Bleeding after adult cardiac surgery: Clinical and genetic risk factors and effect of prophylactic treatment with tranexamic acid

Abstract

Background: Postoperative excessive bleeding is still one of the most common complications in cardiac surgery and is associated with increased morbidity and mortality. Major bleeding is a partially modifiable risk factor, and adequate preventive treatment strategies should be used to reduce the incidence of this complication. Prophylactic drug treatments can reduce postoperative bleeding. Prediction modelling strategies can be used for identifying patients at higher risk of bleeding after cardiac surgery and thereby contribute to implementation of prophylactic therapies and early interventions. Genetic variation may contribute to postoperative bleeding after cardiac surgery, and genetic information could make the prediction of excessive postoperative bleeding more precise. Aims: The overall aim of the thesis was to contribute to better patient care through increased knowledge about postoperative bleeding and about the treatment of this complication in high-risk groups. More specifically, the aim in Study I was to investigate the effects of tranexamic acid on postoperative blood loss and transfusion requirements in elderly patients undergoing complex cardiac surgery. In Study II the aims were 1) to investigate if risk prediction for postoperative bleeding was possible using clinical variables, by validating the Papworth Bleeding Risk Score and by developing local risk prediction models, and 2) to compare the usefulness of the universal definition of postoperative bleeding proposed by Dyke et al with the definition used in the Papworth Bleeding Risk Score. In Study III the main aim was to validate some earlier identified genetic polymorphisms associated with bleeding after cardiac surgery, and to explore other functional polymorphisms central in the coagulation and fibrinolysis systems or in plateletmembrane receptors related to postoperative bleeding. A secondary aim was to evaluate the additional predictive ability with inclusion of genetic information in risk prediction models based only on clinical variables, thereby investigating whether genetic susceptibility is a clinically relevant factor for further studies. Methods: A prospective, randomized, double-blind, placebo-controlled, parallel-group trial was used to investigate the effect of tranexamic acid in patients, 70 years or older, undergoing aortic valve replacement (AVR) or coronary artery bypass grafting (CABG). For Study II and III we used a prospectively recorded local database. All patients included in the studies underwent cardiac surgery with cardiopulmonary bypass at St. Olavs University Hospital, Trondheim, Norway. The predictive ability of the Papworth Bleeding Risk Score was validated, using the universal definition of perioperative bleeding proposed by Dyke et al and the definition used to develop the Papworth Bleeding Risk Score. Clinical prediction models were constructed using multivariate logistic regression analyses, using the two mentioned endpoints. Genetic associations were performed using allelic association tests and logistic regression. Multivariate logistic regression was used to evaluate whether the genetic information could improve the model with clinical variables. Results: The total number of red cell transfusions given was significantly larger in the placebo group of elderly patients compared with the tranexamic acid group (5 vs. 3 respectively, p = 0.049). When using both the Papworth Bleeding Risk Score and the locally developed risk prediction models, the positive predictive value was low (0.15 and 0.12 respectively) in the high-risk group and the negative predictive value was high (0.98 and 0.95 respectively) in the low-risk group. We identified five significant SNPs associated with postoperative bleeding. Addition of the genetic covariates to a logistic regression model with clinical variables significantly improved the model (p<0.01). Conclusions: Tranexamic acid reduced the need for red cell transfusion in elderly patients undergoing combined AVR and CABG surgery. Neither the Papworth Bleeding Risk Score nor the locally developed risk prediction models were suitable for identifying the patients at high risk of severe postoperative bleeding after cardiac surgery, but could more precisely identify patients at low risk of bleeding. None of the two endpoint definitions for postoperative bleeding were clearly preferable. We identified five SNPs associated with postoperative bleeding. Inclusion of genetic information to a risk prediction model with only clinical variables improved the model, indicating that genetic predisposition is relevant for postoperative bleeding after cardiac surgery.