Inflammatory markers in the cerebrospinal fluid linked to mortality in tuberculous meningitis

Abstract

This study examines the role of host inflammation in the high mortality of tuberculous meningitis (TBM) and identifies potential biomarkers associated with improved survival. We conducted a case-control study involving 131 patients in a discovery cohort, 81 TBM patients in a validation cohort, and 43 non-infected controls from a referral hospital in Indonesia. We measured 94 inflammation-related proteins in cerebrospinal fluid (CSF) and performed genome-wide quantitative trait loci (QTL) mapping. Sixty-seven proteins were found to be differentially expressed between TBM patients and controls, with 64 proteins elevated in patients. Five proteins, including vascular endothelial growth factor (VEGF) and matrix metalloproteinase-10 (MMP-10), were identified as predictors of 180-day mortality in TBM patients. The validation cohort confirmed that MMP-10, but not VEGF, was predictive of mortality. Genome-wide QTL mapping identified two genome-wide significant and four suggestive genetic loci associated with CSF MMP-10, which also predicted survival in an additional cohort of 218 patients. High CSF concentrations of MMP-10, along with specific genetic loci, may be associated with survival in TBM patients, suggesting a potential role for MMP-10 in disease pathogenesis and warranting further investigation into its utility in host-directed therapies.

Keywords: MMP-10; biomarkers for survival; cerebrospinal fluid; protein QTL; tuberculous meningitis.

Graphical Abstract

Figure 1

Protein levels show large differences between patients and controls in CSF. (A) Volcano plot for individual proteins in CSF with the fold change between TBM and non-infectious controls on the x-axis, and the FDR on the y-axis. Proteins significantly different between TBM and controls (FDR < 0.05) are coloured in black (N = 67), while the statistically insignificant proteins are coloured in grey (N = 3). Each point represents one protein. (B) Correlation plots for representatives of CSF protein clusters with CSF total protein (as a proxy for barrier disruption) for 131 TBM patients, NSE (as proxy for brain damage), and CSF mononuclear and polymorphonuclear cell counts. Each data point on the x-axis represents the Spearman’s rho value for a single protein.

Figure 2

CSF MMP-10 concentrations predict survival in tuberculous meningitis. Kaplan–Meier graph with survival table for patients of the protein discovery (N = 130, A) and the protein validation cohort (N = 81, B) based on CSF MMP-10 concentrations divided in tertiles.

Figure 3

Genetic variants predict MMP-10 concentrations in CSF from tuberculous meningitis patients. (A) Manhattan plot showing the genome-wide QTL mapping results for CSF MMP-10 concentrations in TBM patients of the genetic discovery cohort (N = 209; three patients were excluded from the analysis because of missing genetic data); horizontal lines correspond to P < 5 × 10⁻⁸ and P < 1 × 10⁻⁵. Each data point indicates the −log₁₀  P-value of a single SNP in relation to MMP-10 protein levels. (B and C) Survival according to the genetic risk score for the genetic discovery cohort (B, N = 209), and the genetic validation cohort (C, N = 218).