Utility of positron emission tomography for the staging of patients with potentially operable esophageal carcinoma

Purpose: A prospective study of preoperative tumor-node-metastasis staging of patients with esophageal cancer (EC) was designed to compare the accuracy of 18-F-fluoro-deoxy-D-glucose (FDG) position emission tomography (PET) with conventional noninvasive modalities. Patients and Methods: Seventy-four patients with carcinomas of the esophagus (n = 43) or gastroesophageal junction (n = 31) were studied. All patients underwent attenuation-corrected FDG-PET imaging, a spiral computed tomography (CT) scan, and an endoscopic ultrasound (EUS). Results: FDG-PET demonstrated increased activity in the primary tumor in 70 of 74 patients (sensitivity: 95%). False-negative PET images were found in four patients with T1 lesions. Thirty-four patients (46%) had stage IV disease. FDG-PET had a higher accuracy for diagnosing stage IV disease compared with the combination of CT and EUS (82% v 64%, respectively; P = .004). FDG-PET had additional diagnostic value in It. (22%) of 74 patients by upstaging 11 (15%) and downstaging five (7%) patients. Thirty-nine (53%) of the 74 patients underwent a 2- or 3-field lymphadenectomy in conjunction with primary curative esophagectomy. In these patients, tumoral involvement was found in 21 local and 35 regional or distant lymph nodes (LN). For local LN, the sensitivity of FDG-PET was lower than EUS (33% v 81%, respectively; P = .027), but the specificity may have been higher (89% v 67%, respectively; P = not significant [NS]). For the assessment of regional and distant LN involvement, compared with the combined use of CT and EUS, FDG-PET had a higher specificity (90% v 98%, respectively; P = .025) and a similar sensitivity (46% v 43%, respectively; P = NS). Conclusion: PET significantly improver the detection of stage IV disease in EC compared with the conventional staging modalities. PET improves diagnostic specificity for LN staging. (C) 2000 by American Society of Clinical Oncology.