Sensitivity of endoscopic biopsy for the detection of gastric and duodenal lesions in dogs and cats

Our aim was to determine how many endoscopic biopsies were necessary to have 99% confidence of finding a gastric/duodenal lesion, and if quality of tissue sample impacted the number of pieces of tissue required for diagnosis. Histopathology slides containing 2,126 tissue samples from 130 animals were solicited from 8 institutions in 5 countries. Slides were randomized using a table of random numbers and coded. Each tissue on each slide was evaluated for diagnosis, severity of pathologic change and quality of sample. Quality of biopsy was assessed as inadequate, marginal, adequate or superior. Severity of lesion was assessed as unreadable, normal, mild, moderate or severe. Sensitivity (Se) was estimated overall and by subsets of species (cat versus dog), biopsy quality (0-3), organ (stomach, duodenum), participating institution and specific lesion type. There was an inverse linear relationship between the number of biopsies needed for diagnosis and the quality of the tissue samples. If tissue quality was not considered, 99% confidence for finding increased cellularity/epithelial lesions required 3 pieces of duodenal tissue in cats and 6 in dogs; and 4 gastric pieces in cats and 3 in dogs. Finding blunt villi required 6 pieces in cats and 9 in dogs; 18 pieces for canine duodenal crypt lesions (this fell to 13 when only adequate tissue samples are considered); and 8 pieces for canine lymphangiectasia. The necessity of taking sufficient numbers of adequate quality biopsies was confirmed, and adequate tissue samples were significantly more likely to find lesions than poor tissue samples.