Aminoglycoside antibiotics

Eyes

Collagen corneal shields pre-soaked with antibiotics are used as a means of delivering drug to the cornea and anterior chamber of the eye. Gentamicin damages the primate retina, particularly by macular infarction, and amikacin can have a similar effect; a subconjunctival injection of tobramycin causes macular infarction [ ]. This potentially devastating consequence suggests that care must be exercised when contemplating instillation of aminoglycosides directly into the eye. Allergic contact dermatitis causing conjunctivitis and blepharitis has been reported with topical ophthalmic tobramycin [ ].

Ears

Ototoxicity is a major adverse effect of aminoglycoside antibiotics [ ]. They all affect both vestibular and cochlear function, but different members of the family have different relative effects ( Table 1 ).

The use of aminoglycosides in Ménière’s disease and the incidence, mechanism, susceptibility factors, and prevention of ototoxicity have been reviewed, based on published literature from 1966 to 2004 [ ].

In a review of studies published between 1975 and January 2008 there was no correlation of serum aminoglycoside serum concentrations with vestibular toxicity but there was an effect of duration of therapy [ ]. Ototoxicity was evaluated in 64 patients with tuberculosis, of whom 34 received amikacin, 26 kanamycin, and 4 capreomycin, with a mean treatment duration of 20 months; 19% developed irreversible hearing loss and 6.3% had involvement of speech frequencies [ ].

Differences between different aminoglycosides

Streptomycin and gentamicin primarily cause vestibular toxicity, whereas amikacin, neomycin, and kanamycin primarily cause cochlear toxicity. However, in comparisons with equipotent dosages, the ototoxicity of amikacin was of the same order as that caused by gentamicin [ ].

In 40 patients tobramycin had little effect on audiometric thresholds, but produced a change in the amplitude of the distortion products, currently considered an objective method for rapidly evaluating the functional status of the cochlea [ ]. In one case, tobramycin caused bilateral high-frequency vestibular toxicity, which subsequently showed clinical and objective evidence of functional recovery [ ].

In a quantitative assessment of vestibular hair cells and Scarpa’s ganglion cells in 17 temporal bones from 10 individuals with aminoglycoside ototoxicity, streptomycin caused a significant loss of both type I and type II hair cells in all five vestibular sense organs [ ]. The vestibular ototoxic effects of kanamycin appeared to be similar to those of streptomycin, whereas neomycin did not cause loss of vestibular hair cells. There was no significant loss of Scarpa’s ganglion cells.

Incidence

The incidence of ototoxicity due to aminoglycosides varies in different studies, depending on the type of patients treated, the methods used to monitor cochlear and vestibular function, and the aminoglycoside used [ ]. Clinically recognizable hearing loss and vestibular damage occur in about 2–4% of patients, but pure-tone audiometry, particularly at high frequencies, and electronystagmography show hearing loss and/or vestibular damage in up to 26% and 10% respectively, despite careful dosage adjustment [ ]. In patients with Pseudomonas endocarditis receiving prolonged high-dose gentamicin, auditory toxicity was found in 44% [ , ].

A review of nearly 10 000 adults suggested rates of 14% for amikacin, 8.6% for gentamicin, and 2.4% for netilmicin. Aminoglycoside toxicity is markedly lower in infants and children, with an incidence of 0–2%. A long duration of treatment and repeated courses or high cumulative doses appear to be critical for ototoxicity, which occurs in high frequency hearing beyond the range of normal speech.

There is a discrepancy between clinical observations, in which very few patients receiving aminoglycosides actually complain of hearing loss, and the reported incidences of ototoxicity in studies of audiometric thresholds. A major reason for this discrepancy relates to the fact that aminoglycosides cause high-frequency hearing loss well before they affect the speech frequency range in which they can be detected by the patient [ ].

Gentamicin also damages the vestibular apparatus at a rate of 1.4–3.7%, resulting in vertigo and impaired balance. This effect is reversible in only about 50% from 1 week to 6 months after administration.

In a retrospective study in 81 men and 29 women, hearing loss of 15 decibels at two or more frequencies, or at least 20 decibels at at least one frequency, was found in 18% of patients treated with aminoglycosides (amikacin, kanamycin, and/or streptomycin) [ ]. In those treated with kanamycin the rate was 16%. Age, sex, treatment duration, total aminoglycoside dose, and first serum creatinine concentration were not associated with hearing loss.

Dose-relatedness

Hearing loss was attributed to repeated exposure to aminoglycosides in 12 of 70 patients with cystic fibrosis (one child) [ ]. There was a non-linear relation between the number of courses of therapy and the incidence of hearing loss. The severity of loss was not related to the number of courses. Assuming that the risk of hearing loss was independent of each course, the preliminary estimate of the risk was less than 2 per 100 courses.

Presentation

Clinically, cochlear ototoxicity is more frequent and easier to detect than vestibular toxicity; combined defects are relatively rare. Symptoms of cochlear damage include tinnitus, hearing loss, pressure, and sometimes pain in the ear. The manifestations of vestibular toxicity are dizziness, vertigo, ataxia, and nystagmus. These are often overlooked in severely ill, bed-ridden patients.

Symptoms of ototoxicity can occur within 3–5 days of starting treatment, but most patients with severe damage have received prolonged courses of aminoglycosides. In some cases, hearing loss progresses after the administration of the causative drug has been interrupted. The ototoxicity is reversible in only about 50% of patients. Permanent deafness is often seen in patients with delayed onset of symptoms, progressive deterioration after withdrawal of treatment, and hearing loss of over 25 db [ ].

There are interesting differences in the toxicity patterns of aminoglycosides in animals. Gentamicin and tobramycin affect the cochlear and vestibular systems to a similar extent, while amikacin, kanamycin, and neomycin preferentially damage the cochlear and streptomycin the vestibular system. Netilmicin appears to be the least toxic [ , ].

In man, differences in the ototoxic risks of the currently used aminoglycosides are difficult to evaluate [ ]. There have been no prospective comparisons of more than two drugs using the same criteria in similar patient populations. However, several controlled comparisons of two aminoglycosides are available and provide some information. A survey of 24 such trials showed the following mean frequencies of ototoxicity: gentamicin 7.7%, tobramycin 9.7%, amikacin 13.8%, netilmicin 2.3% [ ]. There was also a lower incidence of netilmicin-induced inner ear damage compared with tobramycin in two studies [ , ].

Of 20 patients with Dandy’s syndrome, 15 had previously been treated with aminoglycosides (13 with gentamicin and 2 with streptomycin), of whom 10 had symptoms of pre-existing chronic nephrosis or transitory renal insufficiency. In all 13 patients who had gentamicin, peripheral vestibular function was destroyed or severely damaged, whereas there was no hearing loss [ ].

Mechanism

The mechanism of ototoxicity by aminoglycosides is still not fully clarified. Most of the experimental data have been gained in guinea-pigs, which seem to resemble humans.

Animal studies

Traditionally, toxic damage is considered to be the consequence of drug accumulation in the inner ear fluids [ , ]. After a period of reversible functional impairment, destruction of outer hair cells occurs in the basilar turn of the cochlear duct and proceeds to the apex. Similar changes are found in the hair cells of the vestibular system. Gentamicin was detected in the outer hair cells in the cochlea of animals receiving non-ototoxic doses of the drug and continued to increase for several days after withdrawal [ ]. This was followed by a third and very much slower phase of elimination (estimated half-life 6 months), and in the absence of ototoxicity gentamicin could still be detected in the outer hair cells 11 months after treatment. This may explain why patients who receive several courses of aminoglycosides in a year may be more susceptible to ototoxicity, and suggests that the cumulative dose (and by implication the duration of therapy) is the more important determinant of cochlear damage. However, it has been questioned whether ototoxicity correlates with plasma perilymph or whole-tissue concentrations of aminoglycosides [ ].

The effects of aminoglycosides on the medial efferent system have been assessed in awake guinea-pigs [ , ]. The ensemble background activity and its suppression by contralateral acoustic stimulation was used as a tool to study the medial efferent system. A single intramuscular dose of gentamicin 150 mg/kg reduced or abolished the suppressive effect produced by activation of the olivocochlear system by contralateral low-level broadband noise stimulation. This effect was dose-dependent and could be demonstrated ipsilaterally on the compound action potential, otoacoustic emissions, and ensemble background activity of the eighth nerve. Long-term gentamicin treatment (60 mg/kg for 10 days) had no effect, at least before the development of ototoxicity. Single-dose intramuscular netilmicin 150 mg/kg displayed blocking properties similar to gentamicin, although less pronounced, while amikacin 750 mg/kg and neomycin 150 mg/kg had no effect. With tobramycin 150 mg/kg and streptomycin 400 mg/kg a decrease in suppression was usually associated with a reduction of the ensemble background activity measured without acoustic stimulation, which may be a fist sign of alteration to cochlear function. There was no correlation between specificity and degree of aminoglycoside ototoxicity and their action on the medial efferent system.

Possible mechanisms and preventive strategies have also been investigated in pigmented guinea-pigs [ ]. Animals that received alpha-lipoic acid (100 mg/kg/day), a powerful free radical scavenger, in combination with amikacin (450 mg/kg/day intramuscularly) had a less severe rise in compound action potential threshold than animals that received amikacin alone. In a similar study in pigmented guinea-pigs, the iron chelator deferoxamine (150 mg/kg bd for 14 days) produced a significant protective effect against ototoxicity induced by neomycin (100 mg/kg/day for 14 days). The spin trap alpha-phenyl-tert-butyl-nitrone also protected against acute ototopical aminoglycoside ototoxicity in guinea-pigs. These studies have provided further evidence for the hypothesis that aminoglycoside ototoxicity is mediated by the formation of an aminoglycoside-iron complex and reactive oxygen species.

Aminoglycoside-induced ototoxicity may be in part a process that involves the excitatory activation of cochlear NMDA receptors [ ]. In addition, the uncompetitive NMDA receptor antagonist dizocilpine attenuated the vestibular toxicity of streptomycin in a rat model, further stressing that excitotoxic mechanisms mediated by NMDA receptors also contribute to aminoglycoside-induced vestibular toxicity [ ]. In two studies in guinea-pigs, nitric oxide and free radicals, demonstrated by the beneficial effect of the antioxidant/free radical scavenger alpha-lipoic acid, have been suggested to be involved in aminoglycoside-induced ototoxicity [ , ]. In contrast, insulin, transforming growth factor alpha, or retinoic acid may offer a protective potential against ototoxicity caused by aminoglycosides. However, they do not seem to promote cochlear hair cell repair [ ]. In guinea-pigs, brain-derived neurotrophic factor, neurotrophin-3, and the iron chelator and antioxidant 2-hydroxybenzoate (salicylate), at concentrations corresponding to anti-inflammatory concentrations in humans, attenuated aminoglycoside-induced ototoxicity [ , ]. In rats, concanavalin A attenuated aminoglycoside-induced ototoxicity, and kanamycin increased the expression of the glutamate-aspartate transporter gene in the cochlea, which might play a role in the prevention of secondary deaths of spiral ganglion neurons [ , ]. Finally, 4-methylcatechol, an inducer of nerve growth factor synthesis, enhanced spiral ganglion neuron survival after aminoglycoside treatment in mice [ ].

In hatched chicks repeatedly injected with kanamycin, afferent innervation of the regenerated hair cells was related more to the recovery of hearing than efferent innervation [ ].

In an animal model of ototoxicity, the most severe degeneration in the cristae ampullaris, utricle, and saccule was observed after administration of streptomycin. The severity of the vestibular damage in terms of magnitude was in the order streptomycin > gentamicin > amikacin > netilmicin [ ].

Human studies

There is evidence that the site of ototoxic action is the mitochondrial ribosome [ , ]. In some countries, such as China, aminoglycoside toxicity is a major cause of deafness. Susceptibility to ototoxicity in these populations appears to be transmitted by women, suggesting mitochondrial inheritance. In Chinese, Japanese, and Arab-Israeli pedigrees a common mutation was found. A point mutation in a highly conserved region of the mitochondrial 12S ribosomal RNA gene was common in all pedigrees with maternally inherited ototoxic deafness [ ]. A mutation at nucleotide 1555 has been reported to confer susceptibility to aminoglycoside antibiotics, and to cause non-syndromic sensorineural hearing loss. Outside these susceptible families, sporadic cases also have this mutation in increased frequency. In patients bearing this mitochondrial mutation hearing loss was observed after short-term exposure to isepamicin sulfate [ ]. These findings might create a molecular baseline for preventive screening of patients when aminoglycosides are to be used [ ].

Differences between sera from patients with resistance or susceptibility to aminoglycoside ototoxicity have been described in vitro [ ]. Sera from sensitive but not from resistant individuals metabolized aminoglycosides to cytotoxins, whereas no sera were cytotoxic when tested without the addition of aminoglycosides. This effect persisted for up to 1 year after aminoglycoside treatment.

Susceptibility factors

Several factors predispose to ototoxic effects ( Table 2 ). Drug-related toxicity is influenced by the quality of prescribing. Overdosage in patients with impaired renal function, unnecessary prolongation of treatment, and the concomitant administration of other potentially ototoxic agents should be avoided. The exact mechanism of increased toxicity in patients with septicemia and a high temperature is not clear; the possible relevance of additive damage by bacterial endotoxins has been discussed [ ]. Dehydration with hypovolemia is probably the main reason for the increased toxicity experienced when aminoglycosides are given with loop diuretics, but furosemide itself does not seem to be an independent risk factor [ ]. Attempts have been made in animals to protect against ototoxicity by antioxidant therapy (for example glutathione and vitamin C), as well as iron chelators and neurotrophins [ ].

Hereditary deafness is a heterogeneous group of disorders, with different patterns of inheritance and due to a multitude of different genes [ , ]. The first molecular defect described was the A1555G sequence change in the mitochondrial 12S ribosomal RNA gene. A description of two families from Italy and 19 families from Spain has now suggested that this mutation is not as rare as was initially thought [ , ]. The A1555G mutation is important to diagnose, since hearing maternal relatives who are exposed to aminoglycosides may lose their hearing. This predisposition is stressed by the fact that 40 relatives in 12 Spanish families and one relative in an Italian family lost their hearing after aminoglycoside exposure. Since the mutation can easily be screened, any patient with idiopathic sensorineural hearing loss may be screened for this and possible other mutations.

In an Italian family of whom five family members became deaf after aminoglycoside exposure, the nucleotide 961 thymidine deletion associated with a varying number of inserted cytosines in the mitochondrial 12S ribosomal RNA gene was identified as a second pathogenic mutation that could predispose to aminoglycoside ototoxicity [ ]. Molecular analysis excluded the A1555G mutation in this family.

The A1555G mutation in the human mitochondrial 12S RNA, which has been associated with hearing loss after aminoglycoside administration [ ] and has been implicated in maternally inherited hearing loss in the absence of aminoglycoside exposure in some families, can be identified by a simple and rapid method for large-scale screening that uses one-step multiplex allele-specific PCR [ ].

Using lymphoblastoid cell lines derived from five deaf and five hearing individuals from an Arab-Israeli family carrying the A1555G mutation, the first direct evidence has been provided that the mitochondrial 12S rRNA carrying the A1555G mutation is the main target of the aminoglycosides [ ]. This suggests that they exert their detrimental effect through altering mitochondrial protein synthesis, which exacerbates the inherent defect caused by the mutation and reduces the overall translation rate below the minimal level required for normal cellular function.

A second pathogenic mutation that could predispose to aminoglycoside ototoxicity has been identified in an Italian family, of whom five members became deaf after aminoglycoside exposure [ ]. In the mitochondrial 12S ribosomal RNA gene, the deletion of nucleotide 961 thymidine was associated with a varying number of inserted cytosines. Transient evoked otoacoustic emission has been suggested as a sensitive measure for the early effects of aminoglycosides on the peripheral auditory system and may be useful as a tool for the prevention of permanent ototoxicity [ ].

In 87 patients with tuberculosis or non-tuberculous mycobacterial infections randomized to receive intravenous streptomycin, kanamycin, or amikacin, 15 mg/kg/day or 25 mg/kg three times per week, the dose and the frequency of administration were not associated with the incidences of ototoxicity (hearing loss determined by audiography) or vestibular toxicity (determined by physical examination) [ ]. Ototoxicity, which occurred in 32 patients, was associated with older age and with a larger cumulative dose. Vestibular toxicity, which occurred in eight patients, usually resolved. Subjective changes in hearing or balance did not correlate with objective findings.

In a case–control study in 15 children under 33 weeks gestation with significant sensorineural hearing loss and 30 matched controls, the children with sensorineural hearing loss had longer periods of intubation, ventilation, oxygen treatment, and acidosis, and more frequent treatment with dopamine or furosemide [ ]. However, neither peak nor trough aminoglycoside concentrations, nor duration of jaundice or bilirubin concentration varied between the groups. At 12 months of age, seven of the children with sensorineural hearing loss had evidence of cerebral palsy compared with two of the 30 controls. Therefore, preterm children with sensorineural hearing loss required more intensive care in the perinatal period and developed more neurological complications than controls, and the co-existence of susceptibility factors for hearing loss may be more important than the individual factors themselves.

Diagnosis

To recognize auditory damage at an early stage and avoid severe irreversible toxicity, repeated tests of cochlear and vestibular function should be carried out in all patients needing prolonged aminoglycoside treatment. Pure-tone audiometry at 250–8000 Hz and electronystagmography with caloric testing are the standard methods. The first detectable audiometric changes usually occur in the high-tone range (over 4000 Hz) and then progress to lower frequencies. Hearing loss of more than 15 db is usually considered as evidence of toxicity. Brainstem auditory-evoked potentials have been recommended as a means of monitoring ototoxicity in uncooperative, comatose patients [ , ]. This technique is time-consuming and requires some expertise, but may become a useful tool for detecting damage at an early stage. It also provides information on pre-existing changes, which is otherwise rarely available in intensive care patients [ ].

Prevention

In rats, ototoxicity caused by gentamicin or tobramycin was ameliorated by melatonin, which did not interfere with the antibiotic action of the aminoglycosides [ ]. The free radical scavenging agent alpha-lipoic acid has previously been shown to protect against the cochlear adverse effects of systemically administered aminoglycoside antibiotics, and in a recent animal study it also prevented cochlear toxicity after the administration of neomycin 5% directly to the round window membrane over 7 days [ ].

Loss of spiral ganglion neurons can be prevented by neurotrophin 3, whereas hair cell damage can be prevented by N -methyl- d -aspartate (NMDA) receptor antagonists. In an animal study, an NMDA receptor antagonist (MK801) protected against noise-induced excitotoxicity in the cochlea; in addition, combined treatment with neurotrophin 3 and MK801 had a potent effect in preserving both auditory physiology and morphology against aminoglycoside toxicity induced by amikacin [ ].