Matt Harmon

59 Chapter three day 2 18.6 vs 30.8 pg/ml, P =0.001, day 4 17.1 vs 29.0 pg/ml, P =0.001), whereas the other host response biomarkers, including angiopoietin-2 and the ratio of angiopoeietin-2/angiopoietin-1 levels, were not different between groups. There were no differences in soluble ICAM-1, soluble E-selectin and angiopoietin-1 between hypothermic and nonhypothermic patients. Discussion Hypothermia at ICU admission is independently associated with adverse outcome in patients with sepsis. In this extensive evaluation of the immune response in hypothermic sepsis, the host immune response was not altered in patients with hypothermia compared to nonhypothermic patients. The endothelial activation marker fractalkine was persistently higher in hypothermic sepsis, irrespective of disease severity. In addition, low BMI, hypertension and cardiovascular insufficiency were identified as risk factors for hypothermic sepsis. Taken together, this data may suggest that vascular dysfunction could play a role in hypothermic sepsis. An excessive anti-inflammatory response has been proposed as a mechanism for hypothermia. 9 In line with this, a recent study showed increased immunosuppression in hypothermic patients 10 , thereby potentially accounting for the association with adverse outcome. 3-5,10 In contrast, we found no difference in either proinflammatory or anti-inflammatory cytokines between hypothermic and nonhypothermic patients, even after correction for disease severity. These data are in line with a study in hypothermic patients showing no difference in circulating levels of proinflammatory cytokines 5 and extend these data by showing that levels of anti-inflammatory are also not affected by hypothermia. Moreover, whole blood stimulations resulted in similar cytokine release in hypothermic versus nonhypothermic patients. Therefore a mechanism for hypothermia directly involving anti-inflammatory cytokines seems unlikely. Rather, these data suggest that hypothermic patients do not suffer from increased immunosuppression. In support of this, the incidence of ICU- acquired infections was similar between groups, as found before in sepsis. 22 Of note, according to standard cell stimulation protocols experiments were performed ex vivo at an incubation temperature of 37°C. If whole blood stimulations had been performed at a lower temperature to simulate the temperature of hypothermic patients, one could expect that overall the reactivity of cells in terms of cytokine production might have been slightly higher. 23 In this systematic study of risk factors for hypothermia during sepsis we identified several interesting associations. BMI was inversely correlated with hypothermia . A physiological explanation is that increased body mass likely slows the dissipation of heat from the body. As low BMI itself has been associated with