Although complications from cardiopulmonary bypass (CPB) have decreased, excessive perioperative bleeding is still common (1). Such bleeding complications are difficult to characterize because of the complexity of the hemostatic process and the technical and logistical difficulties imposed by operative procedures.
The transfusion of blood and hemostatic blood components remains largely empiric, with considerable variation among institutions (2). The turnaround times of laboratory tests in patients with excessive bleeding after CPB are too long to facilitate targeted blood component therapy. Numerous methods are now available for a rapid, point-of-care assessment of coagulation in the operating room (3–6). The predictive value of the tests for bleeding in cardiac surgery is not clear. With the aid of these rapid point-of-care tests, one should be able to differentiate between a surgical cause for abnormal bleeding and a hemostatic disorder after CPB. An algorithm-based targeted therapy using measurements of point-of-care tests can reduce transfusion of blood products (7–9).
Thromboelastography has been used in several studies to predict blood loss in cardiac surgery. The results from these studies are far from uniform (10–15). Modified computerized thrombelastography (ROTEG™) with activation of coagulation provides results more rapidly than conventional thromboelastography. The traces and calculated variables can be stored for further analysis or research on the attached computer. The ROTEG™ system allows rapid whole blood coagulation testing with different activators of additives reaching the maximal amplitude (MA) typically after 15–20 min. In contrast to the classical thromboelastography method, the probe-containing cup is stationary and the pin is moving, fixed on a rotating shaft, which is guided by a ball bearing system. When a clot is formed inside the cup, it hinders the movement of the pin and the data are processed in real time by the software. Because of the method of activation and the different technique, the results are different than those of the classic thromboelastography, but results between the two methods correlate well. The PFA-100™ (platelet function analyzer) detects impairment of platelet function in cardiac surgical patients taking aspirin preoperatively (16). Slaughter et al. (17) found the PFA-100™ to be useful in reducing unnecessary platelet transfusion after CPB.
In the present prospective and open study, our aim was to assess the predictive values of the two point-of-care tests, ROTEG™ and PFA-100™ platelet function analyzer, alone and in combination, on postoperative bleeding after routine cardiac surgery. To distinguish the influence of preexisting and CPB-acquired coagulopathies, measurements were performed before, during, and after cessation of CPB.
After obtaining ethics committee approval of the Technical University of Munich, Germany, and informed written consent, consecutive patients scheduled for routine cardiac surgical procedures involving CPB were prospectively studied. Our exclusion criteria were emergency operation and missing consent. Preoperatively, all patients were asked about intake of aspirin, ticlopidine, clopidogrel, GPIIb/IIIa inhibitors, or Coumadin.
Opioid-based anesthesia with sufentanil, midazolam, and pancuronium was used in all patients. Monitoring included electrocardiogram, arterial pressure measurement, and pulmonary artery thermodilution catheter. Before the start of CPB, mucosa-heparin was administered in a dose of 375 U/kg body weight. If required, further 125 U/kg were given to maintain activated clotting time (ACT) longer than 480 s during CPB. Based on our previous results (18,19), celite ACT was used to control anticoagulation. Membrane oxygenators (Compactflo, Module 7500; Dideco, Mirandola, Italy) were solely used. A large-dose aprotinin regime (4–6 × 106 KIU per patient) is part of our routine protocol for CPB. After discontinuation of CPB, protamine was administered 1:1. Anesthesiologists and surgeons were blinded to the results of the thromboelastography and PFA measurements. The same person (TR) performed all measurements. Additional protamine or blood products were given based on our clinical guidelines. Blood loss was registered for the first 6, 12, and 24 h postoperatively. Allogeneic blood products used intra- and postoperatively were recorded. The duration of operation, of CPB, and the time interval between the end of CPB and closure of the chest as an expression for the time needed to correct surgical bleeding were recorded.
Because it is known that arterial and venous thromboelastography variables may differ significantly (20), blood samples were exclusively obtained from a radial artery catheter (20 gauge; Arrow, Reading, PA) at three time points: directly after the induction of anesthesia, during CPB after rewarming, and 15 min after protamine administration. For the ROTEG™ and PFA-100™ analyses, we first removed 10 mL of waste blood, which exceeds three times the dead-space volume of the arterial catheter pressure tubing. Blood samples were collected in 3.8-mL tubes containing 0.129 moL/L buffered (pH 5.5) sodium citrate (Sarstedt, Nuermbrecht, Germany). thromboelastography and PFA analyses were performed immediately afterward.
Thromboelastography analysis was performed with the ROTEG™ Coagulation Analyzer (Pentapharm, Munich, Germany). Two different thromboelastography assays were used: 1. activated whole blood thrombelastography—activation of the intrinsic system with 20 μL of Nobi Clot Start thromboelastography (buffered calcium chloride; Nobis, Endingen, Germany), 20 μL of Nobi Clot InTEG-LS Activator (contact activator for the intrinsic system, ellagic acid, and phospholipids; Nobis), and 300 μL of whole blood; and 2. activated whole blood thrombelastography with abciximab—activation of the intrinsic system as in 1 plus 10 μL of ReoPRO® (abciximab; Centocor B.V., Leiden, Netherlands). Every thromboelastography analysis was observed for 60 min and stopped afterward. To eliminate heparin influence, thromboelastography assays containing heparinase (Dade Hepzyme; Behring IBEX Technologies, Marburg, Germany) were used during CPB. Because the values of activated ROTEG™ are different from conventional thromboelastography and only sparse data from healthy volunteers exist, the 10th and 90th percentiles of the preoperative values of our patients not treated with anticoagulants were used as substitute for reference values for ROTEG™.
The PFA test was performed with the PFA-100™ platelet function analyzer (Dade Behring, Schwalbach, Germany). PFA testing was done preoperatively and after protamine simultaneously with the thromboelastography analyses. The PFA-100™ is a device for the simulation of primary hemostasis in vitro. The analyzer provides a viscoelastic measurement. Prolonged closure times can be expected in the majority of samples from patients treated with aspirin when using the epinephrine-coated cartridge but not with the adenosine diphosphate (ADP)-coated cartridge (21). Hematocrit and routine coagulation tests as platelet count, partial thromboplastin time, thromboplastin time, and fibrinogen concentration were also performed by using a whole blood analyzer (Sysmex SE 9000; TOA Medical Electronics, Kobe, Japan) and an automatic coagulation analyzer (CA 6000; TOA Medical Electronics).
Abnormal bleeding was defined in two ways. Based on the literature (22), we defined an “objective” bleeding threshold of 750 mL 6 h postoperatively as increased bleeding tendency. Additionally, to obtain a relation to our own patient population, blood loss exceeding the 75th percentile of our patient population was categorized as enhanced bleeding. The Mann-Whitney U-test was used for statistical analysis, with P < 0.05 indicating statistical significance. Binary logistic regression analysis was done additionally for identifying variables significantly correlated with abnormal bleeding. Receiver operating characteristic curves were drawn to visualize and compare the best predictors (SPSS for Windows 10.0; SPSS Inc., Chicago, IL).
Two hundred fifty-five patients were examined. Three patients (1.2%) had reexploration for major surgical bleeding. These three patients did not have pathologic thromboelastography values and were included for further analysis. Patient demographics, blood loss, and transfusion requirements are given in Table 1. Fifty-nine percent of our patients did not receive allogeneic blood transfusion during the hospital stay; 9% of all patients received hemostatic blood products such as fresh frozen plasma and/or platelets. Blood loss ≥750 mL occurred for 14 patients (5.5%). All their thromboelastography values were not significantly different compared with those with blood loss <750 mL. The PFA test showed a significant difference for preoperative ADP (202 ± 91 versus 132 ± 63 s), postoperative ADP (182 ± 107 versus 147 ± 72 s), and epinephrine (274 ± 35 versus 236 ± 58 s), respectively (P < 0.05).
The 75th percentile of the blood loss in our patient population was 500 mL for 6 h postoperatively. By definition, 69 patients (25%) demonstrated a blood loss ≥500 mL. The thromboelastography and PFA values for all patients and for the groups with blood loss less than or ≥500 mL are listed in Table 2. The routine coagulation variables were not significantly different except for fibrinogen concentration after CPB (218 ± 14 versus 251 ± 92 mg/dL, P < 0.05). Duration of CPB or operation did not differ significantly for both groups.
Solely the measurements after CPB showed significance for abnormal blood loss in all tests (Table 2). All variables, which were significant in the univariate analysis, were then compared with binary logistic regression analysis. This demonstrates angle α after CPB as the best predictor for increased blood loss. A second step revealed the PFA-ADP test after CPB as an additional significant factor (Table 3). The receiver operating characteristic curves (Fig. 1) visualize these results. The area under the curve of the combination of PFA-ADP and angle α after CPB (0.74) is larger than the area under the curve of PFA-ADP (0.66) and angle α (0.69) alone in predicting blood loss ≥500 mL. Abciximab MA after CPB was also significant for abnormal blood loss, which correlates with the fibrinogen measured at the same time (r = 0.63).
The scattergram of angle α versus blood loss (Fig. 2) shows many true negative results. The positive and negative predictive values were calculated. The predictive values for all variables obtained after CPB were comparable, showing a low positive and a high negative predictive value (Table 4). Patients with false-positive results (impaired thromboelastography but without increased blood loss) received neither more blood transfusions or hemostatic components nor was the time to control surgical bleeding longer in them than in patients with true positive results (impaired thromboelastography and increased blood loss).
Patients treated with aspirin within 5 days before operation (n = 104) had significantly more blood loss than those who did not take any anticoagulants preoperatively (412 ± 220 versus 353 ± 224 mL;P < 0.05). Patients with blood loss ≥500 mL and preoperative aspirin did not have different thromboelastography values preoperatively. These patients had different PFA values with PFA-epinephrine 247 ± 72 versus 212 ± 71 s (aspirin and blood loss <500 mL) (P < 0.05) preoperatively and 260 ± 44 versus 225 ± 61 s postoperatively (P < 0.05). The positive predictive value of preoperative PFA-epinephrine for patients taking aspirin was 46%, the negative predictive value 77%; the analogous values for postoperative PFA-epinephrine were 45% and 85%, respectively.
The present study demonstrated that the ROTEG™ variables MA, angle α with and without abciximab, and the PFA tests after cessation of CPB show significant differences between patients who will have normal versus abnormal blood loss in the postoperative period. We defined increased blood loss in two ways: first, we used a general accepted cut-off of 750 mL of blood loss within the first 6 hours postoperatively (22) and, second, we defined in the study protocol the upper quartile of blood loss in our patient population as increased bleeding tendency. This method provides a predictable number of “diseased” patients. Of the variables measured with activated computerized thromboelastography, angle α after CPB and, of the PFA variables, PFA-ADP after CPB, were the best predictors of abnormal blood loss. However, the low positive predictive value indicates that the coagulopathy identified by the test system does not inevitably cause increased bleeding after CPB. The high negative predictive value identifies patients who will tend to not have abnormal bleeding caused by disturbance of hemostasis. This was demonstrated in our study by the three patients with obvious surgical bleeding identified during repeat thoracotomy who showed normal thromboelastography and PFA test results.
Many studies have tried to find a predictive value of thromboelastography for abnormal bleeding after CPB. Spiess et al. (12) presented the first study concerning thromboelastography and blood loss after CPB. In a group of 38 patients, they found that thromboelastography was a significantly better predictor of postoperative hemorrhage and need for reoperation than the ACT or routine coagulation profile. In 42 high-risk patients, Tuman et al. (13) found thromboelastography 100% accurate in predicting bleeding after CPB. In 36 patients, Essell et al. (10) found an abnormal thromboelastography representing an increased risk for hemorrhage. However, these three studies were done in small numbers of patients.
Thromboelastography failed to predict blood loss in other studies. Data from the study by Wang et al. (14) of 101 patients indicated no correlation between the amount of postoperative chest tube drainage and thromboelastography variables. There was a frequent incidence of false-negative thromboelastography in patients who had excessive hemorrhage despite normal thromboelastography. The results of Dorman et al. (15) in 60 patients showed that all components of the thromboelastography failed to predict intraoperative blood loss. Nutall et al. (11) found no correlation between thromboelastography done after CPB and 24-hour blood loss after CPB in 82 patients.
The scatterplots of our results indicate that the relationship between bleeding and the thromboelastography variables, although statistically significant, had little positive predictive value and were mainly significant as a result of the large number of patients in each group, which permitted weak correlations to reach statistical significance (23).
Of particular interest is the significant result of the postoperative abciximab MA. The addition of abciximab to the sample eliminates platelet function completely. Kettner et al. (24) showed a high correlation of abciximab MA with fibrinogen. We were able to confirm this observation. The role of fibrinogen in hemostasis after CPB is not well examined. There is overall consensus regarding the dominant role of platelets in hemostasis after CPB. Probably the most important aspect is the correct interaction of all coagulation components. Modified thromboelastography seems to be ideal for monitoring this interplay. Corresponding to our results, Slaughter et al. (17) found for the ADP-PFA test after CPB a small positive predictive and large negative predictive value (18% and 96%) for postoperative bleeding in 76 cardiac surgical patients.
Our preoperative measurements with thromboelastography and PFA showed less prognostic value than the measurement after CPB in general. The patients taking aspirin preoperatively and with increased blood loss postoperatively had significantly prolonged PFA-epinephrine tests pre- and postoperatively. However, the positive and negative predictive values were in a comparable range to the total patient population.
There are four studies stating a positive effect of thromboelastography-guided transfusion practice for cardiac surgery patients (8,9,25,26). All studies demonstrated fewer transfusions in the thromboelastography-guided group than in the control group. The data of these four investigations support the use of thromboelastography in an algorithm to guide transfusion therapy in complex cardiac surgery. The reduction in transfusions is attributed to improved hemostasis in those patients who had earlier and specific identification of hemostatic abnormalities and consequently received more appropriate intraoperative hemostatic therapy.
The transfusion frequency in the control groups in the studies mentioned is relatively high with 30%–66% of the patients being transfused with hemostatic blood products. In contrast, only 9% of our patients (including reoperations) received allogeneic plasma and/or platelets during the hospital stay. One reason for this may be our routine use of a large aprotinin regime. Restriction of transfusion was not a goal of our study, however.
The results of thromboelastography are dependent on the anticoagulation of the sample with or without citrate and on the duration of storage (27). Camenzind et al. (27) demonstrated that thromboelastography blood coagulation variables in recalcified blood differ from those in native blood and change significantly during 30–60 minutes of storage. For a point-of-care test, the period of interest is between 0 and 15 minutes from sampling. Therefore, we did our measurements in the operating room within five minutes of sampling.
Heparinase itself can influence the thromboelastography measurement and may have affected the results during CPB. Spiess et al. (28) demonstrated higher values for angle α and MA in heparinase-treated heparinized blood than in protamine-treated blood.
Activated thromboelastography with ROTEG™ shows different reference values than conventional thromboelastography. Activation fastens and strengthens clot formation. Values for angle α and MA are higher than in conventional thromboelastography. Therefore, we applied the preoperative measurements of the patients who did not take any anticoagulants preoperatively as reference values. This affects only the calculations of positive and negative predictive values.
Because our observational study protocol was not controlled, our transfusion practice may have influenced postoperative blood loss. However, patients with impaired thromboelastography without increased blood loss did not receive more transfusions or hemostatic drugs.
The use of large-dose aprotinin is possibly responsible for the small prevalence of “disease” in our study. The definition of “excessive bleeding” is not as large as in studies without the treatment with aprotinin. Aprotinin also exerts influence on the activator and prolongs the activated partial thromboplastin time (29), but the clinical relevance of this finding is not known. Because in our study all patients received aprotinin, the use of different activators could not have caused a systematic bias.
In our series of 255 routine cardiac surgical patients, both point-of-care tests, thromboelastography with ROTEG™ and PFA test, showed a large negative predictive value concerning abnormal bleeding after CPB. This is also true for the subgroup of patients taking antiplatelet medication. The largest prognostic value was obtained with the measurements done after the cessation of CPB. thromboelastography allows a better prediction than the PFA test. We conclude that if a patient with a “normal” thromboelastography or PFA test after CPB is bleeding, this is suggestive of a surgical cause, and surgical intervention should be considered early.
Because of the multiple factors influencing blood loss after cardiac surgery, a positive predictive value as large as the negative predictive value cannot be expected. The small positive predictive value with many false-positive results emphasizes that not every hemostatic defect inevitably causes abnormal bleeding after CPB. However, if a pathologic thromboelastography or PFA test coincides with clinical bleeding, this is suggestive for hemostatic bleeding and should be treated to correct the coagulopathy.
We are grateful for the skilled statistical support for the regression analysis provided by K. Ulm, Professor at the Institute for Medical Statistics and Epidemiology of the Technical University, Munich, Germany.
1. Levi M, Cromheecke ME, de Jonge E, et al. Pharmacological strategies to decrease excessive blood loss in cardiac surgery: a meta-analysis of clinically relevant endpoints. Lancet 1999; 354: 1940–7.
2. Stover EP, Siegel LC, Parks R, et al. Variability in transfusion practice for coronary artery bypass surgery persists despite national consensus guidelines: a 24-institution study. Institutions of the Multicenter Study of Perioperative Ischemia Research Group. Anesthesiology 1998; 88: 327–33.
3. Despotis GJ, Levine V, Filos KS, et al. Evaluation of a new point-of-care test that measures PAF-mediated acceleration of coagulation in cardiac surgical patients. Anesthesiology 1996; 85: 1311–23.
4. Despotis GJ, Levine V, Saleem R, et al. Use of point-of-care test in identification of patients who can benefit from desmopressin during cardiac surgery: a randomised controlled trial. Lancet 1999; 354: 106–10.
5. Ereth MH, Nuttall GA, Santrach PJ, et al. The relation between the platelet-activated clotting test (HemoSTATUS) and blood loss after cardiopulmonary bypass. Anesthesiology 1998; 88: 962–9.
6. Wallock M, Jeske WP, Bakhos M, Walenga JM. Evaluation of a new point of care heparin test for cardiopulmonary bypass: the TAS heparin management test. Perfusion 2001; 16: 147–53.
7. Despotis GJ, Santoro SA, Spitznagel E, et al. Prospective evaluation and clinical utility of on-site monitoring of coagulation in patients undergoing cardiac operation. J Thorac Cardiovasc Surg 1994; 107: 271–9.
8. Royston D, von Kier S. Reduced haemostatic factor transfusion using heparinase-modified thrombelastography during cardiopulmonary bypass. Br J Anaesth 2001; 86: 575–8.
9. Shore-Lesserson L, Manspeizer HE, DePerio M, et al. Thromboelastography-guided transfusion algorithm reduces transfusions in complex cardiac surgery. Anesth Analg 1999; 88: 312–9.
10. Essell JH, Martin TJ, Salinas J, et al. Comparison of thromboelastography to bleeding time and standard coagulation tests in patients after cardiopulmonary bypass. J Cardiothorac Vasc Anesth 1993; 7: 410–5.
11. Nuttall GA, Oliver WC, Ereth MH, Santrach PJ. Coagulation tests predict bleeding after cardiopulmonary bypass. J Cardiothorac Vasc Anesth 1997; 11: 815–23.
12. Spiess BD, Tuman KJ, McCarthy RJ, et al. Thromboelastography as an indicator of post-cardiopulmonary bypass coagulopathies. J Clin Monit 1987; 3: 25–30.
13. Tuman KJ, Spiess BD, McCarthy RJ, Ivankovich AD. Comparison of viscoelastic measures of coagulation after cardiopulmonary bypass. Anesth Analg 1989; 69: 69–75.
14. Wang JS, Lin CY, Hung WT, et al. Thromboelastogram fails to predict postoperative hemorrhage in cardiac patients. Ann Thorac Surg 1992; 53: 435–9.
15. Dorman BH, Spinale FG, Bailey MK, et al. Identification of patients at risk for excessive blood loss during coronary artery bypass surgery: thromboelastography versus coagulation screen. Anesth Analg 1993; 76: 694–700.
16. Gibbs NM, Weightman WM, Thackray NM, et al. The effects of recent aspirin ingestion on platelet function in cardiac surgical patients. J Cardiothorac Vasc Anesth 2001; 15: 55–9.
17. Slaughter TF, Sreeram G, Sharma AD, et al. Reversible shear-mediated platelet dysfunction during cardiac surgery as assessed by the PFA-100 platelet function analyzer. Blood Coagul Fibrinolysis 2001; 12: 85–93.
18. Dietrich W, Dilthey G, Spannagl M, et al. Influence of high-dose aprotinin on anticoagulation, heparin requirement, and celite- and kaolin-activated clotting time in heparin-pretreated patients undergoing open-heart surgery: a double-blind, placebo-controlled study. Anesthesiology 1995; 83: 679–89.
19. Dietrich W, Jochum M. Effect of celite and kaolin on activated clotting time in the presence of aprotinin: activated clotting time is reduced by binding of aprotinin to kaolin. J Thorac Cardiovasc Surg 1995; 109: 177–8.
20. Manspeizer HE, Imai M, Frumento RJ, et al. Arterial and venous thrombelastograph variables differ during cardiac surgery. Anesth Analg 2001; 93: 277–81.
21. Mammen EF, Comp PC, Gosselin R, et al. PFA-100 system: a new method for assessment of platelet dysfunction. Semin Thromb Hemost 1998; 24: 195–202.
22. Karski JM, Teasdale SJ, Norman P, et al. Prevention of bleeding after cardiopulmonary by-pass with high-dose tranexamic acid: double-blind, randomized clinical trial. J Thorac Cardiovasc Surg 1995; 110: 835–42.
23. Whitten CW, Greilich PE. Thromboelastography: past, present, and future. Anesthesiology 2000; 92: 1223–5.
24. Kettner SC, Panzer OP, Kozek SA, et al. Use of abciximab-modified thrombelastography in patients undergoing cardiac surgery. Anesth Analg 1999; 89: 580–4.
25. Mongan PD, Hosking MP. The role of desmopressin acetate in patients undergoing coronary artery bypass surgery: a controlled clinical trial with thromboelastographic risk stratification. Anesthesiology 1992; 77: 38–46.
26. Spiess BD, Gillies BS, Chandler W, Verrier E. Changes in transfusion therapy and reexploration rate after institution of a blood management program in cardiac surgical patients. J Cardiothorac Vasc Anesth 1995; 9: 168–73.
27. Camenzind V, Bombeli T, Seifert B, et al. Citrate storage affects thrombelastograph analysis. Anesthesiology 2000; 92: 1242–9.
28. Spiess BD, Wall MH, Gillies BS, et al. A comparison of thromboelastography with heparinase or protamine sulfate added in vitro
during heparinized cardiopulmonary bypass. Thromb Haemost 1997; 78: 820–6.
29. Francis J, Howard C. The effects of aprotinin on the response of the activated partial thromboplastin time (APTT) to heparin. Blood Coagul Fibrinolysis 1993; 4: 35–40.