The ischaemia reperfusion injury

While we have concentrated, in the past, on the ischaemia reperfusion injuries that the myocardium receives as a result of aortic cross-clamping, other organs are by no means immune from this phenomenon. Circulatory arrest, and low flow cardiopulmonary bypass, has demonstrable adverse effects in virtually every tissue, but the effects are most often manifest clinically as dysfunction of the brain, gastrointestinal tract, endothelium, and lungs.²

While many strategies have been shown to modify the initial ischaemic insult, there is no accepted standard of care. There are almost as many protocols for intraoperative protection as there are cardiothoracic units, supporting the notion that we are a long way from a definitive method. Some degree of end-organ dysfunction, therefore, is almost inevitable. For example, Rajiv Chaturvedi, from our group, using a conductance catheter, showed that a significant reduction in left ventricular systolic function is demonstrable even after relatively simple and short-lived surgery requiring cardiopulmonary bypass.³ The Boston group has shown us⁴ that the probability of neurologic abnormality increases linearly with the duration of circulatory arrest, and that pulmonary endothelial dysfunction is common after cardiopulmonary bypass.⁵ Indeed, the pulmonary endothelium probably takes the biggest “hit” of all of the organs, as there is no antegrade flow throughout the entire period of bypass. We have recently shown that the pulmonary endothelial “stunning” that results can be reversed by increased provision of substrate and an endogenous stimulator of nitric oxide release, substance P,⁶ and more recently by intravenous sildenafil.⁷ These data however, serve to illustrate the potential for more widespread endothelial dysfunction throughout all of the organs.²

It would be inappropriate not to acknowledge the major improvements in the techniques of cardiopulmonary bypass that have taken place over its 50 years in clinical practice. Refinements continue. We are already seeing a trend towards additional regional perfusion to protect against brain injury.⁸ A recent study of 30 infants from Japan⁹ shows the potential benefits of continuous perfusion of the lungs during cardiopulmonary bypass. By avoiding pulmonary endothelial ischaemia, the amount of injury to the lung was markedly attenuated, and there was a reduction in the early fall in the circulating neutrophil count. The latter almost certainly reflects reduced sequestration of leukocytes into the lung during the period of reperfusion, and may be a surrogate marker for reduced pulmonary inflammation.

It is worth emphasizing that much of the injury associated with ischaemia and reperfusion occurs during reperfusion. Reperfusion injury results from a complex relationship between the damaged endothelium, the neutrophil, and platelets (Fig. 2), each element being required for full expression of the phenomenon. Experimental depletion of neutrophils, for example, markedly attenuates the myocar-dial injury associated with even prolonged regional ischaemia.¹⁰ The last 10 years or so have had an enormous increase in our understanding of the reperfusion injury and its determinants. There have also been many experimental studies suggesting potential benefit from modification of the key players in the injury (Fig. 3). Antioxidants, l-arginine, donors of nitric oxide, modification of the complement cascade, and blockade of p-selectin, are just a few of the reported strategies.¹¹ While showing huge promise in experimental preparations, none have reached the position of routine use in clinical practice. A brief look at some of the signaling pathways within the leukocyte, for example, might explain why a focused, single pathway approach may ultimately be unsuccessful (Fig. 4). There are multiple potential receptors on the surface of the cell, multiple potential effectors to those receptors, and many intercellular pathways. This redundancy suggests that “knockout” of one pathway may fail significantly to affect the overall cascade of activation of neutrophils and inflammation. Strategies to modify this process may, therefore, need to be more generic in their approach.

Figure 2. The components of ischaemia-reperfusion injury. Activated endothelial cells, circulating leukocytes and platelets are all integral to the process (figure drawn by Dr Igor Konstantinov. Dr Konstantinov also drew Figs 4, 10 and 11).

Figure 3. The process of ischaemia-reperfusion injury. Many different strategies have been proposed to modify the reperfusion injury after cardiac surgery.

Figure 4. Schematic to show examples of the many different neutrophil-signaling pathways involved in postoperative inflammatory responses (see also Fig. 11).

One such treatment is modified ultrafiltration during the final stages of cardiopulmonary bypass. Championed by Martin Elliott at Great Ormond Street in the early 1990s, the technique has been embraced by many units around the world. The evidence for its effectiveness, early after bypass, is compelling. A decrease in the total content of water in the body,¹² reduced total loss of blood, decreased pulmonary resistance, and improvements in pulmonary¹³ and myocardial function,¹⁴ have all been demonstrated beyond doubt. Its impact, on that basis alone, has been enormous. Unfortunately, the promise of modifying the subsequent secondary insult, perhaps as a result of removal of pro-inflammatory cytokines, has not stood the test of time. Although levels may fall acutely, the subsequent profile of release of inflammatory cytokines appears to be unchanged.¹⁵ Furthermore, our clinical experience goes along with the fact that we are far from having abolished the secondary inflammatory insult that occurs after cardiac surgery.

The secondary insult

Anyone who has worked on an Intensive Care Unit, looking after children after cardiac surgery, is well aware of the postoperative “slump” that occurs between 6 and 12 h postoperatively. The data collected by Gil Wernosky concerning the experience at Boston with the arterial switch procedure provided data to mirror our clinical impressions.¹⁶ He showed that, in neonates with transposition, there was an approximate 25% fall in cardiac output after the arterial switch, reaching a nadir at 9 to 12 h, but returning to baseline by 24 h. There may be many different physiologic mechanisms, depending on the underlying disease and intraoperative events, but several studies have indicated diastolic dysfunction, both of the left ventricle,¹⁷ and of the right ventricle¹⁸ as culprit physiologic abnormalities. On the left side, mid-diastolic reversal of flow, measured by pulsed-wave Doppler interrogation, was associated with markedly increased mortality in patients with left-sided obstructive lesions.¹⁷ The phenomenon of restrictive right ventricular diastolic physiology after surgical repair of tetralogy of Fallot is now well-described.¹⁸ Almost certainly having its origins in intraoperative myocardial damage,¹⁹ it is associated with decreased cardiac output, raised central filling pressures, and retention of fluid. Interestingly, it has been more difficult to demonstrate a relationship between postoperative outcomes and systolic dysfunction. This is because the complex haemodynamic milieu conspires against traditional methods for assessing ventricular contractile performance in humans, since most of the methods are as dependent on the conditions of loading as they are on intrinsic myocardial performance. It is clear, nonetheless, that there is a significant direct effect of the products of the postoperative syndrome of the systemic inflammatory response on myocardial performance. For example, in a study of neonatal pigs,²⁰ we were able to show that inhibition of nitric oxide, using N-monomethyl-l-arginine, produced a dosedependant increase in ventricular elastance measured by conductance catheterization. The advantage of this invasive method lies in its ability to dissect out intrinsic myocardial functional properties from the confounding effects of changes in ventricular and pre and afterload. While conductance catheterization is clearly not an appropriate technique for clinical evaluation, there are newer methods that may provide similar insights. Tissue Doppler echocardiography, for example, is likely to allow a more robust assessment of ventricular performance than any previous non-invasive measure. We had been particularly interested in the measurement of isovolumic myocardial acceleration, occurring in the immediate pre-ejection period.^{21, 22} In both the right and left ventricle, isovolumic acceleration appears to be highly sensitive to changes in inotropy, but very resistant to changes in afterload. Perhaps more surprisingly, it is also resistant to pre-load. It is simple to measure, requiring only a four-chamber echocardiographic section, and can be measured repeatedly at the bedside. Importantly, it reacts instantaneously to changing contractile function. Figure 5 shows a tissue Doppler tracing obtained during left ventricular alternans induced by pacing. While other tissue Doppler indexes show no consistent change, isovolumic acceleration varies beat-by-beat in alternating and sequential fashion. This remarkable property has provided us with a non-invasive method of evaluating ventricular force frequency relationships, a fundamental property of myocardial performance. The normal heart develops increased force with increasing heart rate until a critical heart rate is reached, specifically the optimal heart rate, beyond which generation of force starts to fall. Studies in dilated cardiomyopathy, for example, have demonstrated markedly abnormal relationships between force and frequency whereby both the change in force and the optimal heart rate are markedly reduced.²³ Our preliminary studies in children after cardiopulmonary bypass show an enormous variation in the patterns of the postoperative force–frequency relationship. Some children show marked reduction in generation of force with relatively modest tachycardia, while others have a virtually normal response. Neonates, in particular, show a profound attenuation of the normal rise in development of force with decreasing R-R intervals (Fig. 6). While obviously preliminary, and certainly reflecting only one element of the postoperative haemodynamic scene, such data will perhaps allow us better to understand some of the reasons for the rather idiosyncratic responses of individual patients to changes in their heart rate. The ultimate goal, of course, is to use these measurements to “fine tune” and optimize the function of the myocardium, according to its current inotropic sensitivity, and heart-rate dependency.

Figure 5. (A) Typical myocardial velocity map from normal left ventricle. Note the presystolic “spike” of acceleration and deceleration occuring during the isovolumic contraction period. (B) Simultaneous tissue Doppler recording and high-fidelity ventricular pressure trace. Isovolumic acceleration begins at the earliest point of pressure development in the ventricle. (C) Tissue Doppler recording of left ventricular velocities during pacing-induced alternans. While other indices show no consistent change, isovolumic velocity and acceleration (arrowed) show alternating and sequential variations.

Figure 6. Force–frequency curves from children after cardiac surgery. There is marked variation in the pattern and amplitude of responses (see text for discussion).

Treatment of the secondary insult

The postoperative use of inotropic agents, vasodilators, and inodilators has become ubiquitous after cardiac surgery. While it would be unfair to suggest that their effect is not generally positive, there is perhaps more data available in the literature, to the contrary. In an experimental study using Adrenaline, albeit at a high dose of 2 μg/kl/min, there was an initial marked improvement in myocardial contractility, followed rapidly by a spiralling decline as a result of toxic myocytic damage.²⁴ This was particularly prominent in the neonatal myocardium that exhibited destruction of the sarcolemma, and deposition of calcium granules with swelling of the mitochondria. Even under the circumstances of sustained improvements of cardiac output, the affect of beta-sympathomimetic agents may be counterintuitive. Penny and coworkers showed in neonatal lambs that, despite a doubling of cardiac output with dobutamine, and a marked improvement in ventricular contractility as manifested by improved dP/dt, the increase in systemic delivery of oxygen was more than off-set by an increase in systemic consumption.²⁵ The latter, presumably driven by sympathetic effects on metabolically active tissue such as brown fat, led to an adverse ratio of extraction of oxygen at all doses above 10 μg/kg/m², offsetting any benefits of increased cardiac output.

The current “state of the art” for the use of inotropes in postoperative cardiac intensive care comes from the Primacorp Study. Prophylactic milrinone administered in a double blind placebo controlled study of 238 patients led to a reduction in episodes of low cardiac output in the treated patients.²⁶ Unfortunately, overall, the duration of positive pressure ventilation, stay in the intensive care unit, and length of hospital stay, were unchanged by this treatment.

One might reasonably argue, therefore, that the current conceptual approach, which aims to increase delivery of oxygen to the failing circulation, has failed to provide an adequate answer to the problems of the “secondary hit”. Another way of approaching the balance between consumption of and demand for oxyegn is to modify the consumption side of the equation. We have recently shown that postoperative pyrexia increases consumption by approximately 11% for every degree Celsius over 37°.²⁷ A postoperative pyrexia of 40°C will increase delivery of oxygen, and the required cardiac output, by almost one-third. Conversely, one might expect that rigorous attention to euthermia might decrease consumption, and therefore the required cardiac output, under these circumstances. This is indeed the case. Our data recently obtained show, for the first time, a falling consumption of oxygen in the first 24 h postoperatively whenever we are able to maintain euthermia.

To continue this counter-conceptual argument, one might suggest that other measures to reduce consumption of oxygen might obviate the need for some, or all, of the therapies used to increase delivery. Hibernating animals reduce their cardiac output to one tenth of the resting levels seen in the non-hibernating state. This reduced cardiac output, and therefore delivery of oxygen, is appropriately matched to their reduced consumption. Although “science fiction” at present, one might anticipate the day when, as an alternative to extracorporeal membrane oxygenation, we may institute corporal hibernation, reducing the metabolic demands of the body to match the cardiac output achieved by the failing heart after surgery. Hibernation, however, is a complex physiologic process, probably driven by exogenous effects such as hypoxic and hypercapinic cellular anabolism as well as endogenous drivers such as the effects of cerebral and circulating delta- and Leu-enkephalins.²⁸ Further research in this interesting area is more than justified, and recently a tantalising potential link between hibernation and the next topic of discussion, ischaemic preconditioning, has been proposed.²⁹

Ischaemic preconditioning

Hibernation, discussed in the previous section, is a natural mechanism designed to protect against an adverse environmental change. Ischaemic preconditioning is another natural mechanism of defense, in this case the most potent protective mechanism known to reduce ischaemia-reperfusion injury. Its effect has been described in myocardium, brain, lung, liver, kidney, retina and skin flaps.³⁰ A short burst, or bursts, of sublethal ischaemia to the target organ protects that organ against a subsequent prolonged, and potentially lethal, episode of ischaemia. The process is remarkably effective. Taking the myocardium as an example, transient occlusion of the main stem of the left coronary artery protects against subsequent prolonged occlusion of the anterior descending artery, such that myocardial infarction is reduced by over two thirds.³¹ Furthermore, the last 15 years of research has seen an enormous increase in our understanding of this phenomenon. There is an early window of protection that lasts for approximately two hours, which is followed by a second, or late, window of protection some 24 h later. There are many effectors of preconditioning. Physical, neural, and circulating agents such as adenosine, opiates, tumour necrosis factor alpha, and radicals of oxygen have all been shown to mediate ischaemic preconditioning.³² Their effect is ultimately dependant on opening of mitrochondrial potassium channels sensitive to adenosine triphosphate. While there is a burgeoning literature demonstrating the potential beneficial effects of preconditioning, there are very few clinical papers supporting its clinical utility. This is because of the difficulty in producing transient regional ischaemia of the target organ. Thus, while some benefit has been shown with transient myocardial ischaemia prior to myocardial re-vascularization surgery,³³ it has not established a widespread clinical role.

A more interesting, and potentially clinically applicable, form of preconditioning is so-called remote ischaemic preconditioning. This phenomenon was suggested from an experiment where it was found that transient ischaemia of the circumflex coronary artery protected against subsequent potentially lethal ischaemia in the territory supplied by the anterior descending coronary artery, the degree of protection approaching that of local preconditioning.³⁴ Studies in rats then demonstrated that truly remote preconditioning could be provided by both splanchnic and renal ischaemia, providing a similar level of myocardial protection.³⁵ While interesting, none of these stimuluses have a particular clinical relevance. We have recently shown that a stimulus that is easily applicable in a clinical situation may provide a similar level of remote protection. Raj Kharbanda, from our group, showed that ischaemia produced simply by inflating a blood pressure cuff on the upper arm proved to be a powerful protector of human endothelial dysfunction induced by prolonged ischaemia.³⁶ In the same paper, we described the marked attenuation of myocardial infarction in a porcine model subjected to lower limb ischaemia prior to occlusion of the anterior interventricular artery (Fig. 7). These data set the scene for a study of this method of remote ischaemic preconditioning in the setting of cardiopulmonary bypass. In a recently completed study, pigs weighing 15 kg were exposed to 3 h of cardiopulmonary bypass with 2 h of aortic cross-clamping. The animals were followed for six hours postoperatively in order to assess both the primary and secondary effects of cardiopulmonary bypass. The results, in terms of myocardial function, were encouraging. All control animals required inotropic support in order to be weaned from cardiopulmonary bypass, and two of the six died before the end of the study. Conversely, all of the preconditioned animals survived, all weaned from bypass without the need for inotropic support, and only three still needed inotropic agents at the end of the six-hour period. Going along with this, there was a highly significant difference in postoperative measurements of troponin (Fig. 8), and a measurable difference in diastolic function. There were also significant differences in plasma lactate, perhaps reflecting improved the general hemodynamic and metabolic state, and protein S100B, indicative of reduced neural injury. Perhaps the most dramatic effect was in terms of pulmonary function. Airway resistance was significantly improved, as was pulmonary compliance (Fig. 9). Consequently, these animals required lower inspiratory airway pressures, and had improved arterial oxygenation. A randomized, blinded, clinical study is now underway.

Figure 7. Porcine myocardium stained with flourescein (green area, left hand panel) to demonstrate for area at risk (non green-stained area), and tetrazolium to demonstrate extent of infarction (pale staining areas in right hand panel). The control (a) animal suffered infarction in 70% of the area at risk, the animal pretreated by remote ischaemic preconditioning (rIPC, b) suffered infarction in 13% of the area at risk. The graph (c) shows the overall results (see text for details). [squarf ]: Control (8); □: rIPC (9).

Figure 8. Levels of troponin after experimental cardiopulmonary bypass in control animals (squares) and those treated with prior remote ischaemic preconditioning (triangles). Release of troponin is markedly attenuated in the treated animals.

Figure 9. Lung function after experimental cardiopulmonary bypass. Airway resistance (a) and compliance (b) is superior in those animals treated with prior remote preconditioning. [squarf ]: pre-cond; □: control.

How are these effects mediated?

In order to explore the possible role of the neutrophil, the genetic responses to this simple ischaemic preconditioning stimulus have been examined in terms of gene expression. Using the affymetrx 22K gene chip, these responses were examined under control conditions, and at 15 min and 24 h after preconditioning the arm. It is inappropriate to go into detail in this text, but there is a remarkably consistent down regulation of approximately 300 genes. Figure 10 shows the affect on genes within the leukocyte, responsible for trafficking of neutrophils. Figure 11 shows those genes down regulated within the pro-inflammatory intercellular pathway. What is clear is that there is potential for modifying a whole variety of potential pro-inflammatory mechanisms within the neutrophil. This is in contradistinction to the interventions of a single pathway by means of pharmacologic probes, for example. This may explain its widespread and potentially clinically relevant effect. Nonetheless, these data must be taken in context. While they go along with our experimental observations, these genetic modifications have not been correlated with changes in production of proteins, nor relevant functional change. Furthermore, the optimal signal, the duration of its effect, and the ability to modify clinical outcomes, remains to be determined.

Figure 10. Schematic to show effect of the remote ischaemic preconditioning stimulus on genes (highlighted in green) responsible for cell-surface and intracellular proteins in the cellular components of ischaemia reperfusion injury (see also Fig. 2).

Figure 11. Schematic to show effect of the remote ischaemic preconditioning stimulus on genes (highlighted in green) responsible for cell-surface and intracellular proteins in the circulating neutrophil (see also Fig. 3).

Conclusion

Cardiac surgery has come a long way in 50 years. Surgical mortality has reached extremely low levels. Our attention should now turn to modifying early postoperative morbidity and, in turn, late outcomes. It is likely that one will go hand and glove with the other. The more we can protect the heart and other organs during cardiac surgery, the better the later outcome.

Traditional methods and techniques have failed to produce a sea change in the morbidity associated with cardiac surgery. It may be that the future lies in harnessing the natural mechanisms that developed through evolution to protect us.

Acknowledgments

The research presented in this lecture and review was supported by the British Heart Foundation, The Heart and Stroke Foundation of Canada, and the Canadian Institutes of Health Research. I would like to acknowledge the work of Dr Jia Li, Dr Rajiv Chaturvedi, Dr Michael Cheung, Drs Rajesh Karbanda and Raymond MacAllister, Dr Igor Konstantinov, Professor Michael Vogel, and my many surgical colleagues at The Royal Brompton Hospital, Great Ormond Street Hospital, and The Hospital for Sick Children Toronto, without whom very little of the work presented in my lecture, and this manuscript, would have been possible.