文献速递腹部技术与安全CT

基于基线双层光谱CT的栖息地分析用于胰腺癌根治性切除术后复发术前预测及其与肿瘤-间质比的关系

Baseline dual-layer spectral CT-based habitat analysis for preoperative prediction of recurrence in pancreatic cancer after radical resection and its association with tumor-stroma ratio.
PubMed 文献 · Abdom Radiol (NY) · 2025-12-01

回顾性纳入136例可切除胰导管腺癌患者,利用基线双层光谱CT生成细胞外容积分数和动脉强化分数图,经K-means聚类构建栖息地参数并建立复发预测模型。联合模型(栖息地参数+CA19-9+环状强化)在训练和验证队列的c-index分别为0.912和0.899,时间依赖ROC的AUC均高于0.85。栖息地参数f1和f4与病理肿瘤-间质比显著相关,可定量预测肿瘤-间质比。

要点

  • 联合模型整合栖息地参数f1、f4、CA19-9>180 U/mL和环状强化,术前预测PDAC根治术后复发。
  • 联合模型训练队列c-index 0.912、验证队列0.899,时间依赖ROC的AUC均高于0.85。
  • 栖息地参数f1和f4与肿瘤-间质比显著相关,可无创定量预测病理肿瘤-间质比。

核心结论

  • 联合模型整合栖息地定量参数、CA19-9和环状强化,为术前无创预测PDAC复发风险提供了方法;栖息地定量参数可无创定量预测病理肿瘤-间质比。

研究设计与样本

  • 回顾性纳入136例术前接受多期双层光谱CT检查的可切除PDAC患者,随机分配至训练队列和验证队列。

关键影像表现/征象

  • 基于细胞外容积分数和动脉强化分数图,通过K-means算法将肿瘤体素聚类为不同栖息地并提取定量参数。
  • 环状强化被纳入联合模型构建。

诊断标准与关键数值

  • 联合模型在训练队列c-index为0.912,验证队列为0.899。
  • 时间依赖ROC分析显示联合模型预测无复发生存期的AUC均高于0.85。
  • 训练和验证队列中,预测高风险组无复发生存期均显著短于预测低风险组(均p<0.001)。

临床应用与阅片要点

  • 联合模型使用栖息地参数f1、f4、CA19-9>180 U/mL和环状强化进行术前复发风险预测。
  • 栖息地参数f1和f4与肿瘤-间质比显著相关,可用于定量预测肿瘤-间质比。

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: To investigate the value of habitat imaging employing baseline dual-layer spectral CT (DLCT) for preoperative prediction of recurrence in pancreatic ductal adenocarcinoma (PDAC) after radical resection, and explore the relationship with pathological tumor-stroma ratio (TSR).

METHODS: Resectable PDAC patients underwent multiphase DLCT examinations preoperatively were retrospectively enrolled and randomly allocated into training and validation cohorts. Extracellular volume (ECV) fraction and arterial enhancement fraction (AEF) maps were generated using spectral-based images. Voxels of tumor from ECV and AEF maps were clustered into different habitats through K-means algorithm. Habitat quantitative parameters were extracted. Clinical-radiological, habitat, and combined models for predicting recurrence free survival (RFS) were constructed using Cox regression analyses. Model performance was assessed through c-index and time-dependent receiver operating characteristic (ROC) analysis. Kaplan-Meier method was used to assess recurrence rate. Spearman's correlation analysis and multiple linear regression were used to evaluate the relationship between TSR and habitat parameters and build prediction model.

RESULTS: A total of 136 patients were finally included. The fraction of habitat 1 (f1), fraction of habitat 4 (f4), CA19-9 > 180 U/mL, and rim-enhancement were used to construct the combined model. Combined model demonstrated superior performance than clinical-radiological model with c-index of 0.912 and 0.899 in training and validation cohorts, respectively. Time-dependent ROC analysis revealed areas under the curve of the combined model for predicting RFS were all above 0.85. The predicted-high-risk group had significantly shorter RFS than predicted-low-risk group in both training and validation group (both p < 0.001). f1 and f4 were significantly associated with TSR and could be used to predict TSR quantitatively.

CONCLUSION: The combined model, integrating habitat quantitative parameters, CA19-9 and rim-enhancement, provides a noninvasive approach for predicting the risk of recurrence in PDAC preoperatively. Habitat quantitative parameters could be used to quantitative predict pathological TSR noninvasively.

来源:Abdom Radiol (NY),查看原文。版权归原作者所有。

引用

  1. Cai W, Zhu Y, Li D, et al. Baseline dual-layer spectral CT-based habitat analysis for preoperative prediction of recurrence in pancreatic cancer after radical resection and its association with tumor-stroma ratio. Abdom Radiol (NY). 2025. PMID: 41320688. doi:10.1007/s00261-025-05318-8

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。