Plasma extracellular vesicle phenotyping for the differentiation of early-stage lung cancer and benign lung diseases

Abstract

The development of a minimally invasive technique for early-stage lung cancer detection is crucial to reducing mortality. Phenotyping of tumor-associated extracellular vesicles (EVs) has the potential for early-stage lung cancer detection, yet remains challenging due to the lack of sensitive, integrated techniques that can accurately detect rare tumor-associated EV populations in blood. Here, we integrated gold core–silver shell nanoparticles and nanoscopic mixing in a microfluidic assay for sensitive phenotypic analysis of EVs directly in plasma without EV pre-isolation. The assay enabled multiplex detection of lung cancer-associated markers PTX3 and THBS1 and canonical EV marker CD63 by surface-enhanced Raman spectroscopy, providing a squared correlation coefficient of 0.97 in the range of 10³–10⁷ EVs mL⁻¹ and a limit of detection of 19 EVs mL⁻¹. Significantly, our machine learning-based nanostrategy provided 92.3% sensitivity and 100% specificity in differentiating early-stage lung cancer from benign lung diseases, superior to the CT scan-based lung cancer diagnosis (92.3% sensitivity and 71.4% specificity). Overall, our integrated nanostrategy achieved an AUC value of 0.978 in differentiating between early-stage lung cancer patients (n = 28) and controls consisting of patients with benign lung diseases (n = 23) and healthy controls (n = 26), which showed remarkable diagnostic performance and great clinical potential for detecting the early occurrence of lung cancer.