文献速递头颈神经MR

基于MRI的影像组学预测鼻咽癌转移性颈部淋巴结放化疗反应

MRI-based radiomics as response predictor to radiochemotherapy for metastatic cervical lymph node in nasopharyngeal carcinoma.
PubMed 文献 · Br J Radiol · 2021-04-21

本研究纳入145例鼻咽癌患者(训练集102例,验证集43例),从CE-T1WI和T2WI提取2704个影像组学特征构建预测转移性颈部淋巴结放化疗反应的模型。联合序列模型在训练集AUC为0.927,验证集AUC为0.772,敏感度、特异度、准确度在验证集分别为0.792、0.790、0.791。结论提示MRI影像组学模型有望在治疗前预测疗效,辅助个体化治疗。

要点

  • 联合CE-T1WI与T2WI的影像组学模型可预测鼻咽癌转移性淋巴结放化疗反应。
  • 验证集AUC为0.772,敏感度0.792、特异度0.790,具有一定泛化能力。
  • CE-T1WI、T2WI及联合序列模型间AUC无显著差异(P>0.05)。

核心结论

  • 基于MRI的影像组学模型可预测鼻咽癌转移性颈部淋巴结对放化疗的治疗反应,可能有助于治疗前个体化治疗决策。

研究设计与样本

  • 回顾性纳入145例连续鼻咽癌患者,训练集102例、验证集43例。
  • 转移性淋巴结依据影像学标准诊断,治疗反应依据RECIST标准评估。
  • 从CE-T1WI和T2WI分别提取2704个影像组学特征,构建CE-T1WI、T2WI及联合序列模型。

诊断标准与关键数值

  • 联合CE-T1WI和T2WI模型筛选出12个特征构建影像组学标签。
  • CE-T1WI、T2WI及联合序列模型间AUC在训练集和验证集均无显著差异(P>0.05)。

临床应用与阅片要点

  • 治疗前可基于CE-T1WI和T2WI提取影像组学特征,辅助预测转移性淋巴结放化疗反应。
  • 联合序列模型在验证集AUC为0.772,提示有一定预测价值但需进一步外部验证。
  • 注意本研究为单中心小样本,模型泛化能力有限,不应单独用于临床决策。

局限性

  • 验证集样本量较小(43例),可能影响模型性能评估的稳定性。
  • 本研究为单中心回顾性设计,未进行外部多中心验证。
  • 原文未报告影像组学特征的具体名称和阈值,可重复性受限。

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

OBJECTIVE: To establish and substantiate MRI-based radiomic models to predict the treatment response of metastatic cervical lymph node to radiochemotherapy in patients with nasopharyngeal carcinoma (NPC).

METHODS: A total of 145 consecutive patients with NPC were enrolled including 102 in primary cohort and 43 in validation cohort. Metastatic lymph nodes were diagnosed according to radiologic criteria and treatment response was evaluated according to the Response Evaluation Criteria in Solid Tumors. A total of 2704 radiomic features were extracted from contrast-enhanced T1 weighted imaging (CE- T1WI) and T2 weighted imaging (T2WI) for each patient, and were selected to construct radiomic signatures for CE-T1WI, T2WI, and combined CE-T1WI and T2WI, respectively. The area under curve (AUC) of receiver operating characteristic, sensitivity, specificity, and accuracy were used to estimate the performance of these radiomic models in predicting treatment response of metastatic lymph node.

RESULTS: No significant difference of AUC was found among radiomic signatures of CE-T1WI, T2WI, and combined CE-T1WI and T2WI in the primary and validation cohorts (all p > 0.05). For combined CE-T1WI and T2WI data set, 12 features were selected to develop the radiomic signature. The AUC, sensitivity, specificity, and accuracy were 0.927 (0.878-0.975), 0.911 (0.804-0.970), 0.826 (0.686-0.922), and 0.872 (0.792-0.930) in primary cohort, and were 0.772 (0.624-0.920), 0.792 (0.578-0.929), 0.790 (0.544-0.939), and 0.791 (0.640-0.900) in validation cohort.

CONCLUSION: MRI-based radiomic models were developed to predict the treatment response of metastatic cervical lymph nodes to radiochemotherapy in patients with NPC, which might facilitate individualized therapy for metastatic lymph nodes before treatment.

ADVANCES IN KNOWLEDGE: Predicting the response in patients with NPC before treatment may allow more individualizing therapeutic strategy and avoid unnecessary side-effects and costs. Radiomic features extracted from metastatic cervical lymph nodes showed promising application for predicting the treatment response in NPC.

来源:Br J Radiol,查看原文。版权归原作者所有。

引用

  1. Xu H, Liu J, Huang Y, et al. MRI-based radiomics as response predictor to radiochemotherapy for metastatic cervical lymph node in nasopharyngeal carcinoma. Br J Radiol. 2021. PMID: 33882240. doi:10.1259/bjr.20201212

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。