The transfusion of homologous blood products is associated with multiple risks which are cumulative as more blood is transfused. Homologous transfusion requirements for patients undergoing cardiac surgery are costly and impose significant demands on hospital blood banks. Recent fears of HIV transmission have potentiated earlier concerns over risks such as transfusion reactions, transmission of other viral and bacterial diseases, coagulopathy, acute respiratory distress, isoimmunization, and febrile reactions.

The risk of a haemolytic transfusion reaction is 1:6000, and that of a fatal hemolytic transfusion reaction is 1:100000. Honig and Bove analysed deaths resulting from transfusions over a 2-year period and found that all were a result of major ABO incompatibility. Clerical error was the most frequent cause. Although ABO compatibility testing reduces the chances of immediate haemolytic reactions, non-haemolytic reactions occur in 1 to 2 per cent of recipients as a result of sensitization to leucocyte antigens. The incidence of fever, chills, and urticaria in recipients is 1:100.

Although current blood banking techniques make bacterial contamination of blood rare, the threat of transmission of viruses remains. Testing of blood donors for HIV antibody did not begin until 1985; in 1984 an estimated 7200 patients contracted transfusion-associated AIDS. Of these 27 per cent had received the infected unit during cardiac surgery. Although the risks have decreased appreciably since 1985, tests for HIV antibody do not detect all infected donors: HIV antibody may not be present until 6 to 14 weeks after infection, as determined by lymphocyte culture or experimental HIV antigen and antibody assay. The risk of transmission of the virus therefore remains. Although voluntary disqualification by donors at risk of HIV infection reduces the risk of transmission, some high-risk patients continue to donate blood. Current estimates of the risks of HIV transmission from screened blood vary from 1:40000 to 1:1000000.

Other clinically significant illnesses can be transmitted by transfusion. Hepatitis viruses are the most commonly transmitted agents. Screening for hepatitis B antibody began in 1971; however, the risk of infection ranges between 1:200 and 1:300 per unit received. Recent changes in criteria for donor acceptance, in conjunction with the introduction of non-specific tests (antibody to hepatitis B core antigen and alanine aminotransferase) have altered the incidence of transmission. Although the precise incidence is not known, it is estimated at 1:100 or less per unit transfused. In 1988, Houghton and coworkers identified the RNA virus responsible for the majority of cases of non-A, non-B hepatitis, now named hepatitis C virus. A portion of the genome of this virus has recently been cloned, and routine screening of donors for antibody began in 1990. Cytomegalovirus and Epstein-Barr virus are transmitted to previously unexposed recipients; however, infection is usually asymptomatic, except in immunocompromised hosts. Human T-cell lymphotropic virus is present in donor populations in the United States. This virus is associated with adult T-cell leukaemia and a chronic progressive myelopathy. Although no instance has yet been documented as a result of transfusion, the long latency period of these illnesses makes such an association difficult to prove. Malaria, syphilis, and babesiosis have also been transmitted.

Between 1967 and 1970, an average of 12 units of whole blood was transfused to each patient undergoing cardiac surgery. In 1973, the average patient received eight units. Over the last 20 years, advances have been made in perioperative blood salvage techniques; these have reduced the average transfusion requirements to 1 to 3 units per patient. In one series, 90 per cent of patients received no homologous blood.

Blood conservation in cardiac surgery is based on two principles: haemodilution and autologous transfusion. In 1962 Cooley reported the use of a crystalloid pump prime in a patient who refused blood for religious reasons. Verska et al. compared cardiopulmonary bypass with non-blood versus blood prime in 1974; the use of a crystalloid prime is now standard procedure in adult patients. The interest in autologous transfusion has been developing over many years and some form of autologous transfusion is used in most cardiac procedures. In 1978, Moran et al. reported the benefits of centrifugation of the residual extracorporeal circuit contents in order to maximize red blood cell salvage. Other forms of autologous donation currently in use include preoperative autologous blood donation, intraoperative withdrawal of one or more units of autologous blood prior to the institution of cardiopulmonary bypass for reinfusion after bypass, intraoperative autotransfusion of blood suctioned from the operative field, and the postoperative infusion of mediastinal blood collected in chest tube drainage reservoirs. The use of normovolaemic haemodilution has been extended into the postoperative period, with more stringent criteria for transfusion of red blood cells applied to most patients. A haematocrit of 25 per cent or less in completely revascularized patients is considered acceptable in many institutions.

Preoperative donation of autologous blood for intraoperative use in patients undergoing cardiovascular procedures was reported as early as 1967. Interest remained low in the 1970s, due to concerns over administrative difficulties and cost. Current concerns surrounding the risk of transmission of HIV virus, and the realization of the significant morbidity and mortality associated with transfusion-related hepatitis have renewed interest in this practice. Although autologous blood is commonly obtained for use in patients undergoing orthopaedic or other major surgical procedures, acceptance has been slow in the field of cardiac surgery because of concerns over the possible effects of blood donations by patients with coronary artery and valvular heart disease. Several recent reports have documented the safety of preoperative autologous donation by these patients, and have shown a significant reduction in the use of homologous blood. Love et al. reported on 58 patients from whom 1 to 3 units of autologous blood was obtained 1 to 3 weeks preoperatively (patients with unstable angina or aortic stenosis were excluded). Each unit of whole blood was fractionated into its components, and the fresh frozen plasma was also stored for autologous transfusion. There were no reported complications associated with autologous blood transfusion. Thirty-eight per cent of control patients required no homologous transfusion, compared with 64 per cent of patients who donated autologous blood. The administrative costs per unit for autologous blood were the same as for homologous blood. Owings et al. studied 107 patients who donated 1 to 6 units of autologous blood preoperatively. Critical aortic stenosis and medically refractory angina requiring admission to hospital were considered contraindications. Fresh frozen plasma was stored for all patients, and if surgery was delayed red blood cells were also frozen. In a total of 326 donations, no patient reported changes in cardiac symptoms significant enough to require deferral. Eighty-nine preoperative visits by patients led to deferrals, almost universally because of mild anaemia. While 82 per cent of control patients required homologous transfusion, this was the case for only 27 per cent of patients in the autologous transfusion group. A recent study on the effectiveness of recombinant human erythropoietin in increasing the ability of patients to donate autologous blood may alleviate problems with anaemia in the future.

The intraoperative use of a cell scavenging device during cardiac surgery significantly reduces requirements for banked blood. Although such devices have been available for many years, early models were labour-intensive, and their use was associated with complications such as haemolysis, coagulopathies, emboli, renal failure, and pulmonary abnormalities. Current models are completely automated and consist of a regionally heparinized collection system for retransfusion of washed red blood cells with a haematocrit of 50 to 60 per cent. The initial capital outlay is high and disposable parts are expensive. Salvaged red blood cells have normal survival time but higher levels of 2, 3-diphosphoglycerate than banked blood. A remaining problem is that dilutional coagulopathy can occur after several blood volumes have been processed by cell salvage because coagulation factors are discarded during the cell washing process.

Reinfusion of shed mediastinal blood has been shown to be safe. The haematological characteristics of blood shed from the pleural space were first documented by Symbas: haematocrit, fibrinogen, and platelets are decreased, the free plasma haemoglobin is increased, and the red blood cell survival time is normal. The Sorenson Receptaseal autotransfusion system, which consists of a rigid outer reusable canister with a disposable sterile liner, has been evaluated. The liner is made of two plastic bags, each with 170 &mgr;m filters. Shed blood is defibrinated by the action of the mediastinum and anticoagulant is not required. Most protocols require the reinfusion of shed blood every 4 h if drainage exceeds 250 ml; quantities smaller than this are discarded. Evaluation of shed blood showed haematocrits of 18 to 25 per cent, platelets of 60 to 70000, decreased fibrinogen, elevated fibrin split products, and elevated plasma haemoglobin. Platelets are non-functional. Despite these abnormalities, the use of shed blood, even in large volumes, does not adversely affect recipients' coagulation profiles or increase the need for platelets or fresh frozen plasma. Total postoperative blood loss and replacement is similar in autologous recipients and control patients. There is no evidence of renal failure or microembolic complications. Bennett et al. examined the pulmonary effects of massive autotransfusions versus banked blood transfusions in dogs and concluded that autotransfused blood caused fewer microembolic complications. Blood cultures obtained from mediastinal blood were positive after 1 week of incubation in 5 to 25 per cent of samples. Blood cultures obtained from patients following transfusion of this blood were negative and none of the patients who received culture-positive blood demonstrated evidence of clinical infection. The incidence of postoperative systemic and wound infections is no greater in patients receiving mediastinal blood rather than homologous blood. Four per cent of blood cultures from blood bank units have been reported as positive. Although two series, a total of 121 patients, found no benefits of the autotransfusion system, several larger series which studied more than 700 patients demonstrated its efficacy in blood conservation.

The withdrawal of one or more units of autologous blood prior to the initiation of cardiopulmonary bypass for reinfusion following bypass is a common method of blood conservation. Reports on the efficacy of this procedure in decreasing homologous blood requirements are conflicting. Several studies have reported significant decreases in both banked blood requirements and blood loss after bypass; however control groups were not always well matched. Others found no difference in blood requirements compared with well-matched controls, or in the quantity of postoperative bleeding. One study actually reported an increase in banked blood requirements in patients receiving autologous rather than homologous blood.

Although studies have shown that intraoperative ischaemia is common in patients undergoing coronary revascularization, the prognostic significance of these episodes is controversial. In 1518 patients undergoing coronary artery bypass grafting, the occurrence of prebypass ECG evidence of ischaemia correlated with the incidence of postoperative myocardial infarction. Another study found no relationship between either preoperative or prebypass ECG ischaemia with postoperative outcome in 50 patients undergoing coronary artery bypass grafting. The authors noted that postbypass ECG changes suggestive of ischaemia were also common (40 per cent of patients). Leung et al. examined the prognostic significance of intra-operative ischaemia as defined by new regional wall motion abnormalities detected by transoesophageal echocardiography. Although neither prebypass regional wall motion abnormalities nor ECG ischaemia correlated with poor outcome, postbypass ischaemia transoesophageal echocardiography was predictive of adverse events. A new regional wall motion abnormality detected by echocardiography is a more sensitive and specific indicator of ischaemia than ECG changes.

Patients with coronary artery disease are managed medically with &bgr;-adrenergic blocking drugs, calcium entry blocking drugs, and nitrates, often in combination. Several studies have examined the effects of preoperative medications on the incidence of intraoperative myocardial ischaemia. In all studies, preoperative &bgr;-blockade is associated with a significantly lower incidence of intraoperative ischaemia compared to therapy with calcium entry blocking agents. Intraoperative ECG evidence of ischaemia was seen in 34 per cent of 444 patients receiving &bgr;-blocking agents, and in 53 per cent of those not receiving these drugs. Tachycardia (heart rate > 109) was more common in patients not receiving &bgr;-blockers and doubled the incidence of perioperative ischaemia. Other studies have confirmed that tachycardia is the haemodynamic change most commonly associated with intraoperative ischaemia; the majority of intraoperative ischaemic events are not associated with any change in haemodynamics.

All preoperative anti-ischaemic agents should be continued in patients undergoing coronary revascularization. &bgr;-Adrenergic blocking drugs are associated with a significantly lower incidence of perioperative ischaemia, although the prognostic significance of pre- and postbypass ischaemia remains uncertain.

Invasive monitoring during cardiac surgery consists of an arterial catheter and either a central venous catheter or a pulmonary artery catheter. Standard non-invasive monitoring includes 5 lead ECG, pulse oximetry, temperature, and capnography. Transoesophageal echocardiography is a new form of non-invasive monitoring which provides continuous two-dimensional echocardiographic images from a transducer positioned in the oesophagus. This major advance in intraoperative monitoring allows assessment of left ventricular function and detection of myocardial ischaemia.

Controversy exists over the need to place a pulmonary artery catheter in all patients undergoing cardiac surgery. Recent studies have disputed its claim to be the most sensitive and specific early indicator of myocardial ischaemia. In a study comparing 12 lead ECG transoesophageal echocardiography, and pulmonary capillary wedge pressure monitoring in the detection of intraoperative ischaemia, an increase in pulmonary capillary wedge pressure of at least 3 mmHg had a sensitivity of 25 per cent and a positive predictive value of only 15 per cent. Other studies confirm these findings. Advantages of placing a pulmonary artery catheter include the ability to assess pharmacological and fluid therapy by the measurement of filling pressures and cardiac output, and the ability to calculate such parameters as systemic and pulmonary vascular resistance and left ventricular stroke work index. In patients with abnormal right ventricular function or pulmonary hypertension, a pulmonary artery catheter has significant advantages over a central venous catheter, which is not a sensitive monitor in these situations. The principal disadvantages of the pulmonary artery catheter are cost and a low incidence of complications such as pulmonary artery rupture and pulmonary infarction. Although a study in 1094 patients concluded that a pulmonary artery catheter did not favourably alter outcome, serious flaws included a lack of patient randomization, differences in anaesthetic techniques, greater use of pressors in patients electively receiving a pulmonary artery catheter, older age and higher risk categories in patients receiving catheters, and a higher incidence of reoperation in these patients. The effects of the monitoring alone cannot, therefore, be determined. Insertion of a pulmonary artery catheter is preferable to a central venous catheter alone in patients with poor ventricular function, unstable angina, recent infarction, valvular abnormalities, and in those undergoing emergency operations. For stable patients with normal left ventricular function and good distal vessels who are expected to have short cardiopulmonary bypass times, a central venous catheter is adequate.

Since 1982, two-dimensional transoesophageal echocardiography has become increasingly popular for intraoperative monitoring of patients with cardiac disease. The apparatus consists of a crystal mounted on the tip of a gastroscope. This is inserted in the oesophagus under direct vision following anaesthetic induction and intubation (Fig. 1) 1662. Transoesophageal echocardiography allows continuous monitoring of a two-dimensional image of the cardiac structures during surgery. Left ventricular function, volume, and wall motion abnormalities are assessed from the mid-papillary muscle short-axis view of the left ventricle (Fig. 2) 1663. This view is considered optimal because all three major coronary arteries supply a portion of the ventricle at this level.

Numerous studies have reported the superiority of two-dimensional echocardiography over other techniques for the detection of intraoperative ischaemia. Pulse wave and colour flow Doppler capabilities allow evaluation of cardiac output, valvular function, and intracardiac shunts. Transoesophageal echocardiography is considered essential in many centres to detect the presence of residual mitral regurgitation or paravalvular leaks following valve repairs and replacements. The technique is highly sensitive for detecting intracardiac air for the assessment of prosthetic mitral valves. Newer models of the system contain biplane probes, which allow imaging in two planes. This modification allows the evaluation of thoracic aortic aneurysms and improves the assessment of regurgitant lesions. Transoesophageal echocardiography systems with continuous wave Doppler capabilities are currently under investigation. Paediatric-sized probes are also available.

Transoesophageal echocardiography is considered a safe procedure. Early reports documented temporary vocal cord paralysis in patients undergoing sitting craniotomies; this is not a problem in supine patients. A study in dogs found no evidence of either gross or microscopic oesophageal damage after use of the probe. Oesophageal pathology such as varices, prior oesophageal surgery, or the presence of an oesophageal stricture is considered to be a contraindication to placement of the probe.

Limitations to the intraoperative use of transoesophageal echocardiography include the inability to obtain images prior to and during anaesthetic induction, the ability to image in only a single plane in most systems, the requirement for additional training of the anaesthesiologist in echocardiography, and the significant expense of purchasing the equipment.

In the early days of cardiac surgery, the only anaesthetics available were the inhalational agents, principally halothane. In 1969, Lowenstein published a report on the use of morphine in doses of 0.5 to 3 mg/kg as the principal anaesthetic for patients undergoing aortic valve replacement. Patients with poor ventricular function had better haemodynamics compared with those treated by earlier techniques using inhalational agents. Morphine causes histamine release and arterial and venous dilation, however; there was a high incidence of problems such as hypotension and increased fluid requirements and a high incidence of intraoperative recall. With the introduction of fentanyl, a highly lipid soluble, rapidly acting synthetic narcotic, morphine was largely abandoned. The first report of the use of high-dose fentanyl for patients undergoing cardiac surgery was published in 1978.

Fentanyl is approximately 80 times as potent as morphine, and is used in doses of 50 to 100 &mgr;g/kg. Another synthetic opioid, sufentanil, has also become popular as an anaesthetic agent. It is approximately 5 to 10 times as potent as fentanyl, has a slightly more rapid onset, and has a shorter duration of action due to greater lipid solubility. Unlike morphine, fentanyl and sufentanil do not cause histamine release and, in clinical doses, neither is a myocardial depressant. Their effects on the peripheral vasculature remain controversial. Fentanyl has been reported to have no effect or to decrease peripheral vascular resistance in humans. In an isolated canine hindlimb preparation, the effects were dose-dependent, and only reached statistical significance at the highest doses (50 &mgr;g/kg). The synthetic opioids cause a dose-dependent bradycardia, mediated by the vagus nerve. Severe bradycardia or asystole has been reported following administration of the muscle relaxant vecuronium in combination with narcotics, particularly sufentanil. This problem can be minimized by using the muscle relaxant pancuronium which is vagolytic. High-dose, and occasionally low-dose narcotics cause chest wall rigidity, which can be severe enough to make ventilation impossible. This may be due to central nervous system effects although the precise mechanism is unknown. The incidence and severity of this problem may be decreased by the prior administration of a small dose of a nondepolarizing muscle relaxant. All opioid agonists produce dose-dependent respiratory depression through a direct effect on the brain-stem respiratory centres. At the doses used in cardiac surgery, this generally results in a need for postoperative mechanical ventilation. Sufentanil has been reported to allow earlier extubation than fentanyl.

Fentanyl is extensively metabolized to inactive metabolites by the liver with a high hepatic extraction ratio, indicating that metabolism is perfusion-dependent: anything that decreases hepatic blood flow will therefore decrease the rate of metabolism. Metabolites are excreted by the kidney. The elimination half-life is 185 to 219 min, but this is significantly prolonged in elderly patients (945 min). Sufentanil also undergoes hepatic metabolism with a high hepatic extraction ratio: however, some metabolites are active. The elimination half-life is 148 to 164 min. Metabolites are excreted in the urine and faeces. Prolonged respiratory depression has been reported in patients with renal failure. Although large doses of narcotics cause progressive EEG slowing and unconsciousness, intraoperative recall has been reported. The use of amnestic agents such as scopolamine or benzodiazepines as premedication, and intraoperative supplementation is advocated.

The choice of anaesthetic technique for patients undergoing cardiac surgery depends on numerous factors, including the experience and preference of the anaesthesiologist: no data support the superiority of one technique over any other. The goal of the anaesthesiologist is to provide unconsciousness, analgesia, amnesia, and suppression of the neuroendocrine response to surgery. The advantages of inhalational agents such as halothane, enflurane, or isoflurane, include the ability to increase or decrease the dose at will, a decrease in myocardial oxygen consumption as a result of drug-induced myocardial depression, and the lack of prolonged effect which allows extubation whenever it is deemed medically appropriate. Disadvantages include myocardial depression, which may not be tolerated in patients with poor ventricular function, a greater incidence of hypotension and haemodynamic variability than occurs with opioids, the possibility of coronary steal with isoflurane, and a lack of postoperative analgesia. The advantages of high-dose narcotics include ease of administration, haemodynamic stability, even in patients with depressed ventricular function, the provision of a basal level of anaesthesia at all times during the procedure, and extension of analgesia into the postoperative period. Disadvantages include the requirement for postoperative mechanical ventilation, an inconsistent ability to block the hypertensive response to periods of surgical stimulation, and the suggestion that there is a ceiling effect beyond which additional drug is of no benefit. In patients with good ventricular function, it is frequently necessary to supplement the narcotic with low doses of inhalational agents in order to control hypertension and provide stable haemodynamics. High doses of narcotics may induce an acute tolerance. Higher doses have been advocated (above 100 &mgr;g/kg of fentanyl or 20–30 &mgr;g/kg of sufentanil) in order to block hypertensive responses intraoperatively. However, a recent study was unable to find any dose that blocked haemodynamic and endocrine responses in all patients. In other words, there may be no achievable plasma level that consistently provides complete anaesthesia in all patients. Extremely high doses of narcotics, therefore, have little advantage, and some patients will require the addition of either inhalational agents or vasodilators to achieve haemodynamic stability.

The choice of fentanyl or sufentanil also largely depends on personal preference. Although early reports suggested that sufentanil provided greater haemodynamic stability than fentanyl, there are no convincing data to support the superiority of one opioid over another. Proponents of sufentanil argue that it allows earlier extubation. This effect is dose-dependent and patients who receive larger doses may require the same duration of postoperative ventilation as those given fentanyl. One significant difference between the two is cost: sufentanil is significantly more expensive than fentanyl.

In summary, although high-dose narcotics are probably the most popular anaesthetic drugs used in the United States for patients undergoing cardiac surgery, the choice of technique should be based on the patient population and local practices of intensive care management. In institutions where all patients are ventilated for the first postoperative night there is no major advantage, and potential disadvantages, to using pure inhalational techniques. If early extubation is desired, the combination of moderate dose narcotic plus inhalational agents or inhalational agents alone may be preferable. In unstable patients or patients with poor ventricular function, pure narcotic anaesthesia is superior because of the lack of myocardial depression and other undesirable haemodynamic effects. In general, all patients will tolerate a careful, controlled induction with fentanyl without significant difficulty.

As discussed earlier, because narcotics alone can be associated with hypertension and tachycardia during surgical stimulation in patients with normal left ventricular function, it became the practice to use a large dose of fentanyl as a background anaesthetic, with the addition of low to moderate doses of inhalational agents as a supplement to control haemodynamic parameters during these critical events. This technique was successful in producing a smooth, controlled prebypass interval and the agents used include all of the three available: halothane, enflurane, and isoflurane.

Isoflurane controls blood pressure principally as a result of direct vasodilation, while enflurane acts as a negative inotropic agent. Isoflurane therefore became a popular choice as the supplement to narcotics for patients undergoing revascularization. However, in 1983 Reiz et al. published a study of 21 patients undergoing coronary artery bypass grafting and concluded that isoflurane was a direct coronary vasodilator and had caused coronary ischaemia in 10 or 21 patients as a result of a coronary steal phenomenon.

Since that time enormous controversy has been generated over the use of isoflurane in patients with coronary artery disease undergoing both cardiac and non-cardiac operations. Numerous studies in humans and animals have produced varying results. Moffitt et al. measured coronary sinus flow and oxygen content in humans and concluded that isoflurane dilated coronary vessels to a significantly greater degree than equipotent doses of halothane and enflurane. In contrast, Tarnow et al. also studied human beings and found that isoflurane actually protected patients against pacing induced ischaemia compared with other agents. Buffington et al. demonstrated coronary steal with isoflurane compared with halothane in a canine model, using a microsphere technique to measure coronary blood flow. Although equipotent anaesthetic doses of the agents were used, the resulting haemodynamics between groups were significantly different. Halothane reduced mean perfusion pressure from the control of 55 mmHg to 51 mmHg, while isoflurane decreased mean pressure to 41 mmHg: the differing pressures may have obscured the effects of the two agents. Although this is a significant flaw, it is the only study to demonstrate coronary steal directly. However, it is not analogous to the clinical situation where inhalational agents are titrated against haemodynamic effects. In in-vitro preparations using isolated segments of coronary artery, neither halothane nor isoflurane are direct vasodilators.

Two large clinical studies evaluated outcome in coronary artery bypass surgery using various anaesthetic agents. Slogoff and Keats prospectively randomized 1012 patients to receive enflurane, halothane, isoflurane, or sufentanil. There was no difference between groups in the incidence of prebypass ischaemia (28–33.5 per cent), postoperative myocardial infarction (3.6–4.7 per cent), or death (1.2–2.4 per cent). Hypotension occurred twice as often in patients receiving an inhalational agent compared to sufentanil, and hypertension was twice as common in patients given sufentanil. The incidence of tachycardia (heart rate>110 beats/min) was no different with different agents. Tachycardia was the only haemodynamic abnormality associated with a significantly increased incidence of ischaemia: 48 per cent of patients with tachycardia became ischaemic compared to 29.5 per cent of those with no haemodynamic abnormalities (p<0.05). Tuman et al. prospectively studied 1094 patients undergoing coronary revascularization who received fentanyl, sufentanil, diazepam plus ketamine, or halothane. During the intravenous anaesthesia, halothane, enflurane, or isoflurane were used as a supplemental anaesthetic. There was no difference between groups in the incidence of postoperative mycardial infarction, low cardiac output state, or death. Thus, no clinical studies have demonstrated any difference in outcome in patients with coronary artery disease receiving isoflurane rather than other inhalational or intravenous agents. In summary, although there is some experimental evidence to support the concept of isoflurane-induced coronary steal, the data are often contradictory and no study shows it is a clinically relevant phenomenon. Because isoflurane maintains cardiac output compared with the other inhalational agents which are potent mycardial depressants, isoflurane continues to be commonly used as an anaesthetic agent in patients with coronary artery disease.

The subject of cerebral protection is complex and controversial: there are many different models of cerebral ischaemia, and the various protective agents are not generally beneficial in all models. Currently, there are five major areas of interest: hypothermia, barbiturates, isoflurane, nimodipine, and glucose control.

The various models of cerebral ischaemia can be defined as shown in Table 1 483. Focal ischaemia results when portions of the cerebral vascular supply are occluded by emboli, stroke, or a space-occupying lesion. Global ischaemia can be complete or incomplete. Incomplete global ischaemia, where some delivery of O&sub2; continues, occurs as a result of shock or hypotension. Complete global ischaemia occurs following cardiac arrest.

One of the methods nearly certain to provide cerebral protection during times of decreased energy supply is to decrease O&sub2; demand: hypothermia is probably the most effective means of achieving this goal. There is a direct relationship between cerebral metabolic rate and temperature. For every 10°C decrease in temperature, the cerebral metabolic rate decreases by a factor of 2.2: at 28°C, it is approximately 45 per cent of normal. At 18°C the EEG is flat and the cerebral metabolic rate is 20 per cent of normal. Hypothermia decreases the rate of all brain activity, both electrical and metabolic. Profound hypothermia to temperatures of 15 to 18°C with circulatory arrest is used in the repair of complex congenital lesions in children and in adults undergoing aortic arch aneurysm repairs. The incidence of neurological complications varies, but it has been reported to be less than the incidence of focal neurological deficits after cardiopulmonary bypass for coronary artery bypass grafting. In most clinical situations of cerebral ischaemia, however, hypothermia cannot be used because major circulatory support is required.

Barbiturates cause cerebral vasoconstriction, can decrease intracranial pressure, and decrease the cerebral metabolic rate by as much as 50 or 60 per cent by decreasing cerebral electrical activity. Barbiturates are, therefore, only effective when electrical activity is present. Following cardiac arrest, the EEG becomes isoelectric within 20 to 30 s. Although an early primate study suggested that barbiturates were protective following cardiac arrest, other animal studies do not support this and a randomized prospective study failed to show any benefit of barbiturates after cardiac arrest in humans.

Barbiturates are effective in incomplete global or focal ischaemia: animal studies show that barbiturates, given before or during an ischaemic event, can prolong the brain's tolerance to the insult resulting in a smaller infarct. There are fewer data available in humans, but Nusseier et al. reported an improvement in neuropsychiatric function in patients who receive an average dose of 40 mg/kg of thiopental to produce EEG suppression on bypass during cardiac procedures requiring an open ventricle. Patients receiving thiopental required more vasopressor support at the termination of bypass, and experienced delayed extubation. One problem with this study is that normothermic bypass was used, which is not common for these procedures. Intraoperative barbiturate therapy is currently considered to be useful in situations where focal ischaemia may occur, such as carotid endarterectomy, cerebral bypass procedures, or intracerebral aneurysm clipping. Dose regimens vary among studies; however EEG monitoring allows titration of thiopental to EEG suppression.

Most general anaesthetics decrease the cerebral metabolic rate by decreasing the electrical activity of the brain. Inhalational agents increase cerebral blood flow, which can be detrimental in the presence of increased intracerebral pressure. Isoflurane is the only inhalational agent which produces an isoelectric EEG at approximately 2.4 per cent. A study in humans undergoing carotid endarterectomy evaluated the effects of isoflurane on critical cerebral blood flow, defined as the blood flow at which ischaemic EEG changes occur (estimated at 18–24 ml/100 g/min in humans). Isoflurane reduced the critical blood flow to less than 10 ml/100 g/min. However, a study of focal ischaemia in primates reported a better neurological outcome in animals anaesthetized with thiopental rather than isoflurane, although there were differences in haemodynamic measurements as a result of the two agents. In a study of induced hypotension in dogs, isoflurane was superior to trimethaphan, nitroprusside, or halothane for cerebral protection. Although not all studies are in agreement, there is good evidence that isoflurane provides cerebral protection during focal and incomplete ischaemia as a result of metabolic suppression, and it may be the anaesthetic of choice for patients in whom hypotension needs to be induced.

Ischaemia disrupts intracellular calcium homeostasis. There is accumulating evidence that the calcium entry blocking drug nimodipine provides cerebral protection in a variety of circumstances. Allen et al. reported an improvement in neurological outcome in patients with subarachnoid haemorrhage treated with nimodipine compared with controls in a double-blind randomized study. Gelmers et al. performed a prospective, double-blind, randomized, placebo-controlled trial of nimodipine in patients who had suffered an acute stroke. The nimodipine-treated groups experienced a significant reduction in mortality from all causes and a better neurological outcome. Work in primates suggest an improvement in neurological outcome with nimodipine following 17 min of complete cerebral ischaemia. Thus, nimodipine may be effective in preventing damage following cerebral ischaemia in a variety of clinical situations, although the precise mechanism of protection is unknown. There are currently no data on the perioperative use of nimodipine in patients at risk for cerebral ischaemia.

Clinical and laboratory date suggest that glucose may play a role in cerebral ischaemia. In the absence of O&sub2;, glucose is metabolized to lactic acid, which may increase tissue damage. Studies in animals have shown that elevated glucose levels during cerebral ischaemia are associated with a worse neurological outcome. In a primate study, animals were given D51/2NS prior to a period of complete cerebral ischaemia. Control animals received lactated Ringer's solution. There was no difference in serum glucose levels between groups; however; the glucose-treated monkeys suffered more severe neurological compromise. Patients with hyperglycaemia following ischaemic stroke had a poorer neurological outcome than those with normal serum glucose. Current recommendations are to avoid the intraoperative use of glucose-containing solutions, and to monitor serum glucose closely in patients at risk for cerebral ischaemia.

The role of these agents in cerebral protection in the three models of cerebral ischaemia is summarized in Table 2 484. Current intraoperative cerebral protection includes the use of the anaesthetic agents thiopental and isoflurane, with EEG monitoring and glucose control. Few interventions are effective following cardiac arrest, with the possible exception of nimodipine. Clinical studies on the effectiveness of nimodipine for cerebral protection are currently underway.

Abramson NS, et al. Randomized clinical study of thiopental loading in comatose survivors of cardiac arrest. N Engl J Med 1986; 314: 397–403.

Alter MJ, et al. Sporadic non-A, non-B hepatitis: frequency and epidemiology in an urban U.S. population. J Infect Dis 1982: 145: 886.

Breyer RR, Engelman RM, Rousou JA, Lemesshow S. Blood conservation for myocardial revascularization. J Thoracic Cardiovasc Surg 1987; 93: 512–22.

Chung F, et al. Calcium channel blockade does not offer adequate protection from perioperative myocardial ischemia. Anesthesiology 1988; 69: 343–7.

Cohn LH, Fosberg AM, Anderson WP, Collins JJ. The effects of phlebotomy, hemodilution and autologous transfusion on systemic oxygenation and whole blood utilization in open heart surgery. Chest 1975; 68: 283–7.

Committee on Transfusion and Transplantation. General Principles of Blood Transfusion. Chicago: American Medical Association, 1970–17.

Consensus Conference. Perioperative red blood cell transfusion. JAMA 1988; 260: 2700–3.

Cosgrove DM, et al. Determinants of blood utilization during myocardial revascularization. Ann Thoracic Surg 1985; 40: 380–4.

Cuello L, Vazquez E, Rios R, Raffucci FL. Autologous blood transfusion in thoracic and cardiovascular surgery. Surgery 1967; 62: 814.

deLange S, Boscoe MJ, Stanley TH. Comparison of sufentanil–O&sub2; and fentanyl–O&sub2; for coronary artery surgery. Anesthesiology 1982; 56: 112.

Gelmers HJ, Gorter K, deWeerdt CJ, Wiezer HJA. A controlled trial of nimodipine in acute ischemic stroke. N Engl J Med 1988; 318: 203–7.

Goodnough LT, et al. Increased preoperative collection of autologous blood with recombinant human erythropoietin therapy. N Engl J Med 1989; 321: 1163–8.

Gravlee GP, Ramsey RM, Roy RC, Angert KC, Rogers AT, Pauca AL. Rapid administration of a narcotic and neuromuscular blocker: a hemodynamic comparison of fentanyl, sufentanil, pancuronium, and vecuronium. Anesth Analg 1988; 67: 39–47.

Haggmark S, et al. Comparison of hemodynamic, electrocardiographic, mechanical and metabolic indicators of intraoperative myocardial ischemia in vascular surgical patients with coronary artery disease. Anesthesiology 1988; 70: 19–25.

Hallowell P, Bland JH, Chir B, Buckley MJ, Lowenstein ELK. Transfusion of fresh autologous blood in open-heart surgery. J Thoracic Cardiovasc Surg 1972; 64: 941–8.

Honig CL, Bove JR. Transfusion associated fatalities: review of Bureau of Biological reports. Transfusion 1980; 20: 653.

Howie MB, McSweeney TD, Lingam RP, Maschke SP. A comparison of fentanyl–O&sub1; and sufentanil–O&sub2; for cardiac artery surgery. Anesthesiology 1982; 56: 112–8.

Howie MB, et al. A comparison of fentanyl–O&sub2; and sufentanil–O&sub2; for cardiac anesthesia. Anesth Analg 1985; 64: 877.

Johnson RG, Rosenkrantz KR, Preston RA, Hopkins C, Daggett WM. The efficacy of posteropative autotransfusion in patients undergoing cardiac operations. Ann Thoracic Surg 1983; 36: 173–9.

Knight AA, et al. Perioperative myocardial ischemia: importance of the preoperative ischemic pattern. Anesthesiology 1988; 68: 681–8.

Lanier WL, Stangland KJ, Scheithauer BW, Milde JH, Michenfelder JD. Effects of IV dextrose and head position on neurologic outcome after complete cerbral ischemia. Anesthesiology 1985; 63: A110.

Leung JM, O'Kelly B, Browner WS, Tubau J, Hollenberg M, Mangano DT. Prognostic importance of postbypass regional wall-motion abnormalities in patients undergoing coronary artery bypass surgery. Anesthesiology 1989; 71: 16–25.

Love TR, Hendren WG, O'Keefe DD, Daggett WM. Tranfusion of pre-donated autologous blood in elective cardiac surgery. Ann Thoracic Surg 1987; 43: 508–12.

McCarthy PM, et al. Effect of blood conservation efforts in cardiac operations at the Mayo Clinic. Mayo Clin Proc 1988; 63: 2250–9.

Messick JM, Casement B, Sharbrough FW, Milde LN, Michenfelder JD. Correlation of regional cerebral blood flow (rCBF) with EEG changes during isoflurane anesthesia for carotid endarterectomy: critical rCBF. Anesthesiology 1987; 66: 344–9.

Messick JM, Milde LN. Brain protection. Adv Anesthesiol 1987; 4: 47–88.

Moffitt E, et al. Myocardial metabolism and hemodynamic responses with isoflurane anesthesia for coronary arterial surgery. Anesth Analg 1986; 65: 53–61.

Nussmeier NA, Slogoff S. Neuropsychiatric complications after cardiopulmonary bypass: cerebral protection by a barbiturate. Anesthesiology 1986; 64: 164–70.

Ochsner JL, Mills NL, Leonard GL, Lawson N. Fresh autologous blood transfusions with extracorporeal circulation. Ann Surg 1973; 177: 811–7.

Owings DV, Kruskall MS, Thurer RL, Donovan LM. Autologous blood donations prior to elective cardiac surgery. Safety and effect on subsequent blood use. JAMA 1989; 262: 1963–8.

Philbin DM, Rosow CE, Schneider RC, Koski G, D'Ambra MN. Fentanyl and sufentanil anesthesia revisited: how much is enough? Anesthesiology 1990; 73: 5–11.

Roizen MF et al. Monitoring with 2D transesophageal echocardiography. Comparison of myocardial function in patients undergoing supraceliac, suprarenal-infraceliac, or infraceliac aortic occlusion. J Vasc Surg 1984; 1: 300–5.

Sanford TJ, et al. A comparison of morphine, fentanyl, and sufentanil for cardiac surgery: induction, emergence and extubation. Anesth Analg 1986; 65: 259.

Shaw PJ, et al. Early neurological complications of coronary artery bypass surgery. Br Med J 1985; 291: 1384–7.

Slogoff S, Keats AS. Does chronic treatment with calcium entry blocking drugs reduce perioperative myocardial ischemia? Anesthesiology 1988; 68: 676–80.

Smith PC, et al. Cerebral consequences of cariopulmonary bypass. Lancet 1986; I–II: 823–5.

Slogoff S, Keats AS. Randomized trial of primary anesthetic agents on outcome of coronary artery bypass operations. Anesthesiology 1989; 70: 179–88.

Smith JS, et al. Intraoperative detection of myocardial ischemia in high risk patients: electrocardiography vs. 2D transesophageal echocardiography. Circulation 1985; 72: 1015–21.

Starr NJ, Sethna DH, Estafanous RG. Bradycardia and asystole following the rapid administration of sufentanil with vecuronium. Anesthesiology 1986; 64: 521–3.

Thurer RL, Lytle BW, Cosgrove DM, Loop FD. Autotransfusion following cardiac operations: a randomized, prospective study. Ann Thoracic Surg 1979; 27: 500–7.

Tinker JH. Inhalation vs. opioid anesthesia for cardiac surgery. International Anesthesia Research Society, 1990 Review Course Lecture, 1990: 67–73.

Tuman KJ, McCarthy RJ, Spiess BD, DaValle M, Dabir R, Ivankovich AD. Does choice of anesthetic agent significantly affect outcome after coronary artery surgery? Anesthesiology 1989; 70: 189–98.

Tyson GS, Sladen RN, Spainhour V, Savitt MA, Ferguson TB, Wolfe WG. Blood conservation in cardiac surgery. Ann Surg 1989; 209:736–42.

Van Daele M, et al. Do changes in pulmonary capillary wedge pressure adequately reflect myocardial ischemia during anesthesia? A correlative preoperative hemodynamic, electrocardiographic, and transesophageal echocardiographic study. Circulation 1990;81: 865–71.

Ward JW, et al. Transmission of human immunodeficiency virus (HIV) by blood transfusions screened as negative for HIV antibody. N Engl J Med 1988; 318: 473–8.

Wohlgelernter D, et al. Regional myocardial dysfunction during coronary angioplasty: evaluation by 2D echocardiography and 12 lead electrocardiography. J Am Coll Cardiol 1986;7: 1245–54.