Biomarkers' performance in the SEPSIS-3 era

Objectivethe biomarkers performance for diagnosis and severity stratification of sepsis has not been properly evaluated anew using the SEPSIS-3 criteria introduced in 2016. We evaluated the accuracy of 21 biomarkers classically tested in sepsis research to identify infection, sepsis, and septic shock in surgical patients classified using SEPSIS-3. Methodsfour groups of adult surgical patients were compared: post-surgical patients with no infection, patients with infection but no sepsis, patients with sepsis, and patients with septic shock were recruited prospectively from the surgery departments and surgical ICUs from four Spanish hospital. The area under the curve (AUC) to differentiate between groups was calculated for each biomarker. ResultsA total of 187 patients were recruited (50 uninfected post-surgery controls, 50 patients with infection, 47 with sepsis and 40 with septic shock). The AUCs indicated that none of the biomarkers tested was accurate enough to differentiate those patients with infection from the uninfected controls. In contrast, procalcitonin, lipocalin 2, pentraxin 3, IL-15, TNF-, IL-6, angiopoietin 2, TREM-1, D-dimer and C-reactive protein yielded AUCs > 0.80 to discriminate the patients with sepsis or septic shock from those with no infection. C-reactive protein and IL-6 were the most accurate markers to differentiate plain infection from sepsis (AUC = 0.82). Finally, our results revealed that sepsis and septic shock shared similar profiles of biomarkers. ConclusionRevaluation in the "SEPSIS-3 era" identified the scenarios where biomarkers do and do not provide useful information to improve the management of surgical patients with infection or sepsis.