Further Report From The European Conference On Intensive Care Medicine Barcelona 2006
Author: Dr Karen Stuart-Smith, Consultant Anaesthetist, Glan Clwyd Hospital, Denbighshire, Wales.

Haemodynamic Monitoring Part 2: Stable Doors And Bolting Horses
Author: Dr Karen Stuart-Smith, Consultant Anaesthetist, Glan Clwyd Hospital, Denbighshire, Wales.

Further Report From The European Conference On Intensive Care Medicine Barcelona 2006
Author: Dr Karen Stuart-Smith, Consultant Anaesthetist, Glan Clwyd Hospital, Denbighshire, Wales.

PRO-CON DEBATE:

The International Sepsis Forum: Thanks to recent definitive trials we now know:

Patients with sepsis should be managed with tight glycaemic control

Pro: Greet Van den Berghe, Leuven, Belgium
Con: Konrad Reinhart, Jena, Germany

This debate arose from the original study performed by the Van den Berghe group (hereafter Leuven 11). The hypothesis of the study was that the hyperglycaemia frequently observed in ICU patients, previously thought to be a necessary metabolic response to stress², was in fact detrimental to the patient, and could explain many of the complications observed in the critically ill¹. There was certainly sufficient background evidence in the literature at the time of Leuven 1 to justify such a hypothesis³. As I hope to show later, modern understanding of both the process of insulin resistance and the mechanism of hyperglycaemia-induced cell injury further justifies this approach4. However, as with everything in life, the devil is likely to be in the detail.

The Van den Berghe group performed a prospective randomised controlled trial investigating the effect of ‘normalization’ of blood glucose levels on patient outcome in a single-centre intensive care unit. ‘Normalization’ meant intensive insulin therapy (IIT) which maintained the patient’s glucose level within the extremely narrow range of 4.4-6.1 mmol/l (80-110mg/dl). The control group received conventional insulin therapy (CIT) only if their blood glucose exceeded 215 mg/dl (approximately 11.5 mmol/l), to maintain a glycaemic value of 10-11.1 mmol/l (180-200 mg/dl). In all other respects, the patient population was identical. To cut a long story short, patients in the IIT group were significantly more likely to be discharged alive from the intensive care unit and were also more likely to get out of the hospital alive (fig. 1). Morbidity, as defined by: time on the ventilator, inflammatory markers, development of septicaemia and the occurrence of critical illness-related polyneuropathy, were also significantly reduced in the IIT group. The effect of IIT on morbidity and mortality was so marked that the trial was stopped early.

Figure 1 reproduced with permission from reference 1

The apparent success of this 2001 study begs the question, why was a Pro-Con debate on tight glycaemic control considered necessary at a major Intensive Care meeting in 2006? In a packed meeting hall, Van den Berghe had clearly been placed on the defensive. The ‘Con’ argument was put forward by Konrad Reinhart. The arguments he put forward against tight glycaemic control in intensive care were as follows: 1) a firm belief that acute hyperglycaemia is a necessary adaptive stress response, 2) the tight glycaemic control demanded by the Leuven 1 study is impossible to achieve in the standard intensive care unit, 3) the risks of hypoglycaemia, which caused him to abort his own study, are too great, ⁴) the Leuven 1 study used ‘short’ end-points (i.e. getting out of ITU and then getting out of hospital) and 5) the most recent study from the Van den Berghe group (Leuven 2)⁵ apparently reverses many of the Leuven 1 conclusions. When the vote was taken at the end of a rather bruising debate, some members of the audience were definitely ‘pro’ tight glycaemic control, a very few were absolutely against it, most were equivocal, and the only definite conclusion was that everyone wanted another randomised controlled trial. In other words, the debate changed nobody’s mind in any direction.

The appliance of science.
The two questions that should have been really have been addressed are: 1) why did tight glycaemic control improve outcome in the Leuven 1 study, and 2) what part of the glucose control mechanism has caused the doubt so many years subsequently? I will try to answer this question in two parts-the basic science and the clinical science.

The basic science. Why are sick patients hyperglycaemic? The elevated glucose level occurs irrespective of the patient’s previous diabetic status, and is a direct response to metabolic stress (fig 2). However, as with the diabetic patient, high blood glucose levels are no use if the glucose cannot be utilised properly, and as we will see, severe hyperglycaemia is positively cytotoxic⁶. The idea that high glucose provides energy for repair in the sick patient is as unlikely as the concept of Lucozade aiding recovery from a cold.

Figure 2 reproduced with permission from reference 4

Glucose is transported into cells via two primary mechanisms: insulin-dependent and insulin-independent4,7. Insulin-independent glucose transport is mediated via glucose transporter proteins in the cell membrane, known as GLUT-1, GLUT-2 and GLUT-3. These transporters are predominant in vascular endothelium and smooth muscle, neurons, hepatocytes, pancreatic _ cells and gastrointestinal mucosa. The normal response to hyperglycaemia in cells containing GLUT 1-3 is that the transporters are downregulated and internalised, glucose transport into the cell is slowed, and the cell is protected from glucose overload. In the critically ill patient, cellular markers of sepsis such as cytokines and angiotensin II, as well as local hypoxia, upregulate manufacture of the GLUT transporters and facilitates their localisation in the cell membrane⁷. Transport of glucose into the cell is markedly increased in sepsis and cellular glucose overload occurs.

Why is high intracellular glucose cytotoxic? There are several interrelated reasons6, but for the sake of simplicity I am going to the concentrate on the two mechanisms I think are most relevant to hyperglycaemia in the critically ill rather than diabetes per se. The first is increased activation of a ubiquitous intracellular signalling pathway, the DAG (diacylglycerol)/PKC (protein kinase C) system. This regulatory system is the first step in a multiplicity of normal intracellular functions. Hyperglycaemia-mediated PKC activation results in increased cellular production of a number of sepsis-related markers. Fig. 3 summarises these effects. The reader will recognise the pathological consequences of increased expression of these markers as the clinical expression of sepsis6.

Figure 3 reproduced with permission from reference 6

The second form of cellular toxicity caused by hyperglycaemia is the generation of free radicals, mainly reactive oxygen species (ROS). The principal source of ROS in hyperglycaemia is the mitochondrion, specifically, the mitochondrial electron transport chain6. The normal function of the electron transport train is to process the NADH and FADH2 produced by the TCA cycle (aerobic metabolism of glucose) to form ATP to provide energy. For reasons that are not worth going into here, this process always generates small amounts of ROS even in normal cells, but these potential harmful radicals are mopped up by naturally occurring antioxidants. In conditions of hyperglycaemia, far more glucose enters the TCA cycle, far more NADH and FADH2 are produced. Eventually the electron transport chain is ‘overloaded’ and comes to a halt half way along the chain. Since the system is about electron transport to generate energy, the stalled electrons have to go somewhere, so they are helpfully donated to molecular oxygen via the intermediary coenzyme Q. The simple equation is: O2 + electron –O2– –. This free radical is superoxide, and is the most common ROS. If it is produced in excessive quantities intracellular antioxidants are overwhelmed and free radical damage to cellular mechanisms occurs, ultimately leading to cell death if allowed to continue⁸. Superoxide combines with nitric oxide to form peroxynitrite according to the following equation: O2– – + NO – – ONOO – –. Peroxynitrite (commonly known as OhNo! to those of us in the vascular biology field), is an intracellular poison which causes nitration of several cellular messenger molecules, bringing them to a halt. Nitration of DNA also occurs, preventing the manufacture of cellular proteins. I hope it is clear from this explanation that extra glucose cannot provide additional energy to fuel tissue repair as previously thought. On the contrary, high intracellular glucose cannot be processed by the mitochondria and is channelled instead towards damaging free radicals.

With this knowledge in mind let us extrapolate to the clinical situation. In the critically ill patient, the stress response induces hyperglycaemia. Cells relying on GLUT 1-3 insulin-independent transport import more glucose than they can handle because the GLUT transporters are upregulated by the septic process. These cells are then damaged by the processes described above. The specific cell types involved are: 1) vascular endothelial cells (decreased nitric oxide production, necrotic endothelial cells, leading to platelet sticking and microemboli), 2) vascular smooth muscle cells (loss of vascular tone and resultant hypotension), 3) renal mesangial cells (renal failure), 4) neuronal and Schwann cell damage (intensive care associated polyneuropathy). The clinical pattern of the critically ill patient begins to emerge. These factors probably explain why correction of blood glucose to normal levels improves the outlook in sepsis.

The most obvious way to achieve normoglycaemia is to give insulin, but it is here that the trouble starts. The primary issue underlying objections to the Van den Berghe findings is the enormous difficulty of maintaining normoglycaemia without inducing life-threatening hypoglycaemia. The most recently undertaken multi-centre, randomized, clinical trial studying the effect of tight glycaemic control on outcome in the critically ill, the Glucontrol Study⁹, was stopped after the first interim analysis in March 2006 because of unacceptable rates of hypogylycaemia and unintended protocol violations (as reported at the Barcelona meeting).This study included ‘all-comers’ to the enrolled intensive care units. Patients were divided into two groups. Group A underwent an insulin regime intended to achieve a blood glucose level between 4.4 and 6.1 mmol/. The glucose target for group B was 7.8-10 mmol/l. Because the study was stopped early, it is inevitably very underpowered. On the face of it, however, it appears that Group A patients (tight control) showed a greater trend toward a higher mortality in this study, even if they did not suffer a hypoglycaemic episode. Of those that did develop a low glucose, the death rate among group A patients with severe hypoglycaemia was very significantly greater than among group B patients with a similar event. These are very preliminary data, and a full analysis is awaited. One of the most important issues is whether the two groups were strictly comparable in terms of epidemiological parameters, APACHE II scores, etc, or whether one of the seven centres involved had an excess death rate (or survival rate) which might be skewing the figures. I suspect that one of the biggest problems will be the inclusion of every patient with every conceivable pathology no matter what unless they were under 18, did not consent, were pregnant, or were going to die within 24 hours anyway (hopefully not all at the same time). Given the range of ages, disease, physiological scores and centres involved, I’d hate to be the statistician trying to do a sensible linear regression analysis on all of this. The conclusion presented by this speaker, not unreasonably, was that at present, the risks of tight glycaemic control may outweigh the benefits.

Why is maintaining normoglycaemia without precipitous episodes of damaging hypoglycaemia so difficult in critical care patients? Again, it is important to go back to the basic science, to find out why such large doses of insulin are required, and why a sudden tip into hypoglycaemia may occur. There are only a limited number of tissues which rely on insulin-dependent mechanisms of glucose uptake. These are skeletal and cardiac muscle cells (myocytes) and fat cells (adipocytes). In these cells, a fourth type of glucose transporter, GLUT-4, is stored in vesicles in the cytoplasm. Arrival of insulin at the membrane-located insulin receptor sets in motion intracellular signalling mechanisms which cause the GLUT-4 transporter to move to the cell membrane and initiate the transport of glucose into the cell¹⁰ (Fig.4). When normoglycaemia is restored and insulin levels fall, the GLUT-4 transporter moves out of the cell membrane and returns to its cytosolic stores.

Figure 4 reproduced with permission from reference 10

This is an extremely complicated signalling system and in theory, can be disrupted at any point. The most likely cause of insulin resistance in sepsis (i.e. failure of insulin-dependent glucose uptake in skeletal muscle and adipocytes) is disruption of the insulin receptor itself by cytokines such as tumour necrosis factor alpha (TNF-a) and interleukin-6 (IL-6)^4,11 (Fig. 2). This prevents movement of GLUT-4 to the cell membrane, reducing glucose uptake. Taking a step back to the clinical situation, it is therefore not surprising that tight glycaemic control requires intensive, i.e. high dose, insulin therapy. Insulin receptors impaired by sepsis are likely to have a reduced sensitivity to circulating insulin, and it is my (unsupported) guess that higher doses of insulin are required to achieve sufficient receptor activation, movement of GLUT4 to the cell membrane, and increased glucose transport. Under such conditions, it would not be at all surprising if ratcheting up the insulin infusion to very high levels to obtain tight glycaemic control, especially in the face of stress-induced hyperglycaemia, would lead to a sudden overwhelming activation of all the remaining functional insulin levels, and precipitous hypoglycaemia.

The current message seems to be that if you are a critically ill patient, hyperglycaemia is bad for you, but treating it is potentially just as bad or worse. I’m not sure whether another randomized controlled trial will shed light on the situation, unless a simple method of real-time continuous blood glucose analysis that is cheap, simple and reliable is brought to the market. Very frequent blood glucose measurements significantly reduce hypoglycaemic episodes, but are labour intensive¹².

I’m going to suggest a more boring but more practical alternative. The fundamental problems are 1) the stress response and 2) cytokine-induced impairment of the insulin receptor. Why not 1) provide adequate nutrition, analgesia, and pain relief in the perioperative period¹³ and 2) invest much more heavily in Outreach Teams, admitting early to high dependency care so that septic patients may be treated early and aggressively before that nasty TNF-a has had a chance to roll out and spoil the insulin receptors. All that stress-induced hyperglycaemia might never occur in the first place. Just a thought.

References

1. Van den Berghe et al. Intensive insulin therapy in critically ill patients. NEJM 2001;345:1359-67. Copyright © 2001 Massachusetts Medical Society.
2. Mizock BA. Alterations in carbohydrate metabolism during stress: a review of the literature. Am J Med 1995;98:75-84
3. McCowen KC et al. Stress-induced hyperglycaemia. Crit Care Clin 2001;17:107-24
4. Turina M et al. Diabetes and hyperglycaemia: strict glycaemic control. Crit Care Med 2006;34[Suppl 1]:S291-S300. Fig 1, page S293.
5. Van den Berghe et al. Intensive insulin therapy in the medical ICU. NEJM 2006;354:449-61
6. Brownlee M. The pathobiology of diabetic complications. A unifying mechanism. Diabetes 2005;54:1615-25. Copyright © 2005 American Diabetes Association. Reprinted with permission from The American Diabetes Association.
7. Van den Berghe G. How does blood glucose control with insulin save lives in intensive care? J Clin Invest 2004;114:1187-95
8. Stuart-Smith K. Demystified. Nitric oxide. Mol Pathol 2002;55:360-6
9. Glucontrol Study: Comparing the effects of two glucose regimens by insulin in intensive care unit patients. Clinical Trials Identifier: NCT00107601. www.glucontrol.org
10. Watson RT and Pessin JE. Intracellular organization of insulin signalling and GLUT4 translocation. Recent Prog Horm Res 2001;56:175-93. Copyright 2001, The Endocrine Society
11. de Alvaro C et al. Tumour necrosis factor _ produces insulin resist ance in skeletal muscle by activation of inhibitor _B kinase in a p38 MAPK-dependent manner. J Biol Chem 2004;279:17070-78
12. Meijering S et al. Towards a feasible algorithm for tight glycaemic control in critically ill patients: a systematic review of the literature. Critical Care 2006;10:R19
13. Ljungqvist O et al. Metabolic perioperative management: novel concepts. Curr Opin Crit Care 2005;11:295-9

Haemodynamic Monitoring Part 2: Stable Doors And Bolting Horses
Author: Dr Karen Stuart-Smith, Consultant Anaesthetist, Glan Clwyd Hospital, Denbighshire, Wales.

‘No monitoring device can be expected to improve patient outcome if it is not coupled to a treatment that itself improves outcome’¹

‘A large high-risk surgical population accounts for 12.5% of surgical procedures but for more than 80% of deaths. Despite high mortality rates, fewer than 15% of these patients are admitted to the ICU’ ²

‘…critical care is a concept, not a location, which frequently begins with emergency department (ED) intervention, and culminates in ICU admission and continued management’ ³

‘Patients may languish in the ED, as ICU “boarders” until a bed is available, for up to 24 hours in the hands of an understaffed, overloaded work force that provides adverse nurse/patient ratios’ ⁴

Having stepped into the muddy waters of cardiovascular monitoring in my last article (APN Autumn 2006), I am now about to wade in even deeper and try to answer some more fundamental questions, hopefully without drowning.

These questions are:

1. Do we really understand sepsis?
2. Have we even been even been monitoring the right things in septic patients?
3. If the answer to the first two questions is no, where do we go from here?

Allied to these basic questions is a fourth problem, which is: if sick patients go downhill quickly, why don’t we start therapy before the downward slope becomes a headlong rush to the grave? As we shall see, this is not just a clinical (and ethical) question but a political, social and economic question that affects not just the UK National Health Service but all other healthcare services. I hope the reader will be able to see that the quotes at the beginning of this article reflect these concerns.

Shoemaker re-visited

The first critically ill patients whose haemodynamic disorder was extensively studied were the patient cohorts reported by Shoemaker^5,6,7. Initially heralded as a great step forward in the understanding of shock, Shoemaker’s work was increasingly maligned as subsequent studies appeared to contradict his findings. Although there are some partial truths to these criticisms, as acknowledged by Shoemaker and his colleagues⁸, many of the objections to his results are a matter of interpretation and timing of therapy interventions, not to mention his choice of haemodynamic monitoring device, as we shall see.

Shoemaker had been studying the effects of various types of shock on cardiovascular parameters for more than 20 years, both in laboratory studies and in the clinical setting, before the publication of his most quoted clinical trial on the effect of driving the cardiovascular system to produce ‘supranormal’ values as a survival strategy in critically ill patients5. In this study patients were assigned to one of three treatment groups: 1) patients with CVP line (no specific management protocol), 2) pulmonary artery catheter (PAC) placed and used to attain ‘normal’ haemodynamic indices (cardiac index (CI) 2.8-3.5 L/min, global oxygen transport (TO2) 400-500 ml/min, and oxygen consumption (VO2) 120-140 ml/min 3) PAC protocol group, where fluids and vasopressors were used to obtain supranormal values as follows: CI > 4.5 L/min, TO2 > 600 ml/min, and VO2 > 170 ml/min. The PAC protocol group had dramatically fewer ventilator days, ICU days, and postoperative deaths. Not surprisingly, this translated into a substantial cost saving.

Shoemaker attributed the success of the PAC protocol driven therapy to correction of an oxygen debt that occurs as part of the stress response to surgery6. The increased metabolic demand that accompanies surgery obviously requires greater oxygen delivery to the tissues to fuel mitochondrial ATP production via the tricarboxylic acid cycle. The ATP then provides the energy required for post-operative repair (anabolism) and to preserve immune function. Oxygen delivery to the tissues must be adequate to meet these demands, and this means providing enough circulating volume (including red blood cells) at a sufficiently high rate to satisfy tissue requirements. The key to the success of the early Shoemaker studies is the pre-trial decision to begin resuscitation therapy to supranormal values in the preoperative period or as soon as possible after surgery.

The time issue was very well appreciated by Shoemaker from the beginning, because, of course, he read the literature^6,7. Several subsequent studies showing no apparent benefit of using supranormal variables in the management of high-risk surgical patients and/or patients with sepsis have generally not followed a protocol which includes aggressive resuscitation commenced in the emergency room-a marked difference from the Shoemaker studies (for an excellent discussion of these trials, see reference 1). As the trials were generally conducted by intensivists, protocol driven therapy was not usually commenced until the patient was in the ICU. By then, it is probably too late for many patients, who may have followed the pattern of: getting in the ambulance, lying around in A and E waiting for a surgical opinion (i.e. a quick look by the surgical SHO or trans-Atlantic equivalent), being trundled off to CT, and finally deposited on the ward when a porter can eventually be found. The six ‘golden hours’ for adequate resuscitation have long past by then. Colloquially, it is like walking along a beach and seeing a drowning man out to sea, but waiting six hours to call the coastguard. In an inadequately resuscitated patient, oxygen supply to the mitochondria is far below demand, and an oxygen debt develops. If the debt continues to rise, a point of no return will be reached.

An early and telling experiment quoted by Shoemaker is a study of the relationship between oxygen debt and outcome in a dog model of haemorrhagic shock⁹. Dogs with an oxygen debt < 100 ml/kg survived, those with a debt > 140 ml/kg died. Early aggressive resuscitation provides grateful mitochondria with sufficient oxygen to fuel metabolic demand. Late resuscitation (i.e. delayed until entry to the ICU) means that many of these mitochondria are relying on anaerobic metabolism long before the porter arrives to take the patient to the ward. In these latter circumstances, aggressive management to supranormal values may well be detrimental (see below).

A further issue complicating later studies is the use of the PA catheter. The controversy surrounding use of this monitoring device has plagued Intensive Care for 20 years, and is frankly a red herring. Let’s use this article to throw the red herring back out to sea once and for all. As mentioned in my previous article, the PAC was originally designed by a cardiologist to be used by cardiologists. Shoemaker wanted to drive the circulation to specific values determined by his numerous earlier studies. In the 1980s, the PAC was the only device available to do this precisely. Many authors have misinterpreted Shoemaker’s studies to mean that he implied his patients showed increased survival because of the PAC. This was never his intention. I heard Shoemaker speak back in 1988, and I still remember the talk very well. He has always been very clear that they survived because of the therapy. Many of the studies showing an apparent adverse effect of the PAC have been deeply flawed, often because no allowance was made for the illness severity of the patient (for discussion see references 1 and 10).

More recent studies have suggested that at the very least, the PAC does no additional harm^1,11. Whether it is actually of benefit is an entirely different question. The problem is that we can measure all the variables, but because we do not understand the septic process or the relevance of the values obtained, we cannot use PAC-derived results correctly to manage the problem. It is the exact corollary of the situation in the Hitchhiker’s Guide to the Galaxy, where a huge computer was built to find the answer (42 of course), and then another computer had to be built to find out what the question was. My view is that the

PAC is outdated, but if nothing else is available, it can be used to manage the ill patient, provided an appropriate algorithm is followed (Fig. 1)11. My rationale is set out below. In brief, it means concentrating on oxygen transport and demand as the primary indices of care, and forgetting important-sounding but pointless values like pulmonary artery wedge pressure and systemic vascular resistance, which really are of far greater value to the cardiologist than the intensivist. The cardiologist is concerned with pressure. The intensivist is concerned with volume.

Figure 1 reproduced with permission from reference 11

The final Shoemaker–related issue to be settled is whether the induction of supranormal haemodynamic variables actually kills people. The answer again lies in understanding Shoemaker’s original studies. Most obviously, older patients with other co-morbidities may not be able to increase cardiac output sufficiently to meet demand, and myocardial ischaemia is easily unmasked. The peripheral circulation may already be compromised by artherosclerosis, further limiting oxygen supply. There is inevitably a risk that the vulnerable older patient will be compromised by the therapy. Therefore, a ‘one-size-fits-all’ approach to optimisation in standardised clinical trials will inevitably create a bias against optimisation, a fact recognised by Shoemaker^6,7,8. Some common sense trials are required to determine what really is ‘optimisation’ in different patient groups.

However there is no getting away from the fact that early intervention is key. This is what I mean in the title by stable doors and bolting horses. Really sick patients already in the ICU when aggressive therapy is commenced do not benefit from optimisation therapies^8,12. Mitochondria cannot heal themselves. Prolonged tissue hypoxia leads to generation of reactive oxygen species (for mechanism see accompanying article on glycaemic control in this issue).Pharmacological therapies aimed at increasing the cardiac index increase cardiac work, exacerbating hypoxia and acidosis. Vascular endothelial and smooth muscle damage (discussed below) means that all the norepinephrine in the world is not going to overcome the vasoplegia associated with severe sepsis. Gaps open between the dysfunctional endothelial cells, allowing the movement of fluid into the interstitial space, so aggressive fluid management at this juncture simply results in tissue oedema. I feel depressed just writing this.

Sepsis as a disease of the microcirculation

Sepsis is a disease in which oxygen delivery fails to keep pace with oxygen demand in a setting of increased metabolism. As we have seen, the increased oxygen demand derives from mitochondria in the cells of end organs, where increased energy is required to repair tissue damaged by endotoxins, cytokines and free radicals. The oxygen has to arrive via the circulation, and the guardian of end organ perfusion is the microcirculation. The microcirculation is traditionally defined as vessels less than 100 _m in diameter, and consists of arterioles, capillaries and venules13. Microcirculatory blood flow is impaired in sepsis, so that tissue hypoxia may occur in the presence of apparently sterling macrovascular indices^14,15. The disturbance appears to be a patchy distribution of microvascular constriction and/or occlusion in some parts of the circulatory bed, with normal or increased flow in adjacent areas. Large areas of compromised microcirculatory flow are correlated with poor outcome in septic shock¹⁵.

Figure 2 is a summary of the current understanding of the aetiology of microvascular dysfunction in sepsis¹³. If you think it looks a bit vague, you’re right. The exact cause of the microvascular compromise is not known, although damage to the endothelium as a result of the septic process (resulting in vasoconstriction) and the deposition of microemboli (causing occlusion) are reasonable possibilities. However the microcirculation is under complex regulatory control that is quite different from more proximal arteries, and the role of the endothelium in these very small vessels is incompletely understood, even in normal circumstances. I have not found any definitive papers examining the mechanism of microcirculatory dysfunction in sepsis. This work remains to be done.

Figure 2 reproduced with permission from reference 13

Although the exact cause of the microcirculatory disturbance is not known, the clinical consequences are clear. The most important clinical markers of poor tissue perfusion are blood lactate (> 4 mmol/l indicates severe hypoperfusion), and mixed venous oxygen tension (SvO2). Lactate is a conceptually simple monitor of anaerobic metabolism, whereas SvO2 is dependent on several interacting variables. A reduced SvO2 may indicate that demand has increased and oxygen extraction has risen in the periphery. However, if lactate is low, tissue hypoxia has not yet occurred but will do if oxygen delivery is not improved. Measures to improve oxygen delivery include changes in mechanical ventilation and inspired oxygen to raise SaO2, and increases in cardiac output by optimising haemoglobin, increasing fluid administration (decrease in stroke volume variation, see below) and the use of inotropes. Successful manoeuvres will raise the SvO2.

By contrast, a very low SvO2, accompanied by a high lactate, means that oxygen delivery has fallen below oxygen demand, oxygen extraction rates are maximal, and anaerobic metabolism has begun in cells which are far removed from the remaining perfused capillaries and are therefore beyond a reasonable perfusion distance for oxygen^14,16. This is a grey area in terms of patient survival, and is probably at the borderline at which supranormal cardiac indices will improve outcome. In severe (and probably terminal), SvO2 may paradoxically rise in the face of elevated lactate levels. There are two reasons for this. First, oxygen consumption falls as tissue hypoxia gives way to cell death7. Second, there is shunting of oxygenated blood away from closed microcirculatory units to the few remaining open ones. In addition, lactate from the closed units is deposited into the venous circulation. The result is that the larger venules are hyperoxic and acidotic (Fig 3)¹⁴.

Figure 3 reproduced with permission from reference 14

Haemodynamic monitoring in sepsis

It should be clear from all of the above that the high-risk surgical and/or septic patient must be treated early in order to survive. This is the philosophy of Early Goal Directed Therapy (EGDT). Due to lack of space here I direct the reader to an excellent recent review on this subject by the Rivers group4. The basic principle is to provide enough oxygen quickly enough to supply the enhanced needs of the peripheral tissues. A haemodynamic monitoring device is only useful if it helps the clinician achieve that aim. If it is also assumed that treatment of these patients should start in the emergency room, it must have the added benefit of quick set-up and ease of use.

What should the monitor measure? Cardiac output and cardiac index as markers of myocardial performance and the efficacy of inotropes are a basic necessity, as is an accurate measure of mixed venous oxygen saturation. However, we are also interested in optimising volume, so rather than CVP and pulmonary artery wedge pressure, an accurate indicator of volume deficiency is required. Stroke volume variation (SVV) (i.e. the variation in the height of the pulse pressure with respiration) is one such measure. As this is a basic piece of physiology, I will not go into the mechanism here. Anaesthetists have ‘eyeballed’ the SVV on the arterial pressure trace for many years, but recently developed technology has sought to quantify the SVV more accurately. There are many anecdotal case reports of the efficacy of SVV as a physiological marker, but as yet there are no formal studies. A large randomised controlled trial comparing standard fluid therapy via the CVP line as compared with devices utilising SVV is lacking This article considers two haemodynamic monitors which measure these parameters: the FloTrac/Vigileo™ system (Edwards Lifesciences) and the PiCCO plus™ (Pulsion Medical Systems). Both of these systems use patented algorithms to determine cardiac indices from the arterial pressure trace derived from an arterial line. Most of the information provided below has been obtained direct from the relevant companies, to whom the reader should apply for further information.

The FloTrac™ sensor is attached to the patient’s existing arterial line. The sensor is connected to the Vigileo™ monitor, where data is processed and displayed. On the simple principle that the cardiac output is a direct function of heart rate and stroke volume, pulse rate is measured from the upstroke of the arterial pressure wave, and stroke volume is calculated on the basis that it is directly proportional to the pulse pressure. Cardiac output is then calculated using the Edwards algorithm, termed ‘arterial pressure cardiac output’ or APCO. As vascular tone/compliance affect the shape of the arterial wave form, and in theory could compromise cardiac output calculations, further internal arterial waveform analysis using the patients’ age, height and weight (and hence body surface area) is performed. This latter facility obviates the need for external calibration via thermodilution techniques.

In summary, cardiac parameters measured by this system are: cardiac output/index, stroke volume/index, and stroke volume variation (SVV). (Systemic vascular resistance can also be measured). These are the necessary values for assessment of the adequacy of intravascular volume replacement and the ability of the myocardium to deliver this volume to the periphery. Edwards Lifesciences also manufacture a central venous catheter (PreSep) for continuous SvO2 measurement.

The PiCCO (pulse contour cardiac output) System™ is a well-established intensive care device, which continues to undergo technical modifications and additions. Originally designed as alternative to the placement of a pulmonary artery catheter, the PiCCO™ relies on a thermodilution technique utilising cold or room temperature saline to calibrate the system. The saline is injected into the patient’s own central line. However the temperature change is detected downstream via a dedicated long arterial line with a thermistor at the tip, the placement of which is the only potential disadvantage of the system in an emergency situation. The PiCCO™ algorithm determines cardiac output, stroke volume/volume variation and all related indices. Global ejection fraction is used to assess cardiac contractility, and hence myocardial performance. In addition to assessment of volume adequacy via stroke volume variation, global end-diastolic volume (the total diastolic volume of the entire heart) and intra-thoracic blood volume (total volume of blood in the heart and pulmonary system) can be derived. The PiCCO™ system is also able to measure extravascular lung water (EVLW) but I mention this last for the sake of completeness as the relevance of EVLW is a whole other topic beyond the scope of this article.

The company produces two additional pieces of relevant equipment. These are the CeVOX central venous oxygen sensor (attached to the central line, either the ICU’s own or one manufactured by Pulsion), and the LiMON (Liver MONitor). The LiMON is not currently available in the UK. It utilises the dye indocyanine green, which is metabolised by the liver. The dye is injected intravenously, and its rate of disappearance from the circulation determined from a skin sensor attached to the patient) is determined by liver blood flow and parenchymal function. There is a clear potential to measure the adequacy hepato-splanchnic microcirculation, and I await further developments in this area.

There is no question that both the FloTrac/Vigileo™ and the PiCCO™ systems are accurate and well-validated (literature available from the respective companies). The FloTrac™ measures fewer indices, but is more compact, does not require an additional long arterial catheter, and can be used without external calibration. If we are to begin resuscitation therapy at patient admission, this system has the distinct advantage of ease of use and portability. In other words, a relatively handy anaesthetic SHO could take this system with him to the CT scanner, then to theatre, and ultimately to the ICU. With the best will in the world, I cannot see the PiCCO™ system being set up in Accident and Emergency. Nevertheless, for the sick septic patient, it offers valuable additional measurements that should be clear from the rest of this article. Its place is firmly in the ICU, where it can not only inform patient care but is an invaluable tool for research. As with all these things, choice of monitors is partly a ‘horses for courses’ issue.

This last point, the role of research, is extremely important. I should point out that although both monitoring systems are accurate, there are really no papers addressing their effect on patient outcome. This is very important if we are to avoid the situation the PAC got into, which was that it was inserted more for decoration than use. Proper randomised controlled trials are necessary to set the variables which require treatment and the manner in which they should be treated. Otherwise these systems may be perceived either to ‘not work’ or even cause harm-another trap the PAC fell into. A cautionary tale is a study published recently comparing patients receiving fluid therapy via PAC measurements with those using the PiCCO system¹⁷. Surprisingly patients receiving a PiCCO-managed regime had fewer ventilator-free days and a more positive fluid balance, although overall there were no differences in the major end-points for either group. Nevertheless positive fluid balance was in itself correlated with poorer outcome. The principal flaw of the study is that no standard protocol was set, and I suspect that part of the problem was the full potential of the PiCCO was not being utilised properly. The papers’ authors suggest that the easily available readouts from the PiCCO may have tempted clinicians to ‘meddle’ more in the administration of fluid-an unexpected side-effect of designing user-friendly monitors.

Finally, if we are to apply EGDT in its purest form, invasive monitoring has to be started the moment the patient is admitted to hospital (i.e. crosses the threshold of the A and E department). If this is going to happen, resus rooms in UK accident and emergency rooms are going to have to be bigger, tidier (and cleaner), so that a central line and arterial line can be placed and monitoring commence. The medical and nursing staff also need to able to act on the derived parameters, and this means a radical improvement in accident and emergency training generally. Emergency physicians in the United States recognise these problems, and have signed up to the Surviving Sepsis Campaign, which emphasises early and appropriate therapy. We need to do the same here starting with our very own Royal College of Anaesthetists and Association of Anaesthetists as we are the suckers left with the sick patient in the resus room. The application of EGDT is an ethical as well as clinical necessity. Discuss.

References

1. Hadian M and Pinsky MR. Evidence-based review of the use of the pulmonary artery catheter: impact data and complications. Critical Care 2006;10(Suppl 3):S8
2. Pearse RM et al. Identification and characterisation of the high-risk surgical population in the United Kingdom. Critical Care 2006;10:R81
3. Safr P et al. Emergency and critical care medicine: local implementation of national recommendations. Clin Anesth 1974;10:65-125
4. Otero RM et al. Early goal-directed therapy in severe sepsis and septic shock revisited. Concepts, controversies and contemporary findings. Chest 2006;130:1579-95
5. Shoemaker WC et al. Prospective trial of supranormal values of survivors as therapeutic goals in high risk surgical patients. Chest 1988;94:1176-87
6. Shoemaker WC et al. Hemodynamic and oxygen transport responses in survivors and non-survivors of high-risk surgery. Crit Care Med 1993;21:977-90
7. Shoemaker WC et al. Sequence of physiologic patterns in surgical septic shock. Crit Care Med 1993;21:1876-89
8. Velmahos GC et al. Endpoints of resuscitation of critically injured patients: normal or supranormal? Ann Surg 2000;232:409-18
9. Crowell JW and Smith EE. Oxygen deficit and irreversible hemorrhagic shock. Am J Physiol 1964;106:313-9
10. Sakr Y et al. Use of the pulmonary artery catheter is not associated with worse outcome in the ICU. Chest 2005;128:2722-31
11. Vincent J-L. A reappraisal for the use of pulmonary artery catheters. Critical Care 2006;10(Suppl 3):S1. Source Bio Med Central.
12. Poeze M et al. Meta-analysis of hemodynamic optimisation: relationship to methodological quality. Critical Care 2005;9:R771-9
13. Ince C. The microcirculation is the motor of sepsis. Critical Care 2005;9(Suppl 4):S13-19. Source Bio Med Central.
14. Trzeciak S and Rivers EP. Clinical manifestations of disordered microcirculatory perfusion in severe sepsis. Critical Care 2005;9(Suppl 4):S20-28. Source Bio Med Central.
15. Vincent…
16. Huang Y-C T. Monitoring oxygen delivery in the critically ill. Chest 2005;128:554S-560S
17. Uchino S et al. Pulmonary artery catheter versus pulse contour analysis: a prospective epidemiological study. Critical Care 2006; 10:R174