Ho-Jung Shin1,2, Sang-Yong Son1,3, Bo Wang1, Chul Kyu Roh1,3, Hoon Hur1,3, Sang-Uk Han1,3. 1. Department of Surgery, Ajou University School of Medicine, Suwon, Korea. 2. Department of Transplantation and Vascular Surgery, Ajou University School of Medicine, Suwon, Korea. 3. Gastric Cancer Center, Ajou University School of Medicine, Suwon, Korea.
Abstract
OBJECTIVE: To compare long-term outcomes between robotic and LG approaches using propensity score weighting based on a generalized boosted method to control for selection bias. SUMMARY OF BACKGROUND DATA: Minimally invasive surgical approaches for GC are increasing, yet limited evidence exists for long-term outcomes of robotic gastrectomy (RG). METHODS: Patients (n = 2084) with GC stages I-III who underwent LG or RG between 2009 and 2017 were analyzed. Generalized boosted method was used to estimate a propensity score derived from all available preoperative characteristics. Long-term outcomes were compared using the adjusted Kaplan-Meier method and the weighted Cox proportional hazards regression model. RESULTS: After propensity score weighting, the population was balanced. Patients who underwent RG showed reduced blood loss (16 mL less, P = 0.025), sufficient lymph node harvest from the initial period, and no changes in surgical outcomes over time. With 52-month median follow-up, no difference was noted in 5-year overall survival in unweighted [91.5% in LG vs 94% in RG; hazard ratio (HR), 0.71; 95% confidence interval (CI), 0.46-1.1; P = 0.126] and weighted populations (94.2% in LG vs 93.2% in RG; HR, 0.88; 95% CI, 0.52-1.48; P = 0.636). There were no differences in 5-year recurrence-free survival (RFS), with unweighted 5-year RFS of 95.4% for LG and 95.2% for RG (HR, 0.95; 95% CI, 0.55-1.64; P = 0.845) and weighted 5-year RFS of 96.3% for LG and 95.3% for RG (HR, 1.24; 95% CI, 0.66-2.33; P = 0.498). CONCLUSIONS: After balancing covariates, RG demonstrated reliable surgical outcomes from the beginning. Long-term survival after RG and LG for GC was similar.
OBJECTIVE: To compare long-term outcomes between robotic and LG approaches using propensity score weighting based on a generalized boosted method to control for selection bias. SUMMARY OF BACKGROUND DATA: Minimally invasive surgical approaches for GC are increasing, yet limited evidence exists for long-term outcomes of robotic gastrectomy (RG). METHODS:Patients (n = 2084) with GC stages I-III who underwent LG or RG between 2009 and 2017 were analyzed. Generalized boosted method was used to estimate a propensity score derived from all available preoperative characteristics. Long-term outcomes were compared using the adjusted Kaplan-Meier method and the weighted Cox proportional hazards regression model. RESULTS: After propensity score weighting, the population was balanced. Patients who underwent RG showed reduced blood loss (16 mL less, P = 0.025), sufficient lymph node harvest from the initial period, and no changes in surgical outcomes over time. With 52-month median follow-up, no difference was noted in 5-year overall survival in unweighted [91.5% in LG vs 94% in RG; hazard ratio (HR), 0.71; 95% confidence interval (CI), 0.46-1.1; P = 0.126] and weighted populations (94.2% in LG vs 93.2% in RG; HR, 0.88; 95% CI, 0.52-1.48; P = 0.636). There were no differences in 5-year recurrence-free survival (RFS), with unweighted 5-year RFS of 95.4% for LG and 95.2% for RG (HR, 0.95; 95% CI, 0.55-1.64; P = 0.845) and weighted 5-year RFS of 96.3% for LG and 95.3% for RG (HR, 1.24; 95% CI, 0.66-2.33; P = 0.498). CONCLUSIONS: After balancing covariates, RG demonstrated reliable surgical outcomes from the beginning. Long-term survival after RG and LG for GC was similar.