Interobserver reproducibility in the diagnosis of flat epithelial atypia of the breast

Since the early 1980s, a bewildering array of methods for constructing bootstrap confidence intervals have been proposed. In this article, we address the following questions. First, when should bootstrap confidence intervals be used. Secondly, which method should be chosen, and thirdly, how should it be implemented. In order to do this, we review the common algorithms for resampling and methods for constructing bootstrap confidence intervals, together with some less well known ones, highlighting their strengths and weaknesses. We then present a simulation study, a flow chart for choosing an appropriate method and a survival analysis example. Copyright 2000 John Wiley & Sons, Ltd.

In view of the increasing number of patients treated with breast-conserving treatment (BCT) for ductal carcinoma-in-situ (DCIS), risk factors for recurrence and metastasis should be identified. Clinical and pathologic characteristics from patients with DCIS in the European Organization for Research and Treatment of Cancer trial 10853 (excision with or without radiotherapy) were related to the risk of recurrence. Pathologic features were derived from a central review of 863 of the 1,010 randomized cases (85%). The median follow-up was 5.4 years. Factors associated with an increased risk of local recurrence in the multivariate analysis were young age (< or = 40 years) (hazard ratio, 2.14; P =.02), symptomatic detection of DCIS (hazard ratio, 1.80; P =.008), growth pattern (solid and cribriform) (hazard ratios, 2.67 and 2.69, respectively; P =.012), involved margins (hazard ratio, 2.07; P =.0008), and treatment by local excision alone (hazard ratio, 1.74; P =.009). The risk of invasive recurrence was not related to the histologic type of DCIS (P =.63), but the risk of distant metastasis was significantly higher in poorly differentiated DCIS compared with well-differentiated DCIS (hazard ratio, 6.57; P =.01). Patients with poorly differentiated DCIS have a high risk of distant metastasis after invasive local recurrence. Margin status is the most important factor in the success of BCT for DCIS; additionally, young age and symptomatic detection of DCIS have negative prognostic value.

Although the categorization of proliferative breast lesions provides valuable information regarding subsequent risk of breast cancer, the ability of pathologists to classify such lesions in a reproducible fashion has not been adequately evaluated. To assess further interobserver reproducibility in the categorization of proliferative breast lesions, six pathologists each reviewed 24 proliferative ductal lesions and classified them as either usual hyperplasia (H), atypical hyperplasia (AH), or carcinoma in situ (CIS). Before evaluation of the study slides, all the participants were instructed to use the diagnostic criteria of Page and co-workers and were provided with both a written summary of these criteria and a set of teaching slides with representative examples of each type of lesion. Complete agreement among all six pathologists was seen in 14 cases (58%); five or more agreed in 17 cases (71%); and four or more arrived at the same diagnosis in 22 cases (92%). No pathologist consistently rendered more "benign" or "malignant" diagnoses than any other. After assigning numerical values for each diagnostic category (H = 1, AH = 2, CIS = 3), the scores for the group of 24 cases did not differ significantly by pathologist (p = 0.68; average score range, 1.7-2.0). Our results indicate that with the use of standardized criteria, interobserver concordance in the diagnosis of proliferative ductal breast lesions can be obtained in the majority of cases.