Early Prediction of Deep Sternal Wound Infection after Heart Operations by Alpha-1 Acid Glycoprotein and C-reactive Protein Measurements

Abstract

Serum C-reactive protein (CRP) and α 1-acid glycoprotein (AAG) levels were studied in 188 patients undergoing heart operations with cardiopulmonary bypass. Mediastinitis or osteomyelitis of the sternum or both developed in 10 patients on postoperative day 4 to 13 (median, day 9). The mean CRP levels on day 2 were lower in patients with later deep sternal wound infection (9.1 ± 1.5 mg/dl [mean ± standard error]) compared with patients without major infections (14.0 ± 0.8 mg/dl; p = 0.103 [univariate logistic regression]). AAG levels on day 2 reacted in a similar manner, yielding 78.2 ± 5.5 mg/dl and 100.9 ± 2.7 mg/dl, respectively ( p = 0.0004). No correlation was found between CRP or AAG and duration of cardiopulmonary bypass, number of blood transfusions, or total protein levels on day 2. The white blood cell count (WBC) on day 2 was 13.1 ± 1.7 × 10 3/μl for patients with infection and 9.7 ± 0.3 for those without infection. Multivariate logistic regression analysis revealed that AAG, WBC, and CRP on day 2 were significant risk factors sufficiently predicting the probability of a deep sternal infection. After adjustment for these three variables, other variables (age, sex, total protein on day 2, diabetes mellitus, type of operation, duration of cardiopulmonary bypass, length of operation, repeat thoracotomy for bleeding, number of blood transfusions on the day of operation, intraaortic balloon pumping, reoperation, emergency operation, and surgeon's professional status) were not of additional significance. The goodness of fit of the statistical model was confirmed by a high correspondence between predicted and observed cases of deep sternal infection. The classification analysis revealed a correct prediction of 90% of patients with infection when 85.5% of the patients without infection were correctly classified. A defect in the trigger mechanism of the acute-phase response or increased degradation by proteases are discussed as possible causes of the lowered AAG and CRP levels. Early identification of patients at risk for postoperative infection may be a new application of acute-phase protein measurements in clinical practice.