医学
数字聚合酶链反应
多路复用
多重聚合酶链反应
聚合酶链反应
核酸
DNA
肺结核
实时聚合酶链反应
胸膜炎
病理
免疫学
病毒学
核酸扩增试验
结核分枝杆菌
癌症研究
杂交探针
作者
S. Zhang,Ye Xu,Mingxiang Huang,Yuying Pan,J. Shangguan,Rui Peng,Xiaoman Hu,Fang Zheng,Xingqiang Lü,Xiaohong Chen,Qingge Li
出处
期刊:Thorax
[BMJ]
日期:2025-12-09
卷期号:: thorax-2025
被引量:2
标识
DOI:10.1136/thorax-2025-223553
摘要
BACKGROUND: (MTB) in pleural fluid. This study aims to characterise MTB-derived cell-free DNA (cfDNA) in pleural effusion and to develop an optimised molecular assay to improve TBP diagnostic accuracy. METHODS: We quantified MTB cfDNA/genomic DNA (gDNA) concentrations and characterised cfDNA fragment profiles in pleural effusion specimens using ddPCR. A multiplex droplet digital PCR (ddPCR) assay was developed, targeting two insertion sequences (IS6110 and IS1081) using ultra-short amplicons (49-59 bp) to enhance detection efficiency. Diagnostic performance was prospectively evaluated in 356 consecutive adults with radiologically confirmed pleural effusion, using composite microbiological criteria as the reference standard. RESULTS: MTB cfDNA exhibited significantly higher detection rates than gDNA in definite TBP cases (p=0.0006), with dominant fragment lengths of 60-80 bp. The multiplex ddPCR assay demonstrated a limit of detection of 0.2 genome equivalents per reaction. Among microbiologically confirmed TBP cases (n=69), the assay achieved a sensitivity of 94.2%, significantly outperforming Xpert MTB/RIF (52.0%, p<0.0001) and liquid culture (35.3%, p<0.0001). Specificity remained high at 97.0% among cases of non-TB effusion (n=208). CONCLUSIONS: Our findings establish MTB cfDNA as the predominant nucleic acid form in tuberculous pleural effusion. The optimised multiplex ddPCR platform, leveraging multicopy targets and fragment-length-adapted amplification, achieves improved diagnostic sensitivity without compromising specificity in a high-prevalence TB setting. This approach may help address key limitations of TBP diagnostics and shows promise for clinical implementation.
科研通智能强力驱动
Strongly Powered by AbleSci AI