Device Overview

Gravity drainage usually allows collection of blood from the venous circulation into the venous reservoir. Through separate inflows on the reservoir, blood can also be returned from the pericardial well (cardiotomy blood) and from cardiac and aortic vents. A centrifugal or roller pump is used to divert the venous reservoir blood to the oxygenator. After the blood passes through the integrated heat exchanger and oxygenator, it is circulated through an arterial filter and bubble trap and returned to the patient through the arterial cannula (Fig. 62-2). Cardioplegia setups are often intimately incorporated into the CPB circuit.

Figure 62–2 A, Overview of the cardiopulmonary bypass circuit. B, Typical integrated heat exchanger–membrane oxygenator.

Principles of Current Oxygenator Design and Function

Diffusion of gases at the blood–membrane interface can be predicted by Fick’s law, which says that the rate of diffusion is proportional to the partial pressure gradient of the gas in the direction of diffusion. The rate of gas transfer is also inversely proportional to the distance through which it must pass (the thickness of the membrane), and it depends on the property of diffusivity of the membrane biomaterial. Currently used membranes are very permeable to O₂, but they are often less permeable to CO₂. The problem of poor diffusion of CO₂ was solved by the introduction of microporous membranes. These surfaces allow transient direct blood–gas interfacing at pore structures smaller than blood cells; however, the hydrophobic nature of the membrane results in changes in blood surface tension, blocking direct contact between the two phases. As a result, the interface in microporous membranes behaves like a very thin stagnant film of plasma water that offers little resistance to gas exchange. Progressive protein accretion at the pores occurs over time, resulting in a finite functional capacity, detected by worsening efficiency in gas transfer, which is why oxygenators must be replaced with long CPB runs (Fig. 62-3).

Figure 62–3 Scanning electron micrograph of a microporous membrane (Celgard) used in a membrane oxygenator (×20,000).

(Courtesy of Membrana GmbH, Germany.)

As opposed to the relatively facile transfer of O₂ gas through a membrane, when the O₂ is dissolved in plasma or blood, its diffusivity is 25 times less thanthat of CO₂. Two modifications have been introduced to overcome this problem. First, the path length (the distance the blood travels as it passes the gas-exchange surface) has been maximized. This modification is limited by the parallel need to increase the priming volume. Second, disturbed flow patterns are used to promote mixing and to bring deoxygenated blood closer to the exchange surface. Mechanisms that interfere with the development of fully developed flow patterns enhance diffusion by keeping the boundary layer narrow. In an oxygenator, this can be accomplished by making the surface irregular or by positioning elements in the flow stream to disrupt smooth flow and to enhance mixing. This is the rationale for the extraluminal hollow fiber design.

Hypothermia and Acid–Base Balance

The feasibility and applicability of hypothermia for heart surgery was first suggested by Bigelow and colleagues in 1950.⁴ They demonstrated the safety benefit of hypothermia as a means to decrease O₂ consumption during inflow occlusion in an animal model. O₂ consumption decreases by 50% for every 10° C drop in temperature (Fig. 62-4). Lower flows decrease collateral flow and rewarming of the heart from contact with adjacent tissues, and they provide a margin of safety if equipment fails.

Figure 62–4 The effect of hypothermia on oxygen consumption (VO₂).

Hypothermia is associated with marked changes in blood pH and PCO₂ levels. With a rise in temperature, CO₂ becomes less soluble in blood, and there is a greater tendency for the dissolved CO₂ to come out of solution (increased gaseous phase). Thus, CO₂ solubility increases with decreasing temperature. When blood from a hypothermic patient (e.g., at 24° C) is introduced into a CO₂ electrode for measurement, it is first warmed to 37° C (increasing the amount of CO₂ in the gaseous phase), and therefore the measured partial pressure of CO₂ is higher than it was in reality at the cooler temperature. The correction for PCO₂ is based on a calculated decrease of 4.5% per degree Celsius. On the other hand, the pH increases 0.015 units per degree Celsius drop in temperature. This shift in pH is partially related to the influence of buffers such as the imidazole moiety of histidine, but it is also related to the dynamics of the Henderson-Hasselbalch equation, pH = pK + [HCO₃⁻]/(0.03 × PCO₂).

Because alkalosis and hypocarbia trigger a decrease in cerebral blood flow (CBF), some investigators have suggested the addition of CO₂ during hypothermia to compensate and to keep the pH unchanged (pH-stat strategy). In 1987, Murkin and coworkers⁵ confirmed that pH-stat management results in a greater ratio of CBF to cerebral metabolic rate (CMRO₂) than alpha-stat management (i.e., no active correction of pH with hypothermia).⁵ Most agree that a pH-stat strategy is probably preferable in children, where increased CBF does increase the rate of cooling and thus increases the chance of achieving uniform cerebral hypothermia.⁶ The rate of brain oxygen depletion during deep hypothermic circulatory arrest is also considerably slower with pH-stat strategy. As a result, pH-stat substantially prolongs the interval between the onset of arrest and the exhaustion of brain oxygen stores and may be associated with better clinical outcomes in this group.⁷

In contrast, in adults, an alpha-stat strategy may provide greater cerebral protection during hypothermia on CPB. Alpha-stat management is certainly easier to accomplish, and a justification is that most cellular mechanisms are capable of maintaining intracellular pH despite fluctuations in extracellular conditions. Because excess CO₂ is added in a pH-stat strategy, brain acidosis may occur during rewarming, and, combined with decreased O₂ delivery after CPB, this may augment CNS injury in adults. Most importantly, as mentioned, a pH-stat strategy results in excessive CBF, which may increase embolic load. Finally, pH-stat strategy is associated with a decreased ability to maintain autoregulation at low pressures. Three randomized controlled trials have demonstrated a small but present benefit of alpha-stat strategy in terms of neurologic and neurocognitive outcome in adults, especially when CPB time exceeded 90 minutes.⁸

Hematocrit and Priming

The 1960s saw the introduction of crystalloid (dextrose 5% in water, or D5W) in prime as an alternative to routine whole blood prime. An increase in efficiency of oxygenation and a decrease in end-organ complications were seen with this approach.⁹ Now hemodilution is commonly practiced in virtually all adult and pediatric cardiac surgeries, and the hematocrit (Hct) is maintained between 20% and 25% during CPB. Hemodilution has a major effect on blood viscosity, primarily at the capillary level where the radius is small and the shear rate is low. Low flow at the capillary level increases viscosity of the blood, further increasing the resistance, but this effect is counter-balanced by the effect of hemodilution.

The drop in O₂ content at most levels of moderate hemodilution is more than compensated for by augmented cardiac output (CO), so total O₂ delivery (CO × O₂ content of blood) is increased; the optimal Hct is probably in the range of 30% for this effect. Hemodilution results in a significant increase in flow without a parallel increase in perfusion pressure. Hypothermia complicates the effects of hemodilution, as the decreased temperature causes increased viscosity and induces vasoconstriction.

The optimal Hct for CPB remains a topic of significant controversy. A post-hoc assessment of the impact of low Hct on admission of cardiac patients to the intensive care unit (ICU) by Spiess and colleagues¹⁰ found an inverse correlation between admission Hct and the risk of Q-wave myocardial infarction (MI). On the other hand, studies have demonstrated that an Hct of less than 20% may be associated with abnormal distribution of blood flow to organs, and that an Hct of less than 15% may lead to maldistribution of coronary flow away from the subendocardium in the presence of residual coronary stenosis.¹¹ Retrospective observational studies of consecutive patients undergoing isolated CABG have demonstrated that a lower minimum Hct is associated with a significantly increased risk of renal injury¹² and mortality.^13,14 As excessive hemodilution also increases CBF, it may cause a parallel increase in microembolization and thus may theoretically be a contributor to neurologic damage after CPB and the increased risk of stroke observed with lower Hct on CPB.¹⁵

Most centers prime the CPB circuit with a solution of balanced salt with or without the addition of a colloid solution such as pentastarch. Infants and children often need blood added directly to the prime, depending on the anticipated Hct after initiation of bypass (calculated from standard nomograms). Other additives to the prime may include calcium, mannitol, and pharmacologic agents such as heparin and aprotinin. The minimum priming volume is mandated by the patient’s circulating volume and the pump prime volume. Although the prime volume averages 1 L, experienced practitioners can achieve 200 mL in pediatric circuits.

Flow Rates, Perfusion Pressure, and Autoregulation

Flow is generally kept in the range of 2.2 to 2.5 L/min/m² to provide a margin of safety during CPB, as systemic blood flow distribution and O₂ consumption remain normal at this level. At normothermia, a target mean blood pressure of 50 to 70 mm Hg is generally used. The perfusionist can control the pressure by increasing or decreasing flow, or by the addition of vasoconstrictors or vasodilators (inhalational anesthetics). At lower temperatures, a mean pressure of 35 mm Hg is still generally accepted as safe. A number of investigators have recommended greater vigilance to prevent any hypotension (<45 to 50 mm Hg) during CPB,¹⁶ but these findings are controversial.¹⁷

Although the brain comprises only 2% of body weight, its metabolic needs demand 15% of the CO, extracting as much as 25% of the delivered O₂. Temperature is the most important element influencing CBF during CPB. As the temperature is dropped, the CMRO₂ decreases exponentially and the CBF decreases linearly (Fig. 62-5). As a consequence, the ratio of CBF to CMRO₂ increases, resulting in “luxuriant” flow, further facilitated by hemodilution.

Figure 62–5 The effect of hypothermia on cerebral metabolic rate (CMRO₂) and cerebral blood flow (CBF).

Autoregulation of CBF is also related to changes in perfusion pressure. At normothermia, a mean pressure of 50 mm Hg is the threshold at which the brain autoregulates flow, but with hypothermia (26° C), the threshold drops to 30 mm Hg (Fig. 62-6). At deep hypothermia (<20° C), there is a loss of pressure-flow autoregulation, as severe temperature reductions impair cerebral vascular relaxation,¹⁸ and changes in cerebral perfusion pressure alone result in corresponding proportional changes in CBF. Other factors that influence CBF and CMRO₂ include blood viscosity, intracranial pressure and central venous pressure (CVP), and the blood gas status (pH, PaCO₂, PaO₂). CBF varies linearly with the PaCO₂, in the range of 20 to 80 mm Hg, whereas a PaO₂ of less than 50 causes cerebral vasodilation, which overrides pressure-flow autoregulation.

Figure 62–6 Through autoregulation, cerebral blood flow (CBF) is constant from cerebral perfusion pressure (CPP) 40 to 140 mm Hg. Other factors that independently affect CBF include PaCO₂ and PaO₂.

(Modified with permission from Kelly BJ, Luce JM. Current concepts in cerebral protection. Chest 1993;103:1246-54. Copyright 1993, American College of Chest Physicians.)

Pumps for CPB

The two commonly used types of pumps for CPB involve roller and centrifugal fluid propulsion. Noninterrupted contact of the rollers with the tubing in the track results in the nonpulsatile nature of the flow (Fig. 62-7A). A low compression will result in inadequate flow, whereas excessive compression may aggravate hemolysis and tubing wear. Other complications associated with the use of the roller pump include cavitation caused by excess pressure, and spallation (the release of particles from the inner surface).

Figure 62–7 A, A double roller pump has rotating arms oriented at 180º like spokes on a wheel. Spool-shaped rollers are located at the ends of the arms. A length of tubing is locked inside a curved track at the outer circumference of a partial circle of 210 degrees. B, A typical centrifugal pump has a cone-shaped outer housing with an upper inlet and a single lower outlet. The inner chamber contains spinning concentric smooth cones or fins, mounted on a central impeller. Arrows indicate direction of blood flow.

Centrifugal pumps (see Fig. 62-7B) are a popular alternative to the roller pump, particularly in pediatric cardiac surgery and in cases with anticipated long CPB runs. The flow in a centrifugal pump is afterload dependent, and it is not predictable solely on the basis of the calculated revolutions per minute, so an in-line flowmeter is essential. Theoretically, its use should result in less blood trauma, particularly with prolonged CPB times, so this device is preferred for extracorporeal membrane oxygenation (ECMO). In children, the use of the centrifugal pump has been associated with decreased hemolysis, platelet activation, and inflammation and bleeding,¹⁹ although these findings have not been reproduced by other investigators.²⁰ Aside from their expense ($150 per case), these devices are susceptible to air locks, thus requiring extreme vigilance by the perfusionist. On the other hand, this may be protective, as there is less chance of pumping large volumes of air into the patient via an inadvertent leak, and the lines cannot be over-pressurized with distal obstruction, as occurs with roller pumps. These devices are not valved, and if rotation stops without clamping the outflow, rapid retrograde flow from the arterial line occurs in milliseconds, essentially exsanguinating the patient. Finally, roller pumps can be operated manually, but a power outage with a centrifugal pump can be a disaster.

Cardiotomy

Cardiotomy blood is the extravascular blood collected in the thoracic wound during CPB. In general, blood from the wound cavity is aspirated through a sucker device and transferred to a cardiotomy reservoir. Because of the close proximity of tissues in the wound and the potential for particulate matter to be collected, a filter and a defoaming chamber are incorporated into the cardiotomy reservoir. Mechanical injury, such as hemolysis, may result from the air–blood interface at the sucker, as well as from the compressive effects of the roller pump. Other mechanical complications from cardiotomy suction include the formation of particulate emboli including fibrin, macroaggregates of denatured proteins and lipoproteins, fat globules, platelet and leukocyte aggregates, calcium, cellular debris, talc, and suture material.

Cardiac Venting for CPB

Cardiac venting involves the active aspiration of blood from the heart, thus creating a bloodless field to facilitate visualization. Venting in some form is necessary in virtually all cardiac operations involving CPB, primarily to aid visualization but also to avoid chamber distension. When the aorta is cross-clamped and the native coronary circulation is stopped, a variable amount of noncoronary collateralization remains, particularly related to bronchial artery flow and return from the thebesian veins. Blood flow can also occur through the heart as a result of continued transit of blood through the right heart, through the lungs, and into the left heart. Potential sites for cardiac venting include the pulmonary artery, the superior pulmonary vein, the left atrium, the left ventricle, and the ascending aorta.

Cannulation for CPB

Noncellular Response

The pathophysiology of CPB relates to the unique responses that occur when the blood contacts the biomaterial surface (biomaterial-dependent processes) and non–contact-related processes such as cardiotomy blood collection and the effect of nonpulsatile flow. Within milliseconds of the blood’s contacting the synthetic surfaces of the CPB circuit, plasma proteins become adsorbed to the biomaterial. Although the amount, composition, and conformation of protein adsorption may differ between surfaces, there is no surface on which this process is completely inhibited, and each surface has a characteristic blood-adsorption pattern. Further exposure of the surface to blood results in activation of proteins of the contact-activation system (Fig. 62-8). This system comprises four primary plasma proteins: factors XII and XI, prekallikrein, and high-molecular-weight kininogen (HMWK). In the presence of the negatively charged biomaterial, a conformational change occurs in factor XII that permits its activation in the presence of prekallikrein and HMWK. Factor XIIa activates factor XI and initiates the intrinsic coagulation pathway, with the subsequent generation of thrombin and the cleavage of fibrinogen to produce fibrin, which is cross-linked by activated factor XIII. Factor XIIa also activates prekallikrein to form kallikrein within seconds of the start of bypass. Kallikrein catalyzes the conversion of HMWK to bradykinin and plays a role in the activation of the fibrinolytic system. Bradykinin has a very short half-life in the plasma because of its rapid metabolism by angiotensin-converting enzyme in the pulmonary circulation²⁸ and by the vascular endothelium.²⁹ It is believed to be a key mediator of increased capillary permeability and the development of tissue edema. Bradykinin mediates vasodilation by stimulating the release of endothelial nitric oxide,²⁹ and it may also be an important mediator of cerebral ischemia.³⁰

Figure 62–8 Proteins of the contact activation system. FXII and FXI, factors XII, XI; FXIIa and FXIa, activated factors XII and XI; HMWK, high-molecular-weight kininogen.

Activation of the fibrinolytic system during CPB is evidenced by increased levels of tissue plasminogen activator (tPA) and plasmin–antiplasmin complexes. Thrombin and bradykinin contribute to fibrinolysis through the direct activation of endothelial cells and the release of tPA, which further increases plasmin generation.³¹ Kallikrein-mediated activation of the fibrinolytic system includes its role in catalyzing the conversion of plasminogen to plasmin and the activation of pro-urokinase.³² Despite these findings, the contribution of fibrinolysis to post-bypass bleeding is controversial, as clot lysis activity subsides within minutes of terminating bypass.³³ Another mechanism by which plasmin generation may contribute to bleeding is through its direct effect on platelet receptors during CPB.³⁴

The complement system is activated through several mechanisms during CPB (Fig. 62-9). First, the third component of complement (C3), adsorbed to the CPB surface after releasing the potent chemoattractant C3a, is joined by the inactive factor B and properdin to produce the active proteolytic enzyme C3 convertase. C3 convertase cleaves the fifth component C5 to generate the active fragment C5a and the terminal complement complex C5b-9. Other mechanisms for complement generation through alternative pathways include the direct cleavage of C5 by kallikrein to produce C5a and the direct cleavage of C3 by plasmin. Complement generation has also been reported to be induced directly by endotoxin or directly by the cytokines tumor necrosis factor (TNF) and interleukin (IL)-6.³⁵ Heparin–protamine complexes formed at the end of CPB activate the classical complement pathway, as does immunoglobulin bound on the biomaterial surface, which complexes with C1q.³⁶

Figure 62–9 Mechanisms of complement activation during cardiopulmonary bypass (CPB). IL-6, interleukin 6; TNF, tumor necrosis factor; TCC, terminal complement complex.

The generation of complement plays a key role in the recruitment of leukocytes, the upregulation of neutrophil activation markers, and the production of cytokines. Studies by Kirklin’s group³⁷ confirmed that the incidence and degree of deranged function of the heart, lung, and kidney after CPB could be related to the raised plasma concentration of the complement fragment C3a. Furthermore, inhibition of production of terminal complement complex by a protease inhibitor was associated with a significant reduction in the deleterious effects of ischemia-reperfusion to the myocardium after CPB.³⁸

Nitric oxide (NO) is a potent inflammatory mediator whose production is increased after CPB. Endogenously produced NO may cause tissue injury through formation of toxic peroxynitrites, activation of cyclooxygenase, and DNA deamination. Studies have reported a significant increase in inducible nitric oxide synthase (iNOS) expression in human lung after CPB, which may be related to cytokine release (e.g., TNF, IL-1, IL-6).³⁹ This induction has also been shown to be associated with myocardial depression secondary to reperfusion injury.⁴⁰

Finally, CPB initiates a cascade of events that results in cytokine release. This response may be triggered by changes in bowel mucosal blood flow and by bacterial translocation, as lipopolysaccharide (LPS) concentration increases by 100% immediately on CPB institution, with another significant increase seen after aortic cross-clamp release. LPS induces a broad range of immunologic effects and is considered the most potent stimulator of TNFα production from macrophages. TNF induces monocyte IL-1 production, and TNF and IL-1 in concert induce IL-6 production and release.⁴¹ Whereas a rise in TNF and IL-1 as a result of CPB has not been regularly demonstrated in all studies, a marked increase in IL-6 levels appears consistently with CPB. Peak concentration of IL-6 occurs a few hours after the end of CPB, with a gradual decrease toward preoperative levels in the following 24 hours,⁴² as seen in noncardiac surgery.⁴³

Several investigators have demonstrated a correlation between the release of inflammatory mediators and myocardial dysfunction after coronary artery bypass surgery. Peak IL-6 and IL-8 levels correlate with the degree of myocardial dysfunction.⁴⁴ TNF may be released locally in the myocardium, and this may be related to postischemic myocardial stunning.⁴⁵ Jansen and coworkers⁴⁶ demonstrated that a rise in TNF can be detected after release of the aortic cross-clamp. TNF, IL-6,⁴⁷ and IL-8 levels correlated with the duration of the cross-clamping and the degree of myocardial injury as reflected by the creatine kinase (CK) MB isoenzyme levels.⁴⁷

Cellular Activation during CPB: Platelets, Endothelial Cells, Leukocytes

Although the cellular events during CPB are influenced by the dynamic nature of the activation by the unique soluble products produced at the biomaterial, they are more significantly affected by the composition of proteins adsorbed on the nonendothelial surface. Surface-adsorbed fibrinogen is the key mediator of platelet accumulation on foreign materials. Platelet adhesion to artificial surfaces increases with increasing surface concentration of adsorbed fibrinogen if the bound fibrinogen maintains a conformation such that the functional domains of the molecule are recognizable by the activated platelet GP IIb/IIIa receptor. This latter interaction is probably the most important factor mediating platelet consumption during CPB. The three-dimensional conformation of adsorbed fibrinogen also influences the degree of platelet accumulation.

CPB is associated with a consistent increase in the proportion of activated circulating platelets. Agonists of platelet activation during CPB include thrombin, adenosine diphosphate (ADP), heparin, protamine, activated complement, and plasmin. Physical activators of platelets include hypothermia and the process of cardiotomy blood collection. The latter effect may be related to the air–blood interface, the blood–tissue interface, or the exposure to thrombin generation in cardiotomy blood.

The most consistently documented measure of CPB-related activation is an increase in the expression of guanosine monophosphate (GMP)-140 (P-selectin).⁴⁸ Platelet glycoprotein (GP) IIb/IIIa is also likely activated,⁴⁹ but it is controversial whether there is loss of other surface receptors such as GP Ib. Platelet microparticles, which may be highly procoagulant, can be consistently detected by flow cytometry after CPB.⁵⁰ Other platelet products increased after CPB include β-thromboglobulin and platelet factor 4 (PF4).⁵¹

In clinical CPB, thrombocytopenia is seen commonly, with platelet counts dropping more than 50%, resulting not only from platelet adhesion to surfaces but also from hemodilution, platelet aggregate formation, and the formation of platelet–leukocyte complexes. With these changes, it is not surprising that the most predictable alteration in hemostatic function observed after CPB is platelet dysfunction. Clinical reflectors of this process include a universal prolongation of the bleeding time that is directly related to postoperative blood loss.⁵² Platelet aggregation to ADP and epinephrine is consistently abnormal,⁵³ and there is decreased response to thrombin agonist receptor peptide (TRAP), suggesting a decrease in receptor sensitivity.⁵⁴

Because of the intense anti-inflammatory reaction related to CPB, it might be expected that the endothelial layer would be at the front line of many of the cellular changes related to CPB. It is now evident that it is the activation of this axis that mediates many of the injurious processes related to CPB, such as reperfusion injury. The generation of cytokines is a major contributor to endothelial cell activation, primarily because they upregulate receptors necessary for neutrophil–endothelial cell binding.⁵⁵ Endothelial cells undergo upregulation of intercellular adhesion molecule (ICAM) and E selectin. The latter binds to the integrin CD11a/CD18 present on resting polymorphonuclear leukocytes (PMN). PMN and monocytes may be activated with upregulation of the integrin CD11b/CD18 complex,⁵⁶ which also binds ICAM on endothelial cells.⁵⁷ The response to acute injury of the endothelium during CPB is evident by the acute rise in soluble P-selectin, with a concomitant fall in soluble E-selectin, and by increased elastase release.⁵⁸ Leukocytes interact with platelets during CPB as another means of systemic inflammation. Increased GMP140 on the platelet surface is essential to mediate complex formation of platelets and monocytes or PNMs.⁵⁷ Monocyte–platelet conjugates increased from 18% to 44%, whereas PMN–platelet conjugates increased only slightly.⁵⁷