Entropy-driven signal amplification integrated with machine learning in multiplex lateral flow immunoassay for sensitive Point-of-Care colon cancer diagnosis
摘要
Investigations on epithelial-mesenchymal transition (EMT) events occurring on circulating tumor cells (CTCs) are poised to significantly advance nanoliquid biopsy methodologies. This study presented a colorimetric multiplex lateral flow immunoassay strip (mLFIA) for the simultaneous detection of EpCAM and Vimentin, two representative EMT biomarkers, on heterogeneous CTCs. The assay utilized an engineered magnetic probe (pCF-Apt-H1/H2), characterized by a zwitterionic polymer coating on Fe3O4 nanoparticles, which conferred anti-fouling properties. This probe also incorporated dual aptamers specific for EpCAM and Vimentin for recognition, along with two complementary strands (H1/H2) that enabled fluorescent detection. Upon recognition, a cascade of strand displacement and entropy-driven amplification reactions was initiated, generating abundant single-stranded DNA co-labeled with biotin and fluorescein, which were subsequently detected by a gold nanoparticle-based mLFIA strip. Coupled with a custom-fabricated 3D-printed reader, the platform facilitated a rapid colorimetric readout for EpCAM and Vimentin, achieving low limits of detection at 0.22 and 0.16 ng/mL, respectively. Remarkably, this assay also demonstrated the capability to detect these two biomarkers on CTCs down to 10 cells/mL, and to track dynamic EMT processes. Machine learning algorithms were subsequently employed to classify and predict the colorimetric signals. Ultimately, this sensing platform proved effective in differentiating healthy individuals from colon cancer patients.
Graphical Abstract