Intraoperative Low-volume Acute Normovolemic Hemodilution in Adult Open-heart Surgery

Between April 1, 2000 and July 31, 2001, at the Division of Cardiac Surgery of the Policlinico di Monza, 256 consecutive patients underwent elective open-heart surgery and were evaluated for enrollment in the trial. Preoperative exclusion criteria were age less than 18 yr, left ventricular ejection fraction (LVEF) less than 30%, preoperative hematocrit less than 36% or hemoglobin less than 12 g/dl, history of hematologic diseases, chronic renal insufficiency (plasma creatinine > 2 mg/dl), and history of hepatic diseases (e.g. , active hepatitis or cirrhosis). Preoperative treatment with aspirin or subcutaneous low-weight molecular heparin was not a contraindication to enrollment in this study.

Following these criteria, 52 patients were not considered for randomization: 2 patients because they were aged less than 18 yr; 6 because of LVEF less than 30%; 39 because of preoperative anemia; 1 because of active hepatitis C virus; 3 because of chronic renal insufficiency; and 1 because of history of hemorrhagic diathesis.

After Institutional Review Board approval and informed written consent, 204 patients were enrolled in the trial. A standardized protocol for anesthesia was applied. Induction and maintenance of anesthesia were performed with propofol (bolus of 2 mg/kg and infusion of 6–10 mg·kg⁻¹·h⁻¹) and fentanyl (bolus of 5 μg/kg and total dose lower than 20 μg/kg). Muscle relaxation was obtained with pancuronium bromide (bolus of 0.1 mg/kg and further administrations of 0.03 mg/kg as needed). Nitrous oxide or isoflurane was added as indicated. All patients received an intraoperative infusion of tranexamic acid according to a protocol derived by previously published studies. 27–29An intravenous bolus of 1 g in 20 min before sternotomy was followed by a continuous infusion of 400 mg/h until the end of the operation; 500 mg were added to the CPB circuit priming solution.

By using a computer-generated random number sequence, the 204 patients were prospectively randomized, in a nonblinded manner, into two groups: the ANH group (103 patients) and the control group (101 patients). In the ANH group patients, 5–8 ml/kg of whole blood was withdrawn after the induction of anesthesia and before systemic heparinization and collected into sterile blood collection bags containing citrate phosphate dextrose anticoagulant (Fenwal, Baxter Corp., Irvine, CA), using a blood mixer and balance system (Easymix V3, Baxter Corp., Irvine, CA). The blood was withdrawn through a large bore catheter placed into the internal jugular vein, with gravity drainage, to limit shear effects on the platelets. During the withdrawal of the autologous blood, a colloid solution was infused (Emagel, Hoechst, Frankfurt, Germany) in 1:1 ratio. Next crystalloids were infused if hemodynamic status required. Isovolemia was evaluated through standard monitoring (systemic arterial pressure, electrocardiogram, and central venous pressure). Each unit of blood withdrawn was labeled with the patient's identification and stored at room temperature in the operating room. In the control group patients, the collection of the autologous blood was not performed.

The same team of surgeons, with standardized surgical procedures, did all the interventions. All patients were operated on through full median sternotomy. Before cannulation, porcine heparin, 300 U/kg, was administered, and supplemental doses were added, as needed, to maintain a celite-activated clotting time (ACT) above 480 s. Single venous cannulation was used in all cases, with the exception of operations on the mitral valve, where a double cava cannulation was performed. A nonpulsatile blood flow of 2–2.4 l·min⁻¹·m⁻²was obtained with a roller pump (SC Stockert, Stockert Instrumente GmbH, Munich, Germany). Surgery was performed with moderate hypothermia (32–35°C); in all cases, a hollow-fiber membrane oxygenator was used (Affinity, Dideco, Mirandola, Italy). Priming of the CPB circuit consisted in 1,500 ml of a balanced crystalloid–colloid solution (lactated Ringer's solution, 1,000 ml; mannitol 18%, 250 ml; plasma expander, 250 ml). Myocardial protection during aortic cross-clamping time was achieved with cold blood cardioplegia. After termination of CPB, the total dose of heparin was reversed with protamine (1:1 ratio). Further doses of 50 mg were administered if ACT was greater than the basal value. A cell saver circuit (Compact Advanced, Dideco, Mirandola, Italy) was used in all patients to concentrate the remaining cellular content in the circuit for extracorporeal circulation and the blood shed from the surgical field and collected in a cardiotomy before and after CPB. Before chest closure, mediastinal and pleural drains were positioned, and a low-grade suction instituted. The blood drawn before CPB from the ANH group patients was reinfused during extracorporeal circulation if hemoglobin value was less than 6.5 g/dl and if hematocrit value was less than 20%. Otherwise, the autologous blood was reinfused starting at the end of surgery, before transporting patients to the intensive care unit (ICU). Intra- and postoperative criteria for allogeneic transfusions were standardized. These criteria were applied in the ANH group only after the reinfusion of the autologous blood withdrawn. Packed red blood cells (PRBC) were transfused, during CPB if hemoglobin value was less than 6.5 g/dl and hematocrit value was less than 20%, and after CPB and during all the hospitalization if hemoglobin value was less than 8.5 g/dl and hematocrit value was less than 25%. Fresh frozen plasma (FFP) was infused, after protamine administration, if prothrombin time value was 1.5 times the basal in presence of active bleeding. Platelet concentrates (PLTC) were transfused in presence of active bleeding and of a platelet count less than 50,000/mm³. Samples for evaluation of hemoglobin, hematocrit, platelet count, prothrombin time, activated partial thromboplastin time, creatinine, creatine phosphokinase, and creatine phosphokinase myocardial band isoenzyme were performed before the induction of anesthesia (time 1), on arrival in ICU (time 2), 24 h after the arrival in ICU (time 3), 48 h after surgery (time 4), and at discharge (time 5). Further determinations of these variables were made as the clinical situation required. During CPB, hematocrit and hemoglobin values were determined about every 15 min from serial blood arterial samples drawn for blood gas determinations (IL 1640, Instrumentation Laboratory SPA, Milan, Italy).

Blood loss was recorded during the first 24 h. Reinfusion of shed mediastinal blood was not performed during postoperative period.

Surgical reexploration was decided on when bleeding in the first 2 h was greater than 300 ml/h or if greater than 200 ml/h for 4 consecutive h, with normal coagulation data. Intubation time, ICU stay, hospitalization, major complications (e.g ., perioperative myocardial infarction, renal insufficiency, pulmonary embolism, stroke), and mortality were considered.

If we consider, for the type of patients we enrolled in the study, a prevalence of about the 50% of patients transfused with allogeneic blood products, and if we consider as effective a reduction, as a result of ANH, of about 30% of the patients transfused with allogeneic transfusions, for one-sided α of 5% and β of 20%, the number of patients needed to enroll is approximately 100 per group. To test the normality of the distribution of the continuous variables, the Kolmogorov–Smirnov statistic was performed. The normally distributed data were compared between the groups with two-tailed unpaired, Student t  test and expressed as mean ± standard deviation (SD). Nonnormally distributed variables were evaluated with the Mann–Whitney U test and expressed as median (25th–75th percentiles). Categorical data were analyzed with the chi-square test or the Fisher exact test as appropriate. Analysis of variance (ANOVA) for repeated measures was used to evaluate, within groups, the changes of the variables in the time. Statistically significant differences were considered for a P < 0.05.

Of the 204 patients randomized, 2 patients (1 per group) did not complete the study: they died during the first 24 h postoperatively after cardiogenic shock refractory to maximal pharmacologic support and intraaortic counter pulsation. They were excluded by statistical analysis. Two hundred two patients entered in the statistical analysis.

No significant differences were found in demographics (table 1), preoperative hematochemical data (table 2), and operative procedures (tables 3 and 4). Only the amounts of colloids and crystalloids infused before CPB, greater in ANH group, were significantly different (table 4). Minimum hematocrit value during CPB was slightly lower in ANH group but without reaching statistical significance (table 4).

Eight patients required, during CPB, reinfusion of blood previously withdrawn because the values of hemoglobin and hematocrit were lower than the values fixed for transfusion of PRBC. Particularly, one patient required reinfusion of the autologous blood because of intraoperative acute bleeding. Of the remaining seven patients, two had preoperative weight less than 50 kg, and five had preoperative hemoglobin values lower than 12.5 g/dl. The analysis of these patients followed the “intention to treat” rule.

All hematochemical variables evaluated were similar in the two groups at each considered time point (Student t  test,or Mann–Whitney U test). ANOVA for repeated measures showed evidence of significant variations in time for all hematochemical data but without differences between the groups. Also the interaction time between groups did not differ (table 5).

Table 6reports postoperative bleeding and perioperative allogeneic blood components transfused; also in this case, no significant differences were found between the groups, and the number of patients transfused and the number of patients requiring surgical reexploration for excessive bleeding were not different in the two groups. A clear surgical source of bleeding was found in four patients in the control group and in five patients in the ANH group; also in this case, following the intention to treat rule, these patients were included into statistical analysis. The remaining patients had evidence of diffuse bleeding requiring transfusion of FFP or PLTC as indicated by hematochemical data.

The decision to violate protocol was taken in four patients of the control group and in five patients in the ANH group (P = 0.76). These patients were transfused with PRBC even if the hematocrit and hemoglobin values were higher than the established values of the transfusion protocol. In all cases, the decision was made on the basis of the hemodynamic status (hypotension, tachycardia, low values of central venous pressure, with hematocrit and hemoglobin values near to the lower limits for transfusion).

Finally, outcomes and major complications in the two groups were similar (table 7). No complications related to autologous blood transfusion were observed.

The importance of techniques that permit reduction in the need for allogeneic transfusions during major surgical procedures has been recently well reviewed. 30Among these techniques, intraoperative withdrawal of a part of circulating blood volume and the substitution with crystalloid and colloid solution to obtain fresh whole blood to transfuse after the end of surgery was first proposed in the 1960s. 11But up until now, the great number of published trials may derive no uniform conclusions. Controversial results are evident starting from the first studies 11–18and confirmed by most recent reports. 19–25Next to trials, which describe reduction of postoperative bleeding 24or reduction of the need for allogeneic transfusion, 12,17–18,25or both, 13–14,22there are studies that describe partially positive effects only on patients at high risk for excessive postoperative bleeding 23or even the lack of any desirable effect of ANH. 15–16,20–21 

The results of our study seem to confirm these negative findings: the use of low-volume ANH does not determine reduction of postoperative blood loss and perioperative donor blood use in patients undergoing open-heart surgery. These results also confirm the indications of some mathematical analyses of hemodilution, which predicts the inefficacy of low-volume ANH in reducing the need for allogeneic blood components. 31–32 

Further, the authors of a recent meta-analysis, who considered the published works on ANH, have proposed similar conclusions. 26They stressed the fact that most of these studies showed lack of a prospective, randomized method of patient enrollment, but particularly they observed that the studies with a transfusion protocol failed to show statistically significant reduction in either the likelihood of receiving an allogeneic transfusion or the allogeneic units transfused. In the authors’ opinion, this lack of benefits suggests that biased study design may be responsible for the reduction in blood exposure attributed to ANH.

Our trial was designed to be prospective and randomized, with precise limits for transfusions of PRBC, FFP, and PLTC: in this manner, we think that its principal limitation, the fact that it is a nonblinded study, can be partially offset. The number of patients in whom the transfusion protocol was broken was few and was not different between the two groups.

Our desire to study the effects of low-volume ANH on patients undergoing open-heart surgery derives from the fact that these patients are considered to be at increased risk for perioperative excessive bleeding, therefore with theoretical greater benefits deriving by the application of a comprehensive protocol to reduce the need for allogeneic transfusions.

One of the most debated points is the degree of ANH used. Even if some authors consider only high-volume phlebotomy as effective in reducing the need for allogeneic transfusions, 33we chose to perform low-volume ANH to include in the study, as well as patients with a relatively low hematocrit and hemoglobin values (respectively, 36% and 12 g/dl), without considering, as exclusion criteria, a given value of body mass index. This means that smaller patients or patients with hematochemical values close to the lower limits could experience, during CPB, too-low values of hematocrit had high-volume ANH been performed preoperatively. Nuttall et al.  34reported in their patients a high rate of early autologous blood reinfusion (30.3%) for this same reason. Our study seems to confirm these results, although only 8% of our patients required early reinfusion; the majority of these patients showed a preoperative level of hemoglobin near to 12 g/dl, whereas two patients had low body mass index, and only one patient required reinfusion of autologous blood for intraoperative bleeding. Further, Kahraman et al.  25described the same effects of low-volume (5–8 ml/kg) or high-volume (12–15 ml/kg) phlebotomy in terms of amount of perioperative allogeneic transfusions and postoperative bleeding. Moreover, the same study showed, in the high-volume phlebotomy group, a mild increase in total estimated red blood cell volume lost, probably the result of either the transient dilution of coagulation factors or the higher amount of colloidal substitutes administered to these patients.

Another important point of discussion is the fact that previously published studies describing positive effects derived by the use of ANH did not separate the role of this technique from that of other blood-sparing techniques, such as reinfusion of shed mediastinal blood after surgery, 35or the application of pharmacologic hemostatic protocols. 34In our division, the use of cell saver circuit and prophylactic treatment with tranexamic acid, but not postoperative reinfusion of shed mediastinal blood, are routinely performed in all patients. Regarding the use of cell saver, a study by Temple et al.  36demonstrated that simple ANH did not decreased blood bank requirement, whereas the combination of this technique with a cell saver and a membrane oxygenator in the circuit for extracorporeal circulation greatly reduced the need for donor blood. We were not able to confirm these results: the addition of low-volume ANH to the cell saver did not show any improvement in terms of allogeneic blood usage.

Regarding the use of antifibrinolytic therapy, the previously cited study by Nuttall et al.  34evaluated the hemostatic effects of ANH, tranexamic acid, and aprotinin in high-risk patients; they concluded that the combination of ANH with tranexamic acid treatment induced a reduction of the need for allogeneic transfusion requirement and postoperative blood loss similar to that derived by the treatment with high-dose aprotinin, and better than treatment with only tranexamic acid or placebo.

At present, all our patients undergoing surgery, with or without CPB, are treated with tranexamic acid, as a result of a pharmacologic protocol derived by previously published studies. 27–29,37But the results of our study do not seem to confirm the positive results by Nuttall: ANH has no additional effects with tranexamic acid in reducing postoperative bleeding and allogeneic transfusions compared with patients treated solely with tranexamic acid. Similar results were recently described in a study considering patients undergoing total knee replacement who were treated with tranexamic acid or ANH; in these cases, tranexamic acid, but not ANH, showed blood-sparing effects. 38 

The same conclusions were proposed by Vedrinne et al.  21, who described in patients undergoing cardiac surgery, no hemostatic effects of ANH compared with treatment with high-dose aprotinin, which induced, on the contrary, significant reduction of blood loss and transfusion requirement.

One of the major criticisms proposed by some authors regarding the use of ANH implies the possibility of inducing perioperative complications, particularly myocardial ischemia, peripheral edema requiring diuretic therapy, increased lung water content and worsening of pulmonary function in the postoperative period, increased wound drainage and operative blood loss, lower oxygen extraction, and, possibly, elevated lactate levels. 39–41 

In our study, the patients undergoing low-volume ANH did not show increased perioperative complications, but this may simply be the result of the low level of hemodilution obtained and the relatively few patients enrolled.

In conclusion, in our study, low-volume ANH did not show blood-sparing properties in adults undergoing open-heart surgery. Further blinded studies considering larger series of patients and higher volume phlebotomy are needed to better comprehend the effects of ANH in cardiac surgery.