Scholarlink

<p>Patients with chronic obstructive pulmonary disease (COPD) are at increased risk of lung cancer, and the identification of circulating tumor DNA (ctDNA) mutations may help detect lung cancer in them. However, whether the addition of ctDNA information improves lung cancer prediction remains to be elucidated. Blood samples from 236 patients with COPD (119 with lung cancer and 117 without lung cancer) were genotyped using targeted deep sequencing. With 40 clinical and genomic/molecular variables, nine machine learning (ML) models to predict lung cancer were constructed. Prediction performances were compared between models with clinical variables only and those with additional genomic/molecular variables. Three clinical variables (smoking amount, C-reactive protein [CRP], COPD symptom burden) and four genomic/molecular variables (ctDNA mutation detected, lung cancer driver gene, max variant allele frequency, and median duplex) were significantly associated with lung cancer (<i>P</i> &lt; 0.05). Notably, four of the nine ML models, incorporating clinical and genomic/molecular variables, outperformed models using clinical variables alone (AUC of the best models = 0.729 vs. 0.620, <i>P</i> &lt; 0.05). Integration of ctDNA information may enhance risk stratification strategies for lung cancer among COPD patients by providing complementary predictive value to conventional clinical assessment.</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

Integration of circulating tumor DNA data enhances lung cancer prediction in patients with COPD

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月3日

查看全文
  • Soojin Cha,
  • Sun Hye Shin,
  • Seung-Ho Shin,
  • Ho Yun Lee,
  • Yeon Jeong Kim,
  • Donghyun Park,
  • Kyung Yeon Han,
  • You Jin Oh,
  • Woong-Yang Park,
  • Myung-Ju Ahn,
  • Hojoong Kim,
  • Hong-Hee Won,
  • Hye Yun Park

摘要

Patients with chronic obstructive pulmonary disease (COPD) are at increased risk of lung cancer, and the identification of circulating tumor DNA (ctDNA) mutations may help detect lung cancer in them. However, whether the addition of ctDNA information improves lung cancer prediction remains to be elucidated. Blood samples from 236 patients with COPD (119 with lung cancer and 117 without lung cancer) were genotyped using targeted deep sequencing. With 40 clinical and genomic/molecular variables, nine machine learning (ML) models to predict lung cancer were constructed. Prediction performances were compared between models with clinical variables only and those with additional genomic/molecular variables. Three clinical variables (smoking amount, C-reactive protein [CRP], COPD symptom burden) and four genomic/molecular variables (ctDNA mutation detected, lung cancer driver gene, max variant allele frequency, and median duplex) were significantly associated with lung cancer (P < 0.05). Notably, four of the nine ML models, incorporating clinical and genomic/molecular variables, outperformed models using clinical variables alone (AUC of the best models = 0.729 vs. 0.620, P < 0.05). Integration of ctDNA information may enhance risk stratification strategies for lung cancer among COPD patients by providing complementary predictive value to conventional clinical assessment.