Abstract
Background
Pulmonary TB (PTB) patients present with a wide range of pre-treatment Mycobacterium tuberculosis ( Mtb ) burdens, which predict poor treatment outcomes. We sought to identify immune pathways and biomarkers associated with pre-treatment Mtb burden.
Methods
We conducted whole-blood RNA sequencing in 295 Vietnamese adults with PTB, quantifying bacterial load using GeneXpert Ct values. Weighted gene co-expression network analysis (WGCNA) identified gene modules, pathways, and hub genes associated with Mtb burden. Deconvolution analysis assessed contributions of immune cell types. Key findings were validated in independent PTB (n=171) and TB meningitis (TBM, n=281) cohorts, and publicly available animal datasets. We used non-linear regression for variable selection to identify gene predictors of Mtb burden and hurdle regression to model Mtb loads below the detection limit.
Findings
Higher Mtb burden correlated with prolonged symptom duration, elevated neutrophil and monocyte counts, and severe lung pathology. WGCNA identified a 1,093-gene module associated with Mtb burden, characterized by coordinated innate-adaptive pathway interactions. Within this module, IFN-γ signaling participates in modulating the increase of innate signaling (Toll-like, Nod-like receptors, TNF) and the decrease of adaptive signaling (T- and B-cell receptor) pathways in high-burden patients. These responses were primarily driven by neutrophils and classical monocytes. CNIH4 emerged as the strongest hub-gene and a top predictor of bacterial burden, with consistent validation across independent PTB and TBM cohorts.
Interpretation
Our study reveals systemic innate–adaptive immune dynamics underlying bacterial burden in PTB and identifies CNIH4 as a potential biomarker for treatment monitoring as well as a therapeutic target.
Funding
National Institute of Health; Wellcome Trust, UK.
Research in Context
Evidence before this study
We searched PubMed from Jan 1, 2000, to Dec 10, 2025, without language restrictions, for human studies examining the relationship between Mycobacterium tuberculosis (Mtb) burden and host blood transcriptional responses in pulmonary tuberculosis. Search terms were used in combination as follows: (“Tuberculosis, Pulmonary” OR “pulmonary tuberculosis” OR PTB) AND (“Mycobacterium tuberculosis” OR mycobacter*) AND (“bacterial load” OR “bacterial burden” OR “sputum smear” OR “smear grade” OR xpert OR genexpert OR “cycle threshold” OR ct OR “time to positivity” OR TTP OR CFU OR “molecular bacterial load” OR MBLA)) AND (blood OR “whole blood” OR “peripheral blood”) AND (RNA-seq OR “RNA sequencing” OR transcriptom* OR “gene expression” OR microarray). We included studies of active pulmonary tuberculosis that measured quantitative or semi-quantitative bacterial burden and profiled host blood transcriptome-wide responses at baseline, reporting either differential expression by burden strata or associations between bacterial burden and host gene expression. We excluded studies limited to latent tuberculosis, animal or in vitro models, diagnostic or prognostic signature studies without bacterial burden measurement, studies focused on treatment response, and studies using targeted assays without transcriptome-wide profiling.
This search identified 12 articles describing blood transcriptional signatures for tuberculosis diagnosis, prognosis, and treatment response. However, only one study directly examined the relationship between pre-treatment bacterial burden and whole-blood transcriptome-wide profiles. That study demonstrated differences in systemic gene expression between patients with higher and lower sputum mycobacterial load and proposed a 20-gene blood signature associated with bacterial burden. However, the analysis was limited by small sample size, lack of pathway-level and cellular interpretation or assessment of correlation between signature with bacterial load.
Added value of this study
Our study advances existing evidence by leveraging the wide spectrum of pre-treatment bacterial burden observed in routine clinical populations, quantified using GeneXpert Ct values, and integrating this with whole-blood RNA sequencing in large, well-characterized clinical cohorts. Through network-based transcriptomic analysis, immune cell deconvolution, and non-linear modelling, we identify a bacterial burden–associated gene network characterized by enhanced innate inflammatory signaling and relative suppression of adaptive immune pathways, predominantly driven by neutrophils and classical monocytes and modulated by IFN-γ signaling. Within this network, CNIH4 emerges as a central hub gene and a robust predictor of bacterial burden, with consistent validation across independent pulmonary tuberculosis and tuberculous meningitis cohorts.
Implications of all the available evidence
Taken together, the available evidence indicates that host blood transcriptional responses correlate with bacterial burden in pulmonary tuberculosis, but previous studies have provided limited insight into the underlying immune processes. Our findings strengthen the biological link between pre-treatment mycobacterial burden and systemic immune dysregulation, showing that higher bacterial burden is associated with transcriptional state marked by coordinated upregulation of innate immune responses and downregulation of adaptive immune pathways. These results support the use of host transcriptomic profiling as a biologically informative complement to sputum-based measures of bacterial burden and highlight burden-associated immune pathways, particularly CNIH4, as a potential target for treatment monitoring and host-directed therapeutic development.
Full Text Availability
The license terms selected by the author(s) for this preprint version do not permit archiving in PMC. The full text is available from the preprint server.
