3Composition of second-line anti-tuberculosis regimens

Recommendations

3.1.

In the treatment of patients with MDR-TB, a fluoroquinolone should be used (strong recommendation, ⊕○○○/very low quality evidence).

3.2.

In the treatment of patients with MDR-TB, a later-generation fluoroquinolone rather than an earlier-generation fluoroquinolone should be used (conditional recommendation, ⊕○○○/very low quality evidence).

3.3.

In the treatment of patients with MDR-TB, ethionamide (or prothionamide) should be used (strong recommendation, ⊕○○○/very low quality evidence).

3.4.

In the treatment of patients with MDR-TB, four second-line anti-tuberculosis drugs likely to be effective (including a parenteral agent), as well as pyrazinamide, should be included in the intensive phase³(conditional recommendation,⊕○○○/very low quality evidence).

3.5.

In the treatment of patients with MDR-TB, regimens should include at least pyrazinamide, a fluoroquinolone, a parenteral agent, ethionamide (or prothionamide), and either cycloserine or PAS (p-aminosalicylic acid) if cycloserine cannot be used (conditional recommendation, ⊕○○○/very low quality evidence).

Evidence

The evidence used to address the questions on which drugs to include (with or without information on their DST patterns) and the number of drugs to be used in regimens for MDR-TB patients was based on studies published in three major systematic reviews (27–29). All three reviews searched EMBASE and MEDLINE databases as well as the Cochrane Library and the ISI Web of Science. Studies published before 1970 and those including only XDR-TB cases were excluded. The reviewers then pooled individual patient data from studies which had featured in the systematic reviews for a meta-analysis.

The meta-analysis included 32 studies with more than 9000 treatment episodes for which the authors could be contacted and were willing to share their data (30). Patients with XDR-TB (N=410) were excluded, as their treatment regimens were considered not to be comparable with those of other MDR-TB patients. Cohorts included had to have had at least 25 subjects treated for MDR-TB, and one or more of the treatment outcomes meeting the standard definitions (31). Missing values for age, sex, past TB, extent of disease, HIV infection and DST were imputed (>50% of cohort members having an observed value for these variables), but not those for treatment modality or outcome. None of the cohorts was part of randomized controlled trials and thus the quality of evidence was judged to be low or very low. While the odds ratios in the analysis were adjusted for age, sex, HIV-serostatus, past TB treatment, past MDR-TB treatment and extent of disease, there remains a risk of substantial bias (certain drugs may have only been used for sicker patients). Other limitations included incomplete ascertainment of relapse, the under-representation of certain geographical regions, and missing data for some of the variables examined.

Findings from this analysis may not necessarily be generalizeable to all populations in settings with high or low prevalences of drug resistance or different levels of resources. Nonetheless, the results of this analysis represented the best available evidence to date for the group to make recommendations on the composition of treatment regimens.

Summary of findings

Use of drugs to which the strain was reportedly susceptible showed a marginal benefit when compared with their use regardless of susceptibility patterns. Choice of drug would thus depend on the DST of the strain isolated from the patient or close contacts with MDR-TB, previous use of the drug in the patient, and the frequency of its use or documented background drug resistance in the setting. In applying this observation to clinical practice, it is important to underline the uncertainties around the reproducibility and reliability of DST of pyrazinamide (and ethambutol) (32), as well as the second-line drugs other than the parenteral agents and the fluoroquinolones (33).

The analysis showed that in the intensive phase, a regimen with at least four drugs likely to be effective, when adjusted for clinical covariates, all other drugs used concomitantly as well as the total number of susceptible drugs used throughout treatment, was associated with a statistically significant peak in cure with a plateau thereafter.

Data from this analysis did not reveal any second-line parenteral agent – kanamycin, amikacin or capreomycin – to be superior in effect to any other. Given its lower cost, kanamycin would be preferred. Amikacin can be used instead of kanamycin. In an analysis comparing patients who were cured or completed treatment with those who failed or relapsed, capreomycin was shown to be effective if the case was resistant to kanamycin. The use of streptomycin in MDR-TB patients is not recommended.

Fluoroquinolones were significantly associated with cure and this effect was more pronounced in later-generation fluoroquinolones (see Background and methodology for definition). It was highest when used against strains known to be susceptible. Fluoroquinolones should therefore always be used unless there is an important contraindication. Ciprofloxacin, even if it may have some anti-tuberculosis activity, should not be used (34).

Among the oral bacteriostatic drugs, the association with cure was higher with ethionamide than with cycloserine, which was higher than with PAS. Ethionamide or prothionamide should therefore always be included in a regimen unless there is a particular contraindication. Ethionamide showed little effect in patients who had taken prior treatment for MDR-TB. PAS performed the worst, showing no significant effectiveness in the main analysis. Its use would thus be recommended only if an additional drug is needed to achieve a five-drug regimen or if ethionamide or cycloserine cannot be used or are unlikely to be effective. The data did not allow comparison of outcomes between once daily PAS and divided doses, or the formulation of PAS: decisions on how to administer PAS should thus rely on a balance between its tolerance in the patient and the resources available to observe doses.

Patients on Group 5 drugs were observed to have worse outcomes, an effect largely attributed to confounding. When the individual effect of amoxicillin/clavulanate, clofazimine, macrolides⁴ and thioacetazone was analysed, no significant association with cure could be discerned. No separate analysis was possible for linezolid and high-dose isoniazid given the small number of cases treated with these agents.

Pyrazinamide showed a slightly added benefit in one of the analyses in which adjustment was made for other medication used concomitantly. Ethambutol was associated with a marginal but statistically significant reduction in likelihood of cure among patients not previously treated for MDR-TB. As in the case of Group 5 drugs this effect was attributed to confounding rather than a detrimental effect of ethambutol.

The analysis of data from this review bore inconclusive results about the contribution of ethambutol and Group 5 drugs in the treatment of MDR-TB patients and as a result they have not been included among the drugs making up the recommended standard MDR-TB regimen.

The principle of using additional drugs for extensive disease could not be supported by the data used for this review.

As patients with XDR-TB were excluded from the analysis, the current recommendations do not necessarily apply to this subgroup of patients. Until better evidence is available to determine the optimal regimens for treatment of these patients, the same principles used to design MDR-TB regimens should be used, based where possible on the DST pattern of the individual patient, particularly for later-generation fluoroquinolones and second-line parenteral agents. All MDR-TB patients should thus be tested for susceptibility to these two classes of drugs.

The recommended composition of second-line regimens for MDR-TB patients has changed from those in the 2008 emergency update (3) (Table 7). The previous guidelines had likewise recommended designing regimens based on known drug resistance patterns in the country or patient, the history of previous treatment by the patient, and the drugs commonly used in the country. The inclusion of at least four drugs with either certain, or almost certain, effectiveness was previously recommended. The previous recommended regimen was composed of pyrazinamide and/or ethambutol, one fluoroquinolone, one parenteral agent and second-line oral bacteriostatic drugs. Resort to antibiotics from Group 5 was only recommended if additional drugs were needed to bring the total to four. More drugs were recommended in the case of extensive disease or uncertain effectiveness.

Table 7

Changes to the recommendations on regimen composition between the 2008 and 2011 updates of the guidelines.

Benefits

The recommendations contained in this section aim to increase the likelihood of cure and reduce the risk of failure, relapse and death. The decision to recommend an additional drug to the regimen during the intensive phase of treatment – from the minimum of four inferred from the analysis – was based on expert opinion. It is intended to safeguard against the acquisition of additional resistance, particularly in the case of undetected primary resistance to the four drugs considered to be effective given the unreliable nature of DST for drugs other than parenteral agents and fluoroquinolones. Estimates of effects for fluoroquinolones were probably conservative given that patients treated with ciprofloxacin were included in the control group. Studies of the inhA promoter region mutation, although not assessed in this review, may guide treatment by identifying strains that are resistant to ethionamide (35) although the additional costs need to be considered.

Risks

A slight incremental trend in serious adverse events (SAE) was discerned as the number of drugs in the continuation phase increased from two to five. About 14% of patients on oral bacteriostatic drugs had SAE, while for the other drugs evaluated this was much lower (1–6%). An association between the total number of drugs used and the risk of SAE was observed. This association was not observed during the intensive phase.

The risk of additional acquisition of resistance is a concern in cases of unrecognized resistance to some of the drugs used. The long-term potential for SAE, particularly in children and for the later-generation fluoroquinolones, remains unknown. However, a Cochrane review assessing fluoroquinolones as additional or substitute drugs in regimens for drug-sensitive and drug-resistant patients found that substituting or adding fluoroquinolones to a regimen had no demonstrable effect on the occurrence of SAE (34).

Values and preferences

A high value was placed on preventing death and transmission of MDR-TB and a lower value on the potential for SAE resulting from long-term treatment. As a result, the long-term use of fluoroquinolones was considered to outweigh the higher cost and any possible long-term SAE. The recommendation is thus strong. While the use of later-generation fluoroquinolones is generally preferred, a separate recommendation on their use was graded conditional rather than strong because there is uncertainty about the risk of SAE from the long-term use of these agents.

Footnotes

3

The intensive phase is the initial part of a course of treatment during which a parenteral (injectable) agent is used.

4

Azithromycin, clarithromycin and roxithromycin were included in this analysis.