免疫亲和富集的细胞外囊泡揭示肿瘤来源的DNA突变以用于乳腺癌检测

Background: Extracellular vesicle-derived tumor DNA (etDNA) has recently gained attention as a potential analyte for liquid biopsy; however, the biological origin and stability of EV-associated DNA remain controversial. These uncertainties highlight the need for size-independent, molecularly specific EV isolation approaches to rigorously evaluate tumor-derived DNA signals. To address this challenge, we isolated breast cancer-derived EVs (BEVs) using EPCAM-targeted immunoaffinity capture and assessed whether etDNA mutations accurately reflect tumor genotypes in both experimental models and clinical samples. Methods: Mutational targets frequently observed in breast cancer were identified using the TCGA dataset, and four high-prevalence substitutions (PIK3CA p.H1047R, PIK3CA p.E545K, TP53 p.R175H, TP53 p.R273H) were selected. BEVs were isolated from plasma using magnetic beads conjugated with anti-EPCAM antibodies. EV identity was confirmed via nanoparticle tracking analysis and immunoblotting. etDNA was extracted from the isolated BEVs and analyzed by digital droplet PCR (ddPCR). Analytical validation included mutation profiling of breast cancer cell lines, 1D/2D droplet cluster analysis, and limit of detection (LOD) assays. Clinical concordance was assessed using matched tumor tissue sequencing results from breast cancer patients. Results: In vitro studies demonstrated that etDNA mutations corresponded precisely to the genomic alterations of parental breast cancer cell lines. ddPCR yielded clear separation of mutant and wild-type droplets, and the assay showed high analytical sensitivity with an LOD of 0.08 copies/µL. Among 98 breast cancer patients, tumor tissue sequencing identified target mutations in 27 cases (28%). Of these, 19 had available plasma samples, and 11 patients (58%) harbored a concordant etDNA mutation in BEVs. PIK3CA p.H1047R was the most prevalent mutation in both tissue and etDNA. These findings indicate that EPCAM-enriched BEVs retain tumor-specific genetic alterations and that etDNA detection is feasible even at low DNA input, supporting potential applicability to early-stage or low-shedding disease. Conclusion: This study demonstrates that etDNA extracted from immunoaffinity-isolated BEVs reliably reflects the mutational landscape of breast cancer. The combination of EPCAM-based BEV capture and ddPCR provides a sensitive and specific workflow for mutation detection, independent of size-based EV purification. These results support the utility of etDNA as a clinically informative biomarker and highlight its potential to enhance the diagnostic performance of liquid biopsy, particularly in early-stage breast cancer or settings with low variant allele frequency.