Blood Pumps

All extracorporeal flow occurs through processes that incorporate a transfer of energy from mechanical forces to a perfusate, and, ultimately, to the tissue. Methods of achieving this transfer of energy include gravitational and mechanical forces, or a combination of both. It is through the transfer of energy from an electrical power source to the motor of a pumping mechanism and on to the fluid (blood) that causes tissue perfusion.^4,6 Most extracorporeal pumps fall into one of the following categories: positive displacement (PD), centrifugal or constrained vortex (CP), passive filling, pneumatic and electrical pulsation, and axial flow (the latter pumps are used primarily as cardiac assist or replacement devices),^7–9 and are described in Chapter 27.

Positive Displacement Pumps

The PD pump operates by occluding tubing between a stationary raceway and rotating roller(s) or occluder(s) (Figure 29-1). The pumping mechanism is also referred to as the pump head, and the tubing that traverses the raceway is referred to as the pump header. PD pumps were first proposed for use in cardiovascular medicine in the 1930s by Gibbon.² In 1935, an adaptation to the PD pump was described that included tube bushings at the head of the raceway on both inlet and outlet locations to prevent tubing creepage around the roller head.⁹ Melrose¹⁰ later modified the pump to include a grooved raceway, which further reduced tubing shimmy. Both of these adaptations were important in reducing the mobility of tubing during the operation of the pump, which decreased the potential for tubing rupture in the pump head. In a PD pump, fluid is displaced in a progressive fashion from suction to discharge, with the capacity of the displacement dependent both on the volume of the tubing occluded by the rollers and on the number of revolutions per minute (rpm) of the roller. All PD roller pumps (RPs) use the volume in the pump header, which is referred to as a flow constant, and is specific to each size of tubing referred to by the internal diameter of tubing, for calculating the flow of the pump. This is displayed on a digital readout and is referred to as the output (flow) of the pump. It is measured in liters per minute. Although many types of RPs have been used for CPB, the most common PD pump in use today is the twin-RP.

Figure 29-1 Stockert S-3 Twin Roller Pump diagram.

A positive displacement pump with a stationary raceway and rotating twin roller pumps.

(Courtesy of The Sorin Group, Arvada CO.)

There are currently at least five manufacturers of PD pumps used in ECC, with each device consisting of minor variations of the twin-RP design (Box 29-1). A modern heart-lung machine consists of between four and five of these RPs positioned on a base console (Figures 29-2 and 29-3). Most machines are modular in design, permitting the rapid change-out of a defective unit in the case of single-pump failure. It is standard practice of perfusionists to rotate the pumps along the base console in different positions so that mechanical wear is distributed evenly while maintaining equitable time utilization. Each pump is independently controlled by a rheostat that functions to regulate the rpm of the rollers. Each pump is calibrated according to specific flow constants that are calculated from the internal diameter of tubing, as well as the tubing length, placed in the pump raceway. Periodically, PD pumps are calibrated by performing a timed collection of pumped fluid to verify that after proper calibration the pump delivers the volume indicated on the pump flow display. The internal diameters for ECC tubing ranges between 1/8 and 5/8 inch/min. For this reason, a single console can be used to perfuse a wide range of patients whose size may vary from a few kilograms to several hundred. This is accomplished simply by changing the raceway tubing and the shims that hold the tubing in place. It is important to note that the larger the internal diameter of the tubing, the lower the rpm necessary to achieve a desired pump flow. This is especially important because there is a positive correlation between red blood cell (RBC) hemolysis and the rpm of the pump rotation. The magnitude of hemolysis is related to both the time and exposure of the blood to shear forces generated by the pump. A region of high pressure and shear force is created at the leading edge of the roller where the tubing is compressed, which is followed by a period of negative pressure as the tubing expands behind the roller. This momentary negative pressure under certain conditions may induce the cavitation of air dissolved in the solution. A further related concern is particulate emboli that may be generated by microfragmentation, so-called spallation, of the inner surface of the tubing where the roller contacts the tubing and where the fold at the edges of the tubing occurs.¹¹ Studies of tubing wear over time have shown that polyvinylchloride fragments generated from RPs are numerous, frequently less than 20 μm in diameter, and begin to occur during the first hour of use.¹² However, the majority of the hemolysis generated during a routine CPB procedure is not related to the occlusiveness of the arterial pump head but rather by the air-surface interface interaction occurring with the use of suction and “vent” lines components of the circuit.¹³ An underocclusive arterial pump head will result in retrograde flow. This, in turn, will require increased rpm to ensure adequate forward flow, which increases hemolysis.¹⁴ Overocclusive adjustment of the RP results in both hemolysis of RBCs and spallation (particulate fragmentation from the inner walls of tubing) that continues with PD pumps.¹⁵ Kurusz¹¹ identified the erosive and fatiguing action of the RP as a major source for generating tubing particles in CPB circuits.

Figure 29-2 Schematic diagram of cardiopulmonary bypass circuit including four roller pumps (one vent pump, two suction pumps, and a cardioplegia deliver pump). A centrifugal blood pump for systemic blood propulsion is shown on the lower right.

(From Hensley FA, Martin DE, Gravlee GP: A Practical Approach to Cardiac Anesthesia, 4th ed. Philadelphia, Lippincott Williams & Wilkins, 2008, Figure 18.1.)

Figure 29-3 HL20 Heart Lung Machine Console.

(Courtesy of Maquet Cardiovascular, Wayne, NJ.)

The setting of occlusion in the pump head is extremely important and varies among the pumps used on the heart-lung machine console. The arterial pump head occlusion should be set by a water-drop method that incorporates a “30-and-1” rule for setting occlusion. In this method, the occlusion of the arterial pump is set by displacing a column of water (perfusate) 30 cm above the highest water level in the venous or cardiotomy reservoir (whichever is highest) and allowing the perfusate to drop 1 cm/min. The same drop rate can be obtained by setting the fluid height difference at 30 inches and the drop rate at 1 inch/min. Of note, if cardioplegic solution is to be delivered through a separate RP, and/or a left ventricular drainage line (left ventricular vent) placed in a roller head, occlusion for these pumps should be set at 100% (full occlusion) with no drop in fluid movement. This ensures that during the time when cardioplegic solution is not delivered, or the left ventricular vent is turned off, the risk for negative pressure in the ascending aorta or coronary sinus, created by a slowly falling column of fluid, does not create a siphon that causes cavitation or the entrainment of air into the infusion lines. Such aspirated air could be infused directly into the patient by restarting the pump. The heart is vented during CPB to facilitate the removal of ventricular blood that accumulates from noncoronary mediastinal collateral vessels, arteriovenous sinusoids, and thebesian veins, all of which drain directly into the left atrium (LA) or left ventricle. Other anatomic locations of venting the heart include the pulmonary artery and the ascending aorta, with the latter usually drained through an antegrade cardioplegia cannula. The remaining pumps usually are denoted as “suckers” and aspirate shed blood from the operative field. Although debated, it generally is thought that a slightly nonocclusive pump sucker leads to reductions in the amounts of RBC trauma and hemolysis. Additional uses for the peripheral PD pumps include ultrafiltration (UF) or dialysis, for topical myocardial cooling devices, or for removing air from collapsible venous reservoirs.

Centrifugal Pumps

The second type of extracorporeal pump is a resistance-dependent pump termed a centrifugal (CP), or constrained vortex pump.^16–18 The CP conducts fluid movement by the addition of kinetic energy to a fluid through the forced centrifugal rotation of an impeller or cone in a constrained housing (Box 29-2). The greatest force, highest energy, is found at a point most distal to the center axis of rotation (Figures. 29-4 to 29-6). CPs operate as pressure-sensitive pumps, with blood flow directly related to downstream resistance. Blood flow is, therefore, related to both the rpm of the cones or impellers and the total resistance. This represents an important safety feature in coupling blood flow with resistance. During unexpected increases in resistance, the total energy transfer from the CP to blood will not generate forces sufficient to cause arterial line separation. However, when downstream occlusion occurs, either through increases in afterload or through the placement of line clamps, the fluid in the pump head will be heated because of hydrodynamic processes in the magnetic coupling. This increase in temperature could result in increased blood trauma and coagulation defects.¹⁸

Figure 29-4 Rotaflow Centrifugal Pump disposable with low-friction one-point bearing (sapphire ball and PE calotte).

(Courtesy of Maquet Cardiovascular, Wayne, NJ.)

Figure 29-5 Revolution Centrifugal Pump Disposable.

(Courtesy of The Sorin Group Arvada, CO.)

Figure 29-6 Medtronic Bio-Pump Plus Centrifugal Pump.

The manufacturer states that the Bio-Pump laminar flow design is superior to roller pump and impeller pump designs with inherently more turbulent flow characteristics.

(Courtesy of Medtronic Cardiovascular, Eden Prarie, MN.)

The acceptance of these devices in routine CPB has increased tremendously since first being introduced into clinical practice in 1969,¹⁹ and it is the pump of choice during emergency bypass procedures. The CP also has been used as a ventricular assist device (VAD) because of its inherent safety features and pressure sensitivity, as well as relatively low cost. Although these pumps have been used extensively off-label for VADs, none of the CPs has received U.S. Food and Drug Administration clearance for use for systemic circulatory support for more than 6 hours. CP pumps have been used extensively off-label as VADs or in extracorporeal membrane oxygenation (ECMO) circuits. The afterload and preload sensitivity of these pumps make them particularly amenable for use for ECMO in the treatment of reversible respiratory dysfunction and postcardiotomy dysfunction.²⁰ The Levitronix CentriMag Blood Pumping System (Levitronix, Waltham, MA) recently developed a pump with a novel magnetically levitated bearingless motor technology designed to minimize friction and heat generation in the blood path (Figure 29-7), which reduces stasis and minimizes blood trauma. The Centrimag is approved for use for up to 6 hours of support and is undergoing further investigation for prolonged use for patients with heart failure (HF). The Centrimag also recently received approval by the FDA for use as a right ventricular support device, for use up to 14 days to treat patients with right-heart failure—the first approval of this class of pump for use beyond 6 hours. In a recent in vitro study, Guan et al²¹ compared mechanical performance characteristics of the Centrimag pump with a conventional CP, the Rotaflow Centrifugal Pump (Maquet, Wayne, NJ) and reported better mechanical performance characteristics with the rotaflow pump in terms of higher shutoff flow rate, maximal flow, and propensity for retrograde flow.²¹ These findings deserve further study given the magnitude of cost for the Centrimag pump system disposable components (Centrimag costs 20 to 30 times more than other CP disposable components).

Figure 29-7 CentriMag Blood Pump.

Thoratec CentriMag magnetically levitated bearingless blood pump.

(Reprinted with permission from Thoratec Corporation.)

When gross air is introduced into the CP, as in emptying of the venous reservoir, the pump head will deprime, stopping forward flow, which reduces the risk for gas embolization. However, when small quantities of air are aspirated into the pump head, these bubbles will coalesce and be passed into the outlet stream of fluid movement, and potentially into the patient. Although the CP has been described as exerting less trauma to the cellular elements of blood,²² variability in individual pump hemolytic potential has been reported.^23,24 Tamari et al²⁵ have reported that the degree of hemolysis in CP is related to the hemodynamic conditions under which the pump is operated, with lower flows and higher pressure resulting in more hemolysis than similarly operated RPs. There have been reports of thrombus formation when these pumps are used with low anticoagulation or for prolonged periods.²⁶ Later designs possess fins and channels that prevent these areas of stasis. Improved designs have addressed issues of stasis, heat generation, and bearing wear. One contemporary design has minimal contact area for the cone and the outer housing and incorporates a series of magnets to suspend the moving rotor within the pump housing.²⁷ Additional advantages of CP over PD RPs include reduced mechanical trauma to extracorporeal tubing and the generation of high-volume output with moderate pressure development. A potential complication associated with nonocclusive-type pumps involves retrograde flow through the aortic cannula when the pressure in the central aorta exceeds that generated by the pump.²⁸ This may occur during times of power disruption or pump failure when there is an increased risk for drawing air into the arterial line via purse-string sutures placed to secure the arterial cannula (see Safety Mechanisms for Extracorporeal Flow section later in this chapter). Other uses of CPs include supported CPB in high-risk angioplasty patients, left-heart bypass (LHB) during repair of descending thoracic aortic aneurysms or dissections, and veno-venous bypass during hepatic transplantation. Use of CPs to assist venous return for minimally invasive cardiac surgery is described as kinetic-assisted venous return.^29,30

Currently, six manufacturers produce CPs for extracorporeal use: Biomedicus (Biomedicus-Medtronics, Minneapolis, MN), Delphin (3M Health Care, Ann Arbor, MI), Revolution Pump (The Sorin Group, Arvanda, CO), Capiox-SP (Terumo Medical Corporation, Somerset, NJ), Rotoflow Maquet (Wayne, NJ), and the Centrimag Pump. The operational characteristics are similar among the various systems in which the internal smooth cones or vaned impellers are connected to a central magnet (isolated from contact with blood by encasement in a polycarbonate housing), which couples with the console, where electromagnetic forces are produced. The centrifugal console usually is placed in the arterial pump head position on the heart-lung machine, replacing the main drive. All of the consoles currently available include their own battery backup systems in the event of power failure and a manually operated motor in case of drive motor or console failure. The Revolution pump is equipped with an electronic tubing clamp that may be programmed to deploy automatically if low, zero, or retrograde flow is sensed; if the level sensor in the venous reservoir senses a low level; if a high arterial line pressure is sensed; or if the air detector on the arterial line senses air in the circuit (Figure 29-8). This is an especially important feature when using these machines to transfer patients on ventricular assistance or during the conduct of emergency bypass. Each manufacturer markets disposable software that must be purchased in conjunction with the pump.

Figure 29-8 E-Clamp electronic line clamp works in conjunction with the Revolution centrifugal pump. The clamp is automatically deployed to clamp the line when retrograde flow is detected or if the air detector or level detector is triggered.

(Courtesy of The Sorin Group, Arvada, CO.)

A number of investigators have conducted in vitro studies comparing CPs and RPs in terms of blood handling during short-term and long-term use. Oku et al,³¹ Jakob et al,³² Englehardt et al,³³ and Hoerr et al³⁴ reported less hemolysis with the CP when tested in vitro. Kress et al³⁵ showed no difference between the two pump types in a rabbit ECMO model. Tamari et al³⁶ examined hemolysis under various flow and pressure conditions in an in vitro model using porcine blood and concluded that the hemolysis index was related to the duration of blood exposure to shear, the ratio of pump pressure difference between the inflow and outflow, and the flow rate of the pump. From this work they provided guidelines related to pump selection based on the pressure/flow ratio likely to occur in a given application. Rawn et al³⁷ compared an underocclusive RP with a CP and found a significantly higher index of hemolysis in the CP (3.38 to 14.65 vs. 29.58 g/100 L pumped). In a randomized trial, Salo et al³⁸ examined inflammatory response mediators in 16 CABG patients with CPB times of less than 2 hours. These mediators included interleukin-1 beta (IL-1β), IL-2, IL-6, phospholipase A₂, endotoxin, fibronectin, and serum C Group II phospholipase A₂. These researchers found no differences in the levels of these inflammatory markers immediately post-CPB and at 24 hours after surgery. Other randomized clinical trials have been conducted to compare emboli generation, neurocognitive outcome, blood trauma, and patient charges. Wheeldon et al³⁹ conducted a randomized, controlled trial in 16 patients, in which the only difference in equipment and technique was the type of pump used, and found significantly fewer microemboli, less complement activation, and better preservation of platelet count. Parault and Conrad⁴⁰ reported a similar significant improvement in platelet preservation in a retrospective review of 785 cases and further reported that the differences were more profound in patients older than 70 years with CPB times of longer than 2 hours. Klein et al⁴¹ conducted a randomized, prospective clinical study in 1000 adult cardiac patients comparing RPs with the Biomedicus CP (Medtronic, Eden Prairie, MN), using risk stratification methodology, and reported clinical benefits to the CP including blood loss, renal function, and neurologic outcomes, but no significant difference in mortality. Ashraf et al⁴² examined S100 beta levels relative to pump type in a randomized, controlled trial that included 32 patients who had CABG and found no significant difference in S100 beta levels between the groups at 2 and 24 hours after bypass. Dickinson et al⁴³ did a retrospective review of 102 patients examining length of stay, total patient charges, reimbursement, mortality, and major complications but could not identify a single difference. A more recent randomized control trial by Scott et al⁴⁴ subjected 103 patients to a battery of 6 standardized tests and found a trend toward fewer abnormal tests in the CP group; however, it failed to reach statistical significance. DeBois et al⁴⁵ conducted a trial in 200 elective CABG surgery patients who were randomized to either an RP or CP and found similar patient characteristics including platelet counts, hematocrit, transfusion rate, and mortality; however, they observed differences favoring the CP with regard to weight gain, length of stay, and net hospital financial balance. Alamanni et al⁴⁶ evaluated the prevalence of major neurologic complications in 3,438 consecutive patients and found the occurrence of injury to be associated with age and a history of a previous neurologic event. The authors further reported that use of the CP provided a risk reduction for the considered events ranging from 23% to 84%. Babin-Ebell⁴⁷ et al conducted a randomized trial of CABG patients and found a significant reduction in tissue factor in the group supported with a CP; however, this did not translate into a measurable reduction in thrombin formation or other apparent clinical benefit. Baufreton et al⁴⁸ examined cytokine production (tumor necrosis factor-α, IL-6, IL-8) and circulating adhesion molecules (soluble endothelial-leukocyte adhesion molecule-1 and intercellular adhesion molecule-1) in a randomized, controlled trial of 29 CABG patients. They reported greater SC5b-9 and elastase levels in the CP group, suggesting more favorable performance from the RP with regard to complement and neutrophil activation.

Although nearly all of the randomized trials show significant benefit to systems designed with CPs, it is difficult to separate the improved performance conferred from other characteristics, such as lower prime volume, surface coating, more limited surface area, and reduced air-to-blood contact. Current research would suggest that CPs produce less blood damage; however, this improvement may be masked by blood trauma and inflammation related to contact activation of the blood related to cardiotomy suction, the introduction of gaseous and particulate emboli, and related factors. According to the recently published Guidelines on Perioperative Blood Transfusion and Blood Conservation in Cardiac Surgery, jointly endorsed by the Society of Thoracic Surgeons and the Society of Cardiovascular Anesthesiologists, “It is not unreasonable to select a CP rather than a RP but more so for safety reasons rather than blood conservation” (American Heart Association/American College of Cardiology Class IIb level of evidence B).⁴⁹ In 2000, approximately 50% of the cardiac centers in the United States routinely used CPs.⁵⁰

Electromagnetic transducers and Doppler ultrasonic flowmeters are the two methods of measuring CP flow, as compared with the digital display of the PD pumps, which is the product of a flow constant and rpm. Electromagnetic flowmeters operate under Faraday’s principle, that an electric current can be produced in a wire moved through a magnetic field. Voltage is generated when an electrical conductor moves through a magnetic field if the movement is perpendicular to the magnetic lines between the poles of the magnet. Because blood is an electrical conductor, voltage is generated when it passes through a magnetic field and the voltage is directly proportional to the velocity of blood movement. The Doppler technology uses digital signal processing to transform the Doppler analog signal received from the flowmeter into digital format. Fast Fourier transformation then matches the incoming signal to recognizable patterns, which are displayed as flow rates.

Safety Mechanisms for Extracorporeal Flow

Some of the most recent advances in pump design have been a result of a heightened awareness of increasing safety associated with complex operating systems. The PD pumps are pressure independent, which means they will continue to pump regardless of downstream resistance. In a CPB circuit, the summation of resistances against which a pump must function includes the total tubing length, the oxygenator, the heat exchanger, the arterial line filter, the cannula, and the patient’s systemic vascular resistance (SVR). Additional factors that influence SVR include the viscosity of the perfusate, related to the total formed element concentration, which primarily is dependent on the formed elements of blood and the temperature of the solution. According to Poiseuille’s law, the greatest resistance to flow is created at the arterial cannula, where the change in the caliber of the tubing lumen declines the most. Perfusionists routinely monitor the summation of all resistances and record this value as the arterial line, or system, pressure. This always will be greater than the pressure measured at the distal end of the circuit terminating at the cannula tip because the pressure drop across each component in the series circuit will be subtracted from the summation of resistance (resistors) in the entire circuit. Bypass circuitry and components have been designed to incorporate minimal pressure drops; therefore, in routine adult perfusion, the resistance becomes a function of the patient’s SVR and the pump flow rate. Establishing a normal value for arterial line resistance is difficult, although normal limits range between 100 and 350 mm Hg. Any acute change in resistance, such as unexpected clamping or kinking of the arterial line, results in an abrupt increase in arterial line pressure, which can lead to catastrophic line separation or circuit fracture anywhere on the high-pressure side of the circuit. A life-threatening event could occur on the initiation of CPB if the tip of the arterial cannula lodges against the wall of the aorta, undermining the intima of the vessel. Under these conditions, aortic dissection can occur as the vessel intima separates from the media, directing blood flow into a newly created false lumen. This dissection can extend throughout the entire length of the aorta. For this reason, perfusionists routinely check the line pressure after cannulation before the onset of CPB to ensure the presence of a pulsatile waveform, indicating proper cannula placement in the central lumen of the aorta. Either the absence of pulsatility or an extremely high line pressure (> 400 mm Hg when CPB is initiated) should immediately be investigated (see Chapter 28).

All heart-lung machines include a microprocessor-controlled safety interface with their pump consoles. These systems monitor and control pump function and serve as the primary mechanical safety control system for regulating extracorporeal flow. Pressure limits are set by the perfusionist and are determined by patient characteristics and the type of intervention performed. These units consist of early-warning alarms that alert the user to abrupt changes in pressure and will automatically turn off a pump when preset limits are exceeded. These safety devices have been used in both the main arterial pump and the cardioplegia pump; the latter become more important with the utilization of retrograde cardioplegia administration into the coronary sinus.^51,52 Currently, the incorporation of a safety monitor for negative pressure sensing, located on the inflow side of the arterial pump head and during pulsatile perfusion when intermittent occlusion is created, still is lacking.

Electrical failure in the operating room can be especially catastrophic in the conduct of ECC when the native heart and lungs are unable to function. When such an event occurs during CPB, it is imperative that instantaneous actions be instituted to minimize the risk for whole-body hypoperfusion. The perfusionist should be mindful of the power limitations of the electrical outlet used in the cardiac operating room and also be aware of the location of the circuit breaker panel for the room and the specific number of the breaker in the panel for the outlet used for the heart-lung machine and other support equipment. Methods to ensure the safe conduct of CPB involve the incorporation of an emergency power source in the extracorporeal circuit that provides a secondary power source in the event of electrical interruption. Electrical failure during CPB was reported by 42.3% of respondents in a survey on perfusion accidents.⁵³ Although hospitals are equipped with emergency generators for such events, their availability may be limited to certain electrical circuits within the operating suite. Furthermore, these emergency power systems require a brief interruption in power before a generator or backup source of power is initiated. Most heart-lung machines are equipped with uninterrupted backup power, sometimes referred to as the “Uninterrupted Power Source” (UPS), whereby there is a seamless transfer from the wall power source to an internal battery within the pump should the wall power fail. Thus, with this system, there is no loss of flow from the pumps that could result in retrograde flow and entrainment of air or disruption of settings and timers. Cases of primary power failure with concurrent emergency backup failure also have been reported.⁵⁴ As a tertiary fallback measure, emergency hand cranks for CPB pumps are standard features in extracorporeal circuitry, which enable pump operation in the event of total power failure and when emergency systems fail to operate. However, care should be taken to ensure that the direction of blood flow is ascertained because hand cranking in the reverse direction of fluid flow could result in serious patient injury related to exsanguination and the entrainment of air around purse-string sutures at cannulation sites. The Retroguard valve (Quest Medical Incorporated, Allen, TX), a mechanical circuit component to prevent retrograde flow in the arterial outflow of the circuit, is available to prevent retrograde flow in the arterial line and possible entrainment of air into the circuit and into the patient’s arterial circulation. It is composed of a simple duck-bill valve that adds minimal resistance to forward flow in the circuit and will close when downstream pressure is greater (Figure 29-9).

Figure 29-9 Quest Medical RetroGard Valve.

Blood enters from the 3/8-inch inlet, passes through a duck-billed valve, and exits through a 3/8-inch outlet. This valve prevents retrograde flow. This valve may also be placed on an inlet port of the venous reservoir to prevent pressurization of the hard shell venous reservoir.

(Courtesy of Quest Medical, Allen, TX.)

Although the chance of infusing massive air boluses to patients has been reduced dramatically since the early days of CPB,⁵³ this remains a serious potential event during surgery (Box 29-3). Cannulation of the heart with aortic and venting catheters has been identified as the primary cause for air embolization during ECC.⁵⁵ Methods of air-bubble detection have improved tremendously, and the sensitivity for detecting small amounts of air has increased in modern heart-lung machines.⁵⁶ Ultrasonic and capacitance air-detection systems, used for both level sensing and air detection in arterial and cardioplegia circuit lines, represent dramatic improvements over less sensitive (photoelectric) methods.⁵⁷ However, lacking in clinical practice are effective, reliable level-sensing devices that alert the perfusionist to rapid changes in venous reservoir levels, especially during utilization of collapsible venous reservoir systems. Both air-bubble detection and level-sensing devices should be safety techniques used as standards of care in all extracorporeal circuits.

Blood Gas Exchange Devices

The ECC of blood incorporating total heart-lung bypass could not be accomplished were it not for the development of devices that could replace the function of the lungs in pulmonary gas exchange. The technology of pumps to replace the mechanical action of the heart was developed well before their incorporation in ECC. Therefore, the limiting factor hindering the progression of CPB was the development of an artificial lung, or blood gas exchange device (BGED), commonly referred to as a membrane oxygenator (Box 29-4). The term membrane denotes the separation of blood and gas phases by a semipermeable barrier, whereas oxygenator refers to the change in oxygen partial pressure that occurs by the arterialization of venous blood. However, “oxygenator” is a misrepresentation of the functional ability of these systems to perform ventilatory control of carbon dioxide. Numerous engineering challenges hindered the development of BGEDs, but two of the most pressing were the design of high-capacity units for gas exchange with low rates of bioreactivity. The latter requirement, also termed biocompatibility, was imperative to reduce both RBC trauma and activation of the formed elements of blood.

In the 1940s, the first dialyzer membranes were made of cellulose acetate, and although intended for use in dialysis, they also had gas exchange characteristics.¹ In the 1950s, several membrane materials (including polyethylene and ethyl cellulose) were used in a flat sheet or plate configuration. At the same time, rotating disk oxygenators were introduced whereby gas exchange was accomplished by spreading venous blood in a thin film over a rotating disk, which was exposed to an oxygen-rich environment. In the 1960s, the first disposable membrane oxygenators were introduced and were made primarily of silicone rubber in either a plate or spiral wound design. Silicone offered the distinct advantage of separating both the blood and gas phases, facilitating gas exchange through a semipermeable barrier by diffusion. Teflon was introduced in the 1970s as a membrane material, together with microporous polypropylene, which first appeared in Travenol membrane devices. Today, the majority of commercially available oxygenators are made of polypropylene in either a pleated or folded configuration, or as capillary hollow fibers (Figure 29-10). In the United States, manufacturers develop oxygenators that meet federal regulatory guidelines for performance and biocompatibility. Those devices meeting these requirements are “cleared,” approved for use for up to 6 hours of CPB, and represent the majority of oxygenators. Currently, there is only one oxygenator that utilizes silicone membranes that is approved for long-term support such as that occurring for ECMO. However, the “off-label” use of more durable, lower prime, hollow-fiber technology membrane oxygenators and newer polymethylpentane fiber oxygenators is widely reported in the literature.

Figure 29-10 Diagram of blood flow path through the Medtronic Affinity Oxygenator. Heat exchange water path shown in blue. Gas flow path shown in green. Blood flow path shown in red.

(Courtesy of Medtronic Cardiovascular, Eden Prarie, MN.)

Historically, oxygenators were divided into two broad classes based on the method of gas exchange: bubble and membrane systems. Bubble-type devices have been shown to denature plasma proteins, increase RBC fragility,⁵⁸ activate platelets,⁵⁹ and generate substantial gaseous microemboli (GME).^60–62 For these reasons, they are no longer used in most countries and are infrequently encountered in all but a few remaining places throughout the world. Bubbler systems use a direct gas-blood interface, with gas exchange occurring by the dispersion of gas, either 100% oxygen or a mixture of oxygen and carbon dioxide (carbogen), through a column of desaturated blood. Bubble devices are made of two separate compartments: an oxygenating column and a defoaming chamber. The dispersion of gas in a bubbler occurs through a sparger plate, where a thin film of blood comes in direct contact with gas. This direct blood-gas interface results in the production of foam, where gas exchange occurs. Coalescence of the foam is achieved in the defoaming chamber both through the presence of surface tension–reducing substances and by filtration. Gas exchange is affected by several factors, including the quantity of gas and the size of bubbles produced in the gas sparger.⁶³ Small bubbles are extremely efficient at oxygen exchange but poor at carbon dioxide exchange, whereas large bubbles are poor in oxygen but good in carbon dioxide exchange.

Membrane oxygenators are made of three distinct compartments: gas, blood, and water (see Figure 29-10). The latter phase is also termed the heat exchange compartment and is used for temperature control. Gas and blood are partitioned into separate compartments with either a limited or absent gas-blood interface. Microporous membrane oxygenators initially have a blood-gas interface that becomes diminished only after the inner blood contact surface has been exposed to plasma; and a protein layer is deposited, acting as a diffusible barrier to gas exchange. The most common material in use today in membrane oxygenators is microporous polypropylene, which has excellent capacity for gas exchange and good biocompatibility. Membrane devices made of silicone materials transfer gas directly by diffusion across the semipermeable membrane and effectively never have a blood–gas interface.⁶⁴ Despite the improvements made to extracorporeal devices over the past several decades, once blood is exposed to synthetic surfaces, hematologic changes result. Initially, complement is activated mainly through alternative pathways, resulting in the liberation of toxic mediators such as C3a and C5a.^65,66 Both platelets and leukocytes that elicit a complex series of inflammatory and hemostatic reactions that ultimately increase the risk for postoperative complications are activated.⁶³

Gas transfer in membrane oxygenators is a function of several factors that include surface area, the partial pressures of venous oxygen and carbon dioxide, blood flow, ventilation flow (called sweep rate), and gas flow composition. Membrane devices independently control arterial oxygen and carbon dioxide tensions (Pao₂ and Paco₂). Pao₂ is a function of the Fio₂, whereas Paco₂ is determined by the sweep rate of the ventilating gas. This independent control of ventilating gas results in arterial blood gas values more closely resembling normal physiologic blood gas status. However, it is common for perfusionists to maintain Pao₂ levels in the 150- to 250-mm Hg range during CPB because of the limited reserve capacity of membrane oxygenators.

A multitude of factors must be considered in the design of a membrane BGED, including total surface area, blood film thickness, diffusion residence time, gas diffusion rate, blood flow rate, blood flow geometrics, and gas flow characteristics. The most influential factors that affect blood trauma in an oxygenator are related to how blood traverses the device and are termed shear stress and stasis.⁶⁷ Design characteristics that minimize these effects by optimizing flow pattern geometry through extracorporeal devices have been generated through mathematical models termed computational fluid dynamics.⁶⁸ Two of the most important considerations in designing a membrane device are determining the type of membrane material and the handling of water vapor produced in the gas phase of the device. This water vapor would be synonymous with pulmonary exudate and, when excessive, mimics pulmonary edema associated with permeability changes of the alveolar capillary membrane. Another important membrane feature is how blood flows through the membrane. As fluid moves through a conduit, laminae are established, with the highest velocity of flow achieved in the center of the tube. At the same time, the outermost layers, nearest the walls of the conduit, effectively have no velocity because of the drag coefficient of the inside surface. This occurs in both the gas phase and the blood phase of membrane oxygenators. The laminar effect can be disrupted by several techniques that produce a “secondary flow,” facilitating increased gas exchange.⁶⁹ In hollow-fiber membrane oxygenators, incorporating blood flow outside of the fibers, mixing is achieved by winding of the fibers, creating a crossing pattern, increasing blood exposure to the membrane surface. Laminar flow is reduced in hollow-fiber oxygenators with blood flow through the fibers by the expansion and contraction of the capillaries via the movement of blood through them, gently disrupting the boundary layers.

Estimating the total surface area of material necessary for gas exchange is a function of the predicted oxygen demands of the patient and is used as a primary determinant for selecting membrane size. As the surface area of an oxygenator increases, the volume of solution necessary to prime the system increases. Microporous polypropylene hollow-fiber membrane devices come in two classifications: those with blood flow through the fiber and those with blood flow around the fiber. Systems that use the latter design require a lower membrane surface area for gas exchange and hence result in lower prime volumes. Microporous polypropylene membranes have the distinct advantage of a greater gas transfer rate per surface area of membrane than that of silicone membranes.

The oxygenator represents the largest source of nonendothelialized surface area in the extracorporeal circuit, ranging in size between 0.5 and 2.5 M². As a consequence, it is imperative that the device is meticulously primed to remove all residual air before establishing CPB. Oxygenators have been shown to possess different abilities to remove gaseous emboli that vary according to the physical CPB conditions including temperature and pressure decline.^70–73 In an effort to reduce surface exposure and prime volume, several membrane oxygenators are now manufactured that either possess integrated arterial line filters (FX Oxygenator Line; Terumo Cardiovascular, Ann Arbor, MI; Figure 29-11) or systems in which the arterial line filter is sequenced in the oxygenator (Synthesis; Sorin Biomedical, Arvada, CO). Some studies suggest that these devices may result in a reduction in gaseous micromboli.^74,75

Figure 29-11 Hard-shell venous reservoir with integral cardiotomy and membrane oxygenator (Terumo RX15).

(Courtesy of Terumo Cardiovascular, Ann Arbor, MI)

Numerous studies have identified the occurrence of GME during cardiac surgery with CPB.⁷⁶ Weitkemper et al⁷⁷ have shown that currently used microporous membrane oxygenators have widely variable characteristics related to how they handle gas. Furthermore, the design characteristics in some cases cause partial removal of GME, as well as a change in size and numbers of microbubbles. Dickinson et al⁷⁸ conducted an in vitro analysis that showed significant air-handling differences between the oxygenators from four different manufacturers. They demonstrated how a sonar-based system, the embolus Detection and Classification System (EDAC; Lunar Technology, Blacksburg, VA), could be used to evaluate perfusion systems with regard to their ability to handle gas entrained in the circuit.

A new nonporous membrane surface composed of poly-(4-methyl-1-pentene) (PMP) fibers has shown improved diffusion compared with the conventional polypropylene (PPL) hollow fibers. PPL affords improved durability and biocompatibility when used for long-term support^79,80 and for routine CPB.⁸¹ Although oxygen and carbon dioxide gas exchange are comparable between polypropylene hollow fibers and the PMP nonporous fibers, it is important to note that the transfer of volatile anesthetic agents is not the same. Wiesenack et al⁸² demonstrated that the PMP fibers allow only minimal transfer of isoflurane compared with the currently used PPL microporous hollow-fiber oxygenators. During long-term support, it is not uncommon for PMP oxygenators to develop breaches in the surface that lead to plasma leaks after 40 to 90 hours of use, whereas the PPL fibers tend to be more robust and are not prone to plasma leaks, and continue to transfer oxygen and carbon dioxide for many days.

Venous and Cardiotomy Reservoirs

There are two general categories for venous reservoirs: open and closed systems (Box 29-5). Open systems have a hard polycarbonate venous reservoir and usually incorporate a cardiotomy reservoir and defoaming compartment (see Figure 29-11). Closed systems are collapsible polyvinylchloride bags that have a minimal surface area and often a thin single-layer screen filter, and they require a separate external cardiotomy reservoir for cardiotomy suction (Figure 29-12). Filters and defoaming compartments in the venous reservoir and air-trapping ports located at the highest level of the blood flow path within the oxygenator are areas designed to allow passive removal of air. Studies that have examined the air-handling capabilities of oxygenators have shown that all of the currently available oxygenators do not sufficiently remove GME when challenged with air in the inflow.^77,78