The Predictive Value of Inflammatory Biomarkers in Esophageal Anastomotic Leaks

BACKGROUND This study investigated the role of the neutrophil-to-lymphocyte ratio (NLR) in predicting anastomotic leak (AL) after esophagectomy for esophageal adenocarcinoma. METHODS This is a retrospective analysis of a prospectively collected database from the McGill University Health Center. Included were all patients with esophageal adenocarcinoma who underwent esophagectomy between 2005 and 2016. Patients with chronic infections, concurrent active malignancies, and autoimmune conditions were excluded. NLR values were obtained on postoperative days (POD) 0, 1, 2, and 3. Receiver operating characteristic curve study and multivariable logistic analysis were conducted to evaluate the diagnostic value of NLR. RESULTS The study included 330 patients, and AL developed in 16%. Mean NLR values on POD1, 2, and 3 were higher in patients with leaks (20 vs 14 on POD1, P < .001; 20 vs 12 on POD2, P < .001; and 19 vs 10 on POD3, P < .001). The NLR value on POD3 was associated with an area under the curve of 70% and a negative predictive value of 92.4%. Multi-variable analyses identified higher American Society of Anesthesiologists Physical Status Classification, increasing NLR trend (between POD1 and POD3), POD1 NLR, POD2 NLR, and POD3 NLR as independent factors associated with AL. CONCLUSIONS Patients who developed AL demonstrate higher mean NLR values in the early postoperative period with rising trends. Conversely a low NLR is associated with a high negative predictive value for AL. This simple metric allows risk stratification that may guide treatment decisions in esophagectomy patients. (Ann Thorac Surg 2021;112:1790-6) (c) 2021 by The Society of Thoracic Surgeons

Automated summary

The study found that patients who developed anastomotic leak after esophagectomy for esophageal adenocarcinoma had higher neutrophil-to-lymphocyte ratios in the early postoperative period with rising trends, while a low ratio was associated with a high negative predictive value.