Objectives: Intestinal metaplasia (IM) and dysplasia in Barrett's esophagus are recognized surrogates for esophageal adenocarcinoma risk. While few would argue with the "hunt for dysplasia," there is a divide regarding the usefulness of the histological confirmation of intestinal metaplasia in endoscopically apparent long segment Barrett's esophagus. We aimed to assess the frequency of intestinal metaplasia in 125 consecutive patients with columnar-lined esophagus and to determine the optimal biopsy protocol to detect intestinal metaplasia.
Methods: Two-hundred ninety-six endoscopies were performed over a 4-yr period in Barrett's esophagus segments of mean length 4 cm (range 1-11 cm) at a single center and the resulting biopsies were analyzed retrospectively. Biopsies were all processed with routine hematoxylin and eosin (H&E) staining, and a subset (N=92) was subject to alcian blue/periodic-acid Schiff staining.
Results: Using H&E staining, we found that the optimum number of biopsies to diagnose intestinal metaplasia was 8 per endoscopy, mean 67.9% endoscopies having intestinal metaplasia. In contrast, if only four were taken the yield was 34.7% with intestinal metaplasia. Unless more than 16 biopsies were taken (100% yield of intestinal metaplasia), no additional significant detection was achieved. Using additional alcian blue/periodic-acid Schiff staining only had a marginal benefit, with 5.4% of new cases of intestinal metaplasia being identified. There is a proximal cephalo-caudal gradient of intestinal metaplasia, especially with increased chronological age, but doing repeat endoscopies on patients did not increase the detection of intestinal metaplasia.
Conclusions: The data suggest that at least 8 random biopsies is the minimum to be taken and analyzed with conventional H&E staining to diagnose benign intestinal metaplasia. Taking more biopsies did not statistically increase the diagnosis of intestinal metaplasia except when greater than 16 were taken when 100% yield was obtained.