Long-term survival comparison of endoscopic resection versus surgical resection for early gastric cancer: a propensity score-matched analysis based on the SEER database

To compare overall survival (OS) and cancer-specific survival (CSS) between ER and surgical resection for EGC using the Surveillance, Epidemiology, and End Results (SEER) database. Patients diagnosed with T1 gastric adenocarcinoma between 2010 and 2022 were identified from the SEER database. Patients were classified into ER (surgery codes 10–27) and surgical resection (codes 30–80) groups. Propensity score matching (PSM) was performed at a 1:1 ratio using a caliper of 0.2 standard deviations of the logit of the propensity score. Kaplan-Meier survival analysis, log-rank tests, and Cox proportional hazards regression were used to compare OS and CSS. A Fine-Gray competing-risks model was used to distinguish cancer-specific death from non-cancer death. Subgroup and sensitivity analyses, including analyses restricted to patients with known tumour grade and to histologically confirmed endoscopic resections, were conducted to evaluate the robustness of the findings. A total of 5,530 patients were included (ER: 1,422; surgery: 4,108). After 1:1 PSM, 1,096 pairs were well-matched. After PSM, the ER group demonstrated slightly inferior OS compared with the surgery group (HR = 1.17, 95% CI: 1.02–1.34, P = 0.028; 5-year OS: 62.1% vs. 68.7%), while CSS showed no statistically significant difference (HR = 0.83, 95% CI: 0.66–1.04, P = 0.111; 5-year CSS: 85.9% vs. 83.0%). A Fine-Gray competing-risks analysis confirmed that ER was associated with a higher subdistribution hazard of non-cancer death (sHR = 1.47, 95% CI: 1.23–1.74, P < 0.001) but not of cancer-specific death (sHR = 0.80, 95% CI: 0.63–1.01, P = 0.059). Subgroup analysis revealed that in T1a tumours ER was associated with significantly superior CSS (HR = 0.64, 95% CI: 0.42–0.96, P = 0.033). Sensitivity analyses excluding signet ring cell carcinoma confirmed significantly better CSS for ER (HR = 0.77, 95% CI: 0.60–0.98, P = 0.031), and analyses restricted to patients with known grade or to histologically confirmed endoscopic resection were consistent. In this observational analysis, ER provided cancer-specific survival comparable to surgical resection for EGC. Competing-risks analysis indicated that the modest overall-survival disadvantage in the ER group was driven by non-cancer mortality in older patients rather than oncological inferiority. A CSS signal favouring ER in T1a tumours was observed but should be regarded as hypothesis-generating. These findings support ER as a reasonable curative option for appropriately selected EGC patients, particularly those with T1a, well-differentiated adenocarcinoma, although prospective confirmation is warranted given the retrospective design.