Endoscopic and Endosonographic Predictors of Cancer in Patients with Barrett's Esophagus and High Grade Dysplasia

Abstract

Background: Barrett's esophagus (BE) with high grade dysplasia (HGD) may have co-existent esophageal carcinoma and as a result traditional treatment has been esophagectomy. There has been increasing interest in endoscopic therapy as a viable alternative to esophagectomy in the management of HGD but little data exists to correctly guide patients between these treatment options. Aim: We studied patients referred for Endoscopic Ultrasound (EUS)/EGD who had BE with HGD to evaluate the ability of EGD and EUS to determine the presence of co-existing malignancy in HGD before therapy is instituted. Methods: All patients referred for evaluation of HGD in BE from October 2001 to November 2008 for EUS were included. Patients with prior biopsy proven carcinoma were excluded. A total of 53 patients met the above entry criteria. Analysis was performed to see if the findings of endoscopic nodularity, ulcer, stricture, length of BE, focal vs. multifocal HGD, or lesion seen on EUS staging predicted the presence of cancer. Cancer diagnosis was determined by histologic specimen and/or follow-up without evidence of cancer development. Mortality was confirmed for all patients as of November 2008. Results: A total of 30 out of 53 (57%) underwent esophagectomy. 5 patients had continued surveillance, 6 underwent photodynamic therapy, 6 endoscopic mucosal resection, 5 radiofrequency ablation, and 1 chemotherapy and radiation. The average length of follow-up for patients was 43 months (range 1 month to 84 months) with no new cases of adenocarcinoma. 11 patients out of 53 (20%) with HGD had proven esophageal cancer by histology. The mortality rate for the cohort was 2/53 (3.77%), both patients with proven carcinoma. Endoscopic findings of nodularity (p=0.004) and ulceration (p=0.01) predicted the presence of carcinoma. Other findings on endoscopy did not predict carcinoma including stricture (p=0.79), length of BE (p=0.11) or multifocal HGD (p=0.36). EUS stage greater than T0N0 significantly predicted cancer with an odds ratio of 49.4 and p value <0.0000001. One patient with flat HGD and a normal EUS exam was found to have intramucosal carcinoma. Conclusion: 1. The findings of nodularity and ulceration in patients with BE and HGD significantly predicts a co-existent cancer diagnosis. 2. EUS stage is the greatest independent predictor of carcinoma in patients with BE/HGD 3. Patients with flat HGD and negative EUS exam are very unlikely to have invasive cancer present and are thus are the best candidates for endoscopic therapy.