Short-Term Surgical Outcomes of Palliative Gastrectomy According to the Extent of Lymph Node Dissection: A Nationwide Propensity Score-Weighted Cohort Analysis.

PURPOSE: The appropriate extent of lymph node dissection (LND) during palliative gastrectomy (PALG) for stage IV gastric cancer remains undefined. This study evaluated short-term surgical outcomes of limited versus extended LND in PALG. METHODS: We analyzed 855 patients undergoing PALG with pathologically confirmed stage IV gastric cancer from the pooled Korean Gastric Cancer Association nationwide survey registries of 2019 (n = 14,076 and 68 hospitals) and 2023 (n = 12,752 and 72 hospitals). Patients were stratified by the number of retrieved lymph nodes into limited LND (< 15 nodes and n = 62) and extended LND (≥ 15 nodes and n = 793). Confounding by indication was addressed using overlap (ATO)-based propensity score weighting (PSW) with double-robust multivariable regression. RESULTS: Before weighting, the limited LND group had higher 30-day complication (43.5% vs. 29.8% and p = 0.023) and mortality rates (17.7% vs. 6.8% and p = 0.005). After ATO-based PSW (SMD < 0.001 for all covariates), extended LND showed lower odds of 30-day complications (weighted OR 0.54 [95% CI 0.31-0.95]; double-robust OR 0.50 [0.28-0.87]); and mortality (double-robust OR 0.26 [0.10-0.64]). Pneumonia drove the dominant difference (14.8% vs. 3.7% and p < 0.001). Hospital stay remained longer in the limited LND group after weighting (19.9 ± 18.3 vs. 12.8 ± 12.8 days and p < 0.001). CONCLUSION: After rigorous adjustment for confounding by indication, extended LND in PALG was not associated with excess short-term morbidity or mortality. The lower event rates in the extended LND group most likely reflect preferential selection of frailer, more advanced patients for limited LND rather than a causal benefit. In appropriately selected patients at experienced centers, limiting dissection should not be assumed inherently safer.