Marker composition for identifying different types of breast fibroepithelial tumors, kit, diagnostic model construction method and application

Abstract

The invention provides a marker composition for identifying different types of breast fibroepithelial tumors, a kit, a diagnostic model construction method and application, and particularly belongs to the technical field of pathological molecular diagnosis products. The marker composition disclosed by the invention comprises any one or more than two of the following markers: MAL2, KRT8, DSP (Digital Signal Processor), cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381. The invention further discloses a kit for detecting the biomarker. A diagnostic model formed by the marker composition provided by the invention can well distinguish PT from FA and different grades of PT, and is superior to a previously reported diagnostic model.

Description

Technical Field

[0001] The present invention belongs to the technical field of pathological molecular diagnostic products, and particularly relates to a biomarker composition, a kit, a diagnostic model construction method and an application for differentiating different types of breast fibroepithelial tumors. Background Art

[0002] Phyllodes Tumor of Breast (PT) belongs to fibroepithelial lesions (FELs), which are composed of breast fibroconnective tissue and epithelial tissue. PT is not a pure single disease, but a general term for a series of fibroepithelial tumors with different clinical processes and histopathological characteristics. Pathologically, it is divided into three types: benign phyllodes tumor (benign PT), borderline phyllodes tumor (borderline PT), and malignant phyllodes tumor (malignant PT). Different types of PT all have a certain recurrence risk; among them, malignant PT also has a relatively high metastasis risk, about 16 - 25%, and once distant metastasis occurs, the prognosis is extremely poor, with an average overall survival of only 10.7 - 11.5 months.

[0003] The main treatment method for PT is surgery - an extended resection of the tumor with at least 1 cm negative margin. There is no evidence that radiotherapy and chemotherapy can reduce the recurrence or death risk. However, PT has similar clinical manifestations to fibroepithelial tumors that do not require extended resection - fibroadenoma (FA), and it is often impossible to distinguish them through preoperative core needle biopsy (CNB), resulting in doctors being unable to design an accurate surgical scope before surgery. For patients who are diagnosed with PT only after surgery, a second operation is required to meet the margin requirements specified in the guidelines and reduce the recurrence risk. The diagnosis of fibroepithelial tumors, including FA, benign PT, borderline PT, and malignant PT, mainly relies on the morphological characteristics of HE staining, which has a certain degree of subjectivity. The differential diagnosis between fibroadenoma and benign phyllodes tumor and the histological grading of phyllodes tumor are the difficulties in the preoperative and intraoperative diagnosis of PT, and pathologists with different experience levels often draw different diagnostic conclusions. Moreover, the differential diagnosis between benign, borderline, and malignant PT often shows inconsistencies among institutions or pathologists due to the lack of effective molecular markers. Through the analysis of 285 patients with FELs who underwent preoperative puncture and postoperative pathological diagnosis, the results showed that the diagnostic consistency before and after surgery was only 37.6%; among the patients who were preoperatively diagnosed as FA, 31.2% were diagnosed as PT postoperatively, and these patients had to undergo a second operation to meet the margin requirements. Therefore, new diagnostic strategies and methods are urgently needed for the preoperative accurate diagnosis of PT, guiding surgical decisions, optimizing the diagnosis and treatment process, and bringing convenience to clinical work and patient management.

[0004] In recent years, more and more studies have focused on biomarkers for the differential diagnosis of PT. Cani et al. and Tan et al. respectively performed targeted sequencing and exome sequencing on breast FELs and found that FELs is a disease characterized by somatic mutations in exon 2 of recurrent MED12, but this feature cannot distinguish the four types of FELs. Piscuoglio et al. and Otsuji et al. respectively sequenced and verified TERTp mutations by MSK-IMPACT sequencing and ddPCR, and found that TERT alterations may promote PT progression, but this mutation still cannot completely distinguish PT from FA. Zhang et al. found through proteomic sequencing that there are no significant differences in proteomic features between low-grade PT and FA, and its limitation is that malignant PT samples are not included. Hench et al. and Meyer et al. performed DNA methylation microarray detection on PT and found that PT has different methylation patterns from breast cancer and normal breast tissue, and identified methylation features for differentiating malignant PT from metaplastic breast cancer, but did not conduct in-depth research on the differential diagnosis between PT and FA. In summary, single-omics studies have not solved the clinical dilemmas of preoperative differential diagnosis between PT and FA and differential diagnosis between different types of PT. Summary of the Invention

[0005] The purpose of the present invention is to provide a marker composition, a kit, a method for constructing a diagnostic model and an application for differentiating different types of breast fibroepithelial tumors. The diagnostic model constituted by the marker composition provided by the present invention can well distinguish PT and FA and different grades of PT, and this diagnostic model is superior to the previously reported diagnostic models.

[0006] The present invention provides a marker composition for differentiating breast phyllodes tumors and fibroadenomas and / or differentiating different grades of breast phyllodes tumors. The different grades of breast phyllodes tumors include benign breast phyllodes tumors, borderline breast phyllodes tumors and malignant breast phyllodes tumors. The marker composition includes any one or more of the following markers: MAL2, KRT8, DSP, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381; wherein, MAL2, KRT8 and DSP are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381 are DNA methylation biomarkers in the human genome.

[0007] The present invention also provides a marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades, wherein the phyllodes tumors of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast, and malignant phyllodes tumors of the breast, and the marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233, and cg04404381; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233, and cg04404381 are DNA methylation biomarkers in the human genome.

[0008] The present invention also provides a marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades, wherein the phyllodes tumors of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast, and malignant phyllodes tumors of the breast, and the marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, and cg04404381; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, and cg04404381 are DNA methylation biomarkers in the human genome.

[0009] The present invention also provides a marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades, wherein the phyllodes tumors of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast, and malignant phyllodes tumors of the breast, and the marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, and cg09025210; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, and cg09025210 are DNA methylation biomarkers in the human genome.

[0010] The present invention also provides the use of a reagent for detecting the biomarker composition described in the above technical solution in the preparation of a kit for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast.

[0011] The present invention also provides a kit for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast. The kit includes a reagent for detecting the biomarker composition described in the above technical solution. The reagent for detecting the biomarker composition described in the above technical solution includes a reagent for detecting a protein biomarker and a reagent for detecting a methylation biomarker.

[0012] The reagent for detecting a protein biomarker includes a reagent used in any one of the following methods: PRM targeted protein sequencing method, selected reaction monitoring targeted sequencing method, multiple reaction monitoring targeted sequencing method, tandem mass tag quantitative protein sequencing method, isotope tag quantitative protein sequencing method, western blotting method, enzyme-linked immunosorbent assay, and immunohistochemical staining.

[0013] The reagent for detecting a DNA methylation biomarker includes a reagent used in any one of the following methods: MethylTarget targeted methylation sequencing method, methylated DNA immunoprecipitation sequencing method, degenerate representative bisulfite sequencing method, whole genome bisulfite sequencing method, deep target region DNA methylation sequencing method, and pyrosequencing method.

[0014] The present invention also provides a method for constructing a diagnostic model for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast, including the following steps:

[0015] Using the biomarker composition described in the above technical solution to construct a diagnostic model for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast.

[0016] The types of the diagnostic model include any one or more of the following: naive Bayes model, support vector machine model, multiple logistic regression model, decision tree model, random forest model, K-nearest neighbor algorithm model, neural network model, adaptive boosting algorithm model, gradient boosting machine model, extreme gradient boosting algorithm model, least absolute shrinkage and selection operator model, elastic net regression model, and ridge regression model.

[0017] The present invention also provides a diagnostic model device for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast. The device includes: a memory and a processor. The memory is used for storing program instructions. The processor is used for calling the program instructions, and when the program instructions are executed, it is used for performing the following operations:

[0018] Obtain the detection data of the biomarker composition of the sample to be tested, and input the detection data into a diagnostic model constructed by the biomarker composition for differentiating phyllodes tumor and fibroadenoma of the breast and / or differentiating phyllodes tumors of different grades, so as to obtain the diagnostic result of the sample to be tested;

[0019] The biomarker composition is the biomarker composition described in the above technical solution.

[0020] The present invention also provides a diagnostic system for differentiating phyllodes tumor and fibroadenoma of the breast and / or differentiating phyllodes tumors of different grades, including:

[0021] An acquisition unit, configured to obtain the detection data of the biomarker composition of the sample to be tested;

[0022] A processing unit, configured to input the detection data of the biomarker composition into a diagnostic model constructed by the biomarker composition for differentiating phyllodes tumor and fibroadenoma of the breast and / or differentiating phyllodes tumors of different grades, so as to obtain the diagnostic result of the sample to be tested;

[0023] The biomarker composition is the biomarker composition described in the above technical solution.

[0024] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the following method is implemented:

[0025] Obtain the detection data of the biomarker composition of the sample to be tested, and input the detection data of the biomarker composition into a diagnostic model constructed by the biomarker composition for differentiating phyllodes tumor and fibroadenoma of the breast and / or differentiating phyllodes tumors of different grades, so as to obtain the diagnostic result of the sample to be tested;

[0026] The biomarker composition is the biomarker composition described in the above technical solution.

[0027] The present invention provides a marker composition for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades. The present invention determines the optimal diagnostic markers for PT through joint analysis of high-throughput omics data at multiple levels and from multiple sources. Specifically, through multi-omics detection, the present invention identifies molecular markers for differentiating four types of fibroepithelial tumors, including molecular markers of proteins and methylation sites. The diagnostic model composed of them can well distinguish PT and FA as well as PT of different grades, and this diagnostic model is superior to the previously reported diagnostic models. The test results show that the present invention first screens 30 proteins and 20 methylation sites from the high-throughput proteomics and methylomics data of 40 FELs patients as candidate diagnostic markers for differentiating different subtypes of FELs, and then uses the varImpPlot function in the random forest algorithm to rank the importance of these 50 candidate markers, and verifies the top 10 markers by targeted sequencing; through a new independent cohort, 74 FELs samples are simultaneously subjected to PRM targeted proteomics and MethylTarget targeted methylomics sequencing. The results show that the markers described in the present invention can be used alone to achieve FA vs benign PT vs borderline PT vs malignant PT. When the markers of the present invention are combined for the differentiation of FA vs benign PT vs borderline PT vs malignant PT, the effect is better. Among them, when the number of markers is 5, 6, 7, 8, 9, and 10 combinations, they have better diagnostic efficiency; when the number of markers is 7 combinations, the effect is the best. In two independent XJTU cohorts, the diagnostic efficiency of the four-classification diagnostic panel for FELs constructed based on the multiple logistic regression (MLR) machine learning algorithm is the best, and its accuracy, specificity, and sensitivity all reach 100%. In addition, compared with the previously reported two-classification diagnostic panel, the present invention uses the binary logistic regression method to find that the diagnostic panel composed of 7 key markers also has high specificity (100%) and sensitivity (100%) for differentiating PT and FA, benign PT and borderline / malignant PT. Brief Description of the Drawings

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0029] Figure 1The volcano plot of differentially expressed proteins among multiple groups provided by the present invention; wherein, each point in the volcano plot represents a gene, red represents up-regulated proteins (P<0.05, log2FC≥1), blue represents down-regulated proteins (P<0.05, log2FC≤1), and gray represents proteins with no significant change between the two groups;

[0030] Figure 2 The hierarchical clustering heat map of differentially expressed proteins provided by the present invention; wherein, each row represents a protein, the expression level of each protein is subjected to Z-score transformation, and the color of blue gradually deepens from 0 to -2, representing lower protein expression levels, and the color of red gradually deepens from 0 to 3, representing higher protein expression levels; each column represents a sample, and the clinical information corresponding to each sample is shown at the top;

[0031] Figure 3 The volcano plot of differentially methylated sites among multiple groups provided by the present invention; wherein, each red point represents a CpG site, and only DMPs with calibrated P<0.05 and |Δβ|≥0.2 are shown; Δβ≥0.2 are up-regulated DMPs; Δβ≤-0.2 are down-regulated DMPs; Δβ, deltabeta;

[0032] Figure 4 The clustering heat map of differentially methylated sites provided by the present invention; wherein, each row represents a CpG site, the methylation level of each CpG site is subjected to Z-score transformation, and the color of blue gradually deepens from 0 to -3, representing lower methylation levels, and the color of red gradually deepens from 0 to 3, representing higher methylation levels; each column represents a sample, and the clinical information corresponding to each sample is shown at the top;

[0033] Figure 5 The screening flow chart of diagnostic candidate markers provided by the present invention;

[0034] Figure 6 The screening flow chart of diagnostic markers and patient cohorts in the validation cohort provided by the present invention;

[0035] Figure 7 The importance ranking result map of 50 candidate markers by the random forest algorithm in XJTU cohort2 provided by the present invention;

[0036] Figure 8Cumulative AUC value result graph of three machine learning models in XJTU cohort 2 provided by the present invention; wherein, the abscissa represents the number of features included, the ordinate represents the cumulative AUC value, and the data table is the cumulative AUC value corresponding to each machine learning algorithm when including features in sequence; the red box indicates that when the first 7 markers are included, the AUC of the three machine learning algorithms all reaches the highest point; NB, Naive Bayes; SVM, Support Vector Machine; MLR, Multivariate Logistic Regression;

[0037] Figure 9 Cumulative AUC value result graph of three machine learning models in XJTU cohort 1 provided by the present invention; wherein, the abscissa represents the number of features included, the ordinate represents the cumulative AUC value, and the data table is the cumulative AUC value corresponding to each machine learning algorithm when including features in sequence; the red box indicates that when the first 7 markers are included, the AUC of the three machine learning algorithms all reaches the highest point; NB, Naive Bayes; SVM, Support Vector Machine; MLR, Multivariate Logistic Regression;

[0038] Figure 10 Diagnostic efficacy (AUC, specificity and sensitivity) result graph of seven markers for differentiating different classification outcomes provided by the present invention; wherein, extending from the center point 0 to the outermost circle, the value represents from 0 to 1; different colors represent different diagnostic indicators;

[0039] Figure 11 Result graph of the expression differences of two diagnostic proteins in different grades of PT provided by the present invention; wherein, (a) Box plot of the expression of two diagnostic proteins in XJTU cohort 1; (b) Box plot of the expression of two diagnostic proteins in XJTU cohort 2; ns P>0.05, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001;

[0040] Figure 12 Result graph of the methylation level differences of five diagnostic methylation sites in different grades of PT provided by the present invention; wherein, (a) Box plot of the methylation level of five diagnostic methylation sites in XJTU cohort 1; (b) Box plot of the methylation level of five diagnostic methylation sites in XJTU cohort 2; ns P>0.05, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001;

[0041] Figure 13Schematic diagram of the decision tree for the future clinical application of the 7-molecule diagnostic panel provided by the present invention; among them, the marked AUC values are from XJTU Cohort 2, and the red-marked "hyper" indicates that the methylation level of this locus in FA is higher than that in PT; the green-marked "low" indicates that the expression level of this protein in malignant PT is lower than that in other FELs subtypes.

[0042] Figure 14 Comparison result graph of the multi-omics diagnostic model provided by the present invention and the traditional model; among them, (a) Comparison graph of the diagnostic model of the present invention and the previous diagnostic model for differentiating FA and PT; (b) Comparison graph of the diagnostic model of the present invention and the previous diagnostic model for differentiating benign PT and borderline / malignant PT. Detailed implementation manners

[0043] The present invention provides a marker composition for differentiating phyllodes tumor of breast and fibroadenoma and / or differentiating phyllodes tumors of different grades. The phyllodes tumors of different grades include benign phyllodes tumor of breast, borderline phyllodes tumor of breast and malignant phyllodes tumor of breast. The marker composition comprises any one or more than two of the following markers: MAL2 (Myelin and Lymphocyte protein 2), KRT8 (Keratin 8), DSP (Desmoplakin), cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381; wherein, MAL2, KRT8 and DSP are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381 are DNA methylation biomarkers in the human genome. The results show that the methylation sites of cg22449901 (XJTUcohort 1: sensitivity 100% and specificity 100%; XJTU cohort2: sensitivity 93.3% and specificity 95%) and cg24852561 (XJTU cohort 1: sensitivity 100%; specificity 100%; XJTU cohort 2: sensitivity 93.3% and specificity 90%) can be used alone as differential diagnosis markers for FA vs benign PT; the methylation site of cg09025210 (XJTUcohort 1: sensitivity 80% and specificity 100%; XJTU cohort2: sensitivity 90% and specificity 95%) can be used alone as a differential diagnosis marker for benign PT vs borderline PT; KRT8 protein (XJTU cohort 1: sensitivity 100% and specificity 80%; XJTU cohort2: sensitivity 94.7% and specificity 95%) and MAL2 protein (XJTUcohort 1: sensitivity 100% and specificity 100%; XJTU cohort 2: sensitivity 100% and specificity 95%) can both be used alone as differential diagnosis markers for malignant PT vs borderline PT; MAL2 protein (XJTU cohort 1: sensitivity 96.3% and specificity 100%; XJTU cohort 2: sensitivity 100% and specificity 94.5%) can be used alone as a differential diagnosis marker for malignant PT vs Other FELs (FA, BPT and BorPT).When 10 markers are jointly used for the diagnosis of FA vs benign PT vs borderline PT vs malignant PT, the diagnostic AUCs of the three machine learning algorithms are NB = 0.861, SVM = 0.917, MLR = 1 (XJTU cohort 1) and NB = 0.945, SVM = 0.947, MLR = 0.956 (XJTU cohort 2). The differential diagnosis between PT and FA and the histological grading of PT are the difficulties in the preoperative diagnosis of PT. The present invention focuses on this clinical problem of difficult differential diagnosis of PT, and constructs a molecular diagnosis model on the basis of previous multi-omics sequencing data, aiming to improve the diagnostic efficiency and guide surgical decisions. On the basis of the previous multi-omics detection results, the present invention finds that integrin and methylomics can better distinguish different subtypes of FELs, so diagnostic candidate markers for FELs are preliminarily screened.

[0044] The present invention also provides a marker composition for differentiating phyllodes tumor of breast and fibroadenoma and / or differentiating phyllodes tumors of different grades of breast, wherein the phyllodes tumors of different grades include benign phyllodes tumor of breast, borderline phyllodes tumor of breast and malignant phyllodes tumor of breast, and the marker composition comprises any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381 are DNA methylation biomarkers in the human genome. When 9 markers are jointly used for the diagnosis of FA vs benign PT vs borderline PT vs malignant PT, the diagnostic AUCs of the three machine learning algorithms are NB = 0.908, SVM = 0.917, MLR = 1 (XJTU cohort 1) and NB = 0.967, SVM = 0.969, MLR = 0.956 (XJTU cohort 2).

[0045] The present invention also provides a marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades, wherein the phyllodes tumors of different grades include benign phyllodes tumors, borderline phyllodes tumors and malignant phyllodes tumors, and the marker composition comprises any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210 and cg04404381; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210 and cg04404381 are DNA methylation biomarkers in the human genome. When the 8 markers are combined for FA vs benign PT vs borderline PT vs malignant PT diagnosis, the diagnostic AUCs of 3 machine learning algorithms are NB = 0.861, SVM = 0.908, MLR = 1 (XJTU cohort 1) and NB = 0.967, SVM = 0.975, MLR = 0.915 (XJTU cohort 2).

[0046] The present invention also provides a marker composition for differentiating phyllodes tumor of breast and fibroadenoma and / or differentiating phyllodes tumors of different grades, wherein the phyllodes tumors of different grades include benign phyllodes tumor of breast, borderline phyllodes tumor of breast and malignant phyllodes tumor of breast, and the marker composition comprises any one or more than two of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901 and cg09025210; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901 and cg09025210 are DNA methylation biomarkers in the human genome. The present invention further verifies a diagnostic model composed of 7 biological markers, including 2 proteins (MAL2 and KRT8) and 5 methylation sites (cg00515756, cg24852561, cg00686880, cg22449901, cg09025210), in an external cohort (XJTU cohort 2) with an enlarged sample size by using PRM and MethlTraget targeted sequencing technologies and 3 machine learning algorithms. The results show that cg24852561 and cg22449901 have outstanding contributions when applied alone to differentiate FA vs PT or FA vs benign PT, and their AUC, specificity and sensitivity in the discovery cohort and the validation cohort are all greater than 0.9; cg00686880 has an outstanding contribution when applied alone to differentiate malignant PT vs OtherFELs, and its AUC, specificity and sensitivity in the discovery cohort and the validation cohort are all greater than 0.7; cg09025210 has an outstanding contribution when applied alone to differentiate benign PT vs malignant PT, and its AUC, specificity and sensitivity in the discovery cohort and the validation cohort are all greater than 0.8; cg00515756, KRT8 and MAL2 have outstanding contributions when applied alone to differentiate malignant PT vs borderline PT, malignant PT vs OtherFELs, and their AUC, specificity and sensitivity in the discovery cohort and the validation cohort are all greater than 0.778. In addition, the present invention finds that when the 7 markers are combined for the diagnosis of FA vs benign PT vs borderline PT vs malignant PT, the diagnostic AUCs of the 3 machine learning algorithms are NB = 0.944, SVM = 0.95, MLR = 1 (XJTU cohort 1) and NB = 0.971, SVM = 0.975, MLR = 1 (XJTU cohort 2), and the results are better than the differential diagnostic efficacies when 10, 9 or 8 markers are combined.

[0047] The present invention also provides the use of a reagent for detecting the biomarker composition described in the above technical solution in the preparation of a kit for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast.

[0048] The present invention also provides a kit for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades of breast. The kit includes a reagent for detecting the biomarker composition described in the above technical solution. The reagent for detecting the biomarker composition described in the above technical solution includes a reagent for detecting protein biomarker and a reagent for detecting methylation biomarker.

[0049] The reagents for detecting protein markers include the reagents used in any of the following methods: PRM targeted protein sequencing method, selected reaction monitoring targeted sequencing method, multiple reaction monitoring targeted sequencing method, tandem mass tag quantitative protein sequencing method (TMT), isotope-coded affinity tag quantitative protein sequencing method (iTRAQ), western blotting (WB), enzyme-linked immunosorbent assay (ELSIA), and immunohistochemical staining (IHC); the reagents for detecting DNA methylation biomarkers include the reagents used in any of the following methods: MethylTarget targeted methylation sequencing method, methylated DNA immunoprecipitation sequencing method, degenerate representation bisulfite sequencing method, whole genome bisulfite sequencing method, deep target region DNA methylation sequencing method, and pyrosequencing method. In a specific embodiment, when the reagent for detecting DNA methylation biomarkers is the reagent used in the MethylTarget targeted methylation method, the reagent used in the MethylTarget targeted methylation method includes targeted methylation sequencing primers; the targeted methylation sequencing primers include any one or more pairs of the following primers: the forward primer of cg00515756 with the nucleotide sequence shown in SEQ ID NO.1 and the reverse primer of cg00515756 with the nucleotide sequence shown in SEQ ID NO.2, the forward primer of cg24852561 with the nucleotide sequence shown in SEQ ID NO.3 and the reverse primer of cg24852561 with the nucleotide sequence shown in SEQ ID NO.4, the forward primer of cg00686880 with the nucleotide sequence shown in SEQ ID NO.5 and the reverse primer of cg00686880 with the nucleotide sequence shown in SEQ ID NO.6, the forward primer of cg22449901 with the nucleotide sequence shown in SEQ ID NO.7 and the reverse primer of cg22449901 with the nucleotide sequence shown in SEQ ID NO.8, the forward primer of cg09025210 with the nucleotide sequence shown in SEQ ID NO.9 and the reverse primer of cg09025210 with the nucleotide sequence shown in SEQ ID NO.10, the forward primer of cg00819233 with the nucleotide sequence shown in SEQ ID NO.11 and the reverse primer of cg00819233 with the nucleotide sequence shown in SEQ ID NO.12, the forward primer of cg04404381 with the nucleotide sequence shown in SEQ ID NO.13 and the reverse primer of cg04404381 with the nucleotide sequence shown in SEQ ID NO.14.

[0050] The present invention also provides a method for constructing a diagnostic model for differentiating phyllodes tumor of breast and fibroadenoma and / or differentiating phyllodes tumors of different grades of breast, comprising the following steps:

[0051] Construct a diagnostic model for differentiating phyllodes tumor and fibroadenoma of breast and / or differentiating phyllodes tumors of different grades by using the marker composition described in the above technical solution;

[0052] The types of the diagnostic model include any one or more of the following: Naive Bayes model, Support Vector Machine model, Multivariate Logistic Regression model, Decision Tree model, Random Forest model, K-Nearest Neighbors (KNN) model, Neural Networks model, Adaptive Boosting (AdaBoost) model, Gradient Boosting Machine (GBM) model, Extreme Gradient Boosting (XGBoost) model, Least Absolute Shrinkage and Selection Operator (lasso) model, Elastic Network (EN) model, and Ridge model. By detecting proteins and methylation sites of fibroepithelial tumors and using a machine learning model, the present invention can accurately diagnose the tumor type. The diagnostic efficacy evaluation indicators such as specificity, sensitivity, accuracy, and consistency of the molecular diagnostic model are all excellent. The seven markers of the present invention can achieve the discrimination between FA vs. benign PT, FA vs. PT, benign PT vs. borderline PT, malignant PT vs. borderline PT, and malignant PT vs. Other FELs tumors.Specifically as follows: cg00515756, cg00686880, cg22449901, cg24852561 and KRT8 can be combined to distinguish FA from PT; cg00515756, cg00686880, cg09025210, cg22449901, cg24852561 and KRT8 can be combined to distinguish FA from benign PT; cg00515756, cg00686880, cg09025210, cg22449901, cg24852561 and MAL2 can be combined to distinguish benign PT from borderline PT; cg00515756, cg00686880, cg09025210, KRT8 and MAL2 can be combined to distinguish malignant PT from borderline PT; cg00515756, cg00686880, cg09025210, cg22449901, cg24852561, KRT8 and MAL2 can be combined to distinguish malignant PT from Other FELs (the bold markers can also distinguish the two types of tumors alone, that is, cg22449901 and cg24852561 can distinguish FA from PT alone; cg22449901 and cg24852561 can distinguish FA from benign PT alone; cg09025210 can distinguish benign PT from borderline PT alone; cg00515756, KRT8 and MAL2 can distinguish malignant PT from borderline PT alone; KRT8 and MAL2 can distinguish malignant PT from Other FELs alone; but the diagnostic efficacy is the best when multiple markers are combined). In addition, compared with the previous diagnostic models or panels for distinguishing FA from PT or benign PT from borderline PT / malignant PT, the combination of the seven markers of the present invention has higher diagnostic efficacy, and its AUC, specificity and sensitivity are 1, 100% and 100% respectively.

[0053] The present invention also provides a diagnostic model device for distinguishing phyllodes tumor and fibroadenoma of breast and / or for distinguishing phyllodes tumors of different grades, the device comprising: a memory and a processor; the memory is used for storing program instructions; the processor is used for calling the program instructions, and when the program instructions are executed, is used for performing the following operations:

[0054] Obtain the detection data of the marker composition of the sample to be tested, input the detection data into a diagnostic model for distinguishing phyllodes tumor and fibroadenoma of breast and / or for distinguishing phyllodes tumors of different grades constructed by the marker composition, and obtain the diagnostic result of the sample to be tested;

[0055] The marker composition is the marker composition described in the above technical solution.

[0056] The present invention also provides a diagnostic system for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades of the breast, comprising:

[0057] An acquisition unit for acquiring detection data of a marker composition of a sample to be tested;

[0058] A processing unit that inputs the detection data of the marker composition into a diagnostic model constructed by the marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades of the breast, to obtain a diagnostic result of the sample to be tested;

[0059] The marker composition is the marker composition described in the above technical solution.

[0060] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the following method is implemented:

[0061] Acquire detection data of a marker composition of a sample to be tested, input the detection data of the marker composition into a diagnostic model constructed by the marker composition for differentiating phyllodes tumors and fibroadenomas of the breast and / or differentiating phyllodes tumors of different grades of the breast, to obtain a diagnostic result of the sample to be tested;

[0062] The marker composition is the marker composition described in the above technical solution.

[0063] To further illustrate the present invention, the following describes in detail a marker composition, a kit, a diagnostic model construction method and an application for differentiating different types of breast fibroepithelial tumors provided by the present invention with reference to the accompanying drawings and embodiments, but they should not be construed as limiting the protection scope of the present invention.

[0064] Example 1

[0065] Screening of diagnostic molecules for phyllodes tumors of the breast based on high-throughput DIA proteomics and 850k chip methylation sequencing (XJTU Cohort 1)

[0066] 1 Patient clinical data and sample collection

[0067] 1) Sample collection: Paraffin-embedded tumor specimens were collected from 40 patients with FELs (named XJTU cohort 1) who underwent breast mass resection, extended resection of breast mass, or total mastectomy in our hospital from January 2015 to December 2020; among them, there were 10 wax blocks of FA, benign PT, borderline PT, and malignant PT tissues each. Inclusion criteria: Medical records with double-blind reading and consistent diagnosis by two pathologists with senior technical titles were included in the study. Exclusion criteria: Medical records with inconsistent diagnosis results by the two pathologists or samples that did not meet the sequencing quality control requirements. After quality assessment, all 40 samples met the requirements of proteomics, and 36 samples met the requirements of DNA methylation sequencing (8 cases of FA, 9 cases of benign PT, 10 cases of borderline PT, and 9 cases of malignant PT).

[0068] 2) Clinical data collection: The inpatient number, age, height, weight, side of tumor occurrence, tumor size in the ultrasound report at the time of initial diagnosis, pathological diagnosis by puncture at the time of initial diagnosis, surgical conditions after initial diagnosis (surgery time, surgical method, surgical margin), and postoperative specimen pathological report (pathological diagnosis, tumor size, tumor boundary, stromal cell density, cellular atypia, mitotic figure, stromal hyperplasia, whether accompanied by heterologous components, whether there is intratumoral hemorrhage, whether there is intratumoral necrosis) of the patients were collected.

[0069] 3) Patient follow-up: Through telephone follow-up and case review, mainly record whether there is recurrence and metastasis. If there is recurrence / metastasis, it is necessary to record the time, location of recurrence and metastasis, whether the recurrence focus / metastasis focus has undergone pathological puncture or surgery, and record the pathological diagnosis in the pathological report of puncture or surgery to provide clinicopathological and prognostic information for subsequent analysis.

[0070] 2 Downloading and processing of GEO datasets

[0071] The methylation dataset of breast FELs diseases, GSE231574 (including 2 cases of FA, 2 cases of benign PT, 9 cases of borderline PT, 18 cases of malignant PT, and 2 cases of breast metaplastic carcinoma), was downloaded from the official website of the GEO database (https: / / www.ncbi.nlm.nih.gov / geo / ). Since the present invention does not involve the research on breast metaplastic carcinoma, these 2 samples were excluded, and the remaining 31 FELs patients were used as an external validation cohort for screening candidate diagnostic methylation sites. The previous data cleaning (such as removing samples or probes with too many missing values), annotation, and normalization were also performed using the ChAMP R package in R software. Subsequently, based on the probe IDs of the differentially methylated sites with AUC≥0.9 screened in XJTU cohort 1 of the present invention as the unique identifiers, the sample methylation data corresponding to the aforementioned probe IDs were extracted from the GSE231574 dataset, and the diagnostic AUC values of each methylation site were calculated. The 850k methylation data of 40 FELs in the present invention were combined for further screening and validation of candidate diagnostic methylation sites.

[0072] DIA quantitative proteomic detection

[0073] The DIA quantitative proteomic detection of the present invention was provided by Novogene Bioinformatics Technology Co., Ltd.

[0074] Specific experimental steps:

[0075] 1) Tissue protein extraction

[0076] Take out the tissue sample from the -80°C refrigerator, grind it into a powder state while keeping it cold, and then immediately transfer it to a centrifuge tube pre-cooled with liquid nitrogen; add the PASP protein lysate prepared in a specific ratio to the centrifuge tube. This lysate contains 100 mM ammonium bicarbonate and 8 M urea, and its pH value is adjusted to 8. Mix it evenly by shaking, and ultrasonically treat it in an ice-water bath for 5 min to ensure sufficient lysis. Subsequently, centrifuge at 12,000 rpm for 15 min at 4°C.

[0077] 2) Protein reduction and alkylation

[0078] Obtain the supernatant after centrifugation, add 10 mM DTT to it, and let it react at 56°C for 1 h to ensure the reduction reaction of the protein. Next, add a sufficient amount of IAM and continue to react at room temperature in an environment avoiding light for 1 h to complete the protein alkylation process.

[0079] 3) Protein precipitation and purification

[0080] After the reaction is completed, add acetone with a volume four times that of the solution, which has been pre-cooled at -20°C, to the solution. Then place the mixture in an environment at -20°C and allow it to precipitate for at least 2 h. After precipitation, centrifuge at 12,000 rpm for 15 min at 4°C and collect the obtained precipitate. Resuspend and wash the precipitate with 1 mL of acetone that has been pre-cooled at -20°C. Subsequently, perform centrifugation again under the same conditions, then collect the obtained precipitate and let it air-dry naturally. Finally, add an appropriate amount of protein dissolution solution (this solution contains 8 M urea and 100 mM TEAB, and the pH value has been adjusted to 8.5) to dissolve the air-dried protein precipitate.

[0081] 4) Protein concentration determination

[0082] Prepare a BSA standard protein solution (0 - 0.5 μg / μL) according to the instructions of the Bradford protein quantification kit. Add the standard solution and the sample to be measured to a 96-well plate, 20 μL per well, and repeat 3 times. Add the G250 staining solution, let it stand for 5 min, and measure the absorbance at 595 nm. Calculate the protein concentration of the sample to be measured according to the standard curve.

[0083] 5) Gel electrophoresis analysis

[0084] In this invention, 20 μg of the protein sample to be measured is selected and analyzed by 10% SDS-PAGE gel electrophoresis technology. After the electrophoresis is completed, stain with Coomassie Brilliant Blue until the bands are clearly visible.

[0085] 6) Protease digestion treatment

[0086] Take an appropriate amount of protein sample and dissolve it in the DB solution. Add trypsin and TEAB buffer, and digest at 37°C for 4 h; add trypsin and CaCl again 2 Digest overnight; next, adjust the pH < 3, centrifuge at 12,000 rpm for 5 min; take the supernatant after centrifugation and pass it through a C 18 Desalting column, wash 3 times and elute; collect the filtrate after passing through the desalting column and perform freeze-drying treatment.

[0087] 7) DDA spectral library construction and DIA mass spectrometry analysis

[0088] In the spectral library construction stage, the lyophilized samples after dissolution were fractionated by an HPLC system. After merging and lyophilization, they were dissolved in 0.1% formic acid again. Liquid chromatography-mass spectrometry (LC-MS) detection was performed in the data-dependent acquisition (DDA) mode, and the HPLC and mass spectrometer were combined to generate raw data for constructing the DDA spectral library. Subsequently, liquid chromatography-tandem mass spectrometry (LC-MS / MS) analysis in the data-independent acquisition (DIA) mode was carried out. After preparing the mobile phase with a specific ratio, treating the samples and adding iRT reagent, detection was performed using an advanced nano-UHPLC system and a Q ExactiveTM HF-X mass spectrometer. During the detection process, various parameters of the mass spectrometer were set in detail, including the scan range, resolution, C-trap capacity, etc., and MS / MS analysis was carried out, finally generating the original mass spectrometry detection data.

[0089] 4 Search library and information analysis process

[0090] On the one hand, for the Raw files of the offline spectra obtained by the DDA scanning mode, database searching and spectrum deconvolution were performed using Proteome Discoverer to convert the spectrum information into protein information. At the same time, DDA data quality control was carried out, and the quality-controlled data was imported into Spectronaut to construct a real spectral library, DDAlibrary. On the other hand, the Raw files of the offline spectra obtained by the DIA scanning mode were compared to the DDAlibrary by Spectronaut for protein identification.

[0091] Specific experimental steps:

[0092] 1) Database search and parameter setting

[0093] First, based on the homo_sapiens_uniprot_2020_7_2.fasta protein database (containing 192,320 sequences), the mass spectrometry data obtained in the DDA scanning mode was comprehensively searched using the Proteome Discoverer 2.2 (PD2.2, provided by Thermo) software. During the search, strict settings were made for the mass tolerances of precursor ions and fragment ions; the mass tolerance of precursor ions was 10 ppm, while the mass tolerance of fragment ions was 0.02 Da.

[0094] 2) Result filtering and quality control

[0095] To enhance the accuracy and reliability of the analysis results, the present invention uses PD2.2 software to deeply filter and screen the search results. During this process, the present invention only retains the spectral peptides with a credibility exceeding 99%, and those proteins that contain at least one unique peptide segment, regarding them as trustworthy proteins. In addition, the present invention also eliminates the peptides and proteins with FDR>1%. After this series of strict screening and verification processes, 34,379 peptides and 6,870 proteins were successfully identified in the DDA mode.

[0096] 3) Spectral library generation and DIA data analysis

[0097] The search results of PD2.2 software were imported into Spectronaut (version 14.0, provided by Biognosys) software to generate a spectral library. The present invention screened peptides and product ions according to the standard to generate a Target list. Then, the DIA data was processed with Spectronaut software, and the chromatographic peaks were extracted for matching and calculation to complete the qualitative and quantitative analysis of peptides. At the same time, the present invention also corrected the retention time with iRT standards to improve the accuracy. During the analysis process of Spectronaut, the PrecursorQ value cutoff parameter was set to 0.01 to ensure the reliability of the analysis results. Finally, 29,305 peptides and 5,541 proteins were successfully identified in 40 breast FELs samples in the DIA mode.

[0098] 5 Protein differential expression analysis and clustering analysis

[0099] To judge the significance of protein differences between groups, the linear model method built in the limma software package was used to perform a t-test on the relative quantitative data of each protein in the two groups of samples, and the corresponding P-values were calculated one by one. The criteria for screening differential proteins are as follows: if the Fold Change (FC) value of the protein ≥2.0 and P<0.05, it is regarded as an up-regulated protein; if the FC value of the protein ≤0.5 and P<0.05 at the same time, it is determined as a down-regulated protein. The up- and down-regulated proteins between groups screened according to the above conditions were visualized using ggplot2 to draw a volcano plot among multiple groups. The results showed that as the histological grade difference between the two groups of tumors increased, the number of DEPs screened was also more, otherwise vice versa ( Figure 1, volcano plot of differentially expressed proteins among multiple groups; among them, each point in the volcano plot represents a gene, red represents up-regulated proteins (P<0.05, log2FC≥1), blue represents down-regulated proteins (P<0.05, log2FC≤1), and gray represents proteins with no significant change between the two groups). In addition, the hierarchical clustering was implemented and visualized using the hclust function and pheatmap in the R package, and the expression levels of proteins could be intuitively displayed through the color changes. The results showed that FA and benign PT had similar protein expression levels, while borderline PT and malignant PT had similar protein expression levels( Figure 2 , hierarchical clustering heatmap of differentially expressed proteins; among them, each row represents a protein, and the expression level of each protein is transformed by Z-score. The color of blue gradually deepens from 0 to -2, representing lower protein expression levels, and the color of red gradually deepens from 0 to 3, representing higher protein expression levels; each column represents a sample, and the clinical information corresponding to each sample is shown at the top).

[0100] 850k DNA methylomics sequencing

[0101] The 850k DNA methylomics sequencing (Illumina Infinium Methylation EPIC BeadChip) of the present invention was serviced by Novogene Bioinformatics Technology Co., Ltd. in Beijing.

[0102] The specific experimental steps are as follows:

[0103] 1) DNA extraction

[0104] The QIAamp DNA FFPE Tissue Kit from QIAGEN was used to precisely extract DNA from FFPE (formalin-fixed paraffin-embedded tissue) samples. This process includes dewaxing treatment, protease digestion, DNA extraction, and subsequent purification steps to ensure high-quality DNA samples are obtained.

[0105] 2) DNA quantification and quality control

[0106] The concentration and volume of the extracted DNA were measured using a UV spectrophotometer. In addition, agarose gel electrophoresis was used to analyze the degradation degree of DNA and the presence of impurities such as RNA and proteins to ensure the purity and integrity of the samples.

[0107] 3) DNA bisulfite conversion

[0108] According to "EZ DNA Methylation-Gold TMKit》Kit Guide, which converts unmethylated cytosine in DNA into uracil, while methylated cytosine remains unchanged, laying the foundation for subsequent methylation sequencing.

[0109] 4) DNA processing and on-machine testing after transformation

[0110] (1) Starting amount requirement: Ensure at least 200 ng of transformed DNA as the starting sample;

[0111] (2) DNA amplification: denaturation and neutralization pretreatment to prepare for the amplification reaction;

[0112] (3) Isothermal amplification: overnight amplification under appropriate conditions to achieve uniform amplification of the entire genome;

[0113] (4) Enzyme-controlled fragmentation: accurately process the amplified product to avoid excessive fragmentation;

[0114] (5) DNA precipitation and resuspension: DNA was precipitated using isopropanol and resuspended in hybridization buffer;

[0115] (6) Chip hybridization: Add the processed DNA sample to the chip and anneal with the specific probe;

[0116] (7) Chip washing: remove unhybridized and non-specific hybridized DNA to ensure a clean environment;

[0117] (8) Extension and staining: Perform single-base extension reaction on the chip and add detection markers;

[0118] (9) Scanning imaging: Use laser to excite fluorescent groups and record images through a high-resolution scanning system.

[0119] 5) Data analysis

[0120] (1) Data quality control and standardization: After loading the original idat file generated by EPIC, the first step is to strictly quality control the raw data and standardize it with BMIQ to ensure the accuracy and comparability of the data. The probe site quality control includes the following points: First, remove those probes with P ≥ 0.01 and probes with < 3 beads in the sample and > 5% of the total number of beads; second, remove all probes of non-CpG sites in the data set and multi-hit probes with multiple matches; in addition, filter out SNPs within 5 base pairs around the CpG site and all probes located on chromosomes X and Y to eliminate sex-specific effects.

[0121] (2) Calculation and evaluation of methylation level: Using the original signal data, the present invention can calculate the methylation level of each site. Usually, the beta value (β) is used as the evaluation criterion, ranging from 0 to 1. Judgment criterion: β > 0.8 indicates high methylation, β < 0.2 indicates low methylation, and 0.2 ≤ β ≤ 0.8 indicates partial methylation.

[0122] 7 Analysis of differentially methylated sites

[0123] Differentially methylated CpG positions (DMP) are the main part of methylation research and are of great significance for finding subsequent biomarkers. The DMP analysis of the present invention uses the champ.DMP function in the champ package. This function internally calls the limma package to perform linear regression analysis and t-test, and at the same time performs multiple hypothesis testing to obtain the corrected P value. The screening criteria for differentially methylated sites are: P < 0.05 after calibration and |Δβ| ≥ 0.2. The up- and down-regulated methylated sites between groups screened according to this condition are visualized by drawing a volcano plot among multiple groups using the ggplot2 package in R software. The results show that as the histological grade difference between the two groups of tumors increases, the number of DMPs screened is also more, otherwise vice versa ( Figure 3 , volcano plot of differentially methylated sites among multiple groups; among them, each red dot represents a CpG site, and only DMPs with P < 0.05 after calibration and |Δβ| ≥ 0.2 are shown. Δβ ≥ 0.2 are up-regulated DMPs; Δβ ≤ -0.2 are down-regulated DMPs. Δβ, deltabeta). In addition, hierarchical clustering and visualization are implemented using the hclust function and pheatmap in R software, and the methylation level of methylated sites can be intuitively displayed through the change of colors. The results show that benign PT and borderline PT have similar methylation levels, while FA and malignant PT have their own methylation patterns ( Figure 4 , clustering heat map of differentially methylated sites; among them, each row represents a CpG site, the methylation level of each CpG site is transformed by Z-score, and the color gradually deepens from 0 to -3 in blue, representing lower methylation level, and the color gradually deepens from 0 to 3 in red, representing higher methylation level; each column represents a sample, and the clinical information corresponding to each sample is shown at the top).

[0124] 8 Screening of diagnostic markers for phyllodes tumors

[0125] Based on the differential analysis results of proteomics ( Figure 2 ) and DNA methylomics ( Figure 4 ) of the aforementioned 40 FEL samples, the present invention finds that integrating the two can more effectively distinguish FA, benign PT, borderline PT, and malignant PT. Therefore, as followsFigure 5 (Flow chart for screening diagnostic candidate markers) For the screening process shown, in the present invention, among the previously identified 1072 differentially expressed proteins (DEPs) and 94507 differentially methylated positions (DMPs), the DEPs with AUC≥0.9 were ranked, and the top 30 were selected as candidate diagnostic proteins; however, during the screening process of diagnostic methylation positions, the 850k methylation data of GSE231574 was also used to verify the DMPs with AUC≥0.9 in the present invention. The DMPs that simultaneously met the requirement of AUC≥0.9 were ranked, and the top 20 were selected as candidate diagnostic methylation positions. Finally, the varImpPlot function in the random forest algorithm was used to rank the importance of these 50 candidate markers, and the top 10 markers were verified by targeted sequencing.

[0126] Example 2

[0127] Verification of candidate diagnostic molecules for phyllodes tumors of the breast based on PRM and MethlyTarget targeting technologies

[0128] 1 Collection of patient sequencing samples

[0129] Paraffin-embedded tumor specimens were collected from 88 patients with FELs (named XJTU cohort 2) who underwent breast mass resection, extended resection of breast mass, or total mastectomy at the First Affiliated Hospital of Xi'an Jiaotong University from January 2021 to September 2023; among them, there were 23 cases of FA, 21 cases of benign PT, 21 cases of borderline PT, and 23 cases of malignant PT (because the number of patients with FA and benign PT was large, only the patient information of those who had inpatient surgery in the first week of each year was selected); the inclusion and exclusion criteria were the same as those in step 1) of Example 1. The specific screening process is as Figure 6 (Flow chart for patient screening in the validation cohort) shown. Since MethlyTarget targeted methylation sequencing has more stringent sample quality control requirements, after quality control of 88 samples, 82 remaining samples met the MethlyTarget sequencing requirements (18 cases of FA, 21 cases of benign PT, 21 cases of borderline PT, and 22 cases of malignant PT); in addition, 80 FELs samples were selected for PRM targeted protein quantitative sequencing (20 cases of FA, 20 cases of benign PT, 20 cases of borderline PT, and 20 cases of malignant PT). Therefore, due to reasons such as sequencing quality control and sample volume limitations, only 74 FELs samples (including 15 cases of FA, 20 cases of benign PT, 20 cases of borderline PT, and 19 cases of malignant PT) were simultaneously subjected to PRM targeted proteomics and MethylTarget targeted methylomics sequencing to further verify 10 diagnostic candidate markers (including 3 proteins and 7 methylation positions).

[0130] 2 PRM targeted quantitative proteomic detection

[0131] The PRM targeted quantitative proteomics of the present invention was provided by Novogene Bioinformatics Technology Co., Ltd.

[0132] Specific experimental steps:

[0133] 1) Tissue protein extraction

[0134] The steps are the same as those in 1) of step 3 in Example 1.

[0135] 2) Protein reduction and alkylation

[0136] The steps are the same as those in 2) of step 3 in Example 1.

[0137] 3) Protein precipitation and purification

[0138] The steps are the same as those in 3) of step 3 in Example 1.

[0139] 4) Protein concentration determination

[0140] The steps are the same as those in 4) of step 3 in Example 1.

[0141] 5) Gel electrophoresis analysis

[0142] The steps are the same as those in 5) of step 3 in Example 1.

[0143] 6) Protease digestion

[0144] The steps are the same as those in 6) of step 3 in Example 1.

[0145] 7) LC-MS detection (pre-experiment)

[0146] (1) Mobile phase preparation: Prepare solution A (made by mixing water and 0.1% formic acid) and solution B (made by mixing 80% acetonitrile and 0.1% formic acid);

[0147] (2) Sample treatment and injection: Mix an equal amount of peptide segments, take 1 μg for label-free liquid chromatography-mass spectrometry detection, and use a pre-column and an analytical column of specific specifications;

[0148] (3) Mass spectrometry parameter setting: Set the ion spray voltage and temperature, collect data in data-dependent mode, and set parameters such as mass range and resolution;

[0149] (4) MS / MS analysis: Select the top 20 high-intensity precursor ions for fragmentation, set the MS / MS resolution, and optimize the parameters to generate high-quality data;

[0150] (5) Data processing and peptide selection: Search the library using PD2.2 software, select 1-3 unique peptides for each protein, and perform three PRM acquisitions;

[0151] (6) Chromatography and mass spectrometry condition adjustment: Replace the pre-column, repeat the liquid chromatography process, adjust the mass spectrometry settings to full scan plus PRM scan, and optimize the PRM parameters;

[0152] (7) Data analysis: Use Skyline software to analyze the mass spectrometry data and evaluate the applicability of the peptides.

[0153] (8) LC-MS detection (formal experiment)

[0154] (1) Re-prepare the mobile phase: Repeat the preparation of mobile phase A and B to ensure the accurate composition ratio;

[0155] (2) Sample treatment and internal standard addition: Take equal amounts of the peptides after enzymatic digestion of each sample and mix them evenly, and uniformly incorporate a specific amount of heavy-labeled peptides as internal standards;

[0156] (3) Chromatography condition optimization: Use the same EASY-nLCTM 1200 system, pre-column, and analytical column configuration as in the preliminary experiment.

[0157] (4) Fine-tuning of mass spectrometry parameters: Keep the mass spectrometer and ion source settings the same as in the preliminary experiment. Focus on adjusting the mass spectrometry method to full scan + PRM scan, and update the peptide information in the inclusion list based on the results of the preliminary experiment;

[0158] (5) Data acquisition: Perform full scan and PRM scan to ensure that parameters such as resolution, C-trap capacity, and injection time are consistent with the preset values, and keep the peptide fragmentation conditions unchanged to generate accurate raw mass spectrometry data;

[0159] (6) Data calibration and analysis: Finally, use Skyline software to deeply analyze the off-line data, and accurately calibrate the peak area with the internal standard peptides to ensure the accuracy of the quantitative results.

[0160] 3 PRM data analysis

[0161] (1) Relative quantification of target peptides in samples

[0162] Use Skyline to extract the peptide chromatographic peaks of the target peptides from the raw data (.raw) of the PRM formal experiment. Select 3 sub-ions with higher peptide abundances for quantitative analysis, and export the peak area results of each target peptide after Skyline analysis; including the target protein name, target peptide sequence, parent ion charge number, selected sub-ions, sub-ion charge numbers, and the original peak area of each sub-ion used for quantification.

[0163] Integrate the sub-ion peak areas of the peptide segments in the target protein to obtain the original peak areas of the peptide segments in each sample. Use the peak areas of the internal standard peptide segments labeled with heavy isotopes to correct the relative expression levels of each peptide segment in the sample. The correction value is the ratio between the original peak area of the peptide segment and the peak area of the heavy-labeled peptide segment incorporated into the sample. Finally, obtain the relative quantification results of the target peptide segments in the sample.

[0164] 2) Relative expression level of the target protein in the sample

[0165] Based on the relative expression levels of the corresponding peptide segments of each target protein, further calculate the expression level of the target protein in the sample.

[0166] 4MethylTarget targeted methylation sequencing

[0167] The MethylTarget targeted methylation sequencing of the present invention is provided by Shanghai Tianhao Biotechnology Co., Ltd.

[0168] Specific experimental steps:

[0169] 1) DNA sample quality control

[0170] Use NanoDrop 2000 to detect the concentration of the DNA sample. The OD260 / 280 is between 1.7 and 1.9 to ensure that the DNA is intact and free of contamination. In addition, it is required that the concentration of the DNA ≥ 20 ng / μL and its total amount ≥ 400 ng, so that the total amount of DNA can meet the requirements of downstream experiments. Finally, detect by agarose electrophoresis that the main band is clearly visible.

[0171] 2) Primer design and PCR condition optimization

[0172] Use Methylation Fast Target V4.1 software to design primers for methylation site sequencing. The primer design includes Illumina adapter and specific amplification sequences. For the success of downstream experiments, the present invention provides DNA extracted by mixing one paraffin tissue sample of each different PT subtype as a standard product, and then perform bisulfite treatment on it. Subsequently, the product is used to verify the primers to judge the accuracy and uniqueness of the product size amplified by this primer.

[0173] 3) Optimization process of the multiplex PCR primer model

[0174] The primers optimized in 2) are used as the standard multiplex PCR panel, and unique multiplex PCR technology is used to amplify with the bisulfite-treated standard product as the template. After repeated debugging, the balance between the primer amplification efficiency and specificity is achieved, and the composition and concentration of the primers in the model are determined. Use primer3 to design primers for the sequence after Bisulfite treatment, softwarehttp: / / primer3.ut.ee / 。

[0175] 4) Bisulfite treatment process

[0176] Treat the DNA sample in accordance with the guidelines of the 《EZ DNAMethylation-Gold TM Kit》 to accurately convert originally unmethylated cytosine into uracil.

[0177] 5) Multiplex PCR reaction of the sample target fragment

[0178] (1) Amplification step: Use the optimized primer model to amplify the target region, and verify the amplification effectiveness by agarose gel electrophoresis.

[0179] (2) Product mixing and dilution: For different models of the same sample, quantitatively mix and dilute the PCR products as the Index PCR template for adding specific tag sequences. The reaction system is as shown in Table 1 below and the reaction conditions are as shown in Table 2 below.

[0180] Table 1 Multiplex PCR reaction system

[0181] Component Volume (μL) 10× Buffer 2 dNTP (2.5 mM) 24. <![CDATA[MgCl 2 (25mM)]]> 12. Bisulfite-treated Sample DNA 1 HotTaq (5 U / μL) 03. Multiplex PCR Panel Primer (1 μM) 2 Deionized Water 111. Total Volume 20

[0182] Table 2 Multiplex PCR reaction conditions

[0183]

[0184]

[0185] 6) Adding specific tag sequences to the sample

[0186] Configure the reaction system as shown in Table 3 below in a 96-well plate, add primers containing the Index sequence and tag sequences matching the Illumina platform, and perform amplification according to the reaction conditions shown in Table 4 below. Finally, mix the Index PCR products proportionally according to the amplification efficiency.

[0187] Table 3 Reaction system

[0188]

[0189] Table 4 Reaction system

[0190] Step Denaturation Annealing Extension Hold Number of Cycles Step1 95 °C, 2 min 1× Step2 95℃，20s 60℃，30s 72℃，30s 11× Step3 72 °C, 3 min 1× Step4 4℃ Hold

[0191] 7) Sample mixing, gel cutting and recovery

[0192] Prepare 2% agarose gel, take the product in 6) for electrophoresis (conditions: 120V, 35 min), cut the target band with reference to DNA Marker B, and finally use the TIANGEN Gel Extraction kit for gel recovery.

[0193] 8) Library Quantification and Sequencing Preparation

[0194] First, use the Agilent 2100 Bioanalyzer to verify the library fragment length to ensure the library quality. Subsequently, use the Illumina Hiseq or Nova seq platform to perform paired-end sequencing (2×150 bp) to obtain the original FastQ data.

[0195] 5MetlyTarget Data Analysis

[0196] First, use the software FLASH (FLASH: Fast length adjustment of short reads to improve genome assemblies.) to splice the filtered R1 and R2 reads. And use FastX ( http: / / hannonlab.cshl.edu / fastx_toolkit / index.html ) to process the fastq file obtained by splicing to get the fa format sequence. Then use blast+ (Camacho C, (2009) “BLAST+: architecture and applications”) to align all the reads in the fa with the reference sequence of the target region. Select the reads that can cover 90% of its target sequence or 90% of the bases can completely cover its target sequence, record them as valid reads, and count them. Calculate the efficiency of base C converted to T after bisulfite treatment in the effective sequencing data of each sample. Finally, calculate the methylation level of CpG sites in the amplicon.

[0197] 6PRM Protein and MetlyTarget Methylation Data Integration Analysis

[0198] Integrate the data of 3 proteins and 7 methylation sites of 74 patients in XJTU cohort 2, and then use the randomForest function in the randomForest R package to calculate the average Gini index. Generally, the larger this value, the higher the importance of this feature. Use the varImpPlot function to visualize the importance ranking of each feature. The results are as Figure 7(Figure showing the ranking results of the importance of 10 candidate markers by the random forest algorithm in XJTU cohort 2; among them, the mean decrease in Gini index for each feature is calculated based on the reduction in Gini impurity of the feature in the entire model. The larger this value, the higher the importance of the feature), the present invention uses the random forest algorithm to rank the diagnostic importance of 10 candidate markers in XJTU cohort 2.

[0199] Constructing a phyllodes tumor diagnosis model using three machine learning algorithms

[0200] The top 10 ranked markers were sequentially incorporated into three machine learning models. To ensure the reliability of the finally selected diagnostic markers, the present invention uses three machine learning models, namely Naive Bayes (NB), Support Vector Machine (SVM), and Multinomial Logistic Regression (MLR). The present invention uses the Naive Bayes function and svm function in the e1071 R package to construct the NB and SVM models, and uses the multinom function in the nnet R package to construct the MLR model; the multiclass.roc function is used to calculate the AUC of the model. The results show ( Figure 8 , Figure showing the cumulative AUC values of three machine learning models in XJTU cohort 2; among them, the abscissa represents the number of features when the top 10 ranked markers are sequentially incorporated into three machine learning models, such as 5 representing the incorporation of the top 5 ranked markers, the ordinate represents the cumulative AUC value, and the data table is the cumulative AUC value corresponding to each machine learning algorithm when incorporating features in sequence; the red box indicates that when the first 7 markers are incorporated, the AUCs of the three machine learning algorithms all reach the highest point; NB, Naive Bayes; SVM, Support Vector Machine; MLR, Multinomial Logistic Regression), when the model incorporates the first 7 features - including 2 proteins (A0A024R9E4_MAL2 and P05787_KRT8) and 5 methylation sites (cg00515756_DAGLA, cg24852561_CLYBL, cg00686880_RIPK4, cg22449901_PEX19, and cg09025210_ZBTB38), the AUC values of these three machine learning models all reach the best, and the AUCs of NB, SVM, and MLR are 0.971, 0.975, and 1 respectively. In addition, this result was also verified in XJTU cohort 1 ( Figure 9, Cumulative AUC value result graph of three machine learning models in XJTU cohort 1; among them, the abscissa represents the number of features when the top 10 markers are sequentially included in the three machine learning models, and the ordinate represents the cumulative AUC value. The data table is the cumulative AUC value corresponding to each machine learning algorithm when features are sequentially included; the red box indicates that when the first 7 markers are included, the AUC of the three machine learning algorithms reaches the highest point; NB, Naive Bayes; SVM, Support Vector Machine; MLR, Multivariate Logistic Regression), the AUCs of NB, SVM, and MLR are 0.944, 0.950, and 1 respectively. However, other combinations also have excellent diagnostic efficacy. For example, when 10 markers are used for combined diagnosis, the AUCs of the three machine learning algorithms are 0.945, 0.947, and 0.956 (XJTU cohort 2) and 0.861, 0.917, and 1 (XJTU cohort 1) respectively; when 9 markers are used for combined diagnosis, the AUCs of the three machine learning algorithms are 0.967, 0.969, and 0.956 (XJTU cohort 2) and 0.908, 0.917, and 1 (XJTU cohort 1) respectively; when 8 markers are used for combined diagnosis, the AUCs of the three machine learning algorithms are 0.967, 0.975, and 0.915 (XJTU cohort 2) and 0.861, 0.908, and 1 (XJTU cohort 1) respectively; when 6 markers are used for combined diagnosis, the AUCs of the three machine learning algorithms are 0.959, 0.927, and 0.852 (XJTU cohort 2) and 0.894, 0.878, and 0.991 (XJTU cohort 1) respectively; when 5 markers are used for combined diagnosis, the AUCs of the three machine learning algorithms are 0.949, 0.908, and 0.852 (XJTU cohort 2) and 0.885, 0.862, and 0.938 (XJTU cohort 1) respectively.In addition, a single biomarker also showed excellent performance in differential diagnosis. For example, cg24852561 and cg22449901 made outstanding contributions when applied alone in the differentiation between FA and PT or between FA and benign PT, with AUC, specificity, and sensitivity all greater than 0.9 in both the discovery cohort and the validation cohort; cg00686880 made outstanding contributions when applied alone in the differentiation between malignant PT and Other FELs, with AUC, specificity, and sensitivity all greater than 0.7 in both the discovery cohort and the validation cohort; cg09025210 made outstanding contributions when applied alone in the differentiation between benign PT and borderline PT, with AUC, specificity, and sensitivity all greater than 0.8 in both the discovery cohort and the validation cohort; cg00515756, KRT8, and MAL2 made outstanding contributions when applied alone in the differentiation between malignant PT and borderline PT, and between malignant PT and Other FELs, with AUC, specificity, and sensitivity all greater than 0.8 in both the discovery cohort and the validation cohort.

[0201] The peptide sequence information of the 3 candidate diagnostic proteins and the primer information of the 7 methylation sites are shown in Tables 5 and 6.

[0202] Table 5 List of mass-to-charge ratios (m / z) and peptides for PRM targeted sequencing of 3 diagnostic proteins

[0203] Protein Name Target Peptide Sequence (SEQ ID NO.) Mass-to-Charge Ratio (m / z) P05787 (KRT8) ISSSSFSR(1) 435.719424 A0A024R9E4 (MAL2) VTLPAGPDILR(2) 576.34278 P15924 (DSP) AELIVQPELK(3) 570.337164

[0204] Table 6 Primer information for MethlyTarget targeted methylation sequencing of 7 diagnostic methylation sites

[0205] CpGs PrimerF (SEQ ID NO.) PrimerR (SEQ ID NO.) cg00515756 TGGTTYGTTGTTGTGGTTAGGAG(4) TAACACCACCTAAAACTCAACCAC(5) cg00686880 GTGGGTTTTTGGGTTTTGTG(6) CCCACTACCAAACCCRAACC(7) cg09025210 GGAAGATTTAGTYGGTYGAGAGTTTTAGTT(8) CCAAAATCACAAATACCCACAAAA(9) cg22449901 GYGAGAGTAAAGAGTGGTTAATGAATG(10) TTCCTTTTCTTCCTAACACTAAACAA(11) cg24852561 AATTGAGAAATGTTTTGTAGTGTGAGAAG(12) CAACACAATAAACCAAATATCTAATACTCAA(13) cg00819233 GGGAGAGTTTAGTTTTAGGTTTTTATG(14) CCTACCCCACACCTCATTTTT(15) cg04404381 GTTTAAAGGGATTTTTGAGATTATAAATAGGA(16) TCCTACTTACTAAACRCCCCRAAAC(17)

[0206] Furthermore, the confusion matrix function was further used to output diagnostic performance evaluation indicators such as the confusion matrix, accuracy, consistency, specificity, sensitivity, positive predictive value, and negative predictive value of the 3 machine learning diagnostic models. As can be seen more intuitively from Table 7, the confusion matrices of the 3 machine learning models constructed from 7 biomarkers in the two cohorts, as well as the diagnostic performance evaluation indicators such as the specificity, sensitivity, accuracy, and consistency of the models, were all excellent. In summary, the 7-molecule diagnostic model integrating proteins and methylomics can achieve differential diagnosis between FA and PT and different grades of PT in multiple machine learning models.

[0207] Table 7 Evaluation indicators of 3 machine learning models based on 7 biomarkers in two cohorts

[0208]

[0209] Note: NB, Naive Bayes; SVM, Support Vector Machine; MLR, Multinomial Logistic Regression; AUC, Area Under Curve; Accuracy, Accuracy; Kappa, Kappa; PPV, Positive Predictive Value; NPV, Negative Predictive Value. Generally, if the evaluation index is greater than 0.9 or 90%, it is considered that the diagnostic efficiency of the model is good.

[0210] Diagnostic contributions of 8 diagnostic markers to different classification outcomes

[0211] The present invention further clarifies the diagnostic contributions of 7 markers to different classification outcomes and their expression trends in different grades of PT. As Figure 10 (Radar chart of the diagnostic efficiency (AUC, specificity, and sensitivity) of 7 markers in differentiating different classification outcomes; among them, extending from the center point 0 to the outermost circle, the representative value ranges from 0 to 1; different colors represent different diagnostic indicators) shown, among these 7 diagnostic markers, the 3 methylation sites cg24852561, cg00686880, and cg22449901 are mainly used to distinguish FA and PT; cg09025210 is mainly used to distinguish between benign PT and borderline PT; MAL2, KRT8, cg00686880, and cg00515756 can effectively identify malignant PT from FELs. The above results indicate that the 7 diagnostic markers have their own advantages when differentiating different classification outcomes alone. A single marker can achieve differentiation, but it often cannot have both high specificity and sensitivity, resulting in a relatively high rate of missed diagnosis or misdiagnosis of diseases. Therefore, to avoid this shortcoming, a combined diagnostic method using multiple markers is often used clinically to achieve the best sensitivity, specificity, and accuracy. In the present invention, markers that simultaneously meet the specificity or sensitivity ≥ 0.7 in two cohorts are preferably used as the diagnostic panel for certain two tumor subtypes, and the specific combination methods are shown in Table 8.

[0212] Table 8 Diagnostic combinations of 7 diagnostic markers in differentiating different classification outcomes

[0213]

[0214] Note: Taking the diagnostic combination of FA vs PT as an example, for the differential diagnosis of FA and PT, any one or any combination of the markers in the following 5 rows can be selected to achieve. Understand the differential diagnosis of the following four major rows in the same way; "-" indicates that the value is less than 0.7. Other refers to the other three subtypes of FELs, including FA, benign PT, and borderline PT.

[0215] Correlation between diagnostic markers and histological grade of PT

[0216] The t-test was used to compare the statistical differences between the two groups, and the ggboxplot function in the ggplot2 R package was used to visualize the box plots of protein expression or methylation levels between groups. The results showed that the protein expression levels of MAL2 and KRT8 decreased with the increase of PT histological grade ( Figure 11 , the results of the differential expression of the two diagnostic proteins in different grades of PT; among them, (a) the box plot of the expression of the two diagnostic proteins in XJTUcohort 1; (b) the box plot of the expression of the two diagnostic proteins in XJTU cohort2; nsP>0.05, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001); the DNA methylation levels of the three loci cg00515756, cg00686880, and cg09025210 increased with the increase of PT histological grade ( Figure 12 , the results of the differential methylation levels of the five diagnostic methylation sites in different grades of PT; among them, (a) the box plot of the methylation levels of the five diagnostic methylation sites in XJTU cohort 1; (b) the box plot of the methylation levels of the five diagnostic methylation sites in XJTU cohort2; ns P>0.05, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001); however, cg24852561 and cg22449901 showed the opposite trend ( Figure 12 ). The above results indicate that the diagnostic marker protein or methylation level is closely related to the PT histological grade. Combining the diagnostic contributions of the diagnostic markers to different classification outcomes ( Figure 10 ) and their correlations with the PT histological grade ( Figure 11 and Figure 12 ), the present invention drew a schematic decision tree for the future clinical application of the 7-molecule diagnostic panel ( Figure 13 , the AUC values marked here are from XJTU Cohort2, and the red label "hyper" represents that this locus is hypermethylated in FA compared with PT; the green label "low" represents that this protein is lowly expressed in malignant PT compared with other FELs subtypes).

[0217] 10 Comparison between the multi-omics diagnostic model of the present invention and the traditional model

[0218] Previously reported single-omics traditional models focused on differentiating fibroadenoma (FA) from phyllodes tumor (PT), and benign PT from borderline / malignant PT. However, the diagnostic model of the present invention can effectively distinguish FA, benign PT, borderline PT, and malignant PT. To compare with the previously reported single-omics binary classification models, the present invention used a binary logistic regression model to evaluate the diagnostic performance of the 7-molecule diagnostic model of the present invention in differentiating two outcomes: FA and PT, and benign PT and borderline / malignant PT. The results showed that this model could well distinguish FA and PT, as well as benign PT and borderline / malignant PT in both XJTU cohort1 and XJTU cohort2 (binary logistic regression: AUC = 1, sensitivity, specificity, positive predictive value, and negative predictive value were all 100%) (Table 9).

[0219] As Figure 14 (Results comparison chart of multi-omics diagnostic model and traditional model; among them, (a) Comparison chart of the diagnostic model of the present invention and the previously reported diagnostic model for differentiating FA and PT; (b) Comparison chart of the diagnostic model of the present invention and the previously reported diagnostic model for differentiating benign PT and borderline / malignant PT. This figure is a lollipop chart, the abscissa is the AUC value of the model, and the ordinate represents the diagnostic models from different studies. When different methods are used to construct models in the same study, each model constructed by each method is regarded as an independent model and compared with the research model of the present invention; in addition, when multiple indicators in the same study are used to construct diagnostic models separately, each model constructed by each indicator is regarded as an independent model and compared with the research model of the present invention) shows that the multi-omics diagnostic model of the present invention is superior to the reported models, including the somatic mutation model (TERT promoter / MED12 / FLNA / SETD2), the radiomics model based on magnetic resonance imaging (MRI), the whole-slide digital section (WSI) model of pathological HE, and the ultrasound (US) texture feature model (Table 9). In summary, the 7-feature multi-omics diagnostic model of the present invention performs excellently in differentiating FA, benign PT, borderline PT, and malignant PT, and is superior to various previously reported diagnostic models.

[0220] Table 9 Diagnostic models of breast fibroepithelial tumors reported previously

[0221]

[0222]

[0223] Note: NA indicates not reported in the literature; MRI, magnetic resonance imaging; PR, progesterone receptor; WSI, whole slide imaging; SVM, support vector machine; XGB, extreme gradient boosting; RF, random forest; CNN, convolutional neural network; RNN, recurrent neural network; Ridge-RFE, ridge regression combined with recursive feature elimination; AUC, area under the curve; PPV, positive predictive value; NPV, negative predictive value; FA, fibroadenoma; BPT, benign phyllodes tumor; BorPT, borderline phyllodes tumor; MPT, malignant phyllodes tumor; PT, phyllodes tumor.

[0224] Although the above embodiments have described the present invention in detail, they are only a part of the embodiments of the present invention, rather than all embodiments. People can also obtain other embodiments based on these embodiments without creative efforts, and these embodiments all fall within the protection scope of the present invention.

Claims

1. A marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, wherein the phyllodes tumors of the breast of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast and malignant phyllodes tumors of the breast, characterized in that: The marker composition includes any one or more of the following markers: MAL2, KRT8, DSP, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381; wherein MAL2, KRT8 and DSP are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381 ​​are DNA methylation biomarkers in the human genome.

2. A marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, wherein the phyllodes tumors of the breast of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast and malignant phyllodes tumors of the breast, characterized in that: The marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381; wherein, MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210, cg00819233 and cg04404381 ​​are DNA methylation biomarkers in the human genome.

3. A marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, wherein the phyllodes tumors of the breast of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast and malignant phyllodes tumors of the breast, characterized in that: The marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901, cg09025210 and cg04404381; wherein MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901, cg09025210 and cg04404381 ​​are DNA methylation biomarkers in the human genome.

4. A marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, wherein the phyllodes tumors of the breast of different grades include benign phyllodes tumors of the breast, borderline phyllodes tumors of the breast and malignant phyllodes tumors of the breast, characterized in that: The marker composition includes any one or more of the following markers: MAL2, KRT8, cg00515756, cg24852561, cg00686880, cg22449901 and cg09025210; wherein MAL2 and KRT8 are protein markers, and cg00515756, cg24852561, cg00686880, cg22449901 and cg09025210 are DNA methylation biomarkers in the human genome.

5. Use of a reagent for detecting the marker composition according to any one of claims 1 to 4 in the preparation of a kit for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades.

6. A kit for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, characterized in that: The kit comprises a reagent for detecting the marker composition according to any one of claims 1 to 4; the reagent for detecting the marker composition according to any one of claims 1 to 4 comprises a reagent for detecting a protein marker and a reagent for detecting a methylation biomarker; The reagents for detecting protein markers include reagents used in any of the following methods: PRM targeted protein sequencing method, selected reaction monitoring targeted sequencing method, multiple reaction monitoring targeted sequencing method, tandem mass spectrometry label quantitative protein sequencing method, isotope label quantitative protein sequencing method, western blotting, enzyme-linked immunosorbent assay and immunohistochemical staining; The reagents for detecting DNA methylation biomarkers include reagents used in any of the following methods: MethylTarget targeted methylation method, methylated DNA immunoprecipitation sequencing method, degenerate representative bisulfite sequencing method, whole genome bisulfite sequencing method, deep target region DNA methylation sequencing method and pyrophosphate sequencing method.

7. A method for constructing a diagnostic model for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, characterized in that: The following steps are involved: Using the marker composition according to any one of claims 1 to 4 to construct a diagnostic model for distinguishing breast phyllodes tumors from fibroadenomas and / or distinguishing breast phyllodes tumors of different grades; The types of the diagnostic models include any one or more of the following: naive Bayes model, support vector machine model, multivariate logistic regression model, decision tree model, random forest model, K nearest neighbor algorithm model, neural network model, adaptive boosting algorithm model, gradient boosting machine model, extreme gradient boosting algorithm model, least absolute shrinkage and selection operator model, elastic network regression model and ridge regression model.

8. A diagnostic model device for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, characterized in that: The device comprises: a memory and a processor; the memory is used to store program instructions; the processor is used to call the program instructions, and when the program instructions are executed, is used to perform the following operations: Acquiring detection data of the marker composition of the sample to be tested, inputting the detection data into a diagnostic model constructed by the marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, and obtaining a diagnostic result of the sample to be tested; The marker composition is the marker composition according to any one of claims 1 to 4.

9. A diagnostic system for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, comprising: An acquisition unit, used for acquiring detection data of the marker composition of the sample to be tested; A processing unit inputs the detection data of the marker composition into a diagnostic model constructed by the marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, to obtain a diagnostic result of the sample to be tested; The marker composition is the marker composition according to any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following method is implemented: Acquiring detection data of the marker composition of the sample to be tested, inputting the detection data of the marker composition into a diagnostic model constructed by the marker composition for distinguishing phyllodes tumors of the breast from fibroadenomas and / or distinguishing phyllodes tumors of the breast of different grades, and obtaining a diagnostic result of the sample to be tested; The marker composition is the marker composition according to any one of claims 1 to 4.

Images