基于自适应Lasso和Cox回归的食管癌风险建模

Risk modeling for esophageal cancer based on adaptive Lasso and Cox regression.

作者信息

Li Xiaoli, Han Gaoyong, Yang Yudan, Liang Enhao

机构信息

Department of Oncology, the First Affiliated Hospital of Zhengzhou University, Zhengzhou, China.

School of Automation and Electrical Engineering, Zhongyuan University of Technology, Zhengzhou, China.

出版信息

Front Oncol. 2025 Aug 1;15:1609540. doi: 10.3389/fonc.2025.1609540. eCollection 2025.

Abstract

INTRODUCTION

Esophageal cancer (EC) is one of the most aggressive tumor types worldwide, and malnutrition is extremely common among EC patients. By identifying EC biomarkers and conducting risk assessments on patients, more accurate diagnosis and treatment plans can be developed to prolong patients' survival.

METHODS

This study developed a risk assessment model for post-surgical EC patients using clinical data from patients who underwent esophagectomy. Prognostic factors influencing survival were evaluated using Adaptive Lasso for variable selection, followed by Cox proportional hazards regression and Receiver Operating Characteristic (ROC) curve. Among multiple clinical variables, the International Normalized Ratio (INR) emerged as the most significant predictor of survival.

RESULTS

Elevated INR levels were significantly associated with improved 3-year and 5-year survival outcomes compared to the Prognostic Nutritional Index (PNI). Patients with higher INR exhibited notably better postoperative survival rates. Further analysis demonstrated that INR was significantly correlated with the final differentiation degree, final infiltration degree, and final positive/negative status of EC.

DISCUSSION

INR serves as a valuable and independent prognostic biomarker for postoperative survival assessment in EC patients. Incorporating INR into clinical risk models can enhance the accuracy of prognosis and assist clinicians in optimizing individualized therapeutic strategies for surgical EC patients.

摘要

引言

食管癌(EC)是全球最具侵袭性的肿瘤类型之一,营养不良在EC患者中极为常见。通过识别EC生物标志物并对患者进行风险评估,可以制定更准确的诊断和治疗方案,以延长患者的生存期。

方法

本研究利用接受食管切除术患者的临床数据,为术后EC患者建立了风险评估模型。使用自适应套索法进行变量选择,评估影响生存的预后因素,随后进行Cox比例风险回归和受试者工作特征(ROC)曲线分析。在多个临床变量中,国际标准化比值(INR)成为生存的最显著预测因素。

结果

与预后营养指数(PNI)相比,INR水平升高与3年和5年生存结果改善显著相关。INR较高的患者术后生存率明显更高。进一步分析表明,INR与EC的最终分化程度、最终浸润程度以及最终阳性/阴性状态显著相关。

讨论

INR是评估EC患者术后生存的有价值且独立的预后生物标志物。将INR纳入临床风险模型可以提高预后准确性,并协助临床医生为手术EC患者优化个体化治疗策略。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/a7a0/12355212/837283e53e40/fonc-15-1609540-g001.jpg

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索