Usefulness of procalcitonin in the early detection of infection after thoracic surgery

Objective: The twofold aim of this prospective clinical study was to assess the accuracy of procalcitonin as a marker of postoperative infection after thoracic surgery and to compare it with C-reactive protein. Methods: Procalcitonin and C-reactive protein concentrations, clinical symptoms of infection and systemic inflammation were recorded preoperatively and 5 days postoperatively in 157 patients undergoing the following procedures: 52 wedge resections, 28 pneumonectomies and 77 lobectomies (or bilobectomies). Patients were classified as noninfected or infected according to predefined criteria. Results: In non-infected patients (n = 132), procalcitonin peaked on day 1 and C-reactive protein, on day 2. The procalcitonin value was significantly higher in patients having undergone a pneumonectomy (0.73 +/- 0.78 versus 0.54 +/- 0.25 ng/mL for lobectomy and 0.50 +/- 0.35 ng/mL for wedge resection; P = 0.04). The mean value of procalcitonin was significantly higher in patients with postoperative infection (n = 25) than in those with no postoperative infection (3.6 +/- 5.5 versus 0.63 +/- 0.62 ng/mL; P = 0.0001). The onset of infection most frequently occurred on postoperative day 2 (43% of patients); maximum procalcitonin and C-reactive protein concentrations most frequently appeared on postoperative day 1 (56% of patients) and day 2 (63% of patients), respectively. The best cutoff value for detection of infection with procalcitonin was 1 ng/mL and with C-reactive protein, 100 mg/L. Comparing the area under the Receiver Operating Characteristic curves, procalcitonin was better than C-reactive protein for detecting postoperative infection (0.92 versus 0.66; P < 0.0001). Conclusions: Procalcitonin can be used as a reliable diagnostic parameter to detect and to monitor infectious complications in the postoperative period after thoracic surgery, especially in patients felt to be at higher risk (SIRS). It provides more information about the course of the disease than C-reactive protein does, and can be detected before the occurrence of clinical infection. (c) 2005 Elsevier B.V. All rights reserved.