Biomarkers

Abstract

The invention relates to a biomarker related to colorectal cancer prediction in the technical field of biomedicine. Proper biological markers related to colorectal cancer prediction are screened through a glycoproteomics method, healthy people and colorectal cancer patients can be effectively distinguished on the basis of detection of the biological markers related to colorectal cancer prediction, sensitivity and specificity are high, a more reliable and convenient prediction method is provided for colorectal cancer prediction, and the application prospect is wide. Wide application prospects are realized in the field of colorectal cancer.

Description

Technical Field

[0001] The present application relates to the field of biomedical technology, and in particular to a biomarker associated with the prediction of colorectal cancer. Background Art

[0002] In the related technologies, the screening of colorectal cancer mainly relies on colonoscopy, fecal occult blood test (FOBT), fecal DNA detection, and traditional blood biomarkers. The related screening methods mainly include the following categories.

[0003] Colonoscopy: Although it is the most commonly used and effective screening method, colonoscopy is an invasive procedure that may cause discomfort or complications and has low compliance, which makes it difficult for many patients, especially those at high risk, to undergo effective screening at an early stage.

[0004] Fecal occult blood test (FOBT) and stool DNA test: These tests are relatively noninvasive and easy to accept, but their sensitivity and specificity are limited, especially in the detection of early cancers, it is difficult to accurately identify tiny lesions.

[0005] Traditional blood biomarkers: Although traditional biomarkers such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) are widely used in clinical practice, their performance in the diagnosis of early colorectal cancer is less than ideal, with low sensitivity and specificity, and it is difficult to meet the needs of accurate diagnosis. Summary of the invention

[0006] According to various embodiments of the present application, a biomarker is provided.

[0007] A biomarker, comprising one or more peptide structures and / or sugar chain structures, wherein the one or more peptide structures comprise a peptide sequence and a polysaccharide structure, and the peptide sequence comprises a sequence selected from any of the following groups: SEQ ID NO.1-NO.18, the glycan structure is attached to the peptide sequence, and the glycan structure includes structures in which the glycans are abbreviated as N4H5S2, N6H7F1S3, N6H7F1S4, N2H8, N7H7F1S3, N6H7F1S3, N4H4S1, N6H6F1S2, N4H5S1, N4H5F1S2, N5H6S2, N5H6F1S3, N4H5S2, N4H6F1S1, N5H6F1S3, N3H5 and N4H4F1S1; and / or the sugar chain structure includes structures in which the glycans are abbreviated as N6H7F1S3, N6H7S4, N2H6, N6H7S3 and N7H8F1S3.

[0008] In some embodiments, the peptide sequence is selected from SEQ ID NO.1-SEQ ID NO.6, and the glycan structure is selected from structures abbreviated as N4H5S2, N6H7F1S4, N2H8, N7H7F1S3, N6H7F1S3 and N4H4S1.

[0009] In some embodiments, the peptide sequence is selected from SEQ ID NO.9-NO.13, and the glycan structure is selected from the structures in which the glycans are abbreviated as N4H5S1, N4H5H1S2, N5H6S2, N5H6H1S3 and N4H5S2; and / or the peptide sequence is selected from SEQ ID NO.14-NO.18, and the glycan structure is selected from the structures in which the glycans are abbreviated as N4H6H1S1, N5H6H1S3, N6H7H1S3, N3H5 and N4H4H1S1.

[0010] The present application also provides a method for preparing the above biomarker, comprising the following steps:

[0011] S1, enzymatically hydrolyze the clinical sample to obtain an enzymatic solution;

[0012] S2, specifically enriching the peptides in the enzymatic hydrolysate to obtain enriched glycopeptides;

[0013] S3, performing mass spectrometry identification on the enriched glycopeptides, and then performing qualitative and quantitative analysis to obtain corresponding site-specific glycopeptides and / or sugar chains;

[0014] S4, performing random forest modeling on the site-specific glycopeptides and / or sugar chains to screen and obtain biomarkers.

[0015] The present application also provides a method for predicting rectal cancer, comprising the following steps:

[0016] The above biomarkers are selected as input values ​​from the subject's site-specific glycopeptides;

[0017] Inputting the input value into a colorectal cancer prediction model for ROC analysis;

[0018] The sensitivity, specificity and AUC value obtained by ROC analysis were used to obtain auxiliary prediction results of whether the subject had colorectal cancer.

[0019] In some embodiments, the prediction method comprises the steps of:

[0020] Selecting biomarkers of SEQ ID NO.9-NO.13 from the subject's site-specific glycopeptides as input values;

[0021] Inputting the input value into a colorectal cancer prediction model for ROC analysis;

[0022] Obtain auxiliary prediction results of whether the subject suffers from early-onset colorectal cancer based on the sensitivity and specificity obtained by ROC analysis;

[0023] and / or

[0024] Select biomarkers of SEQ ID NO.14-NO.18 from the subject's site-specific glycopeptides as input values;

[0025] Inputting the input value into a colorectal cancer prediction model for ROC analysis;

[0026] Based on the sensitivity and specificity obtained by ROC analysis, auxiliary prediction results of whether the subject has late-onset colorectal cancer are obtained.

[0027] In some embodiments, the construction of the colorectal cancer prediction model comprises the following steps:

[0028] Collect data of colorectal cancer patients as modeling sample data;

[0029] Use modeling sample data to train a machine learning classification model to build a colorectal cancer prediction model;

[0030] External validation of the colorectal cancer prediction model was performed.

[0031] The present application also provides a kit, which comprises one or more glycopeptide standards, a buffer and the above-mentioned biomarkers.

[0032] In some embodiments, the peptide sequence of the biomarker is selected from SEQ ID NO.1-SEQ ID NO.6.

[0033] In some embodiments, the peptide sequence of the biomarker contains at least sequence SEQ ID NO.9-NO.13, which is used to predict early-onset colorectal cancer, and / or the peptide sequence of the biomarker contains at least sequence SEQ ID NO.14-NO.18, which is used to predict late-onset colorectal cancer.

[0034] This application has at least the following beneficial effects:

[0035] High sensitivity and specificity: Compared with existing blood biomarkers, the biomarkers of this application have significantly higher diagnostic sensitivity and specificity, especially in distinguishing early-onset and late-onset colorectal cancer.

[0036] Non-invasive: This application is based on the detection of body fluid glycoproteins, which avoids the invasiveness of colonoscopy, greatly improves patient compliance, and is suitable for wide screening of high-risk populations.

[0037] High-throughput and precise detection: High-throughput detection and quantitative analysis of glycoproteins using liquid chromatography-mass spectrometry (LC-MS / MS) can effectively improve the resolution and accuracy of detection and significantly reduce false positive and false negative results.

[0038] Diagnostic models supported by machine learning: The predictive model built using machine learning algorithms can further improve the predictive performance of biomarkers, facilitate early detection and support individualized treatment plans. Biomarkers are abnormal molecules in the process of tumor occurrence and development, widely present in body fluids such as blood, feces, and urine, and can be used for cancer screening, diagnosis, and prognosis assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 An overview of the glycomics data for this application;

[0040] Figure 2 The multivariate ROC curves of the six biomarkers that distinguished the healthy control group from the colorectal cancer group in all validation groups of plasma samples for this application;

[0041] Figure 3 The multivariate ROC curves of the five biomarkers used to distinguish the healthy control group from the early-onset colorectal cancer group in the early-onset validation group of plasma samples for this application are shown;

[0042] Figure 4 The multivariate ROC curves of the four biomarkers used to distinguish the healthy control group from the late-onset colorectal cancer group in the late-onset validation group of plasma samples in this application are shown;

[0043] Figure 5 Multivariate ROC curves of the five biomarkers used to distinguish between the healthy control group and the colorectal cancer group in all validation groups of plasma samples for this application. DETAILED DESCRIPTION

[0044] In order to facilitate understanding of the present application, the present application will be described in more detail below. However, it should be understood that the present application can be implemented in many different forms and is not limited to the embodiments or examples described herein. On the contrary, the purpose of providing these embodiments or examples is to make the understanding of the disclosure of the present application more thorough and comprehensive.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by those skilled in the art belonging to the technical field of the present application. The terms used herein in the specification of the present application are only for the purpose of describing specific implementation methods or embodiments, and are not intended to limit the present application. The optional range of the term "and / or" used herein includes any one of two or more related listed items, and also includes any and all combinations of related listed items, and the any and all combinations include any combination of two related listed items, any more related listed items, or all related listed items.

[0046] Ⅰ Definition

[0047] The polysaccharide symbol nomenclature (SNFG) used to illustrate polysaccharides is used herein to refer to polysaccharides. The explanation of this graphical system can be obtained on the website of www.ncbi.nlm.nih.gov / glycans / snfg.html, and its entire contents are incorporated herein by reference as a whole for all purposes. The symbol nomenclature of the graphical representation of polysaccharides published in Glycobiology 25:1323-1324,2015 can be obtained on the website of doi.org / 10.1093 / glycob / cwv091. Other information of the diagram of the SNFG system is as follows. In this system, the item HexNAC_i uses i to indicate the quantity of gray solid squares (N-acetylglucosamine), which is abbreviated to N. The item Hex_j: is explained as follows: j indicates the quantity of circular (mannose, gray solid) and the quantity of (galactose, hollow), which is abbreviated to H. The item Fuc_d uses d to indicate the quantity of gray solid triangles (fucose), which is abbreviated to F. The term Neu5AC_l uses l to indicate the number of gray solid diamonds (sialic acid), abbreviated as S. The glycan reference code used herein combines these i, j, d and l terms together to form a 4-5 digit complex glycan reference code, such as HexNAc (6) Hex (7) Fuc (1) NeuAc (3), abbreviated as N6H7F1S3. The graphic representation of N6H7F1S3 contains 7 circles (3 mannose and 4 galactose), 6 squares (N-acetylglucosamine) and 3 diamonds (sialic acid), and also includes 1 triangle (fucose).

[0048] Biomarkers are described herein. These biomarkers can be used for a variety of applications, including but not limited to predicting diseases and conditions. For example, certain biomarkers described herein or a combination thereof can be used to diagnose colorectal cancer. In some embodiments, the biomarkers described herein or a combination thereof can be used to classify patients so that patients receive appropriate medical treatment. In some embodiments, the biomarkers described herein or a combination thereof can be used to treat or improve a patient's disease or condition by, for example, identifying a therapeutic agent for treating a patient. In some embodiments, the biomarkers described herein or a combination thereof can be used to predict the prognosis of a patient's treatment or the likelihood of success or survival of a treatment regimen.

[0049] Described herein are biomarkers selected from the group consisting of glycans, glycopeptides, and combinations thereof.

[0050] In some embodiments, the biomarker is a glycopeptide.

[0051] In some embodiments, the biomarker is a sugar chain.

[0052] Site-specific glycopeptide: A peptide chain that binds a sugar moiety (usually an oligosaccharide or polysaccharide) at a specific amino acid residue position in a glycoprotein. These glycosylation modifications are highly site-specific.

[0053] N-glycosylation: N-glycosylation is a co-post-translational modification process in which sugar chains are attached to aspartic acid residues (Asn) in proteins through nitrogen atoms, usually occurring in the sequence Asn-X-Ser / Thr.

[0054] In some embodiments, the glycopeptides described herein include N-glycosylated peptides. Wherein, the glycan is bound to the peptide via the nitrogen atom of the amino acid. Typically, the amino acid bound to the glycan is asparagine (N).

[0055] In some embodiments, described herein are glycopeptides comprising an amino acid sequence selected from the group consisting of SEQ ID NO. 1-NO. 18 and combinations thereof.

[0056] In some embodiments, described herein are glycopeptides consisting of an amino acid sequence selected from the group consisting of SEQ ID NO. 1-NO. 18 and combinations thereof.

[0057] In some embodiments, the glycopeptide amino acid sequence comprises a glycopeptide consisting of an amino acid sequence selected from the group consisting of SEQ ID NO.1-NO.6 and a group consisting of combinations thereof.

[0058] In a specific example, the glycopeptide according to SEQ ID NO. 1-NO. 6 further comprises a glycan, wherein the glycan is bound to the amino acid residue.

[0059] Ⅱ Methods for screening and validating biomarkers

[0060] In some embodiments, the present invention describes a method for screening and / or validating biomarkers, comprising: obtaining or having obtained a clinical sample from a subject; digesting and / or fragmenting one or more glycoproteins in the sample into one or more glycopeptides; enriching site-specific glycopeptides; and separating, qualitatively and quantitatively analyzing the enriched glycopeptides to obtain differentially expressed site-specific glycopeptides; analyzing the expression levels of the site-specific glycopeptides, and constructing a model based on the analysis.

[0061] In some embodiments, the analysis comprises a multivariate ROC analysis.

[0062] In some embodiments, the site-specific glycopeptides obtained by screening correspond to at least one peptide structure contained in a clinical sample.

[0063] In some embodiments, at least one site-specific glycopeptide structure comprises one or more glycopeptides comprising the amino acid sequence of any one of SEQ ID NO. 1-NO. 18.

[0064] In some embodiments, the method comprises detecting a glycopeptide, a glycan on the glycopeptide, and a glycosylation site residue to which the glycan is bound to the glycopeptide.

[0065] In some embodiments, the methods include detecting glycan residues.

[0066] In some embodiments, the method comprises detecting a glycosylation site on the glycopeptide.

[0067] In some embodiments, detection is accomplished by mass spectrometry coupled to liquid chromatography.

[0068] In the present embodiment, the clinical sample is a plasma sample.

[0069] In some embodiments, the method further comprises digesting and / or fragmenting the glycoprotein in the clinical sample. In certain instances, the method comprises digesting the glycoprotein in the clinical sample.

[0070] In some embodiments, the methods include fragmenting glycoproteins in a sample.

[0071] In some embodiments, the glycoprotein is digested or fragmented in solution phase using a digestive enzyme.

[0072] In some embodiments, the enzyme is trypsin.

[0073] In some embodiments, the methods include enriching digested and / or fragmented glycopeptides.

[0074] In some embodiments, the method comprises performing tandem liquid chromatography-mass spectrometry analysis on the enriched glycopeptide fragments.

[0075] In some embodiments, the method comprises using a machine learning algorithm to model and screen for biomarkers.

[0076] In some embodiments, the machine learning algorithm is selected from the group consisting of: a deep learning algorithm, a neural network algorithm, an artificial neural network algorithm, a supervised machine learning algorithm, a linear discriminant analysis algorithm, a quadratic discriminant analysis algorithm, a support vector machine algorithm, a linear basis function kernel support vector algorithm, a radial basis function kernel support vector algorithm, a random forest algorithm, a genetic algorithm, a nearest neighbor algorithm, a k-nearest neighbor, a naive Bayes classifier algorithm, a logistic regression algorithm, a regularized regression algorithm, or a combination thereof.

[0077] In an embodiment of the present application, the machine learning algorithm is a random forest algorithm.

[0078] In some embodiments, the method comprises using a model derived from machine learning modeling to distinguish between colorectal cancer and healthy controls.

[0079] In some embodiments, the method includes using a model derived from machine learning modeling to distinguish between early-onset colorectal cancer and late-onset colorectal cancer.

[0080] In some embodiments, the method comprises administering a therapeutic regimen to the biological sample.

[0081] In some embodiments, the treatment regimen for a patient having colorectal cancer (CRC) or a patient suspected of having CRC is selected from surgery, an antimetabolite, a chemotherapeutic therapy, a topoisomerase inhibitor, an alkylating agent, a targeted therapeutic agent, an immunotherapeutic agent, an immunotherapy, an antibody, a T cell-related therapy, radiation therapy, or a combination thereof.

[0082] In some embodiments, the method includes using a model derived from machine learning modeling to assess the patient's condition after treatment.

[0083] In some embodiments, the method includes using a model derived from machine learning modeling to predict a patient's response to a treatment method.

[0084] The present invention also provides a kit, comprising one or more glycopeptide standards, a buffer and one or more peptides selected from the group consisting of the amino acid sequences shown in SEQ ID NO.1-NO.18.

[0085] In some embodiments, described herein is a kit for predicting and / or monitoring cancer in a patient by determining a glycan or glycopeptide profile of a clinical sample from the patient and comparing the measured profile with that of a normal patient or a patient with a family history of cancer.

[0086] In some embodiments, the kit comprises one or more glycopeptide standards, a buffer, and one or more peptides selected essentially from the group consisting of the amino acid sequences shown in SEQ ID NO.9-NO.13.

[0087] In some embodiments, described herein is a kit for predicting and / or monitoring cancer in a patient by determining a glycan or glycopeptide profile of a clinical sample from the patient and comparing the measured profile with the profile of a normal patient or a patient with a family history of early-onset colorectal cancer.

[0088] In some embodiments, the kit comprises one or more glycopeptide standards, a buffer, and one or more peptides selected essentially from the group consisting of the amino acid sequences shown in SEQ ID NOs. 14-18.

[0089] In some embodiments, described herein is a kit for predicting and / or monitoring cancer in a patient by determining a glycan or glycopeptide profile of a clinical sample from the patient and comparing the measured profile with the profile of a normal patient or a patient with a family history of late-onset colorectal cancer.

[0090] III. Specific Embodiments

[0091] The raw materials used in the examples are all commercially available and described in detail. The present application is further described in detail below in conjunction with the examples.

[0092] Example 1

[0093] Biomarker screening method for colorectal cancer prediction

[0094] 1. Sample collection:

[0095] The samples collected by the research institute include the discovery cohort and the validation cohort, and both groups of samples were collected by themselves. Among them, the discovery cohort includes 160 blood samples from the early-onset colorectal cancer group (EOCRC) and its healthy control group (EC), and 10 blood samples from the late-onset colorectal cancer group (LOCRC) and its healthy control group (LC); the validation cohort includes 40 blood samples from the early-onset colorectal cancer group and its healthy control group, and 10 blood samples from the late-onset colorectal cancer group and its healthy control group (LC), a total of 40 cases (as shown in Table 1).

[0096] The whole blood samples of the subjects were collected using purple-capped anticoagulant tubes (BD, America) containing ethylene diamine tetraacetic acid (EDTA) anticoagulant. The samples were centrifuged at 1000×g for 10 min at 4°C within 6 h, and the upper plasma was taken into a 1.5 ml sterile centrifuge tube, which was numbered and labeled. The samples were centrifuged at 12000×g for 2 min at 4°C, and the upper clear liquid was taken to obtain the plasma samples, which were stored in a -80°C refrigerator.

[0097] The sample selection criteria are:

[0098] ① Patients with colorectal cancer of all stages diagnosed by histopathology, aged 18-50 years for early-onset colorectal cancer group and over 50 years for late-onset colorectal cancer group;

[0099] ②The control group consisted of healthy volunteers with no history of colorectal cancer or other malignant tumors, and matched the patient group in gender and age;

[0100] ③All subjects signed the informed consent.

[0101] Exclusion criteria:

[0102] ① Failure to obtain informed consent from the subjects or the subjects refuse to join the study;

[0103] ② Patients with incomplete clinical data;

[0104] ③ Those who have received systemic chemotherapy or other targeted therapies in the past;

[0105] ④ Patients with severe medical complications.

[0106] Table 1. Plasma sample information of the discovery cohort and validation cohort

[0107]

[0108] 2. Discovery queue detection method

[0109] (1) Proteolysis and peptide desalting of plasma samples from the discovery cohort:

[0110] Add 9 μL of 8 mol / L urea (urea is prepared in 0.005 mol / L sodium bicarbonate solution) to 1 μL of plasma sample from the discovery cohort, dissolve the plasma sample, add an appropriate amount of 0.2 mol / L dithiothreitol (prepared in 0.05 mol / L sodium bicarbonate solution), control the final concentration of dithiothreitol in the dissolved plasma sample to be 0.01 mol / L, incubate the protein solution at 95°C for 10 min, reduce the disulfide bonds in the protein to obtain the reduced plasma sample. Add an appropriate amount of 0.8 mol / L iodoacetamide (prepared in 0.05 mol / L sodium bicarbonate solution), control the final concentration of iodoacetamide in the reduced plasma sample to be 0.04 mol / L, and place it at room temperature (20-25°C) away from light for 30 min. Add 0.005 mol / L sodium bicarbonate solution until the urea concentration in the sample is diluted to 1 mol / L to obtain the processed plasma sample. Trypsin was mixed with the treated plasma sample at a mass ratio of 1:50, and the mixture was shaken and incubated at 37°C overnight for enzymatic hydrolysis. The next day, formic acid was added to a final concentration of 0.1%, and then centrifuged at 12,000 × g for 10 min. The supernatant was taken and the peptides were desalted using a Strata X C18 protein desalting column (America, Phenomenexcx) according to standard procedures, and the desalted peptides were vacuum dried and stored in a -80°C ultra-low temperature refrigerator for later use.

[0111] (2) Plasma glycopeptide enrichment:

[0112] Fill the C18 gasket (Empore disk C18 solid phase extraction membrane, USA) into the tip column (GELoader tip, eppendorf), prepare a maltose solution (click maltose, homemade) with a concentration of 20 μL / mg with water, and take 30 μL of maltose solution (1.5 mg maltose) to fill the tip column. Dissolve the desalted peptide (~50 μg) with 40 μL of a mixture of 80% acetonitrile and 1% trifluoroacetic acid aqueous solution (80% ACN / 1% TFA), fill the desalted peptide into the tip column filled with maltose solution, collect the sample effluent, and elute the tip column twice with 40 μL of a mixture of 80% acetonitrile and 1% trifluoroacetic acid aqueous solution (80% ACN / 1% TFA), and collect the eluent. The tip column was eluted twice with 20 μL of a mixture of 30% acetonitrile and 1% formic acid (FA) aqueous solution (30% ACN / 1% FA), and the eluate was collected. The collected load effluent and eluate were vacuum dried to obtain enriched glycopeptides, which were stored in a -80°C ultra-low temperature refrigerator for future use.

[0113] (3) Glycoprotein data obtained by high performance liquid chromatography-tandem mass spectrometry analysis:

[0114] Chromatographic conditions: Mobile phase A is a mixture of 0.1% formic acid and 2% acetonitrile aqueous solution; mobile phase B is a mixture of 0.1% formic acid and 90% acetonitrile aqueous solution. Gradient elution, four liquid phase gradient settings: 0-45min, 10%-27% mobile phase B; 45-53min, 27%-37% mobile phase B; 53-57min, 37%-100% mobile phase B; 57-60min, 100% mobile phase B. Flow rate is 0.45ml / min.

[0115] The enriched glycopeptides were analyzed by Orbitrap Exploris 480 high-resolution mass spectrometer (America, Thermo). The high-throughput mass spectrometry detection of glycoproteome was performed using the shotgun strategy and data-dependent acquisition (DDA) data acquisition method.

[0116] Mass spectrometry conditions: electrospray ion source, negative ion scanning, ion source voltage of 2600 V, ion source temperature of 320 ° C. MS / MS scanning analysis settings: primary mass spectrometry scanning range of 350-1500 m / z, scanning resolution of 60000; secondary mass spectrometry scanning range is fixed at 200 m / z, secondary scanning resolution of 30000.

[0117] Data retrieval and analysis: Raw data were processed using Byonic software version 5.4.10 (Protein Metrics), which uses the Byonic built-in database of 309 human N-glycans. The Byonic parameters were set as follows: precursor mass tolerance was 6 ppm; fragment mass tolerance was 20 ppm; digestion specificity was trypsin full specificity. Missed cleavage was set to 2. Carbamidomethylation at C was set as a fixed modification. Oxidation at M and deamidation at N were set as variable modifications. The Homo sapiens protein database (version 2016_02, 20,198 reviewed proteins) was used for searching the data set. Total common maximum and total rare maximum were both set to 1. Intact N-glycopeptides were quantified using Byologic software version 5.4.10 (Protein Metrics), which uses input from MS1 raw data and Byonic search results to determine glycopeptide intensities. Byologic quantification results were filtered for scores >100 and modification site scores >10. For the quantitative analysis of N-glycan modifications, glycopeptides were grouped by their sugar composition or sugar type, and the relative abundance of glycopeptide modifications of each sugar class was determined by the sum of the glycopeptide intensities of that sugar class normalized by the total intensity of all N-glycan-modified glycopeptides in each sample.

[0118] After quality control filtering, data normalization, and protein qualitative and quantitative analysis, a total of 31,336 site-specific glycopeptides from 1,849 glycosylation sites of 518 glycoproteins were identified. Figure 1 shown.

[0119] This step was based on data modeling of all discovery cohort plasma samples (n=160), retaining site-specific glycopeptides detected in at least 70% of the samples in each group as candidate biomarkers, and performing multivariate ROC analysis. 3 / 4 of the modeling group sample data were randomly divided into a training set (training) and 1 / 4 as a test set (test). The training set was used to build and train the machine learning classification model, and the test set was used to verify the discriminative ability of the model. The colorectal cancer prediction model was constructed by iterating the random forest for 1000 times and calculating the average value of the final model accuracy. In this step, the top 8 site-specific glycopeptides were selected as biomarkers for the diagnosis of colorectal cancer. The 8 site-specific glycopeptides include O14638_533_N4H5S2, P00738_241_N6H7F1S3, P02763_72_N6H7F1S4, P0C0L4_226_N2H8, A2VEC9_3063_N7H7F1S3, P02763_72_N6H7F1S3, P04114_1523_N4H4S1, and P04004_242_N6H6F1S2. The 8 site-specific glycopeptides are shown in Table 5.

[0120] The ROC curve is a method to study the relationship between model sensitivity and specificity. It uses sensitivity as the ordinate and 1-specificity as the abscissa to evaluate model performance by comparing the AUC under the curve. The closer the AUC is to 1, the better the model performance and the better the diagnostic effect. If it is less than 0.5, it means that the model accuracy is poor.

[0121] 3. Model Validation

[0122] Model validation was performed using all validation cohort plasma samples:

[0123] This step is the same as the detection and analysis method of the discovery cohort. In order to verify the effectiveness of the constructed colorectal cancer prediction model, the plasma samples of the validation cohort were used as unknown samples for verification in this step. The first 8 site-specific glycopeptides and their combinations shown in Table 5 were used as biomarkers. These 8 site-specific glycopeptides and their combinations in the plasma samples obtained from the validation group analysis were input into the colorectal cancer prediction model for ROC analysis, and all markers and their combinations with AUC values ​​greater than 0.85 were output. The validation results showed that the six glycopeptide combination biomarkers (O14638_533_N4H5S2, P02763_72_N6H7F1S4, P0C0L4_226_N2H8, A2VEC9_3063_N7H7F1S3, P02763-72-N5H6H1S3, and P04114_1523_N4H4S1) had an AUC of 0.88, a sensitivity of 75%, and a specificity of 95%. Figure 2 The results show that the constructed colorectal cancer prediction model shows good performance on unknown samples, can output auxiliary prediction results of whether the subject has colorectal cancer, and can support its application in clinical practice.

[0124] Example 2

[0125] Biomarker screening method for specific early-onset colorectal cancer diagnosis

[0126] The difference between this embodiment and embodiment 1 is that this embodiment is based on the data modeling of the plasma samples (n=80) of the discovery cohort of early-onset colorectal cancer and its control group, retains the site-specific glycopeptides detected in at least 50% of the samples in each group as candidate biomarkers, and performs multivariate ROC analysis. 3 / 4 of the sample data of the modeling group were randomly divided into a training set, and 1 / 4 as a test set. The training set is used to build and train the machine learning classification model, and the test set is used to verify the discriminative ability of the model. The early-onset colorectal cancer model was constructed by iterating the random cycle of the random forest 1000 times and calculating the average value of the final model accuracy. In this step, the top 5 site-specific glycopeptides are selected as biomarkers for the diagnosis of early-onset colorectal cancer. The five site-specific glycopeptides include P02675-394-N4H5S1, P02765-176-N4H5F1S2, P05155-253-N5H6S2, P10909-374-N5H6F1S3, and P19827-285-N4H5S2. The five site-specific glycopeptides are shown in Table 5.

[0127] Model validation was performed using all early-onset validation group plasma samples:

[0128] This step is the same as the detection and analysis method of the discovery cohort. In order to verify the effectiveness of the constructed colorectal cancer prediction model, this step uses the plasma samples of the early-onset validation cohort as unknown samples for verification. Five site-specific glycopeptides and their combinations (P02675-394-N4H5S1, P02765-176-N4H5F1S2, P05155-253-N5H6S2, P10909-374-N5H6F1S3, and P19827-285-N4H5S2) were input into the early-onset colorectal cancer diagnostic model for ROC analysis, and all markers and their combinations with AUC values ​​greater than 0.85 were output. The validation results showed that the five site-specific glycopeptide combination biomarkers (P02675-394-N4H5S1, P02765-176-N4H5H1S2, P05155-253-N5H6S2, P10909-374-N5H6H1S3 and P19827-285-N4H5S2) had an AUC of 0.96, a sensitivity of 80.0%, and a specificity of 100%. Figure 3 The results show that the constructed early-onset colorectal cancer prediction model shows good performance on unknown samples, and can output auxiliary prediction results of whether the subject has early-onset colorectal cancer, which can support its application in clinical practice.

[0129] Example 3

[0130] Biomarker screening method for predicting late-onset colorectal cancer

[0131] The difference between this embodiment and embodiment 1 is that this embodiment is based on the data modeling of the plasma samples (n=80) of the discovery cohort of late-onset colorectal cancer and its control group, retains the site-specific glycopeptides detected in at least 50% of the samples in each group as candidate markers, and performs multivariate ROC analysis. 3 / 4 of the sample data of the modeling group were randomly divided into a training set, and 1 / 4 as a test set. The training set is used to build and train the machine learning classification model, and the test set is used to verify the discriminative ability of the model. The prediction model for late-onset colorectal cancer was constructed by iterating the random cycle of the random forest 1000 times and calculating the average value of the final model accuracy. In this step, the first five site-specific glycopeptides were selected as biomarkers for the diagnosis of colorectal cancer: P02765-176-N4H6F1S1, P02763-72-N5H6F1S3, P02763-93-N6H7F1S3, P01024-85-N3H5, and P01024-85-N4H4F1S1. The five site-specific glycopeptides are shown in Table 5.

[0132] Model validation was performed using all late validation group plasma samples:

[0133] This step is the same as the detection and analysis method of the discovery cohort. In order to verify the effectiveness of the constructed late-onset colorectal cancer prediction model, this step uses the plasma samples of the late-onset validation cohort as unknown samples for verification. The first five site-specific glycopeptides and combinations (P02765-176-N4H6F1S1, P02763-72-N5H6F1S3, P02763-93-N6H7F1S3, P01024-85-N3H5, P01024-85-N4H4F1S1) were used to construct a specific late-onset colorectal cancer diagnostic model for ROC analysis, and all markers and their combinations with AUC values ​​greater than 0.85 were output. The validation results showed that the four glycopeptide combination biomarkers (P02765-176-N4H6F1S1, P02763-72-N5H6F1S3, P01024-85-N3H5 and P01024-85-N4H4F1S1) had an AUC of 0.87, a sensitivity of 100.0%, and a specificity of 70.0%. Figure 4 The results show that the constructed late-onset colorectal cancer prediction model shows good performance on unknown samples, and can output auxiliary prediction results of whether the subject has late-onset colorectal cancer, which can support its application in clinical practice.

[0134] Example 4

[0135] Screening method of sugar chains as biomarkers for colorectal cancer prediction

[0136] The difference between this embodiment and embodiment 1 is that this embodiment is based on data modeling of all discovery cohort plasma samples (n=160), retains sugar chains detected in at least 70% of the samples in each group as candidate biomarkers, and performs multivariate ROC analysis. 3 / 4 of the modeling group sample data were randomly divided into a training set, and 1 / 4 was used as a test set. The training set is used to build and train a machine learning classification model, and the test set is used to verify the discriminative ability of the model. The colorectal cancer prediction model was constructed by iterating the random cycle of the random forest 1000 times and calculating the average value of the final model accuracy. In this step, the first 5 sugar chains (N6H7F1S3, N6H7S3, N6H7S4, N7H8F1S3, N2H6) were selected as biomarkers for diagnosing colorectal cancer, and the 5 sugar chains are shown in Table 6.

[0137] In order to verify the effectiveness of the constructed diagnostic model, all validation group plasma samples (n = 40) were used as unknown samples for verification in this step. The colorectal cancer diagnostic model constructed by combining the first five sugar chains, namely N6H7F1S3, N6H7S3, N6H7S4, N7H8F1S3, and N2H6, was subjected to ROC analysis, and all sugar chain biomarkers and their combinations with AUC values ​​greater than 0.85 were output. Only N6H7F1S3 had AUC = 0.83, sensitivity = 80.0%, and specificity = 83.3%. The results are as follows Figure 5 The results show that the constructed colorectal cancer diagnosis model shows good performance on unknown samples, can output auxiliary prediction results of whether the subject has colorectal cancer, and can support its application in clinical practice.

[0138] IV. Biomarker-related characteristics

[0139] Table 2. Numbers of glycopeptides from the glycopeptide group

[0140]

[0141]

[0142] Table 3. Glycan residue compound number, molecular weight and composition and their symbolic structure

[0143]

[0144]

[0145] Table 4. Glycoproteins associated with healthy controls and disease samples

[0146]

[0147]

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154]

[0155]

[0156] Table 5. Detailed information on site-specific glycopeptides as biomarkers

[0157]

[0158]

[0159] Table 6: Detailed information of sugar chains as biomarkers

[0160]

[0161]

[0162] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0163] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A biomarker, characterized in that: The method comprises one or more peptide structures and / or sugar chain structures, wherein the one or more peptide structures comprise a peptide sequence and a polysaccharide structure, and the peptide sequence comprises a sequence selected from any of the following groups: SEQ IDNO.1-NO.18, the glycan structure is attached to the peptide sequence, and the glycan structure includes structures whose glycans are abbreviated as N4H5S2, N6H7F1S3, N6H7F1S4, N2H8, N7H7F1S3, N6H7F1S3, N4H4S1, N6H6F1S2, N4H5S1, N4H5F1S2, N5H6S2, N5H6F1S3, N4H5S2, N4H6F1S1, N5H6F1S3, N3H5 and N4H4F1S1; and / or the sugar chain structure includes structures whose glycans are abbreviated as N6H7F1S3, N6H7S4, N2H6, N6H7S3 and N7H8F1S3.

2. The biomarker according to claim 1, characterized in that: The peptide sequence is selected from SEQ ID NO.1-SEQ ID NO.6, and the glycan structure is selected from structures abbreviated as N4H5S2, N6H7F1S4, N2H8, N7H7F1S3, N6H7F1S3 and N4H4S1.

3. The biomarker according to claim 1, characterized in that: wherein the peptide sequence is selected from SEQ ID NO.9-NO.13, and the glycan structure is selected from structures in which the glycans are abbreviated as N4H5S1, N4H5H1S2, N5H6S2, N5H6H1S3 and N4H5S2; and / or the peptide sequence is selected from SEQ ID NO.14-NO.18, and the glycan structure is selected from structures in which the glycans are abbreviated as N4H6H1S1, N5H6H1S3, N6H7H1S3, N3H5 and N4H4H1S1.

4. A method for preparing a biomarker, characterized in that: The biomarker is selected from any one of claims 1 to 3, comprising the following steps: S1, enzymatically hydrolyze the clinical sample to obtain an enzymatic solution; S2, specifically enriching the peptides in the enzymatic hydrolysate to obtain enriched glycopeptides; S3, performing mass spectrometry identification on the enriched glycopeptides, and then performing qualitative and quantitative analysis to obtain corresponding site-specific glycopeptides and / or sugar chains; S4, performing random forest modeling on the site-specific glycopeptides and / or sugar chains to screen and obtain biomarkers.

5. A method for predicting colorectal cancer, characterized in that: The following steps are involved: Selecting the biomarker of claim 1 from the subject's site-specific glycopeptides as an input value; Inputting the input value into a colorectal cancer prediction model for ROC analysis; Based on the sensitivity, specificity and AUC value obtained by ROC analysis, auxiliary prediction results of whether the subject has early-onset colorectal cancer are obtained.

6. The prediction method according to claim 5, characterized in that: The steps include: Selecting biomarkers of SEQ ID NO.9-NO.13 from the subject's site-specific glycopeptides as input values; Inputting the input value into a colorectal cancer prediction model for ROC analysis; Obtain auxiliary prediction results of whether the subject has early-onset colorectal cancer based on the sensitivity and specificity obtained by ROC analysis; and / or Select biomarkers of SEQ ID NO.14-NO.18 from the subject's site-specific glycopeptides as input values; Inputting the input value into a colorectal cancer prediction model for ROC analysis; Based on the sensitivity and specificity obtained by ROC analysis, auxiliary prediction results of whether the subject has late-onset colorectal cancer are obtained.

7. The prediction method according to claim 5, characterized in that The construction of the colorectal cancer prediction model comprises the following steps: Collect data of colorectal cancer patients as modeling sample data; Use modeling sample data to train a machine learning classification model to build a colorectal cancer prediction model; External validation of the colorectal cancer prediction model was performed.

8. A kit, characterized in that: The method comprises one or more glycopeptide standards, a buffer and a biomarker, wherein the biomarker is selected from claim 1.

9. A kit according to claim 8, characterized in that: The peptide sequence of the biomarker is selected from SEQ ID NO.1-SEQ ID NO.

6.

10. A kit according to claim 8, characterized in that: The peptide sequence of the biomarker contains at least the sequence SEQ ID NO.9-NO.13, which is used to predict early-onset colorectal cancer, and / or the peptide sequence of the biomarker contains at least the sequence SEQ ID NO.14-NO.18, which is used to predict late-onset colorectal cancer.

Images