Department of Radiation Oncology, Cedars-Sinai Medical Center, Los Angeles, CA, USA.
Samuel Oschin Comprehensive Cancer Institute, Cedars-Sinai Medical Center, Los Angeles, CA, USA.
J Natl Cancer Inst. 2022 Jul 11;114(7):1003-1011. doi: 10.1093/jnci/djac059.
Nodal staging systems vary substantially across solid tumors, implying heterogeneity in the behavior of nodal variables in various contexts. We hypothesized, in contradiction to this, that metastatic lymph node (LN) number is a universal and dominant predictor of outcome across solid tumors.
We performed a retrospective cohort analysis of 1 304 498 patients in the National Cancer Database undergoing surgery between 2004 and 2015 across 16 solid cancer sites. Multivariable Cox regression analyses were constructed using restricted cubic splines to model the association between nodal number and mortality. Recursive partitioning analysis (RPA) was used to derive nodal classification systems for each solid cancer based on metastatic LN count. The reproducibility of these findings was assessed in 1 969 727 patients from the Surveillance, Epidemiology, and End Results registry. Two-sided tests were used for all statistical analyses.
Consistently across disease sites, mortality risk increased continuously with increasing number of metastatic LNs (P < .001 for all spline segments). Each RPA-derived nodal classification system produced multiple prognostic groups spanning a wide spectrum of mortality risk (P < .001). Multivariable models using these RPA-derived nodal classifications demonstrated improved concordance with mortality compared with models using American Joint Committee on Cancer staging in sites where nodal classification is not based on metastatic LN count. Each RPA-derived nodal classification system was reproducible in a large validation cohort for all-cause and cause-specific mortality (P < .001). High quantitative nodal burden was the single strongest tumor-intrinsic variable associated with mortality in 12 of 16 disease sites.
Quantitative metastatic LN burden is a fundamental driver of mortality across solid cancers and should serve as a foundation for pathologic nodal staging across solid tumors.
淋巴结分期系统在实体瘤之间存在很大差异,这意味着淋巴结变量在不同情况下的行为存在异质性。与这一观点相反,我们假设转移性淋巴结(LN)数量是各种实体瘤中普遍存在且主导预后的因素。
我们对 2004 年至 2015 年间在 16 个实体癌部位接受手术的 1304498 例国家癌症数据库患者进行了回顾性队列分析。使用受限立方样条进行多变量 Cox 回归分析,以建立淋巴结数量与死亡率之间的关联模型。递归分区分析(RPA)用于根据转移性 LN 计数为每种实体癌推导淋巴结分类系统。这些发现的可重复性在 Surveillance,Epidemiology,and End Results 注册中心的 1969727 例患者中进行了评估。所有统计分析均采用双侧检验。
在所有疾病部位,死亡率随着转移性 LN 数量的增加而持续增加(所有样条段 P<0.001)。根据 RPA 推导的每种淋巴结分类系统产生了多个预后组,涵盖了广泛的死亡率风险范围(P<0.001)。与不基于转移性 LN 计数的 AJCC 分期模型相比,使用这些 RPA 衍生的淋巴结分类模型的多变量模型与死亡率的一致性得到了改善。在所有部位的全因和特定原因死亡率的大型验证队列中,每个 RPA 衍生的淋巴结分类系统均具有可重复性(P<0.001)。在 16 个疾病部位中的 12 个部位,高定量淋巴结负担是与死亡率最相关的唯一肿瘤内在变量。
定量转移性 LN 负担是实体瘤中死亡率的主要驱动因素,应作为实体瘤中病理淋巴结分期的基础。