Making Primary Diagnosis on Liver Allograft Biopsies With Whole Slide Images - A Validation Study

Abstract

Although the use of whole slide images (WSIs) in diagnostic pathology has been widely studied, its reliability in assessing liver allograft biopsies has not yet been investigated. Due to the time sensitive nature of allograft donor organ biopsies, the use of WSIs may greatly facilitate speedy off-site evaluation. In this study, we examined the accuracy and reproducibility of diagnosis on liver allograft made from of WSIs by pathologists trained with glass slides. 39 cases of allograft liver donors in our institution in the past 10 years were included. WSIs were made using an Aperio ScanScope OS slide scanner. 2 pathologists independently reviewed each case twice: once with glass slides using a microscope, and the second time via WSIs using a computer. The washout time between the 2 reviews was at least 2 weeks. Morphologic features were examined, including fat infiltration (macrovesicular and microvesicular), portal fibrosis, hepatitis, centrilobular necrosis, carcinoma, granulomas, and alcohol induced injuries. Statistical studies based on coefficient of concordance were performed using the SAS “MAGREE” macro. The degree of intraobserver reliability was determined for each individual pathologist. In addition, the degree of interobserver reliability was calculated across the 2 pathologists. We found that there is substantial intraobserver agreement in diagnostic results from the WSIs and the glass slides, with mean concordance greater than 0.7 for all of the morphological features. In addition the interobserver coefficients of concordance are similar between the 2 modalities in regard to the morphological features. These results indicate that WSIs are a reliable diagnostic tool in evaluating liver allograft biopsies.