By definition, chemotherapy of MDR-TB cannot rely upon isoniazid and rifampicin, the two most powerful drugs for the treatment of tuberculosis [9]. Thus, depending on the individual susceptibility pattern, residual first-line oral drugs must be appropriately combined with additional second line drugs comprising injectable aminoglycosides (amikacin, kanamycin, capreomycin), fluoroquinolones (ciprofloxacin, ofloxacin, levofloxacin, moxifloxacin, gatifloxacin), old bacteriostatic second line anti-tuberculosis agents (ethionamide, protionamide, cycloserine, para-amino salicylic acid, thiocetazone) and anti-tuberculosis agents with unclear efficacy (clofazimine, amoxicillin/clavuanate, clarithromycin, linezolid) [2].

A crucial issue related to long-term administration of the injectable group is toxicity. Ototoxicity and nephrotoxicity are well recognized as dose-related adverse effects of aminoglycosides [10]. Ototoxicity and nephrotoxicity have been of major concern because of the narrow therapeutic range of these agents and the wide variability in pharmacokinetics among patients [11]. Amikacin is a semi-synthetic aminoglycoside and shows excellent activity against Mycobacterium tuberculosis and atypical mycobacteria and has been used in the treatment of disseminated atypical Mycobacterium infection in AIDS patients [12].

Kanamycin, an antibiotic elaborated by Streptomyces kanamyceticus has shown activity against Mycobacterium tuberculosis. But as the therapy of this disease is protracted and involves the administration of large total doses of the drug, with the risk of ototoxicity and nephrotoxicity, kanamycin should be used only in infection with organisms that are resistant to the more commonly used agents [12]. It is more toxic to cochlea with well documented ototoxicity [13,14] but is still being commonly used in clinical settings like ours (in developing countries) for MDR-TB where cost considerations are a major factor in patient compliance (because of having one fourth the cost of amikacin and one tenth the cost of capreomycin).

Complete otolaryngologic examination was done as a part of pre-treatment clinical examination in all the patients. Those patients with any pre-treatment evidence of hearing loss on history, clinical assessment (patients with evidence of infective pathology in ear were excluded) or pure tone audiometry, whether conductive (A-B gap > 10 dB) or sensorineural were excluded. Baseline pure tone audiograms between 125 Hz and 8000 Hz were performed for all the patients in a sound treated room. Baseline, in addition to follow-up audiological testing, included both air-and bone-conduction threshold measurements in all the patients. All the patients who had received any ototoxic drug as a part of previous regimen were excluded from the study.

The criteria used for determining ototoxic threshold shift from baseline audiogram were: (1) 20 dB or greater decrease at any one test frequency, (2) 10 dB or greater decrease at any two adjacent frequencies, or (3) loss of response at three consecutive frequencies where responses were previously obtained (ASHA, 1994)[19]. All the changes were confirmed by retest on the same day. These patients were divided into three groups depending upon the ototoxic aminoglycoside used. Group I (n = 34) patients received amikacin (15 mg/kg per day, intramuscular, single daily dose), group II (n = 26) patients received kanamycin (15 mg/kg per day, intramuscular, single daily dose) and group III (n = 4) patients received capreomycin (15 mg/kg per day, intramuscular, single daily dose) as a part of complete regimen comprising of one injectable (aminoglycoside), one quinolone with a minimum of five drugs depending on the drug sensitivity testing and the costs involved. Streptomycin was not used because of high resistance in these patients on drug sensitivity testing. Those patients receiving two ototoxic drugs were excluded from the study. Total duration of therapy in all the patients ranged from 18 to 24 months after sputum smear/culture conversion and aminoglycoside was used for six months after sputum conversion.

Seven patients (20.6%, n = 34) of group I (amikacin) showed sensorineural hearing loss (SNHL) involving the higher frequencies (HFL). Amikacin was stopped on the first report of hearing loss and patient shifted to another of the second line drug. Follow-up audiogram showed development of FLAT loss in two (5.9%, n = 34) of these patients, (Subject # 2) at six months (pure tone audiogram at 6 months, PTA6) and (Subject # 4) at 10 months (pure tone audiogram at 10 months, PTA10) (Table ​(Table1).1). Duration of aminoglycoside use in group I was 235 ± 40 days (n = 34) while duration of aminoglycoside use in seven patients of group I which developed ototoxicity (n = 7) was 163 ± 45 days (as amikacin was discontinued on first report of hearing loss).

Four patients (15.4%, n = 26) of group II (kanamycin) had SNHL involving higher frequencies (HFL). In two (7.7%, n = 26) patients lower frequencies were also involved (FLAT) even when Subject # 8 had injectable drug stopped at 4 months while Subject # 10 had the drug stopped on report of high frequency loss at 6 months (Table ​(Table1).1). Duration of aminoglycoside use in group II was 237 ± 34 days (n = 26) while duration of aminoglycoside use in four patients of group II which developed ototoxicity (n = 4) was 165 ± 30 days (as kanamycin was discontinued on first report of hearing loss).

One of the patients of group III (25.0%, n = 4) developed sensorineural hearing loss involving high frequencies (HFL) at 4 months (Table ​(Table2)2) and capreomycin was substituted with other second line drug based on drug sensitivity testing. Mean duration of aminoglycoside use in group III was 215 ± 64 (n = 4) days. Duration of capreomycin use in single patient with HFL (Subject # 12) was 120 days. None of the patients had any recovery in pure-tone thresholds after stopping the injectable treatment. Group-wise mean loss shown by patients with ototoxic threshold shift (Group I, n = 7; Group II, n = 4, Group III, n = 1) at different frequencies tested in the present study is shown in Table ​Table3.3. Mean loss (average decrease in pure tone threshold from baseline at each frequency tested in patients with ototoxic threshold shift in each group) in seven patients of group I fulfilled the criteria (ASHA' 1994) for ototoxic threshold shift in frequencies range of 4000–8000 Hz (HFL) but not in the frequency range from 250–3000 Hz. Mean loss in four patients of group II fulfilled the criteria for ototoxic threshold shift in both the frequency ranges i.e. 4000–8000 Hz (HFL) and 250–3000 Hz. As sample size for group III was small, the data was not further analyzed but a note of ototoxic threshold shift in 4000–8000 Hz range was made. Studies with larger number of patients using capreomycin are required but higher cost of the therapy with capreomycin limits its use in large number of patients in our setting.

The main constraint to the administration of aminoglycosides are risks of nephrotoxicity and ototoxicity [24]. Ototoxicity is the major irreversible toxicity of aminoglycosides. Cochlear damage can produce permanent hearing loss, while damage to vestibular apparatus results in dizziness, ataxia and/or nystagmus. Aminoglycosides appear to generate free radicals within the inner ear, with subsequent permanent damage to sensory cells and neurons resulting in permanent hearing loss [14]. The other major limitation to the clinical use of aminoglycosides continues to be concern for the development of nephrotoxicity. Nephrotoxicity has been defined as an increase in the baseline serum creatinine concentration of 0.5 mg/dl or a 50% increase, whichever is greater, on two consecutive occasions any time during therapy or up to 1 week after the cessation of therapy [25]. Evidence from studies with animals and humans has demonstrated a correlation between the nephrotoxic effect of aminoglycosides and the accumulation of these drugs in the cortex of kidney [26-28].

The present study evaluates the effect of parenteral second line aminoglycosides namely amikacin, kanamycin and capreomycin on hearing status of MDR-TB patients. We report a hearing loss documented by pure tone audiometry in 18.75% patients of MDR-TB using a single parenteral second line aminoglycoside involving higher frequencies (4000 to 8000 Hz) to start with and progressing to involve lower frequencies (500, 1000, 2000 and 3000 Hz) in 6.25% thus affecting the speech comprehension of the patient (n = 64). Speech comprehension can also be affected with hearing loss in the 4000 Hz range and may adversely affect communication especially in situations like environments with back ground noise [29]. The loss once developed has been found to be irreversible and none of the patients in the present study showed any improvement after stopping the therapy.

Ototoxicity is determined by comparing baseline data, ideally obtained prior to ototoxic drug administration, to the results of subsequent monitoring tests. Detecting changes in pure tone thresholds directly using serial audiograms is considered the most effective indicator of ototoxic hearing loss, particularly when ultra-high frequency thresholds are included [17,18]. Monitoring audiological evaluations after the baseline evaluations have been recommended 1–2 times per week for patients receiving ototoxic antibiotics [17,18,25]. Other approaches to audiologic monitoring for ototoxicity are high frequency audiometry and otoacoustic emissions [30].

In the present study, pure tone audiometry was performed every other month for each patient until the completion of therapy. Because aminoglycoside ototoxicity can progress after discontinuation of the drug [13], we also performed audiometric follow-up in all patients for an average of over one year after drug discontinuation. This long term follow-up confirmed that all aminoglycoside-induced hearing loss in this patient population was permanent and not reversible. Persistence of toxicity of sera has been reported up to one year in patients using aminoglycosides even after stopping the ototoxic drug [31]. Twice weekly audiograms as recommended were not performed in the present study because of cost involved and the inability of the patients from far distant places to report twice weekly at our center where facilities for conventional assessment of hearing are available. It is not common to find equipment for audiometry as well as trained staff at peripheral centers in developing country like ours. Conventional frequency range (250–8000 Hz) was used in the present study as only conventional audiometers with frequency range between 125 and 8000 Hz are available with us owing to low cost compared to high frequency equipment.

Different studies have reported hearing loss as an adverse drug reaction in patients of MDR-TB ranging from 6–18% [21,22,31]. The finding that higher frequencies are involved before the lower frequencies may be used as a monitoring procedure for the detection of ototoxicity and has the potential for minimizing irreversible communication deficits in patients receiving aminoglycoside therapy [32]. In all the patients showing hearing loss, the aminoglycoside was stopped and changed to another second line drug done. Incidence of hearing loss may have been reduced because the aminoglycoside was stopped immediately at the outset of ototoxicity and substituted with another second line drug. All the patients included in the present study completed the remaining part of the therapy. Other authors have also reported changing to other second line drugs and completion of full therapy [21,22].

A number of otoprotective agents are being investigated for protection against hearing loss induced by cisplatin, carboplatin, aminoglycosides or noise exposure. These agents delivered either before or in combination with ototoxic drugs may help to prevent ototoxicity. D-methionine as an otoprotective agent has shown protection against amikacin induced ototoxicity [33].

There is evidence that aminoglycoside accumulation in the kidney may be related to the dosing schedule i.e. administration of larger doses less frequently may reduce the level of drug accumulation in the kidney and has been prospectively shown to reduce the nephrotoxic potentials of aminoglycosides [25]. Conventional multiple daily dosing is being gradually abandoned in favor of once daily dosing and results from meta-analysis of randomized clinical trials show diminished [34] or comparable [35-38] nephrotoxicity, better [35,36,39] or comparable [34,37,38] efficacy and comparable [34-36,38,39] ototoxicity with once daily dosing among adults. Once daily dosing has been used in all the patients in the present study but individualized dosing based on monitoring of serum levels of aminoglycosides has not been used in present study.

Individualized aminoglycoside dosing guided by targeted peak and trough concentrations in serum on the basis of the patient's individual pharmacokinetics parameters and standard equations has been related to decreased toxicity [40]. An association of ototoxicity with nephrotoxicity and with an elevated mean trough aminoglycoside serum level has been observed in patients treated with aminoglycosides [41]. Because of economic constraints and the non-affordability by the patients, serum levels of aminoglycosides during the therapy were not measured in the present study. But with individualized dosing based on patient's individual pharmacokinetic monitoring guided by targeted peak and trough concentrations, side effects could probably be avoided in some cases.