Most clinical studies of the relationship of temperature to safety of a given circulatory arrest time are flawed by lack of information about the temperature of the brain itself and by the variety of sites of measurement of temperature (tympanic membrane, nasopharynx, rectum, midesophagus, bladder, extremity skin) used to estimate safety. There is little consistent correlation between some of these sites, although temperature of the tympanic membrane and nasopharynx most closely resembles mean temperature of the brain. ^, For this reason, these sites should be used whenever possible.

Animal experiments and clinical experiences indicate that when the brain is cooled to 15°C to 20°C, circulatory arrest of 30 minutes or less is tolerated without development of evident structural or functional damage. During such a period, adenosine triphosphate (ATP) concentrations decline to 35% of initial values but return rapidly to normal during reperfusion. Evidence that circulatory arrest of 45 minutes at these temperatures is safe is less secure. ^{, ,}

Considerable information supports the inference that circulatory arrest of 60 minutes or more at temperatures of 15°C to 18°C is associated with irreversible structural or functional damage, although it may be tolerated without evident damage by some subjects under some circumstances. In experimental studies by Folkerth and colleagues and Fisk and colleagues, histologic evidence of anoxic brain damage was found in all animals subjected to hypothermic circulatory arrest for 45 to 60 minutes, although some animals survived without evident functional abnormality. ^{, ,} Half (2 of 4) of the animals studied by Kramer and colleagues showed no recovery of ATP when subjected to 60 minutes of hypothermic circulatory arrest.

Some clinical studies have found few problems associated with 60 minutes or more of hypothermic circulatory arrest. Particularly striking is the experience of Coselli and colleagues, who found no clinical evidence of brain damage attributed to hypothermic circulatory arrest (mean nasopharyngeal temperature 16.9°C; range, 10.1°C–24.1°C) in 56 patients with arrest times ranging from 14 to 109 minutes (median 36 minutes). Comprehensive neuropsychometric studies were not undertaken. Hemiparesis or hemiplegia attributed to cerebral edema developed in 3 of 51 surviving patients (6%; CL 3%–11%). In contrast, Gega and colleagues, in a subsequent study of 394 patients undergoing aortic arch replacement, reported 8 strokes (13%; CL 8.6%–18%) among 61 patients in whom the duration of hypothermic circulatory arrest exceeded 40 minutes. Only 10 strokes (3.3%; CL 2.3%–4.3%) occurred among the remaining 333 patients with shorter intervals of circulatory arrest.

Temporary neurologic dysfunction (TND) manifested as postoperative confusion, agitation, delirium, obtundation, or transient parkinsonism without localizing signs can occur in up to 20% of survivors of operations on the thoracic aorta in which hypothermic circulatory arrest is used. Incremental risk factors associated with developing this complication are duration of hypothermic circulatory arrest and increasing patient age. Prevalence of TND increases substantially among patients in whom duration of circulatory arrest exceeds 60 minutes ( Fig. 2.2 ). Although postoperative delirium is not permanent, it can be an important complication. Among a group of patients undergoing pulmonary thromboendarterectomy, circulatory arrest times of greater than 50 minutes were a powerful risk factor for its occurrence.

Prevalence of temporary (TE) neurologic dysfunction as a function of duration of circulatory arrest time.

(From Ergin MA, Galla JD, Lansman L, Quintana C, Bodian C, Griepp RB. Hypothermic circulatory arrest in operations on the thoracic aorta. Determinants of operative mortality and neurologic outcome. J Thorac Cardiovasc Surg. 1994;107:788.)

Some evidence supports the concept that continuous perfusion of the brain for 60 or more minutes at low temperature also produces neurologic sequelae in a few patients (see “ Evidence of Gross Neurologic Damage ”). However, cold (10°C–15°C) continuous perfusion of brains already at 15°C resulted in no intellectual or other deficit in trained rhesus monkeys.

Uneven cooling of the brain is probably a risk factor for brain damage, although evidence is largely indirect. Almond and colleagues conducted experiments in dogs undergoing hypothermic circulatory arrest for 30 minutes. Results were interpreted to indicate structural and functional brain damage when cooling by CPB was done with the perfusate 20°C colder than the patient. These investigators believed this did not occur when the blood was only 4°C to 6°C cooler than the subject. However, the damage might have been related to the short period of cooling required with the very cold blood, producing uneven brain cooling, as suggested by the work of Zingg and Kantor. The longer period of cooling required with the blood only 4°C to 6°C colder than the subject probably produced more uniform cooling. In their patients, Stewart and colleagues noted a considerably higher prevalence of major neurologic events after circulatory arrest when core cooling by CPB alone was used, compared with surface cooling first to 28°C, followed by core cooling ( Table 2.2 ). A reasonable presumption is that the more rapid core cooling resulted in uneven cooling of the brain. In another study in neonates and infants, rapid core cooling was associated with more evidence of neurologic deficits after hypothermic circulatory arrest than more prolonged core cooling. Again, a reasonable presumption is that prolonged core cooling results in more uniform cooling of the brain.

The relationship of cerebral blood flow during cooling (before establishing hypothermic circulatory arrest) to safety of the arrest period has received little investigation. The earlier literature suggested that reduced arterial blood pressure during CPB without circulatory arrest contributes to postoperative neurologic dysfunction, presumably because the hypotension resulted in reduced cerebral blood flow, but this possibility now appears less certain (see “ Cerebral Blood Flow ” under Distribution of Blood Flow in Section II).

More recently, when cerebral blood flow was measured during operations involving hypothermic circulatory arrest in children, Greeley and colleagues observed that patients with increased oxygen extraction before circulatory arrest may be particularly vulnerable to cerebral injury. In a subsequent study, they demonstrated that a parallel reduction in cerebral oxygen consumption and cerebral blood flow occurred during cooling. However, in three of four patients who were subsequently found to have sustained neurologic injury, oxygen extraction before the period of circulatory arrest was increased, suggesting that cerebral blood flow during this period was inadequate to sustain metabolic requirements. Other studies in children have confirmed the observation that cerebral blood flow generally decreases with temperature during cooling and that coupling with cerebral metabolism is maintained even at low temperatures when ventilation is managed according to the alpha-stat strategy.

Information is available about the magnitude and effect of cerebral blood flow during rewarming. Experimental studies have found that cerebral blood flow is reduced during rewarming after circulatory arrest ( Fig. 2.3 ). ^, With or without circulatory arrest, this phenomenon occurs in humans during cardiac surgery and may affect outcomes. In infants, cerebral blood flow is reduced during rewarming immediately after hypothermic circulatory arrest and after achieving normothermia. Based on jugular venous oxygen saturation measurements, oxygen delivery appears to be adequate during this period of reduced flow. In a study of 255 adult patients undergoing elective coronary artery bypass grafting (CABG) with or without associated cardiac valve replacement, Croughwell and colleagues observed a decline in postoperative cognitive function in 38%. The severity of decline was related to greater arteriovenous oxygen content difference between radial artery and jugular venous blood (Ca vo ₂ ) during rewarming. This increase in oxygen extraction was associated with low jugular venous oxygen saturation and low cerebral blood flow.

Cerebral blood flow (mL · 100 g ⁻¹ · min ⁻¹ ) in gerbils after induction of and recovery from hypothermia by surface cooling. Along the horizontal axis is rectal temperature. Break between 37°C and 18°C represents 48 minutes of circulatory arrest (total bilateral carotid artery occlusion) at 18°C (arrest group) or continuing hypothermic perfusion (no arrest group). Note that cerebral blood flow was lower during rewarming in those animals that had total cessation of cerebral blood flow for 48 minutes.

(From Kirklin JW. The movement of cardiac surgery toward the very young. In: Crupi G, Parenzan L, Anderson RH, eds. Perspectives in Pediatric Cardiology. Vol. 2. Pediatric Cardiac Surgery, part 1 . Mount Kisco, NY: Futura; 1989:3.)

Only incomplete information is available in the area of biochemical milieu. It is uncertain whether some variables are actual risk factors or surrogates for the real risk factor. Arterial blood pH and P co ₂ during cooling may have important direct effects on brain tissue at the beginning of the period of circulatory arrest and thereby on outcome, but they also influence cerebral blood flow, and perhaps its distribution, during cooling (see “ Controlled Variables in Section II). Any effect they may have on neurologic outcome, which is uncertain, could be through either mechanism.

Brunberg and colleagues and Anderson and colleagues suggested that increased tissue glucose, such as is usually present at the beginning of the arrest period, may lead to excessive glycolysis and acidosis during the arrest period, possibly resulting in tissue damage from lactic acid accumulation. ^, This possibility makes it imprudent to use glucose solutions for priming the pump-oxygenator and for intravenous infusion when a period of circulatory arrest is contemplated.

Based on the work of Choi and of Olney and colleagues, evidence has accumulated indicating that the neuroexcitatory amino acids—particularly glutamate, the major transmitter mediating synaptic excitation in the mammalian central nervous system—have potent neurotoxic activity during conditions of depleted cellular energy (e.g., hypoxia, ischemia) when the synaptic reuptake of these amino acids, a highly energy-dependent process, is compromised. ^, Resulting overaccumulation of glutamate leads to excessive excitation of the glutamate receptors, leading to an increase in intracellular calcium and eventual neuronal cell injury and death. This process has been observed in experimental animals after 2 hours of circulatory arrest. Neuronal necrosis is selective and corresponds closely to distribution of excitatory amino acid receptors. The hippocampus, cerebellum, and basal ganglia, which have high concentrations of glutamate receptors, are characteristically most vulnerable to this injury, implying excitation as an underlying mechanism. ^,

Apoptosis, or programmed cell death, has been demonstrated experimentally in the neocortex of piglets following hypothermic circulatory arrest for 90 minutes at 19°C. Damaged neurons were observed between 8 and 72 hours after reperfusion. Caspase 3 and caspase 8, the principal cysteine proteases involved in apoptosis, were substantially elevated in these animals compared to control animals (no CPB or CPB without circulatory arrest). ATP levels were similar to those of control animals. Glutamate excitotoxicity secondary to hypothermic circulatory arrest has been shown to mediate neuronal apoptosis as well as necrosis.

Fessatidis and colleagues’ experimental studies using histopathologic techniques demonstrated that the cerebellum is the most vulnerable area of the brain to prolonged periods (>70 minutes) of hypothermic (15°C) circulatory arrest. ^,

EEG criteria for safe circulatory arrest are conflicting. In a study by Coselli and colleagues, the longest recorded durations of safe circulatory arrest in adults were in situations in which a full formal EEG had recorded electrocerebral silence (no electrical activity of cerebral origin at maximal gain, 2 µV · mm ⁻¹ ) for 3 minutes before the arrest. Mean nasopharyngeal temperature at this point was 16.9°C (range, 10.1°C–23.1°C).

In a subsequent study by Stecker and colleagues of 109 adult patients undergoing hypothermic circulatory arrest, electrocerebral silence was achieved at a mean nasopharyngeal temperature of 17.8°C (range, 12.5°C–27.2°C). Using a standardized protocol, this required cooling for a mean of 27.5 minutes (range, 12–50 minutes). Distributions of times to cool to various EEG events are shown in Fig. 2.4 . The time to cool to electrocerebral silence ( Fig. 2.5 ) was prolonged by high hemoglobin concentration, low arterial partial pressure of carbon dioxide, and slow cooling rates. Only 60% of patients demonstrated electrocerebral silence by either a nasopharyngeal temperature of 18°C or a cooling time of 30 minutes. Although cooling to an endpoint such as electrocerebral silence provides a more reproducible effect of hypothermia on the nervous system than cooling to a specific temperature (e.g., 12.5°C, which was sufficient to produce electrocerebral silence in all patients in this study), the optimal temperature for circulatory arrest could not be determined.

Distribution of nasopharyngeal temperatures at which various EEG landmarks occur. (A) Appearance of periodic complexes. (B) Appearance of burst suppression. (C) Electrocerebral silence. Examples of typical EEG patterns during cooling are also shown: (D) Precooling. (E) Appearance of periodic complexes. (F) Appearance of burst suppression. (G) Electrocerebral silence. Each of the EEG samples represents four channels recorded from the left hemisphere.

(From Stecker MM, Cheung AT, Pochettino A, et al. Deep hypothermic circulatory arrest: I. Effects of cooling on electroencephalogram and evoked potentials. Ann Thorac Surg . 2001;71:14.)

Relationship of interval (seconds) from beginning of circulatory arrest to appearance of electroencephalographic quiescence and nasopharyngeal temperature at time of circulatory arrest. (Note reversal of temperature scale.)

(Redrawn from Harden A, Pampiglione G, Waterston DJ. Circulatory arrest during hypothermia in cardiac surgery: an EEG study in children. Br Med J . 1966;2:1105.)

Ideally, electrocerebral silence on EEG indicates minimal cerebral metabolic demand. However, duration of cooling required to reach electrocerebral silence is variable in adult aortic surgery, and more than 30% of patients exhibit brain activity on EEG even at a temperature of 18°C; in 10% of patients, electrocerebral silence is not obtained even after core temperature below 10°C. Another limitation is that the EEG monitors only the superficial layers of the cerebral cortex on the crests of gyri directly abutting the skull; ischemia in deep brain regions, including the subcortex, may be undetected. Many factors can affect the sensitivity of EEG in detecting brain ischemia, including prior cerebral ischemia, anesthetic medications, hypothermia, CPB, and electrical cautery. Also, the EEG test has a subjective nature of the interpretation of full EEG waveforms. For these reasons, EEG monitoring is used less often in clinical practice nowadays.

Obtained by automated analysis of the electroencephalogram, the BIS is a dimensionless number ranging from 0 (complete brain inactivity) to 100 (indicating an awake and alert patient). The device employs an algorithm to analyze the EEG waveform, considering frequency, phase, and power spectrum to generate a BIS value. The method is mainly used for detecting intraoperative awareness and depth of anesthesia. Some centers use BIS for monitoring cerebral protection during thoracic aortic surgery, but the method has the same limitations as the standard EEG.

TCD is applied to measure the blood flow in the middle cerebral artery and is expected to optimize ACP during circulatory arrest, especially in patients who have an incomplete circle of Willis. However, the TCD measurement requires a sonographer’s expertise and depends on the images acquired through a small window. TCD is a sensitive tool for detecting arterial particles (gaseous and solid), which may embolize and lead to neurologic injury. However, there is no evidence that the routine use of transcranial Doppler can improve clinical outcomes after thoracic aortic surgery, and its use remains largely experimental.

NIRS determines the regional cerebral oxygen saturation by measuring the different absorptive properties of saturated and unsaturated hemoglobin (Hb) in the near-infrared spectrum (600-900 nm). The definition of cerebral desaturation in cardiac surgery has been a decrease in regional cerebral oxygen saturation by over 20% from baseline. NIRS has several limitations, including that it primarily monitors the anterior brain with a maximal depth of 3 to 4 cm of the brain tissue. Usually, NIRS monitoring assumes a fixed distribution of blood volumes between arterial and venous blood (25%-30% arterial to 70%-75% venous), and variability in arteriovenous blood distribution is ignored. NIRS is inaccurate for cerebral oxygenation with RCP during circulatory arrest.

No randomized clinical trials (RCTs) have shown superiority of the NIRS test over standard monitoring, and there are no data to support a threshold level or duration of NIRS-detected brain hypoxia that can be tolerated without detrimental neurologic events. However, in a recent study, Peng and colleagues assessed cerebral autoregulation (CA) by measuring a moving linear correlation coefficient (cerebral oximetry index, COx) between regional cerebral oxygen saturation measured by NIRS and mean blood pressure. A COx value approaching 1 implies that cerebral volume depends on blood pressure and CA is impaired, whereas a value approaching 0 suggests that blood pressure does not correlate with CBV (cerebral blood volume) and CA is functional. An average COx > 0.3 indicates impaired CA. Among 154 patients undergoing aortic arch surgery, impaired CA was associated with increased hospital mortality and morbidity.

Shaggy aorta is a severe arteriosclerotic state that Hollier and colleagues first described in 1991 ¹²⁰ as widespread atheromatous lesions in the aorta. There are few mural thrombi but much protuberant irregular-surfaced atheroma. Preoperative diagnosis is mainly made by direct echography or contrast computed tomography (CT) scans. Several grading scales of aortic shagginess have been proposed, including the thickness of the atheroma and area or length of the lesions. Yokawa and colleagues reported that highly shaggy lesions in the aortic arch or the thoracoabdominal aorta are risk factors for adverse brain or spinal cord neurologic outcomes.

Leukoaraiosis is a patchy punctuate or confluent hyperintensity in the white matter and deep gray nuclei on T2-weighted images. This white matter hyperintensity reflects chronic ischemic damage to myelin and axons. Several studies have demonstrated there is a significant relationship between presence of leukoaraiosis preoperatively and risk of postoperative stroke. ^, In an analysis of 131 patients undergoing elective aortic arch surgery, Morimoto and colleagues described adverse perioperative neurologic events including intraoperative stroke in 8 (6.1%) and TND in 11 (8.4%). Leukoaraiosis (odds ratio [OR], 1.1; P =.02) and aortic arch atheroma (OR 2.4; P =.001) were significantly associated with stroke on multivariate analysis, and TND was significantly associated with leukoaraiosis (OR 1.1; P =.03) and hippocampal atrophy (OR 1.6; P =.01).

Although it has been stated that very young patients experience less brain damage than older patients from hypothermic circulatory arrest, there is little factual support for this concept. Relative to the general population, cognitive, language, and motor performances are importantly reduced at age 4 years in infants younger than 3 months who underwent circulatory arrest. In a randomized trial of 171 neonates with D-transposition of the great arteries who had open repair using either hypothermic circulatory arrest or low-flow CPB, the circulatory arrest group at age 4 years had lower motor scores and more speech abnormalities ( P =.03). They also performed worse on tests of fine motor and visuospatial skills. At 8 years, the circulatory arrest group performed worse on tests of motor function ( P =.003), speech apraxia ( P =.01), visual motor tracking ( P =.01), and phonologic awareness ( P =.0003) than children in whom low-flow CPB was used. In adults in whom circulatory arrest is used, increasing age is an important predictor of both stroke and TND. ^, TND is a marker for long-term functional neurologic deficit.

The neurologic complications associated with aortic arch surgery historically have been categorized as permanent neurologic dysfunction (PND) and TND. PND is defined as either a focal (embolic stroke) or global (coma) deficit with corresponding defects on CT scans that persist at discharge from the hospital. ^, On the other hand, TND or non-focal neurocognitive dysfunction is commonly described as confusion, agitation, obtundation, or delirium that resolves before discharge, and CT scans of the brain in these patients are usually normal. The average rates of neurologic complications range from 7.3% to 12.8% for PND and 8.0% to 10.3% for TND.

Choreoathetosis has occurred early postoperatively in infants and children undergoing hypothermic circulatory arrest. ^{, , ,} When it occurs, it usually develops 2 to 6 days postoperatively. As time passes, the movements usually lessen in severity. If mild, they disappear completely, but if severe, they or hypotonia may persist. Brunberg and colleagues found no correlation between circulatory arrest time or depth of cooling (between 16°C and 20°C) and development of choreoathetosis. These reports suggest that this specific complication occurs in 1% to 12% of patients and that its residual effects are permanent in some. When choreoathetosis occurs, it is often in the setting of prolonged circulatory arrest.

Choreoathetosis has been observed in infants and children subjected to hypothermic CPB without circulatory arrest. There are suggestions that this complication can result from perfusion of the brain with very cold blood for a prolonged period at relatively high flows. This is the basis for the recommendation that arterial temperature not be reduced to less than 15°C.

The cause of choreoathetosis is unclear. Deep hypothermia per se may cause neurologic injury. Egerton and colleagues reported that continuous hypothermic perfusion at 10°C to 12°C produced moderate or severe brain damage, including choreoathetosis, in 10 of 16 patients (63%; CL 46%–77%). Air or particulate embolization to the brain may be a contributing factor. When circulatory arrest is used, choreoathetosis may be related to uneven brain cooling, leading to continued metabolic activity in the white matter and cerebellum (as reported by Reilly and colleagues ), and possibly to uneven brain reperfusion related to vascular changes associated with the no-reflow phenomenon. The latter finding lends support to the rationale for using hemodilution during cooling because absence of red cells in the perfusion used just before circulatory arrest to the brain eliminates the no-reflow phenomenon. Use of the alpha-stat strategy of acid-base balance has also been implicated as a causative factor.

Seizures have occurred in the early postoperative period in 5% to 10% of patients undergoing hypothermic circulatory arrest. ^{, ,} Because seizures are usually transient and followed by uneventful convalescence, they have not been considered major neurologic events. However, in an analysis from the Boston Circulatory Arrest Study involving 171 children with D-transposition of the great arteries, transient postoperative clinical and EEG seizures were associated with worse neurodevelopmental outcomes at ages 1 and 2.5 years, as well as neurologic and MRI-detected abnormalities at age 1 year. At age 4 years, occurrence of perioperative seizures was associated with lower IQ scores ( P =.01) and increased risk of neurologic abnormalities (OR 8.4; P =.05).

In a more recent prospective study of 178 neonates and infants less than age 6 months undergoing CPB with or without hypothermic circulatory arrest for a variety of congenital heart defects, including hypoplastic left heart syndrome and other forms of the single ventricle, EEG-recorded seizures occurred in 20 patients (11.2%; CL 8.8%–14.2%). Patients with duration of circulatory arrest of more than 40 minutes had more seizures (14 of 58, 24%; CL 18%–31%) than those with a duration of 40 minutes or less (4 of 59, 6.8%; CL 3.5%–12%; P =.04). Occurrence of seizures among patients with a duration of circulatory arrest of 40 minutes or less was similar among those in whom circulatory arrest was not used ( P =.38).

The comments concerning possible causes of choreoathetosis apply to seizures. However, it is well known that infants are highly susceptible to seizures from other causes, such as disturbances of thermoregulation and fluid balance, as well as from metabolic disorders, especially those related to glucose and calcium, and many of these factors may be operative in these patients.

Severe gross evidence of brain damage occurs uncommonly after hypothermic circulatory arrest in infants and children, including coma either dating from surgery or developing some hours later, followed by lasting impairment or death . In a study by Stewart and colleagues, 3 such instances (1.4%; CL 0.6%–2.7%) occurred among 218 young patients undergoing repair of the common types of congenital heart disease with hypothermic circulatory arrest; 5 other patients developed choreoathetosis. All these events occurred in the group of patients in whom core cooling alone was used. None occurred in patients whose circulatory arrest duration was less than 45 minutes, and the probability of developing major neurologic events increased as circulatory arrest time increased beyond this ( Fig. 2.6 ).

Relationship between probability of freedom from a major postoperative neurologic event and hypothermic circulatory arrest time in 219 infants younger than 3 months (8 events) undergoing open intracardiac operations. (See Appendix 2A , Equation 2A-5 .)

(Data from Stewart RW, Blackstone EH, Kirklin JW. Neurological dysfunction after cardiac surgery. In: Parenzan L, Crupi G, Graham G, eds. Congenital Heart Disease in the First Three Months of Life. Medical and Surgical Aspects . Bologna, Italy: Patron Editore; 1981:431.)

Focal neurologic damage resulting in serious neurologic impairment (stroke) occurs in adult patients following hypothermic circulatory arrest. Ergin and colleagues demonstrated that this form of injury is related to older age ( P <.0001), particularly beyond 60 years, presence of clot or atheroma in the aortic arch ( P <.0001), as well as longer duration of hypothermic circulatory arrest ( P <.0001). In the series of adult patients reported by Gega and colleagues in whom hypothermic circulatory arrest was used as the sole means of brain preservation, prevalence of stroke was 13.1% (8 of 61; CL 8.6%–19.1%) among patients in whom the duration of circulatory arrest exceeded 40 minutes. CT scans demonstrated that 62% of these strokes were embolic in origin and 38% were related to hypoperfusion.

The effect of hypothermic circulatory arrest on late postoperative intellectual capacity and behavior in infants and children has been difficult to study. Problems in testing infants preoperatively so that each may serve as their own control contributes to the difficulty. Associated congenital developmental disorders, possible adverse effects before operation of severe congenital heart disease, and effects of other perioperative events complicate interpretation of the data.

Results of psychomotor testing in 146 children undergoing cardiac surgery during hypothermic circulatory arrest early in the experience with this technique, obtained by combining the three largest reported series, are summarized in Table 2.3 . ^{, ,} Late postoperatively, 23 of the 146 (16%; CL 13%–19%) had an IQ of 80 or less, more than 1 standard deviation below the test mean. In approximately half these patients, preoperative events were considered likely to account for the low scores. In the remainder, an occasional child suffered an adverse perioperative event, but the low scores were unexplained in nine (6.2%; CL 4.1%–9.0%) patients.

Wells and colleagues obtained data on intellectual and psychological development in children that caused them to question the idea that 60 minutes of circulatory arrest at 18°C is safe. They found that verbal ( P =.06), quantitative ( P =.07), and general cognitive ( P =.003) IQ scores of patients with an arrest time of 50 minutes or more were lower late postoperatively than those of patients with an arrest time of less than 50 minutes.

The first RCT comparing prevalence of brain injury after corrective heart surgery in infants with D-transposition of the great arteries using deep hypothermia, predominantly with circulatory arrest or low-flow CPB, was conducted at Boston Children’s Hospital. This study demonstrated that infants in whom circulatory arrest was used had a higher prevalence of neurologic abnormalities and poorer mental function at age 1 year, and poorer expressive language and motor development at age 2.5 years. Follow-up studies of the same cohort at age 4 showed that circulatory arrest is associated with worse motor coordination and planning but not with lower IQ or worse overall neurologic status. However, neither IQ nor overall neurologic status was correlated with duration of circulatory arrest. In the cohort as a whole, cognitive, language, and motor performance were reduced relative to the general population.

In summary, there is increasing evidence that intervals of hypothermic circulatory arrest of 40 minutes or more are associated with brain injury in infants, children, and adults. Early experience at the Mayo Clinic suggested that 45 minutes was the maximum safe duration even when nasopharyngeal temperature was reduced to 20°C.

At normothermia, at least in rats, 20 minutes of circulatory arrest to the kidney produces no histochemical evidence of cell death, whereas 30 minutes produces extensive cell death in the distal portion of the proximal convoluted tubules, with scattered areas of cell death being seen at 25 minutes. Vogt and Farber identified progressive accumulation of lactic acid during ischemia as a causative factor and rapid decrease of ATP to 20% of control values as an indicator of impending renal death.

Hypothermia prolongs the safe circulatory arrest time for the dog kidney. Ninety minutes of circulatory arrest after surface cooling to 18°C to 20°C produces no late morphologic changes in the kidney, but precise relationships among temperature, duration of circulatory arrest, and morphologic and functional renal damage are unclear. Gowing and Dexter suggest that minimal morphologic changes evolve in the rat kidney after 60 minutes of circulatory arrest at 21°C. It is apparent, however, that at any temperature, the safe circulatory arrest time for the kidney is longer than it is for the brain and shorter than for the liver. In addition, a scattered loss of cells through cell death probably results in no detectable loss of renal function, whereas this may not be true in the brain.

As with other organs, the question of damaging effects of hypothermia per se is not fully resolved. Ward found fewer morphologic and functional derangements of the kidney after 90 minutes of circulatory arrest at 15°C than at either lower or higher temperatures. This suggests that temperatures less than 15°C may damage the kidney.

Important oliguria beginning about 12 hours postoperatively occasionally complicates recovery of infants operated on with hypothermic circulatory arrest for less than 60 minutes. Venugopal and colleagues reported 4 deaths (3%; CL 2%–6%) from renal failure among 130 patients operated on with surface-induced hypothermic circulatory arrest. Among patients who died, renal failure was the mode of death in 14%.

The primary cause of renal failure appears to be low cardiac output after operation. However, in at least some cases, severe oliguria develops when the patient’s hemodynamic state appears to be adequate. Because of the finding in experimental studies that morphologic and functional damage to the kidney does not occur after 60 minutes of circulatory arrest at temperatures of 18°C to 20°C ( Fig. 2.8 ), damaging effects from CPB must be implicated. In part, this may result from low cardiac output preceding and following the interval of circulatory arrest. In part, it may be due to damage to the kidneys by free hemoglobin and circulating toxins that appear during CPB (see Section II ). Free hemoglobin has been found in the renal tubules of some of these patients at autopsy.

Freehand nomogram for the kidney of the relation between probability of safe total circulatory arrest and duration of circulatory arrest at two temperatures. Normothermic relationship is based on Vogt MT, Farber E. On the molecular pathology of ischemic renal cell death. Reversible and irreversible cellular and mitochondrial metabolic alterations. Am J Pathol. 1968;53:1, and the hypothermic one on data presented in the text.

In 161 adult patients undergoing resection of the distal aortic arch and descending thoracic and thoracoabdominal aorta in whom hypothermic circulatory arrest was used (mean nasopharyngeal temperature, 14.5°C; mean interval, 38 minutes; longest interval, 62 minutes), prevalence of postoperative renal failure requiring dialysis among 157 operative survivors was 2.6% (4 patients; CL 1.3%–4.6%). Among the subgroup of 18 operative survivors who had evidence of renal dysfunction preoperatively (serum creatinine level > 1.5 mg · dL ⁻¹ ), none developed renal failure that required dialysis.

The ideal method for creating negative pressure for venous input into the pump-oxygenator is by a regulated and monitored vacuum pressure system coupled to the venous reservoir. The patient and machine can be at or near the same vertical level from the operating room floor, reducing the length of the venous line and hence the priming volume. The negative pressure does not rise above the controlled level if the cannula becomes occluded, and the negative pressure can be varied as needed. Most importantly, the two pressures (i.e., output pressure to the patient and input pressure from the patient) are uncoupled and can be varied independently with an arterial roller pump. If a controlled vortex pump is used, the vacuum pressure in the venous system will reduce the outflow pressure of the nonocclusive vortex pump, thus requiring higher rpm to achieve a constant flow.

Use of a hard-shelled venous reservoir in currently available oxygenators and a vacuum regulator connected to wall suction set at −40 to −60 cm H ₂ O has allowed vacuum-assisted venous return (VAVR) to become widely accepted. ^, VAVR permits use of smaller venous cannulae, smaller reservoirs, considerably shorter tubing, and low priming volume. It is of considerable value for cardiac operations performed in infants and through small incisions in children and adults (see “ Special Situations and Controversies ” in Section III ). ^, In a study by Banbury and colleagues at Cleveland Clinic, VAVR reduced priming volume from 2.0 ± 0.4 L to 1.4 ± 0.4 L ( P <.0001), increased hematocrit both on bypass and immediately postbypass ( P <.0001), and reduced use of blood products both intraoperatively and postoperatively from 39% of patients to 19% ( P =.002). The disadvantage of VAVR is a potential increase in microemboli at higher vacuum settings. It is also important to avoid sudden increases in the volume of air pushed into the hard-shell venous reservoir. For instance, this can happen with removal of a ventricular drainage catheter attached to a roller pump rotating at high rpm. If the air pulled into the venous reservoir is not adequately vented, it can push blood followed by a column of air up the venous return tubing and into the heart. Such complications have occurred. They led to the design of alarms to detect positive reservoir pressure and one-way valves (duckbill design) to relieve positive reservoir pressure.

A common method of generating the negative pressure gradient is through siphonage. Disadvantages of this approach include an imposed difference in the levels of patient and pump-oxygenator, the relatively narrow range of negative pressures generated in the operating room by its use, and its interruption by large boluses of air in the venous line. Most importantly, the need for a reservoir increases the filling (priming) volume of the pump-oxygenator. It is, however, simple, reliable, effective, and inexpensive.

The controlled vortex pump permits direct pumping from the patient’s venous system and is more effective and safer than a roller pump. The potentially large pressure gradient between the tip of the venous cannula and right atrium or venae cavae must be controlled in some way to prevent “fluttering” of their walls around the end of the cannulae. One way of accomplishing this is to use small venous cannulae to impose considerable resistance between the pump and tip of the cannula rather than between the tip of the cannula and the patient’s venous system. This is fortuitously advantageous in percutaneous peripheral cannulation because an 18F or 20F venous cannula of some length can be easily passed into the venous system of a normal-sized adult and provides adequate venous drainage. It also facilitates minimally invasive cardiac surgery. By contrast, 28F to 32F catheters are required when gravity drainage is used.

With present-day oxygenators, maintaining Pao ₂ at about 250 mmHg is easily accomplished. Higher Pao ₂ is unnecessary and theoretically subjects patients to the risk of oxygen toxicity and bubble formation. Pao ₂ lower than about 85 mmHg results in a declining arterial oxygen content (Cao ₂ ) (according to the oxygen dissociation curve of blood that varies according to temperature and pH) and a corresponding reduction of tissue and mixed venous oxygen levels. Shepard demonstrated that when arterial oxygen saturation (Sao ₂ ) fell below 65% in dogs undergoing normothermic CPB, V ˙ o 2 fell, indicating hypoxic cell damage.

Pa o ₂ is related to temperature of the patient, which is related to V ˙ o 2 (see Fig. 2.1 ), blood flow rate ( Q ˙ ), performance of the oxygenator, and, in a complex fashion, to ventilating gas flow rate and composition (see “ Gas Exchange ,” earlier). Reducing the patient’s body temperature reduces V ˙ o 2 and increases P v ¯ o 2 , resulting in increased Pa o ₂ ^, . During rewarming by perfusion from the pump-oxygenator, the increasing V ˙ o 2 and the metabolic debt that has accumulated result in relatively low P v ¯ o 2 . ( Fig. 2.10 ). This period, then, places maximal demands on the oxygen transfer capacity of the oxygenator. Flows that are required for patients with high BSA (body surface area) may surpass the capacity of the oxygenator resulting in low venous oxygen saturation in venous return. Initiating partial CPB usually remedies the problem, although splicing another smaller (e.g., pediatric) oxygenator into the circuit may be necessary.

Hemoglobin saturation (top) and temperature (bottom) during rewarming on CPB. Note sharp decrease in mixed venous oxygen saturation (open circles in upper panel) as rewarming proceeds. “Entering” saturations and temperatures are those in the arterial tubing, and “leaving” saturation and temperatures are those in the venous return tubing.

(From Theye RA, Kirklin JW. Vertical film oxygenator performance at 30°C and oxygen levels during rewarming. Surgery. 1963;54:569.)

Arterial carbon dioxide pressure (Pa co ₂ ) is controllable during CPB by varying the ratio between gas flow rate into the oxygenator ( V ˙ , or ventilation · min ⁻¹ ) and Q ˙ through the oxygenator. This is facilitated by use of microporous or true membrane oxygenators, because V ˙ is not the force driving blood through the oxygenator, as is the case in bubble oxygenators, and Pa o ₂ is well maintained over a wide range of V ˙ . Inline P co ₂ and pH meters facilitate control of Pa co ₂ and pH.

Some clinical perfusions for cardiac surgery are performed at normothermia (≈37°C) and others at various levels of hypothermia: mild (30°C–35°C), moderate (25°C–30°C), or deep (<25°C). Therefore, it is necessary to consider the strategy for controlling Pa co ₂ and, indirectly, pH. The alpha-stat strategy is based on (1) using the pH measured at 37°C and uncorrected for the temperature of the patient’s blood and (2) maintaining this level at pH 7.4. That is, the ventilation of the oxygenator is maintained at a level appropriate for a body temperature of 37°C, no matter how low the temperature. This hyperventilation during hypothermia results in a decrease in Pa co ₂ and an increase in pH when these values are corrected for the temperature of the patient’s blood. Swan and Reeves and Rahn and colleagues have all emphasized that at low temperatures, neutrality exists at a higher pH than at normothermia, because of the change of the dissociation constant of water with temperature. The alpha-stat strategy results in optimal function of a number of important enzyme systems, including lactate dehydrogenase, phosphofructokinase, and sodium-potassium ATPase. Perhaps more importantly, alpha-stat strategy preserves cerebral vascular responses to variations in flow and blood pressure (CA).

In contrast, the pH-stat strategy strives for the same values of pH and Pa co ₂ , corrected to the temperature of the patient’s blood, during hypothermia as at normothermia. This represents a state of respiratory acidosis and hypercarbia. Cerebral blood flow usually increases under these circumstances. This may be considered advantageous in some situations, but so-called luxury perfusion may expose the brain to a larger number of microemboli than would otherwise be the case and, therefore, could be disadvantageous. ^,

At a cellular enzyme level, the alpha-stat strategy may be preferable, but which is preferable in clinical cardiac surgery in neonates, children, and adults is the subject of continued investigation and debate. The alpha-stat strategy results in a lower Pa co ₂ , which may adversely affect cerebral blood flow. This may be of particular importance for patients with cyanotic congenital heart disease (e.g., tetralogy of Fallot with pulmonary atresia) for whom low Pa co ₂ may result in pulmonary vasodilatation in addition to cerebral vasoconstriction. Thus, there can be a steal of blood from the cerebral to the pulmonary vascular bed. ^, Several studies in infants suggest that pH-stat management results in superior neurologic outcome during deep hypothermic CPB and hypothermic circulatory arrest. ^{, ,} The pH-stat technique may depress cardiac function. However, at least in dogs, regional distribution of blood flow during normothermic and hypothermic full-flow CPB is similar with the alpha-stat and pH-stat strategies.

A recent review of 16 best-evidence published papers concluded that better results were achieved with the alpha-stat technique in adult patients and with the pH-stat technique in pediatric patients.

The diluent (which is used to prime the pump-oxygenator system, wholly or in part, and for any erythrocyte-free additions during CPB) is a balanced electrolyte solution with a near-normal pH and an ion content resembling that of plasma. There is some evidence for the concept that in both adults and young patients, it is disadvantageous to include either glucose or lactate in the priming solution ^, (see “ Biochemical Milieu ” under Brain Function and Structure: Risk Factors for Damage in Section I). However, priming solutions containing glucose and lactate are used in some centers.

In pump-oxygenator systems without a venous reservoir or those using VAVR, mixing of blood and prime may be partially or wholly avoided, with residual prime discarded.

In intact humans at 37°C, the normal hematocrit of 0.40 to 0.50 is optimal rheologically and for oxygen transport (assuming a normal red blood cell hemoglobin concentration). This provides sufficient oxygen delivery to maintain normal mitochondrial P o ₂ levels of about 0.05 to 1.0 mmHg, and average intracellular P o ₂ levels of about 5 mmHg, these being reflected in normal P v ¯ o 2 of about 40 (mixed venous oxygen saturation [ S v ¯ o 2 ] of about 75%). When the hematocrit is abnormally high, oxygen content is high, but the increased viscosity tends to decrease microcirculatory blood flow. The rate of oxygen transport varies directly with hematocrit (because oxygen content varies directly with hematocrit, assuming normal red blood cell hemoglobin concentrations and adequate oxygenation) and inversely with blood viscosity (which is also determined primarily by hematocrit). Hypothermia increases blood viscosity; therefore, at low temperatures, a lower hematocrit is more appropriate than at 37°C. This is partially due to the formation of erythrocyte aggregates (e.g., rouleaux) that is enhanced by higher hematocrit, sluggish flow, and lower temperature among other factors.

A lower-than-normal hematocrit appears desirable during hypothermic CPB because the perfusate has a lower apparent viscosity and low shear rates and provides better perfusion of the microcirculation. Thus, a hematocrit of about 0.20 to 0.25 may be optimal during moderately and deeply hypothermic CPB, although a low hematocrit could predispose the patient to neurologic dysfunction, particularly when it exists during a period of low CPB flow and also in elderly and diabetic patients with poor cerebral regulation of blood flow. Several studies of infants suggest that a hematocrit of 0.25 is associated with better neurologic outcome than one of 0.20, but there is no incremental improvement for hematocrit greater than 0.25 (up to 0.35). ^, During rewarming, a higher hematocrit may be desirable because of increased oxygen demands, and the higher apparent viscosity of a higher hematocrit is appropriate during normothermia. This may be achieved by ultrafiltration (see “ Components ” under Pump-Oxygenator in Section III) or by adding packed red blood cells if the blood volume is too low to allow this.

The need for and amount of additional blood or packed red blood cells to achieve a desired hemoglobin concentration during CPB can be determined before the start of CPB ( Box 2.2 ). A blood-free priming solution is used if the calculated hematocrit is in the desired range. If the calculated hematocrit is lower than desired, an appropriate amount of blood (or packed red blood cells) is added.

Banked blood, preferably less than 48 hours old, is preferred, but older blood is accepted for adults when necessary. Banked blood is rendered calcium-free by the anticoagulant solution (citrate-phosphate-dextrose [CPD]) and is acidotic, so additions of heparin, calcium, and buffer may be required before placing it in the pump-oxygenator before or during CPB. However, in at least a few institutions, no calcium is added before placing the blood in the pump-oxygenator, and none is added after that until the patient’s nasopharyngeal temperature reaches about 28°C during the rewarming process. Monitoring the ionized calcium level is important, and calcium is added if necessary. (The normal level is about 1.2 mmol · L ⁻¹ , with total calcium being about 2.5 mmol · L ⁻¹ or 10 mg · dL ⁻¹ .) This practice results in extremely low ionized calcium levels when CPB is first established. Unduly high ionized calcium levels could be more harmful (see “ Damage from Global Myocardial Ischemia ” in Chapter 3 ). A reasonable practice would be to initially add 3 mL of calcium chloride (10%) rather than 5 mL for each unit of banked blood used and then add no more until the ionized calcium is measured.

Concentration of albumin in the mixed patient-machine blood volume, as well as of hemoglobin, is affected by the amount of hemodilution. Theoretically, according to the Starling law of transcapillary fluid exchange (see “ Pulmonary Venous Pressure ” later in this section), a reduction of albumin and, thus, of the colloidal osmotic pressure of the plasma accentuates movement of fluid out of the vascular space into the interstitial space. That this occurs is indicated by the work of Cohn and colleagues, who showed that extracellular fluid volume increases more rapidly when hemodilution is used than when it is not.

During CPB, microvascular permeability to macromolecules is increased ; some of the administered albumin leaks into the interstitial fluid and has an unfavorable effect on the relationships expressed in the Starling law. Homologous albumin may provoke an allergic response, which also increases microvascular permeability and causes albumin leakage into the interstitial fluid.

These complex interrelations probably explain the failure of a randomized trial to find a favorable effect from adding homologous albumin to the prime in adults. It is not uncommon for cardiac surgical patients, particularly the elderly, to present with low-normal or below-normal albumin levels. Whether albumin concentration should be maintained at normal levels in some special situations such as this remains arguable.

Other colloidal solutions (e.g., hydroxyethyl starch) can also be added to the priming solution to attenuate fluid loss from the intravascular space. However, none has been conclusively shown to have a beneficial effect.

Practices vary regarding addition of substances and drugs to the perfusate (by administering them into the priming volume of the pump-oxygenator or patient before CPB, or into the patient or pump-oxygenator during CPB), other than basic balanced salt solution and blood and its required additives.

Use of an osmotic diuretic may be advisable. Mannitol (≈0.5 g · kg ⁻¹ ), a pure osmotic diuretic, can be included as part of the prime. Mannitol also has the advantage of being an effective agent against oxygen free radicals generated during CPB. ^, Glucose (added to the prime of the pump-oxygenator in sufficient quantity to obtain a glucose concentration of about 350 mg · dL ⁻¹ in the prime) also produces diuresis. However, its use in the priming volume and its administration during and early after CPB, employing more than moderate hypothermia, may be unwise because of the strong suggestion that hyperglycemia during cooling and early after hypothermic circulatory arrest increases the probability of brain injury. ^{, ,}

Administration of a potent diuretic during CPB is generally useful. Incorporating furosemide in the pump prime is practiced by many groups. It may be more advantageous to give it as a bolus in a dose of 1 to 2 mg · kg ⁻¹ at the start of rewarming, either after an interval of circulatory arrest or moderately or deeply hypothermic CPB.

The short-acting adrenergic α-receptor blocking agent , phentolamine, can antagonize the vasoconstriction produced by catecholamines and has been shown to produce more uniform body cooling and rewarming and improved tissue perfusion when given during CPB. A 0.2 mg · kg ⁻¹ bolus is administered just after the start of CPB and the initiation of cooling. When circulatory arrest is used, an additional dose of 0.2 mg · kg ⁻¹ is administered with the resumption of CPB for rewarming.

Alternatively, the long-acting adrenergic α-receptor blocking agent , phenoxybenzamine, can be used in infants and children to produce total α-blockade for 8 to 10 hours. It is given in a dose of 1 mg · kg ⁻¹ about 15 minutes before commencing CPB and at the beginning of rewarming after the period of circulatory arrest.

Opinions differ about the advisability of routinely administering (or adding to the perfusate) corticosteroids and the appropriate agent to use. Available evidence suggests that corticosteroids improve tissue perfusion and lessen the increase in extracellular water that usually accompanies CPB. Although some studies have reported improved clinical status when steroids are given in the manner described, this matter remains controversial. ^, Methylprednisolone in a single dose of 30 mg · kg ⁻¹ or dexamethasone in a single dose of 1 mg · kg ⁻¹ given at the onset of CPB and not repeated may be advantageous. These agents do not appear to reduce complement activation, but there is evidence to support the hypothesis that they attenuate complement-mediated leukocyte activation, particularly that associated with reperfusion of the heart and lungs in the latter part of CPB. ^, In piglets, corticosteroids provide brain protection during operations that involve hypothermic CPB and circulatory arrest. ^,

ε-Aminocaproic acid (EACA) and tranexamic acid are two antifibrinolytic agents that can be administered before, during, and after CPB to reduce bleeding and the need for allogeneic blood transfusions. EACA is administered using an empirical dose of 10 g before the skin incision, 10 g during the procedure, and 10 g early postoperatively. Alternatively, it can be given at a dose of 150 mg · kg ⁻¹ at the time of the skin incision, with an additional 30 mg · kg ⁻¹ for 4 hours upon initiation of CPB. Tranexamic acid is given at a dose of 1 g before the skin incision, 500 mg in the pump prime, and 400 mg · h ⁻¹ during the procedure.

A drug that inhibits complement activation during CPB (soluble human complement receptor type 1 [TP10]) was developed in hopes of ameliorating the adverse effects of CPB mediated by inflammation. Trials of TP10 in high-risk patients showed inhibition of complement activation but no consistent improvement in outcome assessed by a composite endpoint. TP10 is not approved for use in CPB.

During CPB for cardiac surgery, blood loss in the operative field and gradual increase in interstitial fluid and urinary output combine to steadily deplete the patient-machine blood volume. Usual practice is for the perfusionist to add increments of a balanced electrolyte solution to maintain the volume at a safe level; in adults, up to 2000 mL may be added. Unless special precautions are taken, such as avoiding return of irrigating fluids to the pump-oxygenator by cardiotomy pump suckers and using ultrafiltration during the final stages of CPB, severe hemodilution results and persists into the postbypass period.

In neonates and infants, ultrafiltration immediately after CPB (before removal of cannulae) is often advisable using the modified ultrafiltration (MUF) technique introduced by Elliot. Others have confirmed its efficacy. ^, In children and adults, ultrafiltration may be performed during the latter part of CPB if the hematocrit is below about 0.25 and there is excess volume in the pump-oxygenator. If not, it may be performed after discontinuing CPB, slowly circulating blood through the patient before any cannulae are removed. A third option, and one that is frequently used, is ultrafiltration of the volume remaining in the pump-oxygenator after CPB is discontinued and the venous cannulae have been removed. Hemoconcentrated pump-oxygenator volume is then infused slowly into the patient before the arterial cannula is removed (see “ Pump-Oxygenator ” in Section III).

Initial response is probably humoral, initiated by the contact of plasma with the foreign surfaces of the tubing and pump-oxygenator and with air. Gas exchange requires a large surface area; therefore, the greatest stimulus to this response occurs in the oxygenator. Humoral response appears to begin with activation of specialized plasma proteins, developed and conditioned through centuries of life to recognize and repel transcutaneous invaders. Whereas previously this invasion has generally been a relatively small, localized, and often extravascular process, in the patient exposed to a pump-oxygenator, it is a massive intravascular process. Even though the patient is heparinized, parts of the coagulation cascade respond virtually immediately to the activating capability of the foreign surface, as do the complement, kallikrein, fibrinolytic, and other cascades. Activation of Hageman factor (factor XII) may be the initial event in activation of these cascades, although platelets appear to be independently activated at about the same time. Nearly all the split products resulting from these multiple activations can be found in the patient’s blood during and, for a time, after bypass. Mechanisms for their disappearance have not been elucidated, but presumably they are to some extent metabolized, taken up by specific cell-surface receptors, dissipated into extravascular fluids, including peritoneal and pleural fluids, and excreted in the urine.

Products of activation of these cascades have powerful physiologic effects, both directly and by activation of other systems and cells. The complement cascade, once activated, results in the production of powerful anaphylatoxins (C3a and C5a) that increase vascular permeability, cause smooth muscle contraction, mediate leukocyte chemotaxis, and facilitate neutrophil aggregation and enzyme release. ^, Complement activation occurs through either the classic or the alternative pathway.

Contact activation of Hageman factor also immediately initiates the kallikrein-bradykinin cascade, resulting in the production of bradykinin. Plasma kallikrein circulates in the blood as a precursor, prekallikrein, 75% of which is bound to HMWK in the plasma. Bradykinin, formed largely from HMWK, increases vascular permeability, dilates arterioles, initiates smooth muscle contraction, and elicits pain. Kallikrein also activates Hageman factor and plasminogen to form plasmin, again demonstrating the complex interactions and feedback loops between the various reactions of blood to nonself.

Once activated, the contact activation system overcomes its normal regulating system, and all the responses are amplified. Because plasma kallikrein leads to conversion of plasminogen to plasmin, whose basic function in the circulation is to digest fibrin clots and thrombi, the fibrinolytic cascade is activated by this and other humoral and cellular mechanisms.

Blood cells and endothelial cells participate in the nonspecific inflammatory response to use of a pump-oxygenator. Lymphocytes (both antibody-forming B cells and T cells) are part of the specific immune system and, as indicated earlier, participate little in the response to CPB in the usual immunologically naive patient. Eosinophilic granulocytes also seem to have limited participation. Basophilic granulocytes (mast cells) may well participate, but the extent to which they do so is not clear, and the same is true of the natural killer (NK) cells within the leukocyte family. Monocytes, once activated, participate in the cellular response.

Neutrophilic granulocytes (polymorphonuclear leukocytes) play a major role in the response to CPB. Neutrophils are activated by complement and other soluble inflammatory mediators. When activated, they migrate directionally toward areas of higher complement concentration (usually in the tissues, but during CPB, probably in blood), change their shape, become more adhesive, and secrete cytotoxic substances, including oxygen-derived free radicals. Of importance—and possibly a clue as to why most patients recover uneventfully from cardiac operations in which CPB is used, despite the strong humoral and cellular response—is the fact that complement can also desensitize neutrophils, thereby reducing their ability to participate in the inflammatory response. Neutrophils are also activated by other humoral agents participating in the cascades in the blood, including kallikrein, as well as by other inflammatory mediators (cytokines) generated by cells, including tumor necrosis factor (TNF) and platelet-activating factor (PAF). These molecules also have been shown to increase in amounts during and early after CPB.

Platelets are strongly affected by CPB using a pump-oxygenator but in a complex manner that Edmunds and colleagues have well summarized. ^, As in the case of neutrophils, platelets must be activated from their normally passive state; this occurs within 1 minute of the start of CPB. The precise initial trigger is uncertain, but possibilities include direct surface contact, abnormal shear stresses, mechanical lysis, exposure to adenosine diphosphate, and unidentified chemical agonists. The mechanism for activation of platelets is exposure on the surface of the platelet of numerous specific membrane receptors. Exposure of the fibrinogen glycoprotein receptors (GPIIb-IIIa complex) and subsequent binding of fibrinogen to them are essential for adherence of platelets to the foreign surfaces of the pump-oxygenator and for their aggregation. Many other specific receptor sites are expressed and exposed by activated platelets. Control, feedback, and amplification mechanisms regulate platelets as well as the humoral systems, all of which are involved in the response to CPB.

Endothelial cells do not pass through the pump-oxygenator, but their complex activities are affected while the patient is connected to it. Triggering mechanisms are not clearly defined, but they probably include abnormal pressures and shear stresses, localized ischemia, and increased concentrations of normal and abnormal substances and cells in the blood. As a result, endothelial surface receptors are exposed, substances are elaborated and extruded, and spaces between the endothelial cells and their membranes are enlarged.

Endothelial and other cells, particularly those in the locally ischemic areas that surely exist during CPB, express phospholipid molecules derived from arachidonic acid (eicosanoids). These are important mediators of inflammation and include prostaglandins, thromboxanes, leukotrienes, and lipoxins. Other cells in areas of acute inflammation that may be present during CPB can produce soluble factors (cytokines) that normally act on other cells to regulate their function; after CPB, they can induce elevation of body temperature, among other things.

Magnitude of the acute elevation of catecholamine levels in the blood that develops during CPB (see “ Catecholamine Response ” later in this section) is a measure of severity of the stress reaction induced by most cardiac surgery using CPB. Thus, in addition to the responses induced by CPB, cardiac operations and CPB induce the important perturbations associated with other major operations and trauma. Characteristics of this “metabolic response to stress” have been intensively studied by a number of investigators and clinicians. Among the first was Cuthbertson in 1930, and among the most prominent, Francis D. Moore. The essence of this process has been well summarized by Wilmore :

“The human body responds to these stresses with dramatic resilience. For example, following injury, clotting mechanisms are immediately activated to reduce blood loss; body fluids shift from the extravascular compartment to restore blood volume; blood flow is redistributed to ensure perfusion of vital organs; and respiratory and renal functions compensate to maintain acid-base neutrality and body fluid tonicity. Following these acute adaptations, other changes occur; these responses are more gradual and prolonged but are apparently necessary for recovery of the injured organism. A variety of immunologic alterations are initiated; leukocytes are mobilized, macrophages and specialized T cells are produced, and ‘acute phase’ plasma proteins are synthesized by the liver. Inflammatory cells invade the injured area, set up a perimeter defense, and engulf the dead and dying cells and other wound contaminants. These initial steps are followed rapidly by ingrowth of blood vessels, appearance of fibroblasts that build collagen scaffolding, and a host of other local changes that aid wound repair. Local changes that occur at the injury site are accompanied by systemic alterations in body physiology and metabolism. Cardiac output is elevated, minute ventilation is increased, and the patient becomes febrile. Lipolysis and skeletal muscle proteolysis are accelerated, providing an ongoing fuel supply and an immediate source of amino acids that are utilized for wound healing and synthesis of “acute phase” proteins and new glucose. The glucose provides essential energy for the brain and other vital organs and for healing of the wound.”

Phenomena associated with CPB not only produce their own damage but also interfere with the metabolic response to stress, a process necessary for recovery. Uneventful recovery of most patients after cardiac surgery means that a vast array of control and counteractive phenomena of both humoral and cellular types is in place, many of which await discovery and exploitation.

During CPB, an initial mild leukopenia develops, which soon returns to baseline values. Similar changes occur without an oxygenator in the system and are in part the result of transient movement of leukocytes out of the vascular system. By the end of CPB, leukocytosis is present, consisting mostly of mature segmented forms of neutrophils (coming primarily from the bone marrow, most of which are activated). ^, Leukocyte count often increases to a peak of 12,000 to 24,000 cells · mL ⁻³ at 24 to 28 hours postoperatively. Both T and B lymphocytes are decreased early after CPB, and T-cell function is decreased.

Pulmonary sequestration of neutrophils occurs during CPB. An inflammatory response follows their disruption and release of proteolytic and vasoactive substances and powerful lysosomal enzymes, contributing to the increased vascular permeability associated with CPB (see “ Complement Activation ” later in this section). ^, Also, activation of neutrophils during CPB by the C3a and C5a complement fragments liberates oxygen-derived free radicals; this contributes to the damaging effects of CPB. ^, Neutrophil elastase, a connective tissue protease and product of neutrophil activation that appears in plasma, is considerably increased by CPB, and the peak concentration correlates positively and closely with the duration of CPB. Such proteases break down elastin, collagen, and fibronectin, destroying extracellular structures, and contribute to the capillary leak that leads to postoperative extracellular volume overload and electrolyte imbalance.

Neutrophils in healthy persons are distinct cells. By inference, these cells are inactive and unprepared for the numerous harmful effects they exert during and early after CPB. However, when stimulated, neutrophils transiently aggregate and cluster with each other and to other cell types, such as vascular endothelial cells. ^, The process of aggregation and clustering is rapid, mediated by cell adhesion molecules (selectins and cell adhesion molecules (CAMs)), and a critical step in development of inflammatory and immune responses. In myocardial infarction, anti-CAM antibodies of specific types, which can be produced by monoclonal techniques and can attenuate or prevent neutrophil aggregation and clustering, have been shown to considerably reduce the extent of cell death produced by ischemia and reperfusion.

Nifedipine, infused during CPB in doses of about 6 µg · kg ⁻¹ · h ⁻¹ , appears to inhibit neutrophil activation. Hypothermia has been shown to delay, although not prevent, the expression of neutrophil adherence molecules. ^, Other interventions that have been evaluated clinically and experimentally to reduce the adverse effects of neutrophil activation include leukocyte filtration and pharmacologic agents (aprotinin, oligosaccharide antagonists, antioxidants, corticosteroids, and omega-3 fatty acids). ^{, ,} Gillinov and colleagues used the antiinflammatory agent NPC15669 to inhibit neutrophil adhesion in a CPB model and found a marked decrease in pulmonary injury.

In vitro test circuits show that the platelet count (corrected for dilution) decreases within 2 minutes of the beginning of extracorporeal circulation to about 80% of the pre-CPB level. By 8 minutes, the count has decreased to about 70% of the pre-CPB level and then stays close to that level during the rest of CPB and the period thereafter. Decrease in platelet count during clinical CPB tends to be greater than this because of hemodilution. Cardiotomy sucker systems substantially reduce platelet count. Interestingly, membrane oxygenators are associated with greater reduction than bubble oxygenators. ^, As a result of these and other factors, the number of platelets in circulating blood post-CPB decreases to about 60% of the prebypass value and does not correlate with duration of CPB. ^,

Sequestration of platelets in the liver and other organs during CPB in humans is slight and, therefore, not a major factor in reducing the platelet count. Something other than mere loss of platelets by adhesion to foreign surfaces is involved, because the platelet count continues to be low in some patients for as long as 72 hours postoperatively. One factor may be reduced survival time of platelets after CPB. Shear stresses likely do not reduce either the number or function of platelets.

Qualitative changes that occur in the platelets of patients undergoing CPB are more complex and probably more important than the change in numbers. Normally, platelets adhere only to cut ends of blood vessels and to subendothelial surfaces (presumably because subendothelial collagen causes them to adhere). Once CPB begins, platelets almost immediately adhere to foreign (nonendothelial) surfaces. Once this process begins, platelets also begin to clump (aggregate), primarily on the foreign surfaces they have already adhered to. There is some evidence that initial aggregation is in small clumps (primary aggregates) capable of disaggregating. However, if the stimulus is strong, these primary aggregates are transformed into larger aggregates. It is believed that only then do platelets begin to release the contents of their granules and become irreversibly activated. ^, Aggregates break off on occasion and become particulate emboli. Either platelets aggregate and disaggregate for many days after CPB or the aggregates formed during CPB persist in the circulation because platelet aggregates can be seen passing through the retinal vessels of patients for days after cardiac surgery with CPB.

As blood circulates normally through endothelially lined tubes, platelets are generally inactive. The stimulus to platelet adherence and aggregation on the surfaces of the pump-oxygenator system, whatever it may be, also activates the platelets, a process that changes their form and internal architecture. This activation causes platelets to expose or assemble specific membrane receptors on their surfaces—for example, membrane glycoproteins IIb and IIIa (which bind fibrinogen) and GPIb (which binds von Willebrand factor)—with a resultant cascade of further platelet adherence to the foreign surfaces and aggregation. The activation process simultaneously affects platelet granules, which are concentrations of selectively sequestered intraplatelet substances. These substances include (1) serotonin, ATP, adenosine diphosphate, pyrophosphate, and calcium in the “dense bodies”; (2) α ₁ -antitrypsin, β-thromboglobulin, platelet factor 4, and platelet-derived growth factor; and (3) lysosomes. When activated, prostaglandin synthesis (arachidonic acid cascade) and other reactions take place in the surface membrane of the platelet as well as within it and lead to external secretion of the highly reactive components of the platelet granules. The entire process may well contribute to a number of the damaging effects of CPB.

As is usual in the humoral and cellular cascades, the same processes that lead to platelet activation lead virtually simultaneously to processes that inhibit it. This proceeds because platelets, like most cells in humans, contain adenylyl cyclase, which converts ATP to cyclic adenosine monophosphate (cAMP). This conversion is greatly stimulated by products of the arachidonic acid cascade, which is known to be accelerated by CPB. In sufficient amounts, cAMP leads to inhibition of platelet adhesion, aggregation, change of shape, and secretion. This is part of the normal autoregulatory process but may be abnormally amplified during CPB.

Although platelet adherence to the foreign surfaces of the pump-oxygenator is the initial feature of the platelet response to CPB and is surely accompanied by other major and complex responses, there remains uncertainty about subsequent events. There are even doubts as to whether platelet depletion and dysfunction are the primary causes of the bleeding tendency usually present after cardiac surgery. In any event, after the first few minutes of CPB, about 60% of circulating platelets have a normal smooth discoid form, as do about 80% 8 minutes after the start of CPB and at the end of CPB. The implication is that either these platelets have never been activated (because they have just been released into the bloodstream or because foreign surfaces of the pump-oxygenator, passivated by absorption and denaturation of fibrinogen and albumin, no longer activate platelets ^{, ,} ), or they have been reversibly activated and returned to an inactive state. The latter is supported by the work of Zilla and colleagues but is contested by Edmunds. ^, Sufficient irreversible activation occurs that partially degranulated platelets, platelets with damaged membranes, and platelet fragments can be recovered both during and at the end of CPB, along with a large number of normal-appearing platelets. ^, Thus, by the end of CPB, platelet aggregability is reduced by 60% and bleeding time is prolonged, abnormalities that may persist more than 24 hours.

Most events during CPB that profoundly depress platelet function appear to occur initially in platelet membranes. During CPB, there appears to be a loss or inactivation of the functionally important glycoprotein-specific surface receptor sites. ^{, ,} It is possible that abnormal shear stresses are partly responsible. The GPIb receptor (to which plasma von Willebrand factor must bind for platelet adhesion) is markedly reduced shortly after the onset of CPB and remains low throughout bypass ( Fig. 2.12 ). The GPIIb and IIIa receptors (which bind fibrinogen in a process that leads to platelet aggregation in the presence of extracellular calcium) are markedly reduced by the end of CPB. ^, Other changes in the platelet membranes may occur. Zilla has postulated, as have others, that the key to preventing loss of platelet function during CPB, and therefore to preventing the sometimes strong bleeding tendency associated with cardiac surgery, is to avert loss of action of platelet membrane-specific receptor glycoproteins.

GPIb antigen on platelet membrane from patients on CPB, untreated (placebo) (closed circles) , and treated with aprotinin (closed squares) . At 5 minutes of CPB and at end of CPB, differences unlikely to be due to chance were observed. Results are given as mean ± standard error.

(From van Oeveren W, Eijsman L, Roozendaal KJ, Wildevuur CR. Platelet preservation by aprotinin during cardiopulmonary bypass. Lancet . 1988;1:644.)

Whatever the mechanisms and despite the high probability that development of platelet abnormalities is inherent in CPB as it is currently used, these alterations can be favorably influenced. The true membrane oxygenator, made from silicone rubber, appears to cause less platelet (and erythrocyte) damage than occurs with bubble oxygenators.

Complement is a group of circulating glycoproteins that function as part of the body’s response to various kinds of injury, such as traumatic, immunologic, or foreign-body insults. The complement system can be activated upon contact of blood with nonbiological surfaces, perhaps by way of Hageman factor, but other substances (e.g., thrombin, plasmin) can also activate it.

Complement activation during CPB was reported by Hairston, Parker, and Hammerschmidt and their colleagues. Complement consumption during CPB was demonstrated by Chiu and Samson. Chenoweth and colleagues identified C3a, a complement breakdown product, in blood shortly after commencing CPB for cardiac surgery, and found that its continuing production was directly related to body temperature and perfusion flow rate. The result is that more than 50% of patients have serum C3a levels above 1000 ng · mL ⁻¹ at the end of operation with CPB ( Fig. 2.13 ). Complement activation has also been demonstrated to occur during hemodialysis from exposure of blood to the dialysis membrane. Complement activation in this setting is through the alternative pathway, with depletion of C3 but not C1. ^, During CPB, activation is also through the alternative pathway. ^, Further complement activation by the classic pathway occurs after administration of protamine at the end of CPB ; this may add to the whole-body inflammatory response in some patients.

C3a levels at end of CPB, expressed in a cumulative percentile plot. Steep vertical (blue) line on the left represents closed cases, 100% of which had near-normal or normal levels. Curve on the right (red line) represents open cases, virtually all of which had increased levels. Fifty percent of patients had levels greater than 1000 ng · mL ⁻¹ , and 25% had levels greater than 1600 ng · mL ⁻¹ .

(From Kirklin JK, Westaby S, Blackstone EH, Kirklin JW, Chenoweth DE, Pacifico AD. Complement and the damaging effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg . 1983;86:845.)

Recent studies suggest that complement activation after CABG with CPB is biphasic, with a second phase occurring between 8 and 48 hours, postoperatively. ^, The second phase appears to be activated by the classical pathway, not by the alternative pathway. Bruins and colleagues demonstrated that higher peak levels of C4b/c on the second postoperative day correlated with increased occurrence of arrhythmia.

Magnitude of complement activation is affected by several factors that probably interact with still other factors in complex ways. ^, The nature of the foreign surface has some effect; nylon is apparently a particularly potent complement activator. True membrane oxygenators are weaker activators of complement than bubble oxygenators. Duration of CPB has a weak positive effect on the final level of C3a, but administration of protamine has a considerably stronger effect. ^{, ,} Pretreatment of the patient with methylprednisolone or other steroids may decrease the amount of complement activation.

Adverse effects of complement activation relate to depletion of a component (complement) necessary for normal immune response and to adverse effects of the intravascular production of anaphylatoxins (C5a and C3a). Hairston and colleagues showed a decreased ability of postbypass serum to inhibit the growth of certain bacteria and related this in part to complement depletion. Adverse effects of anaphylatoxins probably account for the degree of complement activation as a risk factor for morbidity after clinical CPB ( Table 2.4 , Fig. 2.14 ).

Nomogram of probability of morbidity after CPB, according to level of C3a. Relationships are shown for three different CPB times in patients 1 year of age.

(From Kirklin JK, Westaby S, Blackstone EH, Kirklin JW, Chenoweth DE, Pacifico AD. Complement and the damaging effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg . 1983;86:845.)

Pulmonary sequestration of polymorphonuclear leukocytes and neutropenia have been shown to develop during hemodialysis and to be temporally related to complement activation. Similar observations have been made during CPB (see “ Neutrophil Activation ,” earlier). ^, That these changes are functionally significant is evident from the increased alveolar-arterial oxygen difference that develops during hemodialysis and after CPB and from the pulmonary edema observed after CPB. Activation of complement is directly involved in production of pulmonary edema. These findings suggest that neutrophil-mediated pulmonary endothelial injury (see “ Cellular Response ,” earlier) and increased lung vascular permeability, perhaps also mediated by reactive oxygen metabolites, may contribute to the adverse effects of CPB on pulmonary function. Similar sequestrations may take place in other organs.

That important complement activation is dependent on a large proportion of blood in the boundary layer (e.g., in an oxygenator or hemodialysis coil) is evident from the demonstration in sheep that a simple venovenous shunt produces no adverse effects on white blood cells, platelets, or pulmonary artery pressure. Addition of an oxygenator to the circuit results in a decrease in circulating white blood cells and in platelets (presumably from pulmonary sequestration) and a marked increase in pulmonary artery pressure. Fountain and colleagues showed that infusion of complement-activated plasma produces the same result.

Another humoral amplification system involves kallikrein and bradykinin. Several studies have shown important amounts of bradykinin to be present during CPB. Hypothermia itself apparently results in production of bradykinin. Immaturity, such as is present in young infants, results in less effective elimination of bradykinin. Exclusion of the pulmonary circulation probably also reduces the ability of the organism to cope with circulating bradykinin because the lungs are the main site of bradykinin elimination. Bradykinin, a small peptide, is a powerful vasodilator, and this effect is probably important in the overall response of the organism to CPB.

Coagulation, the formation of fibrin clots, is largely inhibited by heparin during CPB (see “ Heparin Levels ,” earlier under Controlled Variables), but the coagulation cascade is in part activated. Related or unrelated to this, coagulation is often defective for some time after CPB.

Normally, platelets and soluble components of the coagulation cascade are activated in the presence of damaged endothelium or an exposed subendothelium. Through the contact phase, intrinsic phase, and extrinsic phase of activation, prothrombin is converted to thrombin, which acts on fibrinogen to produce fibrin monomers that polymerize spontaneously to form a fibrin clot. Were CPB to be started in a nonheparinized patient, the contact phase and intrinsic phase would be rapidly activated, and within a short time, the pump-oxygenator would be filled with clot.

Because of the incomplete blockade of the coagulation cascade by heparin, small amounts of fibrin form even during routine CPB. Many of the soluble coagulation factors are mildly reduced by the end of CPB. Most authorities believe these changes are insufficient by themselves to be responsible for the bleeding tendency following CPB.

The fibrinolytic cascade, another humoral amplification system, is probably activated to some degree in all operations in which CPB is used. Important hyperfibrinolysis was shown to be present in 159 (20%) of 774 patients undergoing CABG.

Naturally occurring inactive plasminogen (normally incorporated within thrombi) is transformed into the active fibrinolytic agent plasmin under certain circumstances, and measurable blood plasmin levels have been demonstrated in patients shortly after initiation of CPB. Because conversion of plasminogen to plasmin is facilitated by kallikrein, which also results from activation of Hageman factor, the fibrinolytic cascade may be initiated during CPB by activation of factor XII. However, extrinsic plasminogen activator expressed by endothelial cells is the major stimulant for conversion of plasminogen to plasmin and thus for the fibrinolytic cascade. A reasonable explanation for this behavior of endothelial cells during CPB is the abnormally high levels of such substances as catecholamines, bradykinins, and other molecules that are generated.

Because plasmin also serves as an activator of complement, prekallikrein, and possibly Hageman factor, the intravascular activation of plasminogen into plasmin (which in intact humans is usually a localized phenomenon) may continue to stimulate cascades of all the humoral amplification systems. Fibrinogen breakdown products (produced to some extent by the coagulation cascade during CPB), when acted upon by plasmin, have been shown experimentally to lead to important pulmonary dysfunction. This is another example of the powerful effects of the intravascular occurrence of events that are usually localized and extravascular in intact humans.

For a time, it was conventional wisdom that excessive bleeding after CPB was primarily the result of platelet depletion and dysfunction. More recently, several lines of information strongly suggest that activation of the fibrinolytic cascade also contributes importantly to postoperative bleeding after cardiac surgery in which CPB is used. ^, One of these is the favorable effect of antifibrinolytic agents on bleeding.

The completely cellular arachidonic acid cascade is activated by a disturbance of cell membranes, which in turn activates phospholipase A ₂ . This releases arachidonic acid from the phospholipid fraction of cells, but the arachidonic acid can also come from intracellular lipid pools. The cascade proceeds through the prostaglandin-endoperoxide (cyclooxygenase) pathway. Stationary, migrating, and intravascular cells are susceptible to the arachidonic acid cascade and liberate the active products, which exhibit a short half-life.

The lung, bypassed during CPB, is a major site of synthesis, release, and degradation of eicosanoids (products of the arachidonic acid cascade), although not necessarily the cellular source of those compounds. Prostacyclin and prostaglandin E ₂ (PGE ₂ ) production appears to be sharply increased shortly after CPB is begun, but later, when CPB becomes partial and some blood again passes through the lung, levels decrease. By contrast, thromboxane B ₂ production (thromboxane B ₂ is a stable metabolite of thromboxane A ₂ ) becomes apparent and reaches peak levels when total CPB becomes partial as the lungs again are perfused. Many researchers believe that most of the thromboxane A ₂ comes from platelets, even though release occurs to a great extent in the lungs. PAF, another product of the cascade, appears to be an important mediator of inflammation.

Leukotriene B ₄ , a product of the arachidonic cascade that promotes plasma leakage and leukocyte adhesion, is also increased during and for a time after CPB.

Details and overall effects of activation of the arachidonic acid cascade during CPB are not completely understood, but there is at least evidence that the magnitude of the release of both of these eicosanoids during CPB is greatest in the very young. The activation releases agents that are somewhat counteracting, including the vasoconstricting agent thromboxane A ₂ , a PAF, and the vasodilating and platelet-inhibiting factors prostacyclin and PGE ₂ .

Cytokines are soluble factors elaborated by cells of the immune system (e.g., T-cell lymphocytes) that normally act on other cells to regulate their function. During CPB, cytokines such as the interleukins (ILs) elaborate other mediators of the inflammatory process (e.g., TNF, leukocyte adhesion molecules, PAF), which in turn continue the process.

IL-1 is an intracellular derivative of stimulated mononuclear phagocytes and a mediator of fever, changes in endothelial cell function and permeability, and decreased vascular resistance. Its concentration in monocytes is increased during CPB and again 24 hours later. A positive correlation has been found between intracellular IL-1 activity and the patient’s temperature 24 hours after CPB. The interrelation between complement activation (which activates monocytes), prostaglandins (which also mediate IL-1 production), and IL-1 illustrates the complexity of the whole-body inflammatory response to CPB and the problems inherent in efforts to prevent its damaging effects.

IL-6 and IL-8 levels rapidly increase after initiation of CPB. Degree of cytokine response appears to correlate with duration of CPB and aortic clamping.

TNF is released by activated monocytes (and macrophages) and is increased in many patients in the later stages of CPB and subsequent hours. It is known to increase endothelial cell permeability and open interendothelial cell spaces, thereby promoting development of interstitial edema. ^,

Endotoxin, a powerful stimulant of complement and endothelial activation, is also a potent agonist of release of TNF from macrophages and is elevated in some patients after CPB. ^{, ,} Endotoxin release may be the result of translocation of bacteria from the gut as the result of splanchnic ischemia and possibly impaired function of Kupffer cells in the liver.

Theoretically, total body V ˙ o 2 during CPB at normothermia (37°C) should be that of an intact human under anesthesia if all parts of the microcirculation are perfused. Yet, in two studies in humans, V ˙ o 2 during normothermic CPB at flows of 1.8 to 2.4 L · min ⁻¹ · m ⁻² was highly variable. Values of 74 to 162 mL · min ⁻¹ · m ⁻² were found in one study ; in the other study, which consisted of 12 patients, mean ± standard deviation of V ˙ o 2 was 131 ± 20 L · min ⁻¹ · m ⁻² .

A combined analysis of experimental animal studies during normothermic CPB indicates a best-fit hyperbolic relationship between the perfusion Q ˙ and V ˙ o 2 ( Fig. 2.15 ). Q ˙ in these studies was expressed in L · min ⁻¹ · m ⁻² . (These units were not used in the excellent study of Andersen and Senning, nor could the data be recalculated in these terms, hence their exclusion. ) The following linear regression equation (see Fig. 2.15 ) was derived from the data:

Relationship of total body V ˙ o 2 to perfusion flow rate ( Q ˙ ) at normothermia during nonpulsatile CPB. Figure contains two depictions. One is a scattergram of data from animal experiments ( n = 213) performed at about 37°C by Cheng and colleagues ( n = 33), Paneth and colleagues ( n = 60), and Starr ( n = 120). Note that scatter of data increases as flow increases. Second depiction is solid and dashed lines (presented as in Fig. 2.1 ), which is a solution of the hyperbolic equation (Appendix Equation 2A-2) derived from these data. The hyperbolic equation is chosen because the correlation coefficient, r , was.69, whereas it was.39,.54, and.52 for the linear equation, log-log equation, and Arrhenius equation, respectively.

(From Starr A. Oxygen consumption during cardiopulmonary bypass. J Thorac Cardiovasc Surg . 1959;38:46; Harris EA, Seelye ER, Squire AW. Oxygen consumption during cardiopulmonary bypass with moderate hypothermia in man. Br J Anaesth . 1971;43:1113; and Paneth M, Sellers R, Gott VL, et al. Physiologic studies upon prolonged cardiopulmonary bypass with the pump-oxygenator with particular reference to (1) acid-base balance, (2) siphon canal drainage. J Thorac Surg . 1947;34:570).

where V ˙ o 2 is oxygen consumption expressed as a percentage of measured control value before bypass, and Q ˙ is flow from the pump-oxygenator, expressed as mL · kg ⁻¹ · min ⁻¹ (correlation coefficient = 0.83). Andersen and Senning noted, however, that Q ˙ and V ˙ o 2 must meet at zero and that the control value for V ˙ o 2 was usually reached at high flows (100-125 mL · kg ⁻¹ · min ⁻¹ ). Visual observation of their scattergram suggests that the hyperbolic model derived from the combined analysis fits their data and ideas well.

Temperature of the patient is also related to V ˙ o 2 during CPB, as it is in intact, anesthetized, nonshivering subjects. Harris and colleagues were first to express mathematically the interrelation among Q, temperature, and V ˙ o 2 , but their data covered a narrow range of temperature. Complete data of the type desired are not available. Using the experimental data at 37°C just described and the relation of V ˙ o 2 to flow at 20°C measured during CPB in humans, a multivariable equation and nomogram have been developed (see Fig. 2.11 ) portraying the relations of V ˙ o 2 to flow at various temperatures.

Pulsatility of the arterial input does not affect V ˙ o 2 , at least during flow rates greater than about 1.4 L · min ⁻¹ · m ⁻² , nor does the strategy (pH-stat vs. alpha-stat) of managing the acid-base balance. Other factors that may affect V ˙ o 2 include amount of developed pericapillary edema and amount and type of catecholamine release.

Cerebral oxygen consumption is important during CPB, particularly hypothermic CPB at low flow, because cerebral V ˙ o 2 that is reduced below the usual (normal) value at a given temperature implies incomplete or uneven cerebral perfusion (decreased effective capillary density). An equation has been developed from the data of Croughwell and colleagues that appears to be the best expression available of the normal relation during CPB at full flow (2.0 L min ⁻¹ · m ⁻² ) between cerebral V ˙ o 2 and temperature during CPB. It is presumably applicable to all ages ( Fig. 2.16 ). Assuming cerebral oxygen consumption at 37°C before CPB is the same as that at 37°C during CPB is inappropriate because whole-body V ˙ o 2 on CPB is slightly less than off CPB. This may be from presumed uneven perfusion on CPB.

Relation between body temperature (nasopharyngeal) and cerebral oxygen consumption (CMRO ₂ ) during full flow (2 L · min ⁻¹ · m ² ) CPB, which is presumed to be applicable to all ages. The equation (van’t Hoff) is:

CMRO 2 = 0.021e 0.1147 ⋅ T r = .85, P <.001

where e is the base of the natural logarithms and T is temperature in °C.

(Based on data from Croughwell N, Smith LR, Quill T, et al. The effect of temperature on cerebral metabolism and blood flow in adults during cardiopulmonary bypass. J Thorac Cardiovasc Surg . 1992; 103:549.)

Cerebral oxygen consumption does not appear to change with variations in cerebral blood flow that occur during clinical CPB. ^, This is consistent with findings in experimental studies. For example, Fox and colleagues found that in monkeys supported with CPB at 20°C, cerebral oxygen consumption was the same at CPB flow rates of 0.5 and 1.6 L · min ⁻¹ · m ⁻² , even though cerebral blood flow at 0.5 was 50% of that at 1.6. Alpha-stat pH management versus pH-stat management may influence cerebral blood flow, because the addition of CO ₂ during hypothermia for pH-stat management results in cerebral vasodilatation and attenuation of autoregulation.

Under conditions that often pertain in adults undergoing cardiac operations (nonpulsatile perfusion; flow 1.6 L · min ⁻¹ · m ⁻² ; temperature ± 25°C), cerebral blood flow (measured by radioactive xenon clearance) is about 25 mL · min ⁻¹ · 100 g brain tissue ⁻¹ , with some variability depending on Pa co ₂ . During CPB in monkeys, a similar value has been found (using microspheres), representing about 6% of total systemic blood flow.

In humans, cerebral blood flow during CPB is no less in elderly patients than in other adult patients, and it appears to be proportionally similar in neonates and infants to that in adults. ^{, ,} Thus, age appears to have little effect on the proportion of total flow represented by cerebral blood flow under usual circumstances of CPB.

During normothermic and moderately hypothermic CPB in adults and elderly patients, cerebral blood flow is not importantly altered with variations of mean arterial blood pressure . ^, This is similar to normal awake adult humans, in whom cerebral blood flow does not vary significantly with variations of arterial blood pressure (mean) from about 60 to 150 mmHg. When arterial blood pressure during CPB falls below about 40 mmHg, cerebral blood flow may decline appreciably, with a concomitant decrease in cerebral oxygen consumption. ^, A reasonable inference is that in adults on CPB, arterial blood pressure need not be manipulated pharmacologically unless it is less than 40 to 50 mmHg.

By contrast, during hypothermic CPB—at least in neonates, infants, and children—cerebral blood flow is dependent on arterial blood pressure. ^{, ,} Corresponding variations in cerebral oxygen consumption have not been established with certainty. In view of this, arterial blood pressure should probably be kept above about 25 mmHg in this setting in young patients.

The effect of decreased CPB flow rate ( Q ˙ ) on cerebral blood flow in humans is incompletely understood, in part because of the interaction between perfusion Q ˙ and arterial blood pressure (see Chapter 4 ). However, during moderately hypothermic CPB at the usual flow rates, there appears to be a direct correlation between CPB and cerebral blood flow rates, despite the poor or absent correlation between arterial blood pressure and cerebral blood flow.

Cerebral blood flow during CPB is affected by Pa co ₂ . Hypercarbia increases cerebral blood flow, whereas hypocarbia decreases it. ^{, ,} In children undergoing hypothermic CPB, the flow increases 1.2 mL · min ⁻¹ · 100 g brain tissue ⁻¹ for every 1-mmHg increase in Pa co ₂ (measured at 37°C) between 33 and 50 mmHg. Infants have a slightly blunted response. Data gathered by Kern and colleagues are compatible with the hypothesis that under usual conditions of hypothermic CPB, metabolic needs of the brain are met with a Pa co ₂ value of 33 mmHg.

Although autoregulation of cerebral blood flow has been described during normothermic and moderately hypothermic CPB, it may well be that it is the remainder of the body, not the brain, that accomplishes autoregulation. This was suggested by the experimental studies of Fox and colleagues.

Cerebral blood flow during CPB may, at times, be excessive in relation to cerebral oxygen consumption. For example, Croughwell and colleagues found that during CPB and reduction of the patient’s body temperature from 37°C to 28°C, cerebral blood flow decreased less than cerebral oxygen consumption. This was referred to as a situation of luxuriant cerebral blood flow accompanied by a narrowing of the cerebral arteriovenous oxygen difference. Similar luxury perfusion can result from hypercarbia, and it has been argued that this increases the risk of cerebral damage by microemboli. ^, This is an argument against use of pH-stat strategy for control of Pa co ₂ during CPB.

Clinical information strongly suggests that blood flow to the skin is severely reduced during nonpulsatile CPB in humans. The small bald spot that develops on the back of the head after CPB in some patients is probably the result of the pressure produced by weight of the head on an area of poorly perfused skin in contact with even a well-padded pillow during operation. Ease with which burns are produced by the cautery pad may also result from poor blood flow to the skin during CPB.

Of interest, a study of the sublingual microcirculation in humans has shown that the proportion of perfused small blood vessels decreases importantly and similarly with induction of anesthesia in patients undergoing thyroidectomy, and cardiac surgery with or without CPB. In the CPB group, the proportion of perfused small vessels decreased after induction, improved slightly thereafter, failed to return to baseline, and persisted after 24 hours ( Fig. 2.17 ). The off-pump cardiac surgery patients had less severe but statistically significant microcirculatory alterations immediately postoperatively. These alterations improved slightly but also persisted after 24 hours. Thus, microvascular perfusion was altered similarly in CPB and off-pump patients 6 to 24 hours after admission to the intensive care unit. In thyroidectomy patients, early changes reversed rapidly in the postoperative period. Severity of microvascular alterations correlated with peak lactate levels after cardiac surgery ( P <.05). These findings suggest that changes in the microcirculation cannot be attributed entirely to CPB.

Evolution of the proportion of perfused small vessels in patients undergoing cardiac surgery with (red boxes, n=9 ) and without (blue boxes, n=6 ) CPB, and in patients undergoing thyroid surgery (green boxes, n=7 ) .

(From De Backer D, Dubois MJ, Schmartz D, et al. Microcirculatory alterations in cardiac surgery: effects of cardiopulmonary bypass and anesthesia. Ann Thorac Surg . 2009;88:1396-1403.)