University of Miami/Jackson Health System, Miami, FL.
Mount Sinai Medical Center of Florida, Miami Beach.
Am J Clin Pathol. 2020 Feb 8;153(3):360-367. doi: 10.1093/ajcp/aqz179.
We aimed to determine the interobserver reproducibility in diagnosing low-grade ductal carcinoma in situ (DCIS). We also aimed to compare the interobserver variability using a proposed two-tiered grading system as opposed to the current three-tiered system.
Three expert breast pathologists and one junior pathologist identified low-grade DCIS from a set of 300 DCIS slides. Months later, participants were asked to grade the 300 cases using the standard three-tiered system.
Using the two-tiered system, interobserver agreement among breast pathologists was considered moderate (κ = 0.575). The agreement was similar (κ = 0.532) with the junior pathologist included. Using the three-tiered system, pathologists' agreement was poor (κ = 0.235).
Pathologists' reproducibility on diagnosing low-grade DCIS showed moderate agreement. Experience does not seem to influence reproducibility. Our proposed two-tiered system of low vs nonlow grade, where the intermediate grade is grouped in the nonlow category has shown improved concordance.
我们旨在确定诊断低级别导管原位癌（DCIS）时观察者间的可重复性。我们还旨在比较使用建议的两级分级系统与当前的三级系统相比的观察者间变异性。
三位专家乳腺病理学家和一位初级病理学家从一组 300 张 DCIS 幻灯片中识别低级别 DCIS。几个月后，要求参与者使用标准的三级系统对 300 例进行分级。
使用两级系统，乳腺病理学家之间的观察者间一致性被认为是中度的（κ=0.575）。包括初级病理学家在内，一致性相似（κ=0.532）。使用三级系统，病理学家的一致性较差（κ=0.235）。
病理学家在诊断低级别 DCIS 方面的可重复性显示出中度一致性。经验似乎不会影响可重复性。我们提出的低级别与非低级别，其中中级别被归类于非低级别组的两级系统显示出了更高的一致性。
