Endoscopic Mucosal Resection Changes Staging for Early-Stage Neoplasia in Barrett's Esophagus Compared to Pinch Biopsies and EUS

Abstract

Background: Endoscopists and surgeons rely on surveillance biopsies to correctly identify dysplasia or carcinoma in Barrett's esophagus (BE). Historically, pinch biopsies (PB) have under staged dysplasia in this setting, with up to 40% of esophagi resected for high-grade dysplasia (HGD) harboring carcinoma. Endoscopic ultrasound (EUS) may improve staging, but its accuracy in mucosal disease has been questioned. Endoscopic mucosal resection (EMR) provides much larger and deeper tissue specimens for histologic analysis and may more accurately identify the level of dysplasia or cancer in BE. Methods: Patients referred to our institution between 12/99 and 11/05 for endoscopic ablative therapy for low-grade dysplasia (LGD), HGD or intra-mucosal carcinoma (IMC) identified by PB (standard or jumbo) who then underwent EMR were identified from a prospectively maintained database. We compared the histologic and tumor stage identified by PB and by EUS with that obtained by EMR. Results: 67 patients underwent at least one EMR for staging or as part of endoscopic ablation. 2 patients were referred with LGD, 43 with HGD and 22 with IMC not believed to extend beyond muscularis mucosa. EUS (12 MHz or 20 MHz Olympus miniprobe) did not agree with PB stage in 17 cases (25%). In 12 cases EUS suggested a higher level of dysplasia while in 5 it suggested a lower level. EMR provided a different histologic diagnosis than PB in 27 cases (40%). 15 cases (22%) were upstaged: From HGD to T1im in 12 cases (18%) and from T1im to T1sm in 3 cases (4%). 12 cases (18%) revealed lower grades of dyplasia: Down-staging from T1im to HGD in 6 pts (9%) and from HGD to LGD in 6 pts (9%). EUS and EMR staging were discordant in 14 cases (21%) with 9 (13%) showing a higher stage on EMR than EUS and 5 (8%) showing a lower stage. All patients were treated endoscopically; as such, no surgical pathology or follow-up biopsies were available to serve as a “gold standard” or final arbiter. The relative accuracy of standard or jumbo forceps biopsies was not discernable from the data. Conclusion: Both EUS and EMR may change the stage of dysplasia in Barrett's esophagus compared to PB. Due to the large sample of tissue available from EMR, EMR probably accurately upstages some patients who are under staged by PB and EUS. It remains unclear whether “down-staging” by EMR is uniformly accurate, since the focal nature of EMR may miss areas of higher stage dysplasia or focal dysplasia may have been removed by the initial PB.