Comparative Analysis of Overall Survival and Cancer-Specific Survival Between Endoscopic Therapy and Surgical...

Few studies have compared endoscopic therapy (ET) with surgical resection for gastrointestinal stromal tumors (GISTs) measuring 2 to 10 cm in terms of overall survival and cancer-specific survival (CSS). This retrospective cohort study aimed to address this gap using data from the Surveillance, Epidemiology, and End Results (SEER) database.

Data were extracted from the SEER database for patients with GIST who underwent ET or surgical resection between 2007 and 2021. Survival differences were assessed using Kaplan-Meier curves and log‑rank tests. Multivariable Cox proportional hazards models and 1 : 1 propensity score matching (propensity score matching [PSM], caliper = 0.02) were used to adjust for potential confounders.

A total of 3471 patients (394 in the ET group and 3077 in the surgery group) were included. Before PSM, the 5- and 10-year overall survival (OS) rates were 84.11% (95% CI: 79.41%-89.10%) and 66.03% (95% CI: 57.96%-75.22%), respectively, in the ET group, compared with 86.00% (95% CI: 84.40%-87.70%) and 69.31% (95% CI: 66.15%-72.61%), respectively, in the surgery group. CSS showed similar patterns. After adjustment for confounders, ET was not significantly associated with OS (hazard ratios [HR] = 1.238, 95% CI: 0.934-1.639, P = .137) or CSS (HR = 0.932, 95% CI: 0.510-1.702, P = .818). PSM yielded comparable OS and CSS between the 2 treatment groups.

In a PSM cohort from the SEER database, ET was associated with overall survival and CSS comparable with that of surgical resection. However, these findings are hypothesis generating and require validation in prospective studies incorporating more detailed oncologic outcomes (R0 resection status, tumor rupture, recurrence, or adverse events, etc.). ET may be considered for carefully selected GISTs when performed by experienced endoscopists after thorough tumor assessment; however, these exploratory findings should not be interpreted as a treatment recommendation.