Abstract
Mortality from surgery to correct congenital heart disease has improved, but high incidences of neurologic injuries and neurodevelopmental abnormalities persist. Neonates with critical heart disease have the highest risk of both neurologic injury and subsequent developmental abnormality. Congenital heart disease places the brain at risk from in utero development, through the perioperative period, and beyond. It has been a challenge of the collective congenital heart disease care team to identify modifiable elements contributing to this important morbidity. This chapter outlines relevant cerebral vascular physiology in the context of hemodynamic challenges posed by congenital heart disease before, during, and after surgical correction.
Key Words
neurodevelopmental outcome, white matter injury, cerebral blood flow
Care of the child with congenital heart disease (CHD) has evolved rapidly in the decades since the 1980s, some adaptations driven by evidence, some not. Changes in operative techniques, cardiopulmonary bypass (CPB) strategies, postoperative care, monitoring, interventional cardiac catheterization, and echocardiography are collectively and nonspecifically associated with a dramatic improvement in mortality for patients with CHD. With improving survival there has also been a trend to perform more complex surgeries on younger patients with more comorbidities. Despite improvement in mortality, many survivors of CHD suffer high rates of long-term neurodevelopmental abnormalities: limitations of executive functioning, impulse control, language, and cognition, as well as attention deficit. The focus of pediatric cardiology and pediatric cardiac surgery has made a pivot from improving survival to improving the quality of life of survivors, including prevention of acquired neurologic injury from CHD.
Neonates with critical CHD who require surgery in the first month of life are particularly vulnerable to neurologic injury. In the last two decades, magnetic resonance imaging (MRI) has been methodically applied to neonates who require cardiac surgery with CPB using both preoperative and postoperative scans. These studies include different surgical and CPB approaches and different MRI scanners and protocols; however, taken together, they collectively describe the nature of presurgical and perioperatively acquired brain lesions seen in neonates with critical CHD. Twenty percent to 40% of neonates have a brain injury before surgery. It is unknown what percentage of these injuries occur in utero or postpartum. Postoperative scans show new injury, acquired in the perioperative period in 35% to 75% of newborns who require heart surgery with CPB, and these are most commonly located in the white matter ( Fig. 17.1 ).
There are two potential susceptibilities seen in preterm infants with similar patterns of white matter injury (WMI) that are invoked to describe the prominent finding of WMI after neonatal cardiac surgery. The white matter is a watershed region in the immature brain, where penetrating arterioles have less overlap than is found in the mature brain. Further, oligodendrocyte precursors in the immature brain have increased susceptibility to hypoxia/ischemia. Consistent with that theory is the finding that infants with critical CHD have immature myelination patterns at term birth, and immature myelination is associated with an increased risk of neonatal WMI after cardiac surgery.
It is widely assumed that the impaired oxygen delivery contributes to WMI and is caused by cyanosis, inadequate cardiac output, CPB strategies, circulatory arrest, and/or low blood pressure with diastolic runoff to pulmonary shunts. However, there are few evidence-based practices that have been shown to modify the incidence of any brain injury in neonates with CHD. Nonmodifiable factors, such as genetic abnormalities, and socioeconomic factors often have greater significance than modified parameters of care in neurodevelopmental studies of CHD patients. This chapter will review the physiology of cerebral vasculature, vulnerabilities to central nervous system (CNS) pathologic conditions seen in CHD patients, perioperatively acquired neurologic injury and relevant perioperative practices, and the neurodevelopmental sequelae of CHD.
Physiology of the Cerebral Vasculature
The patient with CHD is at risk for impairment of cerebral oxygen delivery from multiple factors in the oxygen delivery equation, including both oxygen content and blood flow. Perioperative management of CHD, more than any other category of pediatric critical illness, requires repetitive assessment and modulation of oxygen delivery to the various organ beds. In these patients, interventions do not necessarily align the goals of maintaining oxygenation, cerebral blood flow, coronary blood flow, and splanchnic/renal blood flow. In other words, interventions to improve perfusion of the brain may come at the expense of visceral perfusion and vice versa. In this section the peculiarity of cerebral vascular homeostasis is described and compared to other systemic vascular behaviors.
Perfusion of the brain is unique when compared with other mammalian organs. The brain has minimal glycemic reserves in the form of glycogen storage, and neurons are not tolerant of anaerobic respiration. The brain has an obligate need for an uninterrupted supply of glucose and oxygen. The maintenance of a constant ratio of cerebral substrate delivery and cerebral metabolic demand is the consequence of an intricate overlay of several known vascular responses that have presumably evolved from the selective pressure of the brain’s unique vulnerability. Each of these servomechanisms act on similar targets: modulating cerebral vascular resistance or perfusion pressure to constrain cerebral blood flow. Each of these mechanisms has a different stimulus, regional specificity, and frequency bandwidth (time delay) of operation. With increasing precision the systemic vasoconstrictive response, pressure autoregulation, and neurovascular coupling constitute three principal mechanisms of cerebrovascular control in a dynamic pressure-flow system.
Systemic Vasoconstrictive Response
A decline in cardiac output triggers vasopressin output, the renin-angiotensin-aldosterone system, and the sympathetic nervous system. Activation of this neurohormonal axis promotes systemic vasoconstriction, but the cerebral vessels are relatively spared due to a paucity of receptors for these vasoconstrictors. The overall result is preservation of the cerebral perfusion at the cost of systemic perfusion. The systemic vasoconstrictive response allows the brain to maintain cerebral blood flow across changes in cardiac output and renders cerebral perfusion independent of cardiac output up to a threshold. This response is demonstrated in Fig. 17.2 , which compares cerebral blood flow with renal blood flow during progressive hemorrhagic shock in a newborn swine. In the early stage of low-output shock in this example the mean arterial pressure is preserved between 55 and 65 mm Hg by systemic vasoconstriction. Cerebral blood flow is unaffected by the increased systemic vascular resistance, but renal blood flow is reduced to 25% of its baseline value.
The effect of the systemic vasoconstrictive response can also be seen during CPB. Michler et al. have shown that independent of bypass flow rates, if perfusion pressure is adequate, cerebral blood flow is preserved ( Fig. 17.3 ).
Perioperative and CPB strategy for infants in many centers includes alpha-adrenergic inhibition or other vasodilators to maintain systemic perfusion. This strategy is a blockade of the systemic vasoconstrictive response. Splanchnic and renal perfusion are improved, and this strategy is credited with improved survival, especially in the single-ventricle, shunted patient. Because this strategy results in low perfusion pressures when output is compromised, it is a potential vulnerability to cerebral perfusion pressure. To some degree the vasodilator strategy places renal protection and survival at odds with cerebral protection ( Fig. 17.4 ).
The contribution of this strategy to the burden of neurologic injury is unknown and is likely modulated by the second-tier vascular response of the brain: pressure autoregulation.
Pressure Autoregulation
When the systemic vasoconstrictive response is exhausted, or pharmacologically ablated, arterial blood pressure falls with reductions in cardiac output. Changes in arterial blood pressure, either up or down, engage the servomechanism of pressure autoregulation in the cerebral vasculature. Pressure autoregulation constrains cerebral blood flow fluctuations caused by perfusion pressure changes. Changes in the transmural pressure gradient of muscular arterioles trigger myogenic responses of constriction and relaxation in a dynamic fashion ( Fig. 17.5 ).
These vascular changes have been observed to occur rapidly in animal models and between 2 and 10 seconds in pediatric and adult clinical studies, dependent on the magnitude of the transmural pressure change. The pressure autoregulation response is fully engaged when changes in arterial blood pressure last longer than 30 to 60 seconds. Thus it allows for passive transmission of perfusion pressure changes at the pulse and respiratory frequencies but acts as a high-pass filter preventing the transmission of sustained perfusion pressure changes into cerebral blood flow changes.
Pressure autoregulation allows for tolerance of mild to moderate hypotension when vasodilators are used, without causing a vulnerability for cerebral ischemia. The limit of this tolerance is the blood pressure at which pressure autoregulation is exhausted, termed the lower limit of autoregulation, easily observed in Fig. 17.2 at 40 mm Hg and in Fig. 17.4 at 50 mm Hg. The lower limit of autoregulation was originally estimated in adults to be 50 mm Hg by Lassen and is now understood to have variability between patients and conditions. The lower limits of autoregulation are poorly defined, especially for pediatric populations, so absolute safe arterial blood pressures cannot be recommended. Two pediatric studies of CPB suggest a lower limit on average near 40 mm Hg, but the use of a global threshold ignores wide intersubject variability. A single threshold for all patients places some at risk of inadequate pressure and some at risk of inadequate vasodilation.
Neurovascular Coupling
Metabolic autoregulation is distinct from pressure autoregulation. Metabolic autoregulation, also known as neurovascular coupling, links neuronal activity to cerebral blood flow, mediated at the level of the neurovascular unit: individual astrocytes linking neuronal synapses and penetrating arterioles. These glial cells release both vasoconstrictors and vasodilators on associated muscular arterioles. In contrast with pressure autoregulation, neurovascular coupling is both rapid and regionally specific. Vascular dilation occurs in specific regions of neuronal activity, causing a match between the rate of oxygen consumption and oxygen delivery. Factors that reduce neuronal activity and cerebral metabolism cause reductions in cerebral blood flow, cerebral blood volume, and intracranial pressure. Metabolic autoregulation is an important aspect of therapies such as hypothermia and anesthesia for intracranial pressure reduction and neuroprotection of the injured brain.
Carbon Dioxide, Oxygen, and Glucose Reactivity
In addition to the pressure and flow homeostasis described above, the vasculature of the brain is highly responsive to perturbations of the basic chemistry of the cerebrospinal fluid, with mechanisms that can override pressure and metabolic autoregulation. Hypoxia, hypercarbia, and hypoglycemia all cause cerebral vasodilation.
Acute hypercarbia increases cerebral blood flow, and acute hypocarbia decreases cerebral blood flow, with an effect mediated by changes in cerebrospinal fluid pH. Intentional hypoventilation uses arterial carbon dioxide as a drug to increase cerebral blood flow. Single-ventricle patients with large left-to-right shunts have higher pulmonary vascular resistance and higher cerebral blood flow when they are hypercarbic. The combined effect of hypercarbia is to mitigate pulmonary overcirculation and enhance cerebral oxygen delivery.
However, carbon dioxide as a therapy is likely time limited in its effectiveness. Carbon dioxide diffuses freely across the blood-brain barrier, which is not permeable to hydrogen ions. The spinal fluid compartment has an independent pH buffering system from the blood within the choroid plexus. The choroid plexus can produce appropriately buffered spinal fluid much faster than the kidney buffers the pH of blood, in part because of the high rate of production of cerebrospinal fluid. Spinal fluid pH changes (and thus cerebral blood flow changes) induced by hypocarbia or hypercarbia are limited to 3 to 6 hours, and the serum pH is not an accurate reflection of spinal fluid pH over time.
The cerebral vasculature is also responsive to changes in oxygen tension when oxygen delivery is below the threshold to meet metabolic oxygen consumption demands. Profound vasodilation can occur in the face of arterial oxygen tension of less than 60 mm Hg or anemia. These data come from animal and biventricular subjects, whereas the cerebrovascular effect of chronic exposure to cyanosis in the patient with CHD has not been elucidated.
Causes of Impaired Neurodevelopment in Congenital Heart Disease Outside the Perioperative Period
Identifying modifiable causes of brain injury and neurodevelopmental abnormality for children with CHD is nontrivial. Injuries that occur in an early perioperative period cannot be assessed for impact until years later, when development has occurred. Countless causes of neurodevelopmental abnormality can occur from conception to adulthood, making it difficult to isolate the effect of specific perioperative injuries or care practices. Very few perioperative practices have been shown to associate with long-term developmental outcome, so there has been a proliferation and divergence of practices between centers. This section describes known and potential contributors to impaired neurodevelopment in CHD outside the perioperative period.
Congenital Neurologic Abnormalities
Up to 30% of patients with cardiac defects have genetic abnormalities that affect other organ systems. Aneuploidies such as trisomy 21, 18, and 13; multiple syndromes such as Noonan, Williams, DiGeorge, and CHARGE; and the VACTERL association are all correlated with both CHD and developmental delay. Cohorts of CHD subjects with these recognizable patterns of malformation have developmental testing results that are below population norms and also below the norms of subjects across a wide range of cardiac lesions other than CHD.
Genetic-environmental interactions during illnesses and stressors such as inflammation, hypoxia, and coagulopathy are also postulated to contribute to the neurologic sequelae of CHD. Gaynor et al. studied apolipoprotein E (APOE) alleles for association with neurodevelopment in children with CHD. The APOE ε2 allele was associated with worse scores on multiple parameters of the child behavior checklist indices at the fourth and fifth birthdays, whereas the APOE ε4 allele was associated with better scores. The APOE lipoprotein is thought to have function in lipid transport and neuronal repair in the CNS.
Microcephaly and CNS immaturity are prevalent in patients with complex CHD. The increased incidence of microcephaly at birth persists into childhood and is independently associated with developmental delays. Microcephaly is thought to be caused by in utero hemodynamic perturbations associated with specific heart defects. Shillingford et al. found that in patients with hypoplastic left heart syndrome (HLHS) the degree of microcephaly is correlated with ascending aorta diameter. This finding suggests that interruption of the normal streaming of oxygenated blood from the placenta to the brain via the left ventricular outflow tract leads to reduced brain growth in utero.
CHD is associated with delayed brain maturation, detectable by MRI as both delayed white matter myelination and cortical folding abnormalities. Brain MRIs of fetuses with CHD show significant differences in brain volume that diverge from the normal trajectory in the third trimester. Licht et al. used a validated scoring system of brain maturity to assess four different parameters: myelination, cortical infolding, presence of germinal matrix tissue, and involution of glial cell migration bands. In a cohort of 29 infants with HLHS and 13 patients with transposition of great arteries (TGA), the preoperative brain MRI showed that term infants with CHD have a total maturation score comparable to a 35-week premature infant without CHD. Miller et al. demonstrated similar immaturity patterns using MRI measures of white matter integrity.
Neurologic Injury in Utero
In the normal fetal circulation the eustachian valve and the foramen ovale direct the most highly oxygenated blood to the developing fetal brain via the left ventricular outflow tract. In many complex structural heart defects, especially those with interrupted left-sided outflow, this beneficial streaming is not present, and oxygenated blood flow is directed toward the ductus arteriosus. Studies of patients with TGA, for instance, show that fetal blood with the lowest saturation is directed to the brain and blood with the highest saturation is directed inferiorly to the abdominal organs. Relative cerebral hypoxia in utero may contribute to the neurodevelopmental delays seen among these patients.
Fetal ultrasonography and Doppler ultrasonography of the circle of Willis have demonstrated abnormal responses of the cerebral vasculature in fetuses with systemic outflow obstruction, showing lower than normal cerebral vascular resistance. Donofrio et al. reported this low resistance by measuring the cerebroplacental Doppler ratio (CPR) in a study of fetuses with CHD. CPR is the ratio of the pulsatility index measured in the middle cerebral artery to the pulsatility index measured in the umbilical artery. The pulsatility index is the difference between systolic flow velocity and diastolic flow velocity, divided by the time-averaged flow velocity ( 
), and is related to downstream vascular resistance (higher pulsatility index is associated with higher downstream vascular resistance). Normal CPR is less than 1, meaning that cerebrovascular resistance is greater than placental vascular resistance. Only 48% of patients with HLHS have a measured CPR greater than 1, whereas 95% of normal fetuses have a measured CPR greater than 1—suggesting lower cerebral vascular resistance in fetuses with HLHS.
By contrast, patients with right-sided obstructive lesions can have high cerebral vascular resistance. Kaltman and colleagues reported low middle cerebral artery pulsatility index in HLHS fetuses, and high middle cerebral artery pulsatility index in fetuses with right-sided obstructive lesions compared with fetuses that have a normal heart ( Fig. 17.6 ).
These observed differences in cerebral vascular resistance are consistent with differences in arterial flow patterns at the aortic arch in utero. Normally, cephalad aortic flow in utero is from the left ventricular outflow tract (streamed oxygenated blood from the placenta across the foramen ovale), and caudad flow in the descending aorta is from the right ventricular outflow tract across the ductus arteriosus. With a left-sided obstruction there is loss of streaming of oxygenated blood to the brain. From the physiologic principles outlined, hypoxia may result in a decrease in cerebral vascular resistance. The effect of this change in cerebral vascular resistance on the developing brain is unknown and is a focus of current study.
Innate Vulnerabilities to Acquired Neurologic Injury in Patients With Congenital Heart Disease
Hypercyanotic Spells.
“Tet” spells occur in 10% to 20% of patients with CHD caused by either fixed or dynamic pulmonary outflow tract obstruction. These episodes of cyanosis occur most frequently between the ages of 6 months and 3 years and are associated with sudden drops in arterial oxyhemoglobin saturation. They are caused by marked increases in right ventricular outflow resistance and right-to-left shunting. Unrelenting episodes lead to cerebral hypoxia, loss of consciousness, convulsions, and seizures. Cerebrovascular accidents, hypoxic-ischemic injury, and permeant neurologic sequelae have been reported.
Cerebrovascular Accidents.
Arterial ischemic stroke, as manifested by arterial occlusion seen with vasospasm or clot, is less common than WMI in the perioperative period (see Fig. 17.1 ). However, patients with CHD that involves cyanosis, right-to-left shunting, failing ventricles, or the placement of hardware in the form of valves, conduits, or ventricular assist devices all have an increased risk of cerebrovascular accidents. CHD-related cerebrovascular accidents were more common and more well described in the era preceding surgical repair of CHD. The natural history of unrepaired cyanotic heart disease includes a 75% incidence of cerebrovascular accidents with the majority of these occurring before 2 years of age. Berthrong and Sabistion reported that venous occlusion and thrombus occurs with a higher incidence than arterial embolus. Anemia places infants at higher risk of ischemic events that are arterial in origin, and high hemoglobin concentrations associated with cyanosis increase the risk for venous thrombosis. Most arterial thromboembolic events have a single vessel involvement and are characterized by a focal deficit of the affected vascular territory. If multiple arterial territories or both cerebral hemispheres are involved, then an intracardiac thrombus or vegetation should be suspected and investigated. In the case of venous thrombosis, onset of symptoms is more indolent and often preceded by dehydration, which increases blood viscosity and the polycythemic state. Up to 20% of CHD patients with a frank cerebrovascular accident will suffer severe intellectual disability.
In the more modern era, mechanical cardiac support and the ventricular assist device have increased survival to transplant but have become an important risk factor for frank cerebrovascular accidents. In a trial of 48 pediatric patients less than 16 years of age with chronic heart failure, Fraser et al. compared the Excor Pediatric ventricular assist device against extracorporeal membrane oxygenation (ECMO) for survival to heart transplantation. Although survival to transplant was significantly higher in the ventricular assist device group, the rate of adverse events overall was sizable. The incidence of major bleeding—including intracranial bleeding—was reported at 50%, and thromboembolic events were reported at 29%.
Bacterial Endocarditis and Brain Abscess.
Patients with palliated single-ventricle anatomy and patients with unrepaired ventricular septal defects (VSDs) have the highest documented risk of developing bacterial endocarditis. Endocarditis associated with rheumatic valve disease has become less common with widespread antibiotic treatment for group A streptococci infections. Endocarditis is more likely to result in brain abscess when intracardiac shunt is present. Endocarditis with brain abscess before 2 years of age is still relatively uncommon in children with CHD in the developed world. However, brain abscess associated with CHD is far more prevalent in developing countries, where children with cyanotic cardiac defects often go unrepaired into school age. In this population, morbidity and mortality from endocarditis are correlated with the degree of hypoxia. Most brain abscesses are identified on neuroimaging with a characteristic ring enhancement associated with surrounding cerebral edema. Although they can be solitary, multiple lesions are found in up to 30% of cases, with the frontal lobe being the common site of infection. A study in Taiwan found a relatively low in-hospital mortality rate for brain abscess of 4%. However, another study in Turkey found that more than 50% of survivors had significant neurologic sequelae, including neuromotor deficits, seizures, and hydrocephalus.
Perioperative Neurologic Injury
Perioperative neuroprotection is a field of study with more hypotheses than results, more opinions than facts, and more controversy than standard of practice. Centers of excellence have seen dramatic improvements in outcome with divergent practices that evolve without guiding evidence. Practices that have not been assessed by evidentiary standards are locked in place for fear of disrupting the unidentified elements that caused overall outcome improvement. This section focuses on perioperatively acquired neurologic injury and potentially modifiable aspects of neurodevelopmental outcome for children with CHD. For some of these aspects the quality of data is compelling (e.g., hemodilution and transfusion during CPB). For some of these aspects the available data are either contradictory or leave room for debate (e.g., the use of deep hypothermic circulatory arrest [DHCA] or selective cerebral perfusion). It is reemphasized here that the primary lesion acquired in the perioperative period for newborns with CHD is WMI, which suggests a deficit of oxygen delivery. The link between WMI and neurodevelopmental delay is inferred, but the studies to assess this association have mixed results.
Timing of Surgical Intervention
Longer delays to surgical palliation of critical heart disease in the newborn may increase the likelihood of neurologic injury—specifically WMI. Lynch et al. studied 37 neonates with HLHS. Eight patients suffered preoperative WMI, and 28 patients, or 76%, were found to have postoperative WMI. In that study larger volume of new or worsened WMI was associated with a delay greater than 4 days from birth to Norwood palliation.
Petit el al. reported similar findings in patients with TGA. In that study 26 patients with TGA underwent preoperative and postoperative MRI scans. Ten patients were found to have preoperative WMI, and longer time to surgery was associated with presence of WMI on preoperative MRI. The WMI group underwent surgery at 5.6 ± 2.9 days compared to the no-WMI group, who were operated on at 3.9 ± 2.2 days ( P = .028).
A retrospective analysis of CHD patients undergoing both the arterial switch operation and the stage I repair for HLHS showed that timing of surgery was associated with outcome. For patients with TGA, surgical correction after day of life 3 was associated with increased rates of major morbidity and associated hospital costs. For patients with HLHS, morbidity and hospital costs increases were associated with each day of life that surgery was delayed. The mortality rate in that cohort of 134 neonates with HLHS was 7.4%, and all deaths occurred in subjects operated on after day of life 4, with a median age at operation of 5 days.
Deep Hypothermia and Blood Gas Management
Some degree of hypothermia is applied during CPB at most centers. For surgeries that require periods of time with permissive low-flow or absent-flow states, deep hypothermia is the mainstay of neuroprotection. Hypothermia reduces the metabolic rate of oxygen consumption in the brain. Neurovascular coupling dictates that reduced oxygen consumption concomitantly reduces cerebral blood flow. However, it has been shown that profound hypothermia causes a relative uncoupling of cerebral metabolism and flow, resulting in “luxury perfusion” ( Fig. 17.7 ). In infants and children with CHD, Greeley et al. showed that although cerebral metabolism is reduced exponentially by hypothermia, cerebral blood flow is reduced linearly. From these studies it was estimated that an infant in a state of deep hypothermia could tolerate between 39 and 65 minutes of circulatory arrest before developing an oxygen deficit that would lead to ischemic injury.
During profound hypothermia, either pH-stat or alpha-stat blood gas measurements are used to account for the increased solubility and decreased partial pressure of carbon dioxide in blood. Both are used at centers with good outcomes. The alpha-stat approach seeks to adjust the P a CO 2 to maintain the degree of ionization (alpha) of histidine imidazole groups. The pH-stat method seeks to adjust the P a CO 2 to maintain pH at the temperature of the patient. Changes in P a CO 2 solubility with temperature do not change ionization of histidine but do change pH due to decreased partial pressure of P a CO 2 . Carbon dioxide diffuses freely across the blood-brain barrier to the cerebrospinal fluid, which has a bicarbonate buffer that is independent of serum buffers (and measurements). In contrast to the alpha-stat strategy, the pH-stat strategy corrects for blood temperature, results in higher P a CO 2 , lower pH, and increased cerebral blood flow .
It has been postulated that the “luxury flow” seen in deep hypothermia may be a consequence of a shift in the oxygen-hemoglobin dissociation curve that impairs oxygen unloading to brain tissue (hypoxic vasodilation). Hypercarbia associated with the pH-stat strategy causes a shift in the oxygen-hemoglobin dissociation curve that facilitates oxygen unloading, in theory counteracting the effect of temperature on the oxygen-hemoglobin dissociation curve. Increased cerebral blood flow with pH-stat management facilitates even cooling in the brain, and in animal models there is improved early metabolic recovery from deep hypothermia when compared with alpha-stat management. A comparison of alpha-stat and pH-stat management in piglets showed improvements in both functional disability scores and histopathology when pH-stat was used.
Studies comparing alpha-stat and pH-stat strategies in adults have shown a trend for improved neurocognitive outcome with alpha-stat management. However, the results are not generalizable to the pediatric population in part due to gross differences in the etiology of neurologic injury: embolic lesions are most common in adults during bypass, and ischemic WMI is most common in pediatric patients. A prospective, randomized trial of alpha-stat versus pH-stat management in 182 infants and neonates under deep hypothermia (with and without circulatory arrest) showed no measurable difference in neurodevelopmental outcome. However, subjects in the pH-stat group had more hemodynamic stability, earlier recovery of electroencephalographic activity, and a trend for fewer seizures, which were rare in both groups. The early benefits seen in both clinical and animal models with the pH-stat strategy have not been shown to provide long-term benefit for neonates who require cardiac surgery.
Deep Hypothermic Circulatory Arrest or Selective Cerebral Perfusion
When reconstruction of the aortic arch is required for neonatal cardiac surgery, as in the Norwood operation, interrupted arch repair, or arch advancement, optimal visualization cannot be achieved with normal bypass flows and a standard cannulation technique. Traditionally DHCA has been used to maintain a bloodless field and facilitate these surgeries. As described earlier, the reduced metabolism associated with deep hypothermia provides a window of time that is estimated to be between 39 and 65 minutes to perform the operation and reestablish bypass flow before ischemic injury occurs. However, even when DHCA is restricted to 40 minutes, a characteristic pattern of cerebrovascular and metabolic disturbance has been observed in neonates, including (1) lowered cerebral metabolism that fails to return to baseline values after restoration of full flow, (2) decreased cerebral blood flow (with a normal CBF:CMRO 2 ratio), and (3) a transcranial Doppler flow pattern showing a high pulsatility index (elevated cerebrovascular resistance) and absent flow velocity in diastole ( Fig. 17.8 ). These disturbances may be mitigated by avoiding DHCA and using low-flow bypass, slower rewarming, pH-stat management during cooling, modified ultrafiltration, thromboxane A 2 antagonists, and nitric oxide donors.
Does DHCA contribute to brain injury and neurodevelopmental abnormality in the neonate with CHD? The Hearts and Minds study conducted in Australia and New Zealand demonstrated that longer periods of DHCA were associated with increased severity of WMI ( Fig. 17.9 ). However, that study of 122 neonates with CHD found no association between WMI and neurodevelopmental testing at 2 years. In a separate study at the Children’s Hospital of Philadelphia, neurodevelopmental testing was done at 4 years’ follow-up of 238 subjects after neonatal CHD repair. In that cohort there was no association between duration of DHCA and neurodevelopmental performance. The Boston Circulatory Arrest Study randomized newborns with TGA to DHCA or low-flow bypass and showed, in 170 enrolled patients, that DHCA was associated with postoperative seizures and worse 12-month and 8-year neurodevelopmental testing. However, in the 139 subjects that were tested at age 16, the use of DHCA was not associated with abnormalities on neuropsychologic testing. This study highlights the problem of measuring cause and effect between neonatal surgery and long-term outcome. At the time of mature neuropsychiatric testing the results are difficult to apply to current methodologies. Practice at the time of the Boston trial included hemodilution and profound anemia, which is now known to contribute to brain injury.
