Characteristics of systemic and topical agents implicated in toxicity of the middle and inner ear. (Review).

Toxic injury to the middle and inner ear can be caused by a wide range of substances, including therapeutic drugs, antiseptics, and inert components of the vehicles (table 1). This article reviews those systemic and topical ototoxic agents that otolaryngologists encounter in everyday clinical practice.

Nonantibiotic ototoxic drugs

Chemotherapeutic drugs. Cisplatinum, vinblastine, and vincristine are probably the most commonly used chemotherapeutic agents. Cisplatinum is quite reliably ototoxic, but individual sensitivity varies. Some patients experience ototoxicity after just a single dose, while others seem to tolerate relatively high doses administered over days or weeks. Clinical trials have shown that the incidence of high-frequency hearing loss caused by the platinum group of chemotherapeutic compounds is approximately 62%. (1) The use of sophisticated, objective measurements such as otoacoustic emissions testing will reveal evidence of threshold changes in as many as 85 or 90% of patients who receive a platinum agent. (1) The high-frequency hearing loss experienced by these patients is usually permanent, but it is not associated with any significant vestibular ototoxicity.

Loop diuretics. Ototoxicity caused by the loop diuretics, primarily furosemide and ethacrynic acid, is usually seen in patients with renal failure and in those who receive relatively high doses. For example, furosemide ototoxicity is usually seen in patients who receive 200 to 500 mg in a matter of several hours. Loop-diuretic-induced ototoxicity is marked by edema of the stria vascularis and changes in ionic potentials in the endolymph and perilymph. In experimental animals, strial edema correlates rather nicely with the amount of threshold shift. Fortunately, the strial edema is a transient condition, and ototoxicity caused by a loop diuretic is generally reversible.

Salicylates. The first known report of the therapeutic value of willow bark dates from the 4th century B.C. Americans currently take an estimated 10,000 to 20,000 tons of salicylates annually. The reported incidence of ototoxicity caused by salicylates is as high as 11 per 1,000 patients. (2) Salicylate ototoxicity also occurs in patients who take relatively high doses. Ototoxicity manifests as a 10- to 20-dB flat sensorineural hearing loss that resolves 24 to 72 hours after the agent is discontinued.

NSAIDs. The nonsteroidal anti-inflammatory drugs (NSAIDs) are similar to the salicylates, but the incidence of ototoxicity with NSAIDs is much lower. However, unlike the salicylates, the NSAIDs have been noted in some credible case reports to cause irreversible sensorineural hearing loss.

Quinine. Quinine ototoxicity is often grouped together with salicylate ototoxicity, but they are quite different in that the former manifests not as a flat hearing loss but rather as a high-frequency sensorineural hearing loss. Moreover, quinine toxicity is associated with some meaningful vestibular toxicity, which is not generally seen in salicylate ototoxicity. Quinine-induced vestibular toxicity can be reversed with prompt recognition and treatment, but long-standing toxicity is frequently irreversible.

Initially, quinine-induced ototoxicity was seen primarily in patients with malaria, which is not now a major problem in the United States. However, quinine is being increasingly prescribed by primary care physicians for the treatment of leg cramps, and there are some reports of significant vestibular dysfunction as a result. Because of quinine's effect on the vestibular system, the military now prohibits pilots from flying their planes within 48 hours of taking the agent.

Systemic antibiotic ototoxicity

The two classes of antibiotics known to cause ototoxicity are the macrolides and the aminoglycosides. Vancomycin is usually included on lists of ototoxic antibiotics, but such a designation is difficult to justify. Very few case reports of purported vancomycin-induced ototoxicity are compelling. One reason for this is that the affected patients had often been taking an aminoglycoside either concomitantly with, or shortly before, they started vancomycin. Moreover, other potential causes of ototoxicity were noted in these patients. Therefore, although vancomycin is widely believed to be ototoxic, the evidence to support that claim is weak.

Macrolides. Erythromycin-induced hearing loss has been well recognized for a long time, but its mechanism is not well understood. Erythromycin ototoxicity is generally reversible; of the 30 or 40 case reports I reviewed, I found only two or three in which hearing loss was permanent.

Reports of ototoxicity with azithromycin and clarithromycin are less common, but the percentage of cases that involved permanent hearing loss is probably higher. For the most part, azithromycin ototoxicity and clarithromycin ototoxicity have been seen in patients who received relatively large intravenous doses for the treatment of fairly serious diseases. Reports of ototoxicity following oral administration of either drug are relatively uncommon--unlike the case with erythromycin, where many of the cases of ototoxicity are related to oral use.

Aminoglycosides. The aminoglycosides can be both cochleotoxic and vestibulotoxic. Development of these drugs began in the 1940s as part of a rigorous and systematic attempt to evaluate microorganisms, particularly fungal organisms, for their potential to create antibiotic substances.

The first aminoglycoside was streptomycin, which was created from the fungus-like bacterium Streptomyces griseus. Streptomycin's s ototoxic potential was recognized quite early. In response, researchers modified the drug's molecule and created dihydrostreptomycin. Unlike its predecessor, dihydrostreptomycin is much more cochleotoxic than it is vestibulotoxic; in fact, it is almost exactly as cochleotoxic as streptomycin is vestibulotoxic. Later, investigators tried to reduce the risk of both toxicities by combining equal amounts of the two drugs, but this only resulted in patients experiencing both toxicities rather than neither.

Systemic aminoglycoside ototoxicity

In 1990, Govaerts et al published a study in which they compared the ototoxic potential of various aminoglycosides in animals. (3) They found that gentamicin, kanamycin, and tobramycin were equally cochleotoxic, and all three were more cochleotoxic than amikacin. In terms of vestibular ototoxicity, they reported that streptomycin was more toxic than gentamicin, which was more toxic than tobramycin. They did not study neomycin, because it is so reliably cochleotoxic that it was never approved in the United States for systemic use.

Incidence. Many authors have reported that the incidence of ototoxicity among patients who take a systemic aminoglycoside is only 2 to 3%. However, Govaerts et al found much higher rates. (3) They reported that the incidence of cochlear and vestibular toxicity with gentamicin was approximately 8 and 14%, respectively. The corresponding rates with amikacin were 5 and 13%, and the rates with tobramycin were 14 and 3%, respectively. It is interesting that the incidence of vestibular toxicity with amikacin was nearly as high as that of gentamicin.

The incidence is even higher when objective testing is used to detect ototoxicity. For example, over the course of any antibiotic regimen, the incidence of threshold shift is approximately 50% when ultrahigh-frequency audiometry is performed. (4) Most of this toxicity manifests in the range of 9 to 15 kHz, a range that is not evaluated during routine clinical testing. In addition, studies of rotational-chair testing in asymptomatic patients on aminoglycoside therapy have found that as many as 39% of patients experience statistically significant changes in gain. (5) Findings on objective testing correlate reasonably well with animal pathology.

Cochlear pathology. The pathology of aminoglycoside ototoxicity involves injury to the organ of Corti. Such an injury is always worse in the basal turn of the cochlea than at any other site. In experimental animals, the starting point of degeneration is very clear-cut: 6 to 8mm from the round window in the basal turn. The damage then spreads rapidly to the round window and more slowly toward the apex of the cochlea. The degeneration proceeds progressively and involves higher and higher frequencies.

The outer hair cells are much more susceptible to injury than are the inner hair cells. (6) Among the outer hair cells, those in the inner row are the most susceptible; those in the middle row are less susceptible, and those in the outer row are the least susceptible (figure 1). In more severe cases, damage progresses to the inner hair cells and can even spread to the supporting cells in this order: Deiters' cells, pillar cells, and Claudius' cells. According to Lim, histopathologic findings on light microscopy in aminoglycoside-induced cochleotoxicity are quite similar to those seen in noise-induced hearing loss?

Vestibular pathology. In terms of vestibular pathology, the crista ampullaris of the semicircular canal is more vulnerable than the macula utriculi, which is more vulnerable than the macula sacculi. (6) Type I hair cells are more susceptible than type II hair cells.

Pharmacokinetics. For several decades it was thought that aminoglycosides concentrated in inner-ear fluids--particularly perilymph--to a greater degree than they did in serum. Researchers believed that the serum aminoglycoside concentration peaked within 1 hour of IV administration and then began to decline. Perilymph concentrations, on the other hand, were thought to increase for as long as 12 hours. Similarly, the kinetics of aminoglycoside removal were believed to be just as slow. New studies, however, have posed a compelling challenge to these theories. It now appears that aminoglycosides do not concentrate at all in inner-ear fluids. Aminoglycosides do concentrate in portions of the hair cells and the organ of Corti.

Mechanism of action. The aminoglycosides' mechanism of action at the molecular level remains unknown. Complicated theories surface every 4 or 5 years, but none has been definitively proven. The most current popular theory is that the aminoglycoside itself is not toxic; rather it is a metabolite of the aminoglycoside that causes ototoxicity. (8) Some evidence has been gathered that the formation of an iron-aminoglycoside complex generates free radicals.

Risk factors. The most important risk factors for aminoglycoside ototoxicity are renal insufficiency, the concomitant use of other drugs (especially other aminoglycosides and loop diuretics and, less frequently, the chemotherapeutic agents), and the patient's age. (3)

Researchers have also discovered that a subpopulation of patients has an alteration in the 12S subunit of their mitochondrial DNA (an adenine-to-guanine shift at position 1555) that makes them susceptible to aminoglycoside ototoxicity. Fischel-Ghodsian et al reported that this mutation was found in seven of 41 patients (17.1%) who had aminoglycoside ototoxicity, compared with none of 400 controls who did not. (9) Patients with this mutation can experience significant hearing loss after only one or two doses.

Family history can be key to identifying these at-risk patients. In the study by Fischel-Ghodsian et al, several patients reported that two or three family members had previously experienced significant aminoglycoside ototoxicity. (9) However, this particular mutation certainly does not account for all cases of ototoxicity; other genetic defects might also be responsible. In fact, the degree of individual variability and susceptibility strongly argues that multiple genetic features control the emergence of aminoglycoside ototoxicity.

Dosage. Numerous studies have shown that there is no "safe" dose--that is, no amount of medication and no number of doses that ensure that ototoxicity will not occur. Again, ototoxicity can occur after just a single dose. Therapeutic peak and trough levels do not guarantee that ototoxicity will not occur. Halmagyi et al reported a series of 36 patients with ototoxicity and found that fewer than half had elevated peak levels, fewer than half had elevated trough levels, and two of the 36 had elevations in both peak and trough levels. (10) Most of the patients they evaluated had never exceeded the "safe" dosage of 5 mg/kg/day, and almost none had exceeded the total dose of 50 mg/kg, also considered "safe." It appears that patients who have both received a high total dose and undergone prolonged treatment are at higher risk for ototoxicity, but either one of these factors alone does not seem to increase the risk.

Symptoms. Hearing loss and tinnitus are the symptoms of cochlear injury. The principal hallmark of vestibular injury is gait ataxia that becomes worse when the patient is in darkness or is walking on a soft or uneven surface. These symptoms are not manifest in a patient who is very ill or bedridden. Dizziness or vertigo is rare in vestibular ototoxicity; in the series by Halmagyi et al, none of the 36 patients complained of rotational vertigo of any type. (10)

Signs. The primary sign of cochlear injury is a decrease in hearing threshold on audiometry, stapedius reflex testing, or otoacoustic emissions testing. Otoacoustic emissions testing can be a reasonable method of monitoring these patients during the course of therapy. Signs of vestibular dysfunction include a positive Romberg' s test, a decrease in visual acuity on head shaking, the emergence of compensatory saccades on passive head shaking, and vertical oscillopsia in darkness or on an uneven surface. These signs can be detected at the bedside.

Monitoring. Rigorous maintenance of peak and trough levels is no guarantee that a patient will not experience ototoxicity. Unfortunately, the threat of medicolegal consequences compels the physician to maintain them anyway. We are forced to practice defensive medicine even though it is not necessarily going to help the patient. I have studied a number of cases as a potential expert witness. If a lawyer can find that measurement of even one peak or trough level was made, say, 42 minutes after the last dose of aminoglycoside rather than 30 minutes, he can argue that the physician violated hospital policy and is, therefore, in legal jeopardy. This is a perplexing problem.

On the other hand, we do know that an elevated serum creatinine level is correlated with an increased risk of ototoxicity. Therefore, it is in the best interest of both patients and physicians to closely monitor renal function. Alarm bells should go off if the serum creatinine level begins to rise.

Physicians might consider other types of objective monitoring for cochlear and vestibular function. For cochlear function, physicians should take seriously complaints of tinnitus from patients who are taking an aminoglycoside, and test them. From a strictly technical point of view, the physician would ideally obtain an audiogram, preferably a high-frequency audiogram. Of course, this is not a realistic option in a hospital setting. Another option is to obtain bedside measurements of the stapedius reflex and otoacoustic emissions. Hand-held otoacoustic emissions testing devices are available. As a practical matter, however, I am not aware of any physician who actually does this, and I am not necessarily recommending it.

Several methods are effective in objectively monitoring vestibular function. Rotational-chair testing is the most sensitive and reproducible means of evaluating bilateral vestibular hypofunction. Caloric testing can also be helpful, although dysfunction does not become apparent until there is a 20 to 30% change in saccadic velocity. Vestibular autorotation testing can be useful, although it is expensive and difficult to perform. Again, none of these tests is practical for evaluating a very ill patient. Other possibilities for patients who are well enough to undergo bedside testing include the head-thrust maneuver, evaluation for post-head-shake nystagmus, or testing to detect dynamic changes in visual acuity. Another possibility is a Romberg's test.

Management. Before initiating systemic aminoglycoside therapy, it is imperative to obtain informed consent. The physician must make the patient completely aware of the risks and document having done so. The absence of informed consent is a frequent point of contention in medicolegal cases. Also, discontinue aminoglycoside therapy as soon as ototoxicity is identified or suspected-- assuming that this is possible without endangering the patient's health in some other way. Steroid therapy might be useful, as will be discussed later. Finally, vestibular rehabilitation is very helpful for patients who have experienced significant vestibular injury.

Middle ear ototoxicity from topical agents

Topical aminoglycoside-induced toxicity in the middle ear is not often discussed. Parker and James were the first to address this issue in 1978. (11) They simply evaluated a large number of medications and other substances in an animal model and determined whether they caused mild (slight mucoid secretion or redness), moderate (middle ear filled with mucus), or severe (middle ear filled with pus, mucus, and new bone) toxicity or no toxicity at all.

In this study, several topical medications--carbenicillin, colistin, nystatin, and penicillin G--showed no evidence of toxicity or irritation, but these are agents that otolaryngologists do not use very often. Distilled water and 0.9% sodium chloride also caused no middle ear toxicity. Only moderate middle ear toxicity was seen with topical gentamicin, which is in contrast to the severity of inner ear ototoxicity that can be seen. Moderate toxicity was also seen with bacitracin, chlorhexidine, and tetracycline. Severe middle ear toxicity was seen with topical amphotericin B, benzalkonium chloride, and chloramphenicol and with two common vehicle preparations, polyethylene glycol and propylene glycol. Parker and James concluded that because the middle ear mucosa is so susceptible to irritation and subsequent hyperplasia, eardrops that contain polyethylene glycol and propylene glycol should be avoided because they can damage the cilia. (11)

At my institution--the University of Texas Southwestern Medical Center in Dallas--we looked at a number of these medications in our laboratory animals. (12) One of our findings was that although a single dose of the combination agent sulfacetamide/prednisolone was not toxic to the inner ear, it was fairly irritating to the middle ear (figure 2). One week after administration, this agent had caused a marked inflammation of the mucosa in addition to occasional hemorrhaging and serous effusion. We also saw hemorrhage and hyperplasia of both the lateral and medial surfaces of the tympanic membrane. Four weeks after administration, these changes had substantially resolved.

Another combination agent that was quite harmful to the middle ear was polymyxin/neomycin/hydrocortisone (PNH), possibly because it is delivered in propylene glycol. Administration of this agent led to the formation of mild to severe granulation tissue, and it caused effusion, focal hemorrhage, and osteoneogenesis.

Inner ear ototoxicity from topical agents

Various vehicles, antiseptics, antifungals, and antibiotics have been studied in an attempt to determine if they are toxic to the inner ear when administered topically:

Vehicles. Polyethylene glycol and propylene glycol have shown some evidence of inner ear toxicity, as has benzalkonium chloride.

Antiseptics. Acetic acid, alcohol, chlorhexidine, Cresylate, and gentian violet all have ototoxic potential. Chlorhexidine is quite ototoxic (as Dr. Rutka addresses in more detail in his article beginning on page 9). Gentian violet is so ototoxic that in one study, researchers who had intended to dose 10 animals with it eventually dosed only four, because three of the animals became so ill from vestibular toxicity. Acetic acid's ototoxic potential is worrisome. (13) Alcohol is not nearly as ototoxic as the other antiseptics. The only antiseptic that has been shown not to be ototoxic is povidone iodine. (14)

Antifungals. Antifungals are almost uniformly nontoxic to the inner ear. Clotrimazole, miconazole, nystatin, and tolnaftate have all been investigated for cochlear or vestibular toxicity, and none has been found. A number of published studies and several experts in infectious disease have found that clotrimazole is probably the most effective in eradicating Candida and Aspergillus species. Tom reported that nystatin was not toxic to the inner ear, but it left a thick residue in the middle ear space that might lead to toxicity of the middle ear. (14)

Antibiotics. Among the topical preparations that are severely toxic to the inner ear are polymyxin B, chloramphenicol, ticarcillin/clavulanate, and all of the aminoglycoside-containing agents. At my institution, we found that polymyxin B was actually more toxic than even neomycin.

A single 0.5-ml dose of PNH into the ear of an experimental animal will destroy the organ of Corti. (12) Approximately 3 weeks after administration, it will cause a loss of myelinated nerve fibers in the lamina spiralis ossea; after 3 months, only about one-third of these fibers remain. PNH also causes a loss of hair cells in the crista ampullaris, the macula utriculi, and the macula sacculi, and a marked degeneration of the stria vascularis.

Antibiotics that are not ototoxic include the fluoroquinolones (ciprofloxacin and ofloxacin), ceftazidime, and sulfacetamide. Prior to the quinolone era, our laboratory studied ceftazidime with the idea of formulating it as an otic drop. However, ceftazidime has since been surpassed in bacteriologic efficiency by the quinolones.

Prevention of ototoxicity

A number of agents have been shown to protect against ototoxicity (table 2). Combining fosfomycin with an aminoglycoside has been repeatedly shown to protect from ototoxicity, although the reason it does so is unclear. Also, a small amount of fosfomycin has been shown to completely protect guinea pigs from the ototoxicity associated with PNH. Calcium is another antagonist to aminoglycoside ototoxicity when it is delivered rapidly at the appropriate time. Alpha lipoic acid, antioxidants, and iron chelators have also been shown to be protective in animal models, and this is one basis for the current theory that an iron-aminoglycoside molecule might be responsible for ototoxicity by creating free radicals. (15-17)

Table 1

Agents that have been shown to be ototoxic


Topical antibiotics
  All aminoglycosides (amikacin, gentamicin, kanamycin,
   streptomycin, and tobramycin)
Chloramphenicol
Polymyxin B

Topical combinations
  Polymyxin/neomycin/hydrocortisone
  Ticarcillin/clavulanate

Systemic antibiotics
  All aminoglycosides
  Amphotericin B
  Bacitracin
  Chloramphenicol
  Macrolides (azithromycin, clarithromycin, erythromycin)
  Vancomycin?
  Tetracycline

Antiseptics
  Acetic acid
  Alcohol
  Chlorhexidine
  Cresylate
  Gentian violet

Vehicles
  Benzalkonium chloride
  Polyethylene glycol
  Propylene glycol

Salicylates

Nonsteroidal anti-inflammatory drugs

Quinine

Chemotherapeutic drugs
  Carboplatinum
  Cisplatinum
  Vinblastine
  Vincristine

Loop diuretics
  Ethacrynic acid
  Furosemide

Table 2

Agents that protect against ototoxicity


Alpha lipoic acid
Antioxidants
Calcium
Fosfomycin
Glutathione
Iron chelators

References

(1.) Schweitzer VG. Ototoxicity of chemotherapeutic agents. Otolaryngol Clin North Am 1993;26:759-89.

(2.) Jung TT, Rhee CK, Lee CS, et al. Ototoxicity of salicylate, nonsteroidal antiinflammatory drugs, and quinine. Otolaryngol Clin North Am 1993;26:791-810.

(3.) Govaerts PJ, Claes J, van de Heyning PH, et al. Aminoglycoside-induced ototoxicity. Toxicol Lett 1990;52:227-51.

(4.) Fausti SA, Henry JA, Schaffer HI, et al. High-frequency audio-metric monitoring for early detection of aminoglycoside ototoxicity. J Infect Dis 1992;165:1026-32.

(5.) Black FO, Gianna-Poulin C, Pesznecker SC. Recovery from vestibular ototoxicity. Otol Neurotol 2001;22:662-71.

(6.) Nakashima T, Teranishi M, Hibi T, et al. Vestibular and cochlear toxicity of aminoglycosides--a review. Acta Otolaryngol 2000;120:904-11.

(7.) Lim DJ. Effects of noise and ototoxic drugs at the cellular level in the cochlea: A review. Am J Otolaryngol 1986;7:73-99.

(8.) Minor LB. Gentamicin-induced bilateral vestibular hypofunction. JAMA 1998;279:541-4.

(9.) Fischel-Ghodsian N, Prezant TR, Chaltraw WE, et al. Mitochondrial gene mutation is a significant predisposing factor in aminoglycoside ototoxicity. Am J Otolaryngol 1997;18:173-8.

(10.) Halmagyi GM, Fattore CM, Curthoys IS, Wade S. Gentamicin vestibulotoxicity. Otolaryngol Head Neck Surg 1994;111:371-4.

(11.) Parker FL, James GW. The effect of various topical antibiotic and antibacterial agents on the middle and inner ear of the guinea pig. J Pharm Pharmacol 1978;30:236-9.

(12.) Brown OE, Wright CG, Masaki M, Meyerhoff WL. Ototoxicity of Vasocidin drops applied to the chinchilla middle ear. Arch Otolaryngol Head Neck Surg 1988;114:56-9.

(13.) Aursnes J. Vestibular damage from chlorhexidine in guinea pigs. Acta Otolaryngol 1981;92:89-100.

(14.) Tom LW. Ototoxicity of common topical antimycotic preparations. Laryngoscope 2000;l10:509-16.

(15.) Schacht J. Antioxidant therapy attenuates aminoglycoside-induced hearing loss. Ann N Y Acad Sci 1999;884:125-30.

(16.) Sinswat P, Wu WJ, Sha SH, Schacht J. Protection from ototoxicity of intraperitoneal gentamicin in guinea pig. Kidney Int 2000;58:2525-32.

(17.) Conlon BJ, Smith DW. Topical aminoglycoside ototoxicity: Attempting to protect the cochlea. Acta Otolaryngol 2000;120:596-9.