Moving Toward Evidence-based Management of Coagulopathy in Cardiac Surgery

Moving Toward Evidence-based Management of Coagulopathy in Cardiac Surgery
Kenichi A. Tanaka, MD, MSc1, Michael Mazzeffi, MD, MPH1, Daniel Bolliger, MD2
1Department of Anesthesiology, Cardiothoracic Anesthesia Division, University of Maryland, Suite S8D12, Baltimore, MD, USA
2Department of Anesthesia, University of Basel Hospital, Basel, Switzerland

Key words
Cardiac surgery; coagulopathy; blood component transfusion; coagulation testing

Summary
Major blood loss during cardiac surgery results in anemia and coagulopathy, necessitating blood transfusion. Allogeneic blood components including packed red blood cells (PRBC), platelets, and plasma are most commonly administered as a life saving measure in the cardiac operating room. Increasing uses of potent antiplatelet and anticoagulant agents in the aging population have also brought a major challenge in perioperative hemostasis and transfusion. Blood components are a scarce and expensive resource, and implementing an evidence-based transfusion protocol is important to reduce arbitrary and inappropriate uses. In addition to conventional blood components, plasma-derived and recombinant factor concentrates are indicated for specific perioperative bleeding conditions. It is pivotal to utilize appropriate coagulation tests to monitor dynamic coagulation changes, and therapeutic responses to various hemostatic interventions in the perioperative period. This review focuses on the recent clinical data on: (i) allogeneic blood component products and factor concentrates, (ii) clinical utility and limitations of coagulation testing, and (iii) perioperative management of antithrombotic agents.

Introduction
Blood transfusion has been a crucial resuscitative measure in cardiac surgery for more than 60 years. Anemia, thrombocytopenia, and coagulation factor deficiency are often associated with the use of cardiopulmonary bypass (CPB). Today the majority of transfusion is administered as a component separated from the donated whole blood; packed red blood cells (PRBC), platelets, plasma, and cryoprecipitate. It is thus important to determine the need of each component in a timely fashion in major bleeding situations. Intraoperative diagnosis of anemia is feasible in real time using a blood gas analyzer (CO-oximeter) [1] or a capillary (HemoCue) device [2]. However, standard laboratory coagulation testing usually takes 30 to 90 min [3,4], and blood components are often administered empirically to prevent and/or treat severe coagulopathy [5]. Although this approach can be life saving in a patient with uncontrolled bleeding, imprecise selection and untimely dosing of hemostatic components may result in under- or over-dosing [6,7].

Approximately 5–15% of major cardiac surgical patients with cardiopulmonary bypass (CPB) suffer from major hemorrhage and massive transfusion [8,9,10]. Protracted bleeding after CPB and extracorporeal membrane oxygenation (ECMO) support is associated with increased morbidity and mortality [11,12]. A timely hemostatic intervention could potentially reduce complications secondary to hemodynamic instability, mechanical ventilation, and large amounts of allogeneic blood products [13,14,15,16].

The safety of blood products with regard to pathogen transmission risks has improved over the years, but there are ongoing debates about non-infectious complications including storage lesions of PRBCs [17,18], post-transfusion alloantibody formation [19], and transfusion-related acute lung injury (TRALI) [20,21,22]. There have been clinical efforts to minimize allogeneic blood product usage by utilizing a transfusion algorithm which incorporates point-of-care coagulation testing, and early uses of plasma-derived factor concentrates [10,13,14,15,23].

There are still needs for improving intraoperative monitoring of coagulation, and optimizing the use of allogeneic blood products and pharmacological hemostatic agents.

This review focuses on the current evidence and emerging data in cardiac surgery relating to: (i) allogeneic blood component products and factor concentrates, (ii) clinical utility and limitations of coagulation testing, and (iii) perioperative management of antithrombotic agents.

Laboratory Testing of Coagulopathy
Use of a transfusion algorithm has been shown to be effective in reducing blood usage in numerous studies in cardiac surgery [24,25].

Conventional laboratory tests including prothrombin time (PT) and activated partial thromboplastin time (aPTT) have a long track record as a screening tool for hereditary bleeding disorders. However, a long turnaround time (30–45 min), and sensitivity to heparin during CPB make laboratory-based PT and aPTT impractical for guiding plasma transfusion in the case of serious bleeding immediately after CPB [7]. Further, there is a paucity of data supporting the use of PT/aPTT in predicting multi-factorial coagulopathy of perioperative patients [26,27].

Viscoelastic coagulation tests including thrombelastography (TEG®; Haemonetics, Niles, IL) and thromboelastometry (ROTEM®; TEM Innovations, Munich, Germany) differ from conventional PT and aPTT in several aspects. First, they are primarily performed in the whole blood using citrated whole blood samples, and testing can be immediately started without plasma separation. Second, testing can be performed during CPB because a reagent containing a heparin neutralizer is available. Third, the key end-point of viscoelastic testing is fibrin polymerization, which is not reflected in either PT or aPTT.

Key concepts and uses of TEG and ROTEM have been reviewed elsewhere [7,25,28,29]. The practical aspects of ROTEM are discussed here because this device was used to guide coagulation therapy in several clinical studies referred in the later section of this review. For a standard ROTEM measurement, citrated whole blood sample (300 µL) is placed in a plastic cup using an automated pipette (Figure 1). The sample is recalcified with CaCl2, 0.2 mmol/L (StarTEM; 20 µL) and activated with 20 µL of an EXTEM (tissue factor [TF]) or INTEM (ellagic acid) reagent. Subsequently, the plastic pin is immersed in the blood. Once thrombin is generated in the blood, platelets are activated to express glycoprotein (GP) IIb/IIIa receptors, and fibrin is formed and polymerized. The interactions of GP IIb/IIIa receptors and polymerized fibrin increase the torque (viscoelasticity) between the cup and the rotating pin (at a 4.75° angle). The breakdown of fibrin strands by fibrinolysis decreases the torque. The change in torque is detected optically and is processed by the microprocessor to trace clot formation and a possible breakdown. The commonly used ROTEM variables include coagulation time (CT; seconds), clot formation time (seconds), α-angle (degrees), amplitude at 10 minutes after CT (A10; millimeters), maximum clot firmness (MCF; millimeters), and maximum lysis (ML; maximal % decrease in clot firmness; (Figure 2). Several other tests are available in addition to EXTEM and INTEM for a specific diagnosis of coagulation problems. FIBTEM is a modified EXTEM test with cytochalasin D, which inhibits platelet cytoskeletal reorganization and, thus, fibrin(ogen) binding to platelet GP IIb/IIIa (Figure 2A, B). By combining EXTEM and FIBTEM, the differential diagnosis of thrombocytopenia and/or hypofibrinogenemia is feasible within 20 minutes (Figure 2C). APTEM is also a modified EXTEM, in which aprotinin inhibits plasmin in vitro if systemic fibrinolysis is present (Figure 2D, E). HEPTEM contains a heparinase in addition to the INTEM reagent. It is used as a pair with INTEM for the diagnosis of systemic heparin activity [30] (Figure 2F).

The use of transfusion algorithm incorporating thromboelastometry in combination with factor concentrates has been shown to reduce allogeneic blood product usage in adult patients undergoing complex cardiac surgery with CPB (Figure 3) [10,13,14,15,23].

Evidence-Based Use of Blood Component Transfusion and Prophylactic Antifibrinolytics

Platelet Transfusion
Thrombocytopenia and platelet dysfunction are the most common reasons for platelet transfusion in cardiac surgery. Clinically significant bleeding (>1000 ml in 12 hrs) [31] after CPB is frequently attributed to thrombocytopenia and/or platelet dysfunction [32]. However, the diagnosis of post-CPB bleeding is often subjective, and thus biased by preoperative medication(s), coexisting disease(s) of the patient, type of surgery, and clinical experience of the care team. Platelet count is most commonly used as a trigger for platelet use, but there is no single pre- or post-operative threshold that had been associated with increased perioperative bleeding. It is therefore typical to transfuse platelets when post-CPB platelet count is between 50 x103 and 100 x103 per µL according to the perceived risk of bleeding. A simple transfusion algorithm using the cut-off value of 50x103/µL for coronary artery bypass grafting surgery (CABG) was highly effective in reducing the rate of platelet transfusion when compared to the historical control at one institution [33]. However, a lower platelet threshold may only be applicable to CABG without preoperative P2Y12 inhibitor therapy. Platelets not only play a pivotal role to form a primary hemostatic plug, but they also interact with plasma components of coagulation, augmenting thrombin generation and fibrin polymerization [28]. In more complex cardiac surgical procedures such as multi-valve replacement or replacement of aorta, plasma coagulation factors may be decreased to 20-30% of normal. Disruptions of platelet’s procoagulant activity thus occurs due to insufficient fibrinogen binding to platelet glycoprotein IIb/IIIa receptors, and/or low thrombin generation on aggregated platelets. Multiple units of platelet transfusion are often empirically given in these complex cases [34]. The use of transfusion algorithm including a viscoelastic testing is practical in the simultaneous assessment of platelet count and overall coagulation [25]. Indeed, platelet usage was lower in the thromboelastometry-guided group relative to the control group in the recent studies in aortic surgery using deep hypothermic circulatory arrest (DHCA) (Table 1) [13,15].

Heparin-induced thrombocytopenia (HIT) is a unique condition with underlying thrombophilia. Prophylactic platelet transfusion is not indicated even before an invasive procedure. However, anticoagulation for CPB in HIT patients typically requires direct thrombin inhibitor (e.g., bivalirudin), for which there is no direct antidote. Platelet transfusion may not be avoidable in the management for severe bleeding associated with a direct thrombin inhibitor [35,36,37].

In the case of ongoing P2Y12 inhibitor therapy, preoperative assessment of residual responsiveness to adenosine 5’-diphosphate (ADP) is reasonable because there is a relatively high incidence of clopidogrel non-responsiveness in patients at a high risk of stent thrombosis [38,39]. In a prospective study of clopidogrel-treated patients (n=86) and clopigodrel-naïve patients (n=94), Marla, et al. recently demonstrated that adjusting a waiting period between less than 1 day and 5 days before CABG according to the residual platelet responsiveness to ADP is feasible without increasing bleeding. 24-hr chest tube drainage (median, 650 vs. 780 ml; P=0.08), transfusion rates of PRBC (median, 2 units in both; P=0.54), and 30-day mortality (1 death in each group; P=1.0) were similar between 2 groups [40]. Their study, using residual platelet responsiveness to ADP, showed a 46% shortening of the guideline recommended preoperative waiting period before CABG for clopidogrel-treated patients (mean 2.7 days versus 5 days per patient) with no increased bleeding.

For urgent procedures, preoperative testing for P2Y12 inhibition can assess a potential need for postoperative platelet transfusion. In addition, it may be important to consider therapeutic responses to platelet transfusion because transfused allogeneic platelets are inhibited by circulating residual P2Y12 inhibitor. The recovery of ADP-induced platelet aggregation could be much less than expected from the increased platelet count after transfusion. In general, 2 to 4 units of apheresis platelet transfusion are necessary to reverse dual treatment with aspirin and clopidogrel [41], however allogeneic platelets are extensively inhibited when they are added ex vivo to the blood samples from the patients on aspirin plus prasugrel [42] or ticagrelor [43]. Platelet transfusion is less effective when it is given within 2 hrs from the last dose of prasugrel or ticagrelor, but adding a wait time of 6–24 h would allow natural recovery of platelets from the P2Y12 inhibitors (Table 2).

Bleeding risk from aspirin monotherapy is mitigated by the routine use of antifibrinolytic therapy in cardiac surgery (lysine analogues, and previously aprotinin) [44]. It thus remains controversial if routine testing of platelet ADP response is indicated in cardiac surgical patients without preoperative P2Y12 inhibitor therapy. In a study involving 60 cardiac surgical patients, hematocrit and platelet count were decreased to 73% and 49% of the baseline values after CPB (medina, 108 min). In parallel, platelet aggregation responses to thrombin receptor agonist peptide (TRAP) and ADP were decreased to 50% and 42% of the baseline values on the whole blood aggregometry (Multiplate, Roche, Munich, Germany) [45]. Thrombocytopenia inversely affects TRAP and ADP responses on the whole blood aggregometry [46]. It is thus practical to use platelet count or clot amplitude of TEG or ROTEM to guide platelet transfusion in bleeding patients without prior P2Y12 inhibitor therapy [7,15].

Plasma Transfusion
Coagulation defects after CPB has been treated with the transfusion of fresh frozen plasma (FFP) for many decades. FFP still remains the mainstay therapy at cardiac centers in the United States (US), and a variety of plasma products is interchangeably used with FFP. For practical reasons, we use the term “plasma” to include FP24 (plasma frozen within 24 hrs of collection) and FFP (plasma frozen within 8 hrs of collection) that are used to treat coagulopathy [47].

One of the key mechanisms of coagulation defects after CPB is the occurrence of hemodilution. The priming volume of CPB and fluid replacements (crystalloid and colloid) for bleeding are two major culprits of hemodilution. The extent of hemodilution is particularly high in neonatal and infant CPB cases because the reservoir volume often exceeds the circulating blood volume. Plasma transfusion is generally unavoidable to manage post-CPB bleeding in these cases, but there has been a paucity of clinical data on how plasma transfusion could be optimized. Nakayama, et al. recently reported a prospective randomized study (n=100) of conventional vs. ROTEM-guided hemostatic intervention in cardiac surgery for infants (median age, 10–13 months) [16].

Platelets and plasma (FFP) were only two products available for hemostatic therapy at their institution. In the conventional group, platelets and plasma were transfused when platelet count was below 80,000 per µL, and post-protamine ACT was over 150 sec. Corresponding indications in the ROTEM group were platelets for EXTEM-A10 (10-min amplitude) below 30 mm, and plasma for FIBTEM-A10 below 5 mm. The amount of intraoperative plasma transfusion was higher in the ROTEM group than the conventional (median, 21 ml/kg vs. 14 ml/kg; P<0.005). This resulted in higher fibrinogen levels in the ROTEM group immediately after surgery (165 vs. 125 mg/dL; P<0.001). Reduced postoperative blood loss, PRBC transfusion, and duration of ICU stay in the ROTEM group suggest that early plasma transfusion to maintain fibrin polymerization may be more effective than conventional management focused on clotting time (e.g., PT/INR, aPTT or ACT). Incremental changes in coagulation factors are relatively small after plasma transfusion. Factors (F) II, V, VII, and X were increased by median 10–16% after 12.2 ml/kg, and 28–44% after 30 ml/kg of plasma transfusion. Volume overload is clearly a major challenge to plasma transfusion in patients with limited cardiopulmonary reserve. This problem was highlighted when plasma transfusion (n=81) was compared to prothrombin complex concentrate (PCC) (n=87), a lyophilized concentrate of FII, FVII, FIX and FX, for acute warfarin reversal (INR 2.9) before surgery [48]. Clinical impacts of a lower volume of PCC compared to plasma (mean, 89.7 vs. 819 ml) were evident in the shorter duration of therapy (20.9 min vs. 141 min), the higher factor levels, and the lower incidence of fluid overload (3% vs. 13%; P=0.0478). Clinical experience and data in congenital and acquired bleeding management support the preferred use of factor concentrate over plasma when deficient factors are specifically known, and replacement factors are available as plasma-derived or recombinant freeze-dried protein(s) [49,50]. In massive haemorrhage and extensive factor deficiencies, plasma transfusion remains as the important therapy in replacing FV and FIX as well as endogenous coagulation and fibrinolysis inhibitors [51].

Plasma transfusion is often administered at 2 units (without weight-based dosing) to adult patients in various clinical settings [52]. However, a moderately elevated of INR (1.5–1.85) minimally improves after 2 units of plasma [53]. Emerging clinical data show some promise of further reducing plasma transfusion in cardiac surgery by primarily focusing on fibrinogen replacement [10,14,23]. Clinical implications of fibrinogen-rich components will be discussed in the section below.

Fibrinogen Concentrate and Cryoprecipitate
Plasma fibrinogen is normally in the range of 150 to 350 mg/dL, but in post-CPB bleeding, it is generally below 150 mg/dL. Allogeneic plasma products contain fibrinogen at around 200 mg/dL, but post-transfusion fibrinogen levels achievable in a patient is only about 60% of the original product level even after a massive plasma transfusion (i.e., 100 to 140 mg/dL) [54].

The threshold fibrinogen level for hemostasis after CPB is reported to be >200 mg/dL [55,56] in contrast to 100 mg/dL in isolated fibrinogen deficiency (e.g., afibrinogenemia) [57]. Rannuci, et al. conducted a single-center, prospective randomized study of the plasma-derived fibrinogen concentrate vs. placebo in moderately complex cardiac surgery patients (n=116) with expected CPB time over 90 min [23]. They applied to both groups the rigorous transfusion criteria for PRBC (Hgb <7–8 g/dL), platelet (<50 x103/µL), and plasma (INR >1.5 or TEG reaction time >12 min). The intervention group received the median 4g (IQR, 3–6g) of fibrinogen to the target level of 22 mm on FIBTEM-MCF immediately after protamine administration. Plasma fibrinogen levels were 367 (IQR, 329–410) vs. 242 (IQR, 199–300) mg/dL in the intervention and placebo group, respectively (P=0.001). For the primary endpoint of blood transfusion in 30 days, PRBC use was less frequent in the intervention compared to the placebo (32.8% vs. 55.2%; P=0.015). Four out of 5 patients who bled more than 1 L within the first 12 hrs belonged to the placebo group. None of the fibrinogen-treated patients required platelet or plasma transfusion, but 6.9% and 13.8% of placebo-treated patients received platelet and plasma transfusion, respectively (P=0.006 for plasma). Although their study may be criticized for implementing a hemostatic intervention without the clinical evidence of microvascular bleeding, the results show the advantage of maintaining fibrinogen at mid–high normal ranges to reduce post-CPB bleeding.

Clinical use of fibrinogen concentrate is limited to hereditary afibrinogenemia and hypofibrinogenemia in the US, but cryoprecipitate can be used for perioperative fibrinogen replacement. Plasma-derived fibrinogen concentrate is treated with several pathogen reduction processes, and is free of ABO antibodies. However, cryoprecipitate is not treated with a pathogen reduction procedure, and thawing and blood-type compatibility are prerequisite for transfusion. There is a relative paucity of data on the efficacy of cryoprecipitate, but a preliminary clinical comparison of fibrinogen concentrate (n=30) and cryoprecipitate (n=33) in infants (6–8.4 kg) demonstrated that both products were comparable in post-dose fibrinogen increments in the management of post-CPB hypofibrinogenemia (<100 mg/dL) [58]. The effects of fibrinogen concentrate (60 mg/kg) and cryoprecipitate (10 ml/kg) were similarly demonstrated on FIBTEM-MCF as the median increase of 4 mm. The volume requirement to increase FIBTEM-MCF by 1–1.5 mm would be much higher with plasma (11–13 ml/ kg) as previously shown in the infants after CPB [16]. An exposure to multiple donors (e.g., multiparous female) in the pooled cryoprecipitate remains as a potential concern for serious immune responses including transfusion-related acute lung injury (TRALI) in adults [59]. It is thus important to diagnose intraoperative hypofibrinogenemia in real time, and to promptly replace fibrinogen in the case of bleeding. Thromboelastometry has been used in several fibrinogen replacement studies in cardiac surgery, and the following formula for fibrinogen (in grams) based on FIBTEM (A10, 10 min amplitude or MCF) has been suggested:

Dose = [target FIBTEM-A10 (mm) – current FIBTEM-A10 (mm)] x Weight (kg) / 140

For example, if the target FIBTEM-A10 is set at 10 mm, and current FIBTEM-A10 is 6 mm in a 70-kg patient with clinical bleeding, the dose of fibrinogen can be calculated to be:
Dose = (10 – 6) x 70 / 140 = 2 g

The dose of fibrinogen can be converted to the number of pooled units for cryoprecipitate by using a factor of 5:
Pools of cryoprecipitate = (10 – 6) x 70 / 140 x 5 = 10 U

There is no universal FIBTEM-A10 (or MCF) threshold for fibrinogen replacement that is applicable to all types of patients and cardiac surgical procedures. FIBTEM-MCF between 8 and 10 mm (A10, 5 to 7 mm) is a commonly used cut-off, but fibrinogen replacement to achieve as high as 22 mm has been considered in a few clinical trials (Table 1). It should be cautioned that a higher target may result in overdosing of fibrinogen in some cases in which bleeding is due to coagulation factor(s) other than fibrinogen. In complex cardiac surgical cases, reduced thrombin generation can simultaneously occur with hypofibrinogenemia, and bleeding may not be readily reversed by fibrinogen replacement alone [60,61]. In addition, it is important to consider the timing of fibrinogen replacement because reported positive results were primarily associated with early interventions, and not with a late use in refractory bleeding after conventional therapies [62,63].

There is a paucity of safety data on fibrinogen concentrate and cryoprecipitate in cardiac surgery because most clinical studies involving fibrinogen replacement therapies had not been powered for safety outcomes [10,23,58,61]. A recent propensity-score analysis on the adult cardiac surgical cases showed that the use of fibrinogen concentrate targeting fibrinogen level of 200 mg/dL was not associated with increased mortality, major cardiac events, or thromboembolic events within 1 year compared to the cohort without fibrinogen replacement [15].

Recombinant Activated Factor VII
Recombinant activated FVII (rFVIIa) is a synthetic coagulation factor concentrate that is indicated for the prevention and treatment of bleeding in patients with hemophilia with inhibitors. The enhanced formation of FXa by rFVIIa improves thrombin generation without depending on the FIXa-FXa complex (missing in hemophilia), which normally supports FXa generation. Hemostatic therapy using rFVIIa has become the mainstay bypassing therapy in hemophilia after it was shown to be effective with a relatively low risk of thrombosis [64]. Subsequently, rFVIIa has undergone a wide spread, off-label use in the US and elsewhere as a rescue hemostatic intervention in perioperative bleeding, especially after CPB [65].

Gill, et al. conducted a multi-center, prospective randomized study of the placebo vs. escalating doses of FVIIa (40 and 80 µg/kg) in postoperative cardiac surgery patients (n=172) with major bleeding (> 200 ml/hr or > 2 ml/kg/hr for 2 hrs). The majority of subjects who received rFVIIa underwent double or triple valve replacement procedures, and median CPB durations were 110–115 min. In terms of hemostatic efficacy, rFVIIa-treated groups had significantly fewer re-explorations compared to the placebo (12–14% vs. 25%; P=0.04) for bleeding, and received less allogeneic blood products. However, there was a trend for more frequent serious adverse events (12–14% vs. 7%; P=0.40) in the treatment groups, and in particular, the incidences of cerebral infarction associated with 40 and 80 µg/kg of rFVIIa treatment were 6% and 3%, respectively (none in the placebo). In the Gill, et al.’s study, actual dosing of rFVIIa took place after the median of 140–150 min in the ICU. As rFVIIa is typically used as a last resort for intractable bleeding after CPB, it might not be surprising that increased morbidity and mortality are associated with rFVIIa-treated cardiac surgical patients [66]. Off-label use of rFVIIa in cardiac surgery was also implicated in thromboembolic adverse events, especially in the arterial system [67]. Therefore, current guidelines on the cardiac surgical blood management recommend restricting the use of rFVIIa only in refractory microvascular bleeding after open heart surgery [68], or avoiding the off-label use of rFVIIa in the treatment of bleeding outside hemophilia indications [69].

Antifibrinolytics
Antifibrinolytic therapy is routinely used in cardiac surgery with CPB [8,70,71,72,73]. Lysine analogues, e-aminocaproic acid (EACA) and tranexamic acid (TXA), are two main antifibrinolytic agents used since the use of aprotinin, a direct plasmin inhibitor, was suspended in 2007 due to higher 30-day morbidity and mortality relative to lysine analogs in a Canadian multi-center study of antifibrinolytics in cardiac surgery (BART) [8]. This suspension was ultimately reversed by the Health Canada and the European Medicines Agency in 2011 after it became evident that inappropriate data uses skewed its conclusion in the BART study [74]. However, aprotinin is no longer used as an antifibrinolytic in most countries including the United States.

Hemostatic efficacies of EACA and TXA are highly dependent on fibrin formation. They prevent plasminogen binding to fibrin by occupying the lysine-binding site of plasminogen. Antifibrinolytic therapy is effective in reducing bleeding and PRBC transfusion compared to the placebo in cardiac surgical patients on aspirin [44]. It is presumed that aspirin therapy increases bleeding risks by inhibiting aggregation in cardiac and non-cardiac surgical patients [44,75], but a low-dose aspirin also induces a mild to moderate decrease in thrombin generation [76]. Increased clot susceptibility to fibrinolysis is induced by low thrombin generation [77,78], and antifibrinolytics presumably mitigate this effect in aspirin-treated patients. However, in the case of severe hemodilution such as DHCA cases, antifibrinolytic therapy alone may not be sufficient to control bleeding, and multi-modal hemostatic interventions to restore fibrin polymerization should be considered using viscoelastic testing [15,34].

Lysine analogues are considered to be relatively safe interventions, and EACA and TXA are not associated with an allergic reaction after repeated exposures due to their low molecular weights (131 Da and 157 Da, respectively). Systemic thrombosis is uncommon with EACA or TXA, but their use should be cautioned in patients with a history of thrombosis, or disseminated intravascular coagulation. Lysine analogues are mainly excreted from kidneys, and dose reduction should be considered particularly for TXA in the case of renal dysfunction [79]. Prolonged infusion of TXA is associated with epileptogenic effects [80]. This is presumably due to TXA crossing the blood-brain barrier and antagonizing GABA receptors [81], and glycine receptors [82].

Direct Oral Anticoagulants
Direct oral anticoagulants are increasingly used as an alternative to warfarin for the prevention of stroke due to non-valvular atrial fibrillation, and for the prevention and treatment of deep venous thrombosis and pulmonary thromboembolism [83]. DOACs include the direct thrombin inhibitor dabigatran etexilate (Pradaxa; Boehringer-Ingelheim, Ridgefield, CT) and direct FXa inhibitors, rivaroxaban (Xarelto; Janssen Pharmaceuticals, Titusville, NJ), apixaban (Eliquis; Bristol Myers Squibb, Princeton, NJ), and edoxaban (Savaysa; Daiichi Sankyo, Parsippany, NJ) (Table 3).

Until recently, lack of antidote was a major clinical issue associated in acute bleeding complications, and urgent/emergent surgical procedures. An anecdotal case report of a dabigatran-treated patient undergoing aortic valve replacement with a single vessel CABG highlights a potential risk of severe bleeding after CPB [84]. After massive transfusion including 26 units of PRBC, 22 units of plasma, 5 adult doses of platelets, and 50 units of cryoprecipitate in conjunction with multiple doses of rFVIIa totaling 21.6 mg, hemodialysis was started, bleeding was controlled over the next 4 hrs.

Novel Reversal Agents
The recent approval of idarucizumab (Praxbind, Boehringer-Ingelheim, Ridgefield, CT) by the US Food and Drug Administration (FDA) is a game changer in the management of acute bleeding or urgent invasive procedure/surgery in patients receiving dabigatran. Idarucizumab is a humanized Fab fragment of murine monoclonal antibody, which rapidly binds to dabigatran and neutralizes its anticoagulant effect. In the prospective cohort study, Pollack, et al. evaluated 39 patients (median age, 76; IQR 56–93) who received 5 g of intravenous idarucizumab before an urgent procedure. The normalizations of dilute thrombin time and ecarin clotting time were observed in 93% and 88%, respectively after the infusion of idarucizumab [86]. In 36 patients who underwent actual urgent procedures, normal hemostasis was confirmed in 33 (92%), but mild to moderate coagulation abnormality was found in 3 patients. Overall transfusion rates were 23.1% for both PRBC and plasma, and 5.1% for platelets among 39 patients. Only one patient developed deep vein thrombosis and pulmonary embolism within 72 hrs of idarucizumab, but 3 patients died due to cardiovascular causes within 24 hrs of treatment. These findings suggest that idarucizumab is an effective agent for acute dabigatran reversal, but the target population is aged, and inherently at high risk for complications related to vascular events, bleeding, infection, and coexisting diseases [86].

There are several ongoing trials of antidotes for anti-Xa agents as well, but none have been clinically approved (Table 3). Andexanet alfa (PRT064445; Portola Pharmaceuticals, South San Francisco, CA) is a recombinant factor Xa analogue which lacks procoagulant activity, but reversibly binds to an anti-Xa molecule with a half-life of about 1 hr. This agent is designated as a breakthrough therapy by the FDA, and currently in Phase II and III clinical trials. Aripazine (Ciraparantag, PER977; Perosphere, Inc., Danbury, CT) is a synthetic small molecule (512 Da) which presumably reverses dabigatran and anti-Xa agents as well as heparin and low molecular weight heparin. This agent is still in Phase I and II stages of clinical development.

Non-specific reversal agents including rFVIIa, PCC and activated PCC (factor eight bypassing agent [FEIBA]; Baxter, Westlake Village, CA) are considered in the case of life-threatening bleeding induced by DOAC [87,88], but clinical evidence for their efficacy and dosing in DOAC reversal is lacking.

Conclusion
Transfusion after cardiac surgery is performed after bleeding is clinically confirmed. Laboratory coagulation tests are conventionally of limited value in this setting because of a relatively long turn-around time. On the other hand, there is a paucity of evidence to support preemptive (prophylactic) transfusions of plasma and platelets, and there are even potential increases in morbidity and mortality associated with allogeneic products. Clinical evidence supporting the use of viscoelastic coagulation testing in cardiac surgery has increased in the past decade. Comprehensive coagulation testing toward the end of CPB, and early intervention(s) targeting fibrin polymerization are feasible using thromboelastometry, which allows rational uses of allogeneic and factor concentrate products.

Multi-modal approaches should be considered for blood conservation in increasingly aging and complex cardiac surgical patients. These efforts include limiting hemodilution during CPB, accepting lower hemoglobin thresholds, and optimizing a transition from preoperative antiplatelet and anticoagulant therapy. Future clinical trials need to test a combination of blood conservation efforts on clinical outcomes and costs in cardiac surgery.