Department of Urology, VU University Medical Center, Amsterdam, the Netherlands.
Department of Urology, VU University Medical Center, Amsterdam, the Netherlands.
Clin Genitourin Cancer. 2018 Oct;16(5):e985-e992. doi: 10.1016/j.clgc.2018.05.002. Epub 2018 May 11.
Histologic grade is an important prognosticator in patients with non-muscle-invasive bladder cancer (NMIBC). Currently, 2 classifications for grade are widely used; the World Health Organization (WHO) 1973 and the WHO 2004. We compare inter-observer variability of both classifications and investigate which histologic criteria cause this variability. Furthermore, the prognostic value of both classifications was assessed.
Three pathologists reviewed 328 bladder tissue samples of 232 patients with NMIBC in a blinded manner. WHO 1973 grade, WHO 2004 grade, histologic criteria of both classifications, and T-category were evaluated. Reproducibility was analyzed using the weighted Fleiss κ, association between criteria scores and grade with the χ test, and time-to-recurrence and time-to-progression with the log-rank test and Cox regression.
Reproducibility of both classifications was poor. The WHO 2004 showed better reproducibility (κ = 0.35; 95% confidence interval (CI), 0.29-0.42) compared with the WHO 1973 as a 3-tiered (κ = 0.24; 95% CI, 0.19-0.28), but not as a 2-tiered (G1 + G2 vs. G3) classification (κ = 0.36; 95% CI, 0.29-0.42). Reproducibility of individual criteria was poor (κ range, -0.05 to 0.25). All criteria were associated with grade (P < .05). After a median follow-up of 60 months, 33 of 232 and 112 of 232 patients developed progression and recurrence, respectively. In 1 out of the 3 pathologists, progression was predicted by both the WHO 1973 grade and the WHO 2004 grade in multivariable analysis. Recurrence was not predicted by grade (multivariable).
Reproducibility of the WHO 2004 and WHO 1973 classification for grade are poor. Scoring of individual criteria is poorly reproducible, suggesting that descriptions of these criteria for grade are not specific. The prognostic value of both the WHO 1973 and the WHO 2004 differ per pathologist.
组织学分级是非肌肉浸润性膀胱癌（NMIBC）患者的重要预后指标。目前，广泛使用两种分级系统；世界卫生组织（WHO）1973 年和 2004 年版。我们比较了这两种分级系统的观察者间变异性，并研究了导致这种变异性的组织学标准。此外，还评估了这两种分级系统的预后价值。
3 位病理学家对 232 例 NMIBC 患者的 328 个膀胱组织样本进行了盲法评估。评估了 WHO 1973 年分级、WHO 2004 年分级、两种分级的组织学标准和 T 分期。使用加权 Fleiss κ 分析可重复性，使用 χ²检验评估标准评分与分级之间的关系，使用对数秩检验和 Cox 回归分析无复发生存期和无进展生存期。
两种分级系统的可重复性均较差。与 WHO 1973 年的 3 级分类（κ=0.24；95%置信区间[CI]，0.19-0.28）相比，WHO 2004 年作为 3 级分类（κ=0.35；95%CI，0.29-0.42）显示出更好的可重复性，但作为 2 级分类（G1+G2 与 G3）则不然（κ=0.36；95%CI，0.29-0.42）。个体标准的可重复性均较差（κ 范围，-0.05 至 0.25）。所有标准均与分级相关（P<.05）。中位随访 60 个月后，232 例患者中有 33 例和 232 例患者中有 112 例分别出现进展和复发。在 3 位病理学家中的 1 位中，多变量分析显示，进展既可以由 WHO 1973 分级预测，也可以由 WHO 2004 分级预测。在多变量分析中，分级不能预测复发。
WHO 2004 年和 WHO 1973 年分级系统的分级可重复性均较差。个别标准的评分可重复性较差，提示这些分级标准的描述不具有特异性。两位病理学家对 WHO 1973 年和 WHO 2004 年分级系统的预后价值判断不同。
