Abstract Objectives Individual treatment strategies for esophageal cancer have been investigated based on the anatomical subsite classification. Accurate subsite classification based on these anatomical landmarks is thus important. We investigated the suitability of the existing endoscopic classification and explored alternative landmarks for esophageal subsite classification. Methods Patients who received endoscopic ultrasonography (and computed tomography scans for surveillance of esophageal cancer treatment or esophageal submucosal tumors were included. Distances between anatomical landmarks, including the inferior cricoid cartilage border, superior border of the sternum, and tracheal bifurcation, were measured using a combination of endoscopic ultrasonography, computed tomography, and other information. Results The mean (standard deviation) distances from the superior incisor dentition to the pharynx–esophagus, cervical–upper thoracic esophagus, and upper–middle thoracic esophagus boundaries were 16.9 (1.7), 21.7 (1.9), and 29.0 (1.9) cm, respectively. However, variances in the differences between the mean and individual distances were large (2.8, 3.4, and 3.7, respectively), mainly because of differences in body height. However, variances in the differences between individual distances and novel endoscopic landmarks, including the lower end of the pyriform sinus and lower end of compression of the left main bronchus, were lower (1.7, 1.2, and 0.6, respectively). Conclusions Existing indicators of esophageal subsite boundaries were not consistent with anatomical boundaries. Modification of the distance from the superior incisor dentition based on average distances from anatomical landmarks or the use of alternative endoscopic landmarks is recommended to provide more suitable anatomical boundaries.
