Biomarkers for progression from latent to active tuberculosis identified in multiplexed proteomic assay (SOMAscan™) of human plasma (MPF6P.657)

Abstract

Abstract Background: Biomarkers of progression from latent to active tuberculosis (TB) would be very useful for targeted treatment or priority enrollment in TB vaccine trials. Methods: A highly multiplexed proteomic assay (SOMAscan) was used to measure over 3000 human proteins in plasma from adolescents infected with M. tuberculosis. We compared proteomic data in longitudinally collected samples (six-month intervals over two years) from 97 individuals, 29 of whom developed active TB (progressors) and 68 matched controls who remained healthy. Bioinformatics algorithms were applied to discover biomarkers of TB risk. Results: Among the top host biomarkers distinguishing progressors from controls were 17 proteins, and these factors were used for model building to predict TB risk. An 8-marker model gave a cross-validated performance of 76% sensitivity and 86% specificity (AUC=0.86). The individual markers had variable responsiveness at time-points closer to TB diagnosis. The model identified progressors up to 540 days prior to TB diagnosis, with elevated logOdds for developing active TB. In contrast, a previously obtained host response signature for active pulmonary TB detected progressors up to 180 days prior to diagnosis. Conclusions: The discovery of TB-risk biomarkers indicates that subjects who eventually progress to TB disease can be identified well before clinical diagnosis of TB, and thus enables potential triage for targeted treatment or priority enrollment in TB vaccine trials.