RESUMO
BACKGROUND: The management of early-stage esophageal cancer is nuanced. A multidisciplinary approach may optimize management through selection of candidates for surgical or endoscopic therapies. The objective of this research was to examine long-term outcomes of patients with early-stage esophageal cancer who undergo treatment with endoscopic resection or surgery. METHODS: Data on patient demographics, co-morbidities, pathology results, OS and RFS were obtained for both the endoscopic resection group and esophagectomy group. Univariate analysis of OS and RFS were conducted using the Kaplan-Meier method with calculation of the log-rank test. Multivariate cox-proportional hazards models were created for OS and RFS using a hypothesis-driven approach. A multivariate logistic regression model was created to identify predictors of esophagectomy among patients undergoing initial endoscopic resection. RESULTS: A total of 111 patients were included. The median OS for the surgery group was 67.0 months compared to 74.0 months in the endoscopic resection group (log-rank p = 0.93). The median RFS for the surgery group was 109.4 months compared to 63.3 months in the endoscopic resection group (log-rank p = 0.0127). On multivariable analysis, patients undergoing endoscopic resection had significantly worse RFS (HR 2.55, 95% CI 1.09-6.00; p = 0.032), but equivalent OS (HR 1.03, 95% CI 0.46-2.32; p = 0.941), compared to patients undergoing esophagectomy. High-grade disease (OR 5.43, 95% CI 1.13-26.10; p = 0.035) and submucosal involvement (OR 7.75, 95% CI 1.90-31.40; p = 0.004) were identified as significant predictors of proceeding to esophagectomy. CONCLUSIONS: Through a multidisciplinary approach, patients with early-stage esophageal cancer achieve excellent RFS and OS. Submucosal involvement and high-grade disease place patients at increased risk for local disease recurrence; these patients may undergo endoscopic resection safely if treated with a multidisciplinary approach incorporating endoscopic surveillance and surgical consultation. Further risk-stratification models may enable better patient selection and optimization of long-term outcomes.
