Simple and Isolation-Free Analysis of Tumor-Derived Extracellular Vesicles by Aggregation-Modulated Deterministic Lateral Displacement for Breast Cancer Classification

Abstract Tumor-derived extracellular vesicles (EVs) are increasingly recognized as potential biomarkers for breast cancer (BC) diagnosis and molecular subtyping; however, most current analytical methods rely on laborious EV isolation and sophisticated instrumentation, limiting their routine clinical implementation. Herein, we report an aggregation-modulated deterministic lateral displacement (AM-DLD) platform for simple, label-free analysis of tumor-derived EVs without EV isolation. In this strategy, aptamer-functionalized polystyrene microspheres convert nanoscale EV recognition into microscale microsphere aggregates through target EV-mediated cross-linking, enabling direct analysis using a conventional DLD chip. Because EV-induced aggregation generates microsphere aggregates with heterogeneous sizes, a weighted lateral displacement index (LDI) was developed to integrate particle distributions across multiple DLD outlets into a single quantitative readout. Using EpCAM-positive EVs for BC detection and PD-L1-positive EVs for triple-negative BC (TNBC) classification, the AM-DLD platform achieved limits of detection of 8.20 × 105 and 1.67 × 106 particles mL−1, respectively, together with high linearity, specificity, stability, and recovery. Clinical validation using serum samples from 153 individuals yielded AUC values of 0.9806 for BC detection and 0.9642 for TNBC classification. This work establishes a generalizable signal-conversion strategy that extends conventional DLD technology from microscale particle manipulation to nanoscale biomarker analysis, providing an analytical platform for EV-based liquid biopsy.