电子鼻技术作为纤维化间质性肺疾病诊断工具的国际验证
本研究在国际多中心队列中验证电子鼻(eNose)通过分析呼出气挥发性有机物谱区分不同纤维化间质性肺疾病(fILD)亚型的能力。共分析587例经多学科团队(MDT)诊断且高分辨率CT显示肺纤维化患者的呼出气谱,模型在部分中心训练并在其他中心外部验证。结果显示各ILD亚型与其余ILD比较的AUC在训练集为0.88–0.92、验证集为0.75–0.95,6种亚型间判别AUC训练集0.95–0.98、验证集0.83–0.93。
要点
- eNose可在国际多中心外部验证中准确区分不同纤维化ILD亚型。
- 亚型间判别验证集AUC为0.83–0.93,支持其作为无创分诊工具。
- eNose有望成为fILD诊断的便捷床旁辅助手段。
核心结论
- 国际多中心研究证实eNose技术能准确区分不同纤维化ILD患者的呼出气谱,有潜力作为fILD诊断的便捷床旁工具。
研究设计与样本
- 国际多中心队列研究,纳入5个国际ILD专家中心经MDT诊断且高分辨率胸部CT显示肺纤维化的患者,使用SpiroNose进行呼出气分析。
- 共分析587例患者的呼出气谱,模型在部分中心数据上训练并在其他中心外部验证。
主要发现
- 各ILD亚型与其余ILD比较,AUC在训练集为0.88–0.92,验证集为0.75–0.95。
- 6种ILD亚型间判别AUC在训练集为0.95–0.98,验证集为0.83–0.93。
临床怎么用
- eNose可作为无创床旁工具辅助fILD的早期识别与亚型分诊,减少诊断延迟。
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)。
原文
INTRODUCTION: Fibrotic interstitial lung diseases (fILDs) are a heterogeneous group of rare lung diseases. Symptoms of ILD are nonspecific, and diagnosis requires multiple investigations, including invasive procedures. Therefore, diagnostic delay is common. Previous single-center studies showed that profiling of exhaled volatile organic compounds using noninvasive electronic nose (eNose) sensor technology has potential as a diagnostic tool for ILD. We aimed to validate eNose technology to differentiate various ILDs in an international multicenter cohort. METHODS: We included patients with an ILD diagnosis established in a multidisciplinary team (MDT) discussion and pulmonary fibrosis on high-resolution chest computed tomography scans at 5 international ILD expert centers. An eNose (SpiroNose) was used for exhaled breath analysis. We compared eNose breath profiles of different ILD subtypes with all other ILDs combined and across 6 different ILD subtypes. Breath profiles were analyzed with partial least squares discriminant and receiver operating characteristic analyses. Models were trained on data from a selection of centers and externally validated in other centers. RESULTS: Breath profiles of 587 patients were analyzed. Comparing breath profiles of ILD subtypes with all other ILDs resulted in area under the curve (AUC) values ranging from 0.88 to 0.92 in the training set and 0.75 to 0.95 in the validation set. The ILD subtypes could be discriminated with AUCs ranging from 0.95 to 0.98 in the training set and 0.83 to 0.93 in the validation set. DISCUSSION: This international study demonstrates that eNose technology accurately differentiates breath profiles from -patients with various ILDs. eNose technology holds potential as an easy point-of-care tool for the diagnosis of fILDs.
引用
- Formsma BJ, Van Der Sar IG, Yazbeck L, et al. International validation of electronic nose technology as a diagnostic tool for fibrotic interstitial lung diseases. Am J Respir Crit Care Med. 2026. PMID: 41738130. doi:10.1093/ajrccm/aamaf085
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。