Treatment allocation in patients with early-stage esophageal adenocarcinoma: Prevalence and predictors of lymph node involvement.

BACKGROUND: In considering treatment allocation for patients with early esophageal adenocarcinoma, the incidence of lymph node metastasis is a critical determinant; however, this has not been well defined or stratified by the relevant clinical predictors of lymph node spread. METHODS: Data from the Surveillance, Epidemiology, and End Results database of the National Cancer Institute were abstracted from 2004 to 2010 for patients with early-stage esophageal adenocarcinoma. The incidence of lymph node involvement for patients with Tis, T1a, and T1b tumors was examined and was stratified by predictors of spread. RESULTS: A total of 13,996 patients with esophageal adenocarcinoma were evaluated. Excluding those with advanced, metastatic, and/or invasive (T2-T4) disease, 715 patients with Tis, T1a, and T1b tumors were included. On multivariate analysis, tumor grade (odds ratio [OR], 2.76; 95% confidence interval [95% CI], 1.58-4.82 [P<.001]), T classification (OR, 0.47; 95% CI, 0.24-0.91 [P =.025]), and tumor size (OR, 2.68; 95% CI, 1.48-4.85 [P = .001]) were found to be independently associated with lymph node metastases. There was no lymph node spread noted with Tis tumors. For patients with low-grade (well or moderately differentiated) tumors measuring <2 cm in size, the risk of lymph node metastasis was 1.7% for T1a (P<.001) and 8.6% for T1b (P = .001) tumors. CONCLUSIONS: For patients with low-grade Tis or T1 tumors measuring ≤2 cm in size, the incidence of lymph node metastasis appears to be comparable to the mortality rate associated with esophagectomy. For highly selected patients with early esophageal adenocarcinomas, the results of the current study support the recommendation that local endoscopic resection can be considered as an alternative to surgical management when followed by rigorous endoscopic and radiographic surveillance. Cancer 2016;122:2150-7. © 2016 American Cancer Society.