The Risk Associated with Aprotinin in Cardiac Surgery
http://www.100md.com
《新英格兰医药杂志》
ABSTRACT
Background The majority of patients undergoing surgical treatment for ST-elevation myocardial infarction receive antifibrinolytic therapy to limit blood loss. This approach appears counterintuitive to the accepted medical treatment of the same condition — namely, fibrinolysis to limit thrombosis. Despite this concern, no independent, large-scale safety assessment has been undertaken.
Methods In this observational study involving 4374 patients undergoing revascularization, we prospectively assessed three agents (aprotinin , aminocaproic acid , and tranexamic acid ) as compared with no agent (1374 patients) with regard to serious outcomes by propensity and multivariable methods. (Although aprotinin is a serine protease inhibitor, here we use the term antifibrinolytic therapy to include all three agents.)
Results In propensity-adjusted, multivariable logistic regression (C-index, 0.72), use of aprotinin was associated with a doubling in the risk of renal failure requiring dialysis among patients undergoing complex coronary-artery surgery (odds ratio, 2.59; 95 percent confidence interval, 1.36 to 4.95) or primary surgery (odds ratio, 2.34; 95 percent confidence interval, 1.27 to 4.31). Similarly, use of aprotinin in the latter group was associated with a 55 percent increase in the risk of myocardial infarction or heart failure (P<0.001) and a 181 percent increase in the risk of stroke or encephalopathy (P=0.001). Neither aminocaproic acid nor tranexamic acid was associated with an increased risk of renal, cardiac, or cerebral events. Adjustment according to propensity score for the use of any one of the three agents as compared with no agent yielded nearly identical findings. All the agents reduced blood loss.
Conclusions The association between aprotinin and serious end-organ damage indicates that continued use is not prudent. In contrast, the less expensive generic medications aminocaproic acid and tranexamic acid are safe alternatives.
The mainstay of medical therapy for patients with an acute coronary syndrome — when accompanied by myocardial infarction with ST-segment elevation — includes fibrinolytic and antiplatelet agents to mitigate thrombosis-related events.1 However, if surgical therapy (coronary-artery surgery) is elected, fibrinolytic agents are not used before, during, or after surgery because of concerns regarding excessive bleeding. In fact, these bleeding-related concerns have given rise to the testing, regulatory approval, and widespread use of two classes of agents, both proven to mitigate bleeding: the lysine analogues (aminocaproic acid and tranexamic acid) and the serine protease inhibitors (aprotinin). Consequently, the majority of patients now routinely receive one or more of these agents during and after invasive cardiovascular procedures, including coronary-artery surgery.2,3,4 Thus, at least for patients with ST-elevation myocardial infarction, the surgical approach (with the use of antifibrinolytic agents) is in stark contrast, and may seem counterintuitive, to the medical approach (with fibrinolytic therapy as a mainstay) for thrombosis-related events. (Although aprotinin is a serine protease inhibitor, here we use the term antifibrinolytic therapy to include all three agents.)
The question of the safety of serine protease inhibitors or lysine analogues for thrombosis-related events — though noted in a handful of early case reports and small, single-center experiences involving graft thrombosis5,6,7 and creatinine elevation5,8,9 — has largely been contested by a number of published secondary analyses that have nearly always concluded that antifibrinolytic therapy, as defined here, is safe.10 Unfortunately, however, this "safety evidence" has three important limitations: no prior investigation was adequately powered to assess relatively infrequent, but clinically serious, safety events10; the comparative safety of the three agents has not been assessed within one study — an important consideration, given the large cost differential among agents (aprotinin being far more costly than either aminocaproic acid or tranexamic acid); and nearly all investigations were sponsor-supported10 and therefore carried unavoidable bias.
Addressing these considerations, however, is not straightforward. After a decade of use, antifibrinolytic practice now is embedded and dictated by guidelines,2,4,11 such that safety assessment in independent, placebo-controlled clinical trials with unselected recruitment becomes difficult, if not impossible. In addition, regulatory approval for use of these agents differs among countries, making a large-scale, multicountry, comparative study challenging. Therefore, to address the safety of antifibrinolytic therapy for thrombosis-related cardiac, cerebral, and renal events, we conducted a non–sponsor-supported, prospective, international, multi-institutional study sufficiently powerful (with >800 patients per group) and comprehensive (with hundreds of covariate measurements per patient) to allow comparative safety assessment among the three agents by exacting propensity and multivariable analyses. We hypothesized that the use of either serine protease inhibitors or lysine analogues in patients with acute coronary syndromes presenting for coronary-artery surgery is unsafe.
Methods
After institutional approval and written informed consent had been obtained, patients scheduled for coronary-artery bypass surgery with cardiopulmonary bypass were prospectively enrolled according to a systematic sampling scheme at 69 institutions in North and South America, Europe, the Middle East, and Asia. To be eligible for entry into the study, patients had to be at least 18 years of age, could not be enrolled in another study or trial, and had to be able to engage in a preoperative interview. At each institution, every Rth patient meeting these entry criteria was enrolled, where R=N÷50 (to the closest integer) and where N is the number of patients expected to undergo myocardial revascularization surgery over a one-year period.12 Data were collected throughout hospitalization, with approximately 7500 data fields per patient collected by independent investigators.
Measurement of Outcomes
Each outcome event was prespecified, defined by the protocol, and classified as cardiovascular (myocardial infarction or heart failure), cerebrovascular (stroke, encephalopathy, or coma), or renal (dysfunction or failure). Myocardial infarction required either new Q waves (Minnesota code 1-1-1 or 1-2-7) or new, persistent ST-segment or T-wave changes (Minnesota code 4-1, 4-2, 5-1, 5-2, or 9-2). Heart failure required a cardiac output of less than 2.0 liters per minute associated with a pulmonary-artery occlusion pressure above 18 mm Hg, a central venous pressure above 12 mm Hg, an S3 gallop, or rales. Cerebrovascular events included clinically diagnosed stroke, encephalopathy, and coma. Renal dysfunction required a postoperative serum creatinine level of at least 177 μmol per liter with an increase over preoperative baseline levels of at least 62 μmol per liter; renal failure was defined as dysfunction requiring dialysis or in-hospital death with evidence at autopsy of acute renal failure. Blood loss was assessed as chest-tube output during the first 24 hours after surgery.
Clinical Care and Use of Antifibrinolytic Agents
Clinical decisions were not controlled by the study protocol, and all patients qualifying for enrollment within the prespecified enrollment period were entered into the study (Figure 1). Patients were classified as undergoing primary surgery if the index surgery was elective and involved only coronary-artery revascularization (with no history of cardiac or vascular surgery) or angioplasty. Otherwise, patients were classified as undergoing complex surgery.
Figure 1. Consolidated Standards of Reporting Trials (CONSORT) Diagram of Patient Enrollment.
The diagram shows the numbers of patients who met the criteria for inclusion in, or exclusion from, the study and their distribution among the four study groups and two surgery types. Inclusion in the aprotinin group required administration of a total of more than 2 million kallikrein-inhibitor units intravenously before the end of surgery; inclusion in the aminocaproic acid group, administration of more than 10 g; and inclusion in the tranexamic acid group, administration of more than 1 g.
Statistical Analysis
Baseline medical characteristics were compared statistically (Table 1). The effect of the drugs on outcome was assessed with the use of multivariable logistic regression and propensity-score adjustment. Initially, 97 perioperative risk factors were evaluated for univariate association with outcome (two-tailed P0.20) and then entered stepwise (backward and forward) into multivariable logistic models, with assessment of the association between treatment (aprotinin, aminocaproic acid, or tranexamic acid vs. no treatment) and outcome in the presence of the significant covariates. Comparison between drugs was assessed with the use of contrast functions.
Table 1. Baseline Characteristics of the Patients, According to Study Group.
Selection bias not controlled by multivariable methods was assessed with a propensity-adjustment method. Using nonparsimonious logistic-regression modeling, we developed propensity scores for the use of any antifibrinolytic treatment (vs. no treatment), including 45 treatment-selection covariates, and propensity scores for specific treatments. Covariate interactions proved unnecessary for the balance of baseline characteristics. The discriminate power of the propensity scores was quantified by measurement of the receiver-operating-characteristic area (the C-index). Covariate adjustment was performed with the derived propensity scores and drug-indicator variables. The interaction of the differential drug effect and surgery status (with the propensity score as the adjustment variable13,14) was not significant. Propensity-score analyses according to specific treatment confirmed our findings according to drug class. Finally, the dose response (weight-adjusted) was assessed among 596 patients in the aprotinin group who were receiving either a low-dose regimen (loading dose, 1 million kallikrein-inhibitor units ; total dose, >2 million KIU) or a high-dose regimen (loading dose, 2 million KIU; total dose, >4 million KIU).
SAS statistical software (version 8.2) was used; a P value of less than 0.05 (two-tailed) was considered to indicate statistical significance. Multiple comparison adjustments were assessed.
Results
As expected, patients had evidence of acute and chronic vascular disease. Several imbalances were noted between the treatment groups and the control group before propensity adjustment, but not thereafter (Table 1).
Adverse Safety Outcomes
Overall, the use of aprotinin was associated with an increased risk of renal and nonrenal events when compared with aminocaproic acid, tranexamic acid, or no antifibrinolytic therapy (Figure 2) — a finding confirmed by multivariable logistic regression (Table 2). Interaction by drug group and complexity of surgery was not significant.
Figure 2. Renal and Nonrenal Outcome Events among the 4374 Study Patients.
Panel A shows the incidence of renal events according to study group. All pairwise comparisons between the aprotinin group and any other group were significant (P<0.001 by Bonferroni-adjusted analysis). Comparison of either aminocaproic acid or tranexamic acid with control, or comparison between them, was not significant. Panel B shows the incidence of nonrenal events. In both panels, the numbers above the bars are rounded incidence values, and P values shown are for the comparison between the aprotinin group and any other group (the control group, the aminocaproic acid group, or the tranexamic acid group).
Table 2. Results of Multivariable Logistic Regression for the Renal Composite Outcome in 4374 Patients.
Among the 3013 patients undergoing primary surgery, aprotinin, but not aminocaproic acid or tranexamic acid, was associated with an increased risk of death (2.8 percent vs. 1.3 percent, P=0.02), cardiovascular events (20.4 percent vs. 13.2 percent, P<0.001), cerebrovascular events (4.5 percent vs. 1.6 percent, P<0.001), and renal events (5.5 percent vs. 1.8 percent, P<0.001). Specifically, with regard to cardiovascular events, aprotinin was associated with a 48 percent increase in the risk of myocardial infarction (P<0.001) and a 109 percent increase in the risk of heart failure (P<0.001). After propensity adjustment (C-index, 0.71), multivariable analysis continued to demonstrate a significant association between the use of aprotinin and an increased risk of adverse events (Table 3) as well as an absence of association between either aminocaproic acid or tranexamic acid and such events. Propensity adjustment by drug (C-index for aprotinin, 0.72; for aminocaproic acid, 0.80; and for tranexamic acid, 0.68) yielded similar findings. For example, as compared with control, aprotinin nearly doubled the odds of a renal event (odds ratio, 1.89; 95 percent confidence interval, 1.01 to 3.55; P=0.04), whereas neither aminocaproic acid (odds ratio, 0.85; 95 percent confidence interval, 0.37 to 1.95; P=0.69) nor tranexamic acid (odds ratio, 1.43; 95 percent confidence interval, 0.62 to 3.27; P=0.40) was associated with increased renal risk.
Table 3. Propensity-Adjusted Effect of Treatment on Ischemic Outcome Events.
Among the 1361 patients undergoing complex surgery, aprotinin was associated with increased renal dysfunction and renal failure requiring dialysis, whereas aminocaproic acid and tranexamic acid were not. Propensity adjustment (C-index, 0.73) confirmed these findings (Table 3), and adjustment by specific drug (C-index for aprotinin, 0.78; for aminocaproic acid, 0.78; and for tranexamic acid, 0.76) yielded similar results, as illustrated for renal events: odds ratio with aprotinin, 2.79 (95 percent confidence interval, 1.44 to 5.44; P=0.002); with aminocaproic acid, 0.48 (95 percent confidence interval, 0.17 to 1.34; P=0.16); and with tranexamic acid, 1.01 (95 percent confidence interval, 0.44 to 2.33; P=0.98). A dose–response relationship was found for aprotinin with respect to renal, cardiovascular, and composite outcomes (Figure 3).
Figure 3. Aprotinin Dose Response.
P values shown are for the comparison between a high dose of aprotinin and either a low dose of aprotinin or no antifibrinolytic therapy (control). Pairwise comparison between a high dose of aprotinin and control was significant with respect to renal events, cardiovascular events, and composite outcome events (P<0.001 for all three comparisons by Bonferroni-adjusted analyses) but not for death (P=0.12). Pairwise comparison between a low dose of aprotinin and no antifibrinolytic therapy was significant (in Bonferroni-adjusted analyses) with respect to renal events (P<0.001) and composite outcome events (P=0.003) but not cardiovascular events (P=0.08) or death (P=0.14). Pairwise comparison between a high dose of aprotinin and a low dose of aprotinin was significant (in Bonferroni-adjusted analyses) with respect to renal events (P<0.001) but not cardiovascular events (P=0.04), composite outcome events (P=0.03), or death (P=0.38).
Efficacy
The three medications reduced blood loss to similar extents. As compared with the control group, in which mean (±SD) estimated blood loss was 827±573 ml, blood loss was 753±660 ml in the aprotinin group (P<0.001), 719±578 ml in the aminocaproic acid group (P<0.001), and 676±741 ml in the tranexamic acid group (P<0.001).
Discussion
Our findings raise serious concerns regarding the safety of an approved drug intended to limit blood loss in at-risk patients undergoing surgery. Specifically, the use of aprotinin was associated with a dose-dependent doubling to tripling in the risk of renal failure requiring dialysis among patients undergoing primary or complex coronary-artery surgery. Furthermore, for the majority of patients undergoing primary surgery, we found evidence of multiorgan damage involving the heart (myocardial infarction or heart failure) and the brain (encephalopathy) in addition to the kidneys, suggesting a generalized pattern of ischemic injury. Unlike the serine protease inhibitors, analysis of the less costly lysine analogues aminocaproic acid and tranexamic acid demonstrated no such safety concerns, although these two agents were equally effective in reducing blood loss. Thus, our findings indicate that reconsideration of the safety of aprotinin among patients undergoing cardiac surgery is warranted and indicate replacement of aprotinin with either aminocaproic acid or tranexamic acid.
Blood loss and its replacement merit careful redress — particularly in cardiac surgery, for which large amounts of blood loss occur not only because of direct vascular interruption, but also because of exposure and autotransplantation of blood elements after contact with the foreign surfaces of the bypass circuitry.2,12 Consequently, prostacyclins, desmopressin, lysine analogues, and serine protease inhibitors were developed. The latter two classes are now used in the majority of the 1 million cardiac surgical operations performed annually worldwide2,3,4,10,12,15 — a practice consistent with consensus guidelines.2,4,11
Unlike lysine analogues, aprotinin has high affinity for the kidneys16,17,18,19 — a property that may explain our renal findings. After free glomerular passage, aprotinin binds selectively to the brush border of the proximal tubule membrane, and then, by pinocytosis, it enters into and accumulates within the cytoplasm, inhibiting tubule protease secretion (kallikrein and, secondarily, kinin), prostaglandin and renin synthesis, prostasin secretion, and bradykinin release.18,19,20,21,22 Under normothermic ischemia, hypothermia, and other states of high kallikrein activity, these untoward tubular effects are magnified and are complicated by dose-dependent renal afferent vasoconstriction; deep cortical and medullary perfusion and its autoregulation are thereby impaired, and focal tubular necrosis ensues.20,21,22,23,24,25,26 Furthermore, because of its interference with the synthesis and release of endothelial nitric acid, aprotinin also may instigate macrovascular or microvascular thrombosis.27,28
Despite this in vitro and in vivo evidence, only a minority of the reports of 45 trials of aprotinin in surgical patients even comment on renal function, and of those trials, none were powered to discern renal failure, leaving earlier concerns unchecked.10 However, review of this evidence suggests the presence of several renal "safety signals," including aprotinin-associated alpha1-microglobulin production,8 tubule-cell deposition of protein bands and proteinuria,8 dose-dependent increases in creatinine,5,9 renal dysfunction,9 and platelet-fibrin thrombotic occlusions of the renal arterioles after death.7 Of note, even in a small study (involving 57 patients) that concluded that aprotinin was safe (albeit with concern regarding a possible type 1 statistical error), 90 percent suppression of urinary kallikrein excretion occurred, with arithmetic increases in sodium excretion and osmolar clearance.29
Our data from approximately 1300 aprotinin-treated patients as compared with about 1300 control patients, then, clearly give credence to early concerns stemming from findings in animal models20,21,22,23,24,26 and preliminary findings in patients5,7,8,9 — namely, that aprotinin is associated with severe renal adverse events and that this association is dose-dependent. The lysine analogues, in contrast, are excreted nearly intact within 24 hours after intravenous administration, with their renal clearance approximating creatinine clearance. Moreover, few reports document an association between these agents and renal dysfunction3 — a renal-safety profile validated by our results.
Our findings raise concerns regarding the proclivity of aprotinin, but not the lysine analogues, to instigate cardiovascular and cerebrovascular thrombosis. Although questions regarding each of the agents have been raised, as indicated by early case reports showing a propensity for thrombosis, most compelling is the evidence relating to aprotinin, which distinguishes itself by at least five properties: inhibition of soluble proteases (e.g., kallikreins, plasmin, and trypsin); inhibition of activated protein C; preservation of platelet adhesive and aggregatory properties; impairment of vascular endothelial-cell function (in the coronary and cerebral arteries and aorta)20,21,23,27,28,30,31; and selective impairment of endothelium-derived relaxation by dose-dependent inhibition of nitric oxide synthesis and release.27,32
A possible link between aprotinin and intravascular thrombosis has been observed in several in vivo animal models27,28 as well as in humans — in association with biogenic materials (catheters, cannulas, and oxygenators); within coronary grafts; within the native coronary microcirculation; in the aorta; and disseminated throughout the microvasculature of the heart, lung, brain, and kidneys.5,6,7,33,34,35,36 Randomized clinical trials yielded mixed findings.5,7,37,38,39 Noteworthy, however, are the findings of a larger, more recent, sponsor-supported trial,33 which demonstrated that aprotinin-treated patients (vs. those given placebo) had a significantly greater risk of saphenous-vein occlusion, but even then the results were interpreted as inconclusive. For both aminocaproic acid and tranexamic acid — although reports of related intravascular thrombosis exist — no study has reported a significant association.3
Our findings in patients undergoing primary surgery — namely, that aprotinin-treated patients are at greater risk for ischemic damage to the heart than are either control patients or those receiving aminocaproic acid or tranexamic acid — are thus in agreement with data from the prior in vivo and in vitro studies27,28,30,34 and the majority of the coronary-graft investigations.5,7,33,37 Similarly, with regard to encephalopathy, our findings are consistent with those of earlier studies of microvascular thrombosis28,34 and specifically with those described by Sundt et al.,7 who reported platelet-fibrin thrombi among multiple vessels, including the cerebral arteries, on postmortem examination of patients who had received aprotinin. In contrast, among patients undergoing complex procedures, aside from renal outcomes, we found no other drug associations, probably because the proven blood-sparing salutary effects of antifibrinolytic therapy in patients undergoing complex surgery10 may offset any thrombotic effects. This hypothesis is supported by our secondary finding that among patients with hemorrhage, aprotinin was associated with an increase in the risk of cardiovascular events (34 percent, vs. 19 percent in the control group; P=0.04), whereas no such difference existed among patients without hemorrhage (23 percent and 24 percent, respectively; P=0.79).
Given our findings, especially with regard to serious renal events among patients undergoing either primary or complex surgery, and given the cost of aprotinin therapy, which is at least 10 times that of aminocaproic or tranexamic acid, we estimate that considerable global health care savings would accrue if aprotinin were replaced by either aminocaproic acid or tranexamic acid. Specifically, extrapolating international-use patterns, we estimate that for renal complications alone, the replacement of aprotinin with aminocaproic acid would prevent renal failure requiring dialysis in 11,050 patients per year, yielding an indirect savings (from the saved cost of dialysis) of more than $1 billion per year, in addition to direct savings (from reduced drug costs) of nearly $250 million per year. Replacement of aprotinin with tranexamic acid would prevent 9790 complications necessitating dialysis each year, yielding similar direct and indirect savings.
Regarding potential study limitations, we should note that large-scale, randomized, controlled trials — though ideal for assessing post-marketing drug safety — are difficult (if not impossible) to conduct in the setting of embedded practice, for several reasons: there is inherent bias in the selection of patients to be subjected to "nonroutine" treatments; it is necessary to withhold salutary blood-sparing therapies from those assigned to placebo; there may be reluctance to include the sickest patients, who are those most likely to have adverse events; the required sample size and cost (to detect less frequent safety events) are substantial; and there may be reluctance (a disincentive) on the part of sponsors to discern carefully the risk of a marketed drug. Given that, assessment of safety in observational studies, when sufficiently comprehensive and large, may offer critical insights, even in light of recognized limitations.
Therefore, we assessed safety in a comprehensive, large-scale, observational study based on randomized patient selection, inclusion of more than 800 high-risk patients per group, and measurement of more than 200 covariates (by drug and by outcome) per patient — an approach that permitted nonparsimonious propensity analyses as well as multivariable corrections for the multiple covariates of organ-specific outcomes.40 As such, we believe that our findings, particularly with respect to renal failure, are substantive with respect to effect size, consistency between the primary-surgery and complex-surgery groups, and dose response and that they are notable for their consistency with early in vivo and in vitro animal studies and several suggestive case reports. In addition, our specific analyses of aprotinin as compared with aminocaproic acid and tranexamic acid allowed us to compare patients to whom an antifibrinolytic agent had been administered, thereby mitigating selection bias.
In conclusion, the observed association between aprotinin and serious end-organ damage indicates that continued use is not prudent, whereas the less expensive generic medications aminocaproic acid and tranexamic acid are safe alternatives.
The Ischemia Research and Education Foundation (IREF) is an independent, nonprofit foundation, formed in 1987, that mentors clinical investigators through observational studies and clinical trials addressing ischemic injury of the heart, brain, kidney, and gastrointestinal tract. The Multicenter Study of Perioperative Ischemia (McSPI) Research Group, formed in 1988, is an association of 160 international medical centers located in 23 countries, organized through, and supported by grants from, IREF.
Supported by IREF data collection, which provided all the funding for the study, including site grants, central analysis and data disposition, manuscript grants, and publication of the findings.
No potential conflict of interest relevant to this article was reported.
* Investigators and centers participating in the study are listed in the Appendix.
Source Information
From the Ischemia Research and Education Foundation, San Bruno, Calif.
Address reprint requests to Dr. Mangano at the Editorial Office, Ischemia Research and Education Foundation, 1111 Bayhill Dr., Ste. 480, San Bruno, CA 94066, or at dtb@iref.org.
References
Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1999 Guidelines for the management of patients with acute myocardial infarction). J Am Coll Cardiol 2004;44:E1-E21.
Stover EP, Siegel LC, Body SC, et al. Institutional variability in red blood cell conservation practices for coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth 2000;14:171-176.
Fremes SE, Wong BI, Lee E, et al. Meta-analysis of prophylactic drug treatment in the prevention of postoperative bleeding. Ann Thorac Surg 1994;58:1580-1588.
Faust RJ, Harrison CR, Stehling LC, eds. Questions and answers about transfusion practices. 3rd ed. Park Ridge, Ill.: 1997 Committee on Transfusion Medicine of the American Society of Anesthesiologists, 2002:16-9, 21-2, 4-30.
Cosgrove DM III, Heric B, Lytle BW, et al. Aprotinin therapy for reoperative myocardial revascularization: a placebo-controlled study. Ann Thorac Surg 1992;54:1031-1036.
Umbrain V, Christiaens F, Camu F. Intraoperative coronary thrombosis: can aprotinin and protamine be incriminated? J Cardiothorac Vasc Anesth 1994;8:198-201.
Sundt TM III, Kouchoukos NT, Saffitz JE, Murphy SF, Wareing TH, Stahl DJ. Renal dysfunction and intravascular coagulation with aprotinin and hypothermic circulatory arrest. Ann Thorac Surg 1993;55:1418-1424.
Feindt PR, Walcher S, Volkmer I, et al. Effects of high dose aprotinin on renal function in aortocoronary bypass grafting. Ann Thorac Surg 1995;60:1076-1080.
D'Ambra MN, Akins CW, Blackstone EH, et al. Aprotinin in primary valve replacement and reconstruction: a multicenter double-blind, placebo-controlled trial. J Thorac Cardiovasc Surg 1996;112:1081-1089.
Sedrakyan A, Treasure T, Elefteriades JA. Effect of aprotinin on clinical outcomes in coronary artery bypass graft surgery: a systematic review and meta-analysis of randomized clinical trials. J Thorac Cardiovasc Surg 2004;128:442-448.
Eagle KA, Guyton RA, Davidoff R, et al. ACC/AHA 2004 guideline update for coronary artery bypass graft surgery: a report of the American College of Cardiology American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for Coronary Artery Bypass Graft Surgery). Circulation 2004;110:e340-e437.
Mangano DT. Aspirin and mortality from coronary bypass surgery. N Engl J Med 2002;347:1309-1317.
Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55.
D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265-2281.
Levy JH, Pifarre R, Schaff HV, et al. A multicenter, double-blind, placebo-controlled trial of aprotinin for reducing blood loss and the requirement for donor-blood transfusion in patients undergoing repeat coronary artery bypass grafting. Circulation 1995;92:2236-2244.
Beller FK, Mitchell PS, Gorstein F. Fibrin deposition in the rabbit kidney produced by protease inhibitors. Thromb Diath Haemorrh 1967;17:427-439.
Davis H, Gascho C, Kiernan JA. Effects of aprotinin on organ cultures of the rat's kidney. In Vitro 1976;12:192-197.
Kramer HJ, Moch T, von Sicherer L, Dusing R. Effects of aprotinin on renal function and urinary prostaglandin excretion in conscious rats after acute salt loading. Clin Sci (Lond) 1979;56:547-553.
Vio CP, Oestreicher E, Olivarria V, Velarde V, Mayfield RK, Jaffa AA. Cellular distribution of exogenous aprotinin in the rat kidney. Biol Chem 1998;379:1271-1277.
Seto S, Kher V, Scicli AG, Beierwaltes WH, Carretero OA. The effect of aprotinin (a serine protease inhibitor) on renal function and renin release. Hypertension 1983;5:893-899.
Kramer HJ, Düsing R, Gl?nzer K, Kipnowski J, Klingmüller D, Meyer-Lehnert H. Effects of aprotinin on renal function. Contrib Nephrol 1984;42:233-241.
Maier M, Starlinger M, Zhegu Z, Rana H, Binder BR. Effect of the protease inhibitor aprotinin on renal hemodynamics in the pig. Hypertension 1985;7:32-38.
Nakagawa M, Stewart JM, Vavrek RJ, Nasjletti A. Effects of a kinin antagonist on renal function in rats. Am J Physiol 1990;258:F643-F648.
Pfluger H, Maier M, Binder BR. Effects of antiproteases (aprotinin and gabexate-mesilate) on the postischemic vascular response of the kidney. Urol Int 1983;38:150-157.
Blauhut B, Gross C, Necek S, Doran JE, Sp?th P, Lundsgaard-Hansen P. Effects of high-dose aprotinin on blood loss, platelet function, fibrinolysis, complement, and renal function after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1991;101:958-967.
Fischer JH, Knupfer P. High-dosage aprotinin (Trasylol) therapy -- is it safe for the kidney? Langenbecks Arch Chir 1983;360:241-249.
ülker S, McKeown PP, Bayraktutan U. Aprotinin impairs coronary endothelial function and down-regulates endothelial NOS in rat coronary microvascular endothelial cells. Cardiovasc Res 2002;55:830-837.
Samama CM, Mazoyer E, Bruneval P, et al. Aprotinin could promote arterial thrombosis in pigs: a prospective randomized, blind study. Thromb Haemost 1994;71:663-669.
Schweizer A, Hohn L, Morel DR, Kalangos A, Licker M. Aprotinin does not impair renal haemodynamics and function after cardiac surgery. Br J Anaesth 2000;84:16-22.
Havel MP, Griesmacher A, Weigel G, et al. Aprotinin decreases release of 6-keto-prostaglandin F1 alpha and increases release of thromboxane A2 in cultured human umbilical vein endothelial cells. J Thorac Cardiovasc Surg 1992;104:654-658.
van Oeveren W, Harder MP, Roozendaal KJ, Eijsman L, Wildevuur CR. Aprotinin protects platelets against the initial effect of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1990;99:788-796.
Hill GE, Taylor JA, Robbins RA. Differing effects of aprotinin and episilon-aminocaproic acid on cytokine-induced inducible nitric oxide synthase expression. Ann Thorac Surg 1997;63:74-77.
Alderman EL, Levy JH, Rich JB, et al. Analyses of coronary graft patency after aprotinin use: results from the International Multicenter Aprotinin Graft Patency Experience (IMAGE) trial. J Thorac Cardiovasc Surg 1998;116:716-730.
Saffitz JE, Stahl DJ, Sundt TM, Wareing TH, Kouchoukos NT. Disseminated intravascular coagulation after administration of aprotinin in combination with deep hypothermic circulatory arrest. Am J Cardiol 1993;72:1080-1082.
Alvarez JM, Chandraratna H, Newman MA, Levy JH. Case 3-1999: intraoperative coronary thrombosis in association with low-dose aprotinin therapy. J Cardiothorac Vasc Anesth 1999;13:623-628.
Penkoske PA, Entwistle LM, Marchak BE, Seal RF, Gibb W. Aprotinin in children undergoing repair of congenital heart defects. Ann Thorac Surg 1995;60:Suppl 6:S529-S532.
Lemmer JH Jr, Stanford W, Bonney SL, et al. Aprotinin for coronary bypass operations: efficacy, safety, and influence on early saphenous vein graft patency: a multicenter, randomized, double-blind, placebo-controlled study. J Thorac Cardiovasc Surg 1994;107:543-553.
Bidstrup BP, Underwood SR, Sapsford RN, Streets EM. Effect of aprotinin (Trasylol) on aorta-coronary bypass graft patency. J Thorac Cardiovasc Surg 1993;105:147-153.
Havel M, Grabenwoger F, Schneider J, et al. Aprotinin does not decrease early graft patency after coronary artery bypass graft despite reducing postoperative bleeding and use of donated blood. J Thorac Cardiovasc Surg 1994;107:807-810.
Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342:1887-1892.(Dennis T. Mangano, Ph.D.,)
Background The majority of patients undergoing surgical treatment for ST-elevation myocardial infarction receive antifibrinolytic therapy to limit blood loss. This approach appears counterintuitive to the accepted medical treatment of the same condition — namely, fibrinolysis to limit thrombosis. Despite this concern, no independent, large-scale safety assessment has been undertaken.
Methods In this observational study involving 4374 patients undergoing revascularization, we prospectively assessed three agents (aprotinin , aminocaproic acid , and tranexamic acid ) as compared with no agent (1374 patients) with regard to serious outcomes by propensity and multivariable methods. (Although aprotinin is a serine protease inhibitor, here we use the term antifibrinolytic therapy to include all three agents.)
Results In propensity-adjusted, multivariable logistic regression (C-index, 0.72), use of aprotinin was associated with a doubling in the risk of renal failure requiring dialysis among patients undergoing complex coronary-artery surgery (odds ratio, 2.59; 95 percent confidence interval, 1.36 to 4.95) or primary surgery (odds ratio, 2.34; 95 percent confidence interval, 1.27 to 4.31). Similarly, use of aprotinin in the latter group was associated with a 55 percent increase in the risk of myocardial infarction or heart failure (P<0.001) and a 181 percent increase in the risk of stroke or encephalopathy (P=0.001). Neither aminocaproic acid nor tranexamic acid was associated with an increased risk of renal, cardiac, or cerebral events. Adjustment according to propensity score for the use of any one of the three agents as compared with no agent yielded nearly identical findings. All the agents reduced blood loss.
Conclusions The association between aprotinin and serious end-organ damage indicates that continued use is not prudent. In contrast, the less expensive generic medications aminocaproic acid and tranexamic acid are safe alternatives.
The mainstay of medical therapy for patients with an acute coronary syndrome — when accompanied by myocardial infarction with ST-segment elevation — includes fibrinolytic and antiplatelet agents to mitigate thrombosis-related events.1 However, if surgical therapy (coronary-artery surgery) is elected, fibrinolytic agents are not used before, during, or after surgery because of concerns regarding excessive bleeding. In fact, these bleeding-related concerns have given rise to the testing, regulatory approval, and widespread use of two classes of agents, both proven to mitigate bleeding: the lysine analogues (aminocaproic acid and tranexamic acid) and the serine protease inhibitors (aprotinin). Consequently, the majority of patients now routinely receive one or more of these agents during and after invasive cardiovascular procedures, including coronary-artery surgery.2,3,4 Thus, at least for patients with ST-elevation myocardial infarction, the surgical approach (with the use of antifibrinolytic agents) is in stark contrast, and may seem counterintuitive, to the medical approach (with fibrinolytic therapy as a mainstay) for thrombosis-related events. (Although aprotinin is a serine protease inhibitor, here we use the term antifibrinolytic therapy to include all three agents.)
The question of the safety of serine protease inhibitors or lysine analogues for thrombosis-related events — though noted in a handful of early case reports and small, single-center experiences involving graft thrombosis5,6,7 and creatinine elevation5,8,9 — has largely been contested by a number of published secondary analyses that have nearly always concluded that antifibrinolytic therapy, as defined here, is safe.10 Unfortunately, however, this "safety evidence" has three important limitations: no prior investigation was adequately powered to assess relatively infrequent, but clinically serious, safety events10; the comparative safety of the three agents has not been assessed within one study — an important consideration, given the large cost differential among agents (aprotinin being far more costly than either aminocaproic acid or tranexamic acid); and nearly all investigations were sponsor-supported10 and therefore carried unavoidable bias.
Addressing these considerations, however, is not straightforward. After a decade of use, antifibrinolytic practice now is embedded and dictated by guidelines,2,4,11 such that safety assessment in independent, placebo-controlled clinical trials with unselected recruitment becomes difficult, if not impossible. In addition, regulatory approval for use of these agents differs among countries, making a large-scale, multicountry, comparative study challenging. Therefore, to address the safety of antifibrinolytic therapy for thrombosis-related cardiac, cerebral, and renal events, we conducted a non–sponsor-supported, prospective, international, multi-institutional study sufficiently powerful (with >800 patients per group) and comprehensive (with hundreds of covariate measurements per patient) to allow comparative safety assessment among the three agents by exacting propensity and multivariable analyses. We hypothesized that the use of either serine protease inhibitors or lysine analogues in patients with acute coronary syndromes presenting for coronary-artery surgery is unsafe.
Methods
After institutional approval and written informed consent had been obtained, patients scheduled for coronary-artery bypass surgery with cardiopulmonary bypass were prospectively enrolled according to a systematic sampling scheme at 69 institutions in North and South America, Europe, the Middle East, and Asia. To be eligible for entry into the study, patients had to be at least 18 years of age, could not be enrolled in another study or trial, and had to be able to engage in a preoperative interview. At each institution, every Rth patient meeting these entry criteria was enrolled, where R=N÷50 (to the closest integer) and where N is the number of patients expected to undergo myocardial revascularization surgery over a one-year period.12 Data were collected throughout hospitalization, with approximately 7500 data fields per patient collected by independent investigators.
Measurement of Outcomes
Each outcome event was prespecified, defined by the protocol, and classified as cardiovascular (myocardial infarction or heart failure), cerebrovascular (stroke, encephalopathy, or coma), or renal (dysfunction or failure). Myocardial infarction required either new Q waves (Minnesota code 1-1-1 or 1-2-7) or new, persistent ST-segment or T-wave changes (Minnesota code 4-1, 4-2, 5-1, 5-2, or 9-2). Heart failure required a cardiac output of less than 2.0 liters per minute associated with a pulmonary-artery occlusion pressure above 18 mm Hg, a central venous pressure above 12 mm Hg, an S3 gallop, or rales. Cerebrovascular events included clinically diagnosed stroke, encephalopathy, and coma. Renal dysfunction required a postoperative serum creatinine level of at least 177 μmol per liter with an increase over preoperative baseline levels of at least 62 μmol per liter; renal failure was defined as dysfunction requiring dialysis or in-hospital death with evidence at autopsy of acute renal failure. Blood loss was assessed as chest-tube output during the first 24 hours after surgery.
Clinical Care and Use of Antifibrinolytic Agents
Clinical decisions were not controlled by the study protocol, and all patients qualifying for enrollment within the prespecified enrollment period were entered into the study (Figure 1). Patients were classified as undergoing primary surgery if the index surgery was elective and involved only coronary-artery revascularization (with no history of cardiac or vascular surgery) or angioplasty. Otherwise, patients were classified as undergoing complex surgery.
Figure 1. Consolidated Standards of Reporting Trials (CONSORT) Diagram of Patient Enrollment.
The diagram shows the numbers of patients who met the criteria for inclusion in, or exclusion from, the study and their distribution among the four study groups and two surgery types. Inclusion in the aprotinin group required administration of a total of more than 2 million kallikrein-inhibitor units intravenously before the end of surgery; inclusion in the aminocaproic acid group, administration of more than 10 g; and inclusion in the tranexamic acid group, administration of more than 1 g.
Statistical Analysis
Baseline medical characteristics were compared statistically (Table 1). The effect of the drugs on outcome was assessed with the use of multivariable logistic regression and propensity-score adjustment. Initially, 97 perioperative risk factors were evaluated for univariate association with outcome (two-tailed P0.20) and then entered stepwise (backward and forward) into multivariable logistic models, with assessment of the association between treatment (aprotinin, aminocaproic acid, or tranexamic acid vs. no treatment) and outcome in the presence of the significant covariates. Comparison between drugs was assessed with the use of contrast functions.
Table 1. Baseline Characteristics of the Patients, According to Study Group.
Selection bias not controlled by multivariable methods was assessed with a propensity-adjustment method. Using nonparsimonious logistic-regression modeling, we developed propensity scores for the use of any antifibrinolytic treatment (vs. no treatment), including 45 treatment-selection covariates, and propensity scores for specific treatments. Covariate interactions proved unnecessary for the balance of baseline characteristics. The discriminate power of the propensity scores was quantified by measurement of the receiver-operating-characteristic area (the C-index). Covariate adjustment was performed with the derived propensity scores and drug-indicator variables. The interaction of the differential drug effect and surgery status (with the propensity score as the adjustment variable13,14) was not significant. Propensity-score analyses according to specific treatment confirmed our findings according to drug class. Finally, the dose response (weight-adjusted) was assessed among 596 patients in the aprotinin group who were receiving either a low-dose regimen (loading dose, 1 million kallikrein-inhibitor units ; total dose, >2 million KIU) or a high-dose regimen (loading dose, 2 million KIU; total dose, >4 million KIU).
SAS statistical software (version 8.2) was used; a P value of less than 0.05 (two-tailed) was considered to indicate statistical significance. Multiple comparison adjustments were assessed.
Results
As expected, patients had evidence of acute and chronic vascular disease. Several imbalances were noted between the treatment groups and the control group before propensity adjustment, but not thereafter (Table 1).
Adverse Safety Outcomes
Overall, the use of aprotinin was associated with an increased risk of renal and nonrenal events when compared with aminocaproic acid, tranexamic acid, or no antifibrinolytic therapy (Figure 2) — a finding confirmed by multivariable logistic regression (Table 2). Interaction by drug group and complexity of surgery was not significant.
Figure 2. Renal and Nonrenal Outcome Events among the 4374 Study Patients.
Panel A shows the incidence of renal events according to study group. All pairwise comparisons between the aprotinin group and any other group were significant (P<0.001 by Bonferroni-adjusted analysis). Comparison of either aminocaproic acid or tranexamic acid with control, or comparison between them, was not significant. Panel B shows the incidence of nonrenal events. In both panels, the numbers above the bars are rounded incidence values, and P values shown are for the comparison between the aprotinin group and any other group (the control group, the aminocaproic acid group, or the tranexamic acid group).
Table 2. Results of Multivariable Logistic Regression for the Renal Composite Outcome in 4374 Patients.
Among the 3013 patients undergoing primary surgery, aprotinin, but not aminocaproic acid or tranexamic acid, was associated with an increased risk of death (2.8 percent vs. 1.3 percent, P=0.02), cardiovascular events (20.4 percent vs. 13.2 percent, P<0.001), cerebrovascular events (4.5 percent vs. 1.6 percent, P<0.001), and renal events (5.5 percent vs. 1.8 percent, P<0.001). Specifically, with regard to cardiovascular events, aprotinin was associated with a 48 percent increase in the risk of myocardial infarction (P<0.001) and a 109 percent increase in the risk of heart failure (P<0.001). After propensity adjustment (C-index, 0.71), multivariable analysis continued to demonstrate a significant association between the use of aprotinin and an increased risk of adverse events (Table 3) as well as an absence of association between either aminocaproic acid or tranexamic acid and such events. Propensity adjustment by drug (C-index for aprotinin, 0.72; for aminocaproic acid, 0.80; and for tranexamic acid, 0.68) yielded similar findings. For example, as compared with control, aprotinin nearly doubled the odds of a renal event (odds ratio, 1.89; 95 percent confidence interval, 1.01 to 3.55; P=0.04), whereas neither aminocaproic acid (odds ratio, 0.85; 95 percent confidence interval, 0.37 to 1.95; P=0.69) nor tranexamic acid (odds ratio, 1.43; 95 percent confidence interval, 0.62 to 3.27; P=0.40) was associated with increased renal risk.
Table 3. Propensity-Adjusted Effect of Treatment on Ischemic Outcome Events.
Among the 1361 patients undergoing complex surgery, aprotinin was associated with increased renal dysfunction and renal failure requiring dialysis, whereas aminocaproic acid and tranexamic acid were not. Propensity adjustment (C-index, 0.73) confirmed these findings (Table 3), and adjustment by specific drug (C-index for aprotinin, 0.78; for aminocaproic acid, 0.78; and for tranexamic acid, 0.76) yielded similar results, as illustrated for renal events: odds ratio with aprotinin, 2.79 (95 percent confidence interval, 1.44 to 5.44; P=0.002); with aminocaproic acid, 0.48 (95 percent confidence interval, 0.17 to 1.34; P=0.16); and with tranexamic acid, 1.01 (95 percent confidence interval, 0.44 to 2.33; P=0.98). A dose–response relationship was found for aprotinin with respect to renal, cardiovascular, and composite outcomes (Figure 3).
Figure 3. Aprotinin Dose Response.
P values shown are for the comparison between a high dose of aprotinin and either a low dose of aprotinin or no antifibrinolytic therapy (control). Pairwise comparison between a high dose of aprotinin and control was significant with respect to renal events, cardiovascular events, and composite outcome events (P<0.001 for all three comparisons by Bonferroni-adjusted analyses) but not for death (P=0.12). Pairwise comparison between a low dose of aprotinin and no antifibrinolytic therapy was significant (in Bonferroni-adjusted analyses) with respect to renal events (P<0.001) and composite outcome events (P=0.003) but not cardiovascular events (P=0.08) or death (P=0.14). Pairwise comparison between a high dose of aprotinin and a low dose of aprotinin was significant (in Bonferroni-adjusted analyses) with respect to renal events (P<0.001) but not cardiovascular events (P=0.04), composite outcome events (P=0.03), or death (P=0.38).
Efficacy
The three medications reduced blood loss to similar extents. As compared with the control group, in which mean (±SD) estimated blood loss was 827±573 ml, blood loss was 753±660 ml in the aprotinin group (P<0.001), 719±578 ml in the aminocaproic acid group (P<0.001), and 676±741 ml in the tranexamic acid group (P<0.001).
Discussion
Our findings raise serious concerns regarding the safety of an approved drug intended to limit blood loss in at-risk patients undergoing surgery. Specifically, the use of aprotinin was associated with a dose-dependent doubling to tripling in the risk of renal failure requiring dialysis among patients undergoing primary or complex coronary-artery surgery. Furthermore, for the majority of patients undergoing primary surgery, we found evidence of multiorgan damage involving the heart (myocardial infarction or heart failure) and the brain (encephalopathy) in addition to the kidneys, suggesting a generalized pattern of ischemic injury. Unlike the serine protease inhibitors, analysis of the less costly lysine analogues aminocaproic acid and tranexamic acid demonstrated no such safety concerns, although these two agents were equally effective in reducing blood loss. Thus, our findings indicate that reconsideration of the safety of aprotinin among patients undergoing cardiac surgery is warranted and indicate replacement of aprotinin with either aminocaproic acid or tranexamic acid.
Blood loss and its replacement merit careful redress — particularly in cardiac surgery, for which large amounts of blood loss occur not only because of direct vascular interruption, but also because of exposure and autotransplantation of blood elements after contact with the foreign surfaces of the bypass circuitry.2,12 Consequently, prostacyclins, desmopressin, lysine analogues, and serine protease inhibitors were developed. The latter two classes are now used in the majority of the 1 million cardiac surgical operations performed annually worldwide2,3,4,10,12,15 — a practice consistent with consensus guidelines.2,4,11
Unlike lysine analogues, aprotinin has high affinity for the kidneys16,17,18,19 — a property that may explain our renal findings. After free glomerular passage, aprotinin binds selectively to the brush border of the proximal tubule membrane, and then, by pinocytosis, it enters into and accumulates within the cytoplasm, inhibiting tubule protease secretion (kallikrein and, secondarily, kinin), prostaglandin and renin synthesis, prostasin secretion, and bradykinin release.18,19,20,21,22 Under normothermic ischemia, hypothermia, and other states of high kallikrein activity, these untoward tubular effects are magnified and are complicated by dose-dependent renal afferent vasoconstriction; deep cortical and medullary perfusion and its autoregulation are thereby impaired, and focal tubular necrosis ensues.20,21,22,23,24,25,26 Furthermore, because of its interference with the synthesis and release of endothelial nitric acid, aprotinin also may instigate macrovascular or microvascular thrombosis.27,28
Despite this in vitro and in vivo evidence, only a minority of the reports of 45 trials of aprotinin in surgical patients even comment on renal function, and of those trials, none were powered to discern renal failure, leaving earlier concerns unchecked.10 However, review of this evidence suggests the presence of several renal "safety signals," including aprotinin-associated alpha1-microglobulin production,8 tubule-cell deposition of protein bands and proteinuria,8 dose-dependent increases in creatinine,5,9 renal dysfunction,9 and platelet-fibrin thrombotic occlusions of the renal arterioles after death.7 Of note, even in a small study (involving 57 patients) that concluded that aprotinin was safe (albeit with concern regarding a possible type 1 statistical error), 90 percent suppression of urinary kallikrein excretion occurred, with arithmetic increases in sodium excretion and osmolar clearance.29
Our data from approximately 1300 aprotinin-treated patients as compared with about 1300 control patients, then, clearly give credence to early concerns stemming from findings in animal models20,21,22,23,24,26 and preliminary findings in patients5,7,8,9 — namely, that aprotinin is associated with severe renal adverse events and that this association is dose-dependent. The lysine analogues, in contrast, are excreted nearly intact within 24 hours after intravenous administration, with their renal clearance approximating creatinine clearance. Moreover, few reports document an association between these agents and renal dysfunction3 — a renal-safety profile validated by our results.
Our findings raise concerns regarding the proclivity of aprotinin, but not the lysine analogues, to instigate cardiovascular and cerebrovascular thrombosis. Although questions regarding each of the agents have been raised, as indicated by early case reports showing a propensity for thrombosis, most compelling is the evidence relating to aprotinin, which distinguishes itself by at least five properties: inhibition of soluble proteases (e.g., kallikreins, plasmin, and trypsin); inhibition of activated protein C; preservation of platelet adhesive and aggregatory properties; impairment of vascular endothelial-cell function (in the coronary and cerebral arteries and aorta)20,21,23,27,28,30,31; and selective impairment of endothelium-derived relaxation by dose-dependent inhibition of nitric oxide synthesis and release.27,32
A possible link between aprotinin and intravascular thrombosis has been observed in several in vivo animal models27,28 as well as in humans — in association with biogenic materials (catheters, cannulas, and oxygenators); within coronary grafts; within the native coronary microcirculation; in the aorta; and disseminated throughout the microvasculature of the heart, lung, brain, and kidneys.5,6,7,33,34,35,36 Randomized clinical trials yielded mixed findings.5,7,37,38,39 Noteworthy, however, are the findings of a larger, more recent, sponsor-supported trial,33 which demonstrated that aprotinin-treated patients (vs. those given placebo) had a significantly greater risk of saphenous-vein occlusion, but even then the results were interpreted as inconclusive. For both aminocaproic acid and tranexamic acid — although reports of related intravascular thrombosis exist — no study has reported a significant association.3
Our findings in patients undergoing primary surgery — namely, that aprotinin-treated patients are at greater risk for ischemic damage to the heart than are either control patients or those receiving aminocaproic acid or tranexamic acid — are thus in agreement with data from the prior in vivo and in vitro studies27,28,30,34 and the majority of the coronary-graft investigations.5,7,33,37 Similarly, with regard to encephalopathy, our findings are consistent with those of earlier studies of microvascular thrombosis28,34 and specifically with those described by Sundt et al.,7 who reported platelet-fibrin thrombi among multiple vessels, including the cerebral arteries, on postmortem examination of patients who had received aprotinin. In contrast, among patients undergoing complex procedures, aside from renal outcomes, we found no other drug associations, probably because the proven blood-sparing salutary effects of antifibrinolytic therapy in patients undergoing complex surgery10 may offset any thrombotic effects. This hypothesis is supported by our secondary finding that among patients with hemorrhage, aprotinin was associated with an increase in the risk of cardiovascular events (34 percent, vs. 19 percent in the control group; P=0.04), whereas no such difference existed among patients without hemorrhage (23 percent and 24 percent, respectively; P=0.79).
Given our findings, especially with regard to serious renal events among patients undergoing either primary or complex surgery, and given the cost of aprotinin therapy, which is at least 10 times that of aminocaproic or tranexamic acid, we estimate that considerable global health care savings would accrue if aprotinin were replaced by either aminocaproic acid or tranexamic acid. Specifically, extrapolating international-use patterns, we estimate that for renal complications alone, the replacement of aprotinin with aminocaproic acid would prevent renal failure requiring dialysis in 11,050 patients per year, yielding an indirect savings (from the saved cost of dialysis) of more than $1 billion per year, in addition to direct savings (from reduced drug costs) of nearly $250 million per year. Replacement of aprotinin with tranexamic acid would prevent 9790 complications necessitating dialysis each year, yielding similar direct and indirect savings.
Regarding potential study limitations, we should note that large-scale, randomized, controlled trials — though ideal for assessing post-marketing drug safety — are difficult (if not impossible) to conduct in the setting of embedded practice, for several reasons: there is inherent bias in the selection of patients to be subjected to "nonroutine" treatments; it is necessary to withhold salutary blood-sparing therapies from those assigned to placebo; there may be reluctance to include the sickest patients, who are those most likely to have adverse events; the required sample size and cost (to detect less frequent safety events) are substantial; and there may be reluctance (a disincentive) on the part of sponsors to discern carefully the risk of a marketed drug. Given that, assessment of safety in observational studies, when sufficiently comprehensive and large, may offer critical insights, even in light of recognized limitations.
Therefore, we assessed safety in a comprehensive, large-scale, observational study based on randomized patient selection, inclusion of more than 800 high-risk patients per group, and measurement of more than 200 covariates (by drug and by outcome) per patient — an approach that permitted nonparsimonious propensity analyses as well as multivariable corrections for the multiple covariates of organ-specific outcomes.40 As such, we believe that our findings, particularly with respect to renal failure, are substantive with respect to effect size, consistency between the primary-surgery and complex-surgery groups, and dose response and that they are notable for their consistency with early in vivo and in vitro animal studies and several suggestive case reports. In addition, our specific analyses of aprotinin as compared with aminocaproic acid and tranexamic acid allowed us to compare patients to whom an antifibrinolytic agent had been administered, thereby mitigating selection bias.
In conclusion, the observed association between aprotinin and serious end-organ damage indicates that continued use is not prudent, whereas the less expensive generic medications aminocaproic acid and tranexamic acid are safe alternatives.
The Ischemia Research and Education Foundation (IREF) is an independent, nonprofit foundation, formed in 1987, that mentors clinical investigators through observational studies and clinical trials addressing ischemic injury of the heart, brain, kidney, and gastrointestinal tract. The Multicenter Study of Perioperative Ischemia (McSPI) Research Group, formed in 1988, is an association of 160 international medical centers located in 23 countries, organized through, and supported by grants from, IREF.
Supported by IREF data collection, which provided all the funding for the study, including site grants, central analysis and data disposition, manuscript grants, and publication of the findings.
No potential conflict of interest relevant to this article was reported.
* Investigators and centers participating in the study are listed in the Appendix.
Source Information
From the Ischemia Research and Education Foundation, San Bruno, Calif.
Address reprint requests to Dr. Mangano at the Editorial Office, Ischemia Research and Education Foundation, 1111 Bayhill Dr., Ste. 480, San Bruno, CA 94066, or at dtb@iref.org.
References
Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1999 Guidelines for the management of patients with acute myocardial infarction). J Am Coll Cardiol 2004;44:E1-E21.
Stover EP, Siegel LC, Body SC, et al. Institutional variability in red blood cell conservation practices for coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth 2000;14:171-176.
Fremes SE, Wong BI, Lee E, et al. Meta-analysis of prophylactic drug treatment in the prevention of postoperative bleeding. Ann Thorac Surg 1994;58:1580-1588.
Faust RJ, Harrison CR, Stehling LC, eds. Questions and answers about transfusion practices. 3rd ed. Park Ridge, Ill.: 1997 Committee on Transfusion Medicine of the American Society of Anesthesiologists, 2002:16-9, 21-2, 4-30.
Cosgrove DM III, Heric B, Lytle BW, et al. Aprotinin therapy for reoperative myocardial revascularization: a placebo-controlled study. Ann Thorac Surg 1992;54:1031-1036.
Umbrain V, Christiaens F, Camu F. Intraoperative coronary thrombosis: can aprotinin and protamine be incriminated? J Cardiothorac Vasc Anesth 1994;8:198-201.
Sundt TM III, Kouchoukos NT, Saffitz JE, Murphy SF, Wareing TH, Stahl DJ. Renal dysfunction and intravascular coagulation with aprotinin and hypothermic circulatory arrest. Ann Thorac Surg 1993;55:1418-1424.
Feindt PR, Walcher S, Volkmer I, et al. Effects of high dose aprotinin on renal function in aortocoronary bypass grafting. Ann Thorac Surg 1995;60:1076-1080.
D'Ambra MN, Akins CW, Blackstone EH, et al. Aprotinin in primary valve replacement and reconstruction: a multicenter double-blind, placebo-controlled trial. J Thorac Cardiovasc Surg 1996;112:1081-1089.
Sedrakyan A, Treasure T, Elefteriades JA. Effect of aprotinin on clinical outcomes in coronary artery bypass graft surgery: a systematic review and meta-analysis of randomized clinical trials. J Thorac Cardiovasc Surg 2004;128:442-448.
Eagle KA, Guyton RA, Davidoff R, et al. ACC/AHA 2004 guideline update for coronary artery bypass graft surgery: a report of the American College of Cardiology American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for Coronary Artery Bypass Graft Surgery). Circulation 2004;110:e340-e437.
Mangano DT. Aspirin and mortality from coronary bypass surgery. N Engl J Med 2002;347:1309-1317.
Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55.
D'Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265-2281.
Levy JH, Pifarre R, Schaff HV, et al. A multicenter, double-blind, placebo-controlled trial of aprotinin for reducing blood loss and the requirement for donor-blood transfusion in patients undergoing repeat coronary artery bypass grafting. Circulation 1995;92:2236-2244.
Beller FK, Mitchell PS, Gorstein F. Fibrin deposition in the rabbit kidney produced by protease inhibitors. Thromb Diath Haemorrh 1967;17:427-439.
Davis H, Gascho C, Kiernan JA. Effects of aprotinin on organ cultures of the rat's kidney. In Vitro 1976;12:192-197.
Kramer HJ, Moch T, von Sicherer L, Dusing R. Effects of aprotinin on renal function and urinary prostaglandin excretion in conscious rats after acute salt loading. Clin Sci (Lond) 1979;56:547-553.
Vio CP, Oestreicher E, Olivarria V, Velarde V, Mayfield RK, Jaffa AA. Cellular distribution of exogenous aprotinin in the rat kidney. Biol Chem 1998;379:1271-1277.
Seto S, Kher V, Scicli AG, Beierwaltes WH, Carretero OA. The effect of aprotinin (a serine protease inhibitor) on renal function and renin release. Hypertension 1983;5:893-899.
Kramer HJ, Düsing R, Gl?nzer K, Kipnowski J, Klingmüller D, Meyer-Lehnert H. Effects of aprotinin on renal function. Contrib Nephrol 1984;42:233-241.
Maier M, Starlinger M, Zhegu Z, Rana H, Binder BR. Effect of the protease inhibitor aprotinin on renal hemodynamics in the pig. Hypertension 1985;7:32-38.
Nakagawa M, Stewart JM, Vavrek RJ, Nasjletti A. Effects of a kinin antagonist on renal function in rats. Am J Physiol 1990;258:F643-F648.
Pfluger H, Maier M, Binder BR. Effects of antiproteases (aprotinin and gabexate-mesilate) on the postischemic vascular response of the kidney. Urol Int 1983;38:150-157.
Blauhut B, Gross C, Necek S, Doran JE, Sp?th P, Lundsgaard-Hansen P. Effects of high-dose aprotinin on blood loss, platelet function, fibrinolysis, complement, and renal function after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1991;101:958-967.
Fischer JH, Knupfer P. High-dosage aprotinin (Trasylol) therapy -- is it safe for the kidney? Langenbecks Arch Chir 1983;360:241-249.
ülker S, McKeown PP, Bayraktutan U. Aprotinin impairs coronary endothelial function and down-regulates endothelial NOS in rat coronary microvascular endothelial cells. Cardiovasc Res 2002;55:830-837.
Samama CM, Mazoyer E, Bruneval P, et al. Aprotinin could promote arterial thrombosis in pigs: a prospective randomized, blind study. Thromb Haemost 1994;71:663-669.
Schweizer A, Hohn L, Morel DR, Kalangos A, Licker M. Aprotinin does not impair renal haemodynamics and function after cardiac surgery. Br J Anaesth 2000;84:16-22.
Havel MP, Griesmacher A, Weigel G, et al. Aprotinin decreases release of 6-keto-prostaglandin F1 alpha and increases release of thromboxane A2 in cultured human umbilical vein endothelial cells. J Thorac Cardiovasc Surg 1992;104:654-658.
van Oeveren W, Harder MP, Roozendaal KJ, Eijsman L, Wildevuur CR. Aprotinin protects platelets against the initial effect of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1990;99:788-796.
Hill GE, Taylor JA, Robbins RA. Differing effects of aprotinin and episilon-aminocaproic acid on cytokine-induced inducible nitric oxide synthase expression. Ann Thorac Surg 1997;63:74-77.
Alderman EL, Levy JH, Rich JB, et al. Analyses of coronary graft patency after aprotinin use: results from the International Multicenter Aprotinin Graft Patency Experience (IMAGE) trial. J Thorac Cardiovasc Surg 1998;116:716-730.
Saffitz JE, Stahl DJ, Sundt TM, Wareing TH, Kouchoukos NT. Disseminated intravascular coagulation after administration of aprotinin in combination with deep hypothermic circulatory arrest. Am J Cardiol 1993;72:1080-1082.
Alvarez JM, Chandraratna H, Newman MA, Levy JH. Case 3-1999: intraoperative coronary thrombosis in association with low-dose aprotinin therapy. J Cardiothorac Vasc Anesth 1999;13:623-628.
Penkoske PA, Entwistle LM, Marchak BE, Seal RF, Gibb W. Aprotinin in children undergoing repair of congenital heart defects. Ann Thorac Surg 1995;60:Suppl 6:S529-S532.
Lemmer JH Jr, Stanford W, Bonney SL, et al. Aprotinin for coronary bypass operations: efficacy, safety, and influence on early saphenous vein graft patency: a multicenter, randomized, double-blind, placebo-controlled study. J Thorac Cardiovasc Surg 1994;107:543-553.
Bidstrup BP, Underwood SR, Sapsford RN, Streets EM. Effect of aprotinin (Trasylol) on aorta-coronary bypass graft patency. J Thorac Cardiovasc Surg 1993;105:147-153.
Havel M, Grabenwoger F, Schneider J, et al. Aprotinin does not decrease early graft patency after coronary artery bypass graft despite reducing postoperative bleeding and use of donated blood. J Thorac Cardiovasc Surg 1994;107:807-810.
Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342:1887-1892.(Dennis T. Mangano, Ph.D.,)