Comparison of GLIM and PG-SGA for predicting clinical outcomes of patients with esophageal squamous carcinoma resection

Abstract

Objective: to determine the validity of the Global Leadership Initiative on Malnutrition (GLIM) against the Patient Generated-Subjective Global Assessment (PG-SGA) as a gold standard tool in malnutrition diagnosis, and to assess the impact of malnutrition diagnosed using GLIM and PG-SGA on the clinical outcomes of patients with esophageal squamous carcinoma (ESCC) resection. Methods: we prospectively analyzed 182 patients with ESCC who underwent radical esophagectomy at Zhongshan Hospital, Xiamen University, between October 2018 and December 2019. Preoperative malnutrition was diagnosed using GLIM and PG-SGA, and the postoperative clinical outcomes, including postoperative complications, postoperative chest tube indwelling time, length of stay and total hospitalization cost, were recorded. The association between the prevalence of malnutrition defined by the two tools and postoperative clinical outcomes was evaluated. Results: among the 182 ESCC patients, the incidence of malnutrition before surgery was 58.2 % and 48.4 % defined by PG-SGA and GLIM, respectively. GLIM and PG-SGA had good consistency in nutritional assessment of ESCC patients (k = 0.628, p < 0.001). Malnourished patients had higher TNM stages and older ages (all p < 0.05). Patients with malnutrition as assessed by PG-SGA and GLIM had a higher incidence of postoperative complications, a longer indwelling time of chest tube after esophagectomy, longer hospital length of stay, and higher hospitalization costs than patients with good nutrition (p < 0.001). Comparing the predictive efficiency of postoperative complications, the sensitivity of PG-SGA-and GLIM-defined malnutrition were 81.6 % and 79.6 %, the specificity were 50.4 % and 63.2 %, the Youden index were 0.320 and 0.428, and the Kappa value were 0.110 and 0.130, respectively. The areas under ROC curve of PG-SGA-and GLIM-defined malnutrition and postoperative complications were 0.660 and 0.714, respectively. Conclusions: this study indicates the effectiveness of malnutrition diagnosed according to GLIM and PG-SGA in predicting postoperative clinical outcomes among patients with ESCC. Compared with PG-SGA, GLIM criteria can better predict postoperative complications of ESCC. Follow-up analysis of postoperative long-term survival is needed to explore the association between different assessment tools and postoperative long-term clinical outcomes.