Following M. tuberculosis infection, young children are at high risk of developing the most severe forms of the disease, of which the most devastating form is TBM. This predominantly affects young children with a peak age of onset of 2-4 years (2). Up to 15% of childhood TB may present as TBM (62); with a decreasing incidence of bacterial meningitis attributed to other causes, TB is the leading cause of bacterial meningitis in many settings (63). TBM is associated with significant mortality and morbidity; in a systematic review and meta-analysis published in 2014, the risk of death among children aged 0-14 years with TBM was estimated at 19.3% and the risk of neurological sequelae among TBM survivors was estimated at 36.7% (64). Even among children without severe neurological sequelae, attention deficit and behavioural disorders are common and the human cost and financial burden for families and society are high. Diagnosis in the most advanced clinical stage, which occurred in almost 50% of children with TBM, was associated with worse outcomes based on the findings of the review. WHO currently recommends a regimen of 12 months duration to treat TBM, consisting of isoniazid, rifampicin, ethambutol and pyrazinamide given daily for the first 2 months, followed by isoniazid and rifampicin given daily for 10 additional months (2HRZE/10HR) (8, 56). Recommended doses to be used in this regimen are the same as those for the treatment of PTB (8).
The recommendation on the use of the 12-month regimen was based on a literature review (65) and was first included in the 2010 Rapid advice: treatment of TB in children (56). This review included 46 studies with information on the efficacy of different regimens and dosages for the management of TBM (25 studies included paediatric data and 21 included data on both adults and children). The majority of these studies were non-randomized, non-comparative studies. The quality of the studies ranged from low to very low and no clear conclusions could be drawn from the efficacy studies, given that they differed widely in terms of design, drugs used and patient populations. None of the studies from this review were entered into GRADE, given the lack of comparative data.
Shorter regimens to treat paediatric TBM are used in some settings. The South African National TB Guidelines recommend a regimen composed of daily isoniazid, rifampicin, pyrazinamide and ethionamide, given for 6 months (6HRZEto) (66). The recommendations in South Africa are based on expert opinion, in particular with regards to replacing ethambutol with a drug (ethionamide) that has more efficient blood-brain barrier penetration (65). A clinical trial is underway to compare a regimen consisting of higher-dose isoniazid, higher-dose rifampicin, pyrazinamide and levofloxacin given daily for 6 months, compared to the WHO-recommended regimen, but the results are not expected until 2023 (SURE trial, ISRCTN40829906). Shorter regimens may be as effective and safe, and may improve adherence and reduce the burden on persons with TBM and health care systems, but it is not known how outcomes compare between the shorter regimens and the WHO-recommended 12-month regimen.
PICO question: In children and adolescents with presumed or bacteriologically confirmed drug-susceptible TB meningitis, should a 6-month intensive regimen, compared to the 12-month regimen that conforms to WHO guidelines be used?
Evidence: To inform recommendations on the treatment for child and adolescent TBM, a systematic review and meta-analysis was conducted to compare the effectiveness of shorter regimens versus the current WHO-recommended 12-month regimen (8). The primary intervention of interest was the regimen currently used in South Africa (66); in secondary analyses, outcomes associated with other shorter regimens were examined. The search criteria from an earlier systematic review and meta-analysis conducted in 2014 were updated and the search was run in February 2021 (64). Studies with available information on at least the composition and duration of treatment regimens were included. Ineligible regimens included those without rifampicin, treatment regimens with a duration over 12 months other than the WHO regimen, intermittent regimens and non-intensive short regimens. Pooled proportions were estimated across studies and within regimens through aggregate-level meta-analysis using GLMM with Gauss-Hermite quadrature for the following outcomes: death by end of treatment; loss to follow-up; treatment success; neurological sequelae (among survivors); and survival without neurological sequelae (among those starting treatment). Subgroup analyses were planned but were not feasible, due to insufficient data.
Of the 1820 unique citations screened, 149 full text papers were evaluated. Of these, five met inclusion criteria for the systematic review. In addition, two unpublished cohorts were eligible for inclusion. In total, four studies of intervention regimens (three published and one unpublished) (67-70), and three studies of the comparator regimen (two published and one unpublished) (71-73) were identified. No studies performing head-to-head comparisons of regimens were identified. Three of the four studies of intervention regimens were conducted in a single referral centre in South Africa. As only one study reported on outcomes from an 8-month regimen in Vietnam (69), it was excluded from the meta-analysis. The two published studies of the comparator regimen were conducted at different sites in India (71, 72), while data for the unpublished study were collected in various centres in Europe through the Pediatric Tuberculosis Network European Trials Group (ptbnet) (73). A total of 837 patients with TBM received intervention regimens with the median age in each study ranging from 2.3 to 5.5 years (age range: 2 months to 15 years). Among the 282 patients treated with the comparator regimen, the median age in the European study was 3.3 years, while summary data on age were not reported in the studies from India.
The cumulative number of deaths was recorded at the end of treatment for each regimen (i.e. at 6 months after treatment initiation in studies of the intervention regimen; at 12 months after treatment initiation in studies of the comparator regimen). Among studies of the intervention regimen, 0.0%-9.6% of patients died within six months; most deaths occurred early after hospital admission and primarily involved patients diagnosed in stage 3 at baseline. Among studies of the comparator regimen (i.e. the standard of care), 7.1%-30.0% of patients died. In one of these studies, diagnosis at stage 3 was most strongly associated with mortality, although stage-specific outcome data were not reported. In a random effects meta-analysis, the pooled proportions of deaths were 6.0% (95% confidence interval [CI] 2.0-13.0) and 24.0% (95% CI: 18.0-32.0) for children and adolescents who received the intervention and comparator regimens, respectively. The percentage of patients with treatment success were 78.5%-100.0%, with a random effects pooled proportion of 95.0% (95% CI: 74.0-99.0) among studies of the intervention regimen, and 70%-85.7%, with a random effects pooled proportion of 75.0% (95% CI: 69.0-81.0) among studies of the comparator regimen. Neurological sequelae were defined and assessed differently across studies. Among patients treated with the intervention regimen, 50.0%-66.7% had neurological sequelae, mostly categorized as mild. The vast majority of these were among patients diagnosed at stage 2 or 3. Among patients treated with the comparator regimen, 31.9%-50.0% had neurological sequelae. In one study on the comparator regimen from India, 17 of 29 (58.6%) cases were categorized as mild. The random effects pooled proportions of neurological sequelae among survivors were 66.0% (95% CI: 55.0-75.0) and 36.0% (95% CI: 30.0-43.0) for the intervention and comparator regimens, respectively. Among the 135 HIV-negative and 13 HIV-positive patients who received the intervention regimen in South Africa, none relapsed within two years of post-treatment. Other studies did not report on relapse.
Because of the non-comparative nature of the studies, which were all observational, narrative descriptions reporting pooled proportions were provided rather than estimated measures of effect. The certainty of evidence was deemed to be very low for all outcomes due to very serious risk of bias, serious or very serious inconsistency within regimens, and very serious indirectness. Imprecision could not be assessed due to the lack of comparative data.
The reviewers concluded that pooled estimates need to be interpreted with caution considering the small number of studies, the potential for confounding by indication, other potential residual confounding and between-study heterogeneity regarding the assessment of neurological sequelae.
GDG considerations: While the GDG members discussed evidence of benefits of the intervention regimen and biological plausibility, they noted the very low certainty of the data. The GDG members decided to choose that the balance of effects 'does not favour either the intervention or the comparison'. This decision reflects that both the desirable and undesirable effects varied, as well as the very low certainty of the evidence overall. Recognizing how quickly the condition of children with TBM can deteriorate, GDG members were not comfortable favouring one regimen over the other, based on the very low certainty of the data.
The GDG members also noted that the feasibility of introducing the shorter intensive regimen is dependent on the setting. Acceptability, affordability and access to the component medicines (including the child-friendly ethionamide formulation, a 125 mg dispersible tablet) are important contextual factors as well as any additional implementation considerations such as the necessity and availability of monitoring tests that are needed with the shorter regimen. GDG members thought that the short intensive regimen would probably be acceptable as it includes medicines that have been in use for many years, including first line medicines and ethionamide. The intervention regimen is more costly than the comparator regimen, but refers to the costs of medicines only; GDG members stated that other costs (including to patients, families and the health system) were important.
The GDG acknowledged the limited data on alternative regimens to treat TBM and the need for continued efforts to optimize and better understand treatment options for TBM, including implementation considerations, such as dosing.