Abstract
OBJECTIVE: Endoscopic ultrasound (EUS) evaluation of T2N0 esophageal cancers is associated with high rates of staging inaccuracy, impacting whether induction therapy is appropriately selected for these patients. We aimed to identify whether tumor appearance on endoscopy can predict whether preoperative EUS staging appropriately directs management. METHODS: The impact of tumors being described on endoscopic reports as bulky, ulcerative, or obstructive on the accuracy of EUS staging for patients with cT2N0 esophageal cancer who underwent esophagectomy without induction therapy at a single institution between 2008 and 2024 was evaluated with χ(2) analysis. A receiver operating characteristic curve was constructed using a logistic regression model estimating EUS understaging compared with pathologic reports. RESULTS: Of 35 patients with cT2N0 esophageal cancer who underwent esophagectomy without induction therapy, 57.1% (20/35) patients had pT1-2N0 pathology after esophagectomy. Conversely, 42.9% (15/35) of patients were understaged by EUS. There was a significant association between EUS accuracy and tumors endoscopically described as bulky, ulcerative, or obstructive (P = .002). Of the 15 patients who were understaged, the tumor endoscopic description was bulky, ulcerative, or obstructive in 14 (93.3% sensitivity). Of the 20 patients who were not understaged, 13 were negative for those key words (65.0% specificity). Not having the endoscopic key words was associated with tumors not being understaged in 13 of 14 patients (92.9% negative predictive value), whereas having the key words was associated with understaging in 14 of 21 patients (odds ratio, 26.0; P = .004; 66.7% positive predictive value). CONCLUSIONS: Patients with cT2N0 esophageal cancer by EUS are likely to be understaged when the tumor endoscopic appearance is bulky, ulcerative, or obstructive.