Conservation of blood is a priority during surgery, owing to shortages of donor blood and risks associated with transfusion of blood products.[9,10] However, blood transfusions have been linked to a number of negative postoperative sequelae, including poorer prognosis after cardiac and cancer surgery.[11- 21] In this context, recognition that allogeneic transfusion-associated immunomodulation can increase morbidity in allogeneically transfused patients has become a major concern in transfusion medicine.[9,22,23] In cancer surgery, various studies have documented a positive association between transfusion and death and relapse. In several retrospective analyses of transfusions in colorectal cancer surgery, long-term survival (3 to 5 years) ranged from 60% to 81% for patients not transfused vs 37% to 63% for transfused patients (for all studies, P < .05).[13-15] In one of these studies, the deleterious effect of transfusion was evident in some patients after they received a single unit of blood.[13] In other studies, perioperative blood transfusion was identified as an independent risk factor for colorectal cancer relapse (P = .05).[13,15] Discrepant long-term and short-term survival rates have also been observed in patients with esophageal carcinoma, based on perioperative allogeneic-blood-transfusion status.[17,20] In most of the studies involving esophageal resection, intraoperative allogeneic blood transfusion was an independent predictor of, or prognostic covariate for, patient survival.[ 9,16,18,19,21,24] Reducing the Need for Allogeneic Blood Transfusion Bacterial infection, mistransfusion, and transfusion-related acute lung injury account for most transfusion-related deaths, with bacterial contamination responsible for approximately seven deaths per million units transfused. Although it is the third leading cause of transfusion-related mortality, transfusion-related acute lung injury has been frequently underdiagnosed and underreported, even when it occurred in several patients receiving transfusions from the same frequent plasma donor.[25] Thus, transfusionrelated acute lung injury represents a substantial risk for patients receiving a transfusion. Evidence from animal models indicates that transfusion of platelet concentrates may cause transfusion- related acute lung injury as a result of infusion of bioactive lipids generated during storage.[26] Not unexpectedly, the risk of transfusion- related infection in cardiopulmonary bypass patients depends on the number of units transfused.[11] Multiple linear and logistic regression analyses of data from patients (n = 238) who underwent first-time coronary artery bypass graft surgery indicated that the amount of homologous blood transfused was a significant and independent predictor of postoperative infection.[11] Infections were observed in 3.9% of patients receiving up to 2 units of red blood cells (RBCs) or whole blood, 6.9% of patients receiving 3 to 5 units, and 22% of patients receiving 6 units or more (P = .0022). Although the risk of transmission of infectious agents is well recognized, the dose-response relationship between transfusion and infection may also be attributed to the immunosuppressive effects of homologous blood transfusions.[11] A dose-dependent rate of infection has been seen in other studies of transfused cardiopulmonary bypass populations.[ 27] Higher rates of postoperative bacterial infections occurred in patients transfused with packed cells without buffy coat compared with patients given leukocyte-depleted blood (P = .06 overall and P = .04 for those who received more than three transfusions). Mortality within 60 days was also significantly lower in patients receiving leukocyte-depleted blood (P = .025) compared with recipients of packed cells, and the effect was dose-dependent.[27] Similarly, a recent retrospective cohort study in 23 Canadian hospitals found significantly lower unadjusted in-hospital mortality rates following the introduction of leukoreduction, compared with the control period (6.19% vs 7.03%, respectively; P = .04).[28] Notably, the adjusted odds of death following leukoreduction were decreased (odds ratio [OR]: 0.87; 95% confidence interval [CI]: 0.75 to 0.99), but serious nosocomial infections did not decrease (adjusted OR: 0.97; 95% CI: 0.87 to 1.09) compared with the control period. Although leukocytes in transfused blood can produce immunomodulatory effects that increase the risk of infection, the significant reduction in mortality following leukoreduction cannot be fully explained by marginal changes in infection incidence noted in the previous studies. The precise mechanisms underlying the link between leukoreduction and outcome have not yet been elucidated, but it has been postulated that transfused leukocytes, activated during storage, contribute to an existing inflammatory response and exacerbate tissue damage.[27] Recent studies indicate that not only are short-term mortality and morbidity increased by blood transfusion, but also long-term survival rates are reduced. A review of long-term patient- survival data (n = 1,915) from the United States Social Security Death Index showed that blood transfusion during or after coronary artery bypass graft surgery is associated with decreased long-term survival.[12] In a separate study, transfused cardiopulmonary bypass patients had twice the 5-year mortality of nontransfused patients (15% vs 7%, respectively). After correction for comorbidities and other factors, transfusion was still associated with a 70% increase in mortality (relative risk [RR]: 1.7; 95% CI: 1.4 to 2.0; P = .001). Transfusion remained a significant predictor (P = .04) of long-term (1- to 5-year) mortality in multivariate analysis. Finally, reexploration for bleeding was identified as a strong independent risk factor for operative mortality (P = .005) in a separate multivariate logistic regression analysis of data from cardiopulmonary bypass patients (n = 6,015).[29] Serine Proteases in Coagulation and Inflammation
Clearly, although the amount of perioperative bleeding in cancer surgery can be substantial enough in some cases to warrant a blood transfusion, such transfusions do have the potential to negatively impact patient outcomes by generating immunosuppressive and generalized inflammatory responses. It is interesting that hemostasis and inflammation share several reactants in common and both serve as hostdefense mechanisms.[2,30] The activation of coagulation and inflammation is closely linked through a network of both humoral and cellular components, including proteases of the coagulation and fibrinolytic cascade.[ 5] Serine proteases are essential for virtually all inflammatory and coagulative processes in the extracellular or plasma phase, and they are represented by such ubiquitous molecules as trypsin, thrombin, plasmin, plasminogen activator (PA), kallikrein, and elastase.[2] The normal physiologic response to injury results in the generation of procoagulants, primarily tissue factor, which initiates thrombin generation and clot formation, and in the generation of plasminogen activator, which is central to coordinated cell proliferation and tissue remodeling.[30] Generation of thrombin is key to activation and release of several humoral mediators involved in hemostasis and inflammation.[2] A critical serine protease of the hemostatic system, thrombin is the final common mediator of both the intrinsic and extrinsic coagulation pathways, mediating the proteolytic cleavage of fibrinogen to fibrin and catalyzing the cross-linkage of the fibrin clot.[31-36] Clot formation is typically initiated by a series of platelet-related events that, together with blood trauma and/or the exposure of blood to tissue factor, promote activation of the coagulation system.[2,33-36] Amplification and progression of the hemostatic system requires the presence of an organizing surface, zymogen, cofactor, and serine protease.[ 2] Thrombin, in addition to being a major effector protease in the coagulation cascade (converting fibrinogen to fibrin), has many secondary effects.[ 31,32] For example, this serine protease triggers expression of procoagulant activity on the platelet surface by activating cofactors of tenase and prothrombinase complexes, supporting the generation of additional thrombin.[2] Thrombin is a direct agonist of platelet activation and aggregation through a protease-activated-receptor- mediated series of events.[31,32] It triggers platelet release of platelet agonists such as adenosine(Drug information on adenosine) diphosphate, serotonin, and thromboxane, which further amplify the platelet-activation process, and it triggers release of chemokines and growth factors.[2,37] In addition, thrombin mobilizes adhesive proteins and activates the platelet glycoprotein (GP) IIb/IIIa receptor, which has high affinity for fibrinogen and von Willebrand factor.[32,38-40] Thrombin is integral to angiogenesis and smooth-muscle-cell proliferation, by stimulating secretion of growth factors such as platelet-derived growth factor and vascular endothelial growth factor.[31,32,41-44] Thrombin activates inflammatory processes and is chemotactic for monocytes and mitogenic for lymphocytes.[ 5,32,41] Fibrinolysis and the PlasminogenPlasmin System
Once a fibrin surface is formed, fibrinolysis is initiated by the generation of plasmin, a serine protease with broad trypsin-like specificity.[1,4,45,46] Plasmin not only is responsible for the degradation of fibrin, fibrinogen, and other clotting factors during clot dissolution, but it also is capable of degrading virtually all components of the extracellular matrix (ECM). In addition, it stimulates activation of other proteases, such as MMPs and elastase, which further degrade the extracellular matrix.[1] Plasmin is derived from its precursor plasminogen (zymogen) via the endogenous plasminogen activators urokinase(Drug information on urokinase)-type PA (uPA) and tissuetype PA (tPA)[1,4,45-47] (Figure 1). These two enzymes appear to have different physiologic roles, with tPA being primarily associated with clot lysis and uPA mediating tissue-remodeling processes.[1] Even small amounts of plasminogen activator can result in high local concentrations of plasmin, through the action of uPA and tPA. These activators are opposed by plasminogen-activator inhibitors (PAIs), designated PAI-1, -2, and -3, and the activity of plasmin itself is regulated by naturally occurring serine protease inhibitors, such as alpha2-antiplasmin and alpha2-macroglobulin. Urokinase-type plasminogen activator is secreted by a variety of both normal and neoplastic cells as a singlechain proenzyme (pro-uPA) with virtually no intrinsic enzymatic activity.[ 1,47] However, pro-uPA can be activated by a variety of serine proteases, including plasmin, kallikrein, and trypsin-like enzymes, producing a high-molecular-weight form of uPA that is further degraded into enzymatically active low-molecular-weight uPA. Indeed, trace amounts of plasmin are able to activate pro-uPA, thus generating a feedback mechanism of prou PA and plasminogen activation.[1] The specific cellular receptor for uPA (uPA-R) is found on a variety of cell types and appears to play a central role in mediating pericellular proteolytic activity.[1,46-48] After secretion, pro-uPA binds to uPA-R and is activated by proteolytic cleavage to the enzymatically active uPA form. The interaction of uPA with uPA-R ensures focal localization of enzyme activity on the cell surface, and plasminogen activation is accelerated owing to the juxtaposition of uPA and plasminogen. In addition to maximizing uPA and plasminogen interactions, such binding also impedes inactivation by naturally occurring inhibitors. Thus, the cell surface is the preferential site for plasminogen activation as uPA binds to its specific cellular receptor. Bound uPA is more active than unbound uPA for plasmin generation. This arrangement is optimal for efficient generation of pericellular proteolytic activity.[1,47,48] Multifunctionality of Serine Protease Inhibitors
A single-chain polypeptide comprising 58 amino-acid residues, aprotinin(Drug information on aprotinin) inhibits the action of numerous serine proteases, with decreasing affinity for trypsin, plasmin, kallikrein, elastase, urokinase, and thrombin, respectively. The complex pharmacodynamics of aprotinin translates into a dose-dependent effect on serine protease activity. At low concentrations (eg, about 50 kallikrein-inhibiting units [KIU]/mL), aprotinin is a powerful inhibitor of plasmin, but at higher concentrations (> 200 KIU/mL) it also possesses inhibitory activity against kallikrein, elastase, urokinase, and thrombin[49] (Figure 2). Hemostatic Properties of Aprotinin
Although the source of cardiopulmonary bypass-induced coagulopathy is multifactorial, platelet dysfunction has been implicated as a primary cause of postoperative bleeding in this setting.[ 8,50,51] During extracorporeal circulation of blood, the expression of platelet adhesive receptors, such as glycoprotein (GP) Ib, GP IIb, GP IIa, and GP IIb/IIIa, is reduced. This decline in the numbers of adhesion receptors on the platelet surface is mediated in part by plasmin.[52,53] Dysregulated fibrinolysis also contributes to the hemostatic defect that accompanies extracorporeal circulation.[ 50] During fibrinolysis, platelet receptors bind fibrin degradation products in place of fibrinogen, leading to impaired platelet aggregation and function.[51] Aprotinin acts in a variety of interrelated ways to reduce platelet dysfunction by inhibiting serine proteases, such as plasmin and kallikrein, and preserving platelet receptors (eg, GP Ib and others).[8,51,54] Plasmin is directly inhibited by aprotinin, but aprotinin also blocks contact activation of kallikrein, which is partly responsible for creating enzymatically active uPA that converts plasminogen to plasmin. These antiplasmin activities retard the inhibitory effect of plasmin on the expression of platelet adhesive receptors. Furthermore, the inhibition of plasmin by aprotinin directly diminishes fibrinolysis, in turn causing a reduction in fibrin/fibrinogen degradation products, such as Ddimer, that otherwise would impair platelet function. Thus, the hemostatic effect of aprotinin can be attributed to both its inhibition of fibrinolytic activity and its preservation of platelet membrane-binding functions. Clinical studies have established that the antifibrinolytic and plateletprotective properties of aprotinin can decrease blood loss and transfusions in several subsets of surgical patients.[ 55-60] Subsequent double-blind, randomized, placebo-controlled studies confirmed the transfusion-sparing properties of aprotinin in primary and reoperative cardiac surgery.[57-60] Recent results from randomized, controlled studies have also shown that aprotinin decreases perioperative bleeding and blood-transfusion requirements in a dose-dependent fashion, in orthopedic and transplantation surgery as well as cancer surgery. A study in orthopedic surgery (n = 58), which compared "largedose" (4 × 10⁶ KIU loading dose, followed by 1 * 10⁶ KIU/h infusion) and "small-dose" aprotinin (2 * 10⁶ KIU loading dose, followed by 5 * 10⁵ KIU/ h infusion), showed a significant reduction (P < .05) in postoperative drainage in the two aprotinin groups, compared with placebo.[61] Total measured bleeding and total calculated bleeding decreased significantly (both P < .05) in the large-dose group compared with placebo but did not achieve statistical significance in the smalldose group. The total number of transfused homologous and autologous units was also significantly decreased (P < .05) in the large-dose aprotinin group vs the placebo group. In orthotopic liver transplantation (European Multicentre Study in Aprotinin in Liver Transplantation), aprotinin significantly lowered intraoperative blood loss, with a reduction of 60% in the "high-dose" group and 44% in the "regular-dose" group compared with placebo (P = .03 comparing all three groups).[62,63] The "high-dose" aprotinin regimen consisted of a 2 * 10⁶ KIU loading dose, followed by 1 * 10⁶ KIU/h infusion, plus 1 * 10⁶ KIU before graft reperfusion. The "regular-dose" group received a full Hammersmith regimen. A comparison of these dosing schedules showed that the total amount of homologous and autologous RBC transfusion requirements was 37% lower in "high-dose" recipients and 20% lower in "regular-dose" recipients, compared with patients in the placebo group (P = .02, comparing all three groups). These findings are in line with the significant reduction (P < .03) in transfusion requirements with aprotinin reported in the reoperative heart-transplantation study.[50] Thus, aprotinin has been shown to improve hemostasis in both cardiac and abdominal surgery. Studies in Cancer Patients Importantly, significant blood- and transfusion-sparing effects have also been demonstrated with aprotinin in patients undergoing resection for primary malignant, metastatic, or benign tumors of the liver.[64,65] In a doubleblind, prospective, randomized study, patients (n = 97) undergoing elective liver resection were stratified by diagnosis and assigned to "large-dose" aprotinin (2 * 10⁶ KIU loading dose, followed by 5 * 10⁵ KIU/h infusion, plus a 5 * 10⁵ KIU bolus for every 3 transfused RBC units) or placebo. Results showed a significant overall reduction in intraoperative blood loss with aprotinin, compared with placebo (mean: 1,217 vs 1,653 mL, respectively; P = .048).[64] In stepwise logistic regression analysis, aprotinin treatment remained significantly correlated with blood loss after an adjustment for diagnosis of underlying disease, age, preoperative hematocrit, type of surgery, duration of clamping, repeat surgery, and postoperative Ddimer levels. The percentage of transfused patients (17% vs 39%, respectively; P = .02) and the total transfusion requirement (30 vs 77 RBC units, respectively; P = .015) were also significantly lower in the aprotinin group vs the placebo group. Given the independent prognostic value of PAI-1 levels in at least some tumor types,[66,67] it is noteworthy also that the increase in PAI levels in this study was significantly lower with aprotinin than with placebo (P < .0001).[64] The overall findings of the previous study were reproduced in a subanalysis restricted to patients with colorectal metastasis. In this cohort, intraoperative blood loss (P = .037) and transfusion requirements (P = .03) were significantly reduced in patients treated with aprotinin vs placebo.[65] A similar intraoperative increase in thrombin-antithrombin complexes in aprotinin and placebo groups indicated a comparable activation of coagulation. As in the whole study population, however, aprotinin significantly reduced (P = .01) intraoperative hyperfibrinolysis compared with placebo, as measured by intergroup comparison of D-dimer levels. Most of the safety experience with aprotinin has been outside the oncology setting, in patients undergoing cardiac surgery. Current evidence indicates that clinically relevant doses of aprotinin not only are generally safe and well tolerated,[58,59,68- 71] but also are associated with lower mortality risk in this patient population.[ 71] When considered together, ample evidence indicates that blood transfusions increase the risk of mortality and relapse, and may, in fact, be an independent risk factor for these events following resection of some tumors. The underlying mechanisms for these adverse outcomes have yet to be fully elucidated but may include transfusion- related immunosuppression and inflammation. Immune suppression not only increases the risk of postoperative infections but probably also increases the odds of cancer relapse and recurrence.[9] These immune-system changes take place in a milieu of transfusion-induced inflammation and resulting tissue injury. Accordingly, use of serine protease inhibitors or other transfusion-sparing agents may contribute to improved outcomes after resection of intrathoracic and intra- abdominal malignant disease.