Aprotinin in cardiac surgery patients: is the risk worth the benefit?

BACKGROUND

Aprotinin is the only Food and Drug Administration-approved agent to reduce haemorrhage related to cardiac surgery and its safety and efficacy has been extensively studied. Our study sought to compare the efficacy, early and late mortality and major morbidity associated with aprotinin compared with e-aminocaproic acid (EACA) in cardiac surgery operations.

METHODS

Between January 2002 and December 2006, 2101 patients underwent coronary artery bypass grafting (CABG), valve surgery or CABG and valve surgery in our institution with the use of aprotinin (1898 patients) or EACA (203 patients). Logistic regression and propensity score analysis were used to adjust for imbalances in the patients' preoperative characteristics. The propensity score-adjusted sample included 570 patients who received aprotinin and 114 who received EACA (1-5 matching).

RESULTS

Operative mortality was higher in the aprotinin group in univariate (aprotinin 4.3% vs EACA 1%, p=0.023) but not propensity score-adjusted multivariate analysis (4% vs 0.9%, p=0.16). In propensity score-adjusted analysis, aprotinin was also associated with a lower rate of blood transfusion (38.8% vs 50%, p=0.04), a lower rate of haemorrhage-related re-exploration (3.7% vs 7.9%, p=0.04) and a higher risk of in-hospital cardiac arrest (3.7% vs 0%, p=0.03) and a marginally but not statistically significantly higher risk of acute renal failure (6.8% vs 2.6%, p=0.09). In Cox proportional hazards regression analysis, the risk of late death was higher in the aprotinin compared to EACA group (hazard ratio=4.33, 95% confidence interval (CI)=1.60-11.67, p=0.004).

CONCLUSION

Aprotinin decreases the rate of postoperative blood transfusion and haemorrhage-related re-exploration, but increases the risk of in-hospital cardiac arrest and late mortality after cardiac surgery when compared to EACA. Cumulative evidence suggests that the risk associated with aprotinin may not be worth the haemostatic benefit.