The use of hypothermic circulatory arrest with cardiopulmonary bypass for warming and/or cooling of infants with congenital cardiac disease, was in large part, pioneered in Japan in the 1960s by Mori and Horiuchi and others.² But the impact of the methodology of hypothermic circulatory arrest in a fashion similar to that which is used today for small infants is largely due to the efforts of Sir Brian Barratt-Boyes in Auckland, New Zealand, in the very early 1970s.³ These methods were then adopted, and further advanced, in the United States of America by Aldo Castañeda, working in Boston. Why did they resort to the use of hypothermic circulatory arrest in small infants? Well, if you were training in cardiac surgery in the 1970s, you would of necessity be influenced by the teachings of John Kirklin, the Professor of Cardiothoracic Surgery first at the Mayo Clinic, and then the University of Alabama in Birmingham, who really was the academic leader of cardiac surgery in North America. Kirklin made the strong argument that infants and very small children responded so adversely to the use of the heart-lung machine, and cardiopulmonary bypass, that intracardiac heart repairs in the first few years of life were considered very risky. Hence, they were rarely undertaken. Palliative operations that could be performed without the heart lung machine were the rule, and definitive repairs were generally postponed until palliated infants had reached the age of at least a few years. Kirklin's observations, of course, were accurate. Cardiopulmonary bypass in the form that it existed three decades ago, and in the fashion that it was used three decades ago, did evoke such an enormous inflammatory response in small babies as to be associated with prohibitively high levels of morbidity and mortality. These were related to activation of white blood cells and of the endothelium, and activation of humoral cascades such as the complement system and bradykinin. They also involved activation of factors that, in that era, had not been identified, such as tumor necrosis factor. Exposure to cardiopulmonary bypass was quantitatively related to consumption of coagulation factors, consumption of platelets, hemolysis, embolic load, and injuries related to the cannulation of small vessels. It was unfortunately common in the 1970s, when one did not have a choice but to use cardiopulmonary bypass in the repair of a congenital lesion in a small baby, that these infants became extremely edematous. They typically experienced a profound low cardiac output state after surgery associated with severe hepatic and renal dysfunction. This state, of course, was also associated with adverse neurologic consequences. It was for these reasons that innovators such as Barratt-Boyes and Castañeda pioneered the use of deep hypothermic circulatory arrest in the 1970s and 1980s, minimizing the exposure of infants to cardiopulmonary bypass, avoiding cumbersome requirements for cannulation, minimizing cardiac distention during bypass, and optimizing venous decompression. The technique also gave them a bloodless operative field. This was an era when the use of cardioplegia to render the heart motionless was something relatively new. But a heart that stood still, and was bloodless and flaccid, and did not have blood leaking into the field from the pulmonary veins and the systemic veins, was really revolutionary. The approach led to much more precise intracardiac repairs than had earlier been possible. It also made some operations, and the management of some diseases, possible that had appeared technically unfeasible prior that time. This included definitive therapy for the various forms of interruption of the aortic arch and, of course, the substrate for this Supplement, namely hypoplastic left heart syndrome. The wide application of hypothermic circulatory arrest unequivocally increased the overall safety and effectiveness of repairs undertaken in the neonatal period and early infancy. It was associated with a significant reduction in the generalized edema seen following prolonged periods of cardiopulmonary bypass, and reduced the generalized inflammatory response to cardiopulmonary bypass. Central to the success of techniques relying upon hypothermic circulatory arrest for intracardiac repair was the fact that the adverse effects of cardiopulmonary bypass itself were related to the duration of bypass. So, a large part of the emphasis in the early advocacy for hypothermic circulatory arrest was to reduce the time during which the entire organism was subjected to cardiopulmonary bypass.

As successful reparative cardiac surgery performed in neonates and infants became more common, it also became apparent that hypothermic circulatory arrest has its disadvantages. The most important of these are the potential for hypoxic ischemic injury to the central nervous system, as well as other organs, the potential for reperfusion injury following ischemia, and to a lesser degree the potential for air embolism with reinstitution of cardiopulmonary bypass. With increasing recognition of these issues, most importantly increasing recognition of the potential for neurologic injury in association with deep hypothermic circulatory arrest, two things have taken place during the last decade. One was investigation directed vigorously to explore the use of alternatives to deep hypothermic circulatory arrest, and the other was research to further our understanding of deep hypothermic circulatory arrest, and the potential morbidities associated with it, in order to maximize its utility and safety. Both of these paths of investigation have been of immense importance.

The principal alternatives to hypothermic circulatory arrest that have been explored are deep hypothermia with continuous low flow cardiopulmonary bypass, or moderate hypothermia with continuous high flow cardiopulmonary bypass. There have been important studies comparing the morbidities associated with the various techniques. There has also been an enormous body of research to increase our understanding of the consequences of hypothermic circulatory arrest. This has addressed the mechanisms of injury, the associated morbidities, and ways to increase the safety and optimize the outcomes in association with its use. The comparative studies with which people are most familiar, and which have had a significant impact on the conduct of cardiac surgery in the last decade, is the series of investigations that have together been known as the Boston Circulatory Arrest Trial.⁴ These studies, which began to enroll patients in 1988, and continued through 1992, involved 171 patients at Children's Hospital Boston undergoing arterial switch operations for either transposition with intact ventricular septum, or transposition with ventricular septal defect. The investigations had the features of a well-planned randomized trial, but suffered from some noteworthy flaws in design. For example, all patients experienced at least a brief period of hypothermic circulatory arrest, whether the principle method of support involved hypothermic circulatory arrest or continuous low flow cardiopulmonary bypass. The first report described perioperative findings. The principal feature of that report was the higher incidence of seizures in patients who had been randomized to a strategy relying primarily on hypothermic circulatory arrest rather than on low flow cardiopulmonary bypass. Interestingly, there was also a higher risk of seizures in patients with ventricular septal defect, and patients with older age at surgery, regardless of the primary mode of circulatory support. In addition, the fluid balance was markedly more positive in the patients with continuous low flow cardiopulmonary bypass than in those with circulatory arrest. At this early stage, it looked ominous with regard to the potential for neurologic outcome among patients having hypothermic circulatory arrest, but in some ways it also looked ominous with respect to the total body inflammatory response to cardiopulmonary bypass in those patients who were randomized to continuous low flow. This latter observation was corroborated by laboratory investigations, such as the studies by Skaryak and associates⁵ at Duke in 1-week-old piglets, showing that continuous cardiopulmonary bypass with low flows produced greater pulmonary dysfunction than did hypothermic circulatory arrest. There were also conflicting findings with regard to the pattern of neurologic injury. Scheller and associates,⁶ again in piglets, observed a higher percentage of cells with severe distortion of the Golgi morphology in their animals subjected to low flow cardiopulmonary bypass than they did in those undergoing hypothermic circulatory arrest.

The investigators at Boston followed up periodically with careful reappraisals of the initial patient population. At one year, it was apparent that one-third of all the patients tested had neurologic abnormalities. This was found across both groups. All neurologic abnormalities were assessed to be mild, and neurologic abnormalities tended to be more common in the deep hypothermic circulatory arrest group, at a frequency of two-fifths, than in the patients having continuous low flow cardiopulmonary bypass, where the frequency was three-tenths, but this difference was not statistically significant. They also made the important observations that patients who had experienced seizures around the time of surgery had intelligence quotients that were lower than those of patients without seizures. This observation was again consistent at four years of follow-up. But, very recently, the same group of investigators have described the findings at eight years of follow-up of the same patient cohort, with almost complete follow-up.⁷ At eight years, all neurodevelopmental indexes of the cohort as a whole, were below the expected population means. There was no difference between groups in terms of full-scale intelligence quotient. When I refer to groups, these are the groups that had been randomized primarily to hypothermic circulatory arrest, or to low flow continuous cardiopulmonary bypass. There was no difference between groups with respect to verbal intelligence, neither to performance. There was no difference between groups with respect to any of the 13 subsets of the Wechsler Individual Achievement Test. There was no difference for the overall reading score, or overall math score. There were subtle differences wherein the patients randomized to deep hypothermic circulatory arrest did slightly less well on tests that were indicative of fine motor control.

Not only do these studies involving eight years of follow-up reveal very little difference in indexes of neurodevelopmental outcome between the two groups, but the decade or more that has past since the enrollment of these patients in that study has witnessed an enormous amount of research to improve our understanding of the consequences of hypothermic circulatory arrest, and to increase the safety of this methodology. A lot of this work was done at Harvard in the laboratory of Richard Jonas, and a lot of it was done elsewhere. But there are several areas where there has been steady and important progress in terms of increasing the safety of circulatory arrest. One is with respect to the pH strategy.⁸ With regard to the comparison of alpha-stat versus pH-stat strategies, the latter is felt to provide more optimal flow of blood to the brain during the period of cooling prior to the institution of hypothermic circulatory arrest. This is significant, as it has become evident that the adequacy and uniformity of cooling of the brain is important in terms of minimizing hypoxic ischemic injury with or without periods of complete ischemia. Another area of research concerns the optimal level of hematocrit for bypass and circulatory arrest. Most of the studies done in the late 1980s and early 1990s involved either profound hemodilution down to hematocrits of less than 10%, or at the very least moderate hemodilution down to hematocrits in the range of 16 to 20%. More recent data from both laboratory and clinical studies at Harvard have shown that the safety of hypothermic circulatory arrest is vastly improved by maintaining a milder degree of hemodilution with a hematocrit closer to 30%.⁹ There has been a lot of speculation in the last 10 to 15 years that much of the injury to the central nervous system associated with the use of bypass with or without hypothermic circulatory arrest was related to oxygen free radicals, along with the cellular disruption associated with the presence of these free radicals. The Philadelphia Neuro Protection Trial, of which Robert Clancy is the principal investigator, and in which Gil Wernovsky and myself are co-investigators, looked at the use of pharmacologic protection against injury by free radicals. Specifically, among patients with the diagnosis of hypoplastic left heart syndrome, the use of pharmacologic protection was shown to be efficacious against such injury.¹⁰ As importantly, the hypothesis that the radicals caused a lot of neurologic damage led surgeons to advocate that cardiopulmonary bypass be performed at levels of normoxia rather than hyperoxia. Scientists suggested that the routine use of cardiopulmonary bypass with tensions of arterial oxygen in the range of 400 to 500 mmHg would be associated with injuries due to free radicals. So a whole era evolved where cardiopulmonary bypass was conducted with normoxia, and oxygen tensions between 80 and 120, and those people were quite sure that they had enhanced the safety of their patients. It turns out, however, that when patients are subjected to hypothermic circulatory arrest, the benefits of hyperoxia during cooling and preoxygenation, and during reperfusion and rewarming, outweigh the potential disadvantages associated with liberation of oxygen free radicals. When hypothermic circulatory arrest is utilized, hyperoxic cardiopulmonary bypass results in better neurologic outcomes than does normoxic bypass.¹¹

It is Richard Jonas and associates¹² again who taught us the lesson of the effect of post-circulatory arrest temperature. It was very conventional, and considered very important, a decade ago when the patient came off bypass that they be fully rewarmed to 37°C. And as soon as they reached the Intensive Care Unit, they were placed on a warming table to insure warm fingers and toes, with the belief that this promoted improved cardiac output and improved coagulation. But it turns out that a couple of degrees of mild hypothermia has an important protective effect after cellular ischemia in most organs, including the central nervous system, and it is quite clear that hyperthermic temperature elevation of as little as a degree above normal in the central nervous system markedly increases the potential for post-ischemic injury. Thus, intentional maintenance of mild hypothermia during the initial period of recovery is associated with a lower incidence of adverse neurologic outcomes following hypothermic circulatory arrest.

Finally, referring back again to the Boston Circulatory Arrest Trial, the investigators sought to determine whether late neurodevelopmental outcomes such as intelligence quotient, academic achievement, fine motor function, and speech production were effected by the duration of hypothermic circulatory arrest. They found that the linear effect of the duration of circulatory arrest on neurodevelopmental outcome was quite flat, not significantly different from zero, for durations of circulatory arrest up to 41 min.⁷ Beyond this cut point, assessments of neurodevelopmental outcome declined with an increasing duration of hypothermic circulatory arrest. The authors were careful to point out that this apparent cut point at 41 minutes was reflective of the outcomes using the methodology of hypothermic circulatory arrest that was contemporary over the period from 1988 to 1992, and has been shown to be in many ways inferior to the techniques that have evolved in the decade since then. It is quite clear, nonetheless, that short durations of hypothermic circulatory arrest are not associated with an increased risk of adverse neurologic outcomes as measured by these indexes of neurodevelopmental state.

Despite these encouraging new data with respect to the safety of hypothermic circulatory arrest, investigators have spent much of the last decade figuring out ways to perform intricate cardiac repairs in neonates and infants so as to avoid the use of hypothermic circulatory arrest. Several investigators have advocated selective cerebral perfusion as an alternative to hypothermic circulatory arrest. Its application is discussed in this Supplement in the manuscript by Tchervenkov and associates (see pp 70–74). There are some important unanswered questions with regards to selective cerebral perfusion. Its first principle advocate, Asou at Kitasato University in Japan,¹³ found a difference in pressures between the two sides of the brain when monitoring pressures in the right radial and left superficial temporal arteries during repair of the aortic arch as part of the Norwood operation. Going back to the fundamental scientific literature, they reflected on anatomic studies that showed that the circle of Willis, in nearly one thousand patients, had a perfect pattern only one-fifth of the time, and that hypoplasia of the communicating arteries was present nearly four-fifths of the time. This observation certainly raised a question as to the adequacy of perfusion of the entire brain when provided through a single vessel. Furthermore, Pigula and colleagues,¹⁴ at Pittsburgh, using levels of flow in the range of 20 to 30 ml/kg/min that were monitored and ascertained using near infrared spectroscopy, found considerable somatic perfusion during selective cerebral perfusion using low flows. So, if you are down at a level of flow that is about one-fifth, what you conventionally use to perfuse the entire body, and you imagine that you are optimally perfusing the brain, you find in fact that you are providing a lot of perfusion to organs beneath the diaphragm, including skeletal muscle. This certainly raises a question as to the adequacy of perfusion of the brain at that low level of flow. Hoffman and Tweddell,¹⁵ in Milwaukee, found decreased cerebral oxygenation following selective cerebral perfusion in neonates undergoing the Norwood operation. Using reflectance oximetry of the brain and kidneys during the first stage procedures, they observed maintenance of cerebral oxygenation during selective perfusion, but observed decreased oxygenation in the brain following rewarming on whole body cardiopulmonary bypass. This is the same observation that was made in the Ungerleider's laboratory a decade ago using piglets to study hypothermic circulatory arrest, and was interpreted as an important potential marker for central nervous system injury. Cerebral blood flow was decreased during reperfusion after deep hypothermic circulatory arrest, and here it is very similarly decreased after avoidance of circulatory arrest by means of a period of selective cerebral perfusion.

So, while many surgeons continue further to refine techniques of avoiding hypothermic circulatory arrest, there are those of us who continue to advocate the use of hypothermic circulatory arrest of limited duration, and with constantly improving techniques, in order to optimize and facilitate the most precise repair that can be accomplished. With respect to innovative alternative techniques, there remain many unanswered questions. The animal studies of selective cerebral perfusion are very few and are contradictory. A much larger body of clinical research will ultimately determine the relative safety and efficacy of innovative strategies of perfusion with respect to hypothermic cardiopulmonary bypass and hypothermic circulatory arrest.