Incidence of adenocarcinoma in Barrett's esophagus: Results of a multicenter prospective study

Abstract

4001 Background: Barrett's esophagus (BE) is associated with increased cancer risk. However, estimates of the cancer incidence in BE have generally been based on retrospective studies or retrospective cohort analyses that are strictly not prospective. The aim of this study was to prospectively follow a cohort of cancer-free patients with BE to estimate the incidence of esophageal adenocarcinoma based on follow-up procedures used for the diagnosis of cancer being performed as per standard clinical care. Methods: From January 1994 thru September of 1998, 161 patients were initially enrolled in a BE surveillance program, and had at least two serial endoscopies with biopsies and more than one year of follow up. As usual surveillance, four-quadrant biopsies were taken from the Barrett's epithelium at 2-cm intervals with additional samples obtained from any suspicious lesions. Patients with cancers at diagnosis or those developing cancer within 6 months of the initial visit were excluded as they were considered to have prevalent cancers. Results: To date, 367 endoscopies have been performed in these 161 patients, with mean follow-up of 4.3 years (range 1–16 years), yielding a total of 695 patient-years of follow-up. The mean age at the time of the diagnosis of BE was 62.9 years. Short-segment BE accounted for 52 of the 161 patients (32%). Only two patients developed adenocarcinoma (at year 6 and 12 from diagnosis), yielding an incidence rate of one adenocarcinoma per 347.5 patient-years of follow-up. The length of BE lesion did not significantly change over time. Patients developing high grade dysplasia or adenocarcinoma had a significantly longer length of BE (mean 7.6cm) than those without dysplasia or with low grade dysplasia (mean 4.6cm) (p=0.0193). Conclusions: This is the first prospective follow-up study of cancer risk in BE. The incidence of adenocarcinoma in patients with BE is lower than in previously reported retrospective cohort studies. It also confirms previous findings that the length of BE lesion does not change significantly over time. The development of high grade dysplasia and cancer is strongly correlated with lesion length. No significant financial relationships to disclose.