Gene Expression-Based Diagnosis of Infections in Critically Ill Patients-Prospective Validation of the SepsisMetaScore in a Longitudinal Severe Trauma Cohort.

Thair, Simone; Mewes, Caspar; Hinz, José; Bergmann, Ingo; Büttner, Benedikt; Sehmisch, Stephan; Meissner, Konrad; Quintel, Michael; Sweeney, Timothy E; Khatri, Purvesh; Mansur, Ashham

OBJECTIVES Early diagnosis of infections is pivotal in critically ill patients. Innovative gene expression-based approaches promise to deliver precise, fast, and clinically practicable diagnostic tools to bedside. This study aimed to validate the SepsisMetaScore, an 11-gene signature previously reported by our study group, in a representative longitudinal cohort of trauma patients. DESIGN Prospective observational cohort study. SETTING Surgical ICUs of the University Medical Center Goettingen, Germany. PATIENTS Critically ill patients with severe traumatic injuries. INTERVENTIONS None. MEASUREMENTS AND MAIN RESULTS Paired box gene (PAXgene) RNA blood tubes were drawn at predefined time points over the course of disease. The performance of the SepsisMetaScore was tested using targeted polymerase chain reaction and compared with Procalcitonin using area under the receiver operating characteristics analyses. The SepsisMetaScore showed significant differences between infected and noninfected patients (n = 52). It was able to accurately discriminate infectious from noninfectious acute inflammation with an area under the receiver operating characteristics of 0.92 (95% CI, 0.85-0.99) and significantly outperformed Procalcitonin (area under the receiver operating characteristics curve = 0.53; 95% CI, 0.42-0.64) early in the course of infection (p = 0.014). CONCLUSIONS We demonstrated the clinical utility for diagnosis of infections with higher accuracy using the SepsisMetaScore compared with Procalcitonin in a prospective cohort of severe trauma patients. Future studies should assess whether the SepsisMetaScore may substantially improve clinical practice by accurate differentiation of infections from sterile inflammation and identification of patients at risk for sepsis. Our results support further investigation of the SepsisMetaScore for the development of tailored precision treatment of critically ill patients.