The present study was registered in the public registry ClinicalTrials.gov (number NCT00985920). Approval was obtained from Health Canada, the Research Ethics Board of University Health Network, Toronto, Ontario, Canada, and written informed consent was obtained from all participants.
All adult patients (patients over the age of eighteen years) who were scheduled for a primary unilateral total knee arthroplasty with cement at Toronto Western Hospital, Toronto, Ontario, Canada were eligible for inclusion in the study. The exclusion criteria included allergy to tranexamic acid, acquired disturbances of color vision, preoperative anemia (a hemoglobin value of <11 g/dL in females and <12 g/dL in males), refusal of blood products, preoperative use of anticoagulant therapy within five days before surgery, fibrinolytic disorders requiring intraoperative antifibrinolytic treatment, coagulopathy (as identified by a preoperative platelet count of <150,000/mm3, an international normalized ratio of >1.4, or a prolonged partial thromboplastin time [>1.4 times normal]), a history of arterial or venous thromboembolic disease (such as a cerebrovascular accident, deep-vein thrombosis, or pulmonary embolus), pregnancy, breastfeeding, major comorbidities (such as severe ischemic heart disease [New York Heart Association Class III or IV], previous myocardial infarction, severe pulmonary disease [forced expiratory volume <50% of normal], plasma creatinine of >115 µmol/L in males and >100 µmol/L in females, or hepatic failure), and participation in another clinical trial.
If intraoperative surgical, medical, or anesthetic complications occurred, such as myocardial infarction, intraoperative fracture, or neurovascular injury, the study medication was not administered and the patient was excluded from the study. However, if one of these problems occurred after the application of the study medication, the patient was excluded from outcome measurements and analysis but was followed for the occurrence of complications.
Patients participating in the study were still offered the usual standard of care for blood conservation at our institution. According to a blood-conservation protocol at the University Health Network, preoperative erythropoietin (40,000 to 80,000 IU) was offered to all patients with a hemoglobin level of <13 g/dL and autologous blood donation of two units was allowed two to three weeks before the proposed surgery. Patients who were taking acetylsalicylic acid, anti-platelet agents, or nonselective cyclooxygenase inhibitors were advised to discontinue these medications seven to ten days prior to the scheduled surgery.
The doses of tranexamic acid chosen for the current investigation were based on previous studies of intravenous tranexamic acid in patients undergoing total knee arthroplasty showing efficacy with doses of 15 to 20 mg/kg (i.e., 1.5 g) and a meta-analysis suggesting that doses of >30 mg/kg (i.e., 3 g) were more effective for reducing bleeding15. Therefore, patients undergoing an elective primary unilateral total knee arthroplasty with cement were randomized with use of a computer-generated randomization table with block sizes of 4 to receive a solution containing either (1) 1.5 g (100 mg/mL) of tranexamic acid (Cyklokapron; Sandoz, Boucherville, Quebec, Canada) in 100 mL of normal saline solution, (2) 3 g of tranexamic acid in 100 mL of normal saline solution, or (3) an equivalent volume of placebo (normal saline solution). A stratified randomization for surgeons was used to account for differences in surgical practice. The randomization schedule was kept inaccessible throughout the study period. Patient assignments were placed into sequentially numbered opaque sealed envelopes and were kept by a research pharmacist. On the day of surgery, the envelope was opened and the study medication was prepared under sterile conditions by a research pharmacist who was not involved in the care of the patients. The study medications were identical in appearance. The patients, surgeons, anesthesiologists, health-care providers, research personnel, and outcome assessors were blinded to the randomization.
All patients received a femoral nerve catheter and a sciatic nerve block with ropivacaine (0.2%) for postoperative analgesia. Patients received either a spinal anesthetic with 3 mL of 0.5% bupivacaine and 100 µg of morphine or general anesthesia with propofol, fentanyl, rocuronium, and a volatile agent (i.e., sevoflurane or desflurane with air/oxygen), depending on patient preference for regional or general anesthesia. Standard monitoring was used for all patients. Maintenance fluid requirements and third-space losses were replaced with balanced crystalloid solutions or pentastarch (Pentaspan; Bristol-Myers Squibb, Montreal, Canada) or hydroxyethyl starch (Voluven; Fresenius Kabi, Mississauga, Ontario, Canada). The goal of fluid replacement was to maintain an adequate intravascular volume as indicated by a mean arterial pressure =70 mm Hg or within 20% of normal.
Four orthopaedic surgeons (N.N.M., J.R.D., R.G., and K.A.S.) participated in this study. The surgical procedures were performed as per each surgeon's routine practice. After elevation of the limb and exsanguination of the extremity, a pneumatic tourniquet around the upper part of the thigh was inflated to a pressure of 350 mm Hg. A midline skin and medial parapatellar capsular incision was made to expose the knee joint. Standard surgical techniques for intraoperative hemostasis were used. An appropriate type and size of knee prosthesis (Triathlon [Stryker, Mahwah, New Jersey] or Genesis II [Smith & Nephew, Memphis, Tennessee]) was used. After all components were cemented into place, the joint was thoroughly irrigated and the study medication was applied to the open joint surfaces with use of a bulb syringe and was left in contact with the tissues for five minutes. The surgeon subsequently suctioned away excess study solution by placing the suction tip on the cemented component without touching the joint or the surrounding tissue surfaces. The wound was then closed without any irrigation or manipulation. No surgical drains were used. The tourniquet was released after the application of dressings.
Prophylaxis against venous thromboembolism was administered as per standard practice at our institution with low molecular weight heparin for ten days after surgery. While in the hospital, patients were examined daily for any clinical symptoms of deep-vein thrombosis. A diagnostic Doppler ultrasound examination of both legs was performed on Postoperative Day 2 or 3 for all patients, or earlier if a patient was symptomatic. All surgical and medical adverse events and any thromboembolic events occurring during the six weeks after surgery were recorded at the time of the six-week follow-up visit with the surgeon and/or by means of a telephone interview.
Daily postoperative hemoglobin levels were measured for five days. All patients remained in the hospital for a minimum of three days.
Outcome Measures
The primary outcome was blood loss as calculated from the difference between the preoperative hemoglobin and the lowest postoperative hemoglobin during the hospital stay or the lowest postoperative hemoglobin prior to blood transfusion. Based on hemoglobin balance, the estimated blood loss was calculated according to the formula described by Good et al.7 and Nadler et al.23 (see Appendix).
The secondary outcomes were the rate of perioperative blood transfusion, the number of blood units transfused, the rate of surgical infections, the length of hospital stay, the time until the start of a rehabilitation program, the postoperative changes in joint function (i.e., the range of motion and the severity of pain at rest as determined with use of a visual analog scale [possible range, 0 to 10] on Postoperative Day 2) and the Western Ontario and McMaster Universities Osteoarthritis index (WOMAC) score at six weeks after surgery. The active range of motion of the knee with the patient supine was determined with use of a standard 60-cm clinical goniometer before the operation and on Postoperative Day 2. The patient was told to bend the knee as much as possible while lying in a supine position.
All of the end-point measurements were collected by the research assistants, with the exception of range of motion, which required a clinical examination and therefore was measured by the orthopaedic surgeons.
To evaluate systemic absorption of the drug, the plasma tranexamic acid level was measured, one hour after tourniquet release, with use of Tandem Mass Spectrometry at The Hospital for Sick Children (Toronto, Ontario, Canada).
The criterion for the transfusion of blood products was a hemoglobin level of <8.0 g/dL or a hemoglobin level of <10.0 g/dL if the patient developed intolerable symptoms of anemia or any organ dysfunction that may have been related to anemia and was not attributable to another cause (such as myocardial ischemia or hypoxemia) or if ongoing blood loss was occurring. If transfusion was necessary, one unit of packed red blood cells was transfused at a time to increase the hemoglobin level to =8.0 g/dL. The transfusion guidelines were consistent with the guidelines of the American Society of Anesthesiologists24 and the ONTraC study5.
Venous thromboembolism prophylaxis was begun on the morning after surgery with a standard subcutaneous injection of dalteparin (5000 U) once daily. In November 2008, the Pharmacy and Therapeutics Committee at our institution switched from dalteparin to rivaroxaban because of studies indicating superiority of the latter for prophylaxis against thromboembolism25. Rivaroxaban (10 mg, administered orally) was started on the day after surgery and was continued for ten days. However, in March 2009, the Pharmacy and Therapeutics Committee switched to enoxaparin (40 mg daily for ten days) because of concerns about excessive postoperative bleeding in some patients (not study subjects). The time window of the change in prophylaxis against thromboembolism was examined for the primary outcome.
Statistical Methods
An intention-to-treat analysis was used to compare the study groups. The distribution of potential confounders between the study groups (such as demographic data, baseline data, and surgical and functional characteristics) and the primary and secondary outcomes was assessed with use of summary statistics, including means and weighted means (to adjust for stratification) and standard deviations and 95% confidence intervals for quantitative data and frequencies and percentages for qualitative data. Continuous variables were compared with use of one-way analysis of variance, and categorical variables were compared with use of the chi-square test with the Yates correction factor as necessary. For the one-way analysis of variance and 2 × 3 chi-square analysis, a p value of <0.05 was considered significant. For the pairwise comparisons, the p value was adjusted according to the Bonferroni correction.
Sample Size Calculation
The sample-size estimate was based on the difference in the primary outcome (i.e., postoperative decrease in hemoglobin) among the three study groups. A review of eighty patients who had undergone a primary unilateral total knee arthroplasty at our institution from January to April 2006 showed that the mean reduction in postoperative hemoglobin (and standard deviation) was 4.6 ± 1.7 g/dL, which represented a 32.1% ± 10.6% change in comparison with preoperative levels. In fact, the hemoglobin level dropped from 13.9 ± 2.3 to 9.4 ± 1.3 g/dL, consistent with the published literature26. We performed a meta-analysis (RevMan 4.2; Cochrane Collaboration) of two randomized controlled trials that showed that intravenous tranexamic acid (20 mg/kg) produces a saving of 1 g/dL in the decrease of hemoglobin after total knee arthroplasty as compared with placebo (weighted mean difference, —0.98 g/dL; 95% confidence interval, —1.30 to —0.67 g/dL)27,28. Based on this information and our own review, to prevent a fall of 1 g/dL in postoperative hemoglobin, the sample size required for each arm of the study was twenty-six patients. This sample size was calculated for a fixed effects one-way analysis of variance design. It was assumed that the standard effect size (d) = 0.36, the level of alpha (two-tailed) = 0.05, and power = 0.8. In addition, we assumed the standard deviation within each group to be 14 g/dL. The sample size was increased by 20% to compensate for expected dropouts, resulting in thirty-three patients per group (i.e., the total number of patients was ninety-nine).
Source of Funding
The Physicians’ Services Incorporated Foundation did not play a role in the investigation.
In the present study, topical administration of tranexamic acid to the knee joint before wound closure significantly reduced postoperative blood loss in patients having a primary unilateral total knee arthroplasty. The calculated blood loss was about 20% to 25% lower in patients who were managed with tranexamic acid as compared with placebo, with a mean difference of approximately 300 to 400 mL for the 1.5 and 3-g tranexamic acid groups, respectively. However, no difference in efficacy was seen between the low-dose (1.5-g) and the high-dose (3-g) groups. The patients in the low-dose and high-dose tranexamic acid groups had 16% and 17% higher mean postoperative hemoglobin levels, respectively, compared with the placebo group. With the numbers studied, we did not find a reduction in the frequency of allogeneic blood transfusion between the 1.5-g tranexamic acid group and the placebo group. However, none of the patients in the 3-g tranexamic acid group required transfusion.
The potential mechanism and advantage of topical application of tranexamic acid into the surgical field is to directly target the site of bleeding just before wound closure, but after surgical hemostasis has been achieved, thus attenuating the marked increase in local fibrinolysis associated with release of the tourniquet29. Such inhibited local fibrinolytic activity will help to prevent fibrin clot dissolution and increase its volume and strength at the raw surgical surfaces, thus enhancing microvascular hemostasis30.
Our findings of reduced postoperative bleeding are consistent and comparable with the findings of previous meta-analyses of individual studies of intravenous tranexamic acid administration in patients undergoing total knee arthroplasty, which demonstrated a reduction of blood loss of about 400 mL per case15,17,31. When tranexamic acid is given intravenously, it is widely distributed throughout the extracellular and intracellular compartments32. Tranexamic acid rapidly diffuses into the synovial fluid and synovial membranes, reaching the same concentration in the joint fluid as in the serum33. The biological half-life in the joint fluid is about three hours33. Tranexamic acid is eliminated by glomerular filtration, with excretion being about 30% at one hour, 55% at three hours, and 90% at twenty-four hours after an intravenous dose of 10 mg/kg32. The advantage of topical application of tranexamic acid is minimal systemic absorption. In fact, the plasma levels of tranexamic acid detected in our patients were significantly (~70%) less than an equivalent dose of intravenously administered tranexamic acid. The plasma concentration one hour after intravenous injection of 10 mg/kg of tranexamic acid is 18 mg/L32, whereas the low and high topical tranexamic acid doses used in the present study led to mean plasma levels of 4.5 and 8.5 mg/L, respectively. The minimum therapeutic concentration of tranexamic acid is 5 to 10 mg/L34; therefore, it is possible that the therapeutic effect of topical tranexamic acid in our study may be secondary to some, albeit minimal, systemic absorption.
Our findings are consistent with topical application of tranexamic acid in patients undergoing oral19,34 and cardiac surgery35, in whom the levels of tranexamic acid have been found to be minimal in the systemic circulation and not associated with clinically important thromboembolic manifestations. Therefore, the topical administration of tranexamic acid into the surgical wound in patients undergoing total knee arthroplasty can reduce the risks of a general prothrombotic state.
Topical application of tranexamic acid is simple and inexpensive compared with topical fibrin sealants, which also have been shown to reduce bleeding after total knee arthroplasty36-38. Another disadvantage of fibrin sealants is that they are derived from human plasma; therefore, the risk of transmission of infective agents cannot be completely excluded37,39. The reduction in bleeding in our study is comparable with that in previous studies of fibrin tissue adhesive37,38. A randomized study of topical fibrin spray as compared with intravenous tranexamic acid demonstrated a comparable reduction in blood loss; however, the cost of the topical fibrin group compared with the tranexamic acid group was $585.00 as compared with $6.0037.
Our study is one of the few to examine the safety of tranexamic acid in total knee arthroplasty by systemically screening all study patients with diagnostic imaging to detect deep-vein thrombosis. The rate of detection of pulmonary emboli and asymptomatic deep-vein thromboses with use of Doppler ultrasonography was comparable among all three study groups. Orpen et al.40 reported that no deep-vein thromboses were detected with duplex ultrasound scanning, and Tanaka et al.41 also found there was no increase in deep-vein thromboses or pulmonary emboli on the basis of radioisotope venography and lung scanning in patients receiving intravenous tranexamic acid. Recent findings of increased mortality associated with the use of another antifibrinolytic, aprotinin42, and an advisory issued by the United States Food and Drug Administration about erythropoiesis-stimulating agents because of an increased risk of deep-vein thrombosis in patients following orthopaedic surgery43 emphasize that blood-conservation techniques need to be shown to be both safe and effective before their wide adoption into clinical practice.
Topical administration of tranexamic acid provides a novel therapeutic approach for decreasing bloodshed from the surgical wound after total knee arthroplasty that may be safer than other blood-conservation strategies, such as the preoperative administration of erythropoietin, and more cost-effective than preoperative autologous blood donation44-47. Moreover, several studies of patients undergoing total knee arthroplasty have demonstrated that blood transfusion, even after leukocyte filtration, is associated with an increased prevalence of postoperative surgical infections (noted in 10% to 20% of patients5,48-51).
Postoperative anemia can be an important problem that is associated with adverse effects, including increased mortality and morbidity52,53 and a longer hospital stay due to the associated need for blood transfusion5,46,54. Indeed, in one study, hemoglobin levels on admission to rehabilitation facilities after total knee arthroplasty were inversely correlated with the length of the hospital stay, and higher hemoglobin levels correlated with faster functional recovery at inpatient rehabilitation facilities55. These adverse effects are particularly important in elderly total knee arthroplasty patients, who already have a higher prevalence of anemia56 and comorbid medical conditions. The effect of postoperative anemia on this patient population also may be of greater importance because of diminished hematopoietic reserve57. Recovery from profound postoperative anemia is slow; consequently, many of these patients can experience the adverse effects of postoperative anemia for as long as several weeks after surgery58.
In the present study, the topical application of tranexamic acid to the exposed knee joint did not affect postoperative joint or patient function. This is inferred from the lack of a significant difference between the placebo and tranexamic acid groups with regard to postoperative knee flexion, visual analog scale pain scores, length of stay in the hospital, time to the start of rehabilitation, and the amount of improvement in WOMAC scores at six weeks after surgery.
The present study had several limitations. The study was not powered to examine the assumption that tranexamic acid can decrease the frequency of transfusion, although none of the patients in the 3-g tranexamic acid group required a transfusion. Another limitation of the study is that the type of prophylaxis against thromboembolism was changed partway through the study by our institutional Pharmacy and Therapeutics Committee because of the apparent superiority of rivaroxaban25. However, there was no difference in terms of calculated blood loss between the groups when the type of thromboembolism prophylaxis was compared. In addition, the Doppler ultrasound studies were performed prior to discharge (two to three days after surgery), rather than at the peak of clinically evident thrombosis59 (six to seven days after surgery). Additionally, we did not perform routine screening for pulmonary emboli; however, no other thromboembolic complications were reported during the six-week follow-up period. Finally, we did not specifically assess postoperative functional recovery to investigate the relationship between hemoglobin levels and the outcome of rehabilitation.
In conclusion, topical application of tranexamic acid into the surgical wound of patients undergoing total knee arthroplasty reduces postoperative bleeding by 20% to 25% in comparison with the values in untreated controls, resulting in 16% to 17% higher postoperative hemoglobin values. No increase in thromboembolic or other complications was identified in patients managed with tranexamic acid. Larger trials are needed to further assess whether this promising strategy to reduce bleeding and the need for blood transfusion in patients undergoing total knee arthroplasty is safe with regard to thromboembolic complications.
Note: The authors thank Warren Walsh, BSc, ART, Department of Pediatric Laboratory Medicine, The Hospital for Sick Children, Toronto, Ontario, Canada, for developing the assay for tranexamic acid analysis.