Blackwell Publishing Journal Backfiles 1879-2005
Abstract: Due to the increased efforts of early detection and biopsy of breast lesions, the number of cases with hyperplastic and in situ lesions seen in most pathology laboratories has increased significantly. Pathologists are faced with a greater number of biopsies and a wider spectrum of lobular and ductal lesions including papillary lesions. A set of 31 randomly selected glass slides of ductal, lobular and papillary lesions (benign, borderline, and carcinomas in situ) was circulated among five community- and five academic-based pathologists (median 9 years of experience). The pathologists filled out a standard form for each case which included lesion type (ductal, lobular, or papillary) and diagnostic category (ductal, lobular, or papillary; hyperplasia or carcinoma). We compare the participants' diagnoses, their histologic criteria, and the reproducibility of the diagnoses. Kappa statistics were calculated to measure the levels of agreement among community and academic pathologists, overall for all the pathologists, and within a subgroup of eight participants reporting the use of the same diagnostic criteria (Page's). All pathologists agreed with the diagnosis of cancer in 10 cases (33%), kappa overall 0.347 (fair); with the diagnostic category of lesions in 8 cases (27%), kappa overall 0.537 (moderate); and with the lesion type in 16 cases (53%), kappa overall 0.789 (substantial). Community pathologists were more likely to interpret ductal and papillary lesions as hyperplasias than carcinomas in situ. In contrast, academic pathologists interpreted more cases as carcinomas than hyperplasias (60 versus 45). There was little to no improvement in the level of agreement observed when we considered the subgroup of pathologists using the same standardized criteria. This study demonstrates that the overall agreement in the diagnosis of carcinoma in situ of the breast remains low between pathologists using or not using standardized diagnostic criteria. Papillary and ductal lesions pose a significant problem between observers in characterizing them as benign or malignant. Current diagnostic criteria need to be modified, reemphasized, or replaced, either through the literature or with additional training to decrease interobserver variability.
Type of Medium: