Accurate Diagnosis of Colorectal Cancer Using a Combination of Lectin-Induced Recombinase Polymerase Amplification and CRISPR/Cas12a Assay on a Point-of-Care Testing Platform with Deep Learning Assistant

Specific glycosylation patterns on exosome surfaces represent novel diagnostic biomarkers for cancer liquid biopsy. Lectins can induce exosome aggregation through multiple bindings with exosomal glycoproteins. In this work, we developed a one-pot lectin-induced recombinase polymerase amplification (RPA) and CRISPR/Cas12a-mediated cleavage assay (LI-RPA-CRISPR/Cas12a) for diagnosing colorectal cancer (CRC) through the interactions of abundant α-fucose residues on CRC cell-derived exosome surfaces with Ulex Europaeus Agglutinin I (UEA-I). The combination of a homemade portable isothermal amplification device, the as-proposed LI-RPA-CRISPR/Cas12a exhibits a wide detection range from 2 × 10⁶ to 1 × 10² extracellular vehicles (EVs) μL^-1 with a visual limit of detection (LOD) as low as 1.0 × 10² EVs μL^-1, and has been successfully utilized to dynamically monitor the progression of tumors in mice-bearing SW480 CRC subtype at an early stage. After integration with a long short-term memory (LSTM) deep learning model, the LI-RPA-CRISPR/Cas12a achieves accurate diagnosis of primary colorectal cancer with a drop of blood through a smartphone-based data analysis application, reaching an accuracy of 95% in 100 clinical samples. This rapid, sensitive, and user-friendly approach provides a promising platform for point-of-care testing (POCT) diagnosis of CRC, enabling early detection and monitoring of disease progression through a minimally invasive liquid biopsy.