Intensive Care Foundation Gold Medal Award Presentations Wednesday December 9 2015

摘要

>Purpose: To characterise the temporal stress response to critical illness (encompassing cardiovascular, biochemical, immune, metabolic, bioenergetic and hormonal changes) in an established long-term animal model of sepsis. Comparison was made against data and samples collected from ICU patients. The impact of β-adrenergic blockade on the stress response and mortality was assessed in the animal model.>Background: The biological phenotype of critical illness differs markedly over time. The early acute stress response comprises activation of central and peripheral pathways including immune and neuroendocrine systems, and an associated modulation of haemodynamics, bioenergetics and metabolism. This response is vital in mounting an inflammatory response, trying to maintain perfusion of essential organs, conserving intravascular volume and ensuring availability of substrate for metabolism. Recovery may be slow in some patients who survive this initial phase; this is described as ‘chronic critical illness’ or ‘PICS’ (persistent inflammation, immunosuppression, and catabolism syndrome).¹ Here, patients fail to thrive, remain weak and catabolic, and have a markedly increased susceptibility to new bouts of infection. In such patients mortality and long-term morbidity rates are high.Critical care biology has mainly focused on the inflammatory-immune response. Some studies report modulation of specific hormones or the metabolic response, but there has been no detailed temporal assessment of the ‘systems biology’ of critical illness, encompassing these different systems. This is particularly relevant as good circumstantial evidence supports the notion that both organ failure and PICS may be consequences of excessive stress. The body responds to this allostatic overload by essentially shutting itself down.² We increasingly recognize that patients not only suffer from the (patho)-physiological stress of their underlying illness but also from (i) secondary insults such as nosocomial infection, (ii) numerous psychological stressors related to their stay in intensive care including pain, anxiety and sleep deprivation and (iii) the stressful side-effects of many of the pharmacological interventions inflicted upon them, notably catecholamine therapy. Though the recommended ‘standard-of-care’ for ‘shock’ states,³ catecholamines carry numerous harmful effects, e.g. a markedly pro-thrombotic state, immunosuppression, enhanced bacterial growth and virulence, cardiotoxicity, lipolysis, muscle catabolism and decreased metabolic efficiency.⁴Alleviation of excessive stress may positively impact upon outcomes. β-adrenergic blockade offers such an approach: by attenuating the harmful effects of endogenous and exogenous catecholamines its use may yield significant outcome benefits. A recent randomized study enrolling tachycardic septic shock patients requiring high-dose norepinephrine for >24 hours showed significantly improved recovery of organ function and survival in those receiving an infusion of the β1-blocker esmolol.⁵ Precise mechanisms of benefit, cardiac or extra-cardiac, remain uncertain; arguably, a generalized anti-stressor effect can be postulated. However, this may be a two-edged sword; potential harm cannot be excluded in patients who would otherwise have proceeded to a smooth recovery.My host lab has a well-established fluid-resuscitated, 72-hour, rat model of faecal peritonitis.⁶ Importantly, excellent prognostication (AUC approx. 0.9) can be made as early as 6 h from a low stroke volume and high heart rate, notwithstanding the mild clinical symptomatology at this timepoint.⁷ Animals that die do so between 18 and 36 h while survivors show clear clinical recovery by the 72-h end-point. This model is therefore ideal to assess the temporal stress response in predicted survivors and non-survivors, and to gauge the impact of beta-blockade.>Hypothesis: By counteracting the harmful effects of endogenous catecholamines, β-adrenergic blockade will attenuate an excessive stress response, thereby improving outcomes in critical illness.>Methods: Male Wistar rats (325–375 g), cannulated with tunnelled carotid arterial and internal jugular lines, were injected intraperitoneally with faecal slurry to induce peritonitis. They were then attached to a tether-swivel system, permitting unfettered movement with free access to food and water. Intravenous fluid resuscitation was commenced from 2 h until experiment-end. Transthoracic echocardiography at 6 h classified animals into predicted survivors or non-survivors.⁷In Study 1, 16 animals were observed up to 72 h to confirm the validity of the prognostication.In Study 2, animals (n = 6 per prognostic group per timepoint) were sacrificed at early (6 h), established (24 h) and recovery phases (72 h). Blood samples for hormonal, biochemical, metabolic and inflammatory indices and markers of organ injury were taken to characterise a detailed ‘stress profile’. Organs (kidney, liver, muscle, heart, lung, brain) were removed for metabolomic and transcriptomic analyses. Identically treated control animals were sacrificed at the same timepoints.In Study 3, animals were grouped as predicted survivors (n = 32) or non-survivors (n = 32) and randomised to receive either esmolol or placebo between hours 6 and 24. The dose (500 mcg/kg/min loading dose then 75 mcg/kg/min) was established from pilot studies. Animals were observed until death or study-end (72 h).In Study 4, rats treated with esmolol or placebo were sacrificed at 24 h to examine the effect on their stress profile markers.An observational study in adult ICU patients was undertaken concurrently to characterize the stress response in humans. Patients had either faecal peritonitis or community-acquired pneumonia, within 24 h of hospital admission, with a predicted length of stay >3 days and an initial SOFA score >3. All patients underwent daily blood sampling for the first week, followed by weekly blood sampling until day 28.>Statistics: Previous experience showed that six per group usually demonstrates statistically and clinically significant differences in bio-physiological variables. A sample size of 16 was calculated for survival studies based on an expected mortality rate in non-survivors receiving placebo of 95%, and 45% in those treated with esmolol; power 0.80 and α 0.05.Statistical analyses were performed using SPSS. Normality assessed using Shapiro-Wilk. Parametric data analysed using two-way ANOVA and post-hoc testing. Non-parametric data analysed using Kruskal-Wallis.>Results:Study 1: 7/16 septic rats survived to 72 h. SV and HR at 6 h were good prognosticators (AUC 0.89). id="__p17">Study 2: Significant differences were seen at 6 h between survivors and non-survivors in multiple cardiovascular, inflammatory, hormonal, metabolic and organ injury markers (notably troponin and BNP). The pro-inflammatory response peaked at 6 h while an anti-inflammatory response persisted at 72 h in survivors. A greater shift towards fat metabolism was seen in non-survivors. id="__p18">Study 3: Esmolol halved 72 h mortality in predicted non-survivors and delayed time to death in those that died. However, mortality significantly increased in predicted survivors. id="__p19">Patient study: (samples still being analysed at time of writing). 28-day mortality was 38%, with peak SOFA scores 10.4 (IQR 7–16) in non-survivors and 8.8 (IQR 7–13) in survivors. On Day 1, non-survivors had higher heart rates, higher norepinephrine requirements and, like the animal model, significantly elevated troponin, BNP and cortisol, and reduced HDL-cholesterol. β-blocker use prior to ICU admission was 58% in survivors and 0% in non-survivors. id="__p20" class="p p-last">>Conclusion: Critical illness is associated with early changes in many markers across multiple systems. A similar but accentuated signature was seen in non-survivors compared to survivors, highlighting the effect of an excessive stress response within the first hours of the septic insult. Many similarities were seen between septic rats and patients, supporting the representativeness of this animal model. Esmolol offered significant benefit in prognosticated non-survivors but harm in those predicted to survive; this emphasizes the need to individualize treatment to carefully selected patients.