44 Chapter 3 RNA biomarkers The genomic response to sepsis shows alterations in the transcriptome of peripheral white blood cells with significant differences of their RNA transcripts compared to healthy individuals [7]. Several investigations have reported on the ability of host transcriptome analyses to discriminate between infection and non-infectious acute disease, and even between different causative pathogens [38] (Figure 4). Among these diagnostic RNA biomarkers the molecular host response classifier SeptiCyte™ LAB was recently approved by the US Food and Drug Administration as an aid in differentiating infectionpositive (sepsis) from infection-negative systemic inflammation in critically ill patients on their first day of ICU admission [39]. This four-gene classifier combines CEACAM4, LAMP1, PLA2G7 and PLAC8 to produce a summary area under the receiver operating characteristic (ROC) curve (AUC) of 0.89 (95% CI 0.85-0.93) to differentiate sepsis from noninfectious systemic inflammatory response syndrome [40]. This rapid molecular assay is the first RNA-based clinical diagnostic tool derived from whole blood approved as a diagnostic test in critically ill patients [40]. Another investigation, performed in a large number of independent cohorts, reported an eleven-gene biomarker, named the Sepsis MetaScore, that could reliably distinguish patients with sepsis from patients with sterile inflammation [41] with an AUC of 0.87 (range, 0.70 to 0.98). Furthermore, a recent study, using genome-wide messenger (m)RNA expression profiles, identified a set panel of markers composed of three upregulated transcripts (Toll-like receptor 5, protectin and clusterin) and four down-regulated transcripts (fibrinogen-like 2, IL-7 receptor, major histocompatibility complex class II, carboxypeptidase, and vitellogenic-like) that best described the extent of immune alterations [42]. A gene expression score was created that was greater in patients with definite as well as with possible/probable infection than in those without infection [42]. Our group focused on the development of contextspecific molecular biomarkers, i.e., in patients with a particular clinical presentation, such as suspected community-acquired pneumonia (CAP)[43]. We compared wholegenome mRNA profiles in blood leukocytes of patients treated for suspected CAP on ICU admission, who were designated CAP (cases) and no-CAP patients (control subjects) by post hoc assessment. A 78-gene signature was defined for the diagnosis of CAP, from which a FAIM3:PLAC8 gene expression ratio was derived that outperformed plasma PCT in discriminating between CAP and no-CAP patients [43]. Moreover, other groups have evaluated the FAIM3:PLAC8 score in the context of all-cause adult, pediatric and neonatal sepsis with very favorable ROC AUCs [44, 45]. This indicates that although the FAIM3:PLAC8 score was derived as a context-specific biomarker, that is CAP diagnosis, its applicability may be broadened. These studies have shown that transcriptomics coupled with sophisticated machine learning and statistical tests may provide an invaluable tool to identify diagnostic biomarkers of sepsis. Several studies have indicated that host gene expression signatures can assist in discriminating between causative pathogens in infected patients. In an observational cohort study in adults presenting to the ED host gene expression profiling was used to classify the etiology of suspected community-acquired acute respiratory tract infections (ARI) into bacterial, viral or non-infectious origin. The overall accuracy of gene expression classifiers was 87%, which was significantly better than the use of procalcitonin which could assign patients as having bacterial ARI or nonbacterial ARI with an 78% accuracy [46]. Furthermore, transcriptional profiling of blood leukocytes from hospitalized
RkJQdWJsaXNoZXIy MTk4NDMw