Rising bedaquiline resistance: exploring potential causes and next steps

Extract

Although the estimated incidence of multidrug-resistant (MDR)/rifampicin-resistant (RR) tuberculosis (TB) has been relatively stable in recent years, MDR/RR-TB still represents a significant public health threat, with an estimated number of 400 000 new cases in 2023 [1]. The US Food and Drug Administration granted bedaquiline (BDQ) accelerated approval in 2012, and although the label does not include recommendations for the treatment of RR-TB, BDQ is being incorporated into all guideline-recommended standard regimens for MDR/RR-TB worldwide [1, 2].

Shareable abstract

The availability of drug sensitivity testing upon rollout is pivotal to prevent the loss of novel drugs to resistance shortly after approval, and this challenge will require greater combined effort and collaboration from all stakeholders https://bit.ly/4qthp3v

Although the estimated incidence of multidrug-resistant (MDR)/rifampicin-resistant (RR) tuberculosis (TB) has been relatively stable in recent years, MDR/RR-TB still represents a significant public health threat, with an estimated number of 400 000 new cases in 2023 [1]. The US Food and Drug Administration granted bedaquiline (BDQ) accelerated approval in 2012, and although the label does not include recommendations for the treatment of RR-TB, BDQ is being incorporated into all guideline-recommended standard regimens for MDR/RR-TB worldwide [1, 2]. Recent data shows that by 2026, 78% of all MDR-TB patients are expected to receive BDQ [3]. BDQ, together with a later generation fluoroquinolone and linezolid, were the only drugs shown to reduce mortality in people living with MDR/RR-TB [4]. While these regimens represent a major step forward in the management of MDR/RR-TB by allowing for shorter duration all-oral regimens, they all contain and rely on a BDQ backbone. Therefore, we are particularly concerned in light of multiple recent reports of emerging BDQ resistance, with a recent meta-analysis showing a pooled prevalence of acquired BDQ resistance during treatment of 2.1% [5]. While the development of antimicrobial resistance is an inevitable consequence of antimicrobial use, BDQ has a central role as the backbone of current MDR-TB treatment regimens with limited availability of effective alternative options. Careful consideration is therefore needed to identify strategies to slow the pace of resistance development [5]. Resistance will ultimately lead to a return to use of longer, less potent regimens, likely accompanied by a need for hospitalisation, worse side-effect profiles and clinical outcomes, and potentially a major financial burden both to individuals and national TB programmes [6–9].

Multiple variables have been linked to increased risk of BDQ resistance since programmatic implementation. First, BDQ resistance is more likely to occur in individuals who have failed TB treatment previously, individuals who have previously or are currently receiving clofazimine, and individuals with a fluoroquinolone-resistant strain for which fewer drugs are effective [6]. Secondly, individuals infected with strains containing Rv0678 mutations were found to be more likely to have baseline BDQ resistance and are more likely to fail BDQ-containing regimens; the risk of treatment failure becomes even more likely in strains with pre-existing or co-existing fluoroquinolone resistance [10]. Thirdly, the number of effective drugs in a regimen are key determinants for acquired BDQ resistance on therapy [11].

However, the exact reasons for the development of BDQ resistance remain unclear. There are several factors that are thought to possibly contribute to the development of BDQ resistance, many linked to suboptimal BDQ exposure, facilitating the emergence of resistance. One explanation may be found in the long time it takes BDQ to reach steady concentrations, possibly providing a window for the emergence of resistance [8]. This could be aggravated by suboptimal compliance, something that modelling has shown BDQ at increased risk for due to the thrice-weekly dosing regimen [12]. Furthermore, differences in inter-individual pharmacokinetics of BDQ may lead to suboptimal exposure in certain individuals or populations, and has been proposed as one of the factors contributing to the emergence of drug resistance in TB [13, 14]. Recent work has shown that individuals of black ethnicity achieve lower plasma levels of BDQ than individuals from other ethnic groups, and this is also reflected in BDQ's approval label [15]. It is possible that lower exposure in this group, in addition to widespread addition of BDQ to failing regimens during initial introduction of BDQ, has contributed to the high levels of resistance seen in South Africa and Mozambique. In the Western Cape, within a cohort of 40 people living with RR-TB that had late positive cultures, nearly 50% developed BDQ resistance and in 2021 14% of RR strains in the national TB reference laboratory in Mozambique were BDQ resistant, respectively [8, 15]. A summary of possible drivers of BDQ resistance is shown in figure 1. Further understanding of the contribution of these factors to the emergence of BDQ resistance will facilitate the implementation of preventative measures. Possible strategies – as suggested by mouse models where BDQ resistance was detected during treatment with BDQ, pretomanid and linezolid, but not with BDQ, pretomanid, moxifloxacin and pyrazinamide [16] – could be the co-administration of further rapidly bactericidal drugs in addition to current BDQ-based regimens until BDQ steady state is achieved, and an increase in BDQ dose in populations where decreased exposure is expected, although their effectiveness remains to be proven. In addition, the recent possibility for generic manufacturing of BDQ might help reduce potential local stockouts that may contribute to resistance emergence [17]. Finally, to adequately confirm and understand previous trends, there is a need for population-level surveillance capacity. This understanding is key in preventing further emergence of resistance as it will enable crucial support of individuals undergoing treatment to avoid treatment interruptions.

FIGURE 1.

Possible drivers of bedaquiline resistance.

To monitor the emergence of BDQ resistance, phenotypic antimicrobial susceptibility testing prior to initiation of BDQ-containing regimens is recommended [18]. However, the implementation of phenotypic antimicrobial susceptibility testing in programmatic settings, either by MGIT or broth microdilution, is currently hampered by the complex methodology required, limitations in infrastructure and expertise, and long culture times, meaning individuals will be started on a BDQ-based regimen long before resistance results are obtained. In addition, phenotypic antimicrobial susceptibility testing is further complicated by the presence of strains with low-level resistance and minimum inhibitory concentrations (MICs) around the World Health Organization (WHO)-defined critical concentrations for antimicrobial susceptibility testing [19]. Three mechanisms of drug resistance to BDQ currently meet the criteria for group 1 or 2 grading of association to resistance, including mutations in atpE, resulting in high-level drug resistance, and mutations in either Rv0678 or pepQ, both leading to the previously described low-level resistance [20]. Consequently, several scientists now strongly advocate for the adoption of a composite (genetic and phenotypic) reference standard for BDQ resistance, as reliance on phenotypic testing alone is no longer sufficient [21]. Therefore, there is an urgent need for development and implementation of novel rapid tests to identify BDQ resistance. Targeted next generation sequencing has emerged as a promising method for fast and reliable genetic testing of mutations associated with resistance to multiple drugs simultaneously [22]. Of the currently available assays for Mycobacterium tuberculosis, only the Deeplex Myc-TB assay developed by GenoScreen is endorsed by the WHO for follow-on detection of BDQ resistance in individuals with both drug-susceptible and MDR/RR-TB. However, it currently only targets Rv0678, the most frequent cause of BDQ resistance in clinical settings, meaning cases with other mutations may go unnoticed. In addition, it appears that MmpL5 loss of function mutations in certain strains can also lead to hyper-susceptibility to both BDQ and clofazimine in vitro and can suppress the resistance-conferring effect of Rv0678 loss-of-function mutations; given its frequency, these MmpL5 mutations should be included as target to gather insight on how this translates into clinical settings [19]. A new version called TB XL that, according to the manufacturer, will address these questions, is being developed. Given the wide range of possible mutations that can influence BDQ susceptibility, advances in genotypic susceptibility testing are only possible if whole genome sequencing data continue to be collected from diverse geographical settings. Finally, in addition to the patient-level benefit of improving the methodology and availability of BDQ susceptibility testing, these improvements would allow for ongoing population-level surveillance for BDQ resistance, in addition to data being captured by periodic surveys. This information is crucial to understanding the drivers of BDQ resistance emergence, and allows for timely and targeted interventions to prevent further spread or emergence.

Finally, in addition to resistance monitoring, the global emergence of BDQ resistance and the accompanying increase in unfavourable treatment outcomes in this group (outcomes from recent retrospective cohorts summarised in table 1) highlights the importance of limiting the further development and spread of BDQ resistance to preserve its effectiveness for future use. This starts with improving TB treatment more broadly, including current standard of care for drug-susceptible TB, thereby breaking the chain that leads to further increasing incidence of MDR/RR-TB and the necessity for use of BDQ in the first place. In addition, an increasing percentage of BDQ resistance is transmitted and not acquired, highlighting the need for early detection and treatment initiation to prevent further spread. Finally, investigation to determine the factors determining why BDQ resistance is higher in some areas and lower in others is needed. Together, this will help shape the implementation of approved TB control measures to the needs of each setting. However, judging from history, it is important to simultaneously look ahead and start developing BDQ-sparing regimens. With 18 novel compounds currently in the clinical pipeline, development of MDR/RR-TB regimens without BDQ should be considered as there is the risk of repeating prior mistakes [23]. Moreover, it is important to start learning from our previous errors so that new effective drug regimens can be implemented sooner, rather than single drugs. This is aligned with the basic antimicrobial resistance principle of never adding a single drug to failing regimens [24]. Furthermore, the availability of antimicrobial susceptibility testing upon rollout is pivotal to prevent the loss of novel drugs to resistance shortly after approval. The availability of MIC testing protocols should be a requirement when seeking regulatory approval for novel drugs, a challenge that will require greater combined effort and collaboration from all stakeholders [25].

TABLE 1.

Prevalence of unfavourable treatment outcomes in recent (2025) bedaquiline (BDQ)-resistant cohorts

Author	Country/region	Design/period	Population included	N	Regimen	Antimicrobial susceptibility testing method	Resistance prevalence within cohort at baseline	Unfavourable outcome (BDQ-R individuals)	Unfavourable outcome (BDQ-S individuals)
Singla [26]	India	Retrospective cohort, Dec 2020 to Dec 2022	DR-TB, >1 month BDQ exposure and treatment failure	117	Individualised BDQ regimens	Phenotypic	36% BDQ-R	87% (40% died, 40% failed, 8% LTFU)	42% (13% died, 28% failed, 1.6% LTFU)
Mdlenyani [27]	South Africa	Retrospective matched cohort, Jan 2018 to Jun 2023	Matched case–control individuals (BDQ-R and BDQ-S)	164	RR-TB regimens (88% BDQ-containing in BDQ-R)	Phenotypic	50% BDQ-R	67% (14% died, 43% treatment failure, 10% LTFU)	N/A
Ardizzoni [28]	Armenia	Retrospective cohort, 2013–2015	Individuals with MDR-TB enrolled in BDQ compassionate use programme	39	Individualised BDQ+LZD regimens	Genotypic	12% BDQ-R mutations at baseline, 52% acquired mutations	50% in individuals with baseline resistance, 92% in individuals with acquired resistance	N/A
Kherabi [29]	WHO European region	Retrospective cohort, 2017–2023	All BDQ-R individuals from participating contributors	188	RR-TB regimens (59.7% BDQ-containing in BDQ-R)	Genotypic and phenotypic	66% BDQ-R	66.1% with BDQ-containing regimens; 66.7% with BDQ-free regimens	N/A

DR-TB: drug-resistant tuberculosis; BDQ-R: bedaquiline-resistant; BDQ-S: bedaquiline-susceptible; MDR-TB: multidrug-resistant tuberculosis; RR-TB: rifampicin-resistant tuberculosis; LZD: linezolid; LTFU: lost to follow-up.