Selection of patients for curative or palliative resection of esophageal cancer based on preoperative endoscopic ultrasonography
J. H. Peters, S. F. Hoeft, J. Heimbucher, R. M. Bremner, T. R. DeMeester, C. G. Bremner, G. W. Clark, M. Kiyabu and Y. Parisky
Department of Surgery, University of Southern California School of Medicine, Los Angeles.
OBJECTIVE: To assess the accuracy of pretreatment staging and the potential
of using endosonographic findings to select patients for curative or
palliative resection by comparing the preoperative endosonographic and
computed tomographic (CT) findings with the histology of the surgical
specimen. METHODS: Forty-two patients referred to our clinic with
esophageal carcinoma underwent preoperative upper endoscopy with biopsy,
endosonography, thoracic CT, and abdominal CT. Based on endoscopic
ultrasonographic findings, patients with early-stage disease underwent
en-bloc esophagogastrectomy, whereas those with advanced disease had a
palliative transhiatal esophagectomy. Exceptions included patients with
poor physiologic reserve who were treated by the transhiatal route.
RESULTS: In eight patients, we were unable to pass the ultrasonographic
endoscope. Seven of these eight had transmural tumors with nodal
involvement on histologic study. Tumor length, based on endosonographic
measurements, was correctly predicted in 34 patients (85%). Extent of wall
penetration was accurately predicted in 26 (76%) of the 34, and regional
lymph node status was accurately predicted in 28 (82%) of the 34. Of the
patients with sonographic wall penetration, 80% had histologic evidence of
one or more positive nodes. Using the WNM staging system, endoscopic
ultrasonography correctly staged the cancer in 68% of the patients. Three
patients were treated with an inappropriate procedure. CONCLUSION:
Endosonography is a reliable method for the preoperative staging and
selection of patients for curative or palliative resection. Endosonographic
wall penetration appears to be a critical factor in determining tumor
spread.
