Background: Malignant esophageal masses can be staged with endoscopic
ultrasound (EUS) using the TNM staging classification. Several criteri
a for differentiating between intraesophageal (T1-2) and extraesophage
al (T3-4) masses have been described, but highly accurate staging rema
ins difficult. Methods: This is a blinded evaluation of four specific
EUS criteria in 24 patients with esophageal malignancy who underwent e
sophageal resection after neoadjuvant chemotherapy. Radial EUS was use
d to evaluate the first 12 patients and curved linear EUS was used in
the second half of the group. Using the histology of the resected spec
imens, the sensitivity, specificity, and accuracy of the EUS criteria
after chemotherapy were determined for predicting intraesophageal or e
xtraesophageal invasion. Results: There was no difference in the accur
acy rates with radial or linear EUS. Two ultrasound criteria, muscular
is disruption and irregular mass border, were found to have low accura
cy rates (44% and 50%). The maximal thickness (overall or extraesophag
eal) of the esophageal mass was found to be highly accurate (79% and 8
7%) in predicting intraesophageal or extraesophageal extension. pT3-4
masses had a thickness of 16.0 +/- 2 mm, significantly greater than pT
1-2 masses, 8.2 +/- 1 mm (p < .01). Using receiver operator characteri
stics (ROC) curve analysis, mass thickness was found to be more accura
te (91% and 94%) than a subjective assessment of staging (73%) (p < .0
7). Conclusions: The EUS measurement of a malignant esophageal mass ma
ximal thickness can accurately predict extraesophageal extension.