Inflammation and Risk Prediction of Major Complications Following Cardiac Surgery

Abstract

Background In cardiac surgery improvements in quality of care have been a major focus for several decades. Statistical models for predicting the risk of operative mortality and other adverse outcomes have been an important part of this work. Risk prediction models may be used to inform patients of the risk of the planned operation, to adjust complication rates to enable a valid comparison between institutions or surgeons, and to identify potential fields of improvement. Risk prediction models were often less accurate when applied to other patient populations than the one they were derived from. No risk prediction models for mortality or acute kidney injury following cardiac surgery had been developed using data from Norwegian patients. The surgical trauma and the artificial surfaces in the cardiopulmonary bypass circuit evoke a systemic inflammatory response. The inflammatory response includes activation of leukocytes, endothelium, and plasma cascade systems including the complement system, coagulation and fibrinolysis. Inflammation is thought to play a pivotal role in the development of major complications following cardiac surgery. Cardiac dysfunction following open-heart surgery is a clinical syndrome where reduced cardiac output results in insufficient oxygen delivery to the tissues. It is thought to be induced by ischaemia and reperfusion, as well as inflammation, increasing the myocardial oxidative stress. Several inflammatory biomarkers, including C-reactive protein (general marker of inflammation), lactoferrin (neutrophil activation marker), neopterin (monocyte/macrophage activation marker) and the terminal complement complex (complement activation marker), had previously been associated with adverse cardiac outcomes in ischaemic heart disease. Aims One aim was to develop local risk prediction models for operative mortality and acute kidney injury following cardiac surgery. Another aim was to investigate whether increased preoperative inflammation was associated with the development of cardiac dysfunction following open-heart surgery. Methods For developing the risk prediction models for mortality and acute kidney injury we included all 5029 adult patients who underwent open-heart surgery at St. Olavs University Hospital, Trondheim from 2000 through 2007. We applied multivariable logistic regression for model development, and the models were internally validated using bootstrapping methods. For investigating whether increased preoperative inflammation was associated with cardiac dysfunction following open-heart surgery we included 1018 consecutive patients who underwent open-heart surgery at St. Olavs University Hospital, Trondheim, Norway from 1 April 2008 to 19 April 2010. We applied enzyme immunoassay to measure the preoperative concentration of C-reactive protein, lactoferrin, neopterin and the terminal complement complex in plasma. Logistic regression was used for the statistical analysis, and we adjusted for clinical variables previously associated with postoperative cardiac dysfunction. Results and discussion The mortality risk prediction model consisted of eight preoperative variables easily obtainable in clinical practice: Age, degree of urgency for surgery, female gender, serum creatinine concentration, chronic pulmonary disease, chronic cardiac insufficiency, previous cardiac surgery, and type of operation. The acute kidney injury risk model included eleven easily available preoperative variables: age, body mass index, lipid lowering treatment (protective effect), hypertension, peripheral vascular disease, chronic pulmonary disease, haemoglobin concentration, serum creatinine concentration, previous cardiac surgery, emergency operation, and operation type. Both the mortality and the acute kidney injury risk models displayed good discrimination and calibration in our population. We found that neopterin was associated with cardiac dysfunction after cardiac surgery, and this association remained significant also after adjustment for clinical variables associated with postoperative cardiac dysfunction. Conclusions Our local preoperative risk models predicted mortality and acute kidney injury accurately, and were generally robust. Our findings regarding neopterin and cardiac dysfunction support the hypothesis of the role of inflammation and oxidative stress in the development of postoperative cardiac dysfunction.