Risks associated with transfusion of allogeneic blood have prompted development of methods to avoid or reduce blood transfusions. New oxygen-carrying compounds such as diaspirin cross-linked hemoglobin (DCLHb) could enable more patients to avoid allogeneic blood transfusion.
The efficacy, safety, hemodynamic effects, and plasma persistence of DCLHb were investigated in a randomized, active-control, single-blind, multicenter study in post-cardiac bypass surgery patients. Of 1,956 screened patients, 209 were determined to require a blood transfusion and met the inclusion criteria during the 24-h post-cardiac bypass period. These patients were randomized to receive up to three 250-ml infusions of DCLHb (n = 104) or three units of packed erythrocytes (pRBCs; n = 105). Further transfusions of pRBCs or whole blood were permitted, if indicated. Primary efficacy end points were the avoidance of blood transfusion through hospital discharge or 7 days postsurgery, whichever came first, and a reduction in the number of units of pRBCs transfused during this same time period. Various laboratory, physiologic, and hemodynamic parameters were monitored to define the safety and pharmacologic effect of DCLHb in this patient population.
During the period from the end of cardiopulmonary bypass surgery through postoperative day 7 or hospital discharge, 20 of 104 (19%) DCLHb recipients did not receive a transfusion of pRBCs compared with 100% of control patients (P < 0.05). The overall number of pRBCs administered during the 7-day postoperative period was not significantly different. Mortality was similar between the DCLHb (6 of 104 patients) and the control (8 of 105 patients) groups. Hypertension, jaundice/hyperbilirubinemia, increased serum glutamic oxalo-acetic transaminase, abnormal urine, and hematuria were reported more frequently in the DCLHb group, and there was one case of renal failure in each group. The hemodynamic effects of DCLHb included a consistent and slightly greater increase in systemic and pulmonary vascular resistance with associated increases in systemic and pulmonary arterial pressures compared with pRBC. Cardiac output values decreased more in the DCLHb group patients after the first administration than the control group patients. At 24 h postinfusion, the plasma hemoglobin level was less than one half the maximal level for any amount of DCLHb infused.
Administration of DCLHb allowed a significant number (19%) of cardiac surgery patients to avoid exposure to erythrocytes postoperatively.
RISKS associated with allogeneic blood remain a public concern 1–5and have prompted investigation into methods to avoid blood transfusions. Preoperative treatment to stimulate erythropoiesis, prophylactic treatment with antifibrinolytics, intraoperative cell saving, and the use of crystalloids and other volume expanders have contributed to decreased need for postoperative blood transfusion. 6Nevertheless, some patients undergoing cardiac surgery require an acute increase in blood oxygen-carrying capacity and therefore receive allogeneic blood transfusions.
Diaspirin cross-linked hemoglobin (DCLHb; Baxter Healthcare Corp., Deerfield, IL) is a solution of stabilized human hemoglobin that has been subjected to viral inactivation. 7–8The oxygen binding curve is slightly right shifted (P50= 32 mmHg) compared with fresh blood (P50= 26 to 28 mmHg) to enhance oxygen release to tissues. This has been evidenced by an increase in directly measured oxygen consumption in anemic, critically ill patients given 250 or 500 ml (one or two units, respectively) of DCLHb. 9In addition to its oxygen-carrying ability, DCLHb has demonstrated a pressor effect in both animals and humans. 10–16Infusion of 100–500 ml of DCLHb in intensive care patients who required increasing doses of vasopressor agents to maintain blood pressure allowed the reduction and, in some patients, discontinuation of vasopressor therapy. 16Preclinical studies with DCLHb in models of hemorrhagic shock suggest that it has resuscitation potential comparable with that of autologous blood. 17,18
The purpose of the current study was to assess the efficacy and safety profile, hemodynamic effects, and plasma persistence of DCLHb when used as a substitute for allogeneic blood in post–cardiac bypass surgery patients requiring a blood transfusion.
Materials and Methods
Diaspirin cross-linked hemoglobin is a 10% solution of modified hemoglobin in a balanced electrolyte solution that is similar to Ringer’s lactate. The hemoglobin is acquired from human erythrocytes that are osmotically lysed to release hemoglobin. After ultrafiltration, the purified, stroma-free hemoglobin is reacted with the cross-linking agent, bis(3,5-dibromosalicyl) fumarate, which links the α 99 lysines of the hemoglobin molecule to produce a stabilized tetrameric hemoglobin. 19The solution is heat treated and ultrafiltered to effect viral inactivation and remove extraneous proteins. 7,8Typical characteristics of the resulting hemoglobin solution are listed in table 1. 20The volume of one unit of DCLHb is 250 ml and contains 25 g of hemoglobin. Typed and cross-matched packed erythrocytes (pRBCs) were provided by each institution’s blood bank in the routine manner.
Diaspirin cross-linked hemoglobin was compared with pRBCs in a randomized, active-controlled, single-blind study in post–cardiopulmonary bypass patients at eight European hospitals. The protocol was approved by each local ethics committee, and informed consent was obtained from each patient before participation.
Patient Selection and Randomization
All patients scheduled to undergo elective cardiac surgery were screened for potential study participation according to the study inclusion criteria listed in table 2. After cardiac surgery and within 12 h after removal from cardiopulmonary bypass, patients manifesting a defined transfusion indication (table 3) and meeting treatment inclusion criteria (table 2) were randomized to receive either 250 ml of DCLHb or one unit of pRBCs. Infusions of either solution were given intravenously over a time period ranging from 15 min to 3 h according to the patient’s condition. If necessary, infusion of the randomized solution could be repeated twice (for a total of 750 ml DCLHb or 3 units pRBCs) within 24 h after removal from cardiopulmonary bypass. The same treatment inclusion and exclusion criteria were followed for each subsequent infusion given during the 24-h treatment period. DCLHb treatments could not be given later than 24 h after bypass.
All patients could, at any time, receive any additional treatments deemed necessary, including other blood products. In addition, patients randomized to receive DCLHb could proceed to standard therapy (pRBCs) before complete administration of all three units of DCLHb if the patient’s response to DCLHb was considered unfavorable.
Study End Points
The two prospectively defined primary end points were: (1) the avoidance of banked pRBC transfusion or transfusions, as indicated by the number of patients in the DCLHb group who were spared a transfusion through 7 days after surgery or hospital discharge, whichever came first, and (2) a significant decrease in the total number of units of pRBC or whole blood administered to the DCLHb group compared with the control group.
The safety profile of DCLHb was assessed by comparing the mortality, adverse events, serious adverse events, time to first bowel movement, and clinical laboratory values reported in the DCLHb group with those variables in the control group. Adverse events were deemed serious if they were associated with death, prolonged hospitalization, or morbidity.
The hemodynamic response to DCLHb was determined by measuring mean arterial pressure, pulmonary artery pressure, right atrial pressure (RAP), pulmonary artery wedge pressure, and cardiac output (CO) before and after each infusion, hourly through 12 h postbypass, and then every 4 h through 24 h postbypass. CO measurements were made via the continuous thermodilution technique (Vigilance, Baxter-Edwards, Irvine, CA). The continuous thermodilution method was not used until at least 15 min after discontinuation from cardiopulmonary bypass and the patient’s core temperature exceeded 35.5°C. Systemic vascular resistance and pulmonary vascular resistance were calculated according to the following formulas: systemic vascular resistance =[(mean arterial pressure − mean RAP)/CO]× 80 and pulmonary vascular resistance =[(mean pulmonary artery pressure − mean pulmonary artery wedge pressure) /CO]× 80.
The cardiac enzymes creatine kinase, creatine kinase–myocardial band, lactate dehydrogenase with isozymes, and Troponin I; the renal parameters blood urea nitrogen and creatinine; the liver enzymes aspartate amino transferase, alanine aminotransferase, alkaline phosphatase, γ-glutamyl transpeptidase, and total bilirubin; the pancreatic enzymes amylase and lipase; and the hematologic parameters hemoglobin, hematocrit, and reticulocyte count were assessed before surgery and at 24 h, 72 h, and 7 days postbypass. Plasma hemoglobin was assessed before and after each infusion and at 12 h, 24 h, 72 h, and 7 days postbypass. Arterial blood gases and lactate levels were assessed before and after each infusion and at 12 and 24 h postbypass; lactate levels were also measured at 72 h and 7 days postbypass.
The presence of DCLHb in plasma has the potential to interfere with certain clinical laboratory determinations. Before study initiation, the clinical laboratory at each trial site performed interference testing which established a DCLHb interference cutoff value for each laboratory assay. During the trial, the level of free DCLHb present in laboratory samples was estimated by visual comparison to a set of DCLHb standards. Depending on the amount of free DCLHb present and the equipment being used, samples were either diluted or a calculated correction factor was employed. However, to assure standardized technique, total bilirubin, amylase, pancreatic-specific amylase, lipase, and Troponin I analyses were performed at a central laboratory facility using high-performance liquid chromatography methods. Because DCLHb was administered in varying amounts (250–750 ml) over varying time periods during the 24-h postbypass period, standard pharmacokinetic assessment could not be performed. However, plasma hemoglobin concentration was measured at various time points through 72 h postinfusion.
All analyses were performed using SAS version 6.10 (SAS Inc., Cary, NC) with OS/2 version 2.1. Confidence intervals were constructed for time points day 0 (day of surgery) through day 7 to test the significance of the percentage of patients avoiding transfusion. Daily and total blood volumes transfused were analyzed by analysis of covariance. Clinical laboratory data were tabulated, and mean changes were compared between the two groups using the repeated measures analysis of variance. Vital signs and hemodynamic mean changes from baseline were compared between the groups using analysis of variance. Each statistical test was two-tailed at the 0.05 probability level.
Demographics and Patient Disposition
Of the 1,956 patients screened, a total of 209 patients met the defined criteria (tables 2 and 3) and were enrolled and randomized to receive either DCLHb or pRBCs (fig. 1). Of the 104 patients randomized to receive DCLHb, 17 patients received only one unit, 26 patients received only two units, and 61 patients received three units. Of the 105 patients randomized to receive pRBCs, 34 patients received one unit, 43 patients received two units, and 28 patients received three units. Baseline patient demographics, as listed in table 4, were similar in both the DCLHb and control groups.
Fourteen patients died during the study, and four patients (two from each group) deviated from the protocol; all available data from these patients were included in the analyses. Treatment with DCLHb was withheld or discontinued in seven patients in the DCLHb group due to an adverse event. Reasons for discontinuation included uncontrolled hypertension (n = 2), myocardial infarction (n = 2), electrocardiographic changes (n = 1), severe hemorrhage (n = 1), and gross hematuria (n = 1).
Efficacy End Points
Because randomization occurred at the time a blood transfusion was deemed necessary, all 105 patients in the control group received at least one unit of pRBCs. Of the 104 DCLHb recipients, 20 patients did not receive a blood transfusion through postoperative day 7 or discharge from the hospital, whichever came first (P < 0.05). The number and percentage of DCLHb patients who were not transfused each day after surgery are listed in table 5. On the day of surgery (day 0), 61 of 104 DCLHb patients did not receive a transfusion. However, by day 5, the number of DCLHb-treated patients who avoided a transfusion decreased to 20. The number of DCLHb patients not transfused was statistically different from the control group on each of the 7 days after surgery.
The cumulative average number of units of pRBCs administered during the study period is shown in table 6. Overall, there was no difference between the groups in the average total amount of pRBCs infused, although patients in the DCLHb group received significantly fewer pRBC units on day 0 and day 1. The average daily as well as overall cumulative volume of other blood products was not different between the two groups.
Safety Profile: Adverse Events
Six DCLHb-treated patients and eight pRBC-treated patients died during the course of the study. All DCLHb-treated patients and 102 of 105 pRBC-treated patients experienced at least one adverse event during the study period. There were more serious adverse events reported in the DCLHb group than in the control group (50 vs. 26 events, respectively), and a greater number of patients in the DCLHb group than the control group experienced at least one serious adverse event (33 vs. 21 events, respectively). There were greater incidences of the specific adverse events of hypertension (39 DCLHb vs. 17 pRBC), jaundice (44 DCLHb vs. 0 pRBC), increased liver enzymes (15 DCLHb vs. 3 pRBC), increased pancreatic enzymes (7 DCLHb vs. 1 pRBC), anemia (19 DCLHb vs. 6 pRBC) and hematuria/hemoglobinuria (20 DCLHb vs. 6 pRBC) in the DCLHb group compared with the control group. Jaundice occurred in 1 of 17 patients who received 250 ml of DCLHb, in 10 of 26 patients who received 500 ml of DCLHb, and in 33 of 61 patients who received 750 ml of DCLHb. Seventeen patients (eight DCLHb-treated and nine control-treated) returned to surgery for treatment of bleeding complications.
The mean time interval between surgery and the patient’s first bowel movement was recorded for 64 DCLHb-treated patients and 71 control-treated patients. The mean times did not differ between the two groups (4.4 ± 1.7 vs. 4.2 ± 1.6 days for the DCLHb and control groups, respectively).
Clinical Laboratory Findings
Mean hemoglobin concentrations did not differ between the two treatment groups at any time point (table 7). The mean reticulocyte counts, however, increased more from baseline values in the DCLHb group than in the control group at 24 and 72 h postbypass. All other clinical laboratory parameters showing significant between group differences are provided in table 8.
Changes in mean arterial pressure and systemic vascular resistance, pulmonary artery pressure and pulmonary vascular resistance, and CO and heart rate are shown in figures 2A, 2B, and 2C, respectively. Statistically significant differences were observed in these parameters at several times after infusion, as indicated in figures 2A, 2B, and 2C.
Mean RAP increased minimally (< 1 mmHg) after infusions of DCLHb or pRBCs, with only one significant difference between groups occurring after the third infusion, when the mean RAP was 10 ± 4 mmHg in the DCLHb group compared with 9 ± 5 mmHg in the control group. Mean RAP values ranged from 9 to 10 (± 4) mmHg in both treatment groups throughout the 24-h period immediately after bypass. In both treatment groups, mean changes in pulmonary artery wedge pressure were small (< 2 mmHg) and nonsignificant after all infusions. During the 24-h postbypass period, mean pulmonary artery wedge pressure ranged from 11 to 12 mmHg in both treatment groups.
Plasma Persistence of DCLHb
Plasma hemoglobin levels increased in the DCLHb group after each infusion, with mean values of 474 mg/dl after the first infusion, 927 mg/dl after the second infusion, and 1,315 mg/dl after the third infusion of DCLHb (fig. 3). By 24 h after the start of the first infusion of any dose, the plasma hemoglobin level was less than one half the peak value. At 72 h after the first infusion, plasma hemoglobin levels for all patients had essentially returned to baseline.
One of the two primary efficacy end points of this study was achieved; a significant number of post–cardiac surgery patients who received DCLHb at the time a blood transfusion was deemed necessary subsequently were spared exposure to allogeneic erythrocytes. As these patients had already received various blood conservation and transfusion avoidance treatments such as cell saver and antifibrinolytic therapy, this finding is of clinical significance in that there was additional avoidance of transfusion (19%) beyond that achieved by current practice. The other primary end point, a significant reduction in the total number of units of pRBC or whole blood administered was not achieved, because the DCLHb-treated patients as a group did not receive less blood and blood products through the 7-day study period. However, the significant reduction in the amount of blood transfused in the DCLHb group compared with the control group on day 0 and day 1 suggests that transfusion was effectively delayed by infusion of DCLHb. Given the first 24-h postsurgery dosing regimen used in this study, as well as the relatively short plasma persistence of DCLHb compared with erythrocytes, a transfusion delay might have been a more realistic end point in this particular study.
The design of the study as well as varying transfusion practices 21may be responsible, in part, for both the low transfusion avoidance rate and the lack of a reduction in total amount of blood transfusion at the 7-day time point. Limiting the treatment period to 24 h after cardiopulmonary bypass and the amount of DCLHb to 750 ml may have seriously reduced the DCLHb group’s chances of meeting the efficacy end points of transfusion avoidance and reduced blood utilization. These protocol limitations may have been aggravated by an attempt on the part of some clinicians to either maintain the patients’ postoperative hemoglobin values at a certain level or to achieve hemoglobin values ≥ 10 g/dl before hospital discharge. The mean hemoglobin value of DCLHb-treated patients at 24 and 72 h after surgery was < 10 g/dl, a transfusion trigger still accepted by many clinicians despite current transfusion recommendations. 22–24In contrast, the mean hemoglobin values of the control group at these times were > 10 g/dl. A hemoglobin level < 10 g/dl in the DCLHb group may have prompted some clinicians to transfuse the patient despite reported evidence of improved patient outcomes using a restrictive strategy of transfusion based on lower (7.0 to 9.0 g/dl) hemoglobin values. 25Using the hemoglobin value as the primary transfusion trigger is problematic when comparing oxygen-carrying solutions with blood. Although a unit (250 ml) of DCLHb may provide a positive clinical response comparable to a unit of pRBCs, a unit of DCLHb will not increase whole-blood hemoglobin concentrations to the same extent as a unit of pRBCs. This is due in part to the fact that DCLHb is a 10% solution in which one unit (250 ml) provides 25 g of hemoglobin compared with the ≥ 60 g of hemoglobin provided by one unit of pRBCs. In addition, DCLHb is hyperoncotic (oncotic pressure approximately 42 mmHg), which leads to plasma volume expansion via hemodilution with an apparent decrease in total hemoglobin. Thus, to achieve a hemoglobin concentration of 10 g/dl, most patients would require a transfusion of erythrocytes, as did 81% of the DCLHb-treated patients by the postoperative day 7.
Regarding the safety findings, the 14 deaths in this study (6 in the DCLHb group; 8 in the pRBC group) are high for routine cardiac surgery, although the majority of the patients in this study were considered to be high risk as defined, in part, by the substantial number of patients classified as American Society of Anesthesiologists grade 3, as well as the amount of blood loss and need for transfusion. 26Thus, they represent a more clinically unstable subpopulation of patients selected from a larger group of cardiac surgery patients. Although the number of deaths was balanced, there were more adverse events and serious adverse events reported in the DCLHb recipients.
The myocardial isoenzymes creatine kinase–myocardial band and lactate dehydrogenase-1 did not increase in the DCLHb group, and the Troponin I increase from baseline was less than that in the control group. These results suggest that there was no cost in terms of myocardial stress or damage for using DCLHb at the time a blood transfusion was initially warranted. Because these patients were awakening from anesthesia and rewarming, a solution that did not effectively carry oxygen or a solution that stressed the patient further as a result of toxicity would likely cause more myocardial insult than that observed in the control group. The fact that this did not occur supports the safety of DCLHb in this patient population.
Clinical chemistry results that appear to be related to DCLHb include elevations of lipase and pancreatic-specific amylase. Although there were no cases of pancreatitis in either group and the increases were not indicative of significant injury, DCLHb apparently is directly or indirectly responsible for the elevations in these enzyme levels. Increases in these enzyme levels could be caused by impaired clearance as a result of competitive inhibition rather than tissue injury. Elevations in serum amylase and lipase levels after infusion of low doses of a recombinant human hemoglobin solution have been postulated to be related to inhibition of the nitric oxide pathway. 27
The increase in total bilirubin within 24 h of treatment is consistent with the rapid clearance of the acellular hemoglobin DCLHb. Although the serum creatinine also increased more in the DCLHb group than in the control group, the minimal difference (0.1 mg/dl) does not suggest renal impairment. The observed increases in total creatine kinase and the isoenzyme lactate dehydrogenase-5 are consistent with those observed in the phase I study with DCLHb 13and possibly indicate a skeletal muscle effect; however, there were no cases of rhabdomyolysis and no reports of increase in muscle pain after DCLHb infusion.
The pressor effect of DCLHb observed in this patient population is consistent with that previously reported in normal volunteers, 13hemodialysis patients, 14and intensive care patients, 16and this effect has also been observed with other hemoglobin solutions. 27–29Preclinical studies suggest that binding of nitric oxide, increases in endothelin sensitivity, and upregulation of α1- and α2-adrenergic receptors combine to produce this effect. 30–32In this study, DCLHb caused both the systemic and pulmonary arterial pressures to rise above baseline after the first infusion, but these pressures did not increase further after subsequent infusions, and no adverse clinical sequelae were reported as a consequence of the increase in blood pressure. In particular, the increases in blood pressure and vascular resistance were not associated with decreased tissue perfusion, as indicated by the equivalent effects of DCLHb and pRBCs on gastrointestinal function (time to first bowel movement), base excess, pH, and arterial lactate.
The decline in CO after DCLHb administration, although statistically significant, was not clinically significant and likely is due to the increase in systemic vascular resistance. This finding has been observed in other clinical trials of DCLHb 9,15,16as well as in studies of other right-shifted oxygen carriers. 28,33
The major shortcoming of this study is the lack of blinding of the clinical investigators. The requirement for bedside immunohematology testing by the responsible physician at several of the study sites preempted blinding of the treating physician; by local regulation, the task could not be delegated. The inability to blind the study investigators was compounded by the lack of a fixed transfusion trigger and varying transfusion practices among the sites. Although the study conduct reflects clinical practice, a consistent transfusion trigger employed by all sites would have added more objectivity to the transfusion decision-making process. This is especially important because the efficacy end points were based on the decision to further transfuse after each treatment.
The results of this study demonstrate that infusion of up to 750 ml of DCLHb allowed 19% of postoperative cardiac surgery patients who otherwise would have received allogeneic blood to avoid exposure to erythrocytes. This could have the potential benefits of reducing the risk of transfusion-transmitted infections, immunosuppression, and clerical errors. 1–5However, given the eventual transfusion of most DCLHb recipients, the higher incidence of adverse events, and the solution’s relatively short plasma persistence, the routine use of DCLHb for transfusion avoidance in this population is not supported at this time.
DCLHb is no longer being developed for clinical use.
The authors are indebted to the skills and hard work of the coinvestigators, the study coordinators, and research assistants at all participating sites and to Tony Reppucci and Danielle Gerard of Baxter Healthcare International for study management and administrative support.
The DCLHb Cardiac Surgery Trial Collaborative Group included the following co-investigators: Helene Bouvier, M.D. (Hôpital Louis Pradel, Lyon, France); Jean-Luc Canivet, M.D. (Centre Hospitalier Universitaire, Liège, Belgium); Pascal Chiari, M.D. (Hôpital Louis Pradel, Lyon, France); Pierre-Georges Durand, M.D. (Hôpital Louis Pradel, Lyon, France); Magnus Garrioch, M.D. (The Royal Infirmary of Edinburgh, Edinburgh, United Kingdom); Simone Massonnet-Castel, M.D. (Hôpital Broussais, Paris, France); Armelle Nicolas-Robin, M.D. (Hôpital Broussais, Paris, France); Monique Paris, Pharm.D. (Hôpital Broussais, Paris, France); Edmundo Pereira de Suza, M.D. (Hôpital Louis Pradel, Lyon, France); David H. T. Scott, MBChB, F.R.C.A. (The Royal Infirmary of Edinburgh, Edinburgh, United Kingdom); Didier Sirieix, M.D. (Hôpital Broussais, Paris, France)