A 29 messenger RNA host response signature identifies bacterial and viral infections among emergency department patients

Document Type

Conference Proceeding

Publication Date


Publication Title

Academic emergency medicine


Background and Objectives: Rapid diagnosis of acute infections is a critical need. Current tests show limitations in turnaround time, sensitivity, and/or interpretation. Overtreatment with antimicrobials is well documented and leads to drug-resistant microbes and potential complications. We hypothesized that HostDx™ Sepsis (HDxS), a host response test which measures 29 mRNA targets from peripheral blood could be deployed in the ED to identify and differentiate bacterial and viral infections as well as their severity.

Methods: In a seven-center, prospective, non-interventional clinical trial in EDs across the United States and Europe, we enrolled adult patients with suspected acute infection (of respiratory, urinary, intra-abdominal or skin/soft tissue origin) or suspected sepsis. RNA was isolated from whole blood collected in PAXgene® RNA tubes and quantified remotely using NanoString® nCounter®. Predicted probabilities were calculated using the IMX-BVN-1 machine learning algorithm developed on independent discovery and validation cohorts. Each patient's infection status was adjudicated by two of the three principal investigators in a retrospective chart review using clinical data, standard-of-care tests, procalcitonin (PCT), C-reactive protein (CRP), and respiratory pathogen PCR panel tests, while blinded to HDxS results.

Results: Interim analysis was performed on 268 of 557 total patients enrolled. Physicians consensus adjudication classified patients as bacterial (62; 5 respiratory, 32 urinary, 10 intra-abdominal, 9 skin/soft tissue, 6 other), viral (29; 28 respiratory, 1 skin/soft tissue), co-infected (1; 1 respiratory), noninfected (89), and indeterminant (88). The AUC of HDxS bacterial score for identifying bacterial vs. viral or non-infected patients was 0.81 (95% CI 0.74 - 0.88), compared to 0.74 (95% CI 0.66 - 0.82) for PCT. The AUC of HDxS viral score for identifying viral vs. bacterial or non-infected patients was 0.75 (95%CI 0.64 - 0.86) compared to 0.43 (95%CI 0.32 - 0.54) for PCT.

Conclusion: HDxS diagnosed bacterial infections versus other etiologies with high accuracy in a prospective multicenter trial. HDxS is being developed as a point-of-care instrument with a <30 minute turnaround time. HDxS may help clinicians make earlier and more accurate treatment decisions supporting early sepsis identification and antimicrobial stewardship.



First Page


This document is currently not available here.