Breast Cancer Unit, Department of Surgery, Virgen de La Arrixaca University Hospital, Ctra. Madrid-Cartagena, s/n, El Palmar, 30120, Murcia, Spain.
Breast Cancer Unit, La Fe University and Polytechnic Hospital, Valencia, Spain.
Breast Cancer Res Treat. 2020 Jun;181(2):339-345. doi: 10.1007/s10549-020-05623-4. Epub 2020 Apr 6.
A prognostic model based on the results of molecular analysis of sentinel lymph nodes (SLN) is needed to replace the information that staging the entire axilla provided. The aim of the study is to conduct an external validation of a previously developed model for the prediction of 5-year DFS in a group of breast cancer patients that had undergone SLN biopsy assessed by the One Step Nucleic Acid Amplification (OSNA) method.
We collected retrospective data of 889 patients with breast cancer, who had not received systemic treatment before surgery, and who underwent SLN biopsy and evaluation of all SLN by OSNA. The discrimination ability of the model was assessed by the area under the ROC curve (AUC ROC), and its calibration by comparing 5-years DFS Kaplan-Meier estimates in quartile groups of model predicted probabilities (MPP).
The AUC ROC ranged from 0.78 (at 2 years) to 0.73 (at 5 years) in the training set, and from 0.78 to 0.71, respectively, in the validation set. The MPP allowed to distinguish four groups of patients with heterogeneous DFS (log-rank test p < 0.0001). In the highest risk group, the HR were 6.04 [95% CI 2.70, 13.48] in the training set and 4.79 [2.310, 9.93] in the validation set.
The model for the prediction of 5-year DFS was successfully validated using the most stringent form of validation, in centers different from those involved in the development of the model. The external validation of the model confirms its utility for the prediction of 5-year DFS and the usefulness of the TTL value as a prognostic variable.
需要基于前哨淋巴结 (SLN) 分子分析结果建立预测模型，以取代对整个腋窝分期的信息。本研究的目的是对先前开发的模型进行外部验证，该模型用于预测一组接受 SLN 活检的乳腺癌患者的 5 年 DFS，这些患者接受了 One Step Nucleic Acid Amplification (OSNA) 方法评估。
我们收集了 889 名未接受系统治疗的乳腺癌患者的回顾性数据，这些患者接受了 SLN 活检，并通过 OSNA 评估了所有 SLN。通过 ROC 曲线下面积 (AUC ROC) 评估模型的判别能力，并通过比较模型预测概率 (MPP) 的四分位组的 5 年 DFS Kaplan-Meier 估计值来评估模型的校准。
在训练集中，AUC ROC 范围从 0.78（2 年）到 0.73（5 年），在验证集中，分别从 0.78 到 0.71。MPP 允许区分 DFS 存在异质性的四个患者组（对数秩检验 p<0.0001）。在最高风险组中，训练集的 HR 为 6.04[95%CI 2.70, 13.48]，验证集的 HR 为 4.79[2.310, 9.93]。
该模型成功地使用最严格的验证形式进行了验证，验证中心与模型开发中心不同。该模型的外部验证证实了其对预测 5 年 DFS 的有效性，以及 TTL 值作为预后变量的有用性。
