Salivary extracellular vesicles (EVs) have been recognized as one of the most promising noninvasive sources of biomarkers for early cancer detection. However, the lack of efficient isolation and accurate identification methods for salivary EVs limits their clinical utility in early cancer diagnosis. Here, we report the rapid and precise identification of esophageal cancer using the Saliva Extracellular vesicle-based Early Diagnostic system (SEEDx). In this system, salivary EVs were isolated, purified, amplified, and analyzed from human saliva by integrating the ultrafast-isolation platform EXODUS with tandem mass tag (TMT)-based proteomics analysis (EXODUS-TMT proteomics). Using TCGA esophageal cancer (EC) and GTEx healthy population cohorts, we developed a diagnostic model consisting of 9 markers (HIST2H2BE, TP53BP2, TPM4, CALR, HDGF, LMNA, GUSB, NADSYN1, and AGRN), which demonstrated high diagnostic accuracy with AUC values of 0.996 for EC and 0.991 for early-stage EC. We further established a cost-effective model using only two markers (HDGF and CALR) to diagnose both EC and early-stage EC, achieving AUC values of 0.950 and 0.935, respectively. This model was validated in an independent, prospectively collected cohort of EC patients, yielding AUC values of 0.873 and 0.854, respectively. Our streamlined EV isolation and identification workflow facilitates noninvasive biomarker discovery for early-stage esophageal cancer detection. This study extends the EXODUS platform to salivary EVs for EC detection, documents matrix-specific workflow optimizations, and identifies saliva-derived biomarker signatures integrated into a multimarker diagnostic model.
Huang et al. (Mon,) studied this question.