A method for screening extracellular vesicle tumor markers

Abstract

This invention belongs to the field of biomedical detection and molecular diagnostics technology, and discloses a method for screening extracellular vesicle tumor markers. The method includes the following steps: sample preparation and vesicle identification, proteomics data acquisition, differentially expressed protein screening, biological function and pathway enrichment analysis, microfluidic chip multi-center clinical cohort validation, single-cell transcriptome biological mechanism validation, and machine learning algorithm optimization screening. This invention establishes a systematic and standardized screening process from proteomics screening, microfluidic chip validation to machine learning optimization. The selected combination of extracellular vesicle tumor markers achieves an accuracy of over 90% in early cancer diagnosis, significantly outperforming traditional single markers. This invention also provides a dual-layer microfluidic chip detection platform, enabling high-throughput sample processing and in-situ lysis detection. The screening method of this invention is applicable to various cancer types, has good universality and clinical translation prospects, and provides a new technical means for early cancer screening and precise diagnosis.

Description

Technical Field

[0001] This invention belongs to the field of biomedical detection and molecular diagnostic technology, and specifically relates to a method for screening extracellular vesicle tumor markers. Background Technology

[0002] Cancer is one of the leading causes of death worldwide, with its high mortality rate largely attributed to the tendency of tumors to metastasize. Currently, clinical management of cancer primarily relies on multidisciplinary comprehensive treatment, including surgical resection, chemotherapy, targeted therapy, and immunotherapy. However, significant challenges remain in clinical practice: more than half of early-stage cancer patients experience recurrence or metastasis after radical resection, and approximately 23% of patients already have distant metastases at initial diagnosis, severely limiting treatment options. Despite advancements in cancer detection and treatment technologies in recent years, the lack of efficient and reliable screening methods for early cancer diagnosis remains a key bottleneck restricting improvements in patient survival rates.

[0003] In recent years, extracellular vesicles (EVs) have attracted widespread attention as important mediators of intercellular communication. EVs are nanoscale membrane-bound vesicles released by cells, carrying various biological molecules such as proteins, nucleic acids, and lipids, and can reflect the physiological and pathological state of the source cells. In the dynamic changes of the tumor microenvironment (TME), EV-mediated intercellular communication influences key processes such as tumor cell proliferation, invasion, metastasis, and immune escape. Studies have shown that molecular markers carried by tumor-derived EVs can accurately reflect the real-time status of tumors, providing a new research direction for early cancer diagnosis.

[0004] Proteins, as direct executors of life activities, are abundant and functionally diverse in extracellular vesicles. Compared to free nucleic acids or proteins, proteins in extracellular vesicles are protected by the vesicle membrane, exhibiting greater stability and more accurately reflecting the molecular characteristics of tumor cells. However, current research on tumor markers largely focuses on single markers, whose diagnostic efficacy in liquid biopsies is often limited. Existing reports indicate that single markers cannot comprehensively reflect the heterogeneity and complexity of tumors, leading to insufficient diagnostic sensitivity and specificity.

[0005] Despite the enormous potential of extracellular vesicles in tumor diagnosis, a systematic and standardized screening process for extracellular vesicle tumor markers has yet to be established. How to screen specific combinations of tumor markers from the proteome of extracellular vesicles to achieve precise diagnosis, and how to establish a complete technical system from discovery to clinical validation, are pressing technical challenges that need to be addressed.

[0006] Therefore, developing a systematic method for screening extracellular vesicle tumor markers, and screening out a combination of markers with high sensitivity and specificity through multidimensional validation and clinical cohort evaluation, is of great clinical significance and application value for improving the accuracy of early cancer diagnosis and improving patient prognosis. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method for screening extracellular vesicle tumor markers. The aim is to establish a systematic screening process from proteomics screening and microfluidic chip validation to machine learning optimization, in order to obtain a combination of highly sensitive and specific markers that can be used for the early and accurate diagnosis of cancer.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows: A method for screening extracellular vesicle tumor markers includes the following steps: (1) Sample preparation and vesicle identification: Peripheral blood was collected from cancer patients at different stages and healthy controls. Serum was separated, extracellular vesicles were extracted, and their morphological and particle size characteristics were verified. (2) Proteomics data acquisition: Mass spectrometry was used to perform proteomics analysis on the extracted extracellular vesicle samples to obtain protein expression profile data; (3) Screening of differentially expressed proteins: Proteins that are significantly upregulated in cancer patients are screened with a fold change of expression greater than 1.5 and a corrected P value of < 0.05 as the threshold. (4) Biological function and pathway enrichment analysis: GO functional enrichment analysis was performed on the screened differentially expressed proteins, and a candidate biomarker library was constructed by selecting proteins from the top pathway. (5) Clinical validation of microfluidic chips: The candidate biomarkers were validated in a multi-center clinical cohort using the microfluidic chip platform to evaluate their diagnostic efficacy; (6) Validation of biological mechanisms: Integrate and analyze public single-cell transcriptome databases to validate the tumor origin of candidate biomarkers and their biological association with cancer; (7) Clinical translation screening: The importance of candidate biomarkers is ranked using machine learning algorithms, and biomarker combinations with high diagnostic efficacy and suitable for clinical translation are selected.

[0009] In the above scheme, the extraction of extracellular vesicles in step (1) is carried out by ultracentrifugation combined with size exclusion chromatography, or by a commercial extraction kit based on polymer precipitation.

[0010] In the above scheme, the morphology and particle size verification of the extracellular vesicles in step (1) are verified by transmission electron microscopy and nanoparticle tracking analyzer to verify that they have a double membrane structure or cup-shaped morphology and the particle size distribution is in the range of 30-150 nm.

[0011] In the above scheme, the mass spectrometry technology mentioned in step (2) is 4D-DIA mass spectrometry technology.

[0012] In the above scheme, the differentially expressed protein screening in step (3) is performed using the R language "limma" package for statistical analysis.

[0013] In the above scheme, the microfluidic chip platform mentioned in step (5) includes: The extracellular vesicle capture chip has a CD63 capture antibody deposited in the lower basal layer and multiple sample isolation wells in the upper layer. The detection chip has a lower layer coated with capture antibodies for various candidate proteins and an upper layer with corresponding sample isolation wells. The detection process includes: capturing vesicles using an extracellular vesicle capture chip, lysing and releasing proteins in situ, transferring them to a detection chip, adding detection antibodies and fluorescent markers, and reading the fluorescence signal.

[0014] In a further technical solution, the lower substrate material of the extracellular vesicle capture chip and the detection chip is glass, the upper layer is PDMS material, and antibody modification is performed using a large specific surface area material with π-π bond binding or electrostatic force binding. The material includes graphene quantum dots, graphene oxide quantum dots, or carbon quantum dots.

[0015] In the above scheme, the single-cell transcriptome data analysis in step (6) uses the Seurat R package for dimensionality reduction, clustering and cell type identification to verify the specific expression of candidate proteins in tumor cells.

[0016] In the above scheme, the machine learning algorithm mentioned in step (7) includes one or more of random forest, support vector machine, logistic regression, LightGBM, and CatBoost, which are used to build a diagnostic model and screen the top-ranked biomarkers in terms of contribution rate.

[0017] The above scheme also includes constructing a discovery cohort and an independent validation cohort, using receiver operating characteristic (ROC) curves to assess the sensitivity and specificity of the diagnostic model, and comparing it with traditional biomarkers.

[0018] The method for screening extracellular vesicle tumor markers provided by the above technical solution has the following beneficial effects: (1) Provide a systematic and standardized screening process This invention establishes for the first time a complete and systematic extracellular vesicle tumor marker screening system, from sample preparation, vesicle identification, proteomics analysis, differential protein screening, bioinformatics functional enrichment, multi-center clinical validation of microfluidic chips to single-cell transcriptome biological mechanism validation, filling the current technological gap of lacking standardized screening procedures.

[0019] (2) Significantly improves diagnostic accuracy Through multi-step screening and multiple verifications, the tumor marker combination screened by this invention has an accuracy rate of over 90% in the early diagnosis of cancer, which is significantly better than the diagnostic efficacy of traditional single markers, and effectively solves the technical problems of insufficient sensitivity and specificity of existing markers.

[0020] (3) Combine multiple verification methods to ensure reliability This invention uses 4D-DIA high-precision mass spectrometry to obtain protein expression profiles, utilizes a microfluidic chip platform for multi-center clinical cohort validation, and combines a public single-cell transcriptome database for biological mechanism validation. It ensures the reliability and specificity of the screened biomarkers from three levels: proteomics, clinical detection, and molecular mechanism.

[0021] (4) Innovative microfluidic chip design, resulting in high detection efficiency The bilayer microfluidic chip structure designed in this invention (both the extracellular vesicle capture chip and the detection chip are bilayer structures) enables high-throughput sample processing and in-situ lysis detection, with simple operation and fast detection speed. Antibody modification using large specific surface area materials (such as graphene quantum dots) with π-π bond binding and electrostatic force binding improves antibody capture efficiency and detection sensitivity.

[0022] (5) Machine learning optimizes biomarker combinations to improve the feasibility of clinical translation. By using various machine learning algorithms such as random forest and support vector machine to rank candidate biomarkers by contribution rate, a combination of biomarkers with high diagnostic efficacy and a small number of biomarkers is selected. This reduces detection costs while maintaining high accuracy, thereby improving the feasibility and promotion value of clinical translation.

[0023] (6) It has a wide range of applications and can be extended to various cancers. The screening method of this invention is not limited to specific cancer types, but can be applied to the screening of extracellular vesicle tumor markers for various solid tumors such as lung cancer, liver cancer, breast cancer, and colorectal cancer, and has good universality and prospects for promotion.

[0024] (7) Providing new strategies for liquid biopsy This invention uses serum extracellular vesicles for biomarker screening, which falls under the category of liquid biopsy. It has advantages such as being non-invasive or minimally invasive, convenient for sampling, and capable of real-time dynamic monitoring, providing a new technical means for early cancer screening, efficacy evaluation, and recurrence monitoring. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0026] Figure 1 Volcano plot for screening exosomal protein profiles in serum; Figure 2 This is a schematic diagram of a microfluidic chip platform; where (a) is an extracellular vesicle capture chip and (b) is a detection chip. Figure 3 Flowchart for the detection of proteins in extracellular vesicles; Figure 4 Flowchart for machine learning screening; Figure 5 This is the subject curve.

[0027] 1. Glass substrate layer one; 2. Extracellular vesicle capture layer; 3. Sample isolation well one; 4. Glass substrate layer two; 5. Detection layer; 6. Sample isolation well two; 7. Extracellular vesicle; 8. Capture antibody. Detailed Implementation

[0028] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0029] Example 1: Systematic screening and validation of biomarkers This embodiment aims to illustrate in detail the entire process of discovery, screening, verification, and diagnostic model construction of extracellular vesicle tumor markers described in this invention.

[0030] 1. Sample collection and processing 1.1 Collection of serum samples Peripheral blood was collected from cancer patients at different stages and age- and sex-matched healthy controls using anticoagulant-free vacuum blood collection tubes. After collection, the serum tubes were left to stand upright at room temperature for 30-60 minutes to allow the blood to coagulate and the serum to separate. Once the blood had completely coagulated and the clot had retracted, the supernatant was aspirated and centrifuged at 3000-4000 rpm for 5-10 minutes. The pale yellow, transparent liquid at the top after centrifugation is the serum, which was aliquoted and stored at -80°C for later use, avoiding repeated freeze-thaw cycles.

[0031] 2. Extraction and identification of extracellular vesicles 2.1 Extraction of extracellular vesicles 200 μL of serum was collected and extracted using a commercially available extracellular vesicle extraction kit based on polymer precipitation. The specific procedures were strictly followed according to the kit's instructions. The extracted extracellular vesicle precipitate was resuspended in an appropriate amount of PBS buffer for subsequent identification and analysis.

[0032] 2.2 Identification by transmission electron microscopy 10 μL of vesicle suspension was dropped onto a copper grid, allowed to stand for adsorption, and then negatively stained with phosphotungstic acid. After drying at room temperature, the vesicles were observed under a transmission electron microscope. The results showed that the extracted vesicles exhibited a typical bilayer membrane structure and cup-shaped morphology, with a diameter of approximately 30-150 nm, consistent with the morphological characteristics of extracellular vesicles.

[0033] 2.3 Nanoparticle Tracking Analysis The particle size distribution and concentration of vesicle samples were analyzed using a NanoSight NS300 nanoparticle tracking analyzer. Samples were appropriately diluted with PBS before analysis. Results showed that the main particle size peak of the vesicle particles was located at 142 nm, and the particle size distribution among samples exhibited good consistency, indicating successful extraction of extracellular vesicles and good sample homogeneity.

[0034] 3. Proteomics analysis and differential protein screening 3.1 Proteomics Data Acquisition The extracted extracellular vesicle samples were subjected to proteolytic digestion, and 4D-DIA (four-dimensional data-independent acquisition) mass spectrometry was used for in-depth proteomics analysis to obtain high-reliability protein expression profile data.

[0035] 3.2 Screening of differentially expressed proteins Using a healthy control group as a reference, statistical analysis of proteomics data was performed using the R language package "limma". The screening threshold was set as follows: fold change (FC) ≥ 1.5 and corrected p-value (False Discovery Rate, FDR) < 0.05. Proteins significantly upregulated in extracellular vesicles in the serum of cancer patients were screened, and the results are presented in volcano plot format (e.g., ...). Figure 1 (As shown).

[0036] 3.3 Biological function and pathway enrichment analysis Gene Ontology (GO) functional enrichment analysis was performed on the significantly upregulated differentially expressed proteins identified through screening, encompassing three dimensions: biological process, cellular component, and molecular function. Based on the enrichment analysis results, proteins involved in the top 10 pathways were selected to construct a preliminary candidate biomarker library.

[0037] 4. Quantitative detection and model building on microfluidic chip platforms 4.1 Microfluidic Chip Structure Design The microfluidic chip of this invention comprises two parts: an extracellular vesicle capture chip and a detection chip. (1) Extracellular vesicle capture chip (e.g.) Figure 2As shown in (a): It contains a two-layer structure. The lower layer is a glass substrate layer 1, which is coated with CD63 capture antibody to capture extracellular vesicles 7 in serum. The upper layer is an extracellular vesicle capture layer 2 made of polydimethylsiloxane (PDMS) material, which is provided with 24 (3×8) circular through-holes as sample isolation wells 3 to isolate different samples and avoid cross-contamination.

[0038] (2) Detection chip (e.g.) Figure 2 As shown in (b): It contains a two-layer structure. The lower layer is a glass substrate layer 4, which is coated with capture antibodies 8 containing various extracellular vesicle proteins. The upper layer is a detection layer 5 made of PDMS material, which also has 24 (3×8) circular through-holes as sample isolation wells 6 to isolate different samples.

[0039] To improve antibody capture efficiency, the chip surface is modified with a large specific surface area material that combines π-π bonding with electrostatic bonding. The material includes graphene quantum dots (GQD), graphene oxide quantum dots (GOQD), or carbon quantum dots (CQDs).

[0040] 4.2 Microfluidic chip testing process (e.g.) Figure 3 (As shown) (1) Extracellular vesicle capture: 0.1-1 mg / mL of CD63 capture antibody was spread evenly on a chip substrate (glass material) and incubated overnight (more than 10 hours) at 4°C. Then, an extracellular vesicle capture layer prepared with PDMS was bonded. 10 μL of serum from different clinical samples was added to the corresponding capture layer wells and incubated at room temperature for 1-2 hours. After incubation, excess liquid was removed, and the cells were washed three times with PBS buffer and dried.

[0041] (2) Lysis of extracellular vesicles: Add 5 μL of RIPA lysis buffer to each well and lyse at room temperature for 30 minutes. Gently pipette several times to mix the lysis buffer thoroughly.

[0042] (3) Protein transfer and detection: Pipette 4 μL of the lysed liquid from each well and transfer it to the corresponding well of the detection chip. Incubate at room temperature for 30-60 minutes to allow the protein to bind to the capture antibody. After removing the supernatant, add the corresponding detection antibody (0.1-1 μg / mL) and incubate at room temperature for 30-60 minutes. After removing the supernatant, add 635 nm APC fluorescently labeled molecules (0.01-1 mg / mL) and incubate at room temperature for 20-40 minutes. Remove the supernatant, wash the chip with PBS buffer, dry it, and then place it in a chip scanner for fluorescence signal reading.

[0043] 4.3 Diagnostic Model Construction The multicenter clinical cohort (Cohort 1) was randomly divided into a training set (70%) and a test set (30%). Using the fluorescence signal values ​​of various candidate proteins measured in the training set as features and pathological diagnosis as the gold standard, five machine learning algorithms—Random Forest, Support Vector Machine (SVM), Logistic Regression, LightGBM, and CatBoost—were used to construct diagnostic models. Based on the feature importance scores calculated by each algorithm, the candidate proteins were ranked by contribution rate, and the top 10 proteins were selected as the final tumor marker combination. The specific selection process is as follows: Figure 4 As shown.

[0044] 5. Independent verification queue verification To rigorously evaluate the generalization ability of the diagnostic model, validation was conducted in a completely independent phase 2 clinical cohort (Cohort 2). Using the same microfluidic chip detection process and data analysis methods as in Cohort 1, the diagnostic efficacy of the selected tumor marker combination in distinguishing cancer patients from healthy controls was validated. The diagnostic accuracy of the model was assessed using receiver operating characteristic (ROC) curves and compared with traditional tumor markers (e.g., ...). Figure 5 (As shown). The results show that the biomarker combination screened in this invention exhibits excellent diagnostic efficacy in an independent validation cohort, significantly outperforming traditional biomarkers.

[0045] 6. Validation of biological mechanisms To elucidate the tumor origin and biological basis of the high diagnostic performance of the selected biomarkers, corresponding tissue single-cell RNA sequencing datasets were downloaded from public databases (such as the GEO database). Data preprocessing, dimensionality reduction (UMAP), and cell clustering analysis were performed using the Seurat R package to identify major cell populations such as tumor cells, immune cells, and stromal cells, validating the specific expression of candidate biomarker proteins in tumor cells. The results showed that the selected biomarkers were highly expressed in tumor cells, supporting their rationale as highly specific diagnostic targets at the molecular level.

[0046] 7. Feasibility assessment for clinical translation Taking into account diagnostic efficacy, testing cost, and clinical applicability, the top 10 tumor marker combinations selected using machine learning algorithms effectively reduce the types of antibodies and reagent consumption required for testing while maintaining high accuracy (AUC > 0.9), thus improving the feasibility of clinical translation and its application value.

[0047] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for screening extracellular vesicle tumor markers, characterized in that, Includes the following steps: (1) Sample preparation and vesicle identification: Peripheral blood was collected from cancer patients at different stages and healthy controls. Serum was separated, extracellular vesicles were extracted, and their morphological and particle size characteristics were verified. (2) Proteomics data acquisition: Mass spectrometry was used to perform proteomics analysis on the extracted extracellular vesicle samples to obtain protein expression profile data; (3) Screening of differentially expressed proteins: Proteins that are significantly upregulated in cancer patients are screened with a fold change of expression greater than 1.5 and a corrected P value of < 0.05 as the threshold. (4) Biological function and pathway enrichment analysis: GO functional enrichment analysis was performed on the screened differentially expressed proteins, and a candidate biomarker library was constructed by selecting proteins from the top pathway. (5) Clinical validation of microfluidic chips: The candidate biomarkers were validated in a multi-center clinical cohort using the microfluidic chip platform to evaluate their diagnostic efficacy; (6) Validation of biological mechanisms: Integrate and analyze public single-cell transcriptome databases to validate the tumor origin of candidate biomarkers and their biological association with cancer; (7) Clinical translation screening: The importance of candidate biomarkers is ranked using machine learning algorithms, and biomarker combinations with high diagnostic efficacy and suitable for clinical translation are selected.

2. The method according to claim 1, characterized in that, The extracellular vesicles described in step (1) are extracted using ultracentrifugation combined with size exclusion chromatography, or using a commercially available extraction kit based on polymer precipitation.

3. The method according to claim 1, characterized in that, The morphology and particle size of the extracellular vesicles described in step (1) were verified using a transmission electron microscope and a nanoparticle tracking analyzer to verify that they have a double membrane structure or a cup-shaped morphology, with a particle size distribution in the range of 30–150 nm.

4. The method according to claim 1, characterized in that, The mass spectrometry technique mentioned in step (2) is 4D-DIA mass spectrometry.

5. The method according to claim 1, characterized in that, The differentially expressed protein screening described in step (3) was statistically analyzed using the R language "limma" package.

6. The method according to claim 1, characterized in that, The microfluidic chip platform mentioned in step (5) includes: The extracellular vesicle capture chip has a CD63 capture antibody deposited in the lower basal layer and multiple sample isolation wells in the upper layer. The detection chip has a lower layer coated with capture antibodies for various candidate proteins and an upper layer with corresponding sample isolation wells. The detection process includes: capturing vesicles using an extracellular vesicle capture chip, lysing and releasing proteins in situ, transferring them to a detection chip, adding detection antibodies and fluorescent markers, and reading the fluorescence signal.

7. The method according to claim 6, characterized in that, The lower substrate material of the extracellular vesicle capture chip and detection chip is glass, and the upper layer is PDMS material. Antibody modification is performed using a large specific surface area material with π-π bond binding or electrostatic force binding. The material includes graphene quantum dots, graphene oxide quantum dots or carbon quantum dots.

8. The method according to claim 1, characterized in that, The single-cell transcriptome data analysis described in step (6) uses the Seurat R package for dimensionality reduction, clustering, and cell type identification to verify the specific expression of candidate proteins in tumor cells.

9. The method according to claim 1, characterized in that, The machine learning algorithms mentioned in step (7) include one or more of random forest, support vector machine, logistic regression, LightGBM, and CatBoost, which are used to build diagnostic models and screen biomarkers with the highest contribution rates.

10. The method according to claim 1, characterized in that, It also includes constructing discovery and independent validation cohorts, using receiver operating characteristic (ROC) curves to assess the sensitivity and specificity of the diagnostic model, and comparing it with traditional biomarkers.

Images