Low-dose amikacin in the treatment of Multidrug-resistant Tuberculosis (MDR-TB)

Despite decades of efforts at eliminating tuberculosis (TB), an estimated 10 million people were infected with TB and over 1.2 million people died of TB in 2018 [1]. Multidrug-resistant tuberculosis (MDR-TB), defined by resistance to both isoniazid and rifampin, has emerged as a global epidemic. Until recently, injectable aminoglycosides formed a core component of the treatment regimen used to treat MDR-TB, and both amikacin and streptomycin remain on the list of medicines recommended for use in MDR-TB regimens according to World Health Organization (WHO) guidelines [2].

Updated WHO guidelines for the treatment of drug-resistant TB have classified amikacin as a Group C drug, in part due to the significant toxicity associated with its use [2]. Previously, intravenous amikacin was classified as a Group B drug and more widely used, with a recommended dose of 15 mg/kg to a maximum dose of 1000 mg/day [3]. However, aminoglycosides at high doses are associated with significant toxicity, including ototoxicity, vestibular toxicity, and nephrotoxicity, and the optimal dose of amikacin remains controversial [4].

Although data regarding the pharmacokinetic (PK) and pharmacodynamic (PD) parameters in TB are relatively scarce, there is emerging evidence to support the use of lower-dose aminoglycoside therapy in the treatment of MDR-TB. Aminoglycosides are known to show concentration-dependent activity and thus the C_max/MIC (maximum concentration to minimum inhibitory concentration) ratio is thought to be the most relevant PK/PD parameter to monitor their efficacy [5]. Van Altena and colleagues reported using low-dose amikacin and kanamycin with drug dosing based on therapeutic drug monitoring, using a target C_max/MIC ratio of > 20. Their patients received a median dose of 6.5 mg/kg, achieved a mean weighted C_max/MIC ratio of 31.2 with amikacin, and had lower reported rates of ototoxicity, while still achieving good treatment outcomes [6].

Prior to the wide-spread use of bedaquiline and availability of all-oral regimens for treatment of MDR-TB, amikacin was a standard component of the MDR-TB regimen. In our TB referral hospital in Toronto, Canada, patients received amikacin at a dose of 8–10 mg/kg, with dose adjustment based on therapeutic drug monitoring using 30-min post-dose levels, targeting a C_max/MIC ratio of 25–35. These targets were selected based upon recommended peak serum levels for nontuberculous mycobacterial infections [7] (using extended interval dosing) and gram negative infections [8] (using traditional dosing), given the lack of TB PK/PD data at the time of the study. Here we report our experience with patient tolerability, side-effects and treatment outcomes using this method of dosing and therapeutic drug monitoring.

We conducted a retrospective cohort study of all patients with MDR-TB who received amikacin from 2010 to 2016, inclusive. The study was approved by our institutional research board. We included patients who met the following criteria: age > 18 years; confirmed MDR-TB on genotypic or phenotypic drug susceptibility testing; and received amikacin. Patients who had confirmed amikacin-resistance on drug susceptibility testing were excluded from the study.

Prior to 2018, amikacin was considered a Group B drug and therefore widely used in MDR-TB treatment. The introduction of bedaquiline and all-oral regimens for MDR-TB treatment has relegated amikacin to a Group C drug. However, aminoglycosides remain important drugs in MDR-TB treatment, as they exhibit concentration-dependent bactericidal activity against Mycobacterium tuberculosis [15, 16] and are highly effective in the treatment of drug-resistant TB isolates. In a large individual patient data meta-analysis, the use of injectable amikacin in patients with susceptible strains was associated with greater treatment success [17]. Of all the group C medications classified in the newest WHO guidelines, amikacin has the lowest adjusted odds ratio for treatment failure/relapse versus treatment success; its placement lower down on the list of preferred medications is primarily related to other issues with its use, including difficulty of administration, adverse events, and need for clinical monitoring [2]. In our patient cohort, patients receiving low-dose amikacin with regular clinical monitoring had a lower rate of adverse events compared to that reported in the literature with standard dose therapy. We believe that in situations where an adequate regimen cannot be constructed with Group A and B drugs, low-dose amikacin remains a viable option for MDR-TB treatment.

Extended interval aminoglycoside dosing regimens are used to optimize bactericidal activity by achieving desired peak concentration (C_max) to minimum inhibitory concentration (MIC) ratio [16]. C_max/MIC ratio is determined to be the most relevant PK/PD parameter to monitor amikacin efficacy based on the principle of aminoglycoside concentration-dependent killing [4, 5]. Experimental data demonstrate that a C_max/MIC ratio of at least 10 at the site of infection is necessary to optimize bactericidal activity for amikacin, with MIC ranging from 0.5 to 2 mg/L. [4, 18, 19] At our center, individual MICs for amikacin are not reported. However, the critical concentration for amikacin on the Bactec MGIT 960 system, used by the Public Health Ontario Laboratory, is known to be 1.0 [20]. Since the MIC for amikacin-sensitive isolates will be lower than the critical concentration and therefore less than 1, we conservatively set the MIC at 1 and used a serum drug level drawn 30-min after amikacin dose delivery to estimate the C_max/MIC ratio. By this method, we targeted and achieved a serum C_max/MIC of well over 10 (25–30 with daily dosing and 30–35 with thrice weekly dosing), while still delivering lower than conventional doses of amikacin and avoiding unnecessary toxicity. Van Altena et al targeted a serum C_max/MIC of > 20 with amikacin, and achieved a mean weighted C_max/MIC of 31.2 [6], very similar to those achieved by our method and with a similarly low rate of ototoxicity and good clinical outcomes. However, the penetration of aminoglycosides into lung tissue and other target organs is poorly understood, with variable estimates reported [4]. Therefore, the optimal serum C_max/MIC ratio, or the ratio needed to achieve a C_max/MIC of > 10 at the site of infection (usually the lung), remains unknown. Srivastava et al hypothesized that a C_max/MIC ratio of 67–89 would be necessary to achieve the target C_max/MIC ratio of 10.3 in the lung based on a hollow-fiber model, but this was based upon a reported serum to bronchial secretion ratio of 0.135 [19]. However, Sartre et al found a higher serum to bronchial secretion ratio of 0.3, which would necessitate a serum C_max/MIC ratio of approximately 34, much closer to the targets we utilized [21]. More research into amikacin serum to target organ penetration is needed [4]. Our results, and those of Van Altena et al [6], suggest that the targets we utilized may be adequate in a high-resource setting when amikacin is administered as part of a strong, multi-drug regimen.

Ototoxicity remains one of the most debilitating side effects related to aminoglycoside use. In a similar setting to ours but using standard dose aminoglycosides, ototoxicity based on regular audiogram testing occurred in 18% of patients in the Netherlands [22] 28% in the United Kingdom [23] and 37% of patients in the United States [24] although the US cohort included patients both with TB and nontuberculous mycobacterial disease. Rates of ototoxicty as high as 62% have been reported in MDR-TB patients treated in limited-resource settings [25] and is known to occur more frequently in populations with high rates of HIV co-infection [26]. Hearing loss tends to occur first at high frequencies, higher than those for human speech, and thus requires frequent monitoring with audiograms for early detection. Additionally, it can progress even after aminoglycosides have been stopped, making early detection imperative to avoid significant morbidity [27]. Risk factors for hearing loss with aminoglycoside use include higher cumulative dose of treatment [28], HIV co-infection [29], and older age [30]. In our cohort, 12% of patients experienced hearing loss confirmed on audiometry testing and the majority of these were detected at high frequencies. Our audiometry lab tested patients up to 12,000 Hz, detecting subtle changes at high frequencies that would have otherwise been missed, and only 10% had hearing loss at the conventionally tested frequencies of 8000 Hz or less. None of our patients experienced severe hearing loss. However, we found subjective tinnitus to be a common complaint. The low cumulative dose of aminoglycoside therapy combined with the low rate of HIV co-infection in our cohort likely explains the relatively limited ototoxicity we observed compared to rates reported in the literature, however even with low-dose amikacin, otoxicity remains a concern.

Hair cell injury in the inner ear by aminoglycosides also results in vestibular toxicity, which can manifest as disequilibrium and dizziness [31, 32], and can occur in up to 20% of patients receiving these drugs intravenously [32]. Dizziness was an uncommon complaint experienced by less that 10% of our patients, although our assessments were based on patient-reported symptoms and may therefore have been underestimated.

Aminoglycosides are renally excreted and known to cause nephrotoxicity. This is triggered by uptake of drug by renal tubular cells after glomerular filtration, leading to intra-cellular accumulation and subsequent tubular necrosis [33]. Nephrotoxicity occurs in up to 10% of MDR-TB patients [22, 34], though typically is reversible once the drug is discontinued. Our cohort had a slightly higher rate of nephrotoxicity than that reported in the literature, though this may be related to the definition used, as many patients met our criteria for nephrotoxicity but had serum creatinine and GFR within the normal range at the time of treatment completion. Based on the small number of patients in our cohort, there seemed to be a higher risk of irreversible nephrotoxicity when amikacin was used in combination with other known nephrotoxic medications.

Our study has several limitations that require discussion. As this was a consecutive cohort study with no control group, we could not evaluate the treatment effectiveness of low-dose amikacin in comparison to regimens with standard dosing or regimens excluding amikacin entirely. In addition, our sample size was modest to estimate the proportion of patients with sensorineural hearing loss. Due to the fact that our Public Health laboratory does not report individual MICs for amikacin, we used critical concentration to estimate an MIC of 1 for amikacin susceptible isolates, and therefore likely underestimated our achieved C_max/MIC ratios. Although monthly audiometry was routinely arranged for patients, a median of only 3 audiometry tests were administered despite a median treatment duration of 7 months of amikacin. Although we may have missed some patients with mild hearing loss, we believe that many patients did not attend audiology appointments because they did not have overt concerns with their hearing.

The treatment outcomes in our patient cohort were similar to those in other high-income settings [17] and in settings where standard dose aminoglycosides were routinely utilized [35, 36] and we found a lower rate of aminoglycoside-related adverse events than reported in the literature [22, 23, 25], suggesting that low-dose amikacin, when used with careful clinical monitoring, is relatively safe, yet still effective in the treatment of MDR-TB. When an adequate regimen cannot be constructed with Group A and B drugs, we believe that low-dose amikacin should be considered as a component of a multidrug treatment regimen, as long as adequate resources for monitoring are in place. While there remain significant limitations to the use of low-dose amikacin, including cost, difficulty of administration, and higher than desirable toxicity, our study suggests that a lower dose approach with therapeutic drug monitoring may limit toxicity, and that early detection of high frequency hearing loss enables discontinuation of amikacin, avoiding serious or debilitating hearing loss.

Although recent guidelines have relegated amikacin to a group C drug, we believe that it remains an important treatment option for MDR-TB in patients with complex resistance patterns or those with toxicity to Group A and B drugs. We demonstrate that low-dose amikacin with dose adjustment based on therapeutic drug monitoring has fairly limited toxicity with good treatment outcomes. Future prospective trials evaluating lower dose aminoglycosides and therapeutic drug level monitoring-guided dosing in MDR-TB would be beneficial.