Flooding the Surgical Field With Carbon Dioxide During Open Heart Surgery Improves Segmental Wall Motion

Abstract:

Air embolization to the coronary arteries is a common cause of myocardial ischemia during open heart surgery. Carbon dioxide emboli may be absorbed faster than air emboli. In this randomized, double blind, placebo-controlled trial, we determined that flooding the surgical field with carbon dioxide is associated with improved myocardial function assessed by trans-esophageal echocardiography. Forty-three valve surgeries were randomized to insufflation of 6 L/min of carbon dioxide or placebo through a Jackson Pratt drain into the pericardium during cardiopulmonary bypass. During rewarming, as pulse pressure rose above 10 mmHg, two observers graded severity of bubbles in the left heart. Two other observers evaluated wall motion in the transgastric midpapillary short axis view of the left ventricle using transesophageal echocardiography. Compared with baseline average scores among all walls (carbon dioxide, 1.42 ± 0.46; placebo, 1.39 ± 0.45), worsening of wall motion was less at 1 minute in the carbon dioxide (1.60 ± 0.62) than in the placebo group (1.95 ± 0.54; p = 0.0266). Better wall motion tended to persist in the carbon dioxide group at 10 (1.58 ± 0.59 vs. 1.77 ± 0.6) and 60 minutes (1.61 ± 0.45 vs. 1.66 ± 0.58). Particularly, the inferior wall tended toward transiently better function in the carbon dioxide group (at baseline and 1, 10, and 60 minutes: placebo, 1.62 ± 0.72, 2.68 ± 0.79, 2.48 ± 0.95, 2.10 ± 0.9 vs. 1.88 ± 0.97, 2.33 ± 1.1, 2.18 ± 0.96, 2.20 ± 0.94). Preoperative characteristics, length of bypass, anesthesia time, hospitalization, and intensive care unit stay were not different. We recommend administration of carbon dioxide because it may improve myocardial function. We describe how to avoid adverse effects of giving carbon dioxide by filtering the supply, continuously managing its level during bypass, increasing sweep speeds, continuously analyzing the in-line blood gas, and avoiding suctioning gases in the field into the cardiotomy reservoir.

Despite standard surgical maneuvers to evacuate air from the heart after open heart surgery, significant amounts remain and embolize to the coronary arteries (1–3). Standard surgical maneuvers include filling and venting the left ventricle, ballottement of the heart and lungs, ventilation, Trendelenberg, and needle aspiration. Air embolizes in coronary surgery, albeit less so than in open-heart surgery (2). Flooding the surgical field with carbon dioxide (CO₂) reduces the incidence of intracardiac air by 85%, possibly because of the density and solubility of CO₂ (4). Specifically, the density of CO₂ is 1.5 times that of air, so that CO₂ preferentially fills the dependent parts of the surgical field (5). In addition, CO₂ is 360 times more soluble than air in blood, thus reducing the size, duration, and physiologic impact of emboli (6,7). Indeed, when CO₂ is used, most transesophageal echocardiographic (TEE)-detected intracardiac air disappears within a few minutes (4). Despite evidence that flooding the field with CO₂ reduces the amount of air in cardiac chambers, we are not aware of any studies showing improved myocardial outcome (8).

Air emboli in the coronary circulation cause myocardial ischemia or infarction that can be detected by TEE as segmental wall motion abnormalities (SWMAs) (9). SWMAs are more common in the presence of extensive air, as defined by TEE documentation of more than 30 bubbles in the left atrium or ventricle in the four chamber view (9). In a study of 112 patients undergoing valve surgery, extensive air was associated with a 30% incidence of SWMA, whereas minimal air was associated with a 10% incidence of SWMA (9). Another study showed SWMA are 30 times more common in the presence of extensive air (10). When extensive air is present in the cardiac chambers and a new SWMA occurs, 70% of these new SWMA are in the right coronary artery (RCA) distribution because it is positioned at the anterior portion of the aorta when a patient is supine (9). The RCA supplies the inferior wall of the left ventricle and most of the right ventricle and is important in separation from bypass.

Approximately 10% of patients with extensive air and associated SWMA have difficulty weaning from bypass because of cardiac failure (3,9). Although many of these SWMA resolve, nearly 40% persist for hours (10). Postoperative myocardial infarction occurs five times more often when intraoperative SWMA are detected (11,12). Duration of vasopressor use and intensive care unit stay is prolonged whenever myocardial dysfunction is prolonged, as evidenced by SWMA or serum cardiac troponin I (13,14). Thus, reducing the amount of air with CO₂ may significantly reduce myocardial dysfunction and its consequences after open heart surgery.

The practice of flooding the surgical field with CO₂ has been performed safely since 1957 at many cardiac surgical centers (5). In addition, it has been used during hip surgery to reduce air emboli to the lungs or, through a patent foramen ovale, into the left atrium. Only two case reports challenge the safety of this technique in cardiac surgery. Both cases resulted in elevated arterial partial pressure of CO₂ during cardiopulmonary bypass (CPB), and both resolved promptly by changing the position of the cardiotomy suction, bypass flow rate, or gas flow to the oxygenator (5,15,16).

MATERIALS AND METHODS

After approval from the University of California, San Francisco’s Committee on Human Research, 43 consecutive subjects gave informed consent and were enrolled into this randomized double blind, placebo-controlled trial. Inclusion criteria were age above 18 years, elective open -heart surgery, and (to standardize deairing techniques) operations performed by one surgeon. Bicaval cannulation was always performed, but vacuum-assist venous return was used in less than 20% of the cases (toward the end of the study period) and only when directed by the surgeon. All patients received standard hypothermia and rewarming protocols. After unclamping the aorta, the surgeon created a small hole in the aortic root to allow egress of air. The deairing technique was agitating the lungs during a sustained inflation, rotating the table from side to side with the patient in head down position, and left ventricular venting guided by TEE. Anterograde and retrograde cardioplegia was used in all cases. Whenever a CABG was performed at the time of the valve surgery, side biting aortic clamps were used for proximal anastomoses.

All investigators and clinicians were blinded to group assignment. An unblinded research assistant drew the randomization envelope, masked the identity of the H cylinders of CO₂ and medical grade air, turned on the appropriate gas at 6 L/min for the duration of CPB and monitored the infrared capnograph from a sterile tubing in the open chest cavity just before the conclusion of CPB.

For each surgery, the gas flow meter was connected to a new piece of sterile CPB tubing with a bacterial and particle filter (Intersept Custom Tubing Pack; Medtronic, Minneapolis, MN). The surgeon connected this filter to a sterile Jackson Pratt tube (Allegiance Healthcare Corp., San Jose, CA), which he placed inside the pericardium but outside the heart (4). An arterial blood gas was measured every 30 minutes; continuous calibrated in-line arterial blood gas analysis was performed with the use of a CDI 500 (Terumo Cardiovascular, Ann Arbor, MI). Sweep speed was increased by the perfusionist when arterial CO₂ was greater than 60 mmHg. The percentage of CO₂ in the field was confirmed by a infrared capnograph, before termination of gas delivery, using a validated calculation from percent oxygen. Percent CO₂ = (20.95% inspired oxygen) × 4.77, where inspired was not from the usual location of an endotracheal tube but rather was from a sterile extension tubing attached to this capnograph, when temporarily placed into the pericardial space (4,5).

To determine if the number of bubbles in cardiac chambers correlated with the primary outcome of SWMA, the number of bubbles per chamber was evaluated in a standard four chamber TEE view (4,9). Most emboli occur after aortic unclamping because the heart resumes ejection of blood, after which the bubbles are cleared exponentially(14). Thus, after pulse pressure exceeded 10 mmHg, we counted the maximal number of bubbles of one frozen frame at four time intervals (0–1, 2–2.5, 4–4.5, and 8–8.5 minutes). Grade was assigned, as is standard in the literature, either for either the left atrium or ventricle, whichever was worst; grade 0, uninterpretable; grade 1, 0 bubbles; grade 2, 1–10 bubbles; grade 3, 11–29 bubbles; grade 4, 30 or more bubbles; grade 5, streaming (aggregates) or a pool.

Intraoperatively, one TEE cross-section was acquired for 6 seconds at four time-points by an investigator and stored on S-VHS videotape. The transgastric midpapillary short axis view of the left ventricle was used to assess SWMA prebypass as a baseline and again after 1, 10, and 60 minutes after separation from bypass. Cross-sections were analyzed at a later time by two experts who were blinded to patient identity and time of image acquisition. These standard grades are as follows: grade 1, normal; grade 2, minimal hypokinesis (i.e., good wall thickening, radius of contraction of endocardium to imaginary floating center >30%); grade 3, severe hypokinesis (i.e., slight wall thickening, radius of contraction of endocardium to imaginary floating center 10–30%); grade 4, akinesis (i.e., no wall thickening, no radius of contraction; grade 5, dyskinesis (wall is thin, bulges at systole) for the walls: septal, inferior, posterior, lateral, anterior, anteroseptal.

For sample size determination, we estimated that carbon dioxide treatment reduces the incidence of intracardiac air by 85% (4). The only large study of valve surgery found extensive air was associated with a 30% incidence of SWMAs, whereas minimal air was associated with a 10% incidence of SWMAs (9). Assuming CO₂ would reduce the incidence of extensive air from 38% to 15%, 57 patients would be required per group to have an 80% chance of detecting a 60-minute difference at a significance of 0.05, based on previous studies (9,10). Our results indicated an incidence of SWMA that was unexpectedly several times higher than in these studies. Thus, we stopped after our first interim analysis, which showed a significant difference in SWMA.

Categorical variables (e.g., diabetes) were compared between the treatment and placebo groups using Fisher exact test. Numerical variables (e.g., change in average SWMA) were compared using the Mann-Whitney test. Univariate linear regression was performed for inferior wall change to test the hypothesis that the inferior wall was affected by CO₂ more than the other walls. The changes of inferior SWMA score and the changes of total SWMA score within the treatment and within the placebo groups from baseline to the three later time periods were analyzed by a Wilcoxon signed-rank test. Interobserver reliability for the two readers was analyzed with scatter-plots and interrater correlations of the average SWMAs. Secondary outcomes evaluated include duration of bypass, operating room, intensive care, and hospital stay. Linear regression models were constructed to adjust for age, and cross-clamp time, because these factors might affect bypass times and SWMAs. p values less than 0.05 were considered significant.

RESULTS

Forty-three patients were randomized. Twenty-one (treatment group) received CO₂ and 22 (placebo group) received air. Baseline demographics, including sex, age, diabetes, alcohol consumption, presence or absence of coronary artery bypass, and CPB time, aortic cross-clamp time, and type of surgery were not different between the two groups (Table 1). Rate of resternotomy was significantly higher in the treatment group. Bubble count was higher in the placebo group.

Table 1.

Demographics.

Demographic	Mean for Air	Mean for CO2	p
Age	60	61	.77
Alcohol	2.9	4.1	0.9
Males	12	13	.64
Females	10	8
Diabetes	4	2	.66
CABG + valve	6	8	.53
Aortic	10	14	.22
Mitral	11	7	.36
ASD	1	0	1.0
Reoperation	1	7	.02
Cross-clamp	92	81	.64
Surgical field percent oxygen	21	4.11	.0004
Bubbles at 0–1 min	4.66	2.40	.03
Bubbles at 2–2.5 min	2.92	2.20	.71
Bubbles at 4–4.5 min	2.83	2.10	.50
Bubbles at 8–8.5 min	1.83	0.40	.15
Bubbles average	3.06	1.78	.10

Worsening of average SWMA score from baseline to 1 minute after bypass was less in the CO₂ group compared with the placebo group (p = 0.0266; Table 2; Figure 1). However, between 1 and 60 minutes, the placebo group returned to baseline; the placebo group significantly improved by an average of 0.2857 points per wall compared with the treatment group, which worsened by 0.0044 points (p = .04). Prebypass baseline inferior and average SWMAs were compared with 1, 10, and 60 minutes after bypass. Inferior wall scores started at a similar baseline (placebo, 1.62 vs. treatment 1.88; Figure 1). As an independent variable in a linear regression model, inferior wall SWMAs tended to be worse in the placebo group. If one expert could not interpret a wall but the other expert could, the interpretation used was from the one who could interpret that wall. If one wall could not be interpreted by either expert, an average of the other five walls was taken as the reading for that missing wall, but this only happened in 51 of 1032 walls (i.e., 4.9% of walls). Technical difficulty, often from shadowing by other structures, creates missing walls. Clinically important and correctly interpreted SWMA often occur in two adjacent walls. Averaging the other five walls reflects the overall contractility of the heart and keeps the statistics straightforward. The frequency of uninterpretable walls was not different between the placebo and treatment groups. Secondary outcomes of duration of CPB, anesthesia, intensive care unit, and hospitalization were not different between groups (Table 3).

Table 2.

SWMAs.

Change	Mean for Air	p for Air	Mean for CO2	p for CO2
Baseline SWMA to 1 min	.56	.0002	.18	.08
Baseline SWMA to 60 min	.27	.047	.19	.06
Baseline inferior to 1 min	1	.003	.45	.02
				p between
				Air and CO2
Baseline SWMA to 60 min				.50
SWMA at 1 minute to 60 min				.04
Baseline inferior to 1 min				.10
Baseline inferior to 10 min				.09
Baseline inferior to 60 min				.42

Figure 1.

Mean SWMA scores are listed above each error bar. Open blocks represent average SWMA score. Solid blocks represent inferior wall score. All error bars represent ± SD; SD is 0.45, 0.46, 0.54, 0.62, 0.6, 0.59, 0.58, and 0.45, respectively, for each average SWMA bar, moving left to right. SD is 0.72, 0.97, 0.79, 1.1, 0.95, 0.96, 0.9, and 0.94, respectively, for each bar, moving left to right for each inferior wall bar.

Table 3.

Secondary outcome variables.

Variable	Air mean ± SD	CO2 mean ± SD	p
Hospital stay in days	9.9 ± 5.6	14.3 ± 21.4	.38
Anesthesia minutes	370 ± 65	402 ± 84	.50
Days in intensive care	3.6 ± 2.7	3.8 ± 6.7	.27
Minutes of bypass	129 ± 52	117 ± 33	.36

Interrater variability for bubble counts was calculated to have a Pearson correlation coefficient of 0.99. Interrater variability for SWMAs was calculated to have a Pearson correlation coefficient of 0.67 prebypass, of 0.67 at 1 minute, of 0.83 at 10 minutes, and of 0.66 at 60 minutes. No bias was found toward one observer or the other, thus justifying the method of averaging each wall score between the observers.

DISCUSSION

Two prior studies of open heart surgery showed that the number of air emboli is reduced by flooding the surgical field with CO₂, but neither showed expected clinical benefits, such as improved immediate postbypass myocardial function, when there are fewer covariates (4,17). Because many patients require pacing and/or defibrillation during rewarming, we elected not to measure electrocardiographic changes. For the first 28 subjects, we also studied change in Trailmaking A and B and Rey Auditory Verbal Memory learning at preoperative, 3 days postoperatively, and 2 weeks postoperatively, but found no trends in their transient decline of scores. After collecting pilot data, we focused our study on SWMAs because too many covariates determine late neurologic function. By examining TEE in the four chamber view, the most common view for detecting air, we avoided counting other emboli, such as fat from the aorta or bypass circuit. We also tested a secondary hypotheses that CO₂ is associated with a reduction in myocardial injury assessed by duration of hospitalization and intensive care unit stay.

We found that 1 minute after CPB, SWMAs were worse in the air group (p = .027) compared with the CO₂ group. The 16-segment model would take exponentially longer but might be more revealing than the 6 walls we evaluated. Nevertheless, these six walls do represent all coronary arteries. Because there are six walls per patient, the significant difference we found between treatment and placebo of the average worsening of score of 0.3774 per wall translates to a worsening of score of 2.2644 if it occurs in only one of the six walls. Various rescue inotropes such as calcium or epinephrine or phenylephrine, resting on bypass for several minutes, and other maneuvers were probably urgently used for those with worse function, and thus do not appear in our statistical model. However, baseline ejection fraction was similar, as seen in similar baseline total SWMA scores.

Ischemia rather than infarction is implied by the transient nature of the SWMA. Alternatively, infarction in placebo patients could have been masked by the treatment group’s predominance of patients with more SWMAs at baseline and risk of developing new SWMAs (i.e., resternotomies). A strength of this study is the degree of significance found at 1 minute. Another strength is that one surgeon performed all of the operations, creating a standard deairing technique and surgical management. We linked the etiology to the effect by showing fewer bubbles in cardiac chambers in the CO₂ group, which had better SWMAs.

Before promoting CO₂ as a standard of care, additional studies should evaluate more patients and postoperative serum troponin I levels (13). Our results should apply to the target population of all patients undergoing open heart surgery. We found that insufflating CO₂ over the heart during CPB significantly reduces the impact of air emboli on regional cardiac function 1 minute after separation from bypass. Trends, although not reaching statistical significance, were noted for less SWMAs in most walls at 60 minutes. Even if significant only at 1 minute, possible myocardial benefits may warrant treatment for two reasons. First, few negative effects are associated with insufflation of CO₂. Second, prevention of high SWMA scores (i.e., low ejection fractions) can prevent morbidity (11,12).