Clinical, Endoscopic and Pathologic Factors Combined with EUS Improve Accuracy of Esophageal Cancer Classification

Abstract

Clinical, Endoscopic and Pathologic Factors Combined with EUS Improve Accuracy of Esophageal Cancer Classification Gregory Zuccaro, Thomas Rice, John Vargo, Lisa Rybicki, John Goldblum, David Adelstein, John Dumot, Patricia Trolli, Eugene Blackstone Background: EUS has become a principal tool in staging esophageal cancer (EC), and in guiding therapy based on disease extent. Other potentially useful clinical, endoscopic and pathologic (CEP) factors available to the endosonographer are usually overlooked. Purpose: (1) Determine EUS accuracy in predicting EC disease extent, defined as limited to wall (pTis-pT2, N0, M0) vs advanced beyond wall (pT3pT4, or N1 or M1) (2) Assess if a combination of EUS and CEP factors improves classification accuracy. Methods: 314 pts with EC underwent clinical evaluation, endoscopy with biopsy, and EUS, followed by esophagectomy. No pt received preoperative chemoradiotherapy. Post-operative pathologic classification was the gold standard. Ten CEP factors were assessed along with EUS as potential correlates of advanced disease (age, gender, weight loss, dysphagia, tumor traversability, tumor length, tumor location, tumor morphology, histopathologic type and histologic grade). A logistic regression model was developed to identify correlates of advanced disease (called CEP-EUS model). ROC curves were created for EUS alone and CEP-EUS model. Recursive partitioning analysis was used to identify a cutpoint in CEP-EUS model that best predicted disease extent. From this, referent values for EUS alone and CEP-EUS model were calculated. Results: Pathologic classification at esophagectomy (gold standard): 138 (44%) had disease limited to wall, 176 (56%) advanced beyond wall. EUS correctly predicted disease extent in 265 (84%) pts. Factors correlated with advanced disease in CEP-EUS model: weight loss, presence of dysphagia, tumor length, histologic grade, and EUS result. Area under ROC curves: EUS alone 0.8467, CEP-EUS 0.9401 (p !0.001; Z-test).