Novel Nomograms as Aids for Predicting Recurrence and Survival in Chordoma Patients A Retrospective Multicenter Study in mainland China
Novel Nomograms as Aids for Predicting Recurrence and Survival in Chordoma Patients A Retrospective Multicenter Study in mainland China
复制标题
新型列线图有助于预测脊索瘤患者的复发和生存：中国大陆的一项回顾性多中心研究。
DOI：
10.1097/brs.0000000000003716
复制
发表时间：
2021-01-01
期刊：
影响因子：
3
通讯作者：
Song, Dianwen
中科院分区：
文献类型：
--
作者：
Meng, Tong;Huang, Runzhi;Song, Dianwen
研究设计.进行了回顾性数据分析。Objective.本研究的目的是探讨重要的预后因素，并提出新的诺模图，以促进临床决策。背景数据总结。软骨瘤是一种罕见的骨肿瘤。其临床特征和最佳治疗策略仍不确定。方法.纳入了2015年1月之前在中国大陆四家医疗中心接受治疗的脉络膜炎患者。采用Lasso回归和考克斯比例风险回归模型分析局部无复发生存期（LRFS）和总生存期（OS）的预测因素。然后绘制列线图。分别用C指数、校准曲线和受试者工作特征曲线（ROC）评价其鉴别力、校准和准确性。结果共确定了341例患者，并获得了276例患者的完整预后变量数据。共有179例患者（64.9%）复发，122例患者（44.2%）死于各种原因，中位随访时间为57.5（范围，1-325）个月。我们确定了肿瘤大小、肿瘤位置、组织学亚型和切除方法的复发相关因素，以及肿瘤大小、肿瘤位置、切除方法、并发症和术后复发的死亡相关因素。所构建的LRFS和OS列线图显示出良好的校准和区分能力（C指数分别为0.79和0.76）。ROC显示良好的预测能力，5年曲线下面积（AUC）值分别为0.868和0.786。结论基于一个多中心的病例系列的脉络膜与一个相对较长的后续行动，我们提出了两个诺模图的基础上预测预后的复发和死亡的相关因素。这些发现可作为临床决策过程的参考，并为风险分层提供额外的预后信息。
Study Design. A retrospective data analysis was performed. Objective. The aim of this study is to explore the significant prognostic factors and propose new nomograms to facilitate clinical decision-making. Summary of Background Data. Chordoma is a rare bone tumor. The clinical features and optimal therapeutic strategies are still uncertain. Methods. Chordoma patients treated in four medical centers of mainland China before January 2015 were included. The predictors for local relapse-free survival (LRFS) and overall survival (OS) were identified by the Lasso regression and Cox proportional hazards regression model. Then the nomograms were developed. Their discrimination, calibration, and accuracy were evaluated by the C-index, calibration curve, and receiver operating characteristic curve (ROC), respectively. Results. A total of 341 patients were identified and full prognostic variable data were available for 276 patients. A total of 179 patients (64.9%) experienced recurrence and 122 patients (44.2%) died of all causes with a median follow-up time of 57.5 (range, 1-325) months. We identified recurrence-relevant factors of tumor size, tumor location, histology subtype and resection method, and death-relevant factors of tumor size, tumor location, resection method, complication, and postoperative recurrence. The constructed LRFS and OS nomograms showed good calibration and discriminative ability (C index 0.79 and 0.76, respectively). The ROCs suggested decent prediction ability with the 5-year area under curve (AUC) value of 0.868 and 0.786, respectively. Conclusion. Based on the multicenter case series of chordoma with a relative long follow-up, we proposed two nomograms to predict the prognosis on the basis of recurrence- and death-relevant factors. These findings could be referenced in the clinical decision-making process and provide additional prognostic information for risk stratification.