A Diagnostic Platform Predicts Presymptomatic Exposure to Respiratory Viral Infection.

Abstract

INTRODUCTION Communicable diseases can be a significant burden during military operations in which missions rely on each and every team member for optimal performance. Community-acquired acute respiratory infections are particularly difficult to contain, given their rapid transmissibility and the lack of early diagnostic tests. Disease prevention using vaccination has shown low effectiveness, notably the overall flu vaccine effectiveness estimate for 2014–2015 flu season was 19%. Contemporary approaches to diagnose respiratory infection rely on time-consuming pathogen-specific assessment during the symptomatic window of the infection. In the last decade, however, the genomics revolution—and specifically, systems genomics approaches combined with mathematical modeling, has led to better understanding of disease at the individual level and the emergence of new diagnostic and therapeutic paradigms. The diagnostic limitations to control infections nonetheless remain a medical challenge as evidenced by the most recent H1N1 influenza pandemic outbreak of 2009. As emphasized by a U.S. Department of Health and Human Services retrospective analysis: “The absence of readily available, rapid, simple, and highly sensitive diagnostic tests which could detect the 2009 H1N1 virus made infection control more difficult.” In 2006, the Defense Advanced Research Projects Agency (DARPA) launched an ambitious program to devise models and methods that predict incipient infection after exposure. The program known as Predicting Health and Disease (PHD) aimed to enable the capability to diagnose individuals with a respiratory infection before they become symptomatic. DEVELOPMENT AND VALIDATION OF MODELS THAT DISTINGUISH INFLUENZA INFECTED SYMPTOMATIC PATIENTS To develop novel predictive models of incipient respiratory infectious disease, a group at Duke University led by Dr. Geoff Ginsburg, in collaboration with Dr. Alfred Hero at University of Michigan, performed studies with healthy volunteers experimentally infected with live unattenuated Influenza (H1N1 or H3N2), respiratory syncytial virus, or human rhinovirus, and who underwent serial blood sampling at regular time intervals preand postinoculation. Using factor analysis of mRNA expression arrays with ~22,000 genes, the team identified patterns of “host-derived gene expression signatures” in blood that were highly specific for infection in symptomatic individuals, and distinct from asymptomatic subjects. The discriminatory factors (designated viral factors) encompassed a collection of highly correlated host genes involved in antiviral responses (i.e. Toll-like receptor signaling genes, Interferon response genes, etc.), and were closely associated with disease dynamics and symptom scores.