Blood Transcriptional Phenotypes of Progressive Latent <i>M. tuberculosis</i> Infection Inform Novel Signatures That Improve Prediction of Tuberculosis Risk

Promising blood transcriptional signatures of TB-risk that more precisely identify the 5% of latent M. tuberculosis (Mtb) infection (LTBI) with progressive infection at risk of developing to active tuberculosis (TB), have thus far failed to achieve sensitivity thresholds necessary to support the WHOs global TB elimination strategy. We postulate this reflects transcriptional heterogeneity of progressive infection, not acknowledged in signature development. Applying low-stringency analytical methods we reveal marked transcriptional heterogeneity within progressor LTBI cohorts from diverse TB burden settings along a continuum that is classifiable into three transcriptional subgroups (early, intermediate, advanced) based on transcriptional proximity to active TB. We show that existing candidate TB-risk signatures fail to identify the quarter of progressors with an early transcriptional phenotype. By developing a unique signature optimised to early progressors, we demonstrate a strategy of complementary signatures specifically optimised to transcriptional phenotypes of progressive infection can significantly improve utility and viability of transcriptional biomarker based screening.