Reimagining the IMAGINE-TBM trial: Using metagenomics and host transcriptomics to improve our ability to diagnose tuberculous meningitis and assess response to treatment in a randomized clinical trial - PROJECT SUMMARY Tuberculous meningitis (TBM) is one of the most devastating forms of tuberculosis (TB), with mortality rates as high as 50%. Despite dismal outcomes, few data are available on optimizing the approach to treatment of TBM. While several interventional TBM trials are underway, they face tremendous barriers that could impede the knowledge gained from these essential studies. First, TBM is exceptionally difficult to diagnose, which directly impacts selection of the right trial population on which to test treatment regimens. Second, although the host immune response may be a major driver of poor outcomes in TBM, our understanding of the immunopathogenesis of TBM and how it may be targeted with host-directed therapy is in its infancy. Third, dedicated resources to develop and sustain research capacity are often limited in low- and middle-income countries (LMIC) where TB is endemic and where data-driven treatment strategies for TBM are urgently needed. In this ancillary study, we propose 3 scientific aims and a capacity-building plan designed to address these major challenges. We will perform a secondary analysis of the ACTG IMAGINE-TBM (Improved Management with Antimicrobial AGents Isoniazid rifampiciN LinEzolid for TBM) trial, a phase II randomized trial comparing a 6-month regimen of high dose rifampin, high dose isoniazid, linezolid, and pyrazinamide with a 9-month standard of care regimen in participants at 17 sites in 12 LMIC across 3 continents. In Aim 1, before deploying our previously developed host transcriptomic machine learning classifier (MLC) to diagnose TBM in the diverse IMAGINE-TBM trial population, we will use integrated host gene expression data from stored cerebrospinal fluid (CSF) from two successfully completed TBM trials, TBM-KIDS and ALTER, to retrain and revalidate the MLC. In Aim 2, we will use a combination of direct pathogen detection with CSF metagenomic next generation sequencing (mNGS) and the optimized host transcriptomic MLC developed in Aim 1 to reclassify 300 IMAGINE-TBM participants as definite versus not definite TBM, after which we will re-analyze the efficacy of the investigational regimen versus standard of care. In Aim 3, we will use a case-control study design nested within IMAGINE-TBM to determine if host gene expression profiles differ between participants who have a poor versus good outcome. This study will leverage stored biologic specimens and existing data to augment the knowledge gained from IMAGINE-TBM. By re-analyzing the trial after re-classifying participants using mNGS and host transcriptomics, we could detect a benefit of the investigational regimen that would otherwise be missed. In addition, findings from the study will provide critical insights into immune pathways implicated in TBM pathogenesis and response to treatment, which could support the use of targeted host- directed therapies. In conjunction with the scientific aims, to strengthen capacity for sustainable TBM research programs based in epicenters of the TB epidemic, we will partner with the Chan Zuckerberg Biohub to support training in metagenomics and bioinformatics at 3 regional training hubs in India, Kenya, and Peru.
