Digital Slide Diagnosis of Upper GI Biopsies
D Gui, HM Yang, G Gerney, C Magyar, S Hart, S Dry. UCLA, Los Angeles, CA
Background: Whole slide digital imaging (WSDI) offers an alternative to glass slides for diagnostic interpretation. Prior work has focused on WSDI for frozen section interpretation or second consults. We are not aware of prior studies on use of WSDI for routine GI biopsies. We examined the accuracy and efficiency of WSDI in routine GI biopsies.
Design: Forty-six H&E stained esophageal and stomach biopsies were selected to include diagnoses typical in our GI biopsy service, including: gastric carcinoma (9), reflux/eosinophilic esophagitis (4), Barrett's esophagus (13), Candida esophagitis (4), and chronic gastritis with (11) and without H. pylori (7). Some slides had more than one diagnosis. Slides were anonymized to ensure pathologists were blinded to the diagnosis. Glass slides were scanned at 20X (Aperio XT). Three pathologists with varying diagnostic experience (current GI fellow, 5 years and 10 years experience) and computer skills reviewed the WSDI, then the glass slides, with 3-14 days between reviews. Final diagnosis, time required, problems encountered, additional stains desired and whether 40X magnification was needed were recorded for all cases.
Results: All WSDI cases of carcinoma, Barrett's esophagus, Candida esophagitis and chronic gastritis without H. pylori were correctly diagnosed by all pathologists. For WSDI H. pylori-associated gastritis, all pathologists requested special stains and/or 40X magnification. Two pathologists were willing to diagnose H. pylori gastritis on WSDI, but with less than 100% confidence. In 50% of cases, all pathologists noted difficulties in distinguishing eosinophils from red blood cells (RBCs) on 20X WSDI. Kappa statistics for final diagnosis was almost perfect to perfect for digital to glass interpretation for each pathologist (0.93 - 1.0). Interobserver agreement for WSDI diagnosis was substantial, ranging from 0.78 - 0.82. On average, WSDI took 2-4 times longer than glass slide interpretation for all pathologists.
Conclusions: Our initial data indicates that gastric carcinomas, chronic gastritis and intestinal metaplasia (Barrett's or gastric) is diagnosed accurately and with confidence on 20X WDSI. Small infectious organisms, such as H. pylori, are difficult to see on 20X. Another unexpected difficulty was distinguishing RBCs from eosinophils at 20X WSDI. Significantly more time was required for digital versus glass slide interpretation, which may diminish with experience (studies in progress). Our study suggests 20X WSDI can be used with confidence for most commonly diagnosed upper GI disorders.
Monday, March 22, 2010 9:30 AM
Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 254, Monday Morning