Integrative systems biology identifies shared neuroimmune co-expression programmes across pulmonary tuberculosis and Alzheimer's disease

Abstract Background : There is growing evidence that pulmonary tuberculosis (PTB) exerts long-lasting systemic effects on brain function, whereas Alzheimer's disease (AD) is increasingly recognized as a systemic neuroimmune disorder. BCG vaccination studies suggest that tuberculosis-associated immune pathways influence AD-related biology. However, whether PTB-associated co-expression programmes exhibit concordant activity in AD and display reproducible neuroanatomical organization within the human brain remains unknown. We hypothesized that PTB-associated co-expression programmes would exhibit concordant transcriptional activity in AD, retain aspects of their underlying co-expression architecture across diseases, and identify shared neuroimmune mechanisms relevant to chronic infection and neurodegeneration. We implemented an integrative systems biology framework combining whole-blood transcriptomics, weighted gene co-expression network analysis, functional enrichment, immune-cell deconvolution, and machine learning to identify PTB-associated co-expression programmes and prioritize biomarkers. These programmes were validated in two independent PTB cohorts, evaluated in an AD dataset, and mapped onto the Allen Human Brain Atlas to characterize neuroanatomical organization. Results: Active PTB was associated with widespread transcriptional remodeling, with 1,389 dysregulated genes organized into 16 PTB-associated co-expression modules. Nested machine learning identified a 12-gene biomarker signature that demonstrated excellent diagnostic performance in the discovery cohort (AUC ≈ 0.99) and retained performance in two independent external PTB datasets without model retraining. Majority of PTB-derived immune co-expression modules exhibited concordant activity in the independent AD cohort. Formal module-preservation analysis demonstrated that several PTB-derived networks retained moderate preservation in AD, supporting partial conservation of co-expression architecture across diseases. Brain transcriptomic mapping demonstrated reproducible region-specific organization across adult brain donors. Integrated neuroanatomical contextualization prioritized a subset of PTB-derived neuroimmune modules using cross-disease transcriptional activity, regional specificity, and inter-donor reproducibility, revealing neuroimmune transcriptional programmes shared across infection and neurodegeneration. Conclusions: Our findings establish a systems-level framework for investigating shared neuroimmune processes across chronic infection and neurodegeneration while identifying reproducible co-expression modules and candidate biomarkers for future studies.