Effectiveness and safety of bedaquiline-containing regimens in the treatment of MDR- and XDR-TB: a multicentre study

Discussion

The aim of the present study was to retrospectively evaluate the effectiveness, safety and tolerability of bedaquiline-containing regimens in a large observational cohort of MDR- and XDR-TB patients treated under different conditions (programmatic, expanded access, but not recruited for experimental studies).

To our knowledge, this is the largest study describing how safe, well tolerated and effective bedaquiline is within background regimens and the first in the scientific literature reporting on the nonclinical trial use of bedaquiline for the treatment of MDR- and XDR-TB patients in five continents.

In comparison with the data on individual patient data analysis (where there was a 43% success in XDR-TB) [3], in our bedaquiline-treated XDR-TB cohort the success rate was 71.3%.

In a phase 2 double blind, randomised controlled trial study by Diacon et al. [16] the median time to culture conversion in 79 bedaquiline-treated MDR-TB patients was 83 days; this compares with our median time of 60 days. In the study by Diacon et al. the culture conversion at the end of 24 weeks was 79% and at 120 weeks was 58% versus 91.7% at the end of therapy in our study, and the cure rates were 58% and 62.4%, respectively [16]. Regarding effectiveness, although difficult to attribute to bedaquiline, we can report that bedaquiline-containing regimens achieved culture conversion rates >90% at the end of treatment and treatment success >70% (76.7% when only calculating patients who started their treatment before December 31, 2014), higher than those observed in other MDR-TB cohorts [3, 15, 17].

Pym et al. [15] conducted a phase 2 trial to assess the safety and efficacy of bedaquiline in 233 patients; culture conversion was seen in 72.2% at 120 weeks, 8.6% of patients discontinued treatment and 6.9% of patients died [15].

Regarding safety, in the study by Diacon et al. [16], 13% of patients in the bedaquiline group died (10 out of 79 versus two out of 81 in the placebo group) versus a very similar 33 (13.4%) out of the 247 who had an evaluable outcome in our cohort.

The most frequent adverse events in our study were nausea (31.5%), otovestibular toxicity (23.3%), peripheral neuropathy (23.3%), vomiting (21.2%), anaemia (20.9%) and arthralgia (20.4%), their frequencies being slightly lower than those described during the licensing study by Diacon et al. (41% nausea, 29% vomiting and 37% arthralgia) [17]. Importantly, in the study by Diacon et al. the proportions of adverse events were similar in the treatment group versus placebo patients, suggesting that they were probably due to the background regimen. In this context, other second-line drugs such as fluoroquinolones or clofazimine might contribute to cardiologic or other adverse events [10, 11, 32], and invite caution and ECG monitoring.

The results of our study demonstrate that overall, bedaquiline-containing regimens achieve a relatively higher proportion of treatment success with a relatively lower proportion of adverse events within different settings than previously described, although a standardised approach for the notification of the adverse events was not implemented across the different settings (risk of underreporting).

Of note, culture conversion rates were higher than those reported in cohorts with an analogous degree of disease severity, with time to sputum smear and culture conversion identical or earlier to those observed in comparable cohorts; the proportion of treatment success was higher, and the percentage of adverse outcomes (death or failure) lower than those seen in available study cohorts with matching disease severity; adverse events due to bedaquiline leading to interruption of the drug were relatively uncommon (5.8%) [10, 11].

Our study confirms that bedaquiline-containing regimens are effective, as demonstrated by the fact that a sizeable number of patients were treated with salvage regimens due to previous treatment failure, unfavourable resistance profile, toxicity or all three.

The larger group of patients in different nonclinical trial conditions around the globe reinforces previous findings that bedaquiline is well tolerated and adverse events might be less common than previously thought.

Enthusiasm over bedaquiline and delamanid has been curtailed following concerns of potential cardiotoxicity. Both new drugs are associated with QT prolongation, which may potentially lead to arrhythmia and sudden death, a major reason why their combination has not been recommended. Moreover, the new drugs are likely to be associated with a fluoroquinolone and clofazimine, both known to prolong QT intervals. The specific role of the many drugs with QT-prolonging potential (and their summation or synergistic effect) still needs to be fully understood.

QT prolongation occurred in 9.7% of patients; interruption of bedaquiline due to adverse events occurred in 25 (5.8%) patients.

Although information on QT interval was available in 64% of cases, and the timing of their assessment not standardised, it seems the majority of cases died for non-heart-related reasons. A more accurate assessment of the adverse events (from data collection to detailed clinical investigation) will allow the evaluation of the relationship between death and QT prolongation. Unfortunately, our data are not standardised and therefore heterogeneous.

Close monitoring of drug safety should be implemented widely, particularly for rare adverse events. A comprehensive, population-level pharmacological surveillance in the post-marketing phase might allow a better assessment of the safety and tolerability profile of bedaquiline, alone or in combination with other potentially cardiotoxic anti-TB drugs [45].

The importance of implementing active TB drug-safety monitoring and management when bedaquiline is used programmatically cannot be overemphasised [45].

We underline the importance of using, in future studies, a standardised ECG monitoring protocol allowing to exclude inter- and intra-day variability in QTc measurements. For study purposes, 24-h Holter monitoring is probably the best strategy to assess the true impact of drugs on QTc. Clearly, this is not feasible under programmatic conditions.

The strengths of the study are the large cohort, the inclusion of cases from several countries (ranging from 7.7% of all patients receiving bedaquiline in the UK to 100% of bedaquiline-treated patients in Argentina, Australia (Victoria State), Greece, Peru, Russian Federation (Arkhangelsk oblast) and Spain) and the detailed information collected from the participating centres. The large sample size allows the comparison of treatment outcomes from different settings for the first time.

Some variables, such as the drug-resistance patterns, the number of previous anti-TB treatment cycles and HIV seroprevalence varied between the settings participating in the study.

However, the observational and retrospective design of the study has inbuilt limitations (recent guideline changes, different resource settings, different standards of care and dataset differences), so the research findings need to be confirmed by larger randomised controlled clinical trials and well-designed prospective observational studies.

Furthermore, several sources of bias may confound the reliability of the findings: missing random selectivity of the patients, heterogeneity between recruited geographic settings, different levels of completeness of collected data (especially adverse events and their severity), nonstandardised DST methods for several drugs, background regimen variability between countries which differed over time (mainly later generation fluoroquinolones), different TB/HIV co-infection prevalence and availability of antiretroviral drugs.

The effectiveness of bedaquiline might be overestimated, owing to the additive and/or synergistic role played by other effective drugs (e.g. carbapenems, linezolid, etc.) [7, 23–25, 27–31]. New experimental studies should be implemented to assess the critical effect of bedaquiline, as well as delamanid, in new combination therapies.

The different operating procedures adopted in various settings, as well as heterogeneous drug resistance patterns, may underestimate the real benefits of the bedaquiline-containing regimens.

However, stratified analyses allowed us to better assess the heterogeneity related to time and geographical factors. The new information provided by this observational study allows clinicians managing difficult-to-treat TB cases in programmatic conditions to better understand how to use bedaquiline in case the minimum number of active drugs necessary to design an effective regimen is lacking [1, 5, 11].

Although new compounds will soon become available to support the move towards TB elimination [46], this study confirms the “core drug” characteristics of bedaquiline in the management of MDR- and XDR-TB cases even in field conditions, and eventually for use in newly designed anti-TB regimens of the future.

While it is crucial to recommend the use of bedaquiline in difficult-to-treat cases, WHO guidelines should be followed assiduously in order to limit the emergence and spread of bedaquiline-resistant M. tuberculosis strains.