Immune-related gene metastatic breast cancer patient prognosis model and construction method and application thereof

Abstract

The invention relates to the technical field of prognosis research of metastatic breast cancer. The invention provides a metastatic breast cancer patient prognosis model of an immune-related gene and a construction method and application thereof, the prognosis model is based on differential expression between breast cancer primary lesion and metastatic lesion in GSE102818 and GSE43837, and through clinical information and gene expression of a metastatic breast cancer patient of GSE124648, the prognosis model of the metastatic breast cancer patient of the immune-related gene is established. The TCGA-BRCA data set is divided into an internal training set and a verification set for single-factor and multi-factor Cox analysis, genes related to prognosis are selected, a prognosis model is constructed on the basis of the genes, and the TCGA-BRCA data set serves as an external verification set of the model to detect the accuracy of the model. Meanwhile, by combining the risk score, the progestational hormone receptor state and the visceral organ metastasis state, a column diagram is developed, and the C index is calculated. Products related to diagnosis, treatment and prognosis of metastatic breast cancer can be further developed according to the core gene in the prognosis model.

Description

Technical Field

[0001] The present invention relates to the technical field of prognosis research of metastatic breast cancer, and in particular to a prognosis model of metastatic breast cancer patients of immune-related genes, a construction method and application thereof. Background Art

[0002] Breast cancer is the most common malignancy in women worldwide. In 2020, 2.26 million new cases of breast cancer were reported worldwide, accounting for 11.7% of all new cancer cases, surpassing lung cancer to become the most common malignancy. Approximately 680,000 people died from breast cancer, accounting for 6.9% of cancer-related deaths, making breast cancer the fifth most common cause of cancer death. Although early detection and effective treatment have improved, distant metastasis is still considered the main cause of death in breast cancer patients. MBC is a heterogeneous disease with different prognoses affected by multiple clinical pathological characteristics (such as metastatic sites), resulting in a 5-year overall survival rate of only 25%.

[0003] In recent years, some achievements have been made in the treatment of MBC. The prognosis and survival of MBC patients have improved due to the introduction of chemotherapy, radiotherapy, targeted therapy, and immunotherapy. However, there are considerable differences in survival benefits among individuals with different MBC diseases. Accurate survival time estimation may greatly benefit patients in individualized treatment decisions. Therefore, an effective algorithm that can correctly estimate the prognosis of MBC patients is urgently needed.

[0004] The tumor immune microenvironment is crucial in tumor metastasis. The seed and soil theory proposes that tumor spread results from the interaction between tumor cells and their surroundings. Recent studies have shown that cancer-associated fibroblasts can stimulate tumor cell migration through the NOTCH pathway, as well as secrete CXCL12 and TGF-β. The negative immunoregulatory effects of myeloid-derived suppressor cells may contribute to tumor initiation and progression. Metastatic breast cancer cells can escape immune surveillance through various mechanisms, resulting in an inert immune environment. Compared with primary tumors, BC metastases have a tumor immune microenvironment that is more suitable for breast cancer cell growth. However, there are few studies on the role of immune-related genes that are temporally differentially expressed between metastatic and primary breast cancer lesions and the DEIRGs-based prognostic model for MBC patients. Summary of the invention

[0005] The purpose of the present invention is to provide a prognostic model of metastatic breast cancer patients of immune-related genes and a construction method and application thereof, so as to solve the shortcomings of the prior art of poor prognostic prediction ability and inaccurate results of neutron metastatic breast cancer.

[0006] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0007] The present invention provides a method for constructing a prognostic model of immune-related genes for patients with metastatic breast cancer, comprising the following steps:

[0008] (1) Identify immune-related genes that are differentially expressed between primary and metastatic breast cancer lesions in the public databases GSE102818 and GSE43837, and obtain differentially expressed genes by taking the intersection of the two databases;

[0009] (2) In GSE124648, patients with metastatic breast cancer were divided into an internal training set and a validation set, and a prognostic model was constructed in the internal training set using univariate and multivariate Cox regression analysis;

[0010] (3) Screen out patients with metastatic breast cancer in the TCGA-BRCA dataset and apply the prognostic model constructed in step (2) to the dataset to verify the accuracy of the model.

[0011] Preferably, four data sets are used in the construction, including three microarray data sets and one RNA-seq data set, the three microarray data sets are GSE102818, GSE43837 and GSE124648; the RNA-seq data set is the TCGA-BRCA cohort.

[0012] Preferably, the specific method of constructing in step (2) comprises the following steps:

[0013] A. Preprocessing of the dataset GSE124648: GSE124648 contains clinical data of 140 patients with stage IV BC. The patients were randomly divided into an internal training set of 98 cases and a validation set of 42 cases using R (version 4.1.2);

[0014] B. Univariate Cox regression analysis: Univariate Cox regression analysis was used on the internal training set to find potential prognostic genes;

[0015] C. Multivariate Cox regression analysis: The potential prognostic genes found were intersected with the PPI network genes and included in the multivariate Cox regression analysis to obtain the regression coefficient of each factor related to prognosis;

[0016] D. Model prediction: The model is used to predict the prognosis of patients with multinuclear leukemia in the training set. The risk value is βExp1+βExp2+…+βExpN, where EXP is the gene expression level, β is the regression coefficient of COX regression analysis, and N is the number of genes related to the patient's prognosis.

[0017] Preferably, step (2) further comprises the following steps:

[0018] a. Calculate the risk score: Use the formula to calculate the risk score for each patient in the training set;

[0019] b. Grouping patients: Patients in the training group were divided into high-risk and low-risk groups according to the cutoff values ​​determined by the 'Maxstat' package in R (version 4.1.2);

[0020] c. Kaplan-Meier survival curves: Kaplan-Meier survival curves of high-risk and low-risk groups were evaluated using the R package ‘Survival’ (version 4.1.2);

[0021] d. ROC curve evaluation: The R'time ROC' package (version 4.1.2) was used to generate the ROC curve to evaluate the predictive effect of IPM.

[0022] Preferably, the verification in step (3) includes the following steps:

[0023] ① BC patients with distant metastasis were selected from the TCGA-BRCA cohort as an external validation set to evaluate the prognostic value of the model;

[0024] ②Univariate and multivariate COX regression analysis was performed on the clinical data of MBC patients to test whether the risk score was independent of other clinicopathological variables and to establish a nomogram;

[0025] ③The accuracy and adaptability of the model were analyzed through C index analysis, 1-year, 3-year, and 5-year calibration analysis, and decision curve analysis (DCA).

[0026] The present invention also provides a prognostic model of metastatic breast cancer patients of immune-related genes obtained by the construction method.

[0027] The present invention also provides the use of the prognostic model of metastatic breast cancer patients of the immune-related genes in screening markers for assessing the prognostic risk of metastatic breast cancer.

[0028] The present invention also provides a marker for evaluating the prognostic risk of metastatic breast cancer screened by using the prognostic model of metastatic breast cancer patients using the immune-related genes, wherein the markers are gene CD28 and gene CFB.

[0029] The present invention also provides a reagent set for detecting the marker, the reagent set comprising a liquid reagent or a carrier reagent;

[0030] The liquid reagent comprises a buffer, an antibody or a nucleic acid probe, a marker, and a control substance;

[0031] The carrier reagent comprises a carrier, an antibody or a nucleic acid probe fixed on the carrier, a buffer solution, and a washing solution.

[0032] The present invention also provides a kit for evaluating the prognostic risk of metastatic breast cancer, wherein the kit comprises the reagent set.

[0033] The present invention provides a prognostic model for patients with metastatic breast cancer of immune-related genes and a construction method and application thereof. The prognostic model of the present invention is based on the immune-related genes differentially expressed between primary lesions and metastatic lesions of breast cancer in GSE102818 and GSE43837. The clinical information and gene expression of patients with metastatic breast cancer in GSE124648 are divided into an internal training set and a validation set at 7:3 for univariate and multivariate Cox analysis, and genes related to prognosis are selected and a prognostic model is constructed based on this. The TCGA-BRCA data set is used as an external validation set of the model to detect the accuracy of the model. At the same time, by combining the risk score, progesterone receptor status and organ metastasis status, a nomogram is developed, and the calculated C index is 0.674. This invention can further develop products related to the diagnosis and treatment of sub-metastatic breast cancer and prognosis based on the core genes in the prognostic model.

[0034] The present invention uses the transcriptome data of metastatic breast cancer in the public databases GEO and TCGA, screens differentially expressed genes layer by layer, constructs a prognostic model of metastatic breast cancer of immune-related genes, and verifies that it has good predictive ability for patients with metastatic breast cancer. The prognostic model of the present invention can be used to guide clinical individualized treatment strategies, and can further develop products related to the diagnosis and treatment of metastatic breast cancer and prognosis based on the core genes in the prognostic model. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 The following is a flow chart of the construction method in an embodiment of the present invention.

[0036] Figure 2 The heat map, volcano map and Venn diagram of immune-related differentially expressed genes in metastatic breast cancer obtained in the examples of the present invention; wherein: (a) is the volcano map of DEIRGs in GSE102818; (b) is the heat map of DEIRGs in GSE102818; (c) is the volcano map of DEIRGs in GSE43837; (d) is the heat map of DEIRGs in GSE102818; (e) is the Venn diagram of up-regulated genes in GSE102818 and GSE43837; (f) is the Venn diagram of down-regulated genes in GSE102818 and GSE43837.

[0037] Figure 3The present invention is a screening and confirmation of model genes in the embodiments, wherein (a) is a protein interaction analysis of immune-related genes; (b) is a univariate Cox regression analysis of immune-related genes; (c) is a Venn diagram constructed by protein interaction analysis and univariate Cox regression analysis; (d) is the confirmation of model genes after stepwise multivariate Cox regression analysis.

[0038] Figure 4 Figure 1 is the establishment and validation diagram of the two-gene prognostic model in the embodiment of the present invention, the risk score distribution of the two-gene model in the training set (a), validation set (b) and total set (c) in GSE124648; the heat map of the two model genes in the training set (d), validation set (e) and total set (f) in GSE124648; the Kaplan-Mayer curve of the two-gene model in the training set (g), validation set (h) and total set (I) in GSE124648; the time-dependent ROC analysis of the two-gene model for 1-year, 3-year and 5-year OS in the training set (j), validation set (k) and total set (l) of the GSE124648 data set.

[0039] Figure 5 Figure 1 is an external validation diagram of the dual-gene prognostic model in the embodiment of the present invention, (a) is the risk score distribution of the two genes in TCGA-BRCA and the heat map; (b) is the Kaplan-Mayer curve of the two genes in TCGA-BRCA; (c) is the time-dependent ROC analysis of the dual-gene model for 1-year, 3-year and 5-year OS in the TCGA-BRCA dataset.

[0040] Figure 6 The construction of the nomogram in the embodiment of the present invention, (a) is a nomogram combining the prognostic model, PR status and visceral metastasis constructed in the GSE124648 dataset; 1-year (b), 3-year (c) and 5-year (d) calibration analysis and DCA (e) of the nomogram. DETAILED DESCRIPTION

[0041] The present invention provides a method for constructing a prognostic model of immune-related genes for patients with metastatic breast cancer, comprising the following steps:

[0042] (1) Identify immune-related genes that are differentially expressed between primary and metastatic breast cancer lesions in the public databases GSE102818 and GSE43837, and obtain differentially expressed genes by taking the intersection of the two databases;

[0043] (2) In GSE124648, patients with metastatic breast cancer were divided into an internal training set and a validation set, and a prognostic model was constructed in the internal training set using univariate and multivariate Cox regression analysis;

[0044] (3) Screen out patients with metastatic breast cancer in the TCGA-BRCA dataset and apply the prognostic model constructed in step (2) to the dataset to verify the accuracy of the model.

[0045] In the present invention, four data sets were used in the construction, including three microarray data sets and one RNA-seq data set. The three microarray data sets were GSE102818, GSE43837 and GSE124648; the RNA-seq data set was the TCGA-BRCA cohort.

[0046] In the present invention, the database is the Cancer Genome Atlas (TC GA-BRCA) dataset (https: / / portal.gdc.cancer.gov / ) and the GSE102818, GSE43837 and GSE124648 datasets in the GEO database, and the acquired gene data includes the gene expression and expression information. Relevant clinical information is also obtained. The criteria for sample inclusion are: female patients with stage IV metastatic breast cancer, with follow-up time and complete clinical data.

[0047] In the present invention, the metastatic breast cancer patient information in step (2) uses tumor tissue reported in a public database.

[0048] In the present invention, 730 immune-related genes were obtained from the database GSE10281.

[0049] In the present invention, the screening method is: download the expression data of 730 immune-related genes in GSE102818 (including 31 primary breast cancer samples and 17 metastatic breast cancer samples). At the same time, we cross-compared the differentially expressed genes in GSE43837 (including 19 primary breast cancer and 19 metastatic breast cancer samples) with the former to determine the differentially expressed immune-related genes. The 'LIMMA' package (version 4.1.2) of R software was used to screen candidate genes. Corrected P<0.05 and Fold change<1.5 were considered statistically significant.

[0050] In the present invention, fold change and corrected p-value were used to draw volcano plots, and the expression heat map was displayed using the R software package pheatmap.

[0051] In the present invention, the specific method of constructing in step (2) comprises the following steps:

[0052] A. Preprocessing of the dataset GSE124648: GSE124648 contains clinical data of 140 patients with stage IV BC. The patients were randomly divided into an internal training set of 98 cases and a validation set of 42 cases using R (version 4.1.2);

[0053] B. Univariate Cox regression analysis: Univariate Cox regression analysis was used on the internal training set to find potential prognostic genes;

[0054] C. Multivariate Cox regression analysis: The potential prognostic genes found were intersected with the PPI (protein-protein interaction) network genes and included in the multivariate Cox regression analysis to obtain the regression coefficient of each factor related to prognosis;

[0055] D. Model prediction: The model is used to predict the prognosis of patients with multinuclear leukemia in the training set. The risk value is βExp1+βExp2+…+βExpN, where EXP is the gene expression level, β is the regression coefficient of COX regression analysis, and N is the number of genes related to the patient's prognosis.

[0056] In the present invention, step A divides the GSE124648 dataset into a training set and a validation set, providing a data basis for subsequent model construction, and ensures that the samples of the training set and the validation set have statistically similar feature distributions through random grouping, which helps the generalization ability of the model and the prediction accuracy of the evaluation model. Step A provides a data set for model training and validation, which is the basis for subsequent steps.

[0057] In the present invention, the step B can screen out potential genes related to the patient's prognosis. The Cox regression analysis used in this step is a common survival analysis method for evaluating the impact of multiple variables on survival time. Through univariate Cox regression analysis, genes with statistically significant prognostic significance can be preliminarily screened out. The potential prognostic genes screened out in step B will be used for subsequent multivariate analysis to further determine the impact of each gene on prognosis.

[0058] In the present invention, step C can determine the regression coefficient of each gene in the multivariate model and construct a prognostic model. By intersecting the gene with the PPI network gene, key genes related to prognosis can be screened out. Multivariate Cox regression analysis can further determine the independent prognostic value of these genes in the multifactor model. This step is based on the results of univariate Cox regression analysis to further refine the construction of the model.

[0059] In the present invention, the constructed model is used in step D to predict the prognosis of patients in the training set. This step can quantitatively evaluate the risk level of each patient by calculating the risk value, and then predict their prognosis. This step verifies the predictive ability of the model and provides a basis for the next step of verification.

[0060] In the present invention, step (2) also includes the following steps:

[0061] a. Calculate the risk score: Use the formula to calculate the risk score for each patient in the training set;

[0062] b. Grouping patients: Patients in the training group were divided into high-risk and low-risk groups according to the cutoff values ​​determined by the 'Maxstat' package in R (version 4.1.2);

[0063] c. Kaplan-Meier survival curves: Kaplan-Meier survival curves of high-risk and low-risk groups were evaluated using the R package ‘Survival’ (version 4.1.2);

[0064] d. ROC curve evaluation: The R'time ROC' package (version 4.1.2) was used to generate the ROC curve to evaluate the predictive effect of IPM.

[0065] In the present invention, step a can calculate a risk score for each patient and quantify its prognostic risk. Since the risk score is based on the results of multivariate Cox regression analysis, combined with the gene expression data of each patient, its prognostic risk can be effectively quantified. This step provides a data basis for subsequent patient grouping and survival analysis.

[0066] In the present invention, step b can divide patients into high-risk group and low-risk group according to risk score for survival analysis. Through the 'Maxstat' package, the optimal cutoff value can be determined to divide patients into high-risk group and low-risk group for survival analysis. This step provides grouping data for subsequent survival analysis based on risk score.

[0067] In the present invention, step c can evaluate the survival difference between the high-risk group and the low-risk group. The Kaplan-Meier survival curve is a commonly used survival analysis method for showing the survival probability and survival difference of two groups of patients at different time points. This step is based on the results of patient grouping to further verify the prediction effect of the model.

[0068] In the present invention, step d can evaluate the prediction effect of the model in the validation set. The ROC curve (receiver operating characteristic curve) is used to evaluate the sensitivity and specificity of the model at different thresholds, providing the overall prediction performance of the model. This step is based on the results predicted by the model, and finally evaluates the prediction accuracy of the model.

[0069] In the present invention, the verification in step (3) includes the following steps:

[0070] ① BC patients with distant metastasis were selected from the TCGA-BRCA cohort as an external validation set to evaluate the prognostic value of the model;

[0071] ②Univariate and multivariate COX regression analysis was performed on the clinical data of MBC patients to test whether the risk score was independent of other clinicopathological variables and to establish a nomogram;

[0072] ③The accuracy and adaptability of the model were analyzed through C index analysis, 1-year, 3-year, and 5-year calibration analysis, and decision curve analysis (DCA).

[0073] The present invention also provides a prognostic model of metastatic breast cancer patients of immune-related genes obtained by the construction method.

[0074] The present invention also provides the use of the prognostic model of metastatic breast cancer patients of the immune-related genes in screening markers for assessing the prognostic risk of metastatic breast cancer.

[0075] The present invention also provides a marker for evaluating the prognostic risk of metastatic breast cancer screened by using the prognostic model of metastatic breast cancer patients using the immune-related genes, wherein the markers are gene CD28 and gene CFB.

[0076] The present invention also provides a reagent set for detecting the marker, wherein the reagent set comprises a liquid reagent or a carrier reagent.

[0077] In the present invention, the liquid reagent comprises a buffer, an antibody or a nucleic acid probe, a marker (such as biotin, a fluorescent dye), and a reference substance;

[0078] In the present invention, the antibody can detect the target protein by ELISA (enzyme-linked immunosorbent assay) or Western Blot.

[0079] In the present invention, the carrier reagent comprises a carrier (such as a glass sheet, a silicon sheet), an antibody or a nucleic acid probe fixed on the carrier, a buffer solution, and a washing solution.

[0080] In the present invention, the carrier reagent may be a microarray chip, in which antibodies or nucleic acid probes are fixed on a solid carrier, and can be used for high-throughput detection of multiple markers.

[0081] In the present invention, the liquid reagent should be stored at 2-8°C, protected from light, and avoid repeated freezing and thawing.

[0082] The reagent fixed on the carrier should be stored in a dry, light-proof environment, avoiding high temperature and humidity.

[0083] In the present invention, when the liquid reagent is used, according to the specific detection method (such as ELISA, Western Blot), the sample is reacted with the reagent, and the result is detected by color development, fluorescence or chemiluminescence.

[0084] When the reagent fixed on the carrier is used, the sample is spotted or loaded onto the carrier, and the signal is observed using a fluorescent scanner or a microscope through hybridization or immune reaction.

[0085] The present invention also provides a kit for evaluating the prognostic risk of metastatic breast cancer, wherein the kit comprises the reagent set.

[0086] In the present invention, the kit may contain detection reagents for multiple markers.

[0087] In the present invention, the kit contains reference substances, including positive control and negative control, which are used to verify the accuracy of the detection method.

[0088] In the present invention, the kit comprises a buffer solution and a washing solution, and the buffer solution and the washing solution are used for dilution, reaction and washing of reagents.

[0089] In the present invention, the kit comprises a standard curve, comprising a standard substance of known concentration, for calibrating the detection results.

[0090] In the present invention, the kit includes an operating instruction manual, which details the method of use, precautions and result interpretation of the kit.

[0091] In the present invention, the kit comprehensively evaluates the prognostic risk of patients with metastatic breast cancer by detecting multiple prognostic markers, and provides quantitative or semi-quantitative test results to help clinicians develop personalized treatment plans.

[0092] In the present invention, the method for verifying the test results is to verify the prediction accuracy of the kit by comparing with clinical pathological data.

[0093] In the present invention, the validation data of the test results are cross-validation and external validation data, which prove the stability and reliability of the kit in different data sets and clinical environments.

[0094] In the present invention, the validation indicators include sensitivity, specificity, AUC (area under the ROC curve), etc., to evaluate the predictive performance of the kit.

[0095] The technical solutions provided by the present invention are described in detail below in conjunction with the embodiments, but they should not be construed as limiting the protection scope of the present invention.

[0096] Example 1

[0097] 1. Data collection and preprocessing: Obtain sample matrix data and corresponding clinical information.

[0098] In this step, the matrix data includes 4 breast cancer datasets, including 3 microarray datasets and 1 RNA-seq dataset.

[0099] The three microarray datasets include GSE102818, GSE43837, and GSE124648; the one RNA-seq dataset is the TCGA-BRCA cohort ( Figure 1 ). The specific ways to obtain it are as follows:

[0100] The data in the GSE102818 dataset were obtained from the GPL23249 platform. Specifically, the matrix data and corresponding clinical information of 31 primary breast cancer samples and 17 metastatic breast cancer samples in GSE102818 were downloaded from the GEO database (https: / / www.ncbi.nlm.nih.gov / geo / ).

[0101] The data in the GSE43837 dataset were obtained from the GPL1352 platform. Specifically, the matrix data and corresponding clinical information of 19 primary breast cancer samples and 19 brain metastasis breast cancer samples in GSE43837 were downloaded from the GEO database (https: / / www.ncbi.nlm.nih.gov / geo / ), and 140 metastatic breast cancer samples in GSE124648 were downloaded. The volcano map and heat map ( Figure 2 ad), we intersected the differentially expressed genes from the two databases through a Venn diagram and obtained 2 up-regulated genes and 59 down-regulated genes ( Figure 2 e,f).

[0102] The TCGA-BRCA cohort downloaded transcriptome profiles, mutation data, and clinical follow-up information of metastatic breast cancer samples from The Cancer Genome Atlas (TCGA) database (https: / / portal.gdc.cancer.gov). After the samples that did not contain complete RNA-seq expression profiles and clinical information in the TCGA-BRCA cohort were removed, the number of samples used for subsequent analysis was 21.

[0103] 2. Prognostic model construction: Based on the acquired data, univariate and multivariate Cox analysis was performed to establish a prognostic model.

[0104] In this step, 140 metastatic breast cancer samples in the GSE124648 data were randomly divided into 98 and 42 samples in a ratio of 7:3. 98 samples of metastatic breast cancer were used for model construction, and 42 samples were used for internal validation. Univariate Cox analysis was performed on the training set ( Figure 3a), and then intersected these genes with the PPI network genes ( Figure 3 b,c) and included in multivariate Cox regression analysis ( Figure 3 d), and obtain the regression coefficient of each factor related to prognosis. Therefore, we developed a model to predict the prognosis of patients with multinuclear leukemia in the training set, and the risk value is βExp1+βExp2+…+βExpN. EXP is the gene expression, β is the regression coefficient of COX regression analysis, and N is the number of genes related to patient prognosis.

[0105] 3. Use TCGA-BRCA as external validation.

[0106] The risk score of each sample was calculated according to the unified formula (i.e., the risk score formula in step S2), and the patients in the training group were divided into two groups: high-risk group and low-risk group according to the critical value determined by the 'Maxstat' package in R (version 4.1.2). The Kaplan-Meier survival curves of the high-risk and low-risk groups were evaluated using the R package 'Survival' (version 4.1.2). To evaluate the predictive effect of IPM, the ROC curve was generated by the R 'time ROC' package (version 4.1.2).

[0107] The receiver operating characteristic (ROC) curve is an intuitive method to analyze the accuracy of the classifier, and the area under the curve (AUC) value has always been the standard for judging the accuracy of the prognostic model.

[0108] The prognostic model was constructed by Lasso-Cox regression analysis, and the risk score calculation formula was (-0.3047975*CD28)+(-0.1628303*CFB). The optimal risk cutoff value of IPM was -3.906, based on which the 98 MBC patients in the GSE124648 internal training set were divided into high-risk group and low-risk group. The prognostic curve and scatter plot were used to represent each MBC patient ( Figure 4 a). Heat map of candidate IRGs expression profiles showing that CD28 and CFB were lowly expressed in high-risk groups ( Figure 4 d). Kaplan-Meier analysis confirmed that the OS of the high-risk group was significantly lower than that of the low-risk group ( Figure 4 g). Time-dependent ROC analysis showed that the AUCs for 1-year, 3-year, and 5-year OS were 0.73, 0.75, and 0.68, respectively ( Figure 4 j). We further validated the risk score, and the results of the GSE124648 internal validation set and the total set were consistent with those of the training set ( Figure 4 b, c, e, f). The OS of the high-risk group was significantly lower than that of the low-risk group ( Figure 4h, I). Time-dependent ROC analysis showed that the AUCs for 1-year, 3-year, and 5-year OS in the validation set and the total set were 0.61, 0.69, and 0.77, respectively ( Figure 4 k) and 0.71, 0.72, 0.71 ( Figure 4 l). Scatter plot of MBC in GSE124648 ( Figure 5 a) Survival curve ( Figure 5 b) and time dependence curves ( Figure 5 c) as an external validation set.

[0109] 4. Multivariate analysis was used to evaluate the independence of our prognostic features on the prognosis of metastatic breast cancer. A calibration curve was used to test the prediction efficiency of the nomogram. Univariate and multivariate COX regression analysis was performed on the clinical data of MBC patients to test whether the risk score was independent of other clinical pathological variables and to establish a nomogram. The accuracy and adaptability of the model were analyzed by C-index analysis, 1-year, 3-year, and 5-year calibration analysis, and decision curve analysis (DCA). Cox regression analysis showed that risk score, progesterone receptor (PR) status, and visceral metastasis status can be used as independent prognostic factors for MBC patients. These independent risk factors were integrated into a nomogram model ( Figure 6 a). The c index was 0.674 (95% confidence interval 0.642-0.706). The accuracy and clinical application value of the nomogram model were respectively determined by the calibration curve ( Figure 6 bd) and DCA( Figure 6 e) Detection.

[0110] As can be seen from the above embodiments, the present invention provides a prognostic model for patients with metastatic breast cancer of immune-related genes and its construction method and application. The prognostic model of the present invention is based on the immune-related genes differentially expressed between primary lesions and metastatic lesions of breast cancer in GSE102818 and GSE43837. Based on the clinical information and gene expression of patients with metastatic breast cancer in GSE124648, it is divided into an internal training set and a validation set at 7:3 for univariate and multivariate Cox analysis, and the genes related to prognosis are selected and the prognostic model is constructed. The TCGA-BRCA data set is used as the external validation set of the model to detect the accuracy of the model. At the same time, by combining the risk score, progesterone receptor status and organ metastasis status, a nomogram is developed, and the calculated C index is 0.674. This invention can further develop products related to the diagnosis and treatment of sub-metastatic breast cancer and prognosis according to the core genes in the prognostic model.

[0111] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for constructing a prognostic model of immune-related genes for patients with metastatic breast cancer, characterized in that: The following steps are involved: (1) Identify immune-related genes that are differentially expressed between primary and metastatic breast cancer lesions in the public databases GSE102818 and GSE43837, and obtain differentially expressed genes by taking the intersection of the two databases; (2) In GSE124648, patients with metastatic breast cancer were divided into an internal training set and a validation set, and a prognostic model was constructed in the internal training set using univariate and multivariate Cox regression analysis; (3) Screen out patients with metastatic breast cancer in the TCGA-BRCA dataset and apply the prognostic model constructed in step (2) to the dataset to verify the accuracy of the model.

2. The construction method according to claim 1, characterized in that: Four data sets were used in the construction, including three microarray data sets and one RNA-seq data set. The three microarray data sets were GSE102818, GSE43837, and GSE124648; the RNA-seq data set was the TCGA-BRCA cohort.

3. The construction method according to claim 1, characterized in that: The specific method of constructing in step (2) comprises the following steps: A. Preprocessing of the dataset GSE124648: GSE124648 contains clinical data of 140 patients with stage IV BC. The patients were randomly divided into an internal training set of 98 cases and a validation set of 42 cases using R (version 4.1.2); B. Univariate Cox regression analysis: Univariate Cox regression analysis was used on the internal training set to find potential prognostic genes; C. Multivariate Cox regression analysis: The potential prognostic genes found were intersected with the PPI network genes and included in the multivariate Cox regression analysis to obtain the regression coefficient of each factor related to prognosis; D. Model prediction: The model is used to predict the prognosis of patients with multinuclear leukemia in the training set. The risk value is βExp1+βExp2+…+βExpN, where EXP is the gene expression level, β is the regression coefficient of COX regression analysis, and N is the number of genes related to the patient's prognosis.

4. The construction method according to claim 1, characterized in that: Step (2) also includes the following steps: a. Calculate the risk score: Use the formula to calculate the risk score for each patient in the training set; b. Grouping patients: Patients in the training group were divided into high-risk and low-risk groups according to the cutoff values ​​determined by the 'Maxstat' package in R (version 4.1.2); c. Kaplan-Meier survival curves: Kaplan-Meier survival curves of high-risk and low-risk groups were evaluated using the R package ‘Survival’ (version 4.1.2); d. ROC curve evaluation: The R'time ROC' package (version 4.1.2) was used to generate the ROC curve to evaluate the predictive effect of IPM.

5. The construction method according to claim 1, characterized in that: The verification described in step (3) includes the following steps: ① BC patients with distant metastasis were selected from the TCGA-BRCA cohort as an external validation set to evaluate the prognostic value of the model; ②Univariate and multivariate COX regression analysis was performed on the clinical data of MBC patients to test whether the risk score was independent of other clinicopathological variables and to establish a nomogram; ③The accuracy and adaptability of the model were analyzed through C index analysis, 1-year, 3-year, and 5-year calibration analysis, and decision curve analysis (DCA).

6. A prognostic model for patients with metastatic breast cancer based on immune-related genes obtained by the construction method according to any one of claims 1 to 5.

7. Use of the prognostic model of metastatic breast cancer patients of immune-related genes according to claim 6 in screening markers for assessing the prognostic risk of metastatic breast cancer.

8. A marker for assessing the prognostic risk of metastatic breast cancer screened using the prognostic model of metastatic breast cancer patients of the immune-related genes of claim 6, characterized in that: The markers are gene CD28 and gene CFB.

9. A reagent set for detecting the marker according to claim 8, characterized in that: The reagent set includes a liquid reagent or a carrier reagent; The liquid reagent comprises a buffer, an antibody or a nucleic acid probe, a marker, and a control substance; The carrier reagent comprises a carrier, an antibody or a nucleic acid probe fixed on the carrier, a buffer solution, and a washing solution.

10. A kit for assessing the prognostic risk of metastatic breast cancer, characterized in that: The kit comprises the reagent set according to claim 9.

Images