Sampling error in the standard evaluation of endoscopic colonic biopsies.

It is assumed that the diagnostic information in small gastrointestinal mucosal biopsies is obtained regardless of which side of the biopsy is sectioned. Based on this assumption, mucosal biopsies of the colon are embedded randomly with respect to the side of the biopsy that faces the microtome blade. Our hypothesis is that reorienting ("flipping") the biopsy specimen 180 degrees and sectioning the opposite side might provide new diagnostic information in cases with no specific histopathologic diagnosis. Sixty-one consecutive cases with a clinical impression of a polyp and no histologic abnormality on three hematoxylin and eosin-stained slides (6-15 levels) were selected. Formalin-fixed, paraffin-embedded tissue was reoriented, and three additional hematoxylin and eosin-stained slides (6-15 levels) were evaluated. The side of the specimen initially sectioned was arbitrarily designated "A," and the side of the specimen sectioned after reembedding was designated "B." The slides were sequentially numbered 1-3, and the first slide on which the new diagnosis was evident was recorded. New diagnoses were made for 19 of 61 (31.1%) specimens from a total of 51 patients and included tubular adenomas (12 of 61; 19.7%) and hyperplastic polyps (7 of 61; 11.5%). In 17 of 19 specimens (89.5%), the new diagnosis was evident on the first slide. Reembedding endoscopic colonic biopsies facilitates rendering a definitive diagnosis in cases of discordance between the clinical impression and the histologic impression. Routine examination of at least two sides of randomly oriented biopsy material could provide additional diagnostic, and sometimes critical, information.