文献速递乳腺US

纵向超声Delta影像组学用于乳腺癌新辅助化疗肿瘤反应的早期分层预测

Longitudinal Ultrasound Delta Radiomics for Early Stratified Prediction of Tumor Response to Neoadjuvant Chemotherapy in Breast Cancer.
PubMed 文献 · Acad Radiol · 2025-09-29

本研究纳入233例接受新辅助化疗的乳腺癌患者,基于治疗前及第2周期后超声图像提取影像组学特征,构建Delta_resistance和Delta_PCR两个Delta模型。测试集中两模型均优于单时间点模型,Delta_resistance的AUC为0.917,Delta_PCR的AUC为0.936。SHAP分析揭示了关键影像组学特征,增强了模型的可解释性。

要点

  • Delta_resistance模型测试集AUC 0.917,敏感度0.854,特异度0.846,PPV 0.911。
  • Delta_PCR模型测试集AUC 0.936,敏感度0.884,特异度0.833,NPV 0.857。
  • 两个Delta模型在测试集中均优于单时间点模型。

核心结论

  • 基于纵向超声的Delta影像组学模型可用于早期预测乳腺癌新辅助化疗的肿瘤反应,并可能指导个体化治疗规划。

研究设计与样本

  • 研究纳入233例接受新辅助化疗的乳腺癌患者,分析治疗前及第2周期化疗后超声图像提取影像组学特征,构建Delta_resistance和Delta_PCR两个Delta模型,并以AUC、敏感度、特异度、PPV、NPV和准确率评估模型性能。

诊断标准与关键数值

  • Delta_resistance模型在测试集中预测化疗耐药的AUC为0.917,敏感度0.854,特异度0.846,PPV为0.911。
  • Delta_PCR模型在测试集中预测病理完全缓解的AUC为0.936,敏感度0.884,特异度0.833,PPV为0.864,NPV为0.857。
  • 测试集中两个Delta模型均优于单时间点模型。

临床应用与阅片要点

  • SHAP分析显示关键影像组学特征对模型预测有贡献,增强了模型的透明度和临床可解释性。

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

RATIONALE AND OBJECTIVES: To develop and evaluate Delta-radiomics models based on ultrasound imaging for early prediction and stratification of tumor response to neoadjuvant chemotherapy (NAC) in breast cancer (BC), aiming to support individualized treatment strategies.

METHODS: A total of 233 BC patients undergoing NAC were enrolled. Ultrasound images acquired before treatment and after the second NAC cycle were analyzed to extract radiomic features. Two Delta models were constructed: Delta_resistance for predicting chemotherapy resistance, and Delta_PCR for predicting pathological complete response (PCR). Model performance was evaluated using area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. SHAP (SHapley Additive exPlanations) analysis was applied to improve model interpretability.

RESULTS: In the test set, both Delta models outperformed single time-point models. The Delta_resistance model achieved an AUC of 0.917, sensitivity of 0.854, specificity of 0.846, and PPV of 0.911, demonstrating robust performance in identifying chemoresistant patients. The Delta_PCR model yielded an AUC of 0.936, with balanced sensitivity (0.884), specificity (0.833), PPV (0.864), and NPV (0.857), effectively identifying patients likely to achieve PCR. SHAP analysis further revealed key radiomic features contributing to model predictions, enhancing transparency and clinical interpretability.

CONCLUSION: Delta-radiomics models derived from longitudinal ultrasound imaging provide valuable tools for early prediction of tumor response to NAC. Their strong predictive performance and enhanced interpretability underscore their potential to guide personalized treatment planning and optimize therapeutic outcomes.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Wang M, Huang Z, Tian H, et al. Longitudinal Ultrasound Delta Radiomics for Early Stratified Prediction of Tumor Response to Neoadjuvant Chemotherapy in Breast Cancer. Acad Radiol. 2025. PMID: 41027781. doi:10.1016/j.acra.2025.09.011

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。