A Practical Approach to Cardiac Anesthesia (Practical Approach Series) 5th Ed.

24 Protection of the Brain during Cardiac Surgery

John M. Murkin


 1. The incidence of overt stroke is 2% to 6% for closed-chamber cardiac procedures. For open-chamber procedures, the incidence of stroke is increased to 4.2% to 13%.

 2. Risk factors for early stroke have been found to be advanced age, duration of cardiopulmonary bypass (CPB), high postoperative creatinine, and extensive aortic atherosclerosis, while delayed stroke was associated with female gender, postoperative atrial fibrillation, cerebrovascular disease and requirement for inotropic support.

 3. Within the first postoperative week, up to 83% of all patients undergoing bypass surgery using CPB demonstrate a degree of cognitive dysfunction. Efforts to mitigate against early postoperative cognitive dysfunction are warranted since early postoperative cognitive issues are in part reflective of subclinical brain injury.

 4. Watershed lesions are commonly due to profound hypotensive episodes but may also be the result of cerebral embolism. Embolization and hypotension acting together may magnify CNS injury.

 5. The etiology of CNS damage associated with embolization is from multiple sources that are patient related, procedure related, and equipment related.

 6. Leukocytosis is associated with a higher risk for ischemic stroke. The results strongly implicate inflammation and white cell activation as etiologic factors in both the extent and severity of perioperative cerebral events.

 7. α-Stat blood gas management, which maintains a normal transmembrane pH gradient and maintains cerebral autoregulation of blood flow, should be used in adult patients undergoing bypass. This modality may help prevent cerebral embolization. See Table 24.7 for additional evidence-based practice guidelines.

 8. A wide variety of autoregulation thresholds occurs even with α-stat blood gas management and is likely a consequence of increased age and cerebrovascular disease. There is an emerging consensus during normothermic and tepid bypass to maintain MAP greater than 70 mm Hg. See Table 24.8 for other management strategies to decrease CNS brain injury.

I. Central nervous system (CNS) dysfunction associated with cardiac surgery

   A. Overview. Despite the continuing improvements in surgical and cardiopulmonary bypass (CPB) techniques during cardiac surgery, stroke remains a devastating postoperative complication for patients and their families. In a recent study in which 1,800 patients with three-vessel or left mainstem coronary artery disease were randomized to percutaneous intervention (PCI) or conventional coronary artery bypass (CAB) surgery, there was no difference in mortality at 1 yr but a significantly lower incidence of primary composite endpoint of major adverse cardiac or cerebrovascular event in CAB (12.4%) versus PCI (17.8%) patients [1]. However, while the overall outcome should argue strongly in favor of CAB surgery, the stroke rate was significantly higher in CAB (2.2%) than PCI (0.6%) patients. Given the relatively equivalent risk factors between PCI and CAB patients in this study, the mechanism of perioperative stroke must be better understood if we are to further reduce the risk of CNS morbidity related to cardiac surgery. This chapter will review the current incidence and risk factors for brain damage in cardiac surgery and will outline strategies aimed at protection of the brain during cardiac surgical procedures.


   B. Stroke incidence. In most series reported to date, the incidence of clinically apparent neurologic injury or overt stroke is 2% to 6% for closed-chamber cardiac procedures (e.g., CAB surgery). Up to 25% to 65% of strokes after CAB surgeries are bilateral or multiple, suggestive of an embolic etiology [2]. For open-chamber procedures (e.g., valve surgery), the reported incidence varies from 4.2% to 13%, which could be related to increased risk of embolization, increased hemodynamic instability, or prolonged CPB time. However, more subtle neurologic changes, such as development of primitive reflexes (e.g., snouting, palmomental reflex), visual field defects, and subtle motor-sensory abnormalities, can be demonstrated in the early postoperative period in up to 61% of all patients undergoing CPB. By 2 mos postoperatively, the prevalence of such new subtle neurologic dysfunction decreases to about 20% and persists for at least 1 yr [3].


   C. Early versus delayed stroke. In considering the incidence of perioperative stroke, it is apparent that distinguishing stroke as early (i.e., neurologic deficit apparent on emergence from anesthesia) or delayed (i.e., neurologic deficit developing after awakening from anesthesia) is important to better discriminate etiology and assess potential risk-reduction strategies, as it is apparent that only approximately 50% of perioperative strokes present during the first 24 hrs postsurgery [3], and tend to be more severe with a higher permanent deficit and a higher impact on mortality [35]. The higher mortality could be related to sicker patients, or that stroke is just one manifestation of other concomitant embolic/hypoperfusion-mediated complications. Stroke can also have a long-term effect on mortality (Fig. 24.1). In a prospective study on CAB patients, the adjusted survival rates at 1, 5, and 10 yrs were 94.1%, 83.3%, and 61.9% among patients free from stroke versus 83%, 58.7%, and 26.9% for those suffering perioperative stroke [6]. Risk factors for early stroke have been found to be advanced age, duration of CPB, high preoperative creatinine, and extensive aortic atherosclerosis, while delayed stroke was associated with female gender, postoperative atrial fibrillation, cerebrovascular disease, and requirements for inotropic support [5]. Delayed but not early stroke was associated with long-term mortality.

Figure 24.1 Effect of stroke on 10-yr survival after CAB graft. The crude annual incidence of death was 18.1/100 person-years among patients with strokes and 3.7/100 person-years among patients without strokes (p <0.001). (From Dacey L J, Likosky DS, Leavitt BJ, et al. Perioperative stroke and long-term survival after coronary bypass graft surgery. Ann Thorac Surg. 2005;79:532–536, with permission.)

   D. Cognitive dysfunction. It has been demonstrated that within the first postoperative week, up to 83% of all patients undergoing CAB surgery using CPB demonstrate a degree of cognitive dysfunction. Of these patients, 38% have symptoms of intellectual impairment and 10% are considered to be overtly disabled. Concentration, retention, and processing of new information and visuospatial organization are the most frequently affected domains. At 5-yr follow-up, more than 35% of CAB patients exhibit some degree of neuropsychologic dysfunction [7]. Variable definitions, different measurement techniques, and different intervals of postoperative cognitive testing confound this issue, however, giving rise to reported incidences of perioperative cognitive decline that vary from 4% to 90%. Additional confounders include the variability in performance to repeated neuropsychometric testing even in healthy subjects, and an innate decline in cognitive function associated with both aging and the various comorbidities found in cardiac surgical patients. The challenge lies in discerning whether a specific event, for example, cardiac surgery, is causal or coincidental to deterioration in cognitive function. On the basis of more recent studies, the consensus is that longer-term cognitive dysfunction has a similar incidence whether patients undergo cardiac surgery with or without use of CPB, or instead undergo PCI or are managed medically, with the implication that aging and progression of underlying atherosclerosis and related comorbidities are of most significance [8].

   E. Comparison groups. The incidence of new postoperative CNS dysfunction in CAB patients has been compared with that of patients undergoing major abdominal vascular or thoracic surgical procedures. Most of these patients usually have concomitant disease including hypertension, diabetes mellitus, diffuse atherosclerosis, and chronic lung disease. After adjusting for identified risk factors, patients undergoing any surgical procedure have been found more likely to suffer a cerebrovascular accident (CVA) than nonoperated controls with an odds ratio of 3.9. Even after excluding high-risk surgery (cardiac, vascular, and neurologic), the odds ratio is 2.9, which suggests the perioperative period itself predisposes patients to stroke. This observation is of particular relevance in considering the role of inflammatory processes and the salutary effect of statins as discussed below.

   Studies on patients undergoing CAB surgery have demonstrated minimal difference in long-term cognitive function between patients undergoing on-pump versus off-pump procedures [9]. However, in general, it does appear that compared with other noncardiac surgical groups, the incidence of early postoperative cognitive dysfunction is higher in CAB patients, and since new ischemic lesions on magnetic resonance imaging (MRI) studies in valve surgery patients have been correlated with early postoperative cognitive dysfunction [10], as has intraoperative cerebral oxygen desaturation and early postoperative cognitive dysfunction in CAB patients, it does appear as though efforts to mitigate against early postoperative cognitive dysfunction are warranted since early postoperative cognitive dysfunction is in part reflective of subclinical brain injury.


   F. Risk factors. Table 24.1 shows risk factors for both stroke and cognitive dysfunction. Risk factors have been pooled into various risk prediction models, which, while useful to compare patient groups, are still not predictive of a particular individual’s outcome, although the presence of key risk factors may help in deciding the best procedure for a particular high-risk patient (i.e., surgery vs. angioplasty or valvuloplasty). Specific risk factors (i.e., aortic atherosclerosis, recent stroke) should prompt further preoperative investigations (e.g., carotid scanning, modification of intraoperative management) and may even suggest a change in surgical approach (i.e., off-pump CAB [OPCAB] with no instrumentation of the aorta) to minimize the potential for neurologic complications.

Table 24.1 Risk factors for neurologic complications in cardiac surgery

II. Cerebral physiology

   A. Cerebral autoregulation. In normal subjects, cerebral blood flow (CBF) remains constant at 50 mL/(100 g·min) over a wide range of mean arterial pressure (MAP) from 50 to 150 mm Hg. This autoregulatory plateau reflects the tight matching between cerebral metabolic rate for O2 (CMRO2) and CBF, mediated in part by endothelium-derived relaxing factor (EDRF-nitric oxide [NO]). With decreased metabolic activity resulting from certain anesthetics or hypothermia, lowered CMRO2 produces a resultant reduction in CBF and establishment of a lower autoregulatory plateau. It is apparent and should be considered that rather than a single cerebral autoregulatory curve, there are instead a series of autoregulatory curves. Each autoregulatory curve represents a differing set of metabolic conditions of the brain (e.g., normal metabolic activity) at 37°C versus lowered metabolic activity at 28°C. The autoregulatory plateau is a manifestation of intact cerebral flow and metabolism coupling, and it varies with metabolic rate.

   With intact autoregulation, adequate substrate (blood flow) can be delivered at a lower perfusion pressure during conditions of lowered metabolic rate (e.g., anesthesia, hypothermia) in the absence of cerebral vasodilators (Fig. 24.2). Cerebral autoregulation is lacking in patients with diabetes mellitus and appears to be lost during deep hypothermia (e.g., less than 20°C) and for several hours after deep hypothermic circulatory arrest (DHCA). This results in pressure-passive CBF; in these instances, hypotension may entail increased risk for cerebral hypoperfusion. Similarly, in patients with chronic hypertension, cerebral autoregulation has been reset and higher perfusion pressures might be needed during CPB. In these circumstances, an uncoupling of CBF and cerebral metabolism could be related to neurocognitive decline [11].

   It is notable that most of the initial studies on CBF and CMRo2 specifically excluded patients with overt cerebrovascular disease while more recent studies which included elderly patients and those with previous CVA have indicated a striking variability in the autoregulatory threshold [12]. Using transcranial Doppler (TCD) or cerebral near-infrared spectroscopy (cNIRS), lower limits of cerebral autoregulation ranging from 45 to 80 mm Hg have been demonstrated, and in a study of 127 adult patients during CPB, a correlation has been made between loss of cerebral autoregulation during rewarming and postoperative neurologic events [13,14].

Figure 24.2 Cerebral autoregulatory curves during normothermia and hypothermia. The upper curve demonstrates a higher CBF autoregulatory plateau that is appropriate for the higher CMRO2 in the awake state, versus a lower CBF plateau during hypothermia. With maximal cerebral vasodilation, lower CPP results in lower CBF that is appropriate at a lower CMRO2 (hypothermia), but not at higher CMRO2. (From Murkin JM. The pathophysiology of cardiopulmonary bypass. Can J Anesth. 1989;36:S41–S44, with permission.)

   B. pH management. There is an inverse relationship between solubility of respiratory gases and blood temperature. With cooling of blood, CO2 partial pressure (PaCO2) decreases and arterial pH (pHa) increases, producing an apparent respiratory alkalosis in vivo. To compensate for this condition during hypothermic CPB, total CO2 must be increased by addition of exogenous CO2 to the oxygenator, known as pH-stat pH management.

      1. a-Stat maintains pHa 7.4 and PaCO2 40 mm Hg at 37°C without addition of exogenous CO2. Intracellular pH is primarily determined by the neutral pH (pHN) of water. Since pHN becomes progressively more alkaline with decreasing temperature, intracellular pH becomes correspondingly more alkaline during hypothermia. Since this intracellular alkalosis occurs in parallel with the hypothermia-induced increased solubility of CO2 and increased blood pH, the normal transmembrane pH gradient of approximately 0.6 units remains unchanged, thus preserving optimal function of various intracellular enzyme systems. This preservation of normal transmembrane pH gradient is the crux of a- stat pH theory, and, in fact, we function in vivo according to α-stat principles. Since different tissues have differing temperatures (e.g., exercising muscle at 41°C vs. skin at 25°C), they also will have correspondingly different pHa values (e.g., 7.34 vs. 7.6, respectively), although the net pHa at 37°C will be 7.4. a-Stat management acknowledges the temperature dependence of normal pHa and strives to maintain a constant transmembrane pH gradient by maintaining PaCO2 at 40 mm Hg and pHa at 7.4 as measured in vitro at 37°C. For this strategy during CPB, total CO2 is kept constant by not adding exogenous CO2 and thus not compensating for increased solubility of CO2. Blood samples measured at 37°C will show pHa7.4 and PaCO2 40 mm Hg, but those same samples measured at 28°C would have pHa 7.56 and PaCO2 26 mm Hg (Fig. 24.3).

Figure 24.3 Contrasting arterial blood gas values as seen in vitro at 37°C or in vivo at 28°C when using α-stat or pH-stat management. Using pH-stat, laboratory values in vitro would be pHa 7.26 and PaCO2 56 mm Hg, whereas temperature-corrected values in vivo would be pHa 7.4 and PaCO2 40 mm Hg. If α-stat were used, laboratory values in vitro would be pHa 7.4 and PaCO2 40 mm Hg, whereas temperature-corrected values in vivo would be pHa 7.56 and PaCO2 26 mm Hg.

      2. pH-stat management involves addition of exogenous CO2 to maintain PaCO2 40 mm Hg and pHa 7.4 when corrected for the patient’s body temperature in vivo. Until the mid-1980s, pH-stat management was generally the most common mode of pH management during moderate hypothermic CPB. Since it is a potent cerebral vasodilator, such increases in total PaCO2 associated with pH-stat have been shown to produce cerebral vasodilation, impairing cerebral flow and metabolism coupling and producing loss of cerebral autoregulation (Fig. 24.4). There is evidence that pH-stat management can increase the incidence of postoperative cognitive dysfunction when CPB duration exceeds 90 min [15]. This likely reflects both increased delivery of microemboli into the brain resulting from CO2-induced vasodilation, and impairment of regional cerebral autoregulation. It is notable that in recent studies even with use of a-stat, a wide variability of autoregulatory threshold has been found likely as a consequence of the increased age and presence of overt cerebrovascular disease in current surgical populations [12,14].

Figure 24.4 Linear regression analysis of CBF and CMRO2, or CPP, for patients managed using α-stat (non–temperature-corrected) or pH-stat (temperature-corrected) management during moderate hypothermia (28°C). With pH-stat (A), there is no correlation between CBF and CMRO2, demonstrating loss of cerebral flow and metabolism coupling, whereas with α-stat (C) there is a highly significant (p <0.005) correlation. CBF significantly (p <0.002) correlates with CPP using pH-stat (B), reflecting pressure-passive CBF and loss of autoregulation, whereas with α-stat (D), CBF is independent of CPP. (From Murkin JM, Farrar JK, Tweed WA, et al. Cerebral autoregulation and flow/metabolism coupling during cardiopulmonary bypass: The influence of PaCO2Anesth Analg. 1987;66:825–832, with permission.)

III. Etiology of CNS damage

   A. Embolization. In the context of CPB, focal ischemia is most often a consequence of isolated cerebral arteriolar obstruction by a particulate or gaseous embolus. Emboli vary in size, nature (particulate vs. gaseous), and origin (patient vs. equipment). Embolus factors influencing potential for damage include size, solubility, viscosity, and buoyancy relative to blood. Vessel diameter, anatomical location, and inflammatory responsivity influence tissue vulnerability. Open-chamber procedures generally entail greater risk of embolization than closed-chamber procedures. Calcific or atheromatous macroembolic debris from the ascending aorta or aortic arch appears to be a prime factor in the production of clinical stroke syndromes. It was formerly thought that microembolic elements, either gaseous or particulate, produced cognitive dysfunction. Studies from beating-heart surgery in which CPB is avoided, despite a much lower incidence of embolic events, appear to have a relatively similar incidence of long-term cognitive dysfunction when compared to CAB using conventional CPB. Microgaseous emboli, the numbers of which are greatly reduced by avoidance of CPB, paradoxically appear to be relatively less injurious than otherwise predicted. The use of heparin anticoagulation in cardiac surgical patients has been demonstrated to ameliorate some of the overt effects of cerebral gas emboli [16].

   Areas of brain localized at the boundary limits of major cerebral arteries (e.g., anterior and middle, or middle and posterior cerebral arteries, or superior and posterioinferior cerebellar arteries) are known as arterial boundary zones or watershed zones (Fig. 24.5), and these can manifest as isolated lesions as a consequence of transient global ischemia (see following sections). Although they are commonly due to profoundly hypotensive episodes, watershed lesions are not pathognomonic of a hypotensive episode and may be the result of cerebral emboli. Embolization and hypoperfusion acting together can play a synergistic role and either cause or magnify CNS injury in cardiac surgical patients.


Figure 24.5 Hatched areas showing the most frequent locations of boundary area, or watershed zone infarcts in the brain, situated between the territories of major cerebral or cerebellar arteries. (From Torvik A. The pathogenesis of watershed infarcts in the brain. Stroke.1984;2:221–223, with permission.)

      1. Detection of emboli

        a. Brain histology. Isolated areas of perivascular and focal subarachnoid hemorrhage, neuronal swelling, and axonal degeneration are seen with higher frequency in the brains of patients dying after cardiac surgery than after non-CPB major vascular surgery. After surgery using unfiltered CPB circuits, fibrin and platelet emboli and calcific and atheromatous debris were seen frequently in small arterioles and capillary beds. Small cerebral capillary and arterial dilatations (SCADS) have been demonstrated histologically, occurring in nonsurvivors after proximal aortic instrumentation after either CPB or coronary angiography. These SCADS are increasingly believed to be due in part to lipid microemboli from usage of unprocessed cardiotomy suction blood.

        b. Intraoperative emboli detection. Intraoperative fluorescein retinal angiography has demonstrated that extensive retinal microvascular embolization occurs during CPB. The incidence and extent of retinal obstruction are much greater with bubble than with membrane oxygenators, despite use of 40 μm arterial line filters. Use of TCD insonation enables assessment of blood flow perfusion characteristics through the middle cerebral artery (MCA). TCD insonation permits measurement of blood flow velocity and detection and quantification of emboli, though discrimination of gaseous from particulate emboli remains unreliable. Proximal aortic instrumentation and initiation of CPB have been identified as particularly embologenic events. After open-chamber surgery, cerebral emboli are detected as the heart fills and begins to eject, underscoring the importance of meticulous deairing techniques (see the following sections).


      2. Sources of emboli

        a. Patient-related sources

           (1) Aortic atheroma. Atheromatous debris can be embolized during aortic clamping or cannulation. Intraoperative aortic ultrasonography using either transesophageal echocardiography (TEE; high sensitivity, low specificity) or epiaortic scanning (EAS) (high sensitivity, high specificity) enables visualization of aortic wall and can be used to guide cannulation sites. Ultrasonography has demonstrated that plaque may fracture or shear off and embolize during CPB as a consequence of trauma from aortic clamping and cannulation or from blood “jetting” from the aortic cannula or may result in intimal flap formation with potential for delayed postoperative embolization [17]. Using EAS, Ura et al. compared images before and after CPB in 472 patients undergoing cardiac surgery, and noted new lesions in the ascending aortic intima in 16 patients (3.4%) following decannulation [17]. In 10 patients, 3 of whom suffered postoperative CVA, the new lesions were severe with mobile lesions or disruption of the intima, of which 6 were related to aortic clamping and the other 4 to aortic cannulation. Only the maximal thickness of the atheroma near the aorta manipulation site was a predictor of new lesions. If the atheroma was 3 to 4 mm, the incidence of new lesions was 11.8% and was as high as 33.3% if the atheroma was >4 mm. As such, embolization of plaque or thrombus from such intimal fractures may explain one mechanism of delayed stroke cited above. Proximal aortic atherosclerosis is thus a significant risk factor for neurologic injury.

           (2) Intraventricular thrombi. During closed-chamber procedures in patients with recent mural thrombi, manipulation of the heart can dislodge thrombi that embolize once the heart begins to fill and eject.

           (3) Valvular calcifications. Valve surgery, particularly valve replacement surgery, is associated with increased risk of CVA resulting from embolization of intracavitary valve debris.

           (4) Postoperative atrial fibrillation. Early-onset atrial fibrillation is associated with a variety of adverse outcomes and has been strongly linked to increased perioperative stroke risk, and is particularly associated with increased risk of delayed-onset postoperative stroke [5]. Even transient new-onset atrial fibrillation is associated with increased risk of 30-day and 1-yr cardiovascular events composed of stroke, cardiac death, and myocardial infarction (MI). In cardiac surgical patients a decreased incidence of atrial fibrillation has been associated with perioperative statin therapy as discussed below. Increased efforts should thus be aimed in part at reducing even transient new-onset postoperative atrial fibrillation.

        b. Procedure-related sources

           (1) While open-chamber procedures (e.g., septal repair, ventricular aneurysmectomy, valve surgery) expose the arterial circulation to air or particulate debris, closed-chamber procedures also can be associated with ventricular air. Use of a ventricular vent, particularly if active suction is applied and the heart is empty, produces localized subatmospheric pressure at the vent tip within the left ventricle (LV) and cause air to be entrained retrograde from the vent insertion site (usually through the superior pulmonary vein) into the LV. Use of TEE can assist in visualization and guide the removal of residual intracavitary air (see following discussion). Inadvertent opening of the left atrium (LA) or LV while the heart is beating also caused rapid air entrainment and increased potential for cerebral emboli.

           (2) Aortic cannulation and clamping are associated with cerebral embolization, particularly in the presence of extensive aortic atherosclerosis.

           (3) Duration of CPB is an independent risk for postoperative brain dysfunction. After 90 min of CPB, the incidence of cognitive dysfunction is increased compared to CPB of shorter duration. It is important to note that duration of CPB may be increased by factors (e.g., extensive atherosclerotic disease), which may independently contribute to neurologic injury [3,4].

        c. Equipment-related sources

           (1) Incorporation of a 25 μm filter into the aortic inflow line effectively reduces cerebral embolic load and has been shown to decrease the incidence of postoperative cognitive dysfunction.

           (2) Membrane oxygenators give rise to markedly fewer gaseous microemboli than bubble oxygenators, but this does not entirely eliminate the risk of air emboli. Similarly, air entrained into the venous side of a membrane oxygenator, or gaseous emboli resulting from drug administration via injection directly into CPB circuitry can transit the membrane and appear in the arterial inflow line, despite use of arterial line filters.

           (3) Use of 20 to 40 μm filters in the cardiotomy return line prevents particulate debris from the operative site from entering the CPB circuit. Use of cardiotomy blood washing techniques (cell saver) is associated with reduced amounts of cerebral lipid microemboli but has also been shown to result in greater transfusion requirements and has not been shown to consistently improve CNS outcomes.

           (4) Use of nitrous oxide (N2O) before commencement of CPB has been associated with increased evidence of ischemic damage, likely because residual N2O increases the size of any microgaseous emboli in the cerebral circulation. This is especially true for several hours after CPB when high fractional inspired oxygen (FIO2) should be used to minimize the size of residual gaseous microemboli.

   B. Hypoperfusion

      1. Watershed areas. Collateral perfusion of the brain can occur via extracerebral anastomoses (primarily through the circle of Willis) or by way of intracerebral anastomoses between major cerebral arteries, known as arterial boundary zones (watershed zones; Fig. 24.5). Rapid severe hypoperfusion can produce ischemic lesions within these boundary zones found at the territorial limits of the major cerebral arteries. The most frequently affected area is the parieto-occipital sulci located at the limits of the anterior, middle, and posterior cerebral arteries. Despite a global ischemic stress, these watershed lesions may be focal and asymmetrical. Placement of electroencephalogram (EEG) electrodes using a parasagittal montage (see following) may allow increased sensitivity for border zone ischemia detection.

      2. Cerebral perfusion pressure (CPP). During moderate hypothermia (28 to 30°C) using α-stat pH management, autoregulation is preserved in patients without overt cerebrovascular disease over the CPP range from 20 to 100 mm Hg. However, studies in elderly cardiac surgical patients and those with cerebrovascular disease indicate a wide variability of lower limit of autoregulation between 45 and 80 mm Hg [12]. Additionally, there are several conditions in which autoregulation may be lost (Table 24.2). With profound hypothermia (15 to 20°C), there appears to be loss of autoregulation as a result of hypothermia-induced vasoparesis, while diabetic patients have been shown to have impaired cerebral autoregulation even at moderate hypothermia.

Table 24.2 Factors associated with loss of cerebral autoregulation

          During CPB, there may be dissociation between MAP and CPP as a result of unrecognized cerebral venous hypertension. Particularly with use of a single two-stage venous cannula, cerebral venous drainage may be impaired specially during performance of posterior distal anastomoses (Fig. 24.6). Consequently, jugular venous pressure should be measured proximally within the superior vena cava (SVC; e.g., via introducer port of pulmonary artery or central venous catheter). CPP can also be compromised during performance of off-pump coronary surgery especially while performing multiple-graft procedures. During these operations the patient is often placed head-down and the heart is lifted to expose the distal targets, two factors which can increase central venous pressure and thus decrease CPP. Concurrent systemic arterial hypotension and low cardiac output often occur concomitantly with resultant cerebral hypoperfusion.

Figure 24.6 A: Systolic, mean, and diastolic arterial blood pressures, with commencement of CPB indicated at 3:15 p.m., after which MAP is shown. B: Pulmonary artery systolic, mean, and diastolic pressures with proximal jugular venous pressure (JVP) recorded at 3:15 p.m., with commencement of CPB. A single two-stage venous cannula was used for CPB. With rotation of the heart, venous return to the oxygenator decreased and JVP approached MAP values. (Modified from Murkin JM. Intraoperative management. In: Estaphanous FG, Barash PG, Reves JG, eds. Cardiac Anesthesia: Principles and Clinical Practice. Philadelphia, PA: J.B. Lippincott Company; 1994:326, with permission.)

      3. Circulatory arrest. During circulatory arrest for surgical procedures, profound hypothermia (16 to 18°C) is used to minimize CMRO2 and increase tolerance for ischemia (see following). During circulatory arrest under normothermic conditions, O2 levels are depleted within a few seconds of onset of ischemia, EEG activity is lost (isoelectric EEG) within 30 s, high-energy phosphates are exhausted within 1 min, and ischemic neuronal damage is found after periods of anoxia as brief as 5 min.

          For certain cardiac electrophysiologic procedures (e.g., diagnosis and treatment of certain refractory arrhythmias), transient ventricular fibrillation (VF) is often induced at normothermia and without circulatory support. Duration of VF must be limited to less than 1 min, and prompt hemodynamic resuscitation with at least 4 min of reperfusion should be maintained between episodes of VF [18]. Monitoring and management of these patients should follow the principles outlined herein.

      4. Intracerebral and extracerebral atherosclerosis. Patient-related factors including intracerebral and extracerebral atherosclerosis also modify the impact of perioperative hypotension. Neurologic injury after cardiac surgery is higher in patients with previous stroke, hypertension, advanced age, diabetes, and carotid bruit [25], which are all factors related to more extensive cerebrovascular disease. In one series of 206 perioperative CABG patients, over 50% exhibited concomitant cerebrovascular disease [19]. Such patients are more prone to cerebral ischemia secondary to perioperative hemodynamic instability.

IV. Pathophysiology of neuronal ischemia

Ischemia is defined as diminution of blood flow below a critical level that propagates tissue damage. Whether from embolization or hypoperfusion, neuronal ischemia is the final common pathway leading to cerebral damage. The extent of ischemic changes will depend on the duration of the ischemic insult, the affected vascular territory, the presence of collateral circulation, and factors that either ameliorate (e.g., hypothermia) or increase (e.g., hyperglycemia) the impact of ischemia on neuronal tissue [20].

   A. Lactic acidosis. Glucose is essentially the sole substrate for energy production by the brain, being metabolized to produce 36 moles of adenosine triphosphate (ATP) per mole glucose. Oxygen is essential for oxidative phosphorylation, and in the presence of ischemia, anaerobic glucose metabolism yields only 2 moles of ATP and results in lactate production with accumulation of hydrogen ion (H+). Anaerobic glycolysis is the primary cause of acidosis during ischemia, and the severity of lactic acidosis is directly related to preischemia glucose concentrations. Hyperglycemia is associated with worsening of neurologic injury after cerebral ischemia, and should be avoided in the perioperative interval.

   B. Apoptosis, necrosis, and inflammation. Two distinct phases of cellular death have been described after cerebral ischemia: apoptosis and necrosis. These are related to the intensity and duration of ischemic insult. Apoptosis is programmed cerebral cell death. Its main features include cell shrinkage with preservation of cell membrane and mitochondrial integrity and lack of inflammation and injury to surrounding tissue. There is some evidence that CPB may exacerbate apoptotic processes accelerating neuronal loss manifest as delayed postoperative CNS injury. Necrosis is a nonprogrammed event leading to cellular swelling, disruption of cell membrane and mitochondrial damage with inflammatory reaction, vascular damage, and edema formation [20]. The core of the ischemic tissue will show predominantly necrosis, and apoptosis will be found mostly in the periphery of the ischemic area. The sensitivity of neurons to ischemic insult varies by region with hippocampal areas exhibiting marked vulnerability.

   C. Ion gradients and role of calcium. Neuronal function and structural integrity are dependent on ionic gradients, such that up to 75% of ATP produced by resting neurons is utilized by sodium–potassium ATPase and for extrusion of calcium by calcium-dependent ATPase. With ischemia, decreased ATP production and evolving lactic acidosis impair transmembrane ionic pumps and consequently diminish cellular electrochemical gradients leading to cell depolarization. Extraneuronal leakage of K+ depolarizes adjacent neurons, thereby decreasing synaptic transmission and, along with calcium, promoting vasospasm in adjacent vasculature.

   D. Calcium. With ischemia, ATP depletion causes loss of ionic gradients, resulting in cell membrane depolarization and influx of calcium ion (Ca2+) through voltage-sensitive channels. Intracellular accumulation of Ca2+ is likely the final common pathway leading to neuronal death through enhanced protein and lipid catabolism. Elevated intracellular calcium will activate both phospholipases, which leads to membrane cell breakdown and arachidonic acid and free radical formation, and endonucleases, which induces fragmentation of genomic DNA, mitochondrial dysfunction, and energy failure. The intensity of intracellular calcium overload is the key element leading to irreversible cellular damage. Influx of Ca2+ can be minimized by calcium antagonists. Nimodipine has shown clinical benefit in decreasing vasospasm after subarachnoid hemorrhage, but has been associated with increased bleeding and mortality in cardiac surgical patients.

   E. Free fatty acids. Some of the earliest cell membrane changes with ischemia involve production of free fatty acids (FFAs) from membrane phospholipids. Intracellular Ca2+ activates calcium-dependent phospholipases C and A2, transforming membrane phospholipids into FFAs, which themselves are neurotoxic. FFAs are powerful uncouplers of oxidative phosphorylation and can undergo further oxidation from arachidonic acid, with resultant free radical formation. During cerebral ischemia, FFA production is decreased by administration of calcium antagonists and 21-aminosteroids (lazaroids), potent inhibitors of lipid peroxidation. Despite laboratory promise, clinical trials have so far been disappointing.

   F. Excitotoxicity. Glutamate is the most abundant excitatory amino acid (EAA) in the brain. It serves metabolic, neurotransmitter, and neurotropic functions and is normally compartmentalized in the neuron. Under normal conditions the brain has the ability to quickly uptake extracellular glutamate. Glutamate stimulates two kinds of receptors: Ionophore-linked receptors and metabotropic receptors; the last ones act only as modifiers of the excitotoxic injury. The main excitotoxic role lies in the ionophore-linked receptors, and these include NMDA (N-methyl-D-aspartate), AMPA (alpha-amino-methylisoxazole-propionic acid), and kainate, responsible for mediating transmembrane Ca2+ and Na+K+ passage. Ischemia produces enhanced presynaptic EAA release and decreased reuptake, which causes activation of postsynaptic NMDA and AMPA receptors and produces massive efflux of K+ and influx of Na+ and Ca+ and resultant osmolysis and calcium-related damage. EAAs also are increasingly implicated in free radical formation. Administration of ketamine, an NMDA receptor antagonist, has shown variable efficacy to decrease neuronal ischemic injury (see later).

   G. Nitric oxide. NO is a free-radical gas synthesized from L-arginine by NO synthase (NOS). NO functions as a neurotransmitter and has a role in regulating CBF and inflammation. In brain ischemia, the elevation of intracellular calcium markedly increases the activity of NOS. Increased NO combined with superoxide anion leads to formation of other reactive oxygen species, hydroxyl free radicals, and nitrogen dioxide producing proteolysis and cell damage. NO also mediates activation of ADP-polymerase leading to ATP and nicotinamide consumption and cell death [20]. Experimentally, lazaroids ameliorate neuronal ischemic damage when administered for ischemic stress, but results of clinical trials have not been positive.

   H. Leukocytosis. In a subanalysis of a trial randomizing 18,558 patients with symptomatic vascular disease to aspirin or clopidogrel, it was observed that in the week prior to a second vascular event, the quartile with highest leukocyte counts had higher risks for ischemic stroke, myocardial infarction, and vascular death after adjustment for other risk factors [21]. In the week before a recurrent event, but not at earlier time points, the leukocyte count was significantly increased over baseline levels, suggesting that leukocyte counts and mainly neutrophil counts are independently associated with ischemic events in these high-risk populations. Consistent with this, in a prospective study of 7,483 patients who underwent CAB or valvular surgery or both, leukocyte count was compared with the occurrence of postoperative stroke [22]. There were a total of 125 postoperative strokes and it was demonstrated that leukocyte count was significantly higher preoperatively and directly postoperatively in patients with stroke and that the magnitude of elevation of leukocyte count was correlated with magnitude and extent of stroke. These results strongly implicate inflammation and white cell activation as etiologic in both the extent and severity of perioperative cerebrovascular events.


V. Intraoperative cerebral monitoring

As shown in Table 24.3, there are various confounds for EEG monitoring, which along with the complex etiology of perioperative neurologic morbidity, suggest multimodal neuromonitoring as the best approach for systematic detection and avoidance of perioperative cerebral complications.

Table 24.3 Electroencephalogram confounds

   A. Brain temperature. Accurate monitoring of brain temperature is essential because temperature profoundly influences CMR and thus tolerance for ischemia. Mild hypothermia (less than 35°C) is disproportionately effective in decreasing ischemia-related injury due to inhibition of EAA release (see previous sections). During CPB, thermal gradients exist between various tissues; thus, brain temperature must be measured independent of other sites. Because of the small risk of trauma associated with placement of a tympanic thermistor, nasopharyngeal temperature (NPT) is the preferred site for clinical monitoring of brain temperature. Thermistor insertion should be through the nares to the level of the midpoint of the zygoma, a depth of 7 to 10 cm in an adult. Insertion of the thermistor before heparinization, using lubrication and exerting gentle pressure parallel to the floor of the nose, will prevent epistaxis and trauma to mucosa and turbinates. Esophageal temperature is a poor substitute for NPT because it variously reflects aortic inflow temperature, temperature of surrounding tissue, and the influence of residual ice or cooled fluid within the pericardial sac. For DHCA and high-risk patients, a thermistor/oximetric catheter can be placed retrograde into the jugular bulb, thus providing the most sensitive clinical measure of global brain temperature and oxygenation.

   B. EEG. EEG represents the amplified, summated, spontaneous electrical activity of the superficial cerebral cortex. Each electrode reflects microcurrent (10 to 200 μV) generated by electrical gradients across layers of neurons aligned at right angles to the monitored cortical surface in a 2 to 3 cm radius. Electrode placement should be based on the standard 10- to 20-electrode system and modified according to the number of channels being monitored (Fig. 24.7). EEG activity is commonly divided into four bands according to frequency: δ less than 4 Hz; θ 4 to 8 Hz; α 9 to 12 Hz; and β greater than 13 Hz. In general, slower frequencies indicate a deeper level of anesthesia. Several factors can confound interpretation of intraoperative EEG (Table 24.3) which, along with its technical complexity, have limited its clinical use to monitoring cooling during DHCA. Recordings are potentially made in the presence of various anesthetic agents, during profound changes in body temperature, and in the electrically hostile environment found in an operating room. Although subtle EEG changes may be difficult to interpret, development of asymmetric EEG activity should be considered to represent hemispheric compromise (Table 24.4).

Table 24.4 Causes of electroencephalographic asymmetry

Figure 24.7 Standard bipolar parasagittal montage based on the international 10–20 system. FP1 and FP2 refers to frontal pole; F3F4F7, and F8 refer to frontal; C3 and C4 refer to central; P3 and P4 refer to parietal; O1 and O2 refer to occipital; and T3T4T5, and T6 refer to temporal positions. (From Murkin JM, Moldenhauer CC, Hug CC Jr, et al. Absence of seizures during induction of anesthesia with high-dose fentanyl. Anesth Analg. 1984;63:489–494.)

      1. Processed EEG. After initial electronic filtering, analog EEG voltages are rapidly digitized (150/s) and analyzed over “epochs” (generally 2 to 4 s in duration) using analyses based on either frequency-domain or time-domain processing.

        a. Compressed spectral array (CSA) and density-modulated display of power spectrum analysis (DSA). For frequency-domain processing, many EEG applications use power spectral analysis. In this application, each EEG epoch is converted into a series of sine-wave components using Fourier transformation that treats the digitized EEG as a sum of sine waves of variable frequency and power. The amplitude (power) of each of the sine-wave components is indicated as a function of its frequency, and in the CSA each EEG epoch is shown over time in a three-dimensional representation (frequency vs. power vs. time) with the most current epoch in the foreground. The vertical displacement, representing both power and time, hinders recognition of low-amplitude activity followed by high-amplitude activity in the same frequency band. DSA is a representation in which each epoch is displayed using gray-scale intensity or dot size proportional to the power of the individual frequency band plotted. Consequently, it can be difficult to recognize small changes in frequency using this display.

        b. Spectral edge frequency (aperiodic analysis). Aperiodic analysis is time-domain–based processing and does not use Fourier transformation. Instead it is based on assessing voltage versus time of the raw EEG. For each component EEG wave in an epoch, the frequency is determined as the reciprocal of the time interval measured between zero axis voltage crossings, the zero-crossing frequency (ZXF), while the amplitude is the square root of the sum of squares of the voltages of the wavelets. Fast- and slow-wave components are analyzed separately, then combined for display. This model of analysis is also used to calculate the burst-suppression ratio, which can be an indicator of anesthetic depth and cerebral metabolism depression. Epileptiform activity and artifact have been reported to be most readily identified by time-domain processing. EEG frequency carrying the median power (median frequency power) correlates with plasma levels of several narcotics. The spectral edge frequency (frequency below which 95% of summated EEG power is contained) correlates with clinical assessment of anesthetic depth achieved with barbiturates or volatile anesthetics.

      2. Bispectral (BIS) index. Most frequency-domain processing (CSA, DSA) treats those component waveforms resulting from Fourier transformation as independent. Bispectral (BIS) analysis measures potential interactions between the waves to determine the presence of interactive components (harmonics) indicative of phase coupling (biocoherence), information that is not present in power spectral analysis. It has been recognized that EEG slowing and synchrony often occur in relation to increasing depth of anesthesia. The BIS measurement is the first device specifically for the measurement of the hypnotic effects of drugs approved by the US Food and Drug Administration (FDA).

      3. Evoked potentials. Metabolic and hemodynamic homeostasis determines the state of cerebral functional integrity. The latter can be inferred from EEG changes in response to repeated stimulation of intact afferent pathways. Separated from raw EEG and averaged, these evoked potentials (EPs) are described in terms of latency (time between the stimulus and respective EEG change) and amplitude (cortical microcurrent 1 to 5 μV). Reduction in CBF below 18 mL/(100 g · min) causes progressive decrease of the latter, which disappears at CBF below 15 mL/(100 g · min). In clinical practice, only the response of sensory neurons of gray matter can be tested in this way. More commonly, EPs serve to monitor the function of sensory tracts. Certain anesthetic agents complicate the recognition of specific effects of changing metabolic environment on EPs (e.g., isoflurane increases latency and decreases amplitude of somatosensory EPs) and have opposite effects on different EPs (visual somatosensory). As temperature changes also affect the latency and amplitude of EPs, the net result is potential intraoperative variability in EP during cardiac surgery potentially limiting its clinical applicability.

   C. TCD. Insonation of blood moving within a vessel produces a characteristic shift in signal frequency (Doppler shift) that is proportional to the flow velocity. Use of low-frequency sound waves (2 to 4 MHz) from depth-gated, direction-sensitive probes allows transmission through thin areas of skull (e.g., temporal window located above zygomatic arch between ear and orbit). This transmission enables continuous assessment of blood flow velocity within major intracerebral arteries (e.g., proximal MCA). Cerebral perfusion characteristics also can be assessed using TCD insonation for demonstration of laminar versus pulsatile flow or for detection of emboli. Because dissimilar acoustic echoes reflect inhomogeneities in the insonated substrate, microaggregate or microgaseous emboli can be detected within the bloodstream. Because TCD essentially functions as a microphone, artifactual noise transients can register as emboli. However, certain criteria have been employed to distinguish embolic signals from noise artifact (Table 24.5). Much greater acoustic resonance of gas emboli relative to formed elements creates limits of TCD detection for formed elements greater than 100 μm. In addition, the amplitude of signal is proportional to the size of the embolus, whereas for bubble emboli, limits of resolution are 50 μm and the amplitude of the reflected signal is unrelated to size of the bubble.

   Because of the ability to focus a pulsed U/S beam, however, by using a dual gating technique in which the vessel is insonated at two discrete sites, emboli can be discriminated from artifacts. This reflects the fact that emboli propagate with blood motion and artifact does not and thus emboli but not artifact will be detected sequentially at different depths along the insonated cerebral artery.

   One of the major goals in intraoperative TCD monitoring is discriminating solid versus gaseous cerebral emboli. Solid and gaseous microemboli may be differentiated with a new generation of multifrequency transducers, using both 2- and 2.5-MHz crystals, based on the principle that solid microemboli reflect more ultrasound at the higher than at the lower frequency, whereas the opposite is the case for gaseous microemboli. How robust this will prove in clinical practice remains to be seen and the results to date remain unconvincing.

Table 24.5 Transcranial Doppler characteristics of emboli versus noise

   D. Jugular oximetry. The characteristic attenuation of 650 to 1,100 nm infrared light by a few specific light-absorbing chromophores (primarily oxyhemoglobin, deoxyhemoglobin, and oxidized cytochrome c oxydase) imparts wavelength (color) shift on the incident light. This spectral shift is proportionate to the degree of oxygenation enabling quantification of tissue oxygenation using optical spectroscopic devices. Placement of a fiberoptic oximetric catheter into the jugular bulb provides continuous monitoring of the hemoglobin saturation of effluent cerebral venous blood and reflects global cerebral O2 supply and demand balance. Jugular oximetry may provide an appropriate endpoint for termination of cooling before DHCA (see below). After jugular saturation has increased maximally and stabilized, CMRO2 is at its lowest. Such monitoring has identified an association between rewarming after hypothermic CPB and significant cerebral venous blood desaturation. This indicates mismatching between cerebral O2 supply and metabolic rate, and increasing either hemoglobin concentration or depth of anesthesia (greater metabolic suppression) may be appropriate.

   E. NIRS. Similar principles of light absorbance are used during noninvasive cerebral optical spectroscopy using scalp-attached probes. Most of the currently available commercial devices employ two-channel monitoring using adhesive pads with one or more transmitting and two or more separately spaced receiving optodes. Differential spacing of the receiving optodes enables correction for extracerebral tissues to be made, allowing an assessment of regional oxygen saturation (rSO2) of cerebral cortex (Fig. 24.8). Current studies estimate that there is between 5% and 15% influence of extracerebral tissue on cerebral oximetry values as measured. Advantages and limitations of NIRS cerebral oximetry monitoring are shown in Table 24.6. This device enables indices of cerebral oxygenation to be determined in a continuous manner in a variety of clinical circumstances. There is no requirement for pulsatile blood flow enabling continuous monitoring during CPB, and there are no temperature-related artifacts. A potential limitation is the fact that the cerebral sample volumes are on the order of 1 mL of frontal cortical tissue, thus rendering them highly localized in nature. It is also apparent that since NIRS measures total tissue oxygenation, various factors including patient age, hemoglobin concentration at the measurement site, and sensor location can affect rSO2 values. A prospective study demonstrated that avoidance of low intraoperative cerebral oximetry values decreases major organ morbidity and death in patients undergoing CABG surgery [23], and a strong correlation has also been made between low preoperative baseline cerebral oximetry values and short- and long-term morbidity and mortality in cardiac surgical patients [13]. While there is increasing clinical consensus that cerebral oximetry is beneficial in patients undergoing hypothermic circulatory arrest with direct cerebral perfusion [24], other studies report skeptical results about cerebral oximetry technology [24,25] to counterbalance the argument against routine use in cardiac surgical patients. Large-scale multicenter clinical outcomes studies will be required in order to make a solid recommendation for the widespread use of cerebral oximetry in cardiac surgical patients or other specific subgroups of patients.

Table 24.6 Advantages and limitations of NIRS cerebral oximetry

Figure 24.8 Schematic representation of tissue layers through which light must propagate to reach the brain. Light propagating from source to receiver 1 has a mean tissue path length such that it predominantly samples superficial tissue (scalp and skull), whereas light propagating to receiver 2 has a deeper mean path length into brain. The signal from receiver 1 is used to correct the signal from receiver 2 for superficial tissue contamination. (From McCormick PW, Stewart M, Goetting MG, et al. Noninvasive optical spectroscopy for monitoring cerebral oxygen delivery and hemodynamics. Crit Care Med. 1991;19:89–97, with permission.)

   F. CPP. It represents the difference between driving pressure, or MAP, and downstream pressure, or intracranial pressure. During CPB, direct measure of intracranial pressure is not available; thus, CVP is often used as a surrogate. In the presence of impaired drainage from the SVC, which may occur during dislocation of the heart (particularly with use of a single two-stage cannula), cerebral venous hypertension may occur. Because atrial drainage is unimpaired, CVP measured from the atrium will be low; hence, this condition may be unrecognized. If sustained, cerebral venous hypertension can lead to cerebral edema and substantially decreased CPP, despite apparently adequate MAP (see Fig. 24.6). NIRS cerebral oximetry has been demonstrated to rapidly detect such events when cerebral desaturation occurs. During CPB, cerebral venous pressure should be monitored by a catheter placed proximally (usually pulmonary artery catheter introducer sheath) in the SVC and by visual inspection of the face.


VI. Prevention of CNS injury.

     As discussed in detail below, Table 24.7 shows a series of evidence-based recommendations designed to limit the risk of perioperative cerebral injury in cardiac surgical patients [26]. Table 24.8 outlines specific interventions designed to limit or avoid particular risk factors.

Table 24.7 Evidence-based guidelines for best practice bypass

Table 24.8 Perioperative strategies minimize CNS brain injury

   A. Embolic load

      1. Aortic instrumentation

        a. Although still the standard of care, palpation of the aorta has not proven sensitive to detect aortic atherosclerosis. Direct EAS scan of ascending aorta is the most sensitive technique for assessment of atherosclerotic burden. Alternatively, initial TEE screening of descending aorta, followed by EAS if TEE detects descending aortic atherosclerosis, represents an acceptable screening strategy. With extensive aortic atherosclerosis, distal aortic arch or axillary artery cannulation should be considered.

        b. Minimize the number of aortic clampings. Use of all arterial grafts (e.g., mammary, gastroepiploic) or sutureless proximal anastomotic devices eliminates the need for aortic side clamping for proximal anastomoses. In cases of severe atherosclerosis, “no touch” techniques (OPCAB with zero manipulation of the ascending aorta) have been shown to significantly decrease stroke rate [27].

   B. Off-pump versus conventional CAB

In one of the most recent meta-analyses of risk of stroke in OPCAB versus conventional on-pump CAB (CCAB) surgery, analysis of 59 most recent randomized clinical trials involving 8,961 patients, of whom 4,461 patients underwent OPCAB and 4,500 were randomized to CCAB, determined a significant difference in composite stroke incidence of 1.4% in OPCAB versus 2.1% in CCAB groups [28]. Not inconsistent with this result, no studies have reported a higher stroke risk with OPCAB but have shown either no effect or a trend for decreased stroke rate associated with OPCAB. Accordingly, selective use of OPCAB for patients at increased risk of stroke related to aortic atherosclerosis would appear to be an important strategy to minimize early stroke risk.


      1. Perfusion equipment and techniques

        a. Precirculation of CPB circuit for a minimum of 30 min with a 5 μm filter before usage removes plasticizers and other manufacturing microdebris.

        b. Incorporation of a micropore (20 to 40 μm) filter into the cardiotomy return line keeps tissue and other particulate debris from the surgical field out of the CPB circuit.

        c. Retransfusion of cardiotomy suction blood after processing using cell saver. Of note, while this may decrease embolic load it has not been consistently shown to improve early postoperative cognitive outcomes but has been associated with increased transfusion requirements.

        d. Use of a 40 μm filter on the arterial inflow line decreases delivery of emboli into the arterial circulation.

        e. To minimize gas bubble formation due to decreased solubility with rewarming, the temperature gradient between the arterial inflow blood and the patient must be less than 10°C, particularly with use of a bubble oxygenator.

        f. During rewarming, arterial blood inflow temperature must not exceed 37°C.

        g. Be aware of the possibility of air entrainment from cardiac vents in the surgical field, and ensure meticulous deairing of CPB venous cannulae and syringes for injection into the CPB circuitry to decrease arterial gas embolization.

      2. Open-chamber deairing techniques

        a. Before ventricular ejection, needle aspiration of the LV and LA, combined with manual agitation of the heart, to dislodge air entrapped in trabeculae. This process should be combined with concomitant manual ventilation of the lungs to mobilize residual air within the pulmonary veins.

        b. Use of TEE to detect residual intracavitary air and to direct needle aspirations.

        c. Tilting the patient’s head down. This procedure achieves a dependent position of the great vessels of the head and is thought to minimize cerebral embolization although evidence for its efficacy is lacking.

        d. Transient bilateral carotid compression during defibrillation and initial filling and commencement of heart ejection. This maneuver should be reserved for instances where the risk of intracavitary air remains high and there is no suspicion of carotid atherosclerosis.

   C. Cerebral perfusion. In normal individuals during moderate hypothermia, relative hypotension is well tolerated because cerebral autoregulation is preserved down to CPP 20 mm Hg with α-stat blood gas management. However, in elderly patients or those with cerebrovascular disease, higher pressure should be maintained as the lower autoregulatory threshold has been shown to vary markedly in such patients [12,14]. As assessed by NPT, the brain rewarms rapidly; therefore, hypotension (MAP less than 50 mm Hg) should be avoided after commencement of rewarming in all patients. Inadvertent compromise of CPP should also be avoided by monitoring proximal SVC pressure to detect cerebral venous hypertension. Diabetics and patients with previous CVA have impaired cerebral autoregulation and CBF is directly dependent on MAP. Such patients, as well as those with chronic hypertension, may benefit from close CNS monitoring (see previous discussion) and maintenance of higher perfusion pressures.

   D. Euglycemia. There is considerable evidence from experimental models and from patients with CVA that hyperglycemia increases the magnitude and extent of neurologic injury during ischemia. Hyperglycemia should be avoided as a basic approach. Glucose-free infusions and a glucose-free prime should be used for CPB circuit, because insulin resistance develops during CPB (partially as a result of increased endogenous catecholamines), producing glucose intolerance and increasing the tendency for refractory hyperglycemia. A structured approach to maintain normal values of blood glucose is considered favorable to patients’ outcomes and is a recognized element of best practice CPB guidelines as shown in Table 24.7 [26].

   E. Mild hypothermia. There is increasing evidence showing that EAAs are pivotal in the genesis of ischemic neurologic injury (see preceding discussion). Since EAA synthesis and release are critically temperature dependent and are significantly inhibited below 35°C [20], brain temperature (NPT) should be monitored continuously during rewarming, hyperthermia (NPT greater than 37°C) must be avoided, and brain temperature should be maintained less than 37°C until after separating from CPB and decannulation [26].

   F. Best practice CPB. Recommendations from an evidence-based review for conducting safe, patient-centered CPB as based on a structured MEDLINE search coupled with a critical review of scientific literature and debates stemming from presentations at regional and national conferences, with the level of evidence and findings graded using criteria promulgated by the American Heart Association and American College of Cardiology Task Force on Practice Guidelines, are shown in Table 24.7 [26].

VII. Pharmacologic cerebral protection

Despite a profound and ongoing increase in the understanding of the mechanisms involved in ischemic neuronal injury and the development of new classes of drugs, currently none represent a standard of practice; but these or related compounds may become part of the therapeutic armamentarium in the near future.

   A. Metabolic suppression

      1. Rationale and limitations. Metabolic activity is temperature dependent, and hypothermia produces an exponential decrease in CMR. Unlike pharmacologic metabolic suppressants, hypothermia decreases metabolic activity related both to functional activity (e.g., EEG activity) and basal activity (e.g., ion pumps). Hypothermia prolongs the tolerance for global ischemia (Fig. 24.9) and is undertaken particularly for circulatory arrest (see following discussion). During cardiac surgery, however, greatest risk for cerebral emboli occurs during normothermia with cannulation and decannulation; hence, pharmacologic metabolic suppressants have been investigated. While there was previously some interest in high-dose thiopental, no consistent clinical benefit was demonstrated likely since any potential benefit resulting from such therapy was derived from the suppression of CMR associated with suppression of synaptic activity which would likely result only in prolongation of ischemia tolerance of a very few minutes.

Figure 24.9 Nomogram of probability of “safe” total circulatory arrest according to duration of total arrest time as NPTs of 37°C, 28°C, and 18°C, defined as the duration of total arrest after which no structural or functional damage has occurred. (From Kirklin JK, Kirklin JW, Pacifico AD. Deep hypothermia and total circulatory arrest. In: Arciniegas E, ed. Pediatric Cardiac Surgery. Chicago, IL: Year Book; 1985:79–85, with permission.)

      2. Agents. As shown in Figure 24.10, various anesthetics have the ability to produce EEG burst suppression and result in profound decreases in CMR to approximately 50% of awake CMRO2, averaging 25 mL/(100 g · min).

Figure 24.10 EEG tracings from three patients during normothermic CPB. The top tracing demonstrates characteristic low-voltage activity occurring during high-dose fentanyl anesthesia. The middle tracing shows the burst-suppression pattern resulting from thiopental administration. The lower patterndemonstrates burst suppression occurring during isoflurane administration. (From Woodcock TE, Murkin JM, Gentile PS, et al. Pharmacologic EEG suppression during cardiopulmonary bypass: Cerebral hemodynamic and metabolic effects of thiopental or isoflurane during hypothermia and normothermia. Anesthesiology. 1987;67:218–224, with permission.)

        a. Thiopental. A dosage of 5 to 8 mg/kg results in 5 min of EEG suppression at normothermia. Proportional decreases in both CBF and CMRO2 are produced. An infusion at 0.5 to 1 mg/(kg · min) is required for prolonged EEG suppression, results in prolonged recovery and extubation times, and may increase the need for inotropic support because of myocardial depression.

        b. Propofol. Transient EEG burst suppression is obtained at dosages of 2 to 3 mg/kg and results in proportional decreases in CBF and CMRO2. Infusion at 0.1 to 0.3 mg/(kg · min) produces sustained EEG suppression and is rapidly metabolized; therefore, it does not prolong recovery and extubation times. Hypotension from systemic vasodilation may require administration of phenylephrine or other such vasoconstrictor.

        c. Sevoflurane and Deflurane. At inspired concentrations of 1.5 to 2 MAC, burst suppression is produced. Unlike the intravenous agents, EEG suppression with inhalational agents is not accomplished by any decrease in CBF, although CMRO2 is significantly reduced. Rapid elimination is characteristic of volatile anesthetics. Of particular interest is the evidence of ischemic preconditioning and neuroprotection associated with administration of volatile anesthetics which have been variously demonstrated to decrease glutamate release, modulate calcium flux, and inhibit generation of free radicals as well as modulate apoptosis [29]. Large-scale clinical studies are currently lacking but it does appear as though usage of volatile anesthetics does not increase and may have a salutary effect on postoperative CNS dysfunction.

   B. Calcium-channel blockers. Massive calcium influx is likely the final common pathway of ischemic neuronal injury (see preceding comments). In clinical trials, calcium-channel antagonists (nimodipine) have demonstrated efficacy in decreasing vasospasm after subarachnoid hemorrhage; however, nimodipine also has been associated with increased bleeding and higher mortality in cardiac surgical patients.

   C. Glutamate antagonists. Since excitotoxicity is recognized as central to ischemic neuronal injury (see previous discussion), EAA receptor antagonists are being actively investigated. NMDA and AMPA receptor antagonists have been found to be neuroprotective following cardiac arrest but clinical studies in cardiac surgical patients remain inconclusive [30].

   D. Lidocaine. Given its ability to decrease ischemia-mediated neuronal membrane depolarization and attendant excitotoxic cascades, two small clinical trials of lidocaine versus placebo have demonstrated lowered incidences of cognitive dysfunction in cardiac surgical patients. However, preliminary results from a larger clinical trial have been unable to confirm these results.

   E. Statins: An increasingly promising line of investigation is the role of perioperative statin therapy, either alone or in combination with other medications. Evidence is accruing that when administered prior to CAB surgery, statins reduce the risk of perioperative mortality, stroke, and atrial fibrillation, and when combined with β-blocker administration there is suggestive evidence that statins may result in a significantly decreased risk of perioperative stroke in CAB patients [21]. Whether these results will be borne out in large-scale prospective studies is currently unclear but current evidence suggests that benefits seem to outweigh the risks associated with their use, both in the preoperative and postoperative period.


   A. Clinical indications. DHCA sometimes is used for major surgical procedures because it provides a motionless, cannula-free, bloodless field. By allowing unobstructed surgical access, DHCA facilitates repair of complex congenital anomalies in neonates and infants. In adults, DHCA allows temporary interruption of cerebral perfusion, primarily for aortic arch reconstruction or for resection of giant cerebral aneurysms.

   B. Technique

      1. Core and external cooling. In most North American centers, active external cooling (e.g., ice baths) has been eliminated but packing the head in ice is still recommended to inhibit secondary rewarming prior to onset of reperfusion. Core cooling using CPB allows efficient and controlled onset of hypothermia, and cooling persists until core temperature (bladder, rectal) is stable at 15 to 20°C. Cooling must be continued until stable brain temperature (e.g., NPT, jugular thermistry) has been achieved. Some centers use development of isoelectric EEG as the endpoint for cooling, necessitating careful selection and titration of anesthetic agents.

      2. Decannulation. Before circulatory arrest, administration of long-acting muscle relaxants is essential to ensure profound paralysis in order to minimize systemic O2 consumption. With cessation of perfusion, venous cannulas are unclamped, allowing passive exsanguinations into the CPB circuit and decreasing distention of the heart and bleeding into the surgical site. For pediatric surgery, venous cannulas are usually removed to facilitate surgical exposure. Often passive circulation of blood within the CPB circuit is continued ex vivo to avoid stasis and setting of blood with platelet clumping.

   C. Brain protection

      1. Temperature. Hypothermia is the primary component of brain protection during circulatory arrest. The temperature coefficient (Q10), which is the ratio of metabolic rates at temperatures 10°C apart, has been shown to be 2.3 for human brain such that CMR is still 17% of baseline at 15°C [31]. A 20°C (37 to 17°C) decrease in brain temperature enhances cerebral tolerance for ischemia (see Fig. 24.9). Cooling the patient to a temperature of 10 to 15°C seems to offer the best protection to the brain. Minimizing rewarming of the brain is essential; therefore, external heat sources (e.g., overhead lights, ambient room temperature) should be minimized. Application of external ice packs to the head has been shown experimentally to delay brain rewarming and increase ischemic tolerance. Thiopental and/or steroids are administered before circulatory arrest in some centers, although any beneficial effect is unproven.

      2. Anterograde and retrograde perfusion. Because of cerebral autoregulation with preferential shunting of blood to the brain, even low perfusion rates (e.g., 10 to 25 mL · kg · min) during deep hypothermia have been shown to significantly improve cerebral ischemic tolerance in comparison to total circulatory arrest. A prospective randomized study has shown that continuous low-flow perfusion (0.71/min · m2) at 18°C for pediatric patients younger than 3 months undergoing arterial switch operations results in significantly lower incidences of clinical seizures and brain creatinine kinase isoenzyme versus HCA and is now the standard of care.

          For aortic arch procedures in which arterial inflow is restricted, selective cerebral perfusion (SCP) via brachiocephalic or carotid perfusion is employed in many centers. Since these techniques assume adequacy of circle of Willis, there is increasing clinical interest in bihemispheric NIRS cerebral oximetry to assess adequacy of unilateral SCP. NIRS cerebral oximetry has been reported to detect catheter kinking or inadequacy of flow in a number of cases. Retrograde cerebral perfusion does not provide sufficient substrate supply to maintain brain metabolic demand but may prevent rewarming and decrease cerebral embolization. In a recent case report, RCP with pressure-augmented perfusion guided by cNIRS was felt to provide enhanced cerebral perfusion and was not associated with neurologic deficit despite prolonged duration of perfusion [32]. There is general clinical consensus that cerebral oximetry monitoring is beneficial when SCP or RCP is employed.

      3. pH management. Although in a large randomized clinical trial, use of α-stat versus pH-stat acid–base management strategy during reparative infant cardiac operations with deep hypothermic CPB was not consistently related to either improved or impaired early neurodevelopmental outcomes, experimental and clinical studies have demonstrated more homogeneous brain cooling with pH-stat. Conversely, pH-stat impairs cerebral autoregulation and potentially increases cerebral embolization; therefore, a considered approach is to employ pH-stat during cooling and α-stat during rewarming.

   D. Summary (Table 24.9)

Table 24.9 Summary of deep hypothermic circulatory arrest (DHCA)

IX. Cardiac surgery in patients with cerebrovascular disease

   A. Incidence. Coronary atherosclerosis increases the likelihood of coexisting carotid arteriopathy. Presence or absence of carotid bruits is a poor predictor of carotid stenosis or the risks of perioperative stroke. In a survey of patients undergoing CABG, 5.5% showed significant unilateral carotid stenosis, 2.2% had bilateral stenoses, and 1.5% of patients had unilateral or bilateral carotid occlusion. Another study of over 200 CABG patients demonstrated a 54% incidence of significant cerebrovascular (carotid or intracranial) atherosclerosis preoperatively [19]. In general, however, noninvasive (ultrasonography) and invasive (contrast arteriography) investigations are usually reserved for patients who have had overt symptoms of cerebrovascular insufficiency (e.g., transient ischemic attacks or stroke) during the previous 3 to 6 months [33]. In part, this reflects the clinical recognition that carotid endarterectomy (CEA) does not appreciably reduce risk of perioperative CVA in cardiac surgical patients (see later) [33].

   B. Morbidity. Neurologic injury associated with cardiovascular surgery is a consequence of both cerebral emboli and hypoperfusion. Carotid disease is an important etiologic factor in the pathophysiology of post-CAB stroke but is probably only responsible for, at most, about 50% of all strokes [1,33]. Carotid stenosis is associated with a greater incidence of aortic atherosclerosis and concomitant cerebrovascular disease. The presence of combined carotid and cardiac disease suggests more advanced and severe atherosclerosis and a higher risk of embolization and/or hypoperfusion during cardiac surgery and emphasizes the role of adequate intraoperative management (e.g., neuromonitoring and avoidance of hypoperfusion) in susceptible patients.

   C. Combined carotid and cardiac procedures

      1. Rationale. Indications for CEA and CAB should be considered independently. There is no compelling evidence that, in the absence of significant symptomatology, CEA decreases perioperative stroke risk [1,33]. The role of carotid stenting is promising and remains to be defined. Approximately 9% to 11% of patients undergoing either staged or synchronous procedures will die or suffer a nonfatal stroke/MI in the perioperative period. Prospective community-based studies have reported the worst outcomes [1,33].

      2. Morbidity. In asymptomatic patients CEA does not decrease perioperative stroke risk. The reported combined risk of death, CVA, and MI for synchronous procedures (CEA + CAB) is 11.5%, and for staged procedures (CEA then CAB) it is 10.2%, suggesting that institutional experience is of primary importance. Because of the paucity of natural history data, there is no systematic evidence that staged or synchronous operations confer any benefit over isolated CAB surgery [1,33].

          As such, while there is a strong correlation between risk of perioperative stroke and presence of carotid stenosis, it is not clear that CEA in any way mitigates this risk. In a meta-analysis of 11 studies in which 760 CAB patients underwent either staged or synchronous CAB plus CEA, it was noted that 87% of the patients were neurologically asymptomatic and 82% had unilateral carotid disease [34]. Overall, the 30-day risk of death or stroke was 9.1% and while it was observed that staged CAB then CEA was considered less invasive, the authors concluded that “it remains questionable whether the observed 9% risks of CEA can be justified in any asymptomatic patient with unilateral carotid disease” [34].

X. Summary

There is an evidence-based rationale in support of a number of procedural and technical modifications that have been shown to decrease perioperative CNS complications. Avoidance of instrumentation of ascending aortic atheroma by enhanced screening techniques (e.g., preoperative MRI, intraoperative EAS) and selective use of OPCAB and no-touch techniques for patients with significant aortic atherosclerosis, more judicious use of cerebral monitoring and NIRS for assessing adequacy of cerebral perfusion during CPB, avoidance of cerebral hyperthermia during rewarming on CPB and postoperatively, attenuation of perioperative and CPB-related inflammatory responses, enhanced usage of techniques to decrease postoperative atrial fibrillation, as well as administration of aspirin and statins throughout the perioperative period have all been shown as efficacious in decreasing perioperative CNS complications. The challenge is to better understand which patients are at increased risk and to encourage more widespread assessment and adoption of these various strategies.


 1. Serruys PW, Morice MC, Kappetein AP, et al; SYNTAX Investigators. Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease. N Engl J Med.2009;360:961–972.

 2. Bronster DJ. Neurologic complications of cardiac surgery: Current concepts and recent advances. Curr Cardiol Rep. 2006;8:9–16.

 3. Arrowsmith JE, Grocott H, Reves JG, et al. Central nervous system complications of cardiac surgery. Br J Anaesth. 2000;84: 378–393.

 4. Nishiyama K, Horiguchi M, Shizuta S, et al. Temporal pattern of strokes after on-pump and off-pump coronary artery bypass graft surgery. Ann Thorac Surg. 2009;87:1839–1844.

 5. Hedberg M, Boivie P, Engström KG. Early and delayed stroke after coronary surgery - An analysis of risk factors and the impact on short- and long-term survival. Eur J Cardiothorac Surg.2011;40:379–387.

 6. Dacey LJ, Likosky DS, Leavitt BJ, et al. Perioperative stroke and long-term survival after coronary bypass graft surgery. Ann Thorac Surg. 2005;79:532–536.

 7. Newman MF, Kirchner JL, Phillips-Bute B, et al. Neurological Outcome Research Group and the Cardiothoracic Anesthesiology Research Endeavors Investigators. Longitudinal assessment of neurocognitive function after coronary-artery bypass surgery. N Engl J Med. 2001;344:395–402.

 8. Selnes OA, Grega MA, Bailey MM, et al. Do management strategies for coronary artery disease influence 6-year cognitive outcomes? Ann Thorac Surg. 2009;88:445–454.

 9. Sellke FW, DiMaio JM, Caplan LR, et al. Comparing on-pump and off-pump coronary artery bypass grafting: Numerous studies but few conclusions: A scientific statement from the American Heart Association council on cardiovascular surgery and anesthesia in collaboration with the interdisciplinary working group on quality of care and outcomes research. Circulation. 2005;111:2858–2864.

10. Barber PA, Hach S, Tippett LJ, et al. Cerebral ischemic lesions on diffusion-weighted imaging are associated with neurocognitive decline after cardiac surgery. Stroke. 2008;39:1427–1433.

11. Grigore AM, Grocott HP, Mathew JP, et al. The rewarming rate and increased peak temperature alter neurocognitive outcome after cardiac surgery. Anesth Analg. 2002;94:4–10.

12. Brady K, Joshi B, Zweifel C, et al. Real-time continuous monitoring of cerebral blood flow autoregulation using near-infrared spectroscopy in patients undergoing cardiopulmonary bypass. Stroke.2010;41:1951–1956.

13. Joshi B, Brady K, Lee J, et al. Impaired autoregulation of cerebral blood flow during rewarming from hypothermic cardiopulmonary bypass and its potential association with stroke. Anesth Analg.2010;110:321–328.

14. Murkin JM, Martzke JS, Buchan AM, et al. A randomized study of the influence of perfusion technique and pH management strategy in 316 patients undergoing coronary artery bypass surgery. II. Neurologic and cognitive outcomes. J Thorac Cardiovasc Surg. 1995;110:349–362.

15. Ryu KH, Hindman BJ, Reasoner DK, et al. Heparin reduces neurological impairment after cerebral arterial air embolism in the rabbit. Stroke. 1996;27:303–309.

16. Ura M, Sakata R, Nakayama Y, et al. Ultrasonographic demonstration of manipulation-related aortic injuries after cardiac surgery. J Am Coll Cardiol. 2000;35:1303–1310.

17. Murkin JM, Baird DL, Martzke JS, et al. Cognitive dysfunction after ventricular fibrillation during implantable cardiovertor/defibrillator procedures is related to duration of the reperfusion interval. Anesth Analg.1997;84:1186–1192.

18. Yoon BW, Bae HJ, Kang DW, et al. Intracranial cerebral artery disease as a risk factor for central nervous system complications of coronary artery bypass graft surgery. Stroke. 2001;32:94–99.

19. Auriel E, Bornstein NM. Neuroprotection in acute ischemic stroke–current status. J Cell Mol Med. 2010;14:2200–2202.

20. Grau AJ, Boddy AW, Dukovic DA, et al; CAPRIE Investigators. Leukocyte count as an independent predictor of recurrent ischemic events. 2004;35:1147–1152.

21. Albert AA, Beller CJ, Walter JA, et al. Preoperative high leukocyte count: A novel risk factor for stroke after cardiac surgery. Ann Thorac Surg. 2003;75:1550–1557.

22. Murkin JM, Adams SJ, Novick RJ, et al. Monitoring brain oxygen saturation during coronary bypass surgery: A randomized, prospective study. Anesth Analg. 2007;104(1):51–58.

23. Heringlake M, Garbers C, Käbler JH, et al. Preoperative cerebral oxygen saturation and clinical outcomes in cardiac surgery. Anesthesiology. 2011;114:58–69.

24. Murkin JM, Arango M. Near-infrared spectroscopy as an index of brain and tissue oxygenation. Br J Anaesth. 2009;103(suppl 1): i3–i13.

25. Highton D, Elwell C, Smith M. Noninvasive cerebral oximetry: Is there light at the end of the tunnel? Curr Opin Anaesthesiol. 2010;23(5):576–581.

26. Shann KG, Likosky DS, Murkin JM, et al. An evidence-based review of the practice of cardiopulmonary bypass in adults: A focus on neurologic injury, glycemic control, hemodilution, and the inflammatory response. J Thorac Cardiovasc Surg. 2006;132:283–290.

27. Bergman P, Hadjinikolaou L, Dellgren G, et al. A policy to reduce stroke in patients with extensive atherosclerosis of the ascending aorta undergoing coronary surgery. Interact Cardiovasc Thorac Surg.2004;3:28–32.

28. Afilalo J, Rasti M, Ohayon SM, et al. Off-pump vs. on-pump coronary artery bypass surgery: An updated meta-analysis and meta-regression of randomized trials. Eur Heart J. 2011 Oct 10. [Epub ahead of print]

29. Yu Q, Wang H, Chen J, et al. Neuroprotections and mechanisms of inhalational anesthetics against brain ischemia. Front Biosci (Elite Ed). 2010;2:1275–1298.

30. Bhutta AT, Schmitz ML, Swearingen C, et al. Ketamine as a neuroprotective and anti-inflammatory agent in children undergoing surgery on cardiopulmonary bypass: A pilot randomized, double-blind, placebo-controlled trial. Pediatr Crit Care Med. 2011 Sep 15. [Epub ahead of print]

31. Bouchard D, Carrier M, Demers P, et al. Statin in combination with β-blocker therapy reduces postoperative stroke after coronary artery bypass graft surgery. Ann Thorac Surg. 2011;91:654–659.

32. Murkin JM. Pathophysiological basis of CNS injury in cardiac surgical patients: Detection and prevention. Perfusion. 2006;21:203–208.

33. Kubota H, Tonari K, Endo H, et al. Total aortic arch replacement under intermittent pressure-augmented retrograde cerebral perfusion. J Cardiothorac Surg. 2010;5:97.

34. Naylor AR, Bown MJ. Stroke after cardiac surgery and its association with asymptomatic carotid disease: An updated systematic review and meta-analysis. Eur J Vasc Endovasc Surg. 2011;41:607–624.

35. Naylor AR, Mehta Z, Rothwell PM. A systematic review and meta-analysis of 30-day outcomes following staged carotid artery stenting. Eur J Vasc Endovasc Surg. 2009;37:379–387.