Comparing amikacin and kanamycin-induced hearing loss in multidrug-resistant tuberculosis treatment under programmatic conditions in a Namibian retrospective cohort

Amikacin and kanamycin belong to a group of antibiotics called aminoglycosides, which are used in the treatment of Gram-negative bacterial and mycobacterial infections. These aminoglycosides, in combination with fluoroquinolones, form the backbone for the treatment of multidrug-resistant tuberculosis (MDR-TB), as recommended by the World Health Organization (WHO) [1–3]. A major safety concern of the aminoglycosides is their ability to induce ototoxicity, especially during their long-term use in MDR-TB treatment [4–6]. Depending on the part of the inner ear that is affected as well as the selectivity of the aminoglycoside, the ototoxicity could be auditory or vestibular [7]. The current study focusses on the auditory toxicity (hearing loss or deafness) caused by amikacin and kanamycin. Aminoglycoside-induced hearing loss is permanent, although in some cases; it may be alleviated by the use of hearing aids, cochlear implants or speech rehabilitation, which unfortunately, are costly interventions. By experiencing hearing loss, patients end-up suffering from a distressful yet preventable drug-related disability that may negatively impact on their quality of life and limit their capability to work, for example, in occupations where good hearing ability is a requirement. In children, speech development may be severely compromised [8].

Aminoglycosides have a narrow therapeutic index; hence require careful monitoring of serum levels, particularly during their prolonged use in MDR-TB treatment, to prevent the occurrence of dose-dependent ototoxicity [9, 10]. In addition, regular audiologic assessments may help in the early detection of hearing impairment, before the damage becomes extensive and irreversible [11–13]. Some patients are genetically predisposed to suffering from aminoglycoside-induced hearing loss and genetic typing may be useful in identifying such patients [14–17]. Yet many patients in developing countries do not have access to such interventions or cannot afford them, due to weak public sector health systems and high levels of poverty [18].

Namibia is a developing country situated in the south-western part of Africa. It is classified by the World Bank as an upper middle income country [19]. At the time of this study, there were 13 regional centers for treating patients diagnosed with MDR-TB. One of the centers - the Walvis Bay MDR-TB treatment site - began assessing patients for aminoglycoside-induced hearing loss in 2004. In 2008, Namibia changed the preferred aminoglycoside for MDR-TB treatment from amikacin to kanamycin - which was cheaper and more readily available—and introduced capreomycin as an option for patients prone to hearing loss. Later in the same year, other MDR-treatment sites commenced the systematic audiometric monitoring of patients on MDR-TB treatment for the early detection and management of aminoglycoside-induced hearing loss. The change from amikacin to kanamycin; and the introduction of systematic audiometry provided us with the opportunity of comparing the incidence of hearing loss in patients treated with amikacin and kanamycin-based MDR-TB regimens respectively, in real-life programmatic conditions.

Even though amikacin and kanamycin have been in clinical use for over 50 years, surprisingly to-date, the evidence on their comparative risk of inducing hearing loss is scarce. Moreover, studies often have not been well done, especially in terms of measuring the hearing loss and patients with human immunodeficiency virus (HIV) co-infection have been underrepresented [6]. In sub-Sahara Africa, where the HIV and TB burden are still high [20, 21], HIV co-infection becomes a key consideration in the successful treatment of MDR-TB [22]. Tuberculosis patients co-infected with HIV are an important subgroup because of the potential effect of HIV and antiretroviral treatment on hearing function [23–27].

The aim of this study was to compare the cumulative incidence of hearing loss among patients treated for MDR-TB using amikacin or kanamycin-based regimens, and to identify those that were most-at-risk. The high prevalence of HIV co-infection among patients diagnosed with MDR-TB in Namibia during the period of the study also enabled us to examine the influence of HIV infection on the risk of aminoglycoside-induced hearing loss.

Adverse drug reactions are an important consideration for patients treated for MDR-TB infection where the prolonged treatment with amikacin or kanamycin is likely to result in the development of permanent hearing loss [16]. We report a high incidence of aminoglycoside-induced hearing loss, which was more frequent in patients treated with amikacin-based regimens than in those containing kanamycin. The high cumulative incidence of hearing loss (75 %) in the amikacin-exposed group in our setting is similar to the 70 % that was reported by Reza Javadi et. al. [30], while the 56 % incidence for kanamycin is similar to the 58 % reported by Sataloff and colleagues [31]. This provides compelling evidence that amikacin is more ototoxic than kanamycin, in real-life clinical practice.

To the best of our knowledge, the current study is the first one to quantify the comparative risk of hearing loss of amikacin versus kanamycin in their real-life use for MDR-TB treatment in a low-resource setting. It builds on previous research from other settings, which suggested that amikacin was associated with a greater risk of hearing loss, but did not quantify the magnitude of that risk [16]. Our finding is corroborated by the works of Duggal and Sarkar as well as by Sturdy and colleagues [16, 18] . In Duggal and Sarkar’s study, seven out of 34 patients (20.6 %) treated with amikacin for MDR-TB experienced sensorineural hearing loss involving the higher frequencies while a lesser proportion of four out of 26 patients (15.4 %) treated with kanamycin experienced the same type of hearing loss [18]. Similarly, Sturdy et.al. monitored the occurrence of hearing loss in 50 MDR-TB patients, 29 of whom were treated with amikacin, 11 with capreomycin and 10 with streptomycin, and found that the use of amikacin (P = 0.02) and decreased renal function (P = 0.01) were significantly associated with the development of hearing loss [16]. Although both studies involved small numbers of patients, their findings have been crucial in elucidating on the relative ototoxicity of the aminoglycosides used in MDR-TB treatment. Considering our current findings and those of previous research, we encourage clinicians and managers of the TB control programs that are still using amikacin as the preferred aminoglycoside for treating MDR-TB infection, to consider changing to kanamycin. Switching to kanamycin and implementing other preventive measures, will help to reduce the occurrence of aminoglycoside-induced hearing loss among patients treated for MDR-TB.

The hearing loss seen in our study was sensorineural, mostly bilateral and began by affecting higher frequencies, then progressing to lower conversational-level frequencies as the severity of deafness increased. This finding is consistent with the pathophysiology of aminoglycoside-induced hearing loss [4, 9]. After parenteral administration, aminoglycosides enter the inner ear fluids of the organ of Corti and the sensory hair cells where they are thought to react with heavy metal ions to form highly reactive free radicals that damage the stereocilia of the sensory hair cells [32, 33]. There is emerging evidence that the use of antioxidants like salicylates, ion chelating agents or calcium-binding proteins may prevent aminoglycoside-induced hearing loss [15, 34–37]. As illustrated in the case of patient X in this paper, a patient’s hearing ability could continue deteriorating even after withdrawing the aminoglycoside due to the long half-life or the sequestration of aminoglycosides in the endolymph of the cochlea canals, which continues to cause the loss of sensory hair cells long after stopping the administration of the drug [7, 38].

We, therefore, advocate for MDR-TB treatment programs to implement routine serial audiometry in patients treated with aminoglycosides even in resource constrained settings, so that patients showing early signs of hearing loss can be identified long before the damage is too late to be reversed. When the drugs for preventing aminoglycoside-induced hearing loss become licensed for clinical use, they should be readily made available to patients, as an additional means of protecting patients from developing aminoglycoside-induced hearing loss.

The risk of aminoglycoside-induced hearing loss was greatest in patients with lower baseline body weight (40–59 kg). This could be due to a drug dosing problem whereby clinicians may fail to titrate accurately the aminoglycoside doses according to individual patient body weight. Alternatively, these could be patients who were much sicker of tuberculosis disease than the heavier weighing patients. Since we are unable to ascertain the reason for this observation due to lack of data on serum drug concentrations, we recommend further studies on the long-term pharmacokinetics and pharmacodynamics of aminoglycosides in the context of MDR-TB treatment, taking into consideration patients’ renal function, anthropometric and genetic characteristics.

Patients co-infected with HIV were more at risk of amikacin-induced hearing loss than the HIV uninfected ones. There is emerging epidemiologic and clinical evidence about the association between HIV infection and hearing loss [27, 38, 39]. However, whether antiretroviral medicines also induce hearing loss is a question that is still unanswered because of the mixed findings of previous studies [25, 26, 40]. Besides, the current study doesn’t shed light on this question because of a lack of adequate data on the specific antiretroviral (ARV) drug regimens used by the patients and insufficient patient numbers, by ARV regimen. There is need for continued research in this area to better understand the effect of antiretroviral medicines on hearing ability.

Amikacin, kanamycin and other aminoglycosides are practically not metabolized by the human body and are excreted unchanged almost exclusively by glomerular filtration, hence they require the careful monitoring of their plasma levels during therapy [41, 42]. Unfortunately, therapeutic drug monitoring (TDM) of the aminoglycoside plasma levels was not performed during the treatment of patients for MDR-TB infection in Namibia. This service was not available in the public sector health system in Namibia and is not available in many developing countries [43], perhaps explaining the relatively high incidence of ototoxicity reported among patients on MDR-TB treatment in these countries. We recommend that TB treatment programs in developing countries should consider introducing routine therapeutic drug monitoring for patients treated with aminoglycosides or capreomycin, given the higher cost of correcting permanent hearing loss for the patient and the society. A comparative cost-effectiveness analysis of conducting TDM versus not doing TDM can further inform such a strategy.

Renal clearance may strongly affect the toxicity of aminoglycosides [41, 42, 44]. The lack of data on renal clearance for the patients included in our analysis is an important limitation of the current study. Although we retrieved the serum creatinine levels of 114 of the 353 patients from the laboratory database, the data was of no benefit to this analysis because it represented creatinine values that were measured at time points several months after the initiation of the MDR-TB treatment and baseline data were essentially missing. This, however, does not mean that clinicians in Namibia do not assess patients on MDR-TB treatment for renal function. They do so, but because of some practical challenges in the collection of data for this study, we could not retrieve all the data on serum creatinine levels for the patients in our analysis. Nonetheless, we encourage clinicians in our setting to systematically assess all patients on MDR-TB treatment, or those at the greatest risk, for renal function at baseline and over the course of the treatment, as recommended by the Namibian TB treatment guidelines [2].

Our study was an epidemiologic one, reflecting the real-life usage of amikacin and kanamycin in routine clinical practice. Using this study design, we identified patients that were most-at-risk of developing aminoglycoside-induced hearing loss. Importantly, hearing function was assessed using audiograms, which were part of the routine clinical follow-up of patients treated for MDR-TB. However, the data on the time-to-onset of hearing loss was unreliable because of the “batching” of patients for audiometry, due to the shortage of audiologists and audiology assistants.

Due to practical limitations about the documentation of audiometry in patients treated for MDR-TB at the sites prior to 2007, we could only retrieve information on 51 patients that were treated with amikacin for the period covered by this research. On the other hand, for kanamycin, we retrieved 302 patients, causing an imbalance in the numbers of patients exposed to amikacin and kanamycin, respectively. To assess for possible bias, the 51 patients on amikacin were checked for the risk of hearing loss against 51 randomly selected patients on kanamycin and the results were similar to those of the 353 patient sample.

There were several other limitations of this study. For example, there were too few patients in some sub-groups which limited the power of the study for multiple sub-group analyses. Besides, we were unable to collect data on other potential risk factors like the usage of antiretroviral medicines in HIV infected patients, genetic markers of ototoxicity and other unmeasured confounders including the use of other medicines known to be ototoxic.

The long-term use of amikacin in MDR-TB treatment led to a higher risk of the more severe forms of hearing loss compared to the use of kanamycin for the same indication. Males, patients with low baseline body weight and those co-infected with HIV were most-at-risk. We recommend that managers of MDR-TB treatment programmes should consider using kanamycin instead of amikacin for the treatment of MDR-TB; and invest more resources in building the capacity and skills of health care personnel for routine renal, serum therapeutic drug levels and audiometric monitoring of the most-at-risk patients treated with aminoglycosides. More research needs to be done to better understand the combined risk of hearing loss in patients concomitantly treated for MDR-TB and HIV infections. A better designed and more powered study is needed to confirm the comparative ototoxicity risk of amikacin and kanamycin; and associated risk factors.