Infants and children who undergo repeat sternotomy for repair of congenital cardiac defects are at increased risk of postoperative bleeding . Postcardiopulmonary bypass platelet dysfunction [2,3], dilutional coagulopathy , and abnormal fibrinolysis [5,6] contribute to this bleeding tendency. Tranexamic acid is a lysine analog that competitively binds to the lysine-binding sites of plasmin and plasminogen. It inhibits fibrinolysis with 6- to 10-fold greater potency than aminocaproic acid. Tranexamic acid effectively reduces blood loss and transfusion requirements after cardiac surgery in adults [7-13]. Zonis et al.  recently published the first study evaluating the efficacy of tranexamic acid (single 50 mg/kg dose) in children. However, these authors only demonstrated efficacy in a subgroup of children with preoperative cyanosis. We sought to evaluate the efficacy of tranexamic acid utilizing a larger dose (100 mg/kg after induction, followed by 10 mg [centered dot] kg-1 [centered dot] h-1 and 100 mg/kg in pump prime) in a population of children undergoing repeat sternotomy and cardiopulmonary bypass.
At Children's Hospital in Boston, between February 1994 and May 1996, informed consent was obtained from the parents of 43 children who were scheduled to undergo elective repeat cardiac surgery via sternotomy with cardiopulmonary bypass. All subjects were aged 6 mo to 12 yr and had undergone at least one previous sternotomy. Children with preexisting coagulopathy or preoperative anticoagulation were excluded. Subjects were randomized to receive either tranexamic acid or placebo. The study was conducted with institutional approval.
After induction of anesthesia and before skin incision, patients in the treatment group received tranexamic acid 100 mg/kg diluted to a fixed volume of 20 mL with normal saline, and patients in the placebo group received normal saline 20 mL, both over 15 min. An infusion of tranexamic acid 10 mg [centered dot] kg-1 [centered dot] h-1 or normal saline was then begun and continued until transport to the intensive care unit (ICU). Immediately upon initiation of cardiopulmonary bypass, a second bolus of tranexamic acid 100 mg/kg or normal saline was injected into the pump reservoir. All solutions were prepared by an individual not involved in the clinical management and were presented to the operating room team in a blind manner. The anesthesiologists, surgeons, intensivists, and laboratory personnel were all kept blinded to the randomization group. Seven anesthesiologists and three surgeons were involved in this study.
Immediately prior to aortic cannulation, heparin (200 U/kg for children less than 30 kg and 300 U/kg for children more than 30 kg) was injected into the atrium. The circuit was primed with heparin (250 U/100 mL of prime volume). During bypass, the celiteactivated clotting time (ACT) was maintained for more than 480 s with additional heparin as needed. After separation from bypass and atrial decannulation, heparin was reversed with protamine 4 mg/kg. An ACT measurement was obtained 5 min after protamine administration, and additional protamine, 0.5-1 mg/kg, was administered if the ACT was more than 150 s. The surgeon placed two mediastinal and/or chest tubes in all patients. These drainage tubes were connected to a graduated, water-sealed collection reservoir with a suction of 20 cm H2 O.
Total blood loss volume was defined as the total blood volume collected in the surgical suction and in the mediastinal and/or chest tubes minus irrigation fluid during the time period extending from the initial protamine administration until 24 h after admission to the ICU. This total blood loss volume specifically did not include any blood or irrigation collected prior to protamine administration, nor any blood on the surgical sponges or drapes. Blood components were transfused in the operating room at the discretion of the attending anesthesiologist and in the ICU at the discretion of the attending intensivists according to individual, nonstandardized practice. Total blood transfusion volume included the volume of all whole blood, packed red cells, or reconstituted whole blood that was transfused during the time period extending from the initial protamine administration until 24 h after admission to the ICU. Total red cell unit exposure (whole blood units and packed red cell units) and total nonred cell unit exposure (platelet units, cryoprecipitate units, and fresh-frozen plasma units) included the respective units that were transfused during the time period extending from entry into the operating room until 24 h after admission to the ICU. Reconstituted whole blood units, each of which was composed of one packed red cell unit and one fresh-frozen plasma unit, counted toward the total of both red and nonred cell unit exposures. Thus, blood components that were transfused prior to initial protamine administration were included in the total unit exposure but not in the total blood transfusion volume. Total donor exposure was defined as the sum of total red cell unit exposure and total nonred cell unit exposure.
Arterial blood samples were used to determine hematocrit, prothrombin time (PT), partial thromboplastin time (PTT), and fibrinogen concentration at 10 time periods during the study-immediately before study drug, 10 min after study drug load, 5 min after initial heparin administration, 5 min after onset of cardiopulmonary bypass, 30 min after onset of cardiopulmonary bypass, during rewarming at 32 degrees C esophageal temperature, immediately after separation from bypass, 5 min after heparin reversal, within 30 min of admission to the ICU, and the next morning in the ICU. Platelet quantification was performed only at the latter two time points. Colorimetric assay of unfractionated and low molecular weight heparins (Diagnostica Stago, Asnieres-Sur-Seine, France) and ACT determination were obtained 5 min after protamine administration. The duration of various events before, during, and after cardiopulmonary bypass were also recorded. Closure time was defined as the interval from aortic decannulation to sternal closure minus any interval of hemodynamic instability. Any occurrence of hemodynamic instability after study drug administration was noted. Cases of thrombotic complications, postoperative renal dysfunction, and other major morbidity and mortality were recorded.
The financial cost of all blood components transfused to the subjects was calculated based on the following 1996 list prices of the American Red Cross Blood Services, New England region: whole blood, $160.50 per unit; packed red cells, $98.25 per unit; random donor platelets, $87.25 per unit; apheresis platelets, $646.50 per bag; fresh-frozen plasma, $60.75 per unit; and cryoprecipitate, $46.75 per unit. The pharmacy cost of tranexamic acid at Children's Hospital is $12.94 per 1-g vial.
Statistical analyses were performed using JMP 3.1.5 (SAS Institute, Cary, NC). Categorical data were compared between groups with G2-likelihood ratio chi squared test. Continuous data were evaluated with the Shapiro-Wilk W-test to determine whether they were normally distributed. Several independent variables and most dependent variables were nonnormally distributed. Thus, all univariate analyses on continuous variables were performed with the Mann-Whitney U-test. Descriptive statistics were reported as frequency and percentage for categorical data and as median +/- quartile deviation for continuous data. Quartile deviation is equal to one-half the difference between the first and third quartiles. Selected variables were then analyzed using a multiple linear regression model. All of the blood loss and transfusion data were nonnormally distributed. Prior to applying this model, a natural logarithmic transformation of x + 1 was utilized, which reduced the pronounced positive skew and decreased the variation. The Shapiro-Wilk W-test confirmed that these transformed data were normally distributed with the exception of red cell and nonred cell unit exposure. The analyses for these two variables should be considered approximate. The multiple linear regression analysis was performed with a screening model involving six independent variables. Adjusted and unadjusted means were calculated by reverse transformation of the least-square means with and without covariates in the model. Statistical significance for all tests was accepted as P < 0.05.
Forty-three children were enrolled in the study. One patient was enrolled twice for different surgical procedures, which were separated by 20 mo, and two subjects were excluded from analysis due to profound intraoperative hemorrhage, which was clearly attributable to inadequate surgical hemostasis (i.e., internal mammary artery laceration and disruption of conduit suture line). The 41 subjects who completed the study included 20 in the tranexamic acid group and 21 in the placebo group. Independent variables are shown in Table 1. There was no statistical difference between the two groups with respect to age, weight, height, preoperative hematocrit, platelet count, and arterial hemoglobin oxygen saturation. The three most common surgical procedures were conduit or homograft revision (32%), fenestrated Fontan procedure (29%), and bidirectional Glenn procedure (17%). No statistically significant difference existed in the distribution of anesthesiologist, surgeon, and procedure type between the two groups. Median duration of cardiopulmonary bypass was 18 min greater in the placebo group (112 +/- 19 vs 94 +/- 21 min, P = 0.01). The aortic cross-clamp time, nadir esophageal temperature, and total heparin administered prior to and during bypass were comparable between the groups. Two subjects (one in the placebo group and one in the treatment group) underwent a brief period of deep hypothermic circulatory arrest (15 and 2 min, respectively). Fresh whole blood  (defined as less than 4 days old in our institution) was available 48% of the time in the placebo group and 25% of the time in the tranexamic acid group (P = 0.13).
Tranexamic acid was well tolerated by all subjects in the treatment group. There were no cases of hemodynamic instability, overt thrombotic complications, or other adverse effects associated with the bolus or infusion. Dependent variables are summarized in Table 2. In univariate analyses, total blood loss was reduced by 24% (P = 0.03), and total blood transfusion volume was reduced by 38% (P = 0.04) in the group that received tranexamic acid. The total red blood cell exposure was 2.0 units in the tranexamic acid group versus 3.0 units in the placebo group (P = 0.06), and the total donor exposure was 3.0 versus 4.0 units, respectively, (P = 0.10). Total cost of these blood components ($330 vs $495, P = 0.04) was less in the tranexamic acid group. In addition to these transfusion outcomes, the time interval from aortic decannulation to sternal closure (closure time) was 32% shorter in the tranexamic acid group (15 +/- 6 vs 22 +/- 6 min, P = 0.01). Subjects in the tranexamic acid group also tended to have a 30-min shorter total operating room time (4.8 +/- 1.0 vs 5.3 +/- 1.0 h, P = 0.15). The duration of mechanical ventilation in the ICU (19 +/- 2 vs 17 +/- 9 h, P = 0.90) and the total time spent in the ICU (48 +/- 14 vs 51 +/- 13 h, P = 0.07) was not significantly different between the tranexamic acid and placebo groups.
A multiple linear regression model was utilized to evaluate the contribution of six independent variables (treatment group assignment, extracorporeal bypass duration, anesthesiologist identity, surgeon identity, preoperative cyanosis versus acyanosis, and procedure type) toward the clinical outcome (Table 3). This analysis demonstrated that the treatment group assignment was the only significant factor associated with total blood loss (P = 0.01). Bypass duration had no significant influence upon total blood loss. In this screening model, none of the six independent variables was significantly associated with the transfusion requirements and financial cost outcome. Extracorporeal bypass duration did have an influence (P = 0.06) upon transfusion volume.
Hematologic and coagulation laboratory data are summarized in Table 4. Of the parameters analyzed, only the hematocrit value, which was measured upon arrival in the ICU, was demonstrated to be statistically different between the two groups (40 +/- 4% in the tranexamic acid group versus 33 +/- 5% in the placebo group, P = 0.01). To determine if the initial study drug bolus might alter coagulation parameters, the paired differences in PT and PTT before and after the initial bolus were examined. In the entire study population, PT increased 0.3 +/- 0.3 s (median +/- quartile deviation; one-tailed P < 0.01 when tested against a hypothesis of zero using Wilcoxon signed-rank test), and PTT increased 3 +/- 5 s after study drug bolus (P = 0.03). When these paired differences were compared in the placebo versus the tranexamic acid group, no difference was found (P = 0.44 for PT difference and P = 0.82 for PTT difference). Thus, subjects in both the placebo group and the tranexamic acid group experienced a slight increase in the PT and PTT duration after initial bolus. However, this increase was of equal magnitude when compared between the two groups.
In this prospective, blind, placebo-controlled study, we demonstrated that children who were treated with large-dose tranexamic acid had 24% less total blood loss compared with children who received placebo. Additionally, the total volume transfusion requirements and total unit exposure to banked blood components were less in the tranexamic acid group. Our findings parallel those of similar trials in the adult cardiac surgical population [7-13]. Zonis et al.  were the first to evaluate the hemostatic efficacy of tranexamic acid using a single 50-mg/kg dose in children undergoing cardiac surgery. However, this study only demonstrated a reduction in postoperative blood loss in a subgroup of 18 children with preoperative cyanosis. No significant difference was found in their full group of 88 children. The different results between these two studies is likely related to the different tranexamic acid dose (average total 250 mg/kg in this study versus 50 mg/kg in the study by Zonis and colleagues) and the different patient population (exclusively high-risk subjects undergoing repeat sternotomy in this study versus high- and low-risk subjects in the latter study).
There has been much variation in the tranexamic acid dosing regimen used in prior trials. Horrow and colleagues  conducted a dose-response study in 148 adult cardiac surgical patients. The initial loading doses in this trial were 0, 2.5, 5, 10, 20, and 40 mg/kg, followed by an infusion of one-tenth the bolus per hour. These investigators found that 10 mg/kg decreased bleeding 34% compared with placebo but that larger doses did not provide additional efficacy. On the other hand, Karski et al.  compared 50, 100, and 150 mg/kg tranexamic acid bolus doses and found 100 mg/kg to be significantly more effective than 50 mg/kg. Ralley et al.  compared large-dose (approximately 70 mg/kg) versus small-dose (10 mg/kg) regimens and found only the larger dose to be efficacious over placebo. In an effort to maximize antifibrinolytic activity and observed outcome, we chose to use a larger dose in this clinical trial. Pharmacokinetic data are limited, and pharmacodynamic data are unstudied for tranexamic acid in the pediatric population. Isetta and colleagues  have reported data in children, which show an 80% decline in tranexamic acid plasma concentration between the postbolus peak and the end of cardiopulmonary bypass when a continuous infusion was not used. It is possible that the single 50-mg/kg dose used by Zonis et al. before bypass may result in subtherapeutic plasma concentrations after bypass when the fibrinolytic mechanisms are most significant. It may be particularly important to administer the second bolus after onset of bypass in children because of the large alteration in circulating blood volume when the bypass circuit is introduced . In this study and the study by Zonis et al. , no thrombotic complications or other adverse effects were detected. Complications attributed to tranexamic acid in adult patients are very infrequent. The most critical concern is that tranexamic acid might promote a hypercoagulable state. Cases of cerebral [21-23], pulmonary , mesenteric , and retinal  thrombosis have been reported. A properly powered prospective study seeking either subtle or devastating complications of tranexamic acid has not yet been performed.
In the present study, the duration of cardiopulmonary bypass was 16% greater in the placebo group. We considered this duration to be an independent variable, and, thus, a chance occurrence in the randomization, because we could not conceive of a mechanism whereby presence or absence of antifibrinolytic therapy could affect bypass time. Nevertheless, as extracorporeal duration could potentially affect postbypass bleeding diathesis, multivariate analysis was performed to evaluate the possible influence of this and five additional variables upon the clinical outcome (Table 3). For total blood loss, total transfusion volume, total donor exposure, and total blood component cost, the unadjusted and adjusted means were less in the group that received tranexamic acid. For total blood loss, the difference in adjusted means was statistically significant, which shows that the efficacy of tranexamic acid persists even after accounting for the difference in bypass duration, anesthesiologist, surgeon, cyanosis, and procedure type. For total transfusion requirements, total donor exposure, and total blood component cost, the adjusted means were not significantly different. There are two possible explanations for this. First, the difference in bypass duration and other factors may account for enough of the transfusion differences to make them nonsignificant. Alternately, the complex adjustments in the multiple regression model may cause a loss in power, especially for this sample size. Although it is impossible to definitively rule out the first explanation, we feel that the consistency in the directions of the means, both unadjusted and adjusted, gives some credence to the second explanation.
Of importance, the hematocrit value upon admission to the ICU was significantly greater in the tranexamic acid group (40 +/- 4% vs 33 +/- 5%, P = 0.01). This difference developed between the administration of protamine and admission to the ICU. In this clinical trial, a standardized transfusion protocol was not used. Rather, the intraoperative indications for transfusion were individually determined by the staff anesthesiologist. At this institution and in this patient population, the hematocrit values during rewarming were relatively low. Thus, it has been a common practice to have whole blood immediately available for transfusion during separation from bypass and to use this blood for acute volume expansion and normalization of red cell mass. Assuming that the two groups were euvolemic upon admission to the ICU, these data suggest that those subjects in the tranexamic acid group were relatively "overtransfused" compared with the placebo group. It is quite possible that some patients may have been transfused based on a physician's prior experience rather than on specific indications. For example, if a given anesthesiologist is accustomed to empirically transfusing a certain volume of banked blood for a given patient type and procedure, in the absence of antifibrinolytic therapy, then that physician may overtransfuse a patient who, in fact, did not bleed as much as history dictates. Had standardized transfusion thresholds been used during this clinical trial, it is possible that the transfusion requirements would have been less in the tranexamic acid group or greater in the placebo group, resulting in a greater transfusion difference that would meet statistical significance both in univariate and multivariate analysis.
Recent and continued improvements in donor screening and testing have resulted in a dramatic decrease in transfusion-related disease transmission. A recent publication reports that the risk of acquisition of any virus from a unit of blood is now approaching 1 in 70,000 and that the risk of contracting human immunodeficiency virus is estimated to be approximately 1 in 500,000 units . Nevertheless, pediatric cardiac surgical patients often require multiple staged operations over several years and are exposed to multiple units of blood and blood components. The risks of more frequent noninfectious complications of transfusion, such as transfusion reactions, metabolic complications, and alloimmunization, are all proportional to the number of units to which the patient is exposed . Hence, efforts to minimize the number of units to which a patient is exposed are still warranted.
A financial cost analysis was performed based on 1996 list prices paid to the American Red Cross Blood Services with no additional costs added. The cost of blood components was one-third less in the tranexamic acid group ($330 vs $495, univariate P = 0.04). The median per patient charge for tranexamic acid in the treatment group was $52. This immediate cost analysis does not include the potential cost-savings of reducing blood component exposure and the potential for transfusion-related complications.
We conclude that, in pediatric patients undergoing repeat cardiac surgery, large-dose tranexamic acid effectively reduces blood loss in children undergoing repeat cardiac surgery. The absence of a standardized transfusion protocol in this study limited the conclusions that could be made regarding the reduction in transfusion requirements. Although the safety of large-dose tranexamic acid has not yet been thoroughly established, no thrombotic complications or other adverse events were detected in this clinical trial.
The authors express their appreciation to Nicholas Morana, JD, Lorna Sullivan, RN, and Lynne Zienko, MT (NCA-CLS) for their technical and organizational assistance with this project, and to Stephen D. Simon, PhD (Children's Mercy Hospital, Kansas City, MO), for consultation regarding statistical analysis.
1. DeLeon SY, LoCiceri J III, Ilbawi MM, Idris FS. Repeat median sternotomy in pediatrics: experience with 164 consecutive cases. Ann Thorac Surg 1986;41:184-8.
2. Greeley WJ, Bushman GA, Kong DL, et al. Effects of cardiopulmonary bypass on eicosanoid metabolism during pediatric cardiovascular surgery. J Thorac Cardiovasc Surg 1988;95:842-9.
3. Spiess BD. Cardiac anesthesia risk management. Hemorrhage, coagulation, and transfusion: a risk-benefit analysis. J Cardiothorac Vasc Anesth 1994;8:19-22.
4. Kern FH, Morana NJ, Sears JJ, Hickey PR. Coagulation defects in neonates during cardiopulmonary bypass. Ann Thorac Surg 1992;54:541-6.
5. Kucuk O, Kwaan HC, Frederickson J, et al. Increased fibrinolysis in patients undergoing cardiopulmonary bypass operation. Am J Hematol 1986;23:223-39.
6. Petaja J, Peltola K, Sairanen H, et al. Fibrinolysis, antithrombin III, and protein C in neonates during open heart surgery [abstract]. Anesth Analg 1996;82:SCA20.
7. Horrow JC, Hlavacek J, Strong MD, et al. Prophylactic tranexamic acid decreases bleeding after cardiac operations. J Thorac Cardiovas Surg 1990;99:70-4.
8. Horrow JC, Van Riper DF, Strong MD, et al. Hemostatic effects of tranexamic acid and desmopressin during cardiac surgery. Circulation 1991;84:2063-70.
9. Speekenbrink RG, Vonk AB, Wildevuur CR, Eijsman L. Hemostatic efficacy of dipyridamole, tranexamic acid, and aprotinin in coronary bypass grafting. Ann Thorac Surg 1995;59:438-42.
10. Rousou JA, Engelman RM, Flack JE, et al. Tranexamic acid significantly reduces blood loss associated with coronary revascularization. Ann Thorac Surg 1995;59:671-5.
11. Coffey A, Pittmam J, Halbrook H, et al. The use of tranexamic acid to reduce postoperative bleeding following cardiac surgery: a double-blind randomized trial. Am Surg 1995;61:566-8.
12. Nakashima A, Matsuzaki K, Fukumura F, et al. Tranexamic acid reduces blood loss after cardiopulmonary bypass. ASAIO J 1993;39:M185-9.
13. Karski JM, Teasdale SJ, Norman PH, et al. Prevention of post-bypass bleeding with tranexamic acid and epsilonaminocaproic acid. J Cardiothorac Vasc Anesth 1993;7:431-5.
14. Zonis Z, Seear M, Reichert C, et al. The effect of preoperative tranexamic acid on blood loss after cardiac operations in children. J Thorac Cardiovasc Surg 1996;111:982-7.
15. Manno CS, Hedberg KW, Kim HC, et al. Comparison of the hemostatic effects of fresh whole blood, stored whole blood, and components after open heart surgery in children. Blood 1991;77:930-6.
16. Horrow JC, Van Riper DF, Strong MD, et al. The dose-response relationship of tranexamic acid. Anesthesiology 1995;82:383-92.
17. Karski J, Joiner R, Carroll J, et al. Comparison of the effect of three different doses of tranexamic acid on postoperative bleeding in cardiac surgery performed without active cooling [abstract]. Society of Cardiovascular Anesthesiologists 16th Annual Meeting, Montreal, Quebec, Canada, April 1994.
18. Ralley F, DeVarennes B, Rotitaille M. Comparison of high vs. low dose tranexamic acid on blood loss and blood product requirement after cardiac surgery [abstract]. Anesth Analg 1996;82:SCA8.
19. Isetta C, Garraffo R, Merville C, et al. Pharmacokinetic study of tranexamic acid during cardiac surgery with cardiopulmonary bypass [abstract]. Anesth Analg 1996;82:S199.
20. Gruenwald CE, Andrew M, Burrows FA, Williams WG. Cardiopulmonary bypass in the neonate. In: Karp RB, Laks H, Wechsler AS, eds. Advances in cardiac surgery. vol. 4. Chicago: Mosby Year Book, 1993:137-56.
21. Agnelli G, Gresele P, De Cunto M, et al. Tranexamic acid, intrauterine contraceptive devices and fatal cerebral arterial thrombosis. Case report. Br J Obstet Gynaecol 1982;89:681-2.
22. Davies D, Howell DA. Tranexamic acid and arterial thrombosis. Lancet 1977;1:49.
23. Rydin E, Lundberg PO. Tranexamic acid and intracranial thrombosis [letter]. Lancet 1976;2:49.
24. Woo KS, Tse LK, Woo JL, Vallance-Owen J. Massive pulmonary thromboembolism after tranexamic acid antifibrinolytic therapy. Br J Clin Pract 1989;43:465-6.
25. Razis PA, Coulson IH, Gould TR, Findley IL. Acquired C1 esterase inhibitor deficiency. Anaesthesia 1986;41:838-40.
26. Parsons MR, Merritt DR, Ramsay RC. Retinal artery occlusion associated with tranexamic acid therapy. Am J Ophthalmol 1988;105:688-9.
27. Schreiber GB, Busch MP, Kleinman SH, Korelitz JJ. The risk of transfusion-transmitted viral infections. The Retrovirus Epidemiology Donor Study. N Engl J Med 1996;334:1685-90.
28. Sazama K. Reports of 355 transfusion-associated deaths: 1976 through 1985. Transfusion 1990;30:583-90.