Plasma exosome metabolite screened by combining SEC-LC method and application

Abstract

The invention belongs to the technical field of biomedical detection and tumor molecular diagnosis, particularly relates to a plasma exosome metabolite, and further discloses a method for separating and purifying a plasma exosome based on exclusion chromatography and screening a prostatic malignant tumor biomarker in combination with liquid mass spectrometry metabonomics. The invention also discloses application of the compound in preparation of a prostatic cancer biomarker. According to the method, the SEC method for extracting the exosome and the LC-MS metabonomics analysis technology are organically combined, so that the high-efficiency advantage of the SEC in exosome purification can be fully played, accurate detection and quantitative analysis of trace metabolites in the exosome can be realized by virtue of an LC-MS platform, and then a characteristic metabonomics spectrogram is constructed.

Description

Technical Field

[0001] The present invention belongs to the technical field of biomedical detection and tumor molecular diagnosis, and specifically relates to a plasma exosome metabolite. It further discloses a method for separating and purifying plasma exosomes based on size exclusion chromatography and combining it with liquid chromatography-mass spectrometry metabolomics to screen biomarkers of prostate malignant tumors, as well as the use of the method for preparing prostate cancer biomarkers. Background Art

[0002] In recent years, with the increasing incidence of prostate cancer and the urgent need for early treatment among patients, traditional prostate-specific antigen (PSA) testing and imaging examinations have gradually exposed problems such as insufficient sensitivity, limited specificity, and high invasiveness. These limitations have prompted researchers to continuously explore more accurate, non-invasive detection methods suitable for large-scale clinical application. In recent years, extracellular vesicles (EVs), as important carriers of cellular metabolic information, have gradually become an important direction for tumor biomarker development due to their high stability in biological fluids and their ability to reflect the intrinsic state of tumor cells.

[0003] Currently, the main methods for isolating exosomes include size exclusion chromatography and pre-precipitation, but these methods have limitations in terms of separation efficiency, purity, and cost. In contrast, ultracentrifugation (UC) has become the gold standard for exosome separation technology due to its ease of operation and high separation efficiency. The UC method utilizes centrifugal force to separate particles based on differences in density and mass. This method not only efficiently separates macromolecules (such as proteins and nucleic acids) but also preserves the original biological activity of exosomes, providing a stable and reliable sample foundation for subsequent metabolomics analysis.

[0004] Liquid chromatography-mass spectrometry (LC-MS), the most mature and widely used technology platform in metabolomics analysis, offers significant advantages in biomarker screening and quantitative analysis due to its high sensitivity, high resolution, and broad metabolite coverage. LC-MS enables multidimensional, qualitative, and quantitative metabolite detection in complex biological samples, significantly enhancing the ability to capture subtle metabolic changes under pathological conditions and providing strong technical support for the early diagnosis of diseases such as prostate cancer.

[0005] Although exosome isolation and metabolomics detection have each made progress in the field of tumor diagnosis, their combined application is still in its early stages of exploration, and a systematic and standardized detection process has yet to be established. This field awaits further exploration and research into this approach. Summary of the Invention

[0006] In order to overcome the shortcomings and deficiencies of the prior art, the purpose of the present invention is to provide a plasma exosome metabolite related to the diagnosis of prostate cancer and its use as a biomarker for the diagnosis of prostate cancer. The plasma exosome metabolite related to the diagnosis of prostate cancer has good sensitivity and specificity when used for auxiliary diagnosis of prostate cancer, can be used to prepare a reagent for diagnosing prostate cancer, and has the advantages of easy sample acquisition, non-invasiveness, simplicity and speed.

[0007] In order to solve the above problems, the present invention provides a plasma exosome metabolite, including one or a mixture of the following metabolites: acylcarnitine [C18:0, COOH], lysophosphatidylethanolamine [18:2], lysophosphatidylcholine [22:6], F518.3169901 / ExMrn00096780, F518.3169901 / ExMrn00096781, 2-hydroxystearic acid, 9-hydroxyoctadecanoic acid, [8]-gingerol, adenosine, hexaethylene glycol, acylcarnitine [C18:2], linoleoylcarnitine [C18:2], and lysophosphatidylcholine [18:2 / 0:0].

[0008] It should be noted that the F518.3169901 / ExMrn00096780 refers to a compound with a mass-to-charge ratio of 518.3169901, and the compound number is ExMrn00096780; the F518.3169901 / ExMrn00096781 refers to a compound with a mass-to-charge ratio of 518.3169901, and the compound number is ExMrn00096781.

[0009] The present invention also provides a use of the plasma exosome metabolite for preparing a prostate cancer biomarker.

[0010] The present invention also provides an in vitro diagnostic marker model for prostate cancer, wherein the biological biomarkers of the diagnostic model include the plasma exosome metabolites.

[0011] The present invention also provides a method for screening the plasma exosome metabolites by combining SEC-LC method, comprising the following steps:

[0012] (1) Separation of plasma samples

[0013] Collect blood samples and separate plasma samples and blood cells by centrifugation;

[0014] (2) SEC method for extracting exosomes

[0015] The plasma was centrifuged at 2500-3500×g, the supernatant was collected, and the supernatant was further centrifuged at 8000-12000×g to collect the supernatant as a pretreatment sample; the pretreatment sample was added to the column filter plate of the equilibrium exclusion chromatography column, and when the sample just entered the column and the liquid flow stopped, the elution buffer was added, and the eluted sample was collected to obtain a sample to be tested for later use;

[0016] (3) Metabolic fingerprint detection

[0017] MeOH:ACN was added to the sample to be tested and the mixture was incubated and precipitated; the incubated reaction product was centrifuged, the supernatant extract was collected, evaporated to dryness, and then resuspended in ACN:H2O, centrifuged again, and the supernatant was collected for LC-MS detection;

[0018] (4) Establishment of key metabolic fingerprints

[0019] Combining univariate correlation analysis, t-test (P<0.05) and VIP score, we preliminarily screened the candidate metabolic fingerprints between prostate cancer and benign hyperplasia groups; and used the Lasso regression method to further extract the most predictive variables; and, by intersecting the above key metabolic fingerprints, we screened out key metabolic fingerprints with significant differences and clinical application value.

[0020] Specifically, the method for screening the plasma exosome metabolites by the combined SEC-LC method further includes: (5) constructing a prostate cancer screening model based on machine learning, specifically dividing the plasma sample data collected from patients with prostate cancer and benign prostatic hyperplasia into a training set and a test set, and performing machine learning according to the following process:

[0021] Data collection and processing: plasma mass spectrometry data are cleaned, feature extracted, normalized, and missing value filled;

[0022] Data segmentation: Divide the dataset into training set and validation set according to 8:2;

[0023] Model selection: Choose from ten machine learning algorithms including Adaboost, GBM, Glmnet, kknn, LogitBoost, MLP, NB, RegLogistic, RF, and svmRadiaWeights;

[0024] Modeling: Based on the selected model, use the training set to train the model;

[0025] Hyperparameter optimization: Automatically optimize the model's hyperparameters through algorithms to improve model performance;

[0026] Model Validation: Use the validation set to evaluate the performance of the model.

[0027] Specifically, in the method for screening the plasma exosome metabolites by the combined SEC-LC method, in step (3), the LC-MS detection step includes:

[0028] Octadecylsilane bonded silica gel is used as filler;

[0029] Mobile phase A: water-0.1 v / v% formic acid solution; i.e., 0.1% (v / v) formic acid added to water;

[0030] Mobile phase B: acetonitrile-0.1 v / v% formic acid solution; i.e., 0.1% (v / v) formic acid added to acetonitrile;

[0031] Perform gradient elution according to the following program:

[0032] 0-1min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0033] 1-8min, A% (v / v): B% (v / v) = 99.0%: 1.0% → 1.0%: 99.0%;

[0034] 8-10min, A% (v / v): B% (v / v) = 1.0%: 99.0%;

[0035] 10min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0036] 10-12min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0037] Gradient separation: 12 minutes;

[0038] Injection volume: 2 μL;

[0039] Column flow rate: 500 μL / min;

[0040] Column temperature: 40℃.

[0041] Specifically, in the method for screening the plasma exosome metabolites by the combined SEC-LC method, in step (3), the LC-MS detection step includes:

[0042] The ion spray voltage was set to 3500 V for positive mode and 2800 V for negative mode;

[0043] Full-scan spectra ( m / z 70–1200) were obtained on an Orbitrap at 60,000 resolution;

[0044] The normalized automatic gain control (AGC) target was 100%, and the maximum injection time was 100 ms;

[0045] The precursor ions were selected to enter the collision cell for high-energy collision dissociation (HCD) fragmentation, with collection energies of 20, 30, and 40, respectively;

[0046] The MS / MS resolution was set to 30,000, the normalized AGC target was 100%, the maximum injection time was 60 ms, the isolation window was 1 m / z, and the dynamic exclusion time was 4 s.

[0047] The present invention also provides use of the plasma exosome metabolites in preparing a prostate cancer diagnostic product.

[0048] Specifically, the product includes a reagent for detecting the content of the plasma exosome metabolite.

[0049] The present invention also provides a kit for diagnosing prostate cancer, which comprises a reagent for detecting the content of the plasma exosome metabolite in a sample.

[0050] This invention provides a noninvasive prostate cancer detection method based on extracellular vesicle metabolomics. This method addresses the technical limitations of existing prostate-specific antigen (PSA) tests, which lack specificity, and the invasive nature of imaging studies. Plasma exosomes are isolated and purified using size-exclusion chromatography, and then combined with liquid chromatography-mass spectrometry to construct an exosome metabolomics profile. This method screened 13 characteristic metabolites from 36 clinical samples, and a machine learning diagnostic model was established, achieving an area under the curve of 0.875 for the validation set. By combining a standardized exosome isolation process with metabolic markers, this method achieves high-precision differentiation between prostate cancer and benign hyperplasia.

[0051] The plasma exosome metabolites described in this invention specifically targets the intersection of plasma exosome isolation and purification technologies, small molecule metabolite detection and analysis, and artificial intelligence-assisted diagnostic models. Through standardized exosome isolation procedures and combined identification of metabolite markers, this approach enables molecular differential diagnosis of prostate cancer versus benign prostate diseases. This technical solution encompasses collaborative innovation across multiple technological branches, including clinical laboratory medicine, nanovesicle separation technology, mass spectrometry analysis, and medical big data processing.

[0052] This invention combines exosome isolation with metabolomics detection, organically integrating SEC extraction of exosomes with LC-MS metabolomics analysis. This not only leverages the high efficiency of SEC in exosome purification, but also enables the precise detection and quantitative analysis of trace metabolites within exosomes using the LC-MS platform, thereby constructing a characteristic metabolomics profile. This powerful combination of technologies not only meets the needs of noninvasive, rapid, and highly accurate prostate cancer detection, but also has the potential to effectively distinguish prostate cancer from benign hyperplasia in actual clinical applications.

[0053] The plasma exosome metabolites screened in the present invention were based on plasma samples from 36 prostate tumor groups and benign hyperplasia controls. Exosomes were standardizedly extracted using the SEC method (fast, low sample volume, and high extract purity), and an AI-assisted diagnosis model was constructed in combination with LC-MS metabolomics. The data analysis results showed that the plasma + exosome + metabolic process can provide a more comprehensive analytical perspective (AUC>0.875), effectively solving the problems of traditional prostate cancer detection methods (such as PSA testing and imaging examinations) with insufficient sensitivity, low specificity, and highly invasive procedures, resulting in a high misdiagnosis rate and poor patient compliance.

[0054] The plasma exosome metabolites screened by the present invention were detected based on the SEC+LC-MS platform, and metabolic markers with stable diagnostic performance were screened. 18 plasma exosome metabolic markers for prostate cancer were screened, and the differences between the healthy group and the experimental group were significant (p less than 0.05). The discovery of the prostate cancer markers is crucial for disease diagnosis and disease progression monitoring, but existing metabolic markers cannot achieve high-accuracy diagnostic performance.

[0055] In the screening method for plasma exosome metabolites screened by the present invention, we effectively improve the efficiency of model construction by rationally introducing multi-type machine learning. This can quickly find the optimal model, greatly shorten the model construction time, reduce the probability of misjudgment, and quickly improve the model training speed. It does not require manual parameter adjustment, effectively solving the problem of manual screening and parameter adjustment when selecting the optimal machine learning model, which requires a certain amount of time and difficulty in debugging.

[0056] The present invention uses exosome metabolomics detection and exosome metabolomics to construct a metabolic signature spectrum of prostate cancer, which can accurately distinguish between prostate cancer and benign hyperplasia, thereby greatly improving diagnostic accuracy and specificity, and has higher clinical value than traditional PSA testing and imaging examinations.

[0057] The method of the present invention uses plasma samples for detection, avoiding invasive operations such as tissue biopsy. At the same time, SEC is used to extract exosomes and LC-MS detection technology is used to achieve a rapid and standardized detection process. It is significantly superior to traditional methods in terms of convenience and patient compliance in clinical applications, and has the advantages of being non-invasive, efficient and easy to operate.

[0058] The present invention uses the SEC method to extract exosomes. Compared with traditional methods such as precipitation, the purity and stability of exosome extraction are significantly improved, and it has higher extraction efficiency and sample purity. It provides a more stable and reliable basis for subsequent metabolomics analysis, reduces detection errors, and effectively solves the defects of other exosome extraction methods such as large sample requirements and low separation purity.

[0059] The present invention uses LC-MS technology to achieve precise detection and quantitative analysis of trace metabolites in exosomes, thereby constructing a metabolic spectrum with clinical diagnostic significance. The established standardized detection process not only ensures the repeatability and stability of the detection results, but also has the advantages of standardized detection platform and high throughput. It is also convenient for large-scale clinical screening and promotion and application, can effectively reduce detection costs and operational difficulty, and effectively overcome the limitations of traditional metabolite detection methods in sensitivity and resolution.

[0060] Based on preliminary experimental data and theoretical analysis, the present invention demonstrates higher sensitivity and specificity (AUC: 0.875 and above) in distinguishing prostate cancer from benign diseases, indicating that the detection method has a significant improvement in practical application and provides reliable data support for accurate diagnosis.

[0061] This method combines SEC extraction of exosomes with LC-MS metabolomics analysis. This not only leverages the high efficiency of SEC in exosome purification, but also enables the precise detection and quantitative analysis of trace metabolites within exosomes using the LC-MS platform, thereby constructing a characteristic metabolomics profile. This powerful combination of technologies not only meets the requirements for noninvasive, rapid, and highly accurate prostate cancer detection, but also holds promise for effectively differentiating prostate cancer from benign hyperplasia in actual clinical applications.

[0062] Based on current technological developments and clinical needs, this study explores a method for detecting exosome metabolomics using plasma ultracentrifugation coupled with liquid chromatography-mass spectrometry, aiming to develop a standardized, rapid, and high-throughput detection platform. This platform not only improves the accuracy of early prostate cancer diagnosis and reduces misdiagnosis rates, but also provides patients with a more convenient, non-invasive testing method, effectively promoting the development of precision medicine for prostate cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0064] Figure 1 Schematic diagram of the comparison of exosome characteristic marker detection using different exosome extraction methods in the present invention; PCa: prostate cancer; BPH: prostate hyperplasia; U: ultracentrifugation; P: precipitation; K: kit method; PS: precipitation plus size exclusion chromatography; S: size exclusion chromatography;

[0065] Figure 2 The key metabolic fingerprints identified in the present invention show differential patterns in prostate cancer (PCa) patients and benign prostatic hyperplasia (BPH) patients. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention.

[0067] To fully disclose the entire technical solution of the present invention, this embodiment describes in detail the various operational steps and corresponding effects of the present invention from sample processing to data analysis and model construction, and illustrates the significant advantages brought about by the technical improvements through comparative experiments.

[0068] Example 1

[0069] This embodiment performs test sample processing and exosome separation.

[0070] In this embodiment, a clinically collected blood sample is first pretreated. Immediately after sample collection, the operator centrifuges the sample to obtain plasma, which is then stored at low temperatures to ensure sample stability. Subsequently, exosomes are extracted from the plasma using SEC.

[0071] After collecting blood samples in purple-top anticoagulant tubes, they were immediately centrifuged at 1600 × g for 15 minutes to separate the plasma from the blood cells. The separated plasma was quickly transferred to a -80°C freezer for storage, providing a stable sample for subsequent testing.

[0072] This example uses the SEC method for exosome extraction: First, centrifuge at 3000×g for 10 minutes at 4°C and transfer the supernatant; then, centrifuge at 10,000×g for 30 minutes at 4°C and collect the supernatant as a pre-treated sample. After cleaning and equilibrating the exclusion chromatography column (derived from the sEV 500 Ultrapure Columns kit), 1-2 mL of the supernatant is added to the column filter plate; as soon as the sample enters the column and the liquid flow stops, a predetermined volume of elution buffer is immediately added to exclude components that do not contain exosomes; a new EP tube is placed under the column, and 0.9 mL of elution buffer (PBS) is added to collect the sample, which is finally stored at -80°C for later use.

[0073] In this embodiment, Figure 1 The results shown in Figure 2 show that exosomes extracted using SEC exhibit lower background interference and higher signal consistency in subsequent detection of exosome signature markers, thus ensuring the acquisition of high-precision metabolic fingerprints. Compared with traditional ultracentrifugation methods, SEC is simpler to operate and significantly shortens processing time, while significantly improving the recovery rate and purity of exosomes.

[0074] Example 2

[0075] This example is based on the sample obtained in the above-mentioned Example 1, and performs metabolic fingerprint acquisition and data processing.

[0076] This example uses a mature LC-MS system to perform metabolite analysis on the processed exosome samples.

[0077] Sample pretreatment

[0078] In this embodiment, a suitable organic solvent is used to perform protein precipitation, ultrasonic treatment, and concentration on the sample, so that the metabolite components can be fully released and remain stable.

[0079] Add 240 μL of MeOH:ACN (v:v, 1:1) to the sample, vortex for 30 seconds, and sonicate for 10 minutes. Incubate at 20°C for 1 hour to facilitate protein precipitation. Then, centrifuge the sample at 13,000 rpm for 15 minutes at 4°C, collect the supernatant, and evaporate to dryness in a vacuum concentrator. Resuspend the dried extract in 50 μL of 1:1 ACN:H2O, sonicate for 10 minutes, and centrifuge at 13,000 rpm for 15 minutes at 4°C. Collect the supernatant and store at -80°C.

[0080] LC-MS detection

[0081] In this example, full scan and MS / MS data were collected in positive and negative ion modes under established chromatographic conditions and mass spectrometry parameters to ensure comprehensive capture of metabolite information.

[0082] The supernatant was analyzed by LC-MS using an Orbitrap Exploris™ 480 mass spectrometer coupled to a Vanquish liquid chromatography system (Thermo Fisher Scientific, USA). For Hilic separation, an ACQUITY UPLC BEH C18 column was used. The specific procedure included:

[0083] Mobile phase A: water-0.1 v / v% formic acid solution; i.e., 0.1% (v / v) formic acid added to water;

[0084] Mobile phase B: acetonitrile-0.1 v / v% formic acid solution; i.e., 0.1% (v / v) formic acid added to acetonitrile;

[0085] Perform gradient elution according to the following program:

[0086] 0-1min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0087] 1-8min, A% (v / v): B% (v / v) = 99.0%: 1.0% → 1.0%: 99.0%;

[0088] 8-10min, A% (v / v): B% (v / v) = 1.0%: 99.0%;

[0089] 10min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0090] 10-12min, A% (v / v): B% (v / v) = 99.0%: 1.0%;

[0091] Gradient separation 12 min;

[0092] Inject 2 μL;

[0093] The column flow rate was 500 μL / min;

[0094] The column temperature was 40°C.

[0095] Electrospray ionization mass spectra were acquired in both positive and negative ion modes. Data-dependent acquisition was used to simultaneously acquire full-scan MS and MSMS information. The ion spray voltage was set to 3500 V for positive mode and 2800 V for negative mode. Full-scan spectra (m / z 70-1200) were acquired on an Orbitrap at 60,000 resolution. The normalized automatic gain control (AGC) target was 100%, and the maximum injection time was 100 ms. Precursor ions were selected and entered into the collision cell for high-energy collisional dissociation (HCD) fragmentation, with collection energies of 20, 30, and 40, respectively. The MS / MS resolution was set to 30,000, the normalized AGC target was 100%, the maximum injection time was 60 ms, the isolation window was 1 m / z, and the dynamic exclusion time was 4 s.

[0096] Data processing

[0097] This embodiment performs format conversion, peak detection, alignment, and standardization on the raw data files to ensure the repeatability and accuracy of the data.

[0098] ProteoWizard (version 3.0.6150) was used to convert raw MS data files into mzXML format and MS2 data files into mgf format. Peak detection and alignment of all MS files (mzXML format) were performed using the R package "XCMS" (version 1.46.0). Missing values ​​were imputed using the KNN method, and normalization methods such as MedianNorm, LogNorm, and MeanCenter were performed using the "MetaboAnalystR" package to ensure data quality.

[0099] In this embodiment, under the aforementioned pre-treatment and detection parameters, after a rigorous data processing process, the metabolic fingerprint spectrum obtained has high resolution and high stability, providing a solid data foundation for the subsequent screening of key metabolic markers.

[0100] Example 3

[0101] This embodiment is based on the method of the aforementioned Example 2 to establish key metabolic fingerprints.

[0102] In this example, a combination of univariate correlation analysis, t-tests (P < 0.05), and VIP scores was used to initially screen candidate metabolic fingerprints between prostate cancer and benign hyperplasia groups. Lasso regression was then used to further identify the most predictive variables. Finally, by intersecting these key metabolic fingerprints, key metabolic fingerprints with significant differences and clinical application value were identified.

[0103] Based on the metabolic fingerprint data initially obtained in Example 2, this embodiment uses multiple statistical methods (such as univariate correlation analysis, t-test, VIP score, and Lasso regression, each using R language) to screen for metabolite differences between prostate cancer and benign prostatic hyperplasia samples. The specific steps are as follows:

[0104] (1) Using statistical analysis software, preliminary difference analysis was performed on the sample data of each group, i.e., t-test was performed, and candidate metabolites were initially selected if P < 0.05, VIP > 1, |Log2FC| > 0.5;

[0105] (2) Correlation analysis was performed in combination with clinical indicators to ensure that the screened metabolites had practical diagnostic significance;

[0106] (3) The Lasso regression method was used to further narrow the candidate range and finally determine the key metabolic fingerprints with significant differences and predictive capabilities.

[0107] In this example, the differences in the metabolic fingerprints screened out between the groups are shown in the attached figure. Figure 2 And Table 1-2.

[0108] Table 1 Metabolic biomarker related information

[0109]

[0110]

[0111] Table 2 Correlation analysis between identified key metabolic fingerprints and clinical indicators

[0112]

[0113]

[0114] It can be seen that the metabolic fingerprint screened out in this example has significant differences in the inter-group comparison and has a significant correlation with the commonly used clinical indicators such as PSA, which proves its clinical feasibility as a non-invasive detection indicator.

[0115] Example 4

[0116] This example builds a prostate cancer screening model based on machine learning based on the results of the aforementioned Example 3.

[0117] In this example, the collected plasma sample data (17 cases of prostate cancer and 19 cases of benign prostatic hyperplasia) were divided into training and test sets, and automated machine learning was performed according to the following six main processes:

[0118] 1) Data collection and processing: First, plasma mass spectrometry data were cleaned, feature extracted, normalized, and missing value filled to ensure the integrity and consistency of the data set;

[0119] 2) Data segmentation: The dataset is divided into training set and validation set according to the ratio of 8:2;

[0120] 3) Model selection: Ten machine learning algorithms, including Adaboost, GBM, Glmnet, kknn, LogitBoost, MLP, NB, RegLogistic, RF, and svmRadiaWeights, were selected using the screened key metabolic fingerprints. Models were trained using the training set data, and 10-fold cross-validation was applied to optimize model parameters.

[0121] 4) Modeling: Based on the selected model, use the training set to train the model;

[0122] 5) Hyperparameter optimization: Automatically optimize the model's hyperparameters through algorithms to improve model performance;

[0123] 6) Model Validation: An independent validation set was used to evaluate the performance of the model and compared using multiple metrics (such as AUC, accuracy, etc.). The results are shown in Table 3.

[0124] Table 3 Classification models of different prostate cancer screening models constructed based on plasma metabolic markers

[0125]

[0126]

[0127] In this example, the constructed screening model showed high stability and high accuracy (AUC of 0.95 and above) on the validation set, and the multi-model evaluation indicators were better than the traditional methods, further confirming the practical application value of the method of the present invention in the early diagnosis of prostate cancer.

[0128] It can be seen that the biomarker screening method described in the present invention has various steps that are interconnected, forming a complete detection process from plasma sample acquisition to metabolic fingerprint data analysis to machine learning screening model construction, ensuring the efficiency, accuracy and clinical applicability of the present invention in the non-invasive detection of early prostate cancer.

[0129] In summary, the present invention provides a comprehensive process from plasma sample collection, exosome extraction, metabolic fingerprint acquisition, key metabolic fingerprint screening, to machine learning screening model construction. Each step, validated by experimental data, demonstrates significant technical advantages: improved detection accuracy and specificity, shortened detection time, and reduced sample requirements and operational complexity. Compared with existing technologies, this invention offers significant advantages, providing an efficient, standardized, and scalable technical solution for the noninvasive detection of early prostate cancer.

[0130] The embodiments of the present invention are described in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A plasma exosome metabolite, characterized in that It includes one or more of the following metabolites: acylcarnitine [C18:0, COOH], hemolytic phosphatidylethanolamine [18:2], hemolytic phosphatidylcholine [22:6], F518.3169901 / ExMrn00096780, F518.3169901 / ExMrn00096781, 2-hydroxystearic acid, 9-hydroxyoctadecanoic acid, [8]-gingerol, adenosine, hexaethylene glycol, acylcarnitine [C18:2], linoleoylcarnitine [C18:2], hemolytic phosphatidylcholine [18:2 / 0:0].

2. A use of the plasma exosome metabolite according to claim 1 for preparing a prostate cancer biomarker.

3. A method for screening plasma exosome metabolites according to claim 1 by combining SEC-LC method, characterized in that: The steps include: (1) Separation of plasma samples Collect blood samples and separate plasma samples and blood cells by centrifugation; (2) SEC method for extracting exosomes The plasma was centrifuged at 2500-3500×g, the supernatant was collected, and the supernatant was further centrifuged at 8000-12000×g to collect the supernatant as a pretreatment sample; the pretreatment sample was added to the column filter plate of the equilibrium exclusion chromatography column, and when the sample just entered the column and the liquid flow stopped, the elution buffer was added, and the eluted sample was collected to obtain a sample to be tested for later use; (3) Metabolic fingerprint detection MeOH:ACN was added to the sample to be tested and the mixture was incubated and precipitated; the incubated reaction product was centrifuged, the supernatant extract was collected, evaporated to dryness, and then resuspended in ACN:H2O, centrifuged again, and the supernatant was collected for LC-MS detection; (4) Establishment of key metabolic fingerprints Univariate correlation analysis, t-test (P<0.05), and VIP score were combined to preliminarily screen candidate metabolic fingerprints between prostate cancer and benign hyperplasia groups. Lasso regression was then used to further extract the most predictive variables. In addition, by intersecting the above key metabolic fingerprints, key metabolic fingerprints with significant differences and clinical application value are screened out.

4. The method for screening plasma exosome metabolites by combining SEC-LC method according to claim 3, characterized in that: The method further includes: (5) constructing a prostate cancer screening model based on machine learning, specifically dividing the collected plasma sample data of prostate cancer and benign prostatic hyperplasia patients into a training set and a test set, and performing machine learning according to the following process: Data collection and processing: plasma mass spectrometry data are cleaned, feature extracted, normalized, and missing value filled; Data segmentation: Divide the dataset into training set and validation set according to 8:2; Model selection: Choose from ten machine learning algorithms including Adaboost, GBM, Glmnet, kknn, LogitBoost, MLP, NB, RegLogistic, RF, and svmRadiaWeights; Modeling: Based on the selected model, use the training set to train the model; Hyperparameter optimization: Automatically optimize the model's hyperparameters through algorithms to improve model performance; Model Validation: Use the validation set to evaluate the performance of the model.

5. The method for screening plasma exosome metabolites by combining SEC-LC method according to claim 3 or 4, characterized in that: In the step (3), the LC-MS detection step comprises: Octadecylsilane bonded silica gel is used as filler; Mobile phase A: water-0.1 v / v% formic acid solution; Mobile phase B: acetonitrile-0.1 v / v% formic acid solution; Perform gradient elution according to the following program: 0-1min, A%: B%=99.0%: 1.0%; 1-8min, A%: B%=99.0%: 1.0% → 1.0%: 99.0%; 8-10min, A%: B%=1.0%: 99.0%; 10min, A%: B%=99.0%: 1.0%; 10-12min, A%: B%=99.0%: 1.0%; Gradient separation: 12 minutes; Injection volume: 2 μL; Column flow rate: 500 μL / min; Column temperature: 40℃.

6. The method for screening plasma exosome metabolites by combining SEC-LC method according to claim 5, characterized in that: In the step (3), the LC-MS detection step comprises: The ion spray voltage was set to 3500 V for positive mode and 2800 V for negative mode; Full-scan spectra ( m / z 70–1200) were obtained on an Orbitrap at 60,000 resolution; The normalized automatic gain control (AGC) target was 100%, and the maximum injection time was 100 ms; The precursor ions were selected to enter the collision cell for high-energy collision dissociation (HCD) fragmentation, with collection energies of 20, 30, and 40, respectively; The MS / MS resolution was set to 30,000, the normalized AGC target was 100%, the maximum injection time was 60 ms, the isolation window was 1 m / z, and the dynamic exclusion time was 4 s.

7. An in vitro diagnostic marker model for prostate cancer, characterized in that: The biological biomarkers of the diagnostic model include the plasma exosome metabolites of claim 1.

8. Use of the plasma exosome metabolites according to claim 1 for preparing a prostate cancer diagnostic product.

9. The use according to claim 8, characterized in that The product includes a reagent for detecting the content of the plasma exosome metabolite.

10. A kit for diagnosing prostate cancer, characterized in that: The kit comprises a reagent for detecting the content of the plasma exosome metabolite as claimed in claim 1 in a sample.

Images