基于18F-氟脱氧葡萄糖正电子发射断层扫描/计算机断层扫描影像联合临床特征的列线图预测IASLC分级系统中肺腺癌高级别模式
本研究纳入1327例接受术前18F-FDG PET/CT的浸润性肺腺癌患者,随机分为训练队列(n=928)和测试队列(n=399)。PET/CT影像-临床联合模型预测高级别模式的AUC在训练队列为0.837,优于PET、CT、PET/CT和临床模型,测试队列AUC为0.834。结论认为该联合模型有效,衍生的列线图有助于识别可从辅助治疗获益的患者。
要点
- PET/CT影像-临床联合模型预测IAC高级别模式训练队列AUC=0.837,测试队列AUC=0.834。
- 联合模型在训练队列优于PET(AUC=0.789,P<.001)、CT(AUC=0.780,P<.001)、PET/CT(AUC=0.817,P=.002)和临床模型(AUC=0.744,P<.001)。
- 列线图有助于识别可能从辅助治疗获益的IAC患者。
核心结论
- 18F-FDG PET/CT影像-临床联合模型可有效预测IAC患者的高级别模式,衍生的列线图有助于识别可从辅助治疗获益的患者。
研究设计与样本
- 研究纳入1327例接受术前18F-FDG PET/CT的浸润性肺腺癌患者,随机分为训练队列(n=928)和测试队列(n=399),采用单因素和多因素分析筛选独立临床与影像变量,逻辑回归构建预测模型。
诊断标准与关键数值
- 训练队列中,PET/CT影像-临床联合模型预测高级别模式的AUC=0.837,优于PET模型(AUC=0.789,P<.001)、CT模型(AUC=0.780,P<.001)、PET/CT模型(AUC=0.817,P=.002)和临床模型(AUC=0.744,P<.001)。
- 测试队列中,PET/CT影像-临床联合模型AUC=0.834,优于其他模型(所有P<.05)。
- 校准曲线显示训练队列和测试队列中模型预测与实际观察一致性良好(P>.05)。
临床应用与阅片要点
- 作者认为18F-FDG PET/CT影像-临床联合模型可有效预测IAC高级别模式,衍生的列线图有助于识别可从辅助治疗获益的患者。
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
AIM: The new International Association for the Study of Lung Cancer (IASLC) grading system for invasive lung adenocarcinoma (IAC) has demonstrated significant prognostic efficacy, benefiting many patients with adjuvant chemotherapy. We aimed to develop a nomogram integrating 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) imaging and clinical features for preoperative prediction of IASLC grading.
MATERIALS AND METHODS: This study included 1327 IAC patients who underwent preoperative 18F-FDG PET/CT, who were randomly divided into a training cohort (n=928) and a test cohort (n=399). Univariate and multivariate analyses identified independent clinical and imaging variables. Logistic regression analysis was performed to calculate the independent risk factors and construct models for predicting the high-grade pattern of IAC. Model performance was evaluated using receiver operating characteristic curves and accuracy assessments. Subsequently, a nomogram based on PET/CT imaging-clinical model was established. Decision curve analysis and calibration curves assessed clinical utility.
RESULTS: In the training cohort, PET/CT imaging-clinical combined model effectively predicted patients with a high-grade pattern (area under the curve [AUC] = 0.837), outperforming PET (AUC = 0.789, P<.001), CT (AUC = 0.780, P<.001), PET/CT (AUC = 0.817, P=.002), and clinical (AUC = 0.744, P<.001) models. In the test cohort, similar superiority of PET/CT imaging-clinical combined model (AUC = 0.834) over other models was also observed (all P<.05). The calibration curve suggested good coherence between model predictions and actual observations in both training and test cohorts (P>.05). Good decision curve analysis demonstrated the clinical usefulness of the nomogram.
CONCLUSION: The 18F-FDG PET/CT imaging-clinical combined model is effective in predicting a high-grade pattern in IAC patients. The derived nomogram aids in identifying patients who would benefit from adjuvant therapy.
引用
- Fu S, Sun J, You J, et al. Predicting a high-grade pattern in IASLC grading system of lung adenocarcinoma using 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) imaging combined with clinical features: a nomogram approach. Clin Radiol. 2025. PMID: 41175796. doi:10.1016/j.crad.2025.107112
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。