Scholarlink

<p>Pancreatic adenocarcinoma (PAAD) is a highly lethal malignancy that leads to patients missing optimal treatment opportunities due to its atypical clinical symptoms and the lack of effective diagnostic biomarkers. To develop a biomarker panel based on extracellular vesicle-derived transposable elements (EV-TEs) for non-invasive detection of PAAD, we analyzed 6.75 Tbp sequencing data of 852 EV-derived transcriptomes from two cohorts, and identified 31 EV-TEs features as the biomarker panel using recursive feature elimination. Predictive model constructed using the Support Vector Machine (SVM) algorithm demonstrated excellent performance for PAAD detection in the training set (AUC: 0.90, 95% CI: 0.86–0.93), the test set (AUC: 0.86, 95% CI: 0.79–0.92) and the independent external validation cohort of blood EV-derived samples (AUC: 0.88, 95% CI: 0.84–0.92). This study presents the first EV-TEs based predictive model for PAAD detection, showcasing the immense potential of these ‘junk DNA’ as innovative diagnostic biomarker for cancers.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Development and validation of a predictive model based upon extracellular vesicle-derived transposable elements for non-invasive detection of pancreatic adenocarcinoma

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年4月5日

查看全文
  • Yueting Liang,
  • Xin Sui,
  • Shuai Li,
  • Haoxin Peng,
  • Wenyi Jiang,
  • Minqi Jia,
  • Shaoran Jiang,
  • Weihu Wang,
  • Huajing Teng

摘要

Pancreatic adenocarcinoma (PAAD) is a highly lethal malignancy that leads to patients missing optimal treatment opportunities due to its atypical clinical symptoms and the lack of effective diagnostic biomarkers. To develop a biomarker panel based on extracellular vesicle-derived transposable elements (EV-TEs) for non-invasive detection of PAAD, we analyzed 6.75 Tbp sequencing data of 852 EV-derived transcriptomes from two cohorts, and identified 31 EV-TEs features as the biomarker panel using recursive feature elimination. Predictive model constructed using the Support Vector Machine (SVM) algorithm demonstrated excellent performance for PAAD detection in the training set (AUC: 0.90, 95% CI: 0.86–0.93), the test set (AUC: 0.86, 95% CI: 0.79–0.92) and the independent external validation cohort of blood EV-derived samples (AUC: 0.88, 95% CI: 0.84–0.92). This study presents the first EV-TEs based predictive model for PAAD detection, showcasing the immense potential of these ‘junk DNA’ as innovative diagnostic biomarker for cancers.