Su Yanping, Lu Chenying, Zheng Shenfei, Zou Hao, Shen Lin, Yu Junchao, Weng Qiaoyou, Wang Zufei, Chen Minjiang, Zhang Ran, Ji Jiansong, Wang Meihao
Key Laboratory of Imaging Diagnosis and Minimally Invasive Intervention Research, Institute of Imaging Diagnosis and Minimally Invasive Intervention Research, The Fifth Affiliated Hospital of Wenzhou Medical University, Lishui, Zhejiang, China.
Department of Radiology, Key Laboratory of Intelligent Medical Imaging of Wenzhou, Institute of Aging, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou Medical University, Wenzhou, Zhejiang, China.
Front Oncol. 2023 Mar 16;13:1006172. doi: 10.3389/fonc.2023.1006172. eCollection 2023.
To develop and validate a CT-based radiomics nomogram that can provide individualized pretreatment prediction of the response to platinum treatment in small cell lung cancer (SCLC).
A total of 134 SCLC patients who were treated with platinum as a first-line therapy were eligible for this study, including 51 patients with platinum resistance (PR) and 83 patients with platinum sensitivity (PS). The variance threshold, SelectKBest, and least absolute shrinkage and selection operator (LASSO) were applied for feature selection and model construction. The selected texture features were calculated to obtain the radiomics score (Rad-score), and the predictive nomogram model was composed of the Rad-score and the clinical features selected by multivariate analysis. Receiver operating characteristic (ROC) curves, calibration curves, and decision curves were used to assess the performance of the nomogram.
The Rad-score was calculated using 10 radiomic features, and the resulting radiomics signature demonstrated good discrimination in both the training set (area under the curve [AUC], 0.727; 95% confidence interval [CI], 0.627-0.809) and the validation set (AUC, 0.723; 95% CI, 0.562-0.799). To improve diagnostic effectiveness, the Rad-score created a novel prediction nomogram by combining CA125 and CA72-4. The radiomics nomogram showed good calibration and discrimination in the training set (AUC, 0.900; 95% CI, 0.844-0.947) and the validation set (AUC, 0.838; 95% CI, 0.534-0.735). The radiomics nomogram proved to be clinically beneficial based on decision curve analysis.
We developed and validated a radiomics nomogram model for predicting the response to platinum in SCLC patients. The outcomes of this model can provide useful suggestions for the development of tailored and customized second-line chemotherapy regimens.
开发并验证一种基于CT的放射组学列线图，用于对小细胞肺癌（SCLC）患者铂类治疗反应进行个体化的治疗前预测。
本研究纳入134例接受铂类一线治疗的SCLC患者，其中51例铂类耐药（PR）患者，83例铂类敏感（PS）患者。采用方差阈值法、SelectKBest法和最小绝对收缩和选择算子（LASSO）进行特征选择和模型构建。计算所选纹理特征以获得放射组学评分（Rad-score），预测列线图模型由Rad-score和多变量分析选择的临床特征组成。采用受试者操作特征（ROC）曲线、校准曲线和决策曲线评估列线图的性能。
使用10个放射组学特征计算Rad-score，所得放射组学特征在训练集（曲线下面积[AUC]，0.727；95%置信区间[CI]，0.627 - 0.809）和验证集（AUC，0.723；95% CI，0.562 - 0.799）中均显示出良好的区分度。为提高诊断效能，Rad-score通过结合CA125和CA72-4创建了一种新型预测列线图。放射组学列线图在训练集（AUC，0.900；95% CI，0.844 - 0.947）和验证集（AUC，0.838；95% CI，0.534 - 0.735）中显示出良好的校准和区分度。基于决策曲线分析，放射组学列线图被证明具有临床实用性。
我们开发并验证了一种用于预测SCLC患者铂类治疗反应的放射组学列线图模型。该模型的结果可为制定个性化的二线化疗方案提供有用的建议。
