Fjell CD, Thair S, Hsu JL, Walley KR, Russell JA, Boyd J.; PLoS One. 2013 Nov 14;8(11):e79207.

INTRODUCTION: Inflammatory response during sepsis is incompletely understood due to small sample sizes and variable timing of measurements following the onset of symptoms. The vasopressin in septic shock trial (VASST) compared the addition of vasopressin to norepinephrine alone in patients with septic shock. During this study plasma was collected and 39 cytokines measured in a 363 patients at both baseline (before treatment) and 24 hours. Clinical features relating to both underlying health and the acute organ dysfunction induced by the severe infection were collected during the first 28 days of admission.

HYPOTHESIS: Cluster analysis of cytokines identifies subgroups of patients at differing risk of death and organ failure.

METHODS: Circulating cytokines and other signaling molecules were measured using a Luminex multi-bead analyte detection system. Hierarchical clustering was performed on plasma values to create patient subgroups. Enrichment analysis identified clinical outcomes significantly different according to these chemically defined patient subgroups. Logistic regression was performed to assess the importance of cytokines for predicting patient subgroups.

RESULTS: Plasma levels at baseline produced three subgroups of patients, while 24 hour levels produced two subgroups. Using baseline cytokine data, one subgroup of 47 patients showed a high level of enrichment for severe septic shock, coagulopathy, renal failure, and risk of death. Using data at 24 hours, a larger subgroup of 81 patients that largely encompassed the 47 baseline subgroup patients had a similar enrichment profile. Measurement of two cytokines, IL2 and CSF2 and their product were sufficient to classify patients into these subgroups that defined clinical risks.

CONCLUSIONS: A distinct pattern of cytokine levels measured early in the course of sepsis predicts disease outcome. Subpopulations of patients have differing clinical outcomes that can be predicted accurately from small numbers of cytokines. Design of clinical trials and interventions may benefit from consideration of cytokine levels.

Weblink here