Surgical procedures involving cardiopulmonary bypass

Venous Air in the Bypass Circuit: A Source of Arterial Line Emboli Exacerbated by Vacuum-Assisted Drainage Timothy W. Willcox, CCP(Aust), Simon J. Mitchell, MB, ChB, and Des F. Gorman, PhD Cardiothoracic Surgical Unit, Green Lane Hospital, Auckland, New Zealand Background. Arterial emboli cause neurocognitive deficits in cardiac surgical patients. Carotid artery emboli, detected ultrasonically, have been observed after venous air entrainment into the cardiopulmonary bypass circuit. We investigated in vitro the extent to which venous air affected emboli detected in the arterial line downstream from a 40- m filter. Methods. Using salvaged clinical cardiopulmonary bypass circuits, fixed volumes of air were introduced into the venous return line at unrestricted rates and at fixed rates using gravity venous drainage and vacuum-assisted venous drainage. Emboli counts were recorded distal to the arterial line filter using a 2-MHz pulsed-wave Doppler monitor. Emboli counts were similarly recorded after the introduction of carbon dioxide into the venous return line instead of air. Results. The number of emboli rose with increasing volumes of entrained venous air (p < 0.001), and there was an almost tenfold increase with vacuum-assisted venous drainage (p < 0.0001) compared with gravity venous drainage. Venous air was entrained at a significantly faster rate under vacuum-assisted venous drainage (p < 0.0001). When the entrainment rate of venous air was fixed, the difference in emboli numbers recorded for gravity and assisted venous drainage was not significant. There was a significant reduction in arterial line emboli when carbon dioxide rather than air was entrained under both vacuum-assisted and gravity drainage (p < 0.001). Conclusions. Entrained venous air during cardiopulmonary bypass is a potential hazard, particularly during vacuum-assisted venous drainage. Every effort should be made to avoid venous air entrainment. (Ann Thorac Surg 1999;68:1285 9) 1999 by The Society of Thoracic Surgeons Surgical procedures involving cardiopulmonary bypass (CPB) are associated with more postoperative neurocognitive deficits than those without CPB [1]. Emboli, both gaseous and solid, are an established cause of such deficits [2], and the extent of postoperative neurocognitive deficits worsens as numbers of emboli increase [3]. There are many sources of emboli during CPB [2]. During a recent study involving open heart operations [4], increases in carotid artery Doppler signals typical of gaseous emboli were associated with air seen in the venous return line. Removal of venous line air is considered an advantage of hard-shell venous reservoir designs in CPB circuits [5]. We [6] have previously reported the variable ability of hard-shell venous reservoirs to remove venous line bubbles but the extent to which these still appear in the line downstream from the arterial filter has not been investigated. Vacuum-assisted venous drainage (VAVD) is a relatively new clinical technique. It allows the use of smallerdiameter venous return cannulas for given blood flow rates, thus improving access in the surgical field, especially during minimally invasive surgical procedures. However, the method is based on a vacuum applied to Accepted for publication April 13, 1999. Address reprint requests to Mr Willcox, Clinical Perfusion, Green Lane Hospital, Green Lane West, Auckland 3, New Zealand; e-mail: timw@ahsl.co.nz. the venous reservoir, and this may influence behavior of gaseous emboli in the circuit. The present in vitro study was undertaken to measure the extent to which air introduced into the venous return line affects the numbers of emboli detected in the arterial line downstream from a 40- m filter. The effect of venous air entrainment under VAVD conditions and the impact of changing the gas composition of bubbles entering the venous line were also studied. Material and Methods Experimental Circuit The experiments were performed in vitro using salvaged clinical circuits. After CPB, the arterial line was clamped and the remaining prime recirculated at 1.0 L/min. Each circuit consisted of the following: a venous line with an internal diameter of 12.5 mm ( 1 2 inch); the Maxima Forté hard-shell venous reservoir (Medtronic Inc, Anaheim, CA); a roller pump on a Stockert CAPS or S3 heart-lung machine (Stöckert Instrumente, Munich, Germany); the Maxima Forté hollow-fiber oxygenator (Medtronic Inc); and a Bentley AF-1040D 40- m arterial line filter (Baxter Healthcare Corp, Irvine, CA). The outer arterial reservoir shell of a Bentley Ben 10 bubble oxygenator (Baxter Healthcare Corp) was positioned at a standard height above the venous reservoir and used to simulate the 1999 by The Society of Thoracic Surgeons 0003-4975/99/$20.00 Published by Elsevier Science Inc PII S0003-4975(99)00721-3

1286 WILLCOX ET AL Ann Thorac Surg VENOUS AIR DURING CPB 1999;68:1285 9 patient. The venous return line was connected to the patient reservoir outlet port. The arterial line cannula (24F straight Argyle; Baxter Healthcare Corp) was fixed to the inside of the patient reservoir at the 1,000-mL level obliquely opposing the direction of the outlet port. The circuit prime volume was restored with group O dateexpired resuspended red cells and Plasmalyte 148 (Baxter Healthcare Corp) to a hematocrit of 23% to 26%. The prime was recirculated at 3.5 L/min and a line pressure of 100 mm Hg with the volume of the venous reservoir maintained at 1,000 ml and the patient reservoir at 1,500 ml. The arterial filter purge was open, and the oxygenator was ventilated with a sweep gas flow rate of 2.5 L/min at an inspired oxygen fraction of 0.50. Doppler Emboli Detection In each experiment, the circuit was monitored with a Flowlink 300 color-flow Doppler (Rimed, Tel Aviv, Israel), interfaced with a custom-built analog signal processor that was designed to count emboli signals as described previously [7]. The Doppler monitor was operated in the 2-MHz pulsed-wave mode. A purpose-built clamp was used to mount the Doppler probe on the circuit tubing so that the probe faced the direction of blood flow at 45 degrees. The space between the probe and the tubing was filled with ultrasonic gel. In all experiments, the probe was placed on the arterial line downstream from the filter. Beam depth was set at 27 mm, machine power at 40%, and gain at 6 for optimal audible signal and color-flow display. A sensitivity control on the signal processor was adjusted to obtain correlation between the audible and visible signals typical of emboli and the count from the signal processor. We cite arterial line emboli counts but recognize that there are factors that can confound emboli signal counters [8]. A baseline arterial-line emboli count was recorded over 180 seconds prior to each experimental intervention. The net experimental arterial-line emboli count was obtained in all cases by subtracting the baseline count from the count done after entrainment of venous gas. Venous Air Entrainment A venous air injection site was created by inserting a standard 12.5 mm ( 1 2 inch) Luer-Lok connector into the venous line beyond the venous occluder and fitting an injection plug (B Braun, Melsungen, Germany) into the Luer-Lok. A 37.5-mm (1 1 2-inch) stainless steel 21-gauge hypodermic needle was introduced into the venous line through the injection plug, and a three-way stopcock was attached to the needle. Opening the stopcock enabled entrainment of venous air. The volume entrained was limited by connecting a 1.0-L blood transfer pack (Baxter Healthcare Corp), already filled with the desired volume of gas, to the stopcock. In this mode, there was no restriction on the rate at which gas could be entrained. In experiments requiring control of both gas volume and entrainment rate, gas was introduced from a prefilled syringe at the desired rate. Experiments We conducted the following four experiments: increasing the volume of entrained venous air during gravity venous drainage (GVD) (experiment 1); increasing the volume of entrained venous air during VAVD (experiment 2); a fixed rate of venous air entrainment during VAVD and GVD (experiment 3); and entrainment of carbon dioxide (CO 2 ) into the venous line (experiment 4). For all experiments, a p value of less than 0.05 was considered significant. EXPERIMENT 1. We tested the hypothesis that entrained venous air is incompletely removed by the CPB circuit under gravity drainage and that greater volumes of entrained venous air result in larger numbers of arterial line emboli. Twenty-five milliliters of air was entrained at an unrestricted rate, and the entrainment time was recorded. The number of arterial line emboli counted was recorded over 180 seconds beginning at initiation of air entrainment. The procedure was repeated for 50, 75, and 100 ml of entrained air, and this sequence was repeated three times in each of three separate circuits (a total of nine studies for each volume of venous air). Although air was entrained at an unrestricted rate, the time for entrainment at each volume was virtually identical for each experimental run. One-way analysis of variance was used to assess the significance of any trend in the mean count of arterial line emboli with the increasing volumes of entrained venous air. EXPERIMENT 2. We tested the hypothesis that venous air entrained during VAVD results in larger numbers of emboli in the arterial line than equal amounts of air entrained during GVD. The circuit was converted to VAVD by sealing all luers and inlet ports of the venous reservoir and applying a 60 mm Hg vacuum to the reservoir vent port. The vacuum was controlled using a thoracic low-vacuum regulator (Clements Medical Equipment PTY Ltd, Sydney, Australia), and the pressure within the venous reservoir was continuously monitored using a pressure transducer module on the heart-lung machine. The experimental protocol was identical to that in experiment 1, except that only two volumes of air (25 and 50 ml) were tested. An unpaired t test was used to compare the mean arterial-line emboli count recorded after entrainment of equal volumes of air during VAVD and GVD. The time for these volumes of air to be entrained during both VAVD and GVD were similarly compared. EXPERIMENT 3. We tested the hypothesis that the rate of venous air entrainment accounts for any difference in arterial line emboli count recorded under conditions of vacuum-assisted and gravity drainage. A disposable syringe was filled with 50 ml of air and attached to the three-way stopcock on the venous air injection site. Injection of the air at 2 ml/s and a 180-second arterialline emboli count recording were then begun simultaneously. This was repeated six times in each condition in two circuits. An unpaired t test was used to compare the mean arterial-line emboli count recorded during vacuum-assisted drainage and gravity drainage at a fixed rate of entrainment. EXPERIMENT 4. We tested the hypothesis that entrainment of CO 2 instead of air reduces the resultant arterial-line emboli count under both VAVD and GVD. Arterial

Ann Thorac Surg WILLCOX ET AL 1999;68:1285 9 VENOUS AIR DURING CPB 1287 Fig 1. Mean arterial-line emboli count ( standard deviation) downstream from 40- m filter with increasing volumes of entrained venous air under gravity venous drainage. Fig 2. Mean arterial-line emboli count ( standard deviation) downstream from 40- m filter with entrained venous air under gravity venous drainage (GVD) and 60 mm Hg vacuum-assisted venous drainage (VAVD) (* p 0.0001 compared with GVD.) emboli counts were recorded over 180 seconds from the onset of gas entrainment (unrestricted rate) nine times under the following conditions: first using GVD entrainment of 75 ml of air and 100 ml of air and then 75 ml of CO 2 and 100 ml of CO 2 ; and second using VAVD entrainment of 25 ml of air and 50 ml of air and then 25 ml of CO 2 and 50 ml of CO 2. An unpaired t test was used to compare arterial line emboli count after entrainment of air and CO 2 at both gas volumes under each drainage condition. Results Experiment 1 The mean arterial-line emboli count recorded after entrainment of venous air during GVD is plotted in Figure 1. Emboli were detected in the arterial line downstream from the filter after every entrainment of air into the venous return line. The count increased with increasing volumes of entrained venous air ( p 0.001). Experiment 4 The mean arterial-line emboli count after the entrainment of equal volumes of air and CO 2 under both gravity drainage and vacuum-assisted drainage is shown in Figure 4. At all volumes, under both VAVD and GVD, there was a significant reduction in the arterial line emboli count after entrainment of CO 2 into the venous line ( p 0.001). Comment Gaseous emboli entering the arterial circulation of patients during CPB can originate from the surgical field or the extracorporeal circuit [8]. Sources of arterial emboli from the CPB circuit include bubble oxygenators, cavitation from roller pumps, cardiotomy suction [8], and venous reservoirs [7]. Reductions in gaseous emboli from the CPB circuit have Experiment 2 The mean arterial-line emboli count after entrainment of 25 ml and 50 ml of venous air during GVD and VAVD is plotted in Figure 2. For both volumes of entrained air, an almost tenfold increase in arterial line emboli count was seen under VAVD (p 0.0001). The volume of venous air entrained under VAVD was restricted to 50 ml, as greater volumes resulted in an arterial line emboli count too high to be reliably recorded. The mean time for air to be entrained into the venous line under GVD and VAVD is shown in Figure 3. Air was entrained into the venous line at a significantly faster rate under VAVD (p 0.0001). Experiment 3 The mean arterial-line emboli count after entrainment of 50 ml of venous air at a fixed rate was higher under VAVD than under GVD (266 113.1 [ standard deviation] versus 148.8 73.6). However, the difference did not reach our chosen level of significance. Fig 3. Mean time ( standard deviation) for fixed volumes of venous air to enter the venous line under gravity venous drainage (GVD) or 60 mm Hg vacuum-assisted venous drainage (VAVD) (* p 0.0001 compared with GVD.)

1288 WILLCOX ET AL Ann Thorac Surg VENOUS AIR DURING CPB 1999;68:1285 9 Fig 4. Mean arterial-line emboli count ( standard deviation) downstream from 40- m filter after entrainment of either air or CO 2 into the venous line under gravity venous drainage (GVD) or vacuum-assisted venous drainage (VAVD) (* p 0.001 compared with air.) been achieved by the use of arterial line filters [9] and by improvements in oxygenator and venous reservoir design [6, 10]. The use of membrane oxygenators and arterial line filters has increased in recent years [11]. The contribution of venous air to arterial line emboli has not been investigated. However, small amounts of gas injected into the venous return line result in microemboli downstream from a membrane oxygenator proximal to the arterial filter, as does rapid injection of well-degassed saline solution into the venous sampling port of a venous reservoir [12]. Venous air during CPB is common. The perceived ability of venous reservoirs to remove such air during CPB has led to an acceptance that venous air is consequently benign. Typically, when a hard-shell venous reservoir is used, venous cannulas are not deaired, and initiation of bypass is associated with a bolus of air returning to the CPB circuit. During CPB, air can enter the venous line around the pursestring suture of a venous or retroplegia cannula, especially during retraction of the atrial wall. A single atrial cannula can be used in procedures for some congenital conditions to vent blood, and coincidentally air, from the left heart across a patent foramen ovale or an excised atrial septum. New deairing techniques involve venting air from the left heart and aortic root directly to the venous return line [13]. This study demonstrates that small amounts of venous air result in emboli distal to a 40- m filter in an arterial line when conventional gravity drainage is used. Further, the numbers of arterial line emboli increase as greater volumes of air are entrained (see Fig 1). Whereas we used fixed volumes of venous air, the volume of air entering the venous line clinically is unrestricted. Vacuum-assisted venous drainage is a clinical technique used to enhance venous drainage to the perfusion circuit when narrower cannulas and tubing diameters are used. In a study [14] using an animal model, reductions in circuit priming volume were achieved during VAVD, with the result that the CPB circuit could be simplified and miniaturized. Potential applications for VAVD include minimally invasive surgical procedures, pediatric operations, and femoral-femoral bypass. We have shown that VAVD profoundly increases the number of emboli likely to reach the patient when venous air is entrained into the CPB circuit. Vacuumassisted venous drainage produced an almost tenfold increase in the arterial line emboli count compared with GVD after entrainment of venous air (see Fig 2), despite use of the Medtronic Maxima Forté reservoir, which we [6] have demonstrated to be a comparatively effective venous air filter. Because of the influence of a vacuum, the time for equivalent volumes of air to enter the venous return line was significantly shorter under VAVD than GVD (see Fig 3). However, when the rate of air entry was fixed, the arterial line emboli count under VAVD was not significantly greater than that for GVD. It appears that the passage of venous air through a CPB circuit is deliveryrate dependent. Leaving the venous line unprimed prior to initiation of CPB under VAVD to reduce circuit prime volume has also been advocated [15]. This practice will result in up to 200 ml of air being rapidly delivered to the venous reservoir, potentially increasing arterial line emboli at the onset of CPB. Consequently, deairing the venous cannula during connection to the CPB circuit is recommended. Carbon dioxide flooding of the surgical field to reduce air embolism was first used during CPB in the 1960s [16], and there is renewed interest in this technique [17]. It is reasonable to expect that if CO 2 rather than air were entrained, then given the transit time of these bubbles through the CPB circuit and the increased solubility of CO 2, fewer emboli would be detected in the arterial line. The reduction in arterial line emboli count seen in experiment 4 after entrainment of CO 2 rather than air supports this hypothesis under both VAVD and GVD (see Fig 4). However, it is notable that after 50 ml of CO 2 was entrained under VAVD, the arterial line emboli count was significantly greater than that for 100 ml of air entrained under GVD. Thus, although CO 2 is advantageous, it does not offset the potentially deleterious effect of VAVD on arterial line emboli. These results illustrate the potential hazard of venous air, particularly during VAVD, and suggest that every effort should be made to eliminate this problem when it is detected. Excessive negative pressure caused by gravity can be reduced by partially clamping the venous return line [5], which in turn will reduce the amount of air entrained. However, control of venous air entrainment usually requires intervention by the surgeon. In light of our findings, the priority for such interventions should be emphasized. The support of the staff of the Department of Clinical Perfusion at Green Lane Hospital is greatly appreciated. This work was supported by grants from the English Freemasons of New Zealand and the Health Research Council of New Zealand.

Ann Thorac Surg WILLCOX ET AL 1999;68:1285 9 VENOUS AIR DURING CPB 1289 References 1. Shaw PJ, Bates D, Cartlidge NEF, et al. Neurologic and neuropsychological morbidity following major surgery: comparison of coronary artery bypass and peripheral vascular surgery. Stroke 1987;18:700 6. 2. Stump DA, Roger AT, Hammon JW, Newman SP. Cerebral emboli and cognitive outcome after cardiac surgery. J Cardiothorac Vasc Anesth 1996;10:113 9. 3. Pugsley W, Klinger L, Paschalis C, Treasure T, Harrison M, Newman S. The impact of microemboli during cardiopulmonary bypass on neuropsychological functioning. Stroke 1994;25:1393 9. 4. Mitchell SJ, Pellett O, Gorman DF. Cerebral protection by lidocaine during cardiac operations. Ann Thorac Surg 1999; 67:1117 24. 5. Hessel AE. Cardiopulmonary bypass circuitry and cannulation techniques. In: Gravlee GP, Davis RF, Utley JR, eds. Cardiopulmonary bypass principles and practice. Baltimore: Williams & Wilkins, 1993:55 92. 6. Mitchell SJ, Willcox T, Gorman DF. Bubble generation and venous air filtration by hard-shell venous reservoirs: a comparative study. Perfusion 1997;12:325 33. 7. Mitchell SJ, Willcox T, McDougal C, Gorman DF. Emboli generation by the Medtronic Maxima hard-shell venous reservoir in cardiopulmonary bypass circuits: a preliminary report. Perfusion 1996;11:145 55. 8. Kurusz M, Butler BD. Embolic events and cardiopulmonary bypass. In: Gravlee GP, Davis RF, Utley JR, eds. Cardiopulmonary bypass principles and practice. Baltimore: Williams & Wilkins, 1993:267 90. 9. Treasure T. Interventions to reduce cerebral injury during cardiac surgery the effect of arterial line filtration. Perfusion 1989;4:147 52. 10. Pearson DT, McArdle B, Poslad SJ, Murray A. A clinical evaluation of the performance of one membrane and five bubble oxygenators: gas transfer and gaseous microemboli production. Perfusion 1986;1:15 26. 11. Kurusz M. Lessons from perfusion surveys. Perfusion 1997; 12:221 7. 12. Hill AG, Groom AC, Vinansky RP, Speir AM, Macmanus Q, Lefrak EA. Gaseous microemboli and extracorporeal circulation. Proc Am Acad Cardiovasc Perfusion 1986;7: 131 7. 13. Milsom FP, Mitchell SJ. A dual-vent left heart deairing technique markedly reduces carotid artery microemboli. Ann Thorac Surg 1998;66:785 91. 14. Taketani S, Sawa Y, Masai T, et al. A novel technique for cardiopulmonary bypass using vacuum system for venous drainage with pressure relief valve: an experimental study. Artif Organs 1998;22:337 41. 15. Vacuum assisted venous return procedure. Irvine, CA: Baxter Healthcare Corporation, 1998. 16. Ng WS, Rosen M. Carbon dioxide in the prevention of air embolism during open-heart surgery. Thorax 1968;45: 586 7. 17. Webb WR, Harrison LH Jr, Helmcke FR, et al. Carbon dioxide field flooding minimizes residual intracardiac air after open heart operations. Ann Thorac Surg 1997;64: 1489 91. INVITED COMMENTARY Willcox and colleagues have described an elegant in vitro study examining detection of microbubbles in the cardiopulmonary bypass (CPB) systemic flow line after venous line injections of air when using vacuum-assisted venous drainage (VAVD). The CPB circuit is a conventional one, incorporating a membrane oxygenator and arterial screen filter, both of which have been shown to reduce detectable microbubbles to negligible levels when used with gravity siphon venous drainage. VAVD is a relatively new concept in the management of CPB and has gained popularity with minimally invasive, reoperative, pediatric, and femoral-femoral CPB procedures. The study raises some important questions and potential concerns. There remain, however, some key unresolved issues that either were not part of the Willcox study or that are due to technical shortcomings inherent in theirs and other in vitro evaluations of microbubble activity in extracorporeal systems. The study describes a much greater amount of embolic signals in the systemic flow line with VAVD, when compared with conventional CPB siphon venous drainage, when venous line air is present. This implies a change in the physical characteristics of the circuit and/or behavior of the bubbles. These types of critical issues need to be addressed and study designs need to consider such salient features such as circuit pressures across the membrane oxygenator, arterial line filter, between the filter and the pump, as well as reservoir bubble-handling characteristics under VAVD. The authors employed color-flow Doppler to detect the microbubbles. This device is not customarily used in these applications, and its reliability and accuracy are not well understood or described. Again, while this does not necessarily diminish the message and concern raised by the study, it does detract from the perceived relevance and accuracy of the results. Other issues such as protein components in the circuit perfusate that may affect the stability and characteristics of microbubbles, as well as unique in vitro circuit designs, need to be well understood regarding their effects on study outcomes and conclusions. The Willcox study raises serious concern as to the relative risk of VAVD when bubbles are present in the venous line. Their study did demonstrate that with VAVD, systemic flow line microbubbles were much more prevalent, yet it raises more questions than were answered. Clearly, this topic, now openly raised, needs further careful clinical evaluation. Bruce D. Butler, PhD Mark Kurusz, CCP Department of Anesthesiology University of Texas-Houston Medical School 6431 Fannin, MSMB 5.020 Houston, TX 77030 e-mail: bbutler@anes1.med.uth.tmc.edu. 1999 by The Society of Thoracic Surgeons 0003-4975/99/$20.00 Published by Elsevier Science Inc PII S0003-4975(99)00808-6