CORRELATION BETWEEN PATHOLOGISTS ASSESSING ENDOSCOPIC GASTRIC AND INTESTINAL BIOPSIES USING WSAVA GUIDELINES

We examined agreement between four board-certified pathologists who independently evaluated endoscopic mucosal biopsies of canine and feline stomach and intestine using the WSAVA Guidelines (J Comp Pathol 2008, 138:S1-43). Slides with 2,287 pieces of tissue from 85 dogs and 41 cats were obtained from 7 institutions. Pathologists scored 16 histological parameters as normal, mild, moderate or severe. Spearman rank correlation coefficient was used to determine pair-wise correlation between individual pathologists (i.e., 6 comparisons for each histologic lesion). Correlation was consistently very weak (i.e., < 0.40) for gastric neutrophils (0.06-0.31), duodenal neutrophils (-0.03-0.37) duodenal eosinophils (0.08-0.40), deep gastric injury (5 of 6 comparisons were -0.03-0.38) and duodenal fibrosis (0.11-0.32). Inconsistent correlations that tended to be weak included gastric eosinophils (3 of 6 comparisons were 0.06-0.32) and gastric atrophy (2 of 6 comparisons were 0.18-0.28). All comparisons for gastric intraepithelial lymphocytes (0.44-0.66), gastric lamina proprial lymphocytes (0.48-0.66), gastric fibrosis (0.41-0.69), gastric follicular hyperplasia (0.44-0.61) and duodenal crypt lesions (0.51-0.73) were stronger. Duodenal intraepithelial lymphocytes, lamina proprial lymphocytes, dilated lacteals and blunt villi each had 1 of 6 comparisons that was < 0.40. However, if the pair-wise correlation between one specific set of pathologists were consistently eliminated, then 24 of the 25 remaining comparisons of these duodenal tissues were 0.42-0.73. Our data show that additional work is necessary to accomplish consistency on select histologic lesions (neutrophils, eosinophils, duodenal fibrosis, gastric atrophy, deep gastric injury).We suspect variation in staining and tissue processing contributed to lack of agreement between pathologists on eosinophils and neutrophils