Effectiveness of Acute Normovolemic Hemodilution to Minimize Allogeneic Blood Transfusion in Major Liver Resections

The trial was approved and monitored by the Research Ethics Committee of the Hadassah University Medical Center, and written informed consent was obtained from each patient. Over a 3-yr period, 78 consecutive patients (age > 18 yr) who were scheduled to undergo elective major hepatic resection (hepatectomy or extended hepatectomy 17) for primary or metastatic tumors were studied prospectively. Patients were eligible for the study if they had a preoperative hematocrit greater than 36%, 10were American Society of Anesthesiologists (ASA) physical status I or II, had no cardiovascular or pulmonary disease, and showed no evidence of severe hepatic metabolic disorder (bilirubin > 1.5 times upper limit of normal or aspartate aminotransferase or alanine aminotransferase > 2 times upper limit of normal).

Study patients received a standardized general anesthetic. General anesthesia was induced with fentanyl, thiopental, and vecuronium, and after tracheal intubation was maintained with isoflurane in a 50% oxygen–nitrous oxide balance. Additional fentanyl and vecuronium were administered as appropriate. The surgical technique was also similar for all patients. Minute ventilation was titrated to maintain normocarbia. Hypothermia was prevented during surgery by using a heating blanket and perfusion warming. In addition to the standard monitors, intraoperative monitoring included intraarterial (via  radial artery catheter) and central venous (via  an internal jugular catheter) pressure monitoring and electrocardiogram (5-lead) with continuous ST-segment trends that were reviewed and verified for possible ischemic episodes 18by an investigator who was blinded to patient group assignment.

On admission to the operating room, patients who met inclusion criteria were randomly assigned (random numbers) to one of two groups: control group or ANH group. For patients undergoing ANH, the hemodilution process was initiated immediately after tracheal intubation. ANH was performed via  a high-flow (8.5 French) internal jugular introducer to a target hematocrit of 24%. The volume of blood removed was calculated from preoperative hematocrit, target hematocrit after hemodilution, and the patient's estimated blood volume according to a formula previously described. 7,19The blood, which was removed into standard citrate–phosphate–dextrose blood storage bags, was simultaneously replaced by colloid solutions (6% hydroxyethylstarch [200/0.5], 20 ml/kg, followed by 5% albumin), which was administered via  two large bore peripheral veins, in quantities approximately 1.1 times the volume of blood removed. 20–22The hemodilution process took about 40 min. Additional blood-conserving techniques, i.e. , preoperative autologous blood donation and intraoperative use of cell salvage, were not implemented in the current study.

During the intraoperative period, patients in both groups received continuous crystalloid (Ringer's lactate solution) infusion: 4 ml · kg⁻¹· h⁻¹during the prehepatic resection phase until completion of parenchymal transection (relative fluid restriction was applied to maintain low central venous pressure, i.e. , < 5 cm H²O 11,23) and 10 ml · kg⁻¹· h⁻¹during the posthepatic resection phase until the end of the operation. Intermittent fluid boluses (500 ml) were administered to maintain urine output greater than 0.5 ml · kg⁻¹· h⁻¹and systolic blood pressure greater than 90 mmHg. Subsequent volumes of blood were replaced with Ringer's lactate solution in a 3:1 volume replacement. Subsequent blood loss was estimated by assessment of the suction bottles, sponges, and the surgical drapes and gowns by an anesthesiologist who was not aware of the patient's group assignment. Arterial blood was sampled every 30 min, and when clinically indicated, for measurement of arterial blood gases and hematocrit. Fraction of inspired oxygen was increased from 0.5 to 1.0 when the transfusion trigger was reached. The transfusion threshold for all types of erythrocytes (autologous or allogeneic) was hematocrit less than 20%. To ensure uniformity, transfusion guidelines were also established for administration of fresh frozen plasma (prothrombin time, activated partial thromboplastin time > 1.5 times normal), cryoprecipitate (fibrinogen concentrations < 100 mg/dl), and platelets (< 50 × 10³/μl) in the presence of continuous uncontrolled bleeding with no evidence of clot formation. 2The anesthesiologist making decisions regarding transfusion was not blinded to patient group assignment. Autologous blood was reinfused when a minimal hematocrit value was reached or after completion of liver resection and hemostasis, in the reverse order of collection. In the ANH group, allogeneic blood was administered only after all autologous blood was transfused. Patients were monitored in the postanesthesia care unit until the first postoperative day, and thereafter discharged to the surgical unit.

Postoperative surveillance included hematology, biochemical liver, renal, and standard coagulation tests (prothrombin time, activated partial thromboplastin time), and postoperative morbidity and mortality. Measurements were routinely determined and reviewed on admission to the recovery room and daily until the third postoperative day and every other day until discharge. Additional samples were drawn when clinically indicated. Electrocardiography was performed on admission to the recovery room, on the first and third postoperative day, and on the day of discharge. Additional electrocardiography and measurements of cardiac enzymes were determined when clinically indicated. Adverse outcomes, as listed in Results, were detected by the examining physician and validated by one of the other investigators, both of whom were not aware of the patients’ group assignment.

The two prospectively defined endpoints were (1) the avoidance of exposure to at least 1 unit of allogeneic blood transfusion, as indicated by the proportion of patients in each group who did not receive at least one allogeneic erythrocyte transfusion during the intraoperative period and first 72 h after surgery, and (2) documenting the safety profile of ANH as assessed by comparing mortality, morbidity, and laboratory values reported in the ANH group with those variables in the control group.

Categorical data were analyzed using the chi-square test or Fisher exact test. Differences between the means of two groups and the median units transfused were compared using Student t  test and Mann-Whitney test, respectively. Data within each group were analyzed using analysis of variance for repeated measurements. When appropriate, post hoc  analyses were performed with the Newman-Keuls test. Analysis was performed using Statistical Analysis System software (version 6.12; SAS Institute, Cary, NC). P < 0.05 was considered to represent statistical significance. Results are expressed as the mean ± SD. A power analysis for allogeneic blood transfusion as an outcome, with 80% power to detect a 25% reduction in this outcome, and significance ≥ 0.05, assuming 36% incidence of blood transfusion, indicated that 39 patients were required in each group.

During the study period, 78 patients were included. Both groups (n = 39 per group) were comparable with regard to sex ratio, age, weight, height, and ASA physical status (table 1). In both groups the most common indication for resection was metastatic cancer, and in approximately two third of the patients, a standard right or left hepatectomy was performed. Preoperatively, all patients had normal liver, renal, and standard coagulation test results (data not shown).

In the hemodilution group, the amount of collected blood was 2,020 ± 412 ml, and the mean volume of 6% hydroxyethylstarch plus albumin infused was 2,210 ± 458 ml (table 2). The reduction in hematocrit from 40.5 ± 2.7 to 23.5 ± 1.2% was not associated with significant changes in mean arterial blood pressure, heart rate, central venous pressure, or blood lactate concentrations. There was a slight but statistically significant prolongation in the time from induction of anesthesia to skin incision in the hemodilution group (43 ± 5 vs . 40 ± 6 min in the ANH and control groups, respectively).

There were no significant differences in anesthetic and operative times between the two groups (table 1). The median estimated surgical blood loss was 750 and 890 ml for the ANH and control groups, respectively. Mean respective values were 1,442 ± 1,827 (range, 100–7,000 ml) and 1,528 ± 1,822 (range, 100–7,500 ml). Central venous pressure increased, although not significantly, from preoperative values of 7 ± 4 and 7 ± 5 mmHg to 10 ± 6 and 10 ± 7 mmHg with skin closure in the ANH and control groups, respectively. During parenchymal resection, central venous pressure in the ANH and control groups was 7 ± 2 and 6 ± 3 mmHg, respectively. None of the patients in the two groups required the administration of fluid bolus to keep urine output greater than 0.5 ml · kg⁻¹· h⁻¹. No ST-segment changes were observed during the operation in both groups.

Hematocrit, creatinine, prothrombin time, and arterial pH values were similar in both groups in the postoperative period (tables 3 and 4). Also, the highest postoperative values for bilirubin, alanine aminotransferase, aspartate aminotransferase, and alkaline phosphatase were comparable between the groups (data not shown).

Overall, 14 patients (36%) in the control group compared with 4 patients (10%) in the ANH group received allogeneic blood transfusion (P = 0.014, Fisher exact test). There were no significant differences between the groups in the median number of units of allogeneic blood transfused. The number of units of allogeneic blood transfused for the ANH and control groups and the fractional surgical blood loss in patients receiving allogeneic blood transfusion are shown in figures 1 and 2, respectively. Five patients in the control group compared with three patients in the hemodilution group received fresh frozen plasma, cryoprecipitate, or platelets (P > 0.05). None of the patients died in the perioperative period. The overall complication rate was similar in the two groups.

The major finding of the current study is that ANH in adults undergoing elective major liver resection is a method for significantly reducing exposure to allogeneic blood. Although previous studies have demonstrated the efficacy of this technique in children 24and adults 25,26undergoing liver resection, these studies were not controlled randomized trials. Moreover, criteria for blood transfusion were not defined, 24,25and none of the studies reported central venous pressure at the time of parenchymal resection, a parameter that has been shown to affect surgical bleeding during major hepatic resections. 11,23Finally, recent advances in surgical and anesthetic care made the applicability of findings from earlier studies to current practice uncertain. We therefore undertook this prospective, randomized study that recruited similar patients who were scheduled to undergo major liver resection. Although the anesthesiologist making the decision about the need for transfusion was not blinded to the patients’ group assignment, the indication for blood transfusion was standardized. Also, during the postoperative period, adverse outcomes were detected by the examining physician, who was not aware of the patient's assignment.

Recent mathematical analysis predicted that hemodilution may be efficacious if surgical blood loss is greater than 70% of the patient's blood volume. 10Patients presenting for major liver resections may be appropriate candidates for ANH because expected blood loss may be substantial. 11–16,27The overall mean surgical blood loss in our current series of patients was 1,442 ± 1,827 ml and 1,528 ± 1,822 ml, with a median value of 800 ml (range, 100–7,000 ml) and 900 ml (range, 100–7,500 ml) in the ANH and control groups, respectively, and is comparable to previously reported blood loss associated with major liver resections. 12,25,26,28In a selected group of patients in whom surgical blood loss was greater than 0.7 of the patient's blood volume, the median blood loss was 3,800 and 3,350 ml in the ANH and control groups, respectively. In addition, 7 of the 14 transfused patients in the control group received 1 unit of blood, a blood requirement that could have been accommodated by the use of ANH. 9,20Presumably, the surgical blood loss as a fraction of blood volume in these patients was within “the window of efficacy” as suggested by Weiskopf. 10However, in contrast to the latter mathematical analysis, 10in which the assumption was that hemodilution proceeds until the trigger hematocrit, in the current study a different approach was used: the hemodilution was undertaken to a hematocrit of 24%, while the transfusion trigger was 20%. Although less efficacious, we have chosen this approach, as did other investigators previously, 20,28–32because during liver resections, a sudden extensive bleeding may occur with an acute and rapid decrease in hematocrit below the lowest acceptable value. In addition, the mathematical model also assumes that euvolemia is maintained. This may be difficult in the clinical situation because intraoperative estimates of blood loss are inaccurate. 33Because we did not measure blood volume, it may be argued that isovolemia was not maintained. In the current study, blood removed during the hemodilution process was replaced by colloid solutions in quantities approximately 1.1 times the volume of blood removed, as previously suggested. 20–22Also, surgical blood loss was replaced with crystalloid solution in a 3:1 volume replacement, and patients in both groups received continuous crystalloid infusion with additional intermittent fluid boluses when indicated, according to predetermined criteria. Finally, heart rate did not increase and central venous pressure did not change significantly during hemodilution and thereafter, and urine production was comparable in the two groups and did not decrease to less than 0.5 ml · kg⁻¹· h⁻¹throughout the operation, suggesting that patients in the ANH group were not hypovolemic. In addition, the finding of no change in heart rate during hemodilution in the current study is comparable to previous reports 34–36that showed an increase in cardiac output with hemodilution without a significant heart rate change. This may be caused in part by the effect of anesthetic agents, since previous studies 36,37showed that ANH in awake patients was associated with higher values of heart rate than those measured in anesthetized patients.

Although previous mathematical models 7–10suggest that ANH can save erythrocytes that would otherwise be lost during surgery, they emphasize that limited (posthemodilution hematocrit > 24%) normovolemic hemodilution is of minimal value. In the current study, the amount of blood removed was 2,020 ± 412 ml. A recently published metaanalysis reported that the mean volume of blood removed in 24 randomized controlled trials of ANH was 936 ml. In addition, the investigators also found that, in trials in which the volume of blood withdrawn was less than 1,000 ml, the reduction in the likelihood of transfusion failed to reach statistical significance, whereas in studies in which more than 1,000 ml of blood was removed, a statistically significant reduction in the likelihood of transfusion was found. A protocol that targets a low posthemodilution hematocrit level, in the range from 24% to as low as 20%, might therefore enhance the effectiveness of hemodilution. 7–9 

The perioperative morbidity rate was similar in the two groups and was comparable to that previously reported by other investigators, 11–16indicating that ANH is well tolerated in this patient population. Although Sejourne et al.  26found that alterations in blood coagulation were greater in hemodiluted patients undergoing liver resection than in those without hemodilution, we observed no difference in standard coagulation tests between groups. Rehm et al.  20recently showed that ANH used during certain conditions reduces the surgical loss of erythrocytes. With retransfusion of the removed autologous blood after termination of surgery, a higher hematocrit should therefore be expected.

Previous studies reported higher rates of postoperative infections in patients who received perioperative allogeneic transfusions than in those who were not transfused or who received only autologous blood. 38,39In the current study, the rate of infections was not significantly higher in the control group (8 patients, 3 of whom received transfusion) compared with the hemodilution group (3 patients, none of whom received allogeneic transfusion). With a reduction of only 1 unit of allogeneic blood transfusion (in 7 of 14 transfused patients), no real benefit was to be expected. No adverse cardiac, renal, or neurologic outcomes were observed in the ANH group. However, the occurrence of zero events does not imply the risk is zero. For a sample size of 39 patients, the maximum incidence (with 95% confidence) of an adverse event is 7.4%. Also, in this study we included only ASA I-II patients. This careful patient selection may have contributed to the zero incidence of adverse outcome. ANH has been used previously in the elderly and in patients undergoing cardiac surgery; however, most of the published studies were seriously flawed. 35Moreover, several recent studies suggested that anemia in the presence of cardiovascular disease may be associated with increased risk of organ dysfunction and mortality. 40,41In the present state of knowledge, it seems clear that caution should be exercised in patients with coronary artery disease or aortic stenosis, since their normal compensatory mechanism is impaired. Moreover, ANH is best avoided in patients with vascular disease in whom the prevalence of ischemic heart disease is very high.

In summary, in 39 ASA I or II patients undergoing major liver resection, ANH was used as a method for autologous blood transfusion. By this method, the need for allogeneic blood transfusion was reduced significantly. As ANH is a relatively cheap and simple method of autologous transfusion, these data suggest that ANH should be considered in this patient population. The resultant decrease in allogeneic blood use may help to alleviate the blood shortage that blood banks are facing.