The investigators will conduct a randomized, double blind, placebo controlled trial of adjunctive dexamethasone in the initial (6-8 weeks) treatment of tuberculous meningitis in Vietnamese adults. The trial will address a primary hypothesis in all enrolled patients, and a secondary hypothesis in a sub-group of enrolled patients who develop anti-tuberculosis drug-induced liver injury (DILI). The primary hypothesis is adjunctive dexamethasone increases survival from TBM in HIV co-infected adults. The secondary hypothesis is current guidelines for the management of anti-tuberculosis drug-induced liver injury in those with TBM result in the premature interruption of rifampicin and isoniazid (the critical active drugs in early therapy) and are thereby placing participants at risk of poor outcomes.

Mycobacterium tuberculosis causes ~9 million new cases of tuberculosis and ~1.5 million deaths annually, around 0.4 million of whom are co-infected with HIV. Tuberculous meningitis (TBM) is the most severe form of tuberculosis, killing around 30% of all sufferers despite appropriate anti-tuberculosis chemotherapy. It is especially common in young children, and in those infected with HIV.

There is a longstanding hypothesis that death from TBM results from an excessive intracerebral inflammatory response. The corollary of this hypothesis has been that adjunctive anti-inflammatory treatment with corticosteroids (e.g. dexamethasone) improves survival, which has been demonstrated in predominantly HIV-uninfected individuals in a small number of trials. Yet how corticosteroids improve survival, and whether they do so in HIV-infected patients, remains uncertain. The primary objective of this trial is to determine whether or not adjunctive corticosteroids reduce deaths from TBM in HIV-infected adults.

Adjunctive dexamethasone might improve outcomes from HIV-associated TBM by diverse mechanisms. First, it may control the early intracerebral inflammatory response, reducing cerebral oedema and intra-cranial pressure. Second, it may prevent the potentially life-threatening complications of hydrocephalus, infarction and tuberculoma formation. Third, it may prevent the incidence of anti-retroviral (ARV) treatment-associated neurological immune reconstitution inflammatory syndrome (IRIS). Finally, dexamethasone may help reduce the risk of drug-induced liver injury and thereby improve outcome by enabling uninterrupted anti-tuberculosis treatment.

The current evidence-base for using adjunctive corticosteroids for the treatment of HIV-associated TBM is restricted to 98 adults recruited to a trial in Vietnam published in 2004. This trial randomized a total of 545 subjects (98 of them HIV-positive) and reported an overall reduction in 9-month mortality due to dexamethasone from 41.3% (112/271) to 31.8% (87/274) (hazard ratio of time to death 0.69; 95% CI 0.52-0.92, P=0.01). While there was no clear evidence of treatment effect heterogeneity according to HIV status, the number of included HIV-infected subjects was low and the observed benefit in that subgroup was smaller: 61.4% (27/44) in the dexamethasone group died, compared to 68.5% (37/54) in the placebo group (hazard ratio of time to death 0.86; 95% CI 0.52-1.41; P=0.55).

There are limited data from HIV-infected patients with TBM treated with dexamethasone, but findings from studies using corticosteroids in HIV-infected individuals with other forms of tuberculosis and other opportunistic infections suggest corticosteroids may cause harm in those with advanced HIV infection. There is evidence that corticosteroids may increase the risk of HIV-associated malignancies, especially Kaposi sarcoma. Furthermore, a recent trial of adjunctive dexamethasone for HIV-associated cryptococcal meningitis performed in Southeast Asia and Africa found dexamethasone was associated with worse outcomes, with increased risk of secondary infections, hyperglycaemia and electrolyte abnormalities, and disability.

On the basis of these limited data most international guidelines cautiously recommend dexamethasone should be given for HIV-associated TBM, but all acknowledge the paucity of evidence and the need for additional controlled trial data. Our trial will meet the need for more data and aims to provide definitive evidence as to the risk/benefit of adjunctive dexamethasone in the treatment of this important and very severe disease.

Our secondary objective is to investigate alternative management strategies in a subset of patients who develop drug-induced liver injury that will enable the safe continuation of rifampicin and isoniazid therapy whenever possible. The investigators will perform an open, randomised comparison of three management strategies with the aim of demonstrating which strategy results in the least interruption in R and H treatment. All patients enrolled in the trial will be eligible to take part in this study, with the exception of those known to have TBM caused by isoniazid resistant or MDR M. tuberculosis. Consent will be sought at enrolment, with an option given to patients to enrol in the main study, but not the 'drug-induced liver injury strategy study'.

Eligible patients will be randomised to one of three strategies:

1. Observe: measure transaminases, bilirubin, and INR every 3 days; do not change/stop anti-tuberculosis drugs unless transaminases rise to ≥10x normal, or total bilirubin rises >2.0mg/dl (>34 µmol/L), or INR >1.5 or symptoms of hepatitis worsen (nausea, vomiting, abdominal pain), in which case go to Strategy 3.
2. Stop Pyrazinamide (Z) alone. Observe, measuring transaminases, bilirubin, and INR every 3 days. If transaminases do not fall to < 5x ULN by day 5, or total bilirubin rises >2.0mg/dl (>34 µmol/L), or INR >1.5 or symptoms of hepatitis worsen at any time (nausea, vomiting, abdominal pain), go to Strategy 3.
3. Current standard of care (the current USA CDC guidelines): stop rifampicin (R), isoniazid (H) and Z immediately and add levofloxacin and an aminoglycoside to ethambutol. Restart R (at full dose) once transaminases are <2X ULN and no hepatitis symptoms. If no increase in transaminases after 7 days add isoniazid (at full dose) and stop levofloxacin and aminoglycoside. If transaminases remain normal on full dose R and H, Z was the likely cause and it should not be re-started and treatment duration should be extended to ≥12 months. If transaminases rise ≥ 5x ULN, or ≥3x ULN with symptoms, at any time after re-introduction of R and/or H the physician should stop R and/or H (depending on which was associated with the transaminase rise). If neither R or H can be used, treat with levofloxacin, an aminoglycoside and ethambutol. If R can be used, but not H, treat with R, levofloxacin and ethambutol. If H can be used, but not R, treat with H, levofloxacin and ethambutol.

The primary endpoint is the proportion of time in the 60 days following randomisation during which neither rifampicin nor isoniazid are given (or the subject is dead). For example, if RH is interrupted for 18 days and the participant dies 48 days after randomization, the endpoint will be 50% [(18+(60-48))/60]. Rifampicin and isoniazid are considered critical drugs in early TBM treatment; inability to use these agents (either through bacterial resistance or patient intolerance) is associated with poor outcome. The vast majority of interruptions are expected to be shorter than one month for strategy 3 (standard of care) but as management strategies 1 and 2 delay the time point of the interruption, a longer cut-off of 60 days was chosen.

Inclusion Criteria:

• Adult (18 years or older)
• HIV-infected
• Clinical diagnosis of TBM (≥5 days of meningitis symptoms, and CSF abnormalities) and anti-tuberculosis chemotherapy either planned or started by the attending physician

Note: Published diagnostic criteria will be applied to all enrolled participants at the end of the study when all mycobacterial culture results are available. The criteria will sub-divide all cases into definite, probable and possible TBM, and those with an alternative diagnosis.

Exclusion Criteria:

• An additional brain infection (other than TBM) confirmed or suspected: positive CSF Gram or India Ink stain; positive blood or CSF Cryptococcal antigen test; cerebral toxoplasmosis suspected and attending physician wants to give anti-toxoplasmosis treatment with anti-tuberculosis treatment
• More than 6 consecutive days of two or more drugs active against M. tuberculosis immediately before screening
• More than 3 consecutive days of any type of orally or intravenously administered corticosteroid immediately before randomisation
• Dexamethasone considered mandatory for any reason by the attending physician
• Dexamethasone considered to be contraindicated for any reason by the attending physician
• Previously been randomised into the trial for a prior episode of TBM
• Lack of consent from the participant or family member (if the participant is incapacitated by the disease)

The Oxford University Clinical Research Unit recognizes the ethical obligation to ensure that optimal use is made of the data and specimens that the investigators collect for our research and the value of sharing individual level data. The investigators aim to ensure that data generated from all our research are collected, curated, managed and shared in a way that maximizes their benefit. When sharing data the investigators have an obligation to ensure that the interests of research participants, researchers and other stakeholders are appropriately protected. The Oxford University Clinical Research Unit data sharing policy and the data request form outline the default procedures for data sharing.