Svanholm H, Starklint H, Barlebo H, Olsen S
University Institute of Pathology, Odense Hospital, Denmark.
APMIS. 1990 Mar;98(3):229-36. doi: 10.1111/j.1699-0463.1990.tb01026.x.
The aim of this study was to evaluate the reproducibility, expressed by kappa coefficients, of the histological grading system of prostatic cancer as proposed by Schroeder, Hop, Blom and Mostofi in 1985. This grading system has five prognostic groups based on combinations of growth pattern, mitoses and nuclear anaplasia. All histological slides from 85 cases of prostatic carcinoma were blindly evaluated on two independent occasions by each of four pathologists. The overall INTER-observer agreement and mean INTRA-observer agreement regarding "prognostic group" were 0.55 and 0.69 respectively, whereas the kappa coefficients were 0.38 and 0.57 respectively. An acceptable level of INTER-observer reproducibility of prognostic group I (kappa = 0.70), of the parameter "growth pattern" (kappa = 0.60) and of "slight nuclear anaplasia" (kappa = 0.62) was found. A low overall kappa coefficient of "mitoses" and "nuclear anaplasia" was found. Based on the well reproducible parameters a simplified grading system is proposed.
本研究的目的是评估1985年施罗德、霍普、布洛姆和莫斯托菲提出的前列腺癌组织学分级系统的可重复性，用kappa系数表示。该分级系统基于生长模式、有丝分裂和核间变的组合分为五个预后组。四位病理学家分别在两个独立的时间对85例前列腺癌的所有组织学切片进行了盲法评估。关于“预后组”的总体观察者间一致性和平均观察者内一致性分别为0.55和0.69，而kappa系数分别为0.38和0.57。发现预后组I（kappa = 0.70）、参数“生长模式”（kappa = 0.60）和“轻度核间变”（kappa = 0.62）的观察者间可重复性达到可接受水平。发现“有丝分裂”和“核间变”的总体kappa系数较低。基于可重复性良好的参数，提出了一种简化的分级系统。
