CON: Without more knowledge these attempts will almost certainly continue to be futile.
David Royston MD FRCA Consultant
Nandor Marczin* MD PhD Senior Lecturer and Hon Consultant
Piroska Tarsoly MD Clinical Research Fellow
Royal Brompton and Harefield NHS Trust
Harefield Hospital, Harefield UK
* Heart Science Center, Imperial College, University of London
The question being addressed in this debate is whether a) there is enough scientific or clinical information to b) allow initiation of specific interventions intended to c) inhibit a perceived inappropriate inflammatory process that is judged to be causally related to an adverse outcome after heart surgery.
Logic suggests we need look at this process in reverse. First we need to agree what to consider as a bad outcome. Second we need to have some consensus as to the etiology or genesis of this outcome so that third, specific targeted therapies can be tested. Critical appraisal of various published data suggests that we have not currently fulfilled satisfactorily any of these tasks or conditions.
What is the Problem the Intervention Will Prevent?
Whereas death is a universal end-point for a bad postoperative outcome there is no consensus as to a precise definition of an unwanted morbidity.
Post-operative organ dysfunction is commonly described under an umbrella definition of systemic inflammatory response syndrome or SIRS. Once injury or 'non-self' has been recognized, hemostatic and immune systems integrate to wall off the insult from the rest of the body and then try to repair or consume the injury or invasion1. The term systemic in SIRS implies the focussed and localized innate response to trauma or invasion has become unleashed on the whole body. Historically the extracorporeal system (and sometimes its driver) has taken the blame for this.
The formal definition of SIRS focussed on classifying sepsis, septic shock, and SIRS associated with sepsis prior to embarking on multicenter studies of potential therapeutic interventions2. The definition includes tachycardia and tachypnoea with changes in temperature and white cell count. However, this definition is not robust after cardiovascular surgery. Most patients will be receiving positive pressure ventilation (removing the respiratory component of the definition) and b-adrenoceptor blocking agents (reducing confidence in the definition of tachycardia). Many patients early after cardiac surgery will have temperatures around 38øC and white counts in the 11-12 x 109/L range. However if the patient is orientated and breathing spontaneously, is well-perfused without vasoactive support and passing adequate volumes of urine they are considered as following a normal post-operative course. It is axiomatic that if you, the reader, had leukocytosis, a mild pyrexia, and possibly a degree of nausea then you would consider yourself abnormal, if not downright ill.
Herein lies the crux of the matter. How can we distinguish a normal "physiological" response from an abnormal, possibly pathological process when the variable is likely on a continuum.
Because of a lack of a universally applicable and accepted definition the current choice of factor(s) put under the umbrella of a SIRS after heart surgery depend heavily on the focus of the caring or investigating team. Carers and clinicians recognize SIRS in a minority of patients who do not follow the anticipated recovery pattern described previously. Their patients may be slow to rewarm, possibly with a mild acidosis, require low doses of inotropes and possibly vasopressors, regular diuretics to maintain an adequate urine output and their oxygen transfer may not be appropriate to allow weaning from the ventilator on the day of surgery. These patients rarely die but
cause everyone involved in their care some concern for a day or two.
Common to most investigational definitions is haemodynamic instability. Some (typically the vascular biologist and endothelial doctors) put this instability down to inappropriate vasodilatation. This has logic, as increased blood flow is a Galenic principle. Other groups (the ischaemia ñ reperfusion doctors) champion myocyte dysfunction as the major focus for the haemodynamic component. Capillary leak is another common but not universal component (despite being pivotal in the Galenic model). This abnormality is often equated to some degree of impaired oxygen transfer in the pulmonary system (despite considerable data to show an increase in lung water is one of the least common causes of impaired oxygenation after major surgery). A minority of investigators consider coagulopathy to be part of this process. Finally and most recently cerebral impairment has been reinstalled in a pivotal position as an unwanted morbidity associated with an inflammatory process.
The importance of knowing what we are trying to prevent or improve has far-reaching consequences especially when considering introducing a therapeutic intervention. The following examples illustrate this problem when therapeutic intervention with one of two separate polyvalent anti-inflammatory agents is considered.
First, let us consider the use of steroids to inhibit pulmonary injury. Conventional wisdom suggests that the shunt and ventilation-perfusion mismatch leading to hypoxemia is due to increased lung fluid in the early stages whereas the repair and fibrotic process are the main factors late on in the process. If steroid therapy is to be initiated then the degree of hypoxemia is less important than the etiology or stage of the disease when deciding the appropriate dose, agent and timing of administration. It is, for example, known that high doses of powerful steroids given in the fibrotic phase do not significantly reduce hypoxemia but may increase mortality due to increased infection risk.
As a second example, the role of aprotinin in reducing mortality associated with heart surgery will be used. A meta-analysis of data from 3212 patients in 26 studies having cardiac surgery3 showed an almost two-fold decrease in mortality (from 2.8% to 1.5%)(odds ratio 0.55 (95% CI; 0.34-0.90)) in those patients given high dose aprotinin therapy. Although the voracity of these data has never been challenged the specific reason for a reduced mortality is currently not established. Aprotinin has been shown to inhibit and modulate many aspects of the inflammatory response, however the licensed indication for administration of the drug is prophylactic use to reduce in transfusions. Rates of resternotomy for bleeding are also consistently reduced3. Both transfusion and re-exploration for bleeding are associated with increase post-operative morbidity4 and mortality5. The effect of prophylactic aprotinin therapy on these end-points probably explains much of the observed benefit on mortality.
In contrast to the data of Levi and colleagues3 an increased mortality has recently been reported in those patients given antifibrinolytic agents (including presumably aprotinin)6. How can we explain this dichotomy? One possibility follows from a survey of transfusion practice in the UK7 were 50% of patients who received haemostatic factors and 75% of those returned to theatre for control of bleeding also received "rescue" doses (140-280mg) as opposed to prophylactic high doses (840mg) of aprotinin. This late administration of ineffective doses of aprotinin does not prevent the transfusion or re-exploration associated mortality that is reduced by high dose prophylaxis3. It will therefore not be surprising if administration of low dose aprotinin given after surgery is associated with an increased mortality.
Both these examples show that the dose and timing of an interventions needs to be targeted at a specific physiological abnormality with a precisely know etiology to maximise patient benefit.
Assuming we accept an imperfect definition of SIRS can we define a etiologies for this unwanted clinical scenario? A critical appraisal would need to address many issues but only two will be discussed. The debate as to why the modern extracorporeal systems plays a minor (if any) part in the genesis of an abnormal response is discussed elsewhere8.
What is the root of the problem? Is there consistent evidence that a specific mediator or biochemical system is causally associated with adverse outcome?
The earliest villain identified in bypass-related SIRS was activation of complement. The trend away from investigating the role of complement activation may be because the modern bypass system may not induce such activity. For example, a recent study9 showed no change in activated fragments of C3 or C5 and a decrease in C4 activation with the extracorporeal system used during that study.
If complement activation is no longer an issue how can we be confident that the currently championed biochemical measures are not simply a surrogate for the real villain and therefore will not be an appropriate target for intervention? Cytokine responses will be discussed out of the massive array of potential culprits as they represent a commonly investigated culprit system. At low concentrations cytokines are essential for optimal function of the defence and repair systems of the body and have been found in nature for over 400 million years1. However the high concentrations of cytokines measured in association with sepsis, trauma and heart surgery are singled out as deleterious. The protagonists of this belief (the SIRS Sleuths) suggest modulating this response will be of benefit to the patient. The question is whether the evidence consistently supports this notion?
What should we measure? There are more than 140 cytokines and chemokines recognised in nature. This makes the choice of the "culprit" to measure somewhat of a lottery. The multifaceted activity of TNF (and possibly the fact that the kit for measurement in blood was one of the first commercially available!) made this a focal target for investigation. Unfortunately there are as many studies that show TNF concentrations rise in association with heart surgery as there are those failing to demonstrate this effect.
When should we measure it? In addition to the presence or absence of a response, the timing of that response must be considered. The study by Schmartz9 suggests the peak in IL-8 concentration is some hours after bypass. However, 6 other studies measured the peak in IL-8 concentration at the end of surgery or shortly after the release of the aortic cross clamp8.In the search for causal relationships between inflammatory markers and outcome, this disparity suggests that a peak or trough can be missed.
The next question is whether a rise in a pro-inflammatory or fall in anti-inflammatory marker such as a cytokine will universally precede or produce an adverse outcome? This is currently the weakest of the SIRS Sleuth arguments.
Again Three Examples Serve to Illustrate their Problem.
First a National Library of Medicine search flags over 1500 publications on the topic of cytokines and heart surgery. In only a handful is the biochemical data related to an index of patient outcome and in those studies the biological relevance of the relationship is usually weak. As an example, IL-8 concentrations have been related to cardiac Troponin-I (cTn-I) as an index of myocyte injury10. Although the relationship was statistically significant two thirds of the rise in cTn-I was explained by factors other than the change in plasma IL-8.
Second, based on early evidence showing patients with neutralising antibodies to endotoxin have less morbidity after heart surgery11 then inhibiting any effects of endotoxin may seem a reasonable approach. Administration of low doses of endotoxin to volunteers will produce an inflammatory response that includes increased concentration of TNF and IL-6. If these volunteers also received ibuprofen then the clinical effects of the endotoxin infusion were inhibited and they felt much better. However administration of ibuprofen also significantly (2-3 fold) increased the rise in both TNF (to 627ñ136pg/mL) and IL-6 (to 113ñ66ng/mL) concentrations induced by the administration of endotoxin alone12,13. As an exercise the reader should consider how they would persuade their local review board to use ibuprofen to improve patient well being when concentrations of markers consistently quoted as representing evil and badness are increased further by this intervention!
Finally, and on the same lines the more-is-worse therefore less-is-better thinking may not be appropriate in other circumstances. For example increased mortality with septicaemia and meningococcal disease is associated with a lack of TNF response and an exaggerated IL-10 effect14. This pattern has also been shown in a separate study of 464 patients presenting to hospital with fever (> 38.2 øC)15. Concentrations of IL-10 were significantly higher in the blood of the 33 patients who died compared to survivors. These data suggest that interventions intended to suppress the "proinflammatory" (such as TNF) and augment the "anti-inflammatory" (e.g. IL-10) response may not have the intended effect on outcome.
Where can we go from here?
The first logical step appears to be to persuade investigators (and their sponsors!) to stop considering biochemistry in isolation. Although cytokines have been used as the example the same arguments can be developed for a range of initiators, amplifiers and effectors. If the biochemistry is not in some way related to a clinically relevant outcome measure is it really of any patient, as opposed to intellectual interest? This approach will require a consensus of what makes up appropriate clinical end points. The obvious starting point is defining a patient state that makes their responsible physician feel comfortable. When we have an agreed definition in place it will then be possible to determine those factors considered important in the initiation of these unwanted outcomes.
Second, the more-is-bad/less-is-better model needs to be reassessed. We appear to need investigations that assess the overall balance of a process rather than focussing on a specific mediator. There are, for example, some preliminary data to show that cytokine balance after heart surgery always leans toward the anti-inflammatory or repair process16.
Third, current literature suggests that attempting to inhibit a response (assessed via the investigators choice of an infinite number of mediators) does not easily or universally translate into improved clinical outcome. Without more knowledge these attempts will almost certainly continue to be futile.
- 1. Royston D. The evolution of coagulation and inflammation. In: BD Spiess, editor. The Relationship between Coagulation, Inflammation and the Endothelium-A pyramid towards outcome.A Society of Cardiovascular Anesthesiologists Monograph. Philadelphia; 2000. p. 1-30.
- 2. Muckart DJ, Bhagwanjee S. American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference definitions of the systemic inflammatory response syndrome and allied disorders in relation to critically injured patients. Crit Care Med 1997;25(11):1789-95.
- 3. Levi M, Cromheecke ME, de Jonge E, Prins MH, de Mol BJ, Briet E, et al. Pharmacological strategies to decrease excessive blood loss in cardiac surgery: a meta-analysis of clinically relevant endpoints. Lancet 1999;354(9194):1940-7.
- 4. Speiss BD. Transfusion and outcome in heart surgery. Ann Thorac Surg 2002;74(4):986-7.
- 5. Moulton MJ, Creswell LL, Mackey ME, Cox JL, Rosenbloom M. Reexploration for bleeding is a risk factor for adverse outcomes after cardiac operations. J Thorac Cardiovasc Surg 1996;111(5):1037-46.
- 6. Mangano DT. Aspirin and mortality from coronary bypass surgery. N Engl J Med 2002;347(17):1309-17.
- 7. Wright J, Carbery C, Royston D. Transfusion of Red Cell and Hemostatic Component Therapy Is Highly Variable in UK Cardiac Surgical Units Despite Published Guidelines. Anesthesiology 2003;99.
- 8. Royston D, Kovesi T, Marczin N. The unwanted response to cardiac surgery: Time for a reappraisal? J Thorac Cardiovasc Surg 2003;125(1):32-5.
- 9. Schmartz D, Tabardel Y, Preiser JC, Barvais L, d'Hollander A, Duchateau J, et al. Does aprotinin influence the inflammatory response to cardiopulmonary bypass in patients? J Thorac Cardiovasc Surg 2003;125(1):184-90.
- 10. Wan S, Yim AP. Cytokines in myocardial injury: impact on cardiac surgical approach. Eur J Cardiothorac Surg 1999;16(Suppl 1):S107-11.
- 11. Bennett-Guerrero E, Ayuso L, Hamilton-Davies C, White WD, Barclay GR, Smith PK, et al. Relationship of preoperative antiendotoxin core antibodies and adverse outcomes following cardiac surgery. Jama 1997;277(8):646-50.
- 12. Spinas GA, Bloesch D, Keller U, Zimmerli W, Cammisuli S. Pretreatment with ibuprofen augments circulating tumor necrosis factor-alpha, interleukin-6, and elastase during acute endotoxinemia. J Infect Dis 1991;163(1):89-95.
- 13. McAdam BF, Mardini IA, Habib A, Burke A, Lawson JA, Kapoor S, et al. Effect of regulated expression of human cyclooxygenase isoforms on eicosanoid and isoeicosanoid production in inflammation. J Clin Invest 2000;105(10):1473-82.
- 14. Westendorp RG, Langermans JA, Huizinga TW, Elouali AH, Verweij CL, Boomsma DI, et al. Genetic influence on cytokine production and fatal meningococcal disease. Lancet 1997;349(9046):170-3.
- 15. van Dissel JT, van Langevelde P, Westendorp RG, Kwappenberg K, Frolich M. Anti-inflammatory cytokine profile and mortality in febrile patients. Lancet 1998;351(9107):950-3.
- 16. Marczin N, Hoare J, Al-Ruzzeh S, Amrani M, Royston D, Yacoub M. Despite NFkb activation in circulating monocytes following cardiac surgery with cardiopulmonary bypass the net plasma bioactivity is anti-inflammatory. Anesth Analg 2003;96:SCA 56.