Background.Tuberculous meningitis (TBM) is the most devastating form of tuberculosis, yet very little is known about the pathophysiology. We hypothesized that the genotype of leukotriene A4 hydrolase (encoded by LTA4H), which determines inflammatory eicosanoid expression, influences intracerebral inflammation, and predicts survival from TBM.Methods.We characterized the pretreatment clinical and intracerebral inflammatory phenotype and 9-month survival of 764 adults with TBM. All were genotyped for single-nucleotide polymorphism rs17525495, and inflammatory phenotype was defined by cerebrospinal fluid (CSF) leukocyte and cytokine concentrations.Results. LTA4H genotype predicted survival of human immunodeficiency virus (HIV)–uninfected patients, with TT-genotype patients significantly more likely to survive TBM than CC-genotype patients, according to Cox regression analysis (univariate P = .040 and multivariable P = .037). HIV-uninfected, TT-genotype patients had high CSF proinflammatory cytokine concentrations, with intermediate and lower concentrations in those with CT and CC genotypes. Increased CSF cytokine concentrations correlated with more-severe disease, but patients with low CSF leukocytes and cytokine concentrations were more likely to die from TBM. HIV infection independently predicted death due to TBM (hazard ratio, 3.94; 95% confidence interval, 2.79–5.56) and was associated with globally increased CSF cytokine concentrations, independent of LTA4H genotype.Conclusions. LTA4H genotype and HIV infection influence pretreatment inflammatory phenotype and survival from TBM. LTA4H genotype may predict adjunctive corticosteroid responsiveness in HIV-uninfected individuals.

Adjunctive dexamethasone reduces mortality from tuberculous meningitis (TBM) but not disability, which is associated with brain infarction. We hypothesised that aspirin prevents TBM-related brain infarction through its anti-thrombotic, anti-inflammatory, and pro-resolution properties. We conducted a randomised controlled trial in HIV-uninfected adults with TBM of daily aspirin 81 mg or 1000 mg, or placebo, added to the first 60 days of anti-tuberculosis drugs and dexamethasone (NCT02237365). The primary safety endpoint was gastro-intestinal or cerebral bleeding by 60 days; the primary efficacy endpoint was new brain infarction confirmed by magnetic resonance imaging or death by 60 days. Secondary endpoints included 8-month survival and neuro-disability; the number of grade 3 and 4 and serious adverse events; and cerebrospinal fluid (CSF) inflammatory lipid mediator profiles. 41 participants were randomised to placebo, 39 to aspirin 81 mg/day, and 40 to aspirin 1000 mg/day between October 2014 and May 2016. TBM was proven microbiologically in 92/120 (76.7%) and baseline brain imaging revealed ≥1 infarct in 40/114 (35.1%) participants. The primary safety outcome occurred in 5/36 (13.9%) given placebo, and in 8/35 (22.9%) and 8/40 (20.0%) given 81 mg and 1000 mg aspirin, respectively (p=0.59). The primary efficacy outcome occurred in 11/38 (28.9%) given placebo, 8/36 (22.2%) given aspirin 81 mg, and 6/38 (15.8%) given 1000 mg aspirin (p=0.40). Planned subgroup analysis showed a significant interaction between aspirin treatment effect and diagnostic category (Pheterogeneity = 0.01) and suggested a potential reduction in new infarcts and deaths by day 60 in the aspirin treated participants with microbiologically confirmed TBM (11/32 (34.4%) events in placebo vs. 4/27 (14.8%) in aspirin 81 mg vs. 3/28 (10.7%) in aspirin 1000 mg; p=0.06). CSF analysis demonstrated aspirin dose-dependent inhibition of thromboxane A2 and upregulation of pro-resolving CSF protectins. The addition of aspirin to dexamethasone may improve outcomes from TBM and warrants investigation in a large phase 3 trial.

Prognostic models developed and validated using data from 1699 adults with tuberculous meningitis (TBM), with or without human immunodeficiency virus infection, performed well and could be used in clinical practice to identify patients with TBM at high risk of death.

Many approaches for variable selection with multiply imputed data in the development of a prognostic model have been proposed. However, no method prevails as uniformly best. We conducted a simulation study with a binary outcome and a logistic regression model to compare two classes of variable selection methods in the presence of MI data: (I) Model selection on bootstrap data, using backward elimination based on AIC or lasso, and fit the final model based on the most frequently (e.g. ) selected variables over all MI and bootstrap data sets; (II) Model selection on original MI data, using lasso. The final model is obtained by (i) averaging estimates of variables that were selected in any MI data set or (ii) in 50% of the MI data; (iii) performing lasso on the stacked MI data, and (iv) as in (iii) but using individual weights as determined by the fraction of missingness. In all lasso models, we used both the optimal penalty and the 1‐se rule. We considered recalibrating models to correct for overshrinkage due to the suboptimal penalty by refitting the linear predictor or all individual variables. We applied the methods on a real dataset of 951 adult patients with tuberculous meningitis to predict mortality within nine months. Overall, applying lasso selection with the 1‐se penalty shows the best performance, both in approach I and II. Stacking MI data is an attractive approach because it does not require choosing a selection threshold when combining results from separate MI data sets