Backtranslation of human RNA biosignatures of tuberculosis disease risk into the preclinical pipeline is condition dependent.

It is unclear whether human progression to active tuberculosis disease (TB) risk signatures are viable endpoint criteria for evaluations of treatments in development. TB is the deadliest infectious disease globally and more efficacious vaccines are needed to reduce this mortality. However, the immune correlates of protection for either preventing infection with Mycobacterium tuberculosis or preventing TB disease have yet to be completely defined, making the advancement of candidate vaccines through the pipeline slow, costly, and fraught with risk. Human-derived correlate of risk (COR) gene signatures, which identify an individual's risk of progressing to active TB disease, provide an opportunity for evaluating new therapies for TB with clear and defined endpoints. Though prospective clinical trials with longitudinal sampling are prohibitively expensive, the characterization of COR gene signatures is practical with preclinical models. Using a 3Rs (replacement, reduction, and refinement) approach we reanalyzed heterogeneous publicly available transcriptional data sets to determine whether a specific set of COR signatures are viable endpoints in the preclinical pipeline. We selected RISK6, Sweeney3, and BATF2 human-derived blood-based RNA biosignatures because they require relatively few genes and have been carefully evaluated across several clinical cohorts. These data suggest that in certain experimental designs and in several tissue types, human COR signatures correlate with disease progression as measured by the bacterial burden in the preclinical TB model pipeline. We observed the best performance when the model most closely reflected human infection or disease conditions. Human-derived COR signatures offer an opportunity for high-throughput preclinical endpoint criteria of vaccine and drug therapy evaluations.

Importance: Understanding the strengths or limitations of back-translating human-derived correlate of risk (COR) RNA signatures into the preclinical pipeline may help streamline down-selection of therapeutic vaccine and drug candidates and better align preclinical models with proposed clinical trial efficacy endpoints.