Martini-Klinik Prostate Cancer Center, University Hospital Hamburg-Eppendorf, Hamburg, Germany; Cancer Prognostics and Health Outcomes Unit, Division of Urology, University of Montreal Health Center, Montreal, Quebec, Canada.
Cancer Prognostics and Health Outcomes Unit, Division of Urology, University of Montreal Health Center, Montreal, Quebec, Canada; Department of Urology and Division of Experimental Oncology, URI, Urological Research Institute, IRCCS San Raffaele Scientific Institute, Milan, Italy.
Clin Genitourin Cancer. 2021 Apr;19(2):e120-e128. doi: 10.1016/j.clgc.2020.11.003. Epub 2020 Nov 10.
Intermediate-risk prostate cancer (IR PCa) phenotypes may vary from favorable to unfavorable. National Comprehensive Cancer Network (NCCN) criteria help distinguish between those groups. We studied and attempted to improve this stratification.
A total of 4048 (NCCN favorable: 2015 [49.8%] vs. unfavorable 2033 [50.2%]) patients with IR PCa treated with radical prostatectomy were abstracted from an institutional database (2000-2018). Multivariable logistic regression models predicting upstaging and/or upgrading (Gleason Grade Group [GGG] IV-V and/or ≥ pT3 or pN1) in IR PCa were developed, validated, and directly compared with the NCCN IR PCa stratification.
All 4048 patients were randomly divided between development (n = 2024; 50.0%) and validation cohorts (n = 2024; 50.0%). The development cohort was used to fit basic (age, prostate-specific antigen, clinical T stage, biopsy GGG, and percentage of positive cores [all P < .001]) and extended models (age, prostate-specific antigen, clinical T stage, biopsy GGG, prostate volume, and percentage of tumor within all biopsy cores [all P < .001]). In the validation cohort, the basic and the extended models were, respectively, 71.4% and 74.7% accurate in predicting upstaging and/or upgrading versus 66.8% for the NCCN IR PCa stratification. Both models outperformed NCCN IR PCa stratification in calibration and decision curve analyses (DCA). Use of NCCN IR PCa stratification would have misclassified 20.1% of patients with ≥ pT3 or pN1 and/or GGG IV to V versus 18.3% and 16.4% who were misclassified using the basic or the extended model, respectively.
Both newly developed and validated models better discriminate upstaging and/or upgrading risk than the NCCN IR PCa stratification.
中危前列腺癌（IR PCa）表型可能从有利变为不利。国家综合癌症网络（NCCN）标准有助于区分这些组。我们研究并试图改进这种分层。
从机构数据库（2000-2018 年）中提取了 4048 名接受根治性前列腺切除术治疗的 IR PCa 患者（NCCN 有利：2015 名[49.8%]与不利：2033 名[50.2%]）。建立、验证并直接比较了多变量逻辑回归模型，以预测 IR PCa 的升级和/或升级（Gleason 分级组[GGG] IV-V 和/或≥pT3 或 pN1）。
所有 4048 名患者均随机分为发展队列（n=2024；50.0%）和验证队列（n=2024；50.0%）。发展队列用于拟合基本（年龄、前列腺特异性抗原、临床 T 分期、活检 GGG 和阳性核心百分比[均 P<0.001]）和扩展模型（年龄、前列腺特异性抗原、临床 T 分期、活检 GGG、前列腺体积和所有活检核心内肿瘤百分比[均 P<0.001]）。在验证队列中，基本和扩展模型分别在预测升级和/或升级方面的准确率为 71.4%和 74.7%，而 NCCN IR PCa 分层的准确率为 66.8%。在校准和决策曲线分析（DCA）中，这两种模型均优于 NCCN IR PCa 分层。使用 NCCN IR PCa 分层会将 20.1%的≥pT3 或 pN1 和/或 GGG IV 患者错误分类为 V 级，而使用基本或扩展模型分别会有 18.3%和 16.4%的患者被错误分类。
新开发和验证的模型比 NCCN IR PCa 分层更好地区分升级和/或升级风险。
