Surveillance of adverse events in the treatment of drug-resistant tuberculosis: first global report

Discussion

The aim of the present study was to prospectively evaluate the frequency and severity of adverse events due to anti-TB drugs in a cohort of consecutive patients following the principles and methods of the WHO aDSM project.

The project worked as a “register” according to the WHO proposal to national programmes, aimed at promoting regular monitoring of adverse events, as well as collecting and reporting information on bacteriological status at diagnosis, during and at the end of treatment with final outcomes [7, 29, 30]. WHO recommends that countries use their existing surveillance methodology (electronic registers or existing electronic medical record systems) to extract the data and use them for clinical and public health purposes [29].

National TB programmes face difficulties in implementing aDSM and contributing to the global database. While the amount and type of information to collect is known and there is a sufficient burden of patients to satisfy the need to establish a routine adverse event recording and reporting system, the existing surveillance systems are currently not equipped to collect and analyse relevant variables.

The present project represents the first effort to document the feasibility of the aDSM approach and to collect quality scientific evidence on the adverse events in patients treated with new and repurposed drugs in “field conditions” in countries from all continents. The available scientific evidence on the safety and tolerability profile of anti-TB drugs can be retrieved from single observational and experimental studies. This project provides an international assessment following a register-based methodology.

A first important finding of the study is that when treatment regimens including bedaquiline and delamanid are used, the overall proportion of adverse events is reasonably low (8.7% of patients with serious adverse events (grade 3 and 4, no grade 5 adverse events)).

Notably, the injectables (and ethionamide) are the drugs causing more adverse events (table 2 and 3). With the new WHO all-oral approach [6, 7] and the availability of new drugs, capreomycin will no longer be used, and amikacin as well as ethionamide/prothionamide (and PAS) will be used less. In contrast, linezolid will be used increasingly, and being a drug with frequent and serious adverse events [20] there is a need to balance efficacy and toxicity [34]. Therapeutic drug monitoring may help achieve a therapeutic target of area under the curve/minimal inhibitory concentration >119 [35] while keeping trough concentrations low enough to prevent toxicity [36].

A second important outcome of this study is the possibility to carefully analyse the adverse events caused by bedaquiline and delamanid and by repurposed drugs. While overall 11.1% of the patients had adverse events to bedaquiline and 13.2% to delamanid (table 2), the serious adverse events due to these drugs were few, with only two patients discontinuing bedaquiline (0.35%) and one discontinuing delamanid (0.8%) because of cardiological adverse events [14, 37].

The proportion of patients reporting serious adverse events related to linezolid- and clofazimine-treated patients was 3% and 1.4%, respectively (annex 3).

Overall, 5.8% of the patients experienced an adverse event with levofloxacin and 3.8% with moxifloxacin, while only two patients had serious adverse events with moxifloxacin at the normal dose. None of the 12 patients treated with high-dose isoniazid and high-dose moxifloxacin reported adverse events.

Worryingly, an important proportion of adverse events identified by care providers were not reported to health authorities at the national level. We speculate that the explanations for the adverse event under-reporting include lack of awareness, the administrative burden (need to report to the country and to the aDSM system and to the drug manufacturer with different forms and multiple steps), confidentiality issues, the involvement of different sectors (public and private, prisons, etc.) and the fear of blame.

Furthermore, there were a few discrepancies on grading of the adverse event “QTc prolongation”. In four patients the adverse events were initially categorised as minor, even though they had resulted in the withdrawal of the offending drug. In agreement with the treating physician these adverse events were reclassified as serious. Asymptomatic conditions such as QTc prolongation need clear and well publicised criteria for accurate grading. QTc interval monitoring is usually performed in MDR-TB patients exposed to bedaquiline and delamanid in the WHO European region; although rare fatal events have been recorded, the ECG is a cost-effective preventive intervention for those at risk of developing cardiological adverse events [38].

To avoid premature discontinuation of potent drugs, available national and or international expert panels could be consulted for guidance [5, 31]. Medical conditions which can significantly increase the probability of a cardiological adverse event in MDR-TB patients (i.e. hypokalaemia and AIDS) should be monitored carefully [39].

When compared with the recent individual data meta-analysis performed in five cohorts (Armenia, Georgia, South Africa, France and Janssen Therapeutic cohort) on 537 patients treated with bedaquiline under compassionate use [13], the proportion of adverse events seems rather consistent with those found in our study. For example, 4.9% of patients suffered cardiac adverse events in the five-cohort study similar to the 5.5% in our study (denominator: patients treated with bedaquiline). Similarly, the proportion of interruptions of bedaquiline treatment in our study due to QTcF increase (0.35%) is consistent with that described in a recent systematic review of the literature (0.68%) [26].

The study has several strengths, including the number of countries participating [26] and a large sample size (to our knowledge one of the largest multinational cohorts of MDR-TB patients treated with bedaquiline- and/or delamanid-containing regimens based on WHO aDSM protocol), the prospective design and the accuracy of the information collected in countries with different epidemiological and economic backgrounds. Last, but not least, the majority of countries/states/regions (21 out of 26) provided data on all the consecutive patients treated with bedaquiline and delamanid during the study period.

A limitation is represented by the use of a consensus-based process to attribute adverse events to a specific drug, which included the local expert panel and the aDSM International Group panel. The scientific evidence on the safety and tolerability profile of a single drug or of a pharmacological combination was the driver adopted to identify the drug responsible of an adverse event; the probability of proving a causal relationship in specific patients, where the scientific evidence is poor, is very low. Further studies focused on the anti-TB drugs’ safety, based on the re-challenge methodology (i.e. drug administration after interruption following the occurrence of an adverse event) could help elucidate the adverse event profile of the anti-TB drugs. Furthermore, only a few centres carried out therapeutic drug monitoring to assess the relationship between adverse events and drug exposure (dosage and frequency of administration). Moreover, no variables related to concomitant medications, which could affect drug exposure, were recorded, with HIV therapy in patients with HIV infection the only exception. It was not possible to use approaches like the Naranjo score or the Yale algorithm [40, 41]. A second limitation is that few paediatric patients (four individuals aged <18 years) and people living with HIV (n=29, 4.4%) were included in the cohort to allow specific subanalyses.

The psychological role played by providing information on the risk of treatment failure following drug withdrawal, as well as potential biased communication with migrants and the clinical setting (e.g. ambulatory care), could have affected the patients’ tolerability profile and the reporting of adverse events. Unfortunately, we did not collect any variables which could evaluate those important features.

Furthermore, we evaluated the occurrence of adverse events in both individuals completing their regimen and still on treatment, for whom the cumulative drug toxicity (e.g. from linezolid) may be underestimated. Among patients who completed treatment, where the cumulative toxicity can be adequately assessed, the proportion of adverse events was (nonsignificantly) higher.

We did not collect any genetic/pharmacogenomic data, which could increase the risk of some adverse events. Future studies are needed to better clarify the role played by host and environmental characteristics in the occurrence of adverse events.

Finally, as the majority of countries started their aDSM project with this study, preselection or under-notification of adverse events (particularly minor ones and those not related to the new drugs) cannot be excluded. The under-reporting in a real-world setting can be a key issue in estimating the safety profile of a drug/pharmacological regimen. Healthcare workers and patients should be aware of the importance of reporting the occurrence of adverse events to better understand the pharmacological safety and the benefit/risk ratio of a prescription. A classification bias of some adverse events should be considered: although all clinical centres enrolled in the project followed the WHO protocol on adverse events’ reporting, local audits aimed at assessing the implementation of the standard operating procedures (e.g. regular audiometry) were not carried out because of financial constraints.

Unfortunately, several countries (in America, Asia and sub-Saharan Africa) declined when asked to participate, in view of the voluntary basis of the study perceived as “difficult” or “time-consuming” without provision for additional resources. For this reason, and because of the different entry time in the study (which works as a “register”), the study does not allow us to evaluate the prevalence of drug resistance in the different settings. There is an urgent need to overcome the administrative burden involved in reporting adverse event by easy-to-use e-forms that can be automatically compiled from medical records.

The study will continue to evaluate early and final treatment outcomes as periodic updates occur and the “cohort” is therefore a “living” one. This cohort allows evaluation of novel treatments and combinations in a relatively short time-frame; particularly important given the substantial variation in international practice and guidelines recommending person-centred therapy for MDR-TB [42, 43].

This approach will allow the participating countries to evaluate the “quality” of their treatment services and minimise the risk of post-treatment sequelae responsible of functional damage and impaired quality of life [44–46].

In conclusion, the study results confirm that aDSM for patients undergoing anti-TB regimens with new drugs is feasible. Furthermore, the study reaffirms the relative safety of new drugs recommended by the new WHO guidelines, as the occurrence of serious adverse events in this large cohort of patients from 26 countries was observed in <10% of patients. Greater adoption of the recommended aDSM at a local, national and international level is possible by improving the quality of the process (i.e. standardised, active and regular recording and reporting based on shared standard operating procedures).