Abstract
BACKGROUND AND METHODS: Mice are used for modelling the biology of many human diseases, including colorectal cancer (CRC). Mouse models recapitulate many aspects of human disease and are invaluable tools for studying the biology, treatment and prevention of CRC. Unlike humans, many mouse models develop lesions primarily in the small intestine, which necessitates removal and examination of this organ in order to evaluate treatment efficacy. Commonly, the small intestine is visually examined for gross lesions and then selectively embedded in paraffin blocks for further microscopic analysis. Unfortunately, this method suffers from inherent bias toward counting large lesions and simultaneously missing smaller lesions. Even more, this method leaves no permanent record of diagnosed and measured lesions. We evaluated inter-observer variability in a mouse model of CRC using visual examination, and directly compared the visual, gross examination with a histologic analytic method using digital slides of hematoxylin and eosin stained tissue sections. RESULTS: Using visual examination, there was a high degree of inter-observer variability. As this method does not provide a permanent record of measurements, there is no capability to arbitrate between differing observations. In contrast, histologic analysis allowed for the creation of a permanent record of lesion measurements taken. When compared directly, histologic analysis of annotated digital images has significantly improved accuracy. Using this method we were able to distinguish mutant mice from wild type littermates even at a very young age. With gross visual examination, this distinction was not possible. CONCLUSION: Histologic analysis of digital images of murine intestinal tissue provides a vital improvement over the commonly used visual, gross examination method. Unlike visual gross examination, histologic analysis is not biased by the size of intestinal adenoma, misdiagnosis of another lesion type, or presence of a Peyer's patch. It also provides accountability in the form of a permanent record of lesions counted. Histologic analysis using digital slides represents a critical improvement over the current, widely used method of visual gross examination and should be considered for future studies using mouse models of CRC.