Turning a new page in sepsis molecular diagnostics necessitates context-specific biomarkers

Abstract

In a recent issue of Critical Care Medicine, Sweeney and Khatri (1) recently reported performance comparisons of three blood gene expression diagnostics for sepsis and noninfectious patient discrimination, namely their 11 gene Sepsis MetaScore (SMS) (2), the 2 gene FAIM3/PLAC8 ratio (3), and the 4 gene SeptiCyte LAB (4). Using publicly available data, the authors generated a “unified” repository of gene expression data gathered from studies investigating a variety of clinically predefined critically ill patient groups. Despite the evident bias toward the 11 gene SMS (2), which was derived from multiple datasets included in the repository presented in this study, the authors concluded that on the basis of receiver operator characteristic area-under-the-curve analysis, there was no significant difference between the three gene expression diagnostics (1). We disagree with the approach used by the authors because the FAIM3/PLAC8 ratio was taken out of the context for which it was derived and validated, that is, assisting the clinician in discriminating adult patients with community-acquired pneumonia from those patients with noninfectious respiratory distress at ICU admission using whole blood gene expression (3). The FAIM3/PLAC8 ratio was validated only for this specific clinical and analytical scenario, not for discriminating sepsis from a noninfectious condition in postoperative patients, patients after trauma or with Dengue fever, and infants with a variety of infections, in different test samples (blood, buffy coats, neutrophils).