Procalcitonin and C-Reactive Protein as Early Predictors of Anastomotic Leak in Colorectal Surgery: A Prospective Observational Study

BACKGROUND: Although the early diagnosis of anastomotic leak is a key point in reducing its clinical consequences, in daily practice, anastomotic leak diagnosis is often late. OBJECTIVE: The aim of this study was to determine whether procalcitonin and C-reactive protein are good predictors of anastomotic leak in colorectal surgery. DESIGN: This is a prospective observational study. SETTINGS: This study was conducted by a specialized colorectal multidisciplinary team of a tertiary teaching hospital. PATIENTS: A series of 205 consecutive patients who underwent elective colorectal surgery in a specialized unit was prospectively analyzed. The following data were collected: demographic, surgical, ASA class, POSUM, and morbidity. During the first 5 postoperative days, procalcitonin, C-reactive protein, leukocytes, platelets, and vital signs were evaluated daily. INTERVENTIONS: Daily assessment of clinical variable and serological data were conducted in the first 5 postoperative days. MAIN OUTCOME MEASURES: The primary outcome measure was the area under the curve at receiving operating characteristic curve analysis of the different variables in relation to the anastomotic leak. RESULTS: Anastomotic leak was detected in 17 (8.3%) patients; 11(5.4%) of the patients had a major anastomotic leak (need for drainage or reoperation). None of the variables evaluated were shown to be reliable in the early detection of anastomotic leak, considering both minor and major (maximum area under the curve <0.80). In contrast, when considering only major anastomotic leaks, procalcitonin and C-reactive protein were reliable predictors on postoperative days 3 to 5 (p < 0.0001, area under the curve >0.80). The best combination was procalcitonin at postoperative day 5 (area under the curve = 0.86), with a cutoff of 0.31 ng/mL, resulting in a 100% sensitivity, 72% specificity, 100% negative predictive value, and 17% positive predictive value. LIMITATIONS: Only symptomatic patients were investigated to rule out anastomotic leakage. CONCLUSIONS: Procalcitonin and C-reactive protein are both reliable predictors of major anastomotic leak after colorectal resection, although procalcitonin is more accurate. Raised procalcitonin and C-reactive protein serum concentration on postoperative days 3 to 5 renders necessary a careful evaluation of the patient before discharge.