Data driven identification of divergent host immune subgroups in tuberculosis

Abstract BACKGROUND Immunity is a critical factor of clinical outcomes after Mycobacterial tuberculosis (Mtb) infection, however, the lack of identification of immune correlates of protection has prevented development of host directed therapeutics. TB studies have identified incongruous immune responses that can lead to a similar TB disease phenotype. Therefore, instead of envisioning susceptibility to TB to follow a unique path, we sough evidence by which divergent host immunity could result in TB. RESULTS Inspired by multiomics identification of cancer subtypes improving clinical care, we implemented unbiased, RUV clustering from 12 publicly available datasets consisting of data from 717 TB patients and 527 controls, including 12,468 common genes, and identified 4 distinct immune responses among TB patients. The two largest sub-groups displayed divergent metabolic, epigenetic and immune pathways. Cluster A, consisting of 333 TB patients was characterized by increased glycolysis; up-regulation of epigenetic-modifying genes and increased IFN-γ and TNF signaling. In contrast, Cluster B, consisting of 313 TB patients enriched for gene expression signatures similar to patients with sepsis and immune exhaustion and was characterized by up-regulated NFAT and hormone metabolism and decreased glycolysis, IFN-γ and TNF signaling. In silico evaluation suggests that drugs beneficial for TB patients in Cluster A are detrimental in Cluster B type disease, and vice versa. Multiplex ELISA using a validation cohort of TB patients confirmed divergent immune sub-types in TB. CONCLUSION Host immunity to TB is heterogenous and applying precision medicine may enabled sub-stratification guided host directed therapies for management of TB.