Six months therapy for tuberculous meningitis

Description of the condition

Tuberculosis (TB) is caused by infection with one of several mycobacterial species that belong to the Mycobacterium tuberculosis complex. TB is estimated to be the leading infectious cause of death worldwide alongside HIV, and approximately one‐third of the world's population is thought to have latent TB. TB is associated with poverty. Low‐ and middle‐income countries carry the greatest burden of the disease, and more than three‐quarters of cases are reported in South‐East Asia, the Western Pacific, and Africa (WHO 2015). Tuberculous meningitis (TBM) is a severe form of TB, which affects the meninges that cover the brain and spine. TBM is the main form of TB that affects the central nervous system (CNS), which accounts for approximately 1% of all cases of active TB, and 5% to 10% of extrapulmonary TB cases (TB involving organs other than the lungs) (Thwaites 2009; Török 2015). TBM prevalence is higher in populations with high overall TB prevalence. The World Health Organization (WHO) reported 0.8 million new extrapulmonary TB cases worldwide in 2013 out of 5.4 million people with a first episode of TB (WHO 2014). TBM mainly affects children and immunosuppressed people (Principi 2012).

TBM is a major contributor to the death and disability associated with TB. Death occurs in around 30% of people with TBM and 50% of people who survive TBM suffer from neurological deficits that cause disability despite antituberculous treatment (ATT) (Principi 2012). Early diagnosis and prompt treatment with ATT and corticosteroids are the main determinants of outcome in TBM (Prasad 2016; Thwaites 2013).

TBM is diagnosed clinically by recognition of meningitis together with findings suggestive of M. tuberculosis infection (Principi 2012). A "bacteriologically confirmed TBM case" implies confirmation of the diagnosis by identification of M. tuberculosis in the person's cerebrospinal fluid (CSF) by microscopy, cell culture, or molecular methods. A case diagnosed as TBM by a health worker without laboratory confirmation is defined as a "clinically diagnosed TBM case" (WHO 2013). The disease severity at presentation is classified into three stages according to the British Medical Research Council (MRC) (MRC 1948), based on the Glasgow Coma Scale (GCS) (Teasdale 1974) and neurological signs.

• Stage I: non‐specific signs and symptoms but no neurological focal signs and GCS of 15.
• Stage II: minor neurological focal signs such as cranial nerve palsies or GCS of 11 to 14.
• Stage III: severe neurological deficits such as paresis or GCS equal to or less than 10.

Such classification facilitates comparison between findings from different studies and is useful to predict prognosis.

Description of the intervention

Without treatment, TBM leads to death. To effectively treat people with TBM, ATT must eliminate active TB bacilli to prevent neurological sequelae and death, eliminate dormant bacilli to prevent relapse, and prevent the emergence of drug resistance (Woodfield 2008). In contrast to pulmonary TB, there are no standardized international recommendations for treating people with TBM. There is a wide range of differing recommendations and practices for treating people with TBM worldwide (Appendix 1). This is partly due to the limited existing evidence regarding the optimal choice and dose of antituberculous drugs, as well as the most appropriate duration of treatment for people with TBM. Most data on treating TBM have been extrapolated from pulmonary TB (Heemskerk 2011).

Regarding treatment duration, two main concerns have led to the perception that treatment longer than the six months regimen for pulmonary TB is needed for people with TBM to ensure microbiological cure and prevent relapse. Firstly, the blood‐brain barrier hinders the penetration of antituberculous drugs to reach adequate drug concentration in the infected site, meaning that it may take longer to eradicate viable bacilli in the person's CNS (Thwaites 2013). The second argument concerns relapse rates. In 2013, 0.3 million cases of TB relapse after previous cure of the disease were reported (WHO 2014). Although relapse rates of 5% are generally considered acceptable for pulmonary TB regimens, this may not be the case for TBM, in which relapse can lead to severe neurodisability and death (Donald 2010a). Furthermore, reliable parameters for monitoring response to treatment in TBM are lacking, in contrast to pulmonary TB where sputum microscopy and culture are used. Consequently, some specialists recommend 12 months treatment for drug‐sensitive TBM, consisting of a two months intensive phase with isoniazid, rifampicin, pyrazinamide, and either ethambutol or streptomycin, followed by a 10 months continuation phase with isoniazid and rifampicin (Török 2015).

All international guidelines, including the WHO guidelines (WHO 2010a), recommend the use of isoniazid, rifampicin, and pyrazinamide, usually with a fourth drug such as ethambutol or streptomycin, as first‐line antituberculous drugs in people with TBM. Isoniazid is highly active against rapidly replicating mycobacteria and has good CSF penetration. Rifampicin kills both rapidly replicating and slow or non‐replicating bacilli, while pyrazinamide has strong bactericidal effect against intracellular mycobacteria. Thus, both rifampicin and pyrazinamide are critical drugs to sterilize lesions by killing dormant bacteria and those in remote sites such as the CSF, which is essential to reduce the risk of relapse (Donald 2010b; Thwaites 2005a). The introduction of rifampicin allowed the shortening of pulmonary TB treatment (Zumla 2014). Although rifampicin has poor CSF penetration, its importance in TBM treatment has being emphasized in a trial in which mortality rate was higher among participants with rifampicin‐resistant TBM (Thwaites 2005b).

How the intervention might work

There is uncertainty about the effectiveness of short‐course regimens of six months ATT in people with TBM. Some study authors report that six months were adequate for treating TBM (Donald 1998; van Toorn 2014), while prolonged regimens have disadvantages. Prolonged regimens may increase the risk of drug‐related adverse events and are associated with poor adherence to treatment and default. Poor adherence increases the risk of relapse leading to potential neurological disability and death, and facilitates the development of drug‐resistant bacterial strains (van Loenhout‐Rooyackers 2001). Finally, long regimens lead to greater expenditure of resources and greater workload for TB control programmes. Short‐course regimens may not be long enough to eliminate dormant bacilli from the CNS, and may lead to higher relapse rates. Higher relapse rates may increase neurodisability and death after completion of ATT (Figure 1).

Figure 1

Possible disadvantages of prolonged and short‐course antituberculous regimens for treating people with TBM.

Relapse is the most daunting possible consequence of using short treatment regimens. There is uncertainty around the lag time between completion of treatment and relapse. However, most relapses seem to occur within the first six months after completing ATT (Jullien 2015). Thus, at least six months of follow‐up are required to adequately assess this outcome, and longer follow‐up periods are desirable for detecting all cases of relapse. Defaulters may be at increased risk of relapse due to inadequate dosing with ATT, particularly during the intensive phase of treatment, although direct evidence for this association is lacking in people with TBM (Conde 2009). The number of deaths and neurological deficits are reduced by early diagnosis and prompt ATT. Indeed, most of the deaths associated with TBM will occur during the first weeks of disease (Yaramiş 1998). In preliminary work for this Cochrane Review, Jullien 2015 attributed deaths occurring after six months of treatment mainly to complications of TBM and comorbidity, such as secondary infection. Few deaths were associated with relapse. Finally, longer regimens expose patients to the risk of adverse effects of antituberculous drugs for longer.

Why it is important to do this review

The most appropriate duration of treatment for people with TBM is uncertain. Although several studies have reported safe outcomes in adults and children with TBM treated with six months regimens (Donald 1998; van Toorn 2014), there are concerns about whether six months regimens would be sufficient to sterilize CSF and thus prevent relapse. Most international guidelines recommend prolonged‐course ATT of nine to 12 months for people with drug‐sensitive TBM. In practice, more than four antituberculous drugs are sometimes used, and regimens may be prolonged up to 24 months for fear of relapse. Treatment is also prolonged in people for whom supervision and follow‐up are doubtful (Principi 2012). On the one hand, TBM treatment duration must be long enough to eliminate dormant bacilli to prevent relapse. On the other hand, treatment should be as short as possible to avoid disadvantages related to prolonged regimens. Reducing the duration of TBM therapy from prolonged‐course regimens to short‐course regimens could have several benefits, including reduced drug toxicity and improved adherence to treatment, with fewer consequent relapses and lower risk of developing drug‐resistant strains. Additionally, short‐course regimens would reduce the use of resources and workloads in TB control programmes, which are often overburdened in settings where TB incidence is high (Conde 2009; van Loenhout‐Rooyackers 2001).

A study that reviewed treatment duration for people with TBM by comparing case series of adults and children, showed similar completion and relapse rates between six months regimens including at least isoniazid, rifampicin, and pyrazinamide and longer regimens (van Loenhout‐Rooyackers 2001). In this review, we have included more recent studies, and performed 'Risk of bias' assessments and appraised the quality of the evidence.

This review was prompted as part of work with the Central TB Division, Ministry of Health and Family Welfare in India, and the All India Institute of Medical Sciences in New Delhi in preparation for the Indian Extra‐Pulmonary TB (INDEX‐TB) guidelines (INDEX‐TB 2016). There was uncertainty and debate as to whether six to nine months of treatment is as safe as longer regimens for treating people with TBM (Jullien 2015), with varying views on what the literature showed. As we knew there were no randomized controlled trials (RCTs), we sought to summarize and critically appraise evidence from observational studies. The initial systematic review found several observational studies that reported low numbers of relapse with short regimens, but did not include all the available evidence for longer regimens. In this Cochrane Review, we examined the effects of six months (short‐course) and longer than six months (prolonged‐course) regimens on TBM outcomes.

Criteria for considering studies for this review

Search methods for identification of studies

We attempted to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and ongoing).

Data collection and analysis

Description of studies

Results of the search

We conducted the literature search up to the 31 March 2016 and identified 1541 studies. We identified five additional records through other resources. By screening titles, abstracts, and keywords, we selected 154 records and attempted to retrieve the full‐text articles of these studies. By contacting the first author of one of these potentially eligible studies for clarification, we were granted access to the database of an ongoing prospective observational cohort of participants with tuberculous meningitis (TBM), which met our inclusion criteria (Alvarez‐Uria 2012; NCT02454569). Although some of the participants included in our analysis have also been presented in Alvarez‐Uria 2013 and Alvarez‐Uria 2015, all the data we present in this review come from the larger database provided by the study author. Overall, 18 studies presented in 21 records met the inclusion criteria of this review (see the 'Characteristics of included studies' table). We had insufficient information to decide inclusion or exclusion of 11 studies (see the 'Characteristics of studies awaiting classification' table). We presented the reasons for excluding the remaining 122 reports in the 'Characteristics of excluded studies' table. We illustrated this process in the study flow diagram (Figure 2).

Figure 2

Study flow diagram.

Risk of bias in included studies

See Figure 3 for a summary of the 'Risk of bias' assessment of both group of cohorts receiving six months and more than six months of ATT. The 'Risk of bias' tables provide further details for supporting evidence in the 'Characteristics of included studies' table.

Figure 3

'Risk of bias' summary for the cohorts treated for six months (left) and the cohorts treated for longer than six months.

Effects of interventions

See Figure 4 for the 'Summary of findings' table. We summarized the findings of the Alvarez‐Uria 2012 cohort separately in Table 7.

Figure 4

'Summary of findings' table.

We have provided more details in Table 8 and Table 9 for cohorts treated for six months, and in Table 10 and Table 11 for cohorts treated for longer than six months.

Summary of main results

The seven cohorts treated for six months included 458 participants from Thailand, South Africa, Ecuador, and Papua New Guinea, and only one study was conducted after the year 2000. The 12 cohorts treated for longer (eight to 16 months) included 1423 participants from India, Thailand, South Africa, China, Romania, Turkey, and Vietnam, with four conducted between 2001 and 2011. The proportion of participants classified as having stage III disease (severe) was higher in the cohorts treated for six months, but the proportion with known concurrent HIV was higher in the cohorts that were treated for longer. The diagnosis was confirmed on culture in 56/254 (22%) participants treated for six months versus 470/1264 (37%) treated for longer (in studies where this was clearly stated). Although there were variations in the treatment regimens, most cohorts received isoniazid, rifampicin, and pyrazinamide during the intensive phase.

Three out of the seven cohorts treated for six months and five of the 12 cohorts treated for longer achieved follow‐up beyond 18 months after completing treatment. All studies had potential sources of bias in their estimation of the relapse rate, and comparisons between the cohorts could be confounded.

Relapse was an uncommon event in both groups of cohorts of participants treated for six months and longer than six months, and only one death was attributed to relapse in each group.

Overall, the proportion of participants who died was higher in the cohorts treated for longer than six months, but most deaths occurred during the first six months regardless of treatment duration. This difference is not due to the treatment duration, but rather indicates that there is likely to be substantial clinical heterogeneity between the six months cohorts and the longer treatment cohorts. The disease severity is unlikely to be the main cause of this difference in the mortality rate in these cohorts, as the proportion of participants with severe tuberculous meningitis (TBM) (grade III) was higher in the cohorts treated for six months. The reason for the higher mortality rate in the group of cohorts treated for longer than six months is likely multifactorial. Delay in starting antituberculous treatment (ATT), drug resistance, drug‐induced hepatitis, adherence to ATT, quality of supportive care, HIV status, and drug regimens could all contribute to this difference between the groups of cohorts.

As there was a higher proportion of deaths during ATT in the longer treatment cohorts, the proportion of participants that achieved clinical cure was higher in the group of cohorts treated for six months.

Few participants defaulted from treatment in both groups of cohorts, and adherence was poorly reported.

Most of the study authors reported on adverse events, but the level of detail in reporting varied significantly, and we were unable to draw any clear conclusions about rates of adverse events between the two groups of cohorts.

The main difference between the Alvarez‐Uria 2012 cohort and the rest of the included cohorts was that all participants in the Alvarez‐Uria 2012 trial were HIV‐positive, and that they were managed in a "real‐world" setting, contrary to the other cohorts in which participants were under trial or study conditions. The relapse rate was higher, with no cases among the 18 survivors who completed six months of ATT, and seven cases among the 71 survivors treated for longer than six months. In this cohort, it would be misleading to compare participants treated for six and more than six months, as we do not know whether the baseline characteristics of the participants were comparable. The mortality rate was high (131/217, 60.4%), and 71.8% of the deaths occurred during the first six months of treatment, similarly to the rest of the cohorts included in the review. The proportion of defaulters was higher than in the rest of the included studies (13.8%). This could represent a more accurate picture of the real proportion of defaulters among HIV‐positive people in an operational setting who must take a high number of prescribed drugs.

Overall completeness and applicability of evidence

This Cochrane review includes adults and children from a variety of geographical settings, including countries in South‐East Asia and Africa with a high tuberculosis (TB) burden (WHO 2015).

Only one cohort in the group of studies treating for six months reported the HIV status of the participants, and in this study HIV‐positive people were all treated for nine months (van Toorn 2014). Five cohorts where participants received more than six months mentioned HIV status: two cohorts explicitly excluded HIV‐positive people (Iype 2014; Sharma 2013a), and three cohorts included 122/376 HIV‐positive people (Lau 2005; Török 2011a; van Toorn 2014). In addition, the cohort from Alvarez‐Uria 2012 reported 217 participants co‐infected with HIV. Further evidence relating to the duration of ATT in HIV‐positive people with TBM is needed.

Six of the seven cohorts treated for six months and 11 of the 12 cohorts treated for longer than six months presented the three stages of clinical severity, with around half of participants in stage II in all cohorts. Results from this review are therefore applicable for all stages of clinical severity of the disease at presentation.

In this review, we have restricted the inclusion of studies to those that included ATT regimens that contained rifampicin. All six months course regimens contained at least isoniazid, rifampicin, and pyrazinamide during the intensive phase, and 1250/1423 participants treated for longer also received these three drugs, which reflect the current first‐line regimens. The range of single daily doses used is mostly concordant with current recommendations.

We also restricted study inclusion to studies that reported participants with drug‐sensitive TBM. van Toorn 2014 excluded children with multidrug‐resistant TBM. However, this study included those with isoniazid‐monoresistance. For these participants, the planned treatment consisted in a fluoroquinolone and terizidone in addition to the standard ATT for nine months. We decided to include this study as data on isoniazid‐monoresistant cases (three children) were disaggregated. Across all studies, most cases did not have a M. tuberculosis‐positive culture. Thus, the drug‐sensitivity pattern was unavailable and we cannot be sure that some cases were not caused by drug‐resistant strains. However, if any, this would have affected few cases, as prevalence of drug‐resistant strains was lower in the period when most of the studies were conducted. Hence, results from this review are not applicable to drug‐resistant TBM.

Adjunctive treatment with corticosteroids is currently recommended in all cases of TBM to reduce death (Prasad 2016). Most of the recruited participants received corticosteroids, which means that applicability of the results is unlikely to be affected by this co‐intervention.

Quality of the evidence

We derived the results from single‐arm studies and not direct randomized comparisons, although some were arms within randomized controlled trials (RCTs) performed for other purposes. As such, participant characteristics, comorbidities, the severity of the disease, differences in ATT regimens, drug resistance, drug‐related adverse events, quality of supportive care, and method of follow‐up could be different between the two groups, and making causal inferences on the basis of potentially confounded cohorts is problematic. Because of this substantial clinical and methodological heterogeneity between the studies, we did not attempt to perform a meta‐analysis or perform a Grading of Recommendations, Assessment, Development and Evaluation (GRADE) assessment. Nevertheless, relapse rates were low after treatment completion in both groups.

Potential biases in the review process

We attempted to limit bias by following the rigorous methods provided by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The Cochrane Infectious Diseases Group Information Specialist performed the literature search without language restrictions, and as such it is unlikely that we missed any large studies. We attempted to identify unpublished studies by contacting relevant researchers in this area, but only one study author replied. Thus, we cannot rule out the possibility that we missed some unpublished data. Also, we were unable to assess publication bias because no study directly compares six versus longer than six months of ATT. We attempted to limit bias in the selection process, data extraction, and assessment of the quality of the included studies by conducting these processes independently and by comparing results between at least two review authors.

Agreements and disagreements with other studies or reviews

The findings of this Cochrane Review are similar to those of a systematic review conducted by van Loenhout‐Rooyackers 2001, who concluded, based on the results of 11 included studies, that six months regimens with at least isoniazid, rifampicin, and pyrazinamide were sufficient for participants with fully drug sensitive TBM. This review was similar to our review in that the included studies were all non‐randomized observational studies. Completion of treatment was 81% in participants that received six months regimens and 85% in those that received longer regimens. There were 2/131 and 0/591 participants with relapse in the six months regimen group and longer ATT group respectively. Another systematic review, which belongs to a series that addressed the scientific evidence behind the World Health Organization (WHO)’s recommendations, reviewed the most appropriate ATT for TBM and drew similar conclusions regarding length of treatment, based on four prospective and retrospective observational cohorts (Woodfield 2008).

Maria‐Inti Metzendorf, 8 February 2017