Does Hemoglobin Concentration Affect Perioperative Myocardial Lactate Flux in Patients Undergoing Coronary Artery Bypass Surgery?

Abstract

Increasing concern over complications related to blood transfusions has prompted a reevaluation of what constitutes an "adequate" perioperative hemoglobin concentration, particularly in patients undergoing coronary artery bypass graft (CABG) surgery.Data from 224 patients with preserved ventricular function (ejection fraction > 50%), undergoing CABG surgery, previously studied under a variety of anesthetic protocols, were reexamined to determine the effect of hemoglobin (HGB) concentration on myocardial lactate flux (MLF) (as an index of ischemia). The interaction of MLF and HGB concentration, anesthetic technique (ANES), and hemodynamic variables (including systemic and pulmonary arterial pressures (SAP and PAP), cardiac output (CO), and myocardial oxygen consumption (MVO₂) was determined from a pool of 1598 data sets obtained from 224 patients. Data were collected from just prior to induction of anesthesia until 24 h postoperatively. Univariate analysis revealed a statistically significant relationship between MLF and HGB concentration (P < 0.001) but the correlation coefficient was only 0.09. Multiple regression analysis did not determine HGB concentration to be a significant independent term affecting MLF in either the overall group or in a subgroup of 22 patients having an adverse outcome (myocardial infarction, stroke, or death). For patients undergoing CABG surgery, HGB concentrations within the range of 58-172 g/L were not a significant variable in production of global myocardial ischemia as evidenced by MLF. This suggests that HGB concentrations as low as 60-70 g/L in the perioperative period are well tolerated and are not associated with an increased incidence of myocardial ischemia.

(Anesth Analg 1995;80:910-6)

Anemia is often encountered in patients undergoing surgery. Increasing concern over the risks of blood transfusion has promoted the evolution of strategies to minimize the likelihood of receiving a perioperative blood transfusion and has prompted a reevaluation of what constitutes an "adequate" hemoglobin (HGB) concentration (i.e., one that permits adequate tissue oxygenation at realistic flow rates) ^[1]. One blood conservation strategy involves acceptance of a lower HGB concentration than previously deemed necessary ^[2,3], either in association with preoperative isovolemic hemodilution, or in the postoperative phase.

Perioperative anemia, whether due to organic disease or associated with intentional isovolemic hemodilution, involves potential morbidity. Concern over impaired wound healing and wound infection contributing to increased hospital stay does not appear to be justified unless the anemia is severe ^[1]. Although uremic patients with anemia often demonstrate bleeding abnormalities, there is no evidence that anemia contributes to increased bleeding in nonuremic patients ^[1]. Severe anemia may predispose patients to myocardial ischemia ^[3], and although rare, neurologic symptoms have been reported in association with severe euvolemic anemia ^[4].

Euvolemic anemia may be associated with a compensatory increase in organ blood flow and increased oxygen extraction ^[5]. Patients undergoing cardiopulmonary bypass (CPB) usually have some degree of coronary artery stenosis limiting increases in regional blood flow. Furthermore, the near maximal oxygen extraction of the myocardium at baseline leaves little room for compensation. Attempts have been made to establish lower limits of acceptable HGB concentration in patients with coronary artery disease but the results are not definitive ^[6-9].

In this study, the influence of several factors, including HGB concentration, hemodynamics, coronary sinus blood flow (CSBF), myocardial oxygen consumption (MVO (₂)), and anesthetic technique (ANES) on myocardial lactate flux (MLF, i.e., the product of myocardial lactate extraction and CSBF--a measure of myocardial ischemia) was determined from a preexisting data base of 224 patients undergoing coronary artery bypass grafting (CABG) at the Maritime Heart Centre since 1984 ^[10-17].

Methods

Data from 224 patients (28 female) with preserved ventricular function (i.e., ejection fraction > 50%), who had undergone CABG and participated in clinical studies at the Maritime Heart Centre since 1984 ^[10-17] were reexamined to determine the relationship between MLF and HGB. The mean age of patients was 61 yr, ranging from 36 to 84 yr. Indications for surgery were intractable angina with significant angiographic disease, three-vessel involvement, or left main coronary artery stenosis. All data were collected with approval of the institutional review committee and consent of the patients.

The anesthetic technique used varied according to the protocols as previously described ^[10-17]. CPB was conducted using a crystalloid (Plasmalyte-A Registered Trademark; Baxter Corp., Missisauga, Ontario, Canada) and 5% albumin priming solution, resulting in a hematocrit of 20%-25% while on bypass. During the period when CABG was performed, the heart was arrested using cold, high potassium cardioplegia. Hypothermia with core temperatures maintained at 28-30 degrees C while on bypass was used routinely. Anesthesia was terminated at the time of skin closure. In all cases, the surgeon considered revascularization to have been adequate. Postoperatively, patients received intermittent morphine or pethidine for analgesia and benzodiazepines for sedation. Tracheal extubation was performed routinely on the morning after surgery.

For each sample, the following data were available: 1) arterial lactate concentration, 2) arterial oxygen tension, 3) coronary sinus lactate concentration, 4) coronary sinus oxygen tension, 5) CSBF (by thermodilution), 6) MVO₂, 7) heart rate (HR), 8) systemic arterial pressure (systolic [SBP], diastolic, and mean), 9) pulmonary arterial pressure (systolic, diastolic, and mean) and pulmonary artery occlusion pressure, 10) right atrial pressure, 11) thermodilution cardiac output, 12) systemic vascular resistance index, and 13) HGB. Conduct of the anesthetic was left to the discretion of the attending anesthetist based on clinical signs, hemodynamic data, and arterial blood sampling. Data derived from the use of the coronary sinus catheter (coronary sinus lactate concentration, coronary sinus oxygen tension, and CSBF) were not made available until all patients in a particular study group were completed. During surgery, every attempt was made to maintain vital signs within a clinically acceptable range Figure 1, A-B.

Figure 1: A and B, Hemodynamic variables (mean +/- SD) for 224 patients undergoing coronary artery bypass graft surgery (TOTAL 1598 data sets). Series 1-10 represent measurements taken prior to induction (sedated but awake, n = 193), postinduction (n = 195), postintubation (n = 189), skin incision (n = 192), poststernotomy (n = 184), just prior to cardiopulmonary bypass (CPB) (n = 29), immediately post-CPB (n = 171), skin closure (n = 173), 1 h postoperative (n = 166), and 24 h postoperative (n = 106), respectively. MAP = mean arterial pressure (mm HG); HR = heart rate (bpmin). Differences in protocols are responsible for differences in the number of data sets at any measurement point.

Data were obtained for 1598 separate sample points. These points represented measurements taken from 10 intervals from the time of insertion of the coronary sinus catheter prior to induction of anesthesia until just prior to the onset of CPB, and from the time of separation from CPB until up to 24 h postoperatively Figure 1. No data were obtained while on CPB. Data were excluded if a complete data set could not be obtained (e.g., coronary sinus catheter malfunction) for any particular sample point. Because of differences in protocol design, the number of data sets for each interval is not necessarily the same.

Data were analyzed as a single group (1598 samples) and after separation into pre-CPB (982 samples, series 1-6) and post-CPB (616 samples, series 7-10) groups. The separation into pre- and post-CPB groups was undertaken since it was anticipated that the post-CPB group might contain a greater proportion of anemic patients. A separate analysis was undertaken on data from the subset of 22 patients known to have major perioperative complications such as death, stroke, or myocardial infarction (MI). Stroke was defined as a new onset motor deficit with evidence of cerebral infarction on computed tomography scan, occurring within the first postoperative week. MI was determined on the basis of new Q-waves on the electrocardiogram and/or increase of plasma creatinine kinase-MB (CK-MB) to > 5 micro gram/L where CK-MB was > 4% of total CK.

Using a simple linear regression model MLF = alpha + beta times HGB, the coefficient beta was calculated and the percentage variation (r²) in MLF explained by changes in HGB was determined. In addition, each group was analyzed using a similar linear regression technique to determine the importance of the following factors as predictors of reduced MLF: ANES, cardiac index, MVO₂, SBP, diastolic arterial pressure, HR, and pulmonary artery occlusion pressure. Those factors (including HGB) found to be significant predictors of reduced MLF (P < 0.05) were then combined and subjected to a multivariate regression model (SYS-TAT Registered Trademark 5.0; Systat Inc., Evanston, IL) to determine the best overall model for prediction of MLF from the data available ^[18]. Minimum tolerance for entry into the model was 0.01. Forward step-wise regression with alpha-to-enter = 0.15 and alpha-to-remove = 0.15 was used.

Results

The regression line for MLF versus HGB, with 95% confidence limits for all 1598 data points and for the pre- and post-CPB subsets, is shown in Figure 2 and Figure 3. Table 1 summarizes the analysis of the data points for the pre- and post-CPB groups and the combined total. Shown are the mean +/- SD for MLF and HGB, and the Pearson correlation coefficients (r). A statistically significant relationship (P < 0.05) was determined between HGB and MLF in the combined data sets of pre- and post-CPB groups and post-CPB group Figure 2 and Figure 3B; however, the slope is nearly horizontal (r = 0.090 and 0.099, respectively) and the values for r² are correspondingly small Table 1. The relationship between HGB and MLF in the pre-CPB group was not significant, with r = 0.042, but P = 0.186. Table 2 summarizes the results of the univariate analysis and Table 3 summarizes the results of the multiple regression model including MLF, HGB, ANES, and hemodynamic variables. As a significant predictive variable, HGB was excluded from the model (P < 0.05, two-tail).

Figure 2: Linear regression analysis and 95% confidence limits for myocardial lactate flux (MLF) versus hemoglobin (HGB) in 224 patients from prior to induction of anesthesia until 24 h postoperative (n = 1598 data sets; MLF = 6.59 times 10² + 1.75 times 10^-4 times HGB; P < 0.001; r = 0.090).

Figure 3: A, Linear regression analysis and 95% confidence limits for myocardial lactate flux (MLF) versus hemoglobin (HGB) in patients prior to cardiopulmonary bypass (CPB) (n = 982 data sets; MLF = 1.51 times 10^-1 + 1.06 times 10^-4 times HGB; P < 0.186; r = 0.042). B, Linear regression analysis and 95% confidence limits for MLF versus HGB in patients after CPB up to 24 h postoperative (n = 616 data sets; MLF = -4.45 times 10^-2 + 2.98 times 10^-4 times HGB; P < 0.014; r = 0.099).

Table 1: Summary of Statistical Analysis for the Linear Regression of Myocardial Lactate Flux (MLF) Versus Hemoglobin Concentration (HGB) for Precardiopulmonary Bypass (Pre-CPB), Post-CPB, Overall (TOTAL), and in Patients with Complications (Comp)

Table 2: Summary of Univariate Linear Regression Analysis (P Value and Correlation Coefficient, r) of Myocardial Lactate Flux Versus Variables for Precardiopulmonary Bypass (Pre-CPB), Post-CPB, Overall (TOTAL), and Patients with Complications (COMP)

Table 3: Summary of Multiple Linear Regression Analysis

We were unable to determine the clinical outcome from the postoperative records for 24 patients. A review of the postoperative records from the remaining 200 patients determined that 22 patients experienced major complications (11%) including 20 MIs (10%), 1 stroke (0.5%), and 1 early postoperative death (0.5%). From these 22 patients, 161 data sets from the data base were obtained and reanalyzed as before. The relationship between HGB and MLF was found to be significant (P = 0.025), with the correlation coefficient r = 0.176 and r² = 0.031. Multiple regression analysis Table 3 excluded HGB as a significant variable. The incidence of global ischemia (MLF < 0) was 14.3% overall and 11.8% in patients with complications.

Discussion

We determined the relationship between HGB concentration and the occurrence of global myocardial ischemia as defined by alterations in MLF. Although a statistically significant relationship between MLF and HGB was found Table 1, a reflection of the large sample size, the correlation was minimal (e.g., r = 0.090) for the combined pre- and post-CPB group data sets. When we attempted to fit hemodynamic variables into the regression analysis, a small improvement in the overall correlation ("multiple r", Table 3) was obtained but HGB was excluded as a significant covariable in MLF determination. We speculated that the 22 patients having major perioperative complications might represent a subgroup particularly vulnerable to hemodilution. However, in that group, the multiple regression analysis did not show HGB concentration to be an important contributor to MLF.

Taken together, this suggests that in this group of patients, HGB concentration is not predictive of myocardial lactate production. Although other variables, such as ANES, HR, cardiac index, MVO₂, and SBP were more important determinants of MLF Table 3, a large proportion of the variation in MLF is not explained by this model. MLF is reflective of global myocardial metabolism and may not be sensitive to areas of regional myocardial ischemia, where lactate production may be obscured by net lactate uptake in the myocardium as a whole. Despite this limitation, MLF remains a sensitive index of myocardial ischemia, especially global ischemia, and a negative lactate flux (i.e., net lactate production by the myocardium) clearly indicates overt ischemia ^[19].

The ability of the body to tolerate progressive anemia is directly related to its capacity to increase blood flow and maintain adequate oxygen supply to the tissues. Because the heart itself has tremendous metabolic requirements, it is both a consumer and provider of this increased flow. As such, the heart is likely the principle organ at risk, and the first to fail in the presence of severe anemia. The presence of significant coronary artery disease introduces an additional variable, the complexity of which is not fully understood. It can be reasoned that because of the flow-limiting nature of a severe coronary stenosis, the ability of the heart to compensate for anemia by increased blood flow distal to the stenosis is less than normal. At the same time, progressive anemia results in decreasing blood viscosity and improved flow characteristics, particularly through stenosed vessels ^[20]. The benefits of modest anemia in patients with cerebral vasospasm or severe peripheral vascular disease are well known ^[21,22]. Data for what constitutes the minimum HGB required for oxygen delivery in patients with coronary artery disease are incomplete.

Our results differ from those of the study by Weisel et al. ^[7] who concluded that in the early postoperative period, a normal lactate extraction was obtained only when the HGB was more than 100 g/L. Although most of our data comes from the earlier perioperative period, a significant portion comes from 1 h and 24 h postoperatively Figure 1. We found no clinically significant relationship between MLF and HGB Table 1 and Table 3, Figure 2 and Figure 3. Using MLF as an index of ischemia, our results show that HGB, within the range of 58-172 g/L Table 1, did not affect MLF to a clinically important degree (r² = 0.008). This suggests that these values of HGB lie within the range where the coronary circulation can adequately compensate (in awake, sedated patients with preserved ventricular function, during anesthesia, and in the early postoperative period after revascularization), provided that clinically acceptable hemodynamic control is maintained Figure 1. This is consistent with the data of Kim et al. ^[6] who studied the effects of hemodilution and hypothermia in 14 CABG patients postoperatively and found no differences in coronary sinus O₂ content between the markedly hemodiluted group (hematocrit 23% +/- 2%) and the moderately hemodiluted group (hematocrit 34% +/- 3%) at 34 degrees C or at normothermia. More recently, Mathru et al. ^[8] found no evidence of ischemia (based on electrocardiograph changes and lactate production) in anesthetized patients after CABG surgery, even with hematocrits as low as 17.4% +/- 3.7%.

Data from animal species suggests impaired myocardial oxygenation with hematocrits below 17%-42.5%, depending on the species and the presence of coronary stenosis ^[20,23]. Differences in species and methodology may explain the wide variation in critical hematocrits below which myocardial function deteriorates in the presence of coronary artery stenosis and anemia. However, these studies ^[20,23] point to a range of hematocrits within which the changing rheologic effects and coronary vasodilation (autoregulation) compensate enough to maintain the oxygen supply/demand ratio. When the lower limit of this autoregulatory range is exceeded, further (limited) compensation may occur by increasing the myocardial oxygen extraction ratio ^[23], but this probably heralds the onset of ischemia. Clinically, we would expect that this lower threshold would vary according to individual hemodynamic variables, metabolic requirements (e.g., depth of anesthesia), and extent of coronary disease.

A large proportion of variation in MLF was not explained by the linear regression model Table 3. The reasons that this may be so are unknown, but two possibilities are suggested. First, MLF may not be sensitive to regional ischemia where small areas of lactate production may have been obscured by global lactate extraction. As a measure of global left ventricular ischemia, however, MLF is quite sensitive ^[19]. Secondly, factors other than those included in the regression model (i.e., HGB and hemodynamic variables) may be more important determinants in producing myocardial ischemia in the perioperative period. Such factors may include vascular hyperreactivity, local and systemic catecholamine levels ^[24], and platelet function ^[25].

Conclusions that may be drawn from this study are tempered by several factors. The data were analyzed in a retrospective fashion from an existing data base of 224 patients examined over a 7-yr period. The patients were studied under a variety of different anesthetic protocols ^[10-17], differences among which were correlated to MLF. Patients undergoing CABG surgery are highly susceptible to myocardial ischemia, but this risk should be diminished after completion of the coronary grafts. Therefore, the results of this study may not be applicable to similarly high-risk patients having noncardiac surgery.

The data from the present study, the largest to date, suggest that in the immediate perioperative period for patients undergoing CABG surgery, HGB concentrations as small as 70 g/L were well tolerated in the majority of patients. A low HGB concentration did not contribute in a major way to the incidence of myocardial ischemia (based on MLF) in patients undergoing CABG surgery, provided normal hemodynamics were maintained. Even in patients suffering major perioperative complications which could be attributed to impaired oxygen delivery (e.g., MI, stroke, death), there was not a strong correlation between HGB concentration and MLF. Further study into factors responsible for perioperative myocardial ischemia are warranted.