基于CT的影像组学预测根治性切除后局部晚期非小细胞肺癌患者脑转移作为首次失败模式
本研究纳入124例根治性切除的IIB-IIIB期非小细胞肺癌患者,基于术前胸部CT提取影像组学特征构建无脑转移生存预测模型。整合列线图在训练集和验证集的C-index分别为0.889和0.853,显著优于临床列线图,校准曲线和决策曲线分析显示其预测性能更优。
要点
- 术前胸部CT影像组学特征与根治性切除后LA-NSCLC患者的无脑转移生存显著相关。
- 整合列线图预测无脑转移生存的区分度优于临床列线图,训练集和验证集均达统计学显著。
- 整合列线图在验证集中较影像组学列线图的改善未达统计学显著(p=0.0554)。
核心结论
- 与临床或影像组学列线图相比,整合列线图的预测性能显著提高,最适合用于预测根治性切除后LA-NSCLC患者的无脑转移生存。
研究设计与样本
- 本研究为预测模型开发与验证研究,纳入本机构2014年1月至2018年6月间124例根治性切除的IIB-IIIB期NSCLC患者。
- 患者按4:1比例随机分为训练集和验证集,从术前胸部CT图像中筛选影像组学特征。
- 基于训练集采用LASSO算法构建影像组学标签,采用Cox比例风险模型开发临床模型,并构建临床、影像组学和整合列线图。
- 通过区分度、校准度和临床实用性评估模型预测性能。
诊断标准与关键数值
- 影像组学标签在总体队列中与无脑转移生存显著相关。
- 整合列线图在训练集的C-index为0.889(95%CI 0.872-0.906),在验证集为0.853(95%CI 0.788-0.918)。
- 整合列线图区分度显著优于临床列线图,训练集p<0.0001,验证集p=0.0008。
- 整合列线图较影像组学列线图有不同程度改善,训练集p=0.0007,验证集p=0.0554(未达统计学显著)。
- 校准曲线和决策曲线分析显示整合列线图在预测无脑转移生存方面优于临床或影像组学列线图。
临床应用与阅片要点
- 作者认为整合列线图最适合用于预测根治性切除后LA-NSCLC患者的无脑转移生存。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
PURPOSE: Brain metastasis (BM) is the primary first failure pattern in patients with curatively resected locally advanced non-small cell lung cancer (LA-NSCLC). It is not yet possible to accurately predict the occurrence of BM. The purpose of the research is to develop and validate a prediction model of BM-free survival based on radiomics characterising the primary lesions combined with clinical characteristics in patients with curatively resected LA-NSCLC.
METHODS: This study consisted of 124 patients with curatively resected stage IIB-IIIB NSCLC in our institution between January 2014 and June 2018. Patients were randomly divided into training and validation cohorts using a 4:1 ratio. Radiomics features were selected from the chest CT images before surgery. A radiomics signature was constructed using the LASSO algorithm based on the training cohort. Clinical model was developed using the Cox proportional hazards model. The clinical, radiomics, and integrated nomograms were constructed. The prediction performance of the models was assessed based on its discrimination, calibration, and clinical utility.
RESULTS: The radiomics signature is significantly associated with BM-free survival in the overall cohort. The discrimination performance of the integrated nomogram, with the C-indexes 0.889 (0.872-0.906, 95 % CI) and 0.853 (0.788-0.918, 95 % CI) in the training and validation cohorts, respectively, is significantly better than the clinical nomogram (p < 0.0001 for the training cohort, p = 0.0008 for the validation cohort). Compared with the radiomics nomogram, the integrated nomogram is also improved to varying degrees, but not apparent in the validation cohort (p = 0.0007 for the training cohort, p = 0.0554 for the validation cohort). The calibration curve and decision curve analysis demonstrated that the integrated nomogram exceeded the clinical or radiomics nomograms in predicting BM-free survival.
CONCLUSIONS: Compared with the clinical or radiomics nomograms, the predictive performance of the integrated nomogram is significantly improved. The integrated nomogram is most suitable for predicting BM-free survival in patients with curatively resected LA-NSCLC.
引用
- Sun F, Chen Y, Chen X, et al. CT-based radiomics for predicting brain metastases as the first failure in patients with curatively resected locally advanced non-small cell lung cancer. Eur J Radiol. 2020. PMID: 33246270. doi:10.1016/j.ejrad.2020.109411
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。