Model-Based Insulin Sensitivity as a New Biomarker of Sepsis Diagnosis in the Intensive Care Unit

Abstract

INTRODUCTION: Currently, there is a lack of real-time biomarker to diagnose sepsis. Insulin sensitivity (SI) may be determined in real-time using mathematical glucose-insulin models, but its effectiveness as a diagnostic test of sepsis remains unexplored. We aimed to explore the diagnostic value of model-based SI as a new biomarker of sepsis in a mixed cohort of diabetic and non-diabetic patients newly admitted to the intensive care unit (ICU). MATERIALS AND METHODS: In this cross-sectional study, we analysed SI levels derived from the Intensive[1]Control-of-Insulin-Nutrition-Glucose model in septic (n=45) and non-septic (n = 41) patients upon their ICU admission. The diagnostic value of model-based SI for sepsis was determined through analysis of the area under the curve (AUC) of the receiver operating characteristic curve. RESULTS: Baseline SI levels were significantly lower in patients with sepsis than those without sepsis (0.560 (SD=0.676) vs. 1.097 (SD=1.473) x 10-4 L/mU/min, P = 0.037). However, the AUC of 0.588 revealed that model-based SI was a poor diagnostic test of sepsis in the mixed cohort of diabetics and non-diabetics. In a separate analysis among the non-diabetics (n=19), model-based SI predicted sepsis with clinically valid performance (AUC 0.911). CONCLUSION: Presence of sepsis significantly reduced SI in the critically ill patients but a low SI could predict sepsis only in the non-diabetic cohort.