432554 A Mechanistic Model of Stress Hyperglycemia Evolution in the Intensive Care Unit

Wednesday, November 11, 2015: 10:00 AM
150G (Salt Palace Convention Center)
Ari Pritchard-Bell1, Timothy Knab2, Gilles Clermont3 and Robert S. Parker2, (1)University of Pittsburgh, Pittsburgh, PA, (2)Chemical and Petroleum Engineering, University of Pittsburgh, Pittsburgh, PA, (3)Department of Critical Care Medicine, University of Pittsburgh, Pittsburgh, PA

Many patients in the intensive care unit (ICU) experience elevated blood glucose levels, so-called stress hyperglycemia. It has been shown that controlling glucose levels using intensive insulin therapy (IIT) can reduce mortality [1, 2]. However, a prospective multi-center study [3] of over 6,000 patients showed increased mortality for those receiving IIT. Retrospective analysis [4] shows that IIT caused increased occurrence of hypoglycemia, which consequently reduced survival within the IIT cohort. The conclusion drawn from the aforementioned studies is the existence of a glucose range that, when patients are controlled therein, yields lower mortality, approximately 80 to 120 mg/dL. A model-based decision support system (DSS) can provide patient-specific treatment and improve blood glucose control to range without increasing risk of hypoglycemia. The critical component of a patient-tailored DSS is a mathematical model that can resolve the dynamic changes resulting in a unique individuals metabolic state.

The underlying mechanism of stress hyperglycemia is a complex network of biological signaling pathways that decrease sensitivity to insulin [5] and increase endogenous glucose production (EGP). In this work, mathematical modeling is used to identify and characterize the complex biological pathways leading to stress hyperglycemia. The two modeled metabolic regulatory processes involved in dynamic modification of metabolism in the ICU are: (i) the hypoglycemia counterregulatory response and (ii) the acute inflammatory response. The counterregulatory response occurs following a hypoglycemic event, common in the ICU [4], and increases blood glucose via increased EGP and insulin resistance leading to reduced insulin-mediated glucose uptake (IMGU). Similarly, the acute inflammatory response includes cytokines such as TNF-α and hormones such as cortisol, which also decrease IMGU in humans [7, 8]. These two metabolic regulatory pathways are coupled with an model of glucose and insulin homeostasis [6] from literature to resolve patient specific variations in metabolic state.

Patient glucose measurements are fit by adjusting a single, time-varying insulin sensitivity parameter. The time-varying parameter profile used to match model glucose values with recorded glucose values for each patient is subsequently used to predict explanatory biomarker concentrations belonging to the two metabolic regulatory pathways previously described. Specifically, TNF-α, cortisol, epinephrine, and glucagon dynamics are simulated to match the insulin sensitivity profile for each patient. Mathematical models of acute inflammation and counterregulation are used as explanatory mechanisms driving insulin sensitivity, which enforce pathway specific dynamics on overall glucose dynamics. Both pathways are evaluated to determine whether they act individually or concurrently to produce observed insulin resistance and stress hyperglycemia in ICU patients.

This model of glucose and insulin, combined with mechanistic counterregulatory and inflammatory dynamics, serves as a simulation platform to generate clinically-relevant critically ill patient metabolic profiles. A library of time-varying metabolic parameter profiles is fit to a subset of the 215 trauma victims from two hospital centers in the Cologne area participating in the German Trauma Registry effort, to provide a virtual patient cohort matching the dynamics of clinical response. Such a virtual patient platform is useful for developing DSS control strategies, as well as to better understand possible patient differentiation metrics for separating treatment cohorts (e.g., counterregulation driven, inflammation driven, or both) for which treatment strategies may differ as a result of their metabolic upset.


1. Van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, et al. (2001) Intensive insulin therapy in critically ill patients. New England journal of medicine 345: 1359-1367.

2. Krinsley JS (2004) Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clinic Proceedings. Elsevier, volume 79, pp. 992-1000.

3. Finfer S (2009) Intensive versus conventional glucose control in critically ill patients. New England Journal of Medicine 360: 1283-1297.

4. Finfer S (2012) Hypoglycemia and risk of death in critically ill patients. New England Journal of Medicine 367: 1108-1118.

5. Brealey D, Singer M (2009) Hyperglycemia in critical illness: A review. Journal of Diabetes Science and Technology 3: 1250-1260.

6. Lin J, Razak NN, Pretty CG, Le Compte A, Docherty P, et al. (2011) A physiological intensive control insulin-nutrition-glucose (ICING) model validated in critically ill patients. Computer Methods and Programs in Biomedicine 102: 192-205.

7. Plomgaard P, Bouzakri K, Krogh-Madsen R, Mittendorfer B, Zierath JR, et al. (2005) Tumor necrosis factor-α induces skeletal muscle insulin resistance in healthy human subjects via inhibition of akt substrate 160 phosphorylation. Diabetes 54: 2939-2945.

8. Rizza RA, Mandarino LJ, Gerich JE (1982) Cortisol-induced insulin resistance in man: Impaired suppression of glucose production and stimulation of glucose utilization due to a postreceptor defect of insulin action. The Journal of Clinical Endocrinology & Metabolism 54: 131-138.

Extended Abstract: File Not Uploaded