|
| 基于 LASSO-Cox 回归构建并验证局部晚期鼻咽癌远处转移风险列线图模型：一项基于468例病例的回顾性研究 |
| Development and Validation of a LASSO-Cox Regression-Based Nomogram for Predicting Distant Metastasis in Locally Advanced Nasopharyngeal Carcinoma: A Retrospective Study of 468 Cases |
| 投稿时间：2025-12-25 修订日期：2026-02-06 |
| DOI： |
|
 |
| 中文关键词: 鼻咽癌 局部晚期 远处转移 LASSO-Cox回归 预测模型 |
| 英文关键词:nasopharyngeal carcinoma locally advanced distant metastasis LASSO-Cox regression prediction model |
| 基金项目:国家自然科学(82172804)；江苏省卫健委重点项目(K2019028)；南京市科技计划项目(2022SX00001663)；南京医科大学鼻咽癌专病队列研究(NMUC2021011A) |
|
| 摘要点击次数: 2 |
| 全文下载次数: 0 |
| 中文摘要: |
| [目的] 构建并验证一个整合了传统临床分期、营养状况、炎症指标及治疗方法的预测模型，旨在于精准识别放化疗后的高风险鼻咽癌患者，以指导临床个体化治疗决策的制定。[方法] 本研究回顾性收集2018年1月至2021年12月期间初诊的468例Ⅲ–Ⅳa期鼻咽癌患者资料，按7:3比例随机划分为训练集（n=327）与验证集（n=141）。对所有连续变量（如年龄、血小板与淋巴细胞比值、白蛋白等）进行z-标准化处理，以消除量纲影响。通过最小绝对收缩和选择算子（least absolute shrinkage and selection operator, LASSO）筛选预测变量，并在构建Cox比例风险模型前对连续变量与Log(风险)关系进行检验及必要转换，同时对高度相关的炎症指标进行共线性处理。采用Schoenfeld残差检验验证Cox模型的比例风险假定。采用受试者工作特征（receiver operating characteristic, ROC）曲线确定风险评分截断值，并据此将患者分为高风险组和低风险组。采用Kaplan-Meier法绘制生存曲线，并用Log-rank检验比较两组间无远处转移生存（distant metastasis-free survival, DMFS）的差异。通过校准曲线及决策曲线分析（decision curve analysis, DCA）对模型的准确性和临床实用性进行验证。[结果] 年龄、治疗前血小板与淋巴细胞比值、治疗前白蛋白、治疗前淋巴细胞与中性粒细胞比值、T分期和靶向治疗是DMFS的独立预后因素。构建的列线图模型在训练集和验证集2年、2.5年、3年的ROC曲线下面积（AUC值）分别为0.778、0.747、0.747和0.731、0.724、0.724，校准曲线显示预测与实际结果具有良好一致性，临床决策曲线证实模型具临床实用性。基于风险评分进行分层，识别出高风险与低风险患者，Kaplan-Meier生存曲线表明两组生存差异显著（P<0.0001），表明该模型具有优越的风险分层能力。[结论]基于 LASSO-Cox 回归构建的列线图模型，能够利用局部晚期鼻咽癌患者的治疗前基线特征有效预测接受放化疗患者的远处转移风险，具有良好的预测准确性和潜在临床应用价值。 |
| 英文摘要: |
| [Objective] To develop and validate a predictive model integrating conventional clinical staging, nutritional status, inflammatory markers, and treatment modalities, aimed at precisely identifying high-risk nasopharyngeal carcinoma patients after chemoradiotherapy, thereby guiding individualized clinical decision-making. [Methods] We retrospectively collected data from 468 newly diagnosed stage III–IVa nasopharyngeal carcinoma patients between January 2018 and December 2021, and randomly divided them into a training cohort (n = 327) and a validation cohort (n = 141) at a 7:3 ratio. Continuous variables, including age, platelet-to-lymphocyte ratio, and serum albumin, were standardized using z-transformation to eliminate dimensional effects. Predictive variables were selected using the least absolute shrinkage and selection operator (LASSO). Prior to constructing the Cox proportional hazards model, the relationships between continuous variables and log(relative risk) were assessed and transformed as necessary, and multicollinearity among highly correlated inflammatory markers was addressed. The proportional hazards assumption of the Cox model was verified using Schoenfeld residuals. Risk score cutoffs were determined using receiver operating characteristic (ROC) curves, and patients were stratified into high-risk and low-risk groups accordingly. Kaplan-Meier survival curves were generated, and the log-rank test was used to compare distant metastasis-free survival (DMFS) between groups. Model accuracy and clinical utility were evaluated using calibration curves and decision curve analysis (DCA). [Results] Age, pre-treatment platelet-to-lymphocyte ratio, pre-treatment serum albumin, pre-treatment lymphocyte-to-neutrophil ratio, T stage, and targeted therapy were identified as independent prognostic factors for DMFS. The constructed nomogram demonstrated area under the ROC curve (AUC values) of 0.778, 0.747, and 0.747 at 2, 2.5, and 3 years in the training cohort, and 0.731, 0.724, and 0.724 in the validation cohort, respectively. Calibration curves indicated good agreement between predicted and observed outcomes, and DCA confirmed the model's clinical applicability. Risk stratification based on the risk scores effectively distinguished high-risk from low-risk patients, with Kaplan-Meier survival analysis showing significant differences between groups (P<0.0001), indicating excellent risk stratification capability. [Conclusion] The LASSO-Cox-based nomogram model effectively predicts the risk of distant metastasis in locally advanced NPC patients receiving chemoradiotherapy using baseline pre-treatment features, demonstrating good predictive accuracy and potential clinical applicability. |
|
在线阅读
查看/发表评论 下载PDF阅读器 |
|
|
|