Background & Aims: The aim of this study was to assess the validity of frozen section analysis of endoscopic mucosal resection (EMR) specimens from Barrett's esophagus as compared with permanent sections for the detection of neoplasia. Frozen sections help to give immediate feedback for surgical procedures. It has not been determined whether EMR can be adequately interpreted by using frozen sections to aid endoscopists in completely resecting neoplastic lesions. Methods: EMR specimens from Barrett's esophagus with high-grade dysplasia (HGD) and/or carcinoma were tested by frozen section. Pathologists evaluated EMR specimens for the depth of invasion as well as the appearance of clear margins of resection. The κ statistic was calculated to assess the degree of agreement between the frozen section and permanent section diagnoses. Results: Twenty-three consecutive patients underwent 30 EMRs with frozen section diagnosis. Frozen section revealed a carcinoma in 7 specimens (23%) and dysplasia in 20 (66%). Permanent sections found carcinoma in 8 specimens (26%), dysplasia in 19 specimens (63%), and normal or nondysplastic Barrett's esophagus in the remainder. The κ statistic for the depth of invasion of EMR specimens was 0.93 (near perfect agreement). The κ statistic for the margins of the EMR specimens was 0.80 (excellent agreement). Conclusions: This study indicated that frozen section analysis of esophageal EMR specimens is valid as compared with permanent section. This technique might allow rapid evaluation about the degree and depth of involvement of cancers. This allows physicians to make decisions regarding further therapy if margins are involved or decrease the use of EMR for histologically benign-appearing lesions.
ASJC Scopus subject areas