Extracellular vesicles (EVs) and EV proteins are promising biomarkers for cancer liquid biopsy. Herein, we designed a case–control study involving 100 controls and 100 patients with esophageal, stomach, colorectal, liver, or lung cancer to identify common and type-specific biomarkers of plasma-derived EV surface proteins for the five cancers. EV surface proteins were profiled using a sequencing-based proximity barcoding assay. In this study, five differentially expressed proteins (DEPs) and eight differentially expressed protein combinations (DEPCs) showed promising performance (area under curve, AUC > 0.900) in pan-cancer identification [e.g., TENM2 (AUC = 0.982), CD36 (AUC = 0.974), and CD36-ITGA1 (AUC = 0.971)]. Our classification model could properly discriminate between cancer patients and controls using DEPs (AUC = 0.981) or DEPCs (AUC = 0.965). When distinguishing one cancer from the other four, the accuracy of the classification model using DEPCs (85–92%) was higher than that using DEPs (78–84%). We validated the performance in an additional 14 cancer patients and 14 controls, and achieved an AUC value of 0.786 for DEPs and 0.622 for DEPCs, highlighting the necessity to recruit a larger cohort for further validation. When clustering EVs into subpopulations, we detected cluster-specific proteins highly expressed in immune-related tissues. In the context of colorectal cancer, we identified heterogeneous EV clusters enriched in cancer patients, correlating with tumor initiation and progression. These findings provide epidemiological and molecular evidence for the clinical application of EV proteins in cancer prediction, while also illuminating their functional roles in cancer physiopathology.

中文翻译：

---

基于单一细胞外囊泡表面蛋白的血液检测确定了用于检测和筛查五种癌症的潜在生物标志物

细胞外囊泡 (EV) 和 EV 蛋白是癌症液体活检有前途的生物标志物。在此，我们设计了一项病例对照研究，涉及 100 名对照者和 100 名食管癌、胃癌、结直肠癌、肝癌或肺癌患者，以确定这五种癌症的血浆来源 EV 表面蛋白的常见和类型特异性生物标志物。使用基于测序的邻近条形码分析对 EV 表面蛋白进行分析。在本研究中，五种差异表达蛋白（DEP）和八种差异表达蛋白组合（DEPC）在泛癌鉴定中表现出良好的性能（曲线下面积，AUC > 0.900）[例如，TENM2（AUC = 0.982）、CD36（AUC） = 0.974) 和 CD36-ITGA1 (AUC = 0.971)]。我们的分类模型可以使用 DEP (AUC = 0.981) 或 DEPC (AUC = 0.965) 正确区分癌症患者和对照。当区分一种癌症与其他四种癌症时，使用 DEPC 的分类模型的准确性 (85-92%) 高于使用 DEP 的分类模型 (78-​​84%)。我们在另外 14 名癌症患者和 14 名对照者中验证了其性能，DEP 的 AUC 值为 0.786，DEPC 的 AUC 值为 0.622，这凸显了招募更大队列进行进一步验证的必要性。当将 EV 聚类为亚群时，我们检测到在免疫相关组织中高表达的簇特异性蛋白。在结直肠癌的背景下，我们发现了癌症患者中富含的异质 EV 簇，与肿瘤的发生和进展相关。这些发现为 EV 蛋白在癌症预测中的临床应用提供了流行病学和分子证据，同时也阐明了它们在癌症病理生理学中的功能作用。