Jeffrey A Hagen, MD1, Vivek Kaul, MD2, Michael S Smith, MD, MBA3. 1University of Southern California, 2University of Rochester School of Medicine, 3Temple University School orf Medicine
Background: Multiple endoscopic methods are utilized to ablate Barrett’s esophagus (BE) and esophageal dysplasia (ED). The current standard for post-ablation surveillance involves four quadrant forceps biopsies (FB) taken throughout the original BE segment. This technique leaves a significant proportion of mucosa unsampled, decreasing the likelihood of detecting residual or recurrent intestinal metaplasia (IM) and ED. Wide Area Transepithelial Sampling with Computer-Assisted 3-Dimensional Tissue Analysis (WATS3D) has been shown to improve detection of both metaplasia and dysplasia when added to FB in screening/surveillance of BE. Our aim was to further explore the benefit of WATS3D when used for post-ablation surveillance at multiple centers treating BE.
Methods: Patients undergoing surveillance endoscopy following previous BE ablation underwent same session FB and WATS3D tissue sampling. Sampling technique, plus the frequency and location of biopsies, were at the discretion of the endoscopist. All WATS3D samples were analyzed at a central laboratory using a neural network to highlight potentially abnormal cells in a computer synthesized three dimensional image for review by the pathologist. FB were read as per each site’s standard protocol. Procedures performed between June 2013 and September 2015 where WATS3D found IM or ED in the post-ablation setting were identified. Each site provided a de-identified, summarized data set for these cases, prior to aggregation and analysis.
Results: During the study period, a total of 354 patients underwent 611 procedures where FB and WATS3D were performed during the same upper endoscopy for post-ablation surveillance. Most patients were male (75%) with an average age of 65.5 years (25-89). WATS3D identified non-dysplastic IM that was undetected by FB in 27 cases (4%) and ED not detected by FB in 12 cases (2%). Therefore, the number needed to test (NNT) to identify an additional case of either IM or ED not detected by FB was 15.7 (611/39). To identify an additional case of ED, the NNT was 50.9 (611/12). No complications from WATS3D use were reported.
Conclusions: Following BE ablation, adjunctive use of WATS3D with FB increases the detection of residual or recurrent IM as well as ED. With a NNT of only 15.7 to identify any residual or recurrent IM or ED missed by FB, WATS3D provides an effective method to decrease the sampling error inherent in 4 quadrant FB in this setting.