Systemic Inflammation Score as a Predictor of Pneumonia after Radical Resection of Gastric Cancer: Analysis of a Multi-Institutional Dataset
Michita Shoka, Mitsuro Kanda, Seiji Ito, Yoshinari Mochizuki, Hitoshi Teramoto, Kiyoshi Ishigure, Toshifumi Murai, Takahiro Asada, Akiharu Ishiyama, Hidenobu Matsushita, Chie Tanaka, Daisuke Kobayashi, Michitaka Fujiwara, Kenta Murotani, Yasuhiro Kodera
Abstract
BACKGROUND: Curative treatment for gastric cancer (GC) comprising gastrectomy with systematic lymph node dissection can result in postoperative complications. Postoperative pneumonia is sometimes fatal, like surgery-related complications such as anastomotic leakage. In this retrospective study, we analyzed a multi-institutional collaborative dataset with the aim of identifying predictors of postgastrectomy pneumonia. METHODS: From a retrospective database of 3,484 patients who had undergone gastrectomy for GC at nine Japanese institutions between 2010 and 2014, 1,415 patients who met all eligibility criteria were identified as eligible for analysis. Predictive values of 31 candidate variables for postoperative pneumonia were assessed. RESULTS: Forty-two patients (3.0%) had grade II or higher postoperative pneumonia. Preoperative systemic inflammation score (SIS) had the greatest area under the curve (0.655) for predicting postoperative pneumonia (optimal cutoff value = 2). The odds ratio (OR) of high SISs associated with postoperative pneumonia was 3.10 (95% confidence interval [CI], 1.54-6.07; p < 0.001). Multivariate binomial logistic analysis identified high SIS as an independent risk factor for postoperative pneumonia (OR, 2.31; 95% CI, 1.19-4.48; p = 0.013). A forest plot revealed that ORs of high SISs were highest in female patients. CONCLUSIONS: Our findings indicate that the preoperative SIS may serve as a simple predictor of postgastrectomy pneumonia, assisting physicians' efforts to take preventive measures against this complication.