Congenital heart disease (CHD) is the most common birth defect in children in the United States, occurring in 0.3-1.2% of live-born neonates.  Over the last 30 years, advances in surgical and interventional cardiology have greatly improved, and many centers now achieve 30-day surgical mortality rates of less than 1%. Indeed, survival into adulthood has become the expectation for most patients with cardiac lesions. Approximately 26,000 infants are born with CHD annually in the United States; approximately 23,000 of these infants reach adulthood. With dramatic improvements in survival, minimizing morbidity has taken a prominent role in the management of these children. A growing body of work examining and refining techniques aimed at neuroprotection throughout the perioperative period is at the forefront of these efforts.
Studies in the late 1980s and early 1990s indicated that the incidence of acute neurological complications in children undergoing cardiac surgery was as high as 25%,  but more recent surveys estimate the incidence to be lower.  Menache et al reported an incidence of 2.3% in a retrospective review.  As moderate and long-term studies examine the functional and neurodevelopmental status of these children, the frequency of both mental and psychomotor deficits appears to be much higher than previously believed. [5, 6] Although most children’s intelligence quotient (IQ) scores are within the reference range, as many as one third of school-aged children require some form of special education. 
Risk factors for abnormal neurological outcomes in children with CHD begin in the preoperative and prenatal periods.  These children are at risk for structural CNS malformations. Specifically, children with hypoplastic left heart syndrome (HLHS) have been shown to have as much as a 30% risk of brain dysgenesis.  Children with microdeletions of chromosome 22, commonly associated with various types of CHD such as tetralogy of Fallot and truncus arteriosus, have structural brain abnormalities.  Indeed, Gaynor et al have found that the presence of a genetic syndrome and polymorphisms of certain genes (presence of the apolipoprotein E e2 allele) may influence neurodevelopmental outcomes. 
In addition to structural CNS abnormalities, these children are subjected to a host of other preoperative risk factors. The hemodynamic consequences of many types of CHD result in impaired tissue oxygen delivery from both hypotension and hypoxemia. Poor cardiac output, particularly in children with left sided obstructive lesions, may result in acidosis. Those with left sided obstructive lesions, such as coarctation of the aorta, have an increased risk of cerebral hemorrhage. 
Many children who require neonatal cardiac surgery have ongoing feeding and nutritional issues. In those with ductally dependent, left-sided obstructive lesions, impaired splanchnic perfusion is noted preoperative. The managing clinician often limits the volume of enteral feeds, necessitating that most nutrition is delivered in parenteral form. Similarly, in children with large left-to-right shunts and volume overload lesions, the child’s respiratory and overall clinical status may limit the volume of enteral feeds given.
Intraoperative Risk Factors
Depending on the underlying cardiac anatomy, pediatric cardiac surgery frequently requires complex intracardiac and intravascular repairs. To accomplish this goal, cardiopulmonary bypass (CPB), often coupled with either deep hypothermic circulatory arrest (DHCA) or antegrade cerebral perfusion (ACP), is usually required.
In its simplest form, the goal of CPB is to circulate blood that supports the tissues of the body during cardiac surgery. Although delivery of oxygen and removal of carbon dioxide is fundamental to this goal, in order to maintain an optimal physiological condition for operating many other factors must be controlled.
From a mechanical standpoint, cannulation, including cannula size and proper placement, is vital. Since most pediatric CPB systems use gravity to siphon the venous blood back to the reservoir, larger cannulae are necessary. If the cannulae are too small, resistance increases, causing direct trauma to the blood, hemolysis, inflammatory cytokine release, and protein denaturation. If placement of the venous cannula is not correct, venous drainage may be suboptimal, and cerebral congestion can occur. CPB flow rates can be adjusted and must be closely matched to the rate of venous return to the reservoir to avoid the circuit “running dry” and entraining air. Most centers advocate high flow rates (approximately 150 mL/kg/min for smaller children) to enhance tissue perfusion and minimize local acidosis.
Control of the child’s temperature is predominantly accomplished through use of a heat exchanger. At the onset of CPB, the patient’s blood is cooled, and the child’s core temperature is reduced. For more complex operations that require a bloodless field, or in neonates in whom placement of venous cannulae and a left atrial vent is prohibited by the child’s size, more profound hypothermia is used, as is discussed below.
Most centers advocate the use of moderate hypothermia (25-33 º C), which has been shown to protect vital organs from the effects of ischemia, for cases that require routine CPB. The mechanisms by which hypothermia is protective are not completely understood. It reduces the metabolic demand of tissues in an exponential manner, so that at 20 º C (a level frequently used for deep hypothermia in cardiac surgery) the body’s metabolic rate is about 20% of that at 37 º C.
Hypothermia has been shown in animal models to cause a favorable shift in the intramyocardial anti-inflammatory cytokine balance, with both a decrease in the release of the pro-inflammatory cytokine tumor necrosis factor (TNF)-α and an increase in expression of the anti-inflammatory cytokine interleukin (IL)-10. 
From a CNS perspective, hypothermia has additional advantages. Although cooling causes cerebral blood flow to decline at a linear rate, it decreases metabolic rate in an exponential fashion, as mentioned above. Thus, cooling the brain increases the blood flow to metabolic needs ratio, and may help prevent ischemic injury. Cerebral autoregulation is generally well preserved at moderate hypothermia;  cerebral blood flow does not significantly vary over a wide range of blood pressures. In adults, some evidence suggests that moderate hypothermia is protective against postoperative neurocognitive deficits,  although no comparable studies have been performed in pediatrics settings.
When hypothermia is used, both the rate and extent of cooling and rewarming is important. For moderate hypothermia, cooling typically occurs over 10-15 minutes and is accomplished predominantly by precooling the perfusate to approximately 25 º C. At the conclusion of the repair, rewarming commences and generally proceeds over at least 20 minutes. During rewarming, the temperature of the heat exchanger must be no higher than 10 º C warmer than the child’s temperature because rapid rewarming can cause gaseous cavitation within the solution.
Deep hypothermic circulatory arrest
For children in whom the surgical repair requires a bloodless field or in whom full cannulation including a left atrial vent is prohibited by the patient’s size, circulatory arrest may be necessary. When this is required, deep hypothermia is generally used to further protect the vital organs.
Deep hypothermia is accomplished by further reducing the temperature of the heat exchanger, placement of a cooling blanket, and packing the head and heart in ice so that the core temperature reaches approximately 18 º C. In children, data suggest that deep hypothermia permits a longer time period with reduced or absent cerebral perfusion.  Because smaller children have a larger ratio of body surface area to volume, external cooling results in more efficient brain cooling; thus, neonates and infants may tolerate longer ischemic periods than older children or adults.
Cerebral autoregulation is generally preserved at levels of moderate hypothermia, but this is not the case when deep hypothermia is required.  Reactivity to carbon dioxide is reduced, and these effects may be more pronounced in neonates than older children. Evidence suggests that hypothermia increases cerebral vascular resistance [16, 17] and mitochondrial dysfunction. [18, 19]
Increasing duration of DHCA is associated with an increased risk of postoperative seizures,  which has similarly been shown to be associated with worse neurological outcomes.  The length of “safe” DHCA is unknown, but multiple studies have shown periods longer than 41-60 minutes are likely more detrimental from a CNS perspective. [21, 22, 23] However, even shorter periods of DHCA may be associated with some neurological risk; thus, antegrade cerebral perfusion (also known as regional low flow perfusion) is being increasingly used at many centers. [24, 25]
Antegrade cerebral perfusion
Traditionally, neonatal cardiac repairs, particularly those requiring extensive work on the aortic arch (eg, stage I Norwood palliations, aortic arch advancements), require a prolonged period of DHCA. An alternative strategy, antegrade cerebral perfusion (ACP), involves cannulation of either the innominate or subclavian artery using a Gore-Tex graft to allow selective perfusion of the carotid artery. [26, 27] Collateral flow, including via the circle of Willis, allows perfusion of the entire cerebral circulation, with minimal detectable flow to the remainder of the somatic circulation. 
ACP is often conducted at temperatures between standard CPB and DHCA, around 20-25 ºC. Although only limited studies have examined the neurological outcomes of children repaired using ACP, several studies have shown that it allows reduction and sometimes elimination of the need for DHCA  ; therefore, it may improve the neurological risk profile, approaching comparable outcomes to children undergoing CPB alone. 
The optimal treatment strategy for controlling blood pH and PCO2 is the subject of considerable controversy. The exact amount of carbon dioxide, and therefore the pH of the blood, can easily be manipulated by controlling the composition of the gas exposed to the membrane oxygenator as well as the gas flow rate. Blood is analyzed throughout the surgical case using either the “alpha stat” or “pH stat” methods. The fundamental difference between these techniques is that the pH stat method corrects for temperature, whereas alpha stat does not. Thus, use of an alpha stat method results in a more alkalotic and hypocapnic management of the child.
The alpha stat method more closely mimics the human body’s natural response to hypothermia; thus, the technique inherently seems more physiologic. However, an alkaline pH causes cerebral vasoconstriction and shifts the oxyhemoglobin dissociation leftwards, which may limit oxygen delivery to the vulnerable brain. Both animal and human studies are mixed regarding the merits of each method. Early work from Jonas et al demonstrated that the alpha stat strategy was associated with worse cognitive outcomes;  however, subsequent work from the same group showed no differences in 1-year and 2-year to 4-year developmental and neurological outcomes in children managed with the different strategies. 
Prevention of microemboli
Gaseous and particulate microemboli must be prevented because they may cause terminal vessel dilation and aneurysm, as well as cerebral microinfarcts.  Heparin-bonded circuits minimize the proinflammatory response and fibrinolytic activity caused by CPB.  Arterial line filters can protect against emboli down to around 37 mm in diameter and have been shown to reduce a large fraction of microemboli in both animals and human studies. [35, 36] They should routinely be used. Membrane oxygenators help to further filter gaseous microemboli.
Hemodilution and hematocrit management
The concept of hemodilution during CPB was first used in the 1950s and was thought to improve microcirculatory flow, particularly because the viscosity of blood increases with hypothermia, leading to increased systemic vascular resistance. However, hemodilution also decreases cerebral perfusion pressure and increases cerebral flow, which may increase the microembolic load.
Finally, hemodilution reduces the blood’s oxygen carrying capacity. When coupled with the effects of hypothermia and an alkalotic pH strategy as mentioned above, the decreased oxygen carrying capacity can limit oxygen delivery to vulnerable neurons. In the past, perfusion strategies targeting a hematocrit level around 20-25% were thought to be optimal, but recent studies have shown that children randomized into a lower hematocrit strategy (21%) compared with those in a higher hematocrit strategy (28%) performed worse at 1-year follow-up on the Psychomotor Developmental Index of the Bayley Infant Scales of Development.  Further studies from the same group have found that the increases in psychomotor development reach a plateau around a hematocrit level of 24%.  When children were randomized to a strategy of a hematocrit level of 25% versus a hematocrit level of 35%, no differences in developmental outcomes were noted at age 1 year. 
Glucose and electrolyte balance
Management of glucose and other electrolyte concentrations is accomplished using an ultrafiltration system while on CPB. In general, all electrolytes levels are kept within the reference range, with several important exceptions. An increasing body of work suggests that hyperglycemia is detrimental in various laboratory and clinical situations. In particular, hyperglycemia can be toxic to the CNS when subjected to ischemia. [40, 41] The priming solution used for CPB is therefore prepared without added glucose, particularly for cases with anticipated CNS ischemia (circulatory arrest).
Ultrafiltration of prime solution in conjunction with zero-balance ultrafiltration and modified ultrafiltration during CPB appears to have a statistically significant effect on improved pulmonary function in the early postoperative period relative to conventional plus modified ultrafiltration.  However, overall outcomes appear to be similar.
In addition to glucose, cellular calcium concentrations must be carefully controlled. Calcium homeostasis is involved in ischemia-reperfusion injuries, and massive increases of intracellular calcium have been found in tissues reperfused after lethal ischemia. Hypothermia appears to induce intracellular calcium accumulation; thus, achieving cardiac arrest prior to the institution of deep hypothermia may be beneficial. Maintaining low normal or reduced levels of calcium may be advantageous in the preischemic period and may reduce the degree of ischemic damage; however, these measures are purely hypothetical at the present time.
With the broad range of potential neurological insults discussed above facing the surgical and anesthetic team, careful intraoperative monitoring to ensure continual oxygen delivery to the brain is crucial. Despite this need, data examining the effectiveness of neurologic monitoring in pediatric cardiac surgery is relatively scarce. [43, 44]
An electroencephalogram (EEG) can provide a rough estimate of the depth of anesthesia. Unfortunately, standard EEGs are difficult to place and require a dedicated technician be present for interpretation, both of which make their routine use impractical. The Bispectral Index (BIS) monitor (Aspect Medical Systems; Newton, MA) has been approved by the US Food and Drug Administration (FDA) and provides real-time, unprocessed EEG data that is simple to apply and interpret. The monitor uses Fourier transforms to produce a single numerical output, the BIS, which ranges from 0 (isoelectric EEG) to 100, with mean awake levels of 90-100.  Similarly, the Patient State Index (PSI) (Physiometrix, Inc; North Billerica, MA) is also FDA approved for assessing the depth of anesthesia; however, the sensors are larger than those for the BIS, which may interfere with the ability to place additional monitors. Both the BIS and PSI have only around 70% accuracy at predicting both loss and return to conscious. 
Transcranial Doppler ultrasound
Transcranial Doppler ultrasound (TCD) provides real-time monitoring of the cerebral blood flow velocities. Various probes are available and when placed on the temporal window, the angle of insonation and the depth can be adjusted to sample both the middle and anterior cerebral arteries. Alternatively, in neonates, the probe can be placed over the lateral edge of the anterior fontanel. Although normal values have been evaluated in infants and children, those values were obtained in awake children without congenital heart disease (CHD) under ideal conditions. Thus, the clinician often must establish a baseline on the child at the start of the case and use the TCD monitoring more as a trend than an absolute value. TCD, sometimes coupled with near infrared spectroscopy, can be used to assess the effectiveness of cerebral perfusion in low flow cardiopulmonary bypass (CPB) with or without antegrade cerebral perfusion (ACP) to help guide bypass flow rates. 
As mentioned above, assessment of cerebral oxygen delivery is vital to cerebral protection during cardiac surgery. Near-infrared spectroscopy (NIRS) uses optical wavelengths of near-infrared light where iron-porphyrin complexes of oxyhemoglobin and deoxyhemoglobin have different absorption spectra. The NIRS probe is placed on the child’s forehead with a diode light emitter and several detectors. Light is transmitted in a banana-shaped curve through the child’s cerebrum. The absorption of both oxyhemoglobin and deoxyhemoglobin are measured, allowing the cerebral saturation to be calculated. Although anatomical models predict that the volume of blood within the light path is approximately 75% venous and 25% arterial, the actual ratio in pediatric patients widely varies and averages 85% venous and 15% arterial. 
Several NIRS monitors are commercially available; thus, various terms have been applied for the output of NIRS monitors. One such monitor, the Somanetics INVOS system (Somanetics, Inc; Troy, MI) has an output termed the regional cerebral saturation index or rSO2 i, which is a numeric value ranging from 15-95%.
Like TCD, NIRS monitors can be used to help assess the adequacy of bypass flow rate at providing sufficient cerebral perfusion when using ACP. [26, 49] Because NIRS monitors show a good deal of variation between patients’ baseline levels, NIRS outputs are more helpful as a trend monitor than as an absolute number. Studies have shown that a decline of 20% from the patient’s baseline may represent a clinically important change. 
Children with congenital heart disease (CHD) face a multitude of risk factors for neurological morbidities throughout the preoperative, intraoperative, and postoperative periods. As intraoperative care is refined, including assurance of adequate oxygen delivery to vulnerable neurons, minimizing deep hypothermic circulatory arrest (DHCA) through the use of antegrade cerebral perfusion (ACP), and careful control of temperature, hematocrit, glucose, calcium, pH, and carbon dioxide levels, both short-term and long-term outcomes are optimized.
Additionally, the expanding array or cerebral monitoring allows intraoperative care to be modified on a minute-by-minute basis. In the future, large multicenter studies to examine long-term neurocognitive and developmental outcomes in these children will be necessary to fully evaluate the efficacy of our field’s efforts.