The potential of serum extracellular vesicles (EVs) as non-invasive biomarkers for diagnosing colorectal cancer (CRC) remains elusive. We employed an in-depth 4D-DIA proteomics and machine learning (ML) pipeline to identify key proteins, PF4 and AACT, for CRC diagnosis in serum EV samples from a discovery cohort of 37 cases. PF4 and AACT outperform traditional biomarkers, CEA and CA19-9, detected by ELISA in 912 individuals. Furthermore, we developed an EV-related random forest (RF) model with the highest diagnostic efficiency, achieving AUC values of 0.960 and 0.963 in the train and test sets, respectively. Notably, this model demonstrated reliable diagnostic performance for early-stage CRC and distinguishing CRC from benign colorectal diseases. Additionally, multi-omics approaches were employed to predict the functions and potential sources of serum EV-derived proteins. Collectively, our study identified the crucial proteomic signatures in serum EVs and established a promising EV-related RF model for CRC diagnosis in the clinic.

Objectives: The absence of non-invasive biomarkers for the early diagnosis of colorectal cancer (CRC) has contributed to poor prognosis. Extracellular vesicles (EVs) have emerged as promising candidates for cancer monitoring using liquid biopsy. However, the complexity of EVs isolation procedures and the absence of clear targets for detecting serum-derived EVs have hindered the clinical application of EVs in early CRC diagnosis. Methods: In the discovery phase, we conducted a comprehensive 4D-DIA proteomic analysis of serum-derived EVs samples from 37 individuals, performing an initial screening of EVs surface proteins. In the technical validation phase, we developed an extraction-free CRC-EVArray microarray to assess the expression of these potential EVs surface proteins in a multi-centre study comprising 404 individuals. In the application phase, the authors evaluated the diagnostic efficacy of the CRC-EVArray model based on machine-learning algorithms. Results: Through 4D-DIA proteomic analysis, the authors identified seven potential EVs surface proteins showing significantly differential expression in CRC patients compared to healthy controls. Utilizing our developed high-throughput CRC-EVArray microarray, we further confirmed the differential expression of three EVs surface proteins, FIBG, PDGF-β and TGF-β, in a large sample population. Moreover, we established an optimal CRC-EVArray model using the NNET algorithm, demonstrating superior diagnostic efficacy with an area under the curve (AUC) of 0.882 in the train set and 0.937 in the test set. Additionally, we predicted the functions and potential origins of these EVs-derived proteins through a series of multi-omics approaches. Conclusions: Our systematic exploration of surface protein expression profiles on serum-derived EVs has identified FIBG, PDGF-β, and TGF-β as novel diagnostic biomarkers for CRC. The development of CRC-EVArray diagnostic model based on these findings provided an effective tool for the large-scale CRC screening, thus facilitating its translation into clinical practice.

Extracellular vesicles (EVs) were promising circulating biomarkers for multiple diseases, but whether serum EVs-derived proteins could be used as a reliable tumor biomarker for colorectal cancer (CRC) remained inconclusive. In this study, we identified CXCL4 by a 4D data-independent acquisition-based quantitative proteomics assay of serum EVs-derived proteins in 40 individuals and subsequently analyzed serum EVs-derived CXCL4 levels by ELISA in 2 cohorts of 749 individuals. The results revealed that EVs-derived CXCL4 levels were dramatically elevated in CRC patients than in benign colorectal polyp patients or healthy controls (HC). Furthermore, receiver operating characteristic curves revealed that EVs-derived CXCL4 exhibited superior diagnostic performance with area under the curve of 0.948 in the training cohort. Additionally, CXCL4 could effectively distinguish CRC in stage I/II from HC. Notably, CRC patients with high levels of EVs-derived CXCL4 have shorter 2-year progression-free survival than those with low levels. Overall, our findings demonstrated that serum EVs-derived CXCL4 was a candidate diagnostic and prognostic biomarker for CRC.