABRs), 19 an animal model for evaluating potential ototoxicity of topical agents has been developed recently in our laboratory. 20 The animal chosen for study was the fat sand rat, because of its unique middle and inner ear anatomy, 21 allowing the performance of the delicate middle and inner ear surgical procedures required for implementing this model. The efficacy of this model was demonstrated by the abolishment of the VsEPs, as well as the ABR, after topical application of gentamicin, an agent with known cochlear and vestibular ototoxicity when given locally 20 and systemically. 22 Similar topical application of saline solution did not affect the recorded potentials.
The purpose of the present study was to evaluate and compare the effect of chlorhexidine gluconate, povidone-iodine, and alcohol—three antiseptic agents used in ear surgery—on the inner ear, with emphasis on potential damage to vestibular function. This study is distinguished by two main features: the use of an objective test (VsEP recordings) for evaluating the functional damage to the vestibular end-organ itself and the fact that this is a controlled comparative study of three disinfectants in regular use, assessing their effects using the same animal model and study protocol.
MATERIALS AND METHODS
The study was conducted on 25 adult fat sand rats (Psammomy obesus) with mean body weight of 179 ± 24 g, who survived all stages of the experiment. This rodent species was chosen for study because of its unique middle and inner ear anatomy consisting of a large bulla cavity, a thin otic capsule, and an inner ear that clearly projects into the middle ear cavity. 21 This anatomy allowed the performance of delicate middle and inner ear procedures, which are described later in this section. In addition, this laboratory has extensive experience in induction and recording of short-latency auditory and vestibular evoked potentials in this rodent. 14
For experimentation, the animals were anesthetized by intraperitoneal injection of pentobarbital 25 mg/kg, and additional doses were given intraperitoneally as needed. While the animal was under anesthesia, during the surgical procedures and recording of evoked potentials, rectal temperature was monitored and maintained at 37°C ± 0.5°C (using heating pads). Initially, right-side total labyrinthectomy was performed; then a polyethylene tube was introduced into the left middle ear cavity 20 (as described in detail later in this section). The experiments were conducted on unilateral labyrinthectomized animals to simplify the interpretation of the VsEPs and to avoid the need for masking the contralateral ear while recording ABR. As a result of destruction of the right-side inner ear, we could be certain that the evoked potential recordings originated only from the tested left ear. After these surgical procedures, baseline recordings of VsEPs in response to linear acceleration and ABR in response to clicks were conducted. After the baseline recordings, the agent under investigation was injected once daily into the left middle ear cavity through the polyethylene tube for 5 consecutive days (0.1 mL per daily injection).
The 25 animals were randomly assigned to five groups consisting of five sand rats each. Two groups were used as controls, the first receiving saline solution and the second receiving gentamicin 40 mg/mL. The saline and gentamicin groups were considered as controls because it has already been shown in this laboratory that saline is without effect and that gentamicin has both vestibular and cochlear ototoxicity. 20 The sand rats in the three remaining groups were injected with one of the following antiseptic agents: 1) povidone-iodine 10% equal to 1% iodine in aqueous solution; 2) chlorhexidine gluconate 0.5% in aqueous solution; 3) ethyl alcohol 70% in water. On the eighth day (3 days after the last injection), the sand rats were anesthetized again and repeat measurements of VsEP and ABR were carried out. Subsequently, a lethal dose of pentobarbital was injected intraperitoneally and 5 minutes after respiratory arrest, postmortem recordings of VsEPs were performed to rule out possible electromagnetic or electromechanically induced artifact in the measurement. A postmortem examination of the middle ear was conducted to assess the effect of the different agents under investigation on the middle ear and confirm that the polyethylene tube was still in place. All experiments were carried out in accordance with the guidelines published by the Hebrew University–Hadassah Medical School Animal Care and Use Committee.
The skin was incised behind the pinna (right ear). The bulla was opened, and the middle ear exposed by drilling the thin bony wall. The incus and malleus were carefully removed. The projecting cochlea (behind the removed ossicles), the three semicircular canals with their ampullae, and the vestibule were drilled out and completely destroyed. The skin was sutured with 3-0 silk thread.
A small incision was made behind the pinna (left ear) and the bone was exposed. A small hole in the bone of the cortex was drilled out between the superior and inferior horizontal septa. After visualizing the round window, a 1.5-cm polyethylene tube was inserted through the small hole in the bone with its distal end in a position opposite to the round window. The tube was fixed to the bone with glue and to the skin with a 3-0 silk suture.
Techniques for Induction and Recording of Evoked Potentials
Vestibular evoked potentials.
A detailed description of the stimulator apparatus has been reported in previous publications. 13,14 In general, the linear acceleration stimulator consisted of a solenoid that repeatedly delivered acceleration impulses to a sliding device restricted to move in one axis. The moving sliding device was attached to the head of the animal by a head holder that firmly gripped the upper jaw in a plane that is optimal for utricle stimulation in rodents (head forward). 23 The magnitude of the acceleration was measured with a Bruel and Kjaer 4393 accelerometer (Naerum, Denmark) mounted on the sliding device. The stimuli produced displacements of approximately 50 μm, with acceleration of 3 g and a short rise time of 1 to 1.5 milliseconds. This intensity is in the mid region of the VsEP input-output function. 13 Acceleration impulses were given at a rate of 2.06 per second.
Auditory evoked potentials.
The response was elicited by alternating polarity click stimuli at a rate of 20.6 per second from an intensity of 120 dB peak equivalent (pe) sound pressure level (SPL) down to threshold in 5-dB steps. If no response could be recorded at 120 dB pe SPL, an intensity of 135 dB pe SPL was used. The earphone was placed 0.5 cm from the left ear, taking care not to obstruct the external meatus by the pinna.
The electrical activity in response to the different stimuli was recorded by needle electrodes (Grass Instruments, Astro-Med, Inc., West Warwick, RI) inserted subdermally into the vertex referred to the left pinna, with the right pinna serving as ground. The activity was band-pass filtered (300–1500 Hz), amplified, and averaged (n = 128) by standard evoked potential equipment (Microshev 4000, Microshev, Efrath, Israel) and displayed "vertex positive up." Each response was obtained three times to ensure reproducibility. Comparison of the peak latency and peak-to-peak amplitude of the first wave of VsEP in response to a 3 g stimulus before and after application of the agents was carried out using a nonparametric paired test (paired Wilcoxon rank test). The latency and amplitude of the first wave of ABR in response to an 80 dB pe SPL click were similarly analyzed. We chose to analyze the ABR in response to 80 dB pe SPL because this intensity is approximately comparable to the 3 g acceleration stimulus for the VsEP, since both are in the mid region of the input-output function curve of the response. The 80 dB pe SPL intensity is approximately 15 to 25 dB greater than the ABR threshold of the normal fat sand rat (55–65 dB pe SPL), allowing the detection of relatively small changes in auditory function. Therefore the latency and the amplitude of the first waves of both responses (VsEP and ABR) reflect the relevant end-organ function.
Baseline Recordings of Vestibular and Auditory Evoked Potentials
After the initial surgery, clear baseline responses to vestibular and auditory stimuli were recorded from all 25 sand rats. The recorded response to vestibular stimuli consisted of five to six waves. The mean peak latency ± SD of the first wave was 1.87 ± 0.22 milliseconds after the onset of acceleration with a mean peak-to-peak amplitude of 1.27 ± 0.36 μV. These baseline parameters are similar to those previously reported. 14 The response to auditory stimuli consisted of four waves with a threshold between 55 and 65 dB pe SPL. The mean latency of the first wave in response to 80 dB pe SPL intensity click was 1.03 ± 0.08 milliseconds with an average amplitude of 1.31 ± 0.61 μV.
Control Groups After Application of Topical Agents
Vestibular evoked potentials could be clearly recorded in all animals 3 days after the final (fifth) day of saline application. Mean latency and amplitude of the first wave were not significantly different from baseline recordings before treatment (Table I). ABR waves were clearly recorded in all animals with thresholds between 55 and 65 dB pe SPL, similar to baseline thresholds. No statistically significant difference was found between peak latency and peak-to-peak amplitude of the first wave in response to 80 dB pe SPL clicks before and after saline application (Table II). After respiratory arrest, all waves disappeared and the middle ear cavity was inspected under the microscope; no evidence of disease was found.
*P < .05 (nonparametric test).
After topical application of gentamicin (40 mg/mL), all VsEP waves disappeared and could not be recorded in any of the animals treated. The traces were similar to postmortem recordings. Also, ABR waves could not be obtained even with maximal intensity clicks of 135 dB pe SPL. Postmortem inspection of middle ear cavity revealed no disease.
Experimental Groups After Application of Topical Agents
Vestibular evoked potential and ABR waves were abolished in all five animals after treatment with chlorhexidine gluconate 0.5%. Responses could not be obtained in any animal even to maximal intensity stimuli, and the traces were similar to postmortem recordings. No disease was detected on inspection of any of the middle ears of these animals. Recordings from a typical sand rat before and after chlorhexidine application and postmortem are shown on the right in Figure 1 (VsEP) and Figure 2 (ABR). Thus the findings with chlorhexidine were similar to the gentamicin group.
Vestibular evoked potentials were clearly recorded in all five animals 3 days after the last (fifth) povidone-iodine (10%) application. No significant statistical difference between latency and amplitude of the first wave before and after application was found (Table I). Auditory evoked potentials were detected in all animals. In four animals thresholds were between 60 and 65 dB pe SPL, similar to baseline recordings. One animal had a threshold of 70 dB pe SPL (15 dB elevation in comparison to its baseline threshold). Postmortem inspection of this individual sand rat showed granulation tissue in the middle ear cavity and external auditory canal. The mean latency of the first wave in these five animals in response to 80 dB pe SPL click was significantly prolonged (P = .01) in recordings after application of the topical agent, whereas no significant difference in amplitude was found (Table II). Typical VsEP and ABR recordings before and after iodine application are shown on the left in Figures 1 and 2, respectively. Except for the one animal with the granulation tissue (described earlier), mild mucosal thickening in the middle ear cavity was seen in only one additional animal.
Neither ABR nor VsEP could be recorded in two animals, 3 days after the final (fifth) application of alcohol (70%). ABR responses were present in only two of the other three animals, both with elevated thresholds (70 and 80 dB pe SPL) and prolonged latency of the first wave (not statistically significant, because of the small number of animals). Of particular note, in one animal, responses to vestibular stimuli were recorded but there was no response to maximal-intensity auditory stimuli. Erythema and mild edema were noted in the mucosa of the middle ear cavities of the five sand rats receiving alcohol. One sand rat from which auditory and vestibular evoked potentials were successfully recorded had abundant granulation tissue in the middle ear cavity, which might have covered the round window and prevented the penetration of alcohol into the inner ear.
The results of this study demonstrate the vestibular and cochlear ototoxicity of chlorhexidine gluconate and alcohol by showing their clear effect on recordings of vestibular and auditory evoked potentials. This effect is comparable to that of topical gentamicin, which was used as a control. In contrast, povidone-iodine solution appeared not to have any effect on vestibular evoked potentials, with a minimal effect, if any, on auditory responses. The recordings after iodine application were comparable to those after application of saline, which was used as a control.
The minimal effect of povidone-iodine consisted of statistically significant (P = .01) prolongation of the latency of the first wave of ABR, without significant change in amplitude or elevation in threshold. This effect is typical of minimal conductive hearing loss and could be caused by middle ear disease such as mucosal thickening or granulation tissue observed in postmortem examination of two sand rats from this group.
Whereas chlorhexidine completely abolished the vestibular and auditory potentials in all treated animals, alcohol had the same effect on only some animals in the group. Based on this result, we might speculate that chlorhexidine has a greater toxic potential than alcohol, even though they are both ototoxic. However, a possible explanation for the different effect of alcohol on individual animals may be the formation of granulation tissue in some of the animals which covered the round window and may have protected the inner ear from toxic effects of the agent. Also of interest, one animal had a normal vestibular response after alcohol application but did not respond to auditory stimuli. This might imply that alcohol has greater cochlear than vestibular ototoxicity.
Relatively little study has been directed to the potential ototoxicity of these agents, and most of the studies refer to their effect on the cochlear part of the inner ear. With respect to vestibular toxicity, our findings correlate with the reports by Aursnes 4,6 of vestibular damage resulting from chlorhexidine exposure and the absence of damage after application of iodine in aqueous solution. While both of these reports are based on description of histological findings, the present study demonstrates the agents' effect on vestibular function using an objective test. We are not aware of any reports of vestibular ototoxicity of topical alcohol, which is described in the present study. With respect to cochlear ototoxicity, previous studies 3,8 have demonstrated it after chlorhexidine application, as does the present study. Regarding alcohol, Morizono and Sikora 2 published a systematic study in the early 1980s on its potential cochlear ototoxicity. The authors described variability among individual animals but concluded that the agent has cochlear ototoxicity. The potential ototoxicity of iodine disinfectants has also previously been studied. 5–7 One of these studies 6 evaluated damage by means of histological inspection and did not find pathological changes in guinea pigs that were exposed to iodine in aqueous solution. Another one of these published reports 5 investigated povidone-iodine solution and scrub and demonstrated that the scrub, which included a detergent, was more ototoxic than the aqueous solution. In contrast to the findings in the present study, it was shown that povidone-iodine, in increasing concentrations and without the scrub, caused changes in threshold of the compound action potentials of the auditory nerve.
One of the benefits of the present study is the comparison of chlorhexidine gluconate, povidone-iodine, and alcohol using the same animal model and study protocol. We chose these three agents because they are frequently used independently and also in conjunction with each other (povidone-iodine in alcohol, chlorhexidine in alcohol). In using the three agents for this study, we were careful to use pure substances that were always in aqueous solution to ensure that effects on the inner ear were due to the active agents, not to different additives.
From a clinical point of view, two issues should be addressed. First, our protocol involved the application of the agents for 5 consecutive days. This does not mimic the clinical situation of the use of disinfectants in ear surgery when the inner ear is subject to exposure to a small amount of the investigated agents on a single occasion. This protocol was used because it is the same protocol that was used in the well-controlled model developed previously in our laboratory. 20 Also, the purpose of this study was to investigate the absolute ototoxicity of these agents. We can assume that, because povidone-iodine was not found to be toxic in five daily applications, it will not damage the inner ear in small amounts of single exposure during surgery. In addition, these findings should be taken into account in regard to topical treatment of ears with tympanic membrane perforations because some of the commonly used eardrops contain alcohol as a solvent. Second, a discrepancy exists between ototoxicity of topically applied agents in animal models and in humans. For example, topical aminoglycosides are shown to be very ototoxic in animals, whereas in humans there is a relatively small number of reports of hearing loss after treatment with eardrops containing these agents. This possible difference between laboratory animals and humans may be explained by the fact that in humans the round window is deeper and thicker and frequently is covered by another mucous membrane, 24 and these factors may be protective.
Taking these two issues into consideration, one should be cautious with regard to applying the findings of this study in a clinical setting. However, in view of these findings it is suggested that povidone-iodine solution might be preferable to chlorhexidine or alcohol for use during ear surgery with a perforated tympanic membrane.
It appears to be clear that both chlorhexidine and alcohol have a toxic effect on the vestibular and cochlear function of the inner ear of the sand rat, as demonstrated by the disappearance of vestibular and auditory evoked potentials after topical application of these agents. In contrast, it appears that povidone-iodine does not have this adverse effect, and the recordings of evoked potentials obtained after topical application of this agent are similar to those obtained after saline application. Thus, taking into consideration the fact that this is an animal study, it would be reasonable to conclude that the use of povidone-iodine solution is safer in disinfecting ears with perforated tympanic membranes than the other agents tested.
- 12.A surface recorded vestibular evoked response to acceleration stimuli in cats. J Laryngol Otol 1984;(Suppl 9):111–119., , .
- 19.Auditory nerve and brainstem responses (ABR): physiological basis and clinical uses. In: DesmedtJE, ed. Neuromonitoring in Surgery. Amsterdam: Elsevier Science Publishers, 1989: 23–47..