Evaluation of cross-resistance gene set, scoring model and construction method and application thereof for her2-positive gastric cancer

Abstract

The application belongs to the field of biological information, and discloses a cross-resistance gene set for evaluating HER2-positive gastric cancer, a scoring model, a construction method and application thereof. The construction method comprises the following steps: screening a taxane drug resistance gene set and a trastuzumab drug resistance gene set; screening a preliminary cross-resistance gene set; screening 13 cross-resistance genes to construct a cross-resistance scoring model. The application also proposes a cross-resistance gene set for evaluating the prognosis and drug sensitivity of a HER2-positive gastric cancer patient and the use of the cross-resistance gene set in constructing a cross-resistance scoring model; and proposes a cross-resistance scoring model and a kit for evaluating the prognosis and drug sensitivity of a HER2-positive gastric cancer patient. The model can not only better evaluate the prognosis of a HER2-positive gastric cancer patient, but also has significant significance in predicting the drug sensitivity of a trastuzumab-resistant HER2-positive gastric cancer patient, and can realize individualized treatment.

Description

Technical Field

[0001] This invention belongs to the field of bioinformatics, specifically relating to the evaluation of cross-drug resistance gene sets, scoring models, their construction methods, and applications in HER2-positive gastric cancer. Background Technology

[0002] Human epidermal growth factor receptor 2 (HER2 or ERBB2) is a receptor tyrosine kinase whose overexpression / amplification occurs in 10–15% of invasive gastric cancers. Previous studies have found that HER2 overexpression can activate signaling pathways such as PI3K / Akt / mTOR and MAPK, which are associated with cell proliferation, survival, and angiogenesis in gastric cancer.

[0003] Trastuzumab is a recombinant humanized monoclonal antibody that effectively inhibits the proliferation of tumor cells overexpressing HER2. The ToGA trial found that trastuzumab combined with chemotherapy significantly improved overall survival in patients with HER2-positive gastric cancer. However, many patients with HER2-positive gastric cancer still fail to experience significant prognosis after trastuzumab treatment, primarily due to the development of trastuzumab resistance. Studies have identified trastuzumab resistance mechanisms including HER2-positive deletion / mutation and bypass activation. Currently, the treatment regimen for trastuzumab resistance remains chemotherapy, primarily taxane-based drugs, with an efficacy rate of less than 10%.

[0004] Meanwhile, studies have found a potential for cross-resistance between trastuzumab and chemotherapy. Cross-resistance means that resistance to the original drug by tumor cells may reduce their sensitivity to other drugs with different structures and mechanisms of action. Increasing evidence suggests that targeted therapies such as trastuzumab exhibit widespread cross-resistance with other anti-tumor drugs, negatively impacting patient survival. Therefore, accurately identifying patients with cross-resistance and developing individualized treatment plans is a pressing issue in cancer treatment. Summary of the Invention

[0005] The present invention aims to at least solve one of the technical problems existing in the prior art. To this end, the present invention proposes a method for constructing a cross-resistance scoring model for assessing the prognosis and drug sensitivity of HER2-positive gastric cancer patients. The constructed cross-resistance scoring model not only effectively assesses the prognosis of HER2-positive gastric cancer patients, but also has significant predictive value for drug sensitivity in trastuzumab-resistant HER2-positive gastric cancer patients, enabling individualized treatment for HER2-positive gastric cancer patients.

[0006] This invention also proposes a cross-resistance gene set for assessing the prognosis and drug sensitivity of HER2-positive gastric cancer patients.

[0007] The present invention also proposes the use of the aforementioned cross-resistance gene set in constructing a cross-resistance scoring model for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer.

[0008] This invention also proposes a cross-resistance scoring model for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer.

[0009] This invention also proposes a kit for evaluating the prognosis and drug sensitivity of patients with HER2-positive gastric cancer.

[0010] According to one aspect of the present invention, a method for constructing a cross-resistance scoring model for assessing prognosis and drug sensitivity in patients with HER2-positive gastric cancer is provided, comprising the following steps:

[0011] S1: Screening for taxane resistance gene sets and trastuzumab resistance gene sets;

[0012] S2: Through correlation analysis, explore the co-expression pattern of the taxane drug resistance gene set and the trastuzumab resistance gene set, screen out co-resistance genes and co-sensitive genes, and obtain a preliminary cross-resistance gene set; assign 1 as the scoring coefficient of the co-resistance gene and -1 as the scoring coefficient of the co-sensitive gene.

[0013] S3: A cohort of HER2-positive gastric cancer patients was obtained from the TCGA (The Cancer Atlas) database. Univariate Cox regression analysis was used to identify genes with a preliminary cross-resistance gene set that were prognostically correlated with the HER2-positive gastric cancer patient cohort and had a P-value < 0.05. Subsequently, 13 cross-resistance genes were selected: one with a hazard ratio (HR) > 1 and a resistance score coefficient of 1, and another with an HR < 1 and a resistance score coefficient of -1. These were used to construct the final cross-resistance scoring model, CR score. CR score = (-1*SNTB1 + CNTNAP3B + MTMR9 + NCE H1 + RIMKLB - 1*SLCO4A1 - 1*ETS2 - 1*MAP2K6 - 1*NR6A1 + ARHGAP29 + UPP1 - 1*ANKZ F1 - 1*FAM211A) / 13, where SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, and SLCO4A1 are cross-resistance genes. CO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A represent the gene expression levels of each gene in the HER2-positive gastric cancer patients described above; -1 and 1 represent drug resistance score coefficients.

[0014] In some embodiments of the present invention, the method for screening taxane resistance gene sets in step S1 includes: screening taxane-sensitive cell lines and taxane-resistant cell lines, and using an algorithm to screen for genes that are sensitive to or resistant to taxanes; taking the intersection of the taxane-resistant genes with taxane resistance-related genes screened through differential gene analysis to obtain the taxane resistance gene set.

[0015] Specifically, based on cell lines in the CTRP and PRISM databases, taxane-sensitive cell lines and taxane-resistant cell lines are screened according to the AUC value of the taxane drug.

[0016] More specifically, cell lines with AUC values ​​in the top 30% for the taxane drug are designated as taxane-sensitive cell lines, and those with AUC values ​​in the bottom 30% are designated as taxane-resistant cell lines.

[0017] Specifically, genes related to the binary outcome of sensitivity or resistance to the taxane drugs are screened using the Blasso algorithm.

[0018] Specifically, the differential gene analysis between drug-resistant and drug-sensitive cell lines was performed using the limma R package, and genes with log2(fold change) > 0 and corrected P value < 0.05 were selected as taxane drug resistance-related genes.

[0019] In some embodiments of the present invention, the method for screening the trastuzumab resistance gene set in step S1 includes: performing differential gene analysis between trastuzumab resistant group cell lines and sensitive group cell lines using the DESeq2 R package, screening out genes with log2(fold change) > 1.5 and corrected P value < 0.05, and obtaining the trastuzumab resistance gene set.

[0020] In some embodiments of the present invention, the correlation analysis in step S2 is Spearman correlation analysis, which explores the co-expression patterns of the taxane drug resistance gene set and the trastuzumab resistance gene set by creating a correlation matrix, thereby screening out the co-resistance genes and the co-sensitivity genes.

[0021] Specifically, by creating the correlation matrix, the correlation coefficients between the taxane drug resistance gene set and the trastuzumab resistance gene set are analyzed, and genes with a correlation coefficient > 0.5 and a P value < 0.05 are screened out.

[0022] Specifically, the co-resistance gene is a gene that is highly expressed in both the taxane drug resistance gene set and the trastuzumab resistance gene set in the drug-resistant group cells.

[0023] Specifically, the co-sensitive genes are genes that are highly expressed in both the taxane drug resistance gene set and the trastuzumab resistance gene set, and are simultaneously expressed in the sensitive group cells.

[0024] In some embodiments of the present invention, the method for obtaining the HER2-positive gastric cancer patient cohort in step S3 includes: obtaining transcriptome sequencing data and copy number variation data of gastric adenocarcinoma from the TCGA database, and screening patients with an ERBB2 site copy number of 2 from them, thereby obtaining the HER2-positive gastric cancer patient cohort.

[0025] In some embodiments of the present invention, the construction method further includes verifying the accuracy of the cross-drug resistance scoring model and the accuracy of using the cross-drug resistance scoring model to assess the prognosis.

[0026] In some embodiments of the present invention, methods for verifying the accuracy of the cross-resistance scoring model include verifying the accuracy by transcriptome sequencing, immunohistochemistry, or single-cell sequencing.

[0027] Specifically, the verification method of the transcriptome sequencing method includes: performing transcriptome sequencing on tissues of patients who progressed after receiving trastuzumab treatment and patients who were first diagnosed with HER2-positive gastric cancer; calculating the CR score for each patient according to the formula for calculating the CR score; and verifying the accuracy by comparing the CR scores of the two groups of patients.

[0028] Specifically, the immunohistochemical verification method includes: using immunohistochemistry to detect the protein expression of the most significantly different resistance genes (resistance score coefficient of 1) and sensitive genes (resistance score coefficient of -1) in tissues from patients who progressed after trastuzumab treatment and patients newly diagnosed with HER2-positive gastric cancer, and verifying the accuracy by comparing the protein expression of the two.

[0029] Specifically, the verification method of the single-cell sequencing method includes: performing single-cell sequencing on tumor cell populations of patients who have progressed after receiving trastuzumab treatment and patients who have been diagnosed with HER2-positive gastric cancer for the first time, identifying specific subpopulations, calculating the CR score of each tumor cell sample according to the CR score calculation formula, and verifying the accuracy by comparing the CR scores of tumor cells from the two groups of patients in each tumor cell subpopulation.

[0030] In some embodiments of the present invention, the accuracy of using the cross-resistance scoring model to assess the prognosis is verified by using the Kaplan-Meier plotter database.

[0031] Specifically, due to database discrepancies, the Kaplan-Meier plotter database does not include CNTNAP3B sequencing data. Therefore, based on the CR score calculation formula, the CR scores of HER2-positive gastric cancer patients from the GEO cohort were obtained using the remaining 12 cross-resistance genes. Patients were divided into a high CR score group and a low CR score group using the median CR score as the cutoff point. Kaplan-Meier plot survival curves were plotted for both groups, and the Log-rank test was used to compare the survival times of the high- and low-risk groups. A Log-rank P-value < 0.05 was used to determine if there was a statistically significant difference between the two groups, thus verifying the accuracy of the cross-resistance model in assessing the prognosis.

[0032] According to a second aspect of the invention, a set of cross-resistance genes for assessing prognosis and drug sensitivity in patients with HER2-positive gastric cancer is proposed, including SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A.

[0033] According to a third aspect of the invention, the use of the aforementioned cross-resistance gene set in constructing a cross-resistance scoring model for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer is proposed.

[0034] According to a fourth aspect of the present invention, a cross-resistance scoring model for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer is proposed. The cross-resistance scoring model is a CR score, which is characterized by summing the products of the expression levels of each gene in the cross-resistance gene set and the corresponding resistance score coefficients, and then dividing by the number of genes in the cross-resistance gene set. The CR score is defined as: CR score = (-1*SNTB1+CNTNAP3B+MTMR9+NCEH1+RIMKLB-1*SLCO4A1-1*ETS2-1*MAP2K6-1*NR6A1+ARHGAP29+UPP1-1*ANKZF1-1*FAM211A) / 13, where SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29+UPP1-1*ANKZF1-1*FAM211A ... HGAP29, UPP1, ANKZF1, and FAM211A represent the gene expression levels of each gene in the HER2-positive gastric cancer patients described above; -1 and 1 represent drug resistance score coefficients.

[0035] According to a fifth aspect of the invention, a kit for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer is provided, comprising primers or probes for detecting the expression levels of each gene in the cross-resistance gene set.

[0036] In some embodiments of the present invention, the kit is used to detect the expression levels of each gene in the cross-resistance gene set by real-time quantitative PCR, sequencing, or gene chip technology.

[0037] According to a preferred embodiment of the present invention, at least the following beneficial effects are achieved:

[0038] Mounting evidence suggests widespread cross-resistance between targeted therapies like trastuzumab and other anti-tumor drugs, limiting the therapeutic efficacy of different classes of anti-tumor drugs and significantly reducing patient survival. Therefore, developing methods for screening CS drugs based on the understanding and research of collateral sensitivity (CS) is crucial for drug-resistant patients. This invention provides, for the first time, a method for constructing a quantifiable cross-resistance scoring model between trastuzumab and chemotherapy drugs (taxanes), and identifies 13 cross-resistance genes that can assess drug sensitivity in HER2-positive gastric cancer patients. This invention further demonstrates that this cross-resistance scoring model not only effectively assesses the prognosis of HER2-positive gastric cancer patients but also has significant predictive value for drug sensitivity in HER2-positive gastric cancer patients resistant to trastuzumab. This invention attempts to stratify patient drug resistance by constructing a cross-resistance scoring model and predict sensitive drugs after cross-resistance using gene expression in drug-resistant cells, further expanding the drug treatment spectrum for cancer patients and enabling personalized treatment. This has significant implications and clinical application value for personalized medicine. Attached Figure Description

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments, wherein:

[0040] Figure 1 This is a schematic diagram illustrating the construction process of a cross-resistance scoring model for evaluating the prognosis and drug sensitivity of HER2-positive gastric cancer patients in this embodiment of the invention, as well as the verification of the model's accuracy and the method for predicting drug sensitivity using the model.

[0041] Figure 2 This is a graph showing the cross-resistance score test results for each patient in Example 2 of the present invention;

[0042] Figure 3The image shows the immunostaining results of the drug resistance genes (MTMR9 and ARHGAP29) and the sensitive genes (ANKZF1 and MAP2K6) in Example 3 of this invention.

[0043] Figure 4 This is a graph showing the cross-resistance score of tumor cell subsets in the treatment progression group (drug-resistant group) and the initial diagnosis group (sensitive group) in Example 4 of the present invention; where 1-sensitive group, 2-drug-resistant group;

[0044] Figure 5 This is a KM plot survival curve for patients in the high CR score group and the low CR score group in Example 5 of the present invention;

[0045] Figure 6 This is a graph showing the predicted drug sensitivity of paclitaxel, fluorouracil, gemcitabine, and cabazitaxel in patients with high CR score and low CR score in Example 6 of the present invention.

[0046] Figure 7 This is a graph showing the predicted drug sensitivity of saplatin, cisplatin, platinum, and 5-fluorouracil in patients with high CR score and low CR score in Example 6 of the present invention.

[0047] Figure 8 This is a graph showing the predicted drug sensitivity of gemcitabine, capecitabine, and mecitabine in patients with high CR score and low CR score in Example 6 of the present invention. Detailed Implementation

[0048] The embodiments of the present invention are described in detail below. These embodiments are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0049] In the description of this invention, unless otherwise explicitly defined, terms such as "co-expression" and "screening" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0050] In the description of this invention, references to terms such as "one embodiment," "some embodiments," etc., indicate that a specific feature, material, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment. Furthermore, the specific features, materials, or characteristics described may be combined in any suitable manner in one or more embodiments.

[0051] Unless otherwise specified, the experimental methods used in the examples are conventional methods; unless otherwise specified, the materials and reagents used are commercially available.

[0052] The following examples illustrate the construction process of a cross-resistance scoring model for assessing the prognosis and drug sensitivity of HER2-positive gastric cancer patients, the validation of the model's accuracy, and the method for predicting drug sensitivity using this model. Figure 1 As shown.

[0053] Example 1

[0054] This embodiment constructs a cross-resistance scoring model for assessing the prognosis and drug sensitivity of HER2-positive gastric cancer patients. The specific process is as follows:

[0055] 1. Screening of Taxane Resistance Gene Sets: Cell lines with the highest AUC values ​​for taxanes in the CTRP (https: / / portals.broadinstitute.org / ctrp / ) and PRISM (https: / / www.theprismlab.org / ) databases were designated as the susceptible cell lines, while those with the lowest AUC values ​​were designated as the resistant cell lines. The Blasso algorithm was then used for 300 iterations to identify the top 20 important genes associated with either susceptible or resistant outcomes for each taxane. Simultaneously, differential gene analysis was performed between the resistant and susceptible cell lines using the limma R package, identifying genes with log2(fold change) > 0 and corrected P-value < 0.05 as taxane resistance-related genes. The intersection of the 20 taxane resistance-related genes identified by the Blasso algorithm and the results of differential gene analysis yielded the taxane resistance gene set (59 genes).

[0056] 2. Screening of Trastuzumab Resistance Gene Set: Transcriptome data from three N87 cell lines in the trastuzumab-resistant group and three N87 cell lines in the sensitive group were analyzed by BGI Genomics. Differential gene analysis was performed on the transcriptome data of the trastuzumab-resistant and sensitive cell lines using the DESeq2 R package. Genes with log2(foldchange) > 1.5 and corrected P value < 0.05 were screened to obtain the trastuzumab resistance gene set (number of genes = 2097).

[0057] 3. Exploring the co-expression patterns of taxane resistance genes and trastuzumab resistance genes: A correlation matrix was created using Spearman correlation analysis to analyze the correlation coefficients between the taxane resistance gene set and the trastuzumab resistance gene set. Genes with a correlation coefficient > 0.5 and a P value < 0.05 were screened. Genes highly expressed in both the taxane and trastuzumab gene sets in the resistant group cells were defined as co-resistance genes, assigned a coefficient of 1 as the resistance score coefficient. Conversely, genes highly expressed in both the taxane and trastuzumab resistance gene sets in the sensitive group cells were defined as sensitive genes, assigned a coefficient of -1 as the resistance score coefficient. A total of 794 preliminary cross-resistance genes were obtained, pending further analysis.

[0058] 4. Identification of Cross-Resistance Gene Set Based on TCGA Database: Transcriptome and copy number variation data of gastric adenocarcinoma were obtained from The Cancer Genome Atlas (TCGA) database. Fifty-one patients with an ERBB2 copy number of 2 were selected to form a cohort of HER2-positive gastric cancer patients. Univariate Cox regression analysis was used to analyze the expression matrix of this cohort in TPM format and identify genes associated with prognosis. Genes present in the preliminary cross-resistance gene set with a P-value < 0.05 were identified. Further screening revealed 13 genes with HR > 1 and resistance score coefficient = 1, and 13 genes with HR < 1 and resistance score coefficient = -1, which were used to construct the final cross-resistance scoring model. The results are shown in Table 1.

[0059] Table 1

[0060]

[0061]

[0062] The constructed cross-resistance scoring model CR score = (-1*SNTB1+CNTNAP3B+MTMR9+NCEH1+RIMKLB-1*SLCO4A1-1*ETS2-1*MAP2K6-1*NR6A1+ARHGAP29+UPP1-1*ANKZF1-1*FAM211A) / 13, where SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A represent the gene expression levels of each gene in the HER2-positive gastric cancer patients; -1 and 1 represent the resistance score coefficients.

[0063] Example 2

[0064] This embodiment uses first-line cohort CR score (transcriptome sequencing) to verify the accuracy of the cross-resistance scoring model constructed in Example 1. The specific process is as follows:

[0065] Fresh tumor tissue or frozen specimens were collected from 25 patients with HER2-positive gastric cancer for transcriptome sequencing. Among them, 8 patients had progressed after trastuzumab treatment, and 17 patients were newly diagnosed with HER2-positive gastric cancer. CR scores were calculated based on the gene expression profiles provided by sequencing to further validate the correlation between cross-resistance scores and cross-resistance phenomena. The expression level of CR genes in each patient in this cohort was multiplied by the resistance score coefficient, weighted, summed, and then divided by the number of CR genes to obtain the CR score for each patient. The results are as follows: Figure 2 As shown. From Figure 2 It can be seen that the CR score of patients with treatment progression (drug-resistant group) was significantly higher than that of patients with initial diagnosis (sensitive group). These results further validate that the model constructed in Example 1 can predict the possibility of patients developing cross-drug resistance through cross-resistance scores.

[0066] Example 3

[0067] This embodiment uses immunohistochemistry to verify the accuracy of the cross-drug resistance scoring model constructed in Example 1. The specific process is as follows:

[0068] Forty-four paraffin-embedded tumor tissue sections were collected from patients with HER2-positive gastric cancer. Among them, nine patients had progressed after trastuzumab treatment, and 35 patients were newly diagnosed with HER2-positive gastric cancer. The protein expression of the most significant resistance genes (resistance score coefficient of 1) and sensitive genes (resistance score coefficient of -1) in the cross-resistance gene cluster were detected. The results are as follows: Figure 3 As shown, immunohistochemical staining results revealed increased protein expression of resistance genes (MTMR9 and ARHGAP29) and decreased protein expression of sensitive genes (ANKZF1 and MAP2K6) in patients who progressed after trastuzumab treatment (resistant group). Conversely, increased protein expression of sensitive genes (ANKZF1 and MAP2K6) and decreased protein expression of resistance genes (MTMR9 and ARHGAP29) were observed in patients newly diagnosed with HER2-positive gastric cancer (sensitive group). This indicates that cross-resistance scoring based on cross-resistance gene sets can indirectly reflect the cross-resistance status in patients who progressed after trastuzumab treatment.

[0069] Example 4

[0070] This embodiment uses single-cell sequencing to verify the accuracy of the cross-drug resistance scoring model constructed in Example 1. The specific process is as follows:

[0071] (1) Data comparison and quality control: Fresh tumor tissues were collected from 6 patients with HER2-positive gastric cancer, including 3 patients newly diagnosed with HER2-positive gastric cancer and 3 patients whose cancer progressed after trastuzumab treatment. Gene expression matrices for each cell were constructed using CellRanger software (version 6.0.1) based on the raw data from the above 6 samples. The SoupXR package (v1.5.2) was used with default parameters to remove environmental RNA contamination. Then, the DoubletFinder software (v2.0.3) was used to remove double cells with nExp set to 0.05, pK value estimated using mean-variance normalized bimodal coefficients for each dataset, and pN value set to 0.25. The data after removing double cells were then imported into the Seurat R package (v4.0.4). Low-quality cells were removed based on the distribution map of gene count (nFeature_RNA) for each cell. Cells with fewer than 300 or more than 7000 genes (7000 for F68 and F161, and 7500 for F166) and mitochondrial gene content greater than 50% were removed. To exclude the influence of mitochondrial ribosome genes on clustering, mitochondrial ribosome genes were removed from the expression matrix. The expression matrix of the remaining cells was used for subsequent merge clustering analysis.

[0072] (2) Gene Expression Standardization: After filtering low-quality cells, the following operations were performed on the samples: First, to remove the influence of library depth, the NormalizeData function was used with the LogNormalize global scaling normalization method to normalize the gene expression value of each cell using the total expression value, and then multiplied by a scaling factor (default is 10000). The result was then logarithmically transformed. Next, the FindVariableFeatures function was used to obtain 1000 highly variable genes from each sample for downstream analysis. To eliminate the influence of highly expressed genes on subsequent analysis, the ScaleData linear transformation function was used to process the data. By changing the expression of each gene, the average expression between cells was reduced to 0, and the expression of each gene was scaled to reduce the difference between cells to 1.

[0073] (3) Dimensionality Reduction and Clustering: The RunPCA function was used to perform PCA (Principal Component Analysis) dimensionality reduction on the scaled data after standardization, reducing it to 50 dimensions by default. Next, to eliminate batch effects between samples, Harmony's ensemble method was used for data ensemble. For the ensembled data, the FindNeighbors function was used to perform cell clustering analysis in the 50 PC dimensions based on a graph-based clustering approach. Simultaneously, the FindClusters function was used to set the cluster resolution to 0.3 for cell clustering. Finally, the RunUMAP function was used for non-linear dimensionality reduction, further reducing the data in the 50 PC dimensions to two dimensions for data visualization.

[0074] (4) Identification of tumor epithelial cells: The markers for each cluster were obtained using Seurat's FindAllMarkers function with default parameters, thus obtaining the clusters expressing epithelial cell markers. Subsequently, with endothelial cells as the background cells, tumor epithelial cells were identified from the epithelial cells using inferCNV.

[0075] (5) Cell type subpopulation subdivision: To determine the specific subpopulations of tumor epithelial cells, a second round of UMAP dimensionality reduction was performed on the tumor epithelial cell subpopulations. The number of principal components of each major subpopulation was independently determined by the Elbowplot function implemented in Seurat v3. In addition, doublets in each cell cluster were identified and filtered out by graph-based clustering and the R package DoubletFinder50, finally obtaining the subdivided subpopulations of tumor epithelial cells.

[0076] (6) Cross-resistance scoring: Cross-resistance scores were performed on each cell in the subgroups of tumor epithelial cells, and the cross-resistance scores of each subgroup in the treatment progression group were compared with those in the initial diagnosis group. Results are as follows: Figure 4 As shown, the CR score of tumor cells in patients with treatment progression (drug-resistant group) was higher than that in patients at first diagnosis (sensitive group). Furthermore, the CR scores of tumor cell subsets (subsets 0, 1, 2, 3, 4, and 6) were also higher than those in patients at first diagnosis. This indicates that the CR score can reflect cross-resistance between trastuzumab and chemotherapy drugs.

[0077] Example 5

[0078] This embodiment utilizes the online database Kaplan-Meier plotter to explore the guiding significance of the cross-resistance scoring model constructed in Example 1 for the prognosis and treatment of HER2-positive gastric cancer patients. The specific process is as follows:

[0079] A sample size of 202 HER2-positive gastric cancer patients was obtained from the Kaplan-Meier plotter online database. The complete response (CR) score (CR) was calculated for each patient. Patients were then ranked from lowest to highest CR score, and the median CR score was used as the split point to divide them into a high CR score group and a low CR score group. Kaplan-Meier plot survival curves were plotted for both groups. The Log-rank test was used to compare the survival times of the two groups, with a Log-rank P-value < 0.05 considered statistically significant. Due to database differences, CNTNAP3B was not sequenced in the Kaplan-Meier plotter online database; therefore, the CR scoring gene set included the remaining 12 genes. Results are as follows: Figure 5 As shown, the survival prognosis of patients in the high CR score group was worse than that of patients in the low CR score group. The Log-rank P value was 0.0046 < 0.05, indicating that the CR score can assess the patient's prognosis.

[0080] Example 6

[0081] This embodiment utilizes ridge regression analysis to predict drug sensitivity in HER2-positive gastric cancer patients using the cross-resistance scoring model constructed in Example 1. The specific process is as follows:

[0082] The cross-resistance scoring model constructed in Example 1 was applied to HER2-positive gastric cancer patients in the TCGA database. Patients were then ranked from lowest to highest according to their CR score, and the median CR score was used as the dividing point to separate them into a high CR score group and a low CR score group. Subsequently, a ridge regression model based on cell line expression profiles from the CTRP and PRISM databases was constructed using the pRRophetic R package. The pRRophetic algorithm was used to predict the IC50 of chemotherapy drugs, TKIs, and other therapeutic agents in the CTRP and PRISM databases for both the high and low CR score groups. The log2 (fold change) value and p-value were calculated for the resistance group compared to the sensitive group. A log2 (fold change) value > 0 indicated a trend towards drug resistance, a log2 (fold change) value < 0 indicated a trend towards drug sensitivity, and a p-value < 0.05 was considered statistically significant. Results are as follows: Figures 6-8As shown, common chemotherapy drugs such as taxanes (paclitaxel, cabazitaxel), platinum-based drugs (saplatin, cisplatin, platinum), and pyrimidine-based drugs (fluorouracil, gemcitabine, capecitabine, mecetabine) all showed a trend of drug resistance in patients with high CRscore.

[0083] The embodiments of the present invention have been described in detail above. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention. Furthermore, the embodiments of the present invention and the features thereof can be combined with each other unless otherwise specified.

Claims

1. A method for constructing a cross-resistance scoring model to assess the prognosis and drug sensitivity of patients with HER2-positive gastric cancer, characterized in that, Includes the following steps: S1: Screening for taxane resistance gene sets and trastuzumab resistance gene sets; S2: Through correlation analysis, explore the co-expression pattern of the taxane drug resistance gene set and the trastuzumab resistance gene set, screen out co-resistance genes and co-sensitive genes, and obtain a preliminary cross-resistance gene set; assign 1 as the scoring coefficient of the co-resistance gene and -1 as the scoring coefficient of the co-sensitive gene. S3: Obtain a cohort of HER2-positive gastric cancer patients from the TCGA database, and use univariate Cox regression analysis to determine the correlation between the preliminary cross-resistance gene set and the prognosis of the HER2-positive gastric cancer patient cohort. P Genes with a value < 0.05 were then selected; subsequently, 13 cross-resistance genes with HR > 1 and resistance score coefficient = 1, and HR < 1 and resistance score coefficient = -1 were selected to construct the final cross-resistance scoring model CR score; CR score = (-1*SNTB1+CNTNAP3B+MTMR9+NCEH1+RIMKLB-1*SLCO4A1-1*ETS2-1*MAP2K6-1*NR6A1+ARHGAP29+UPP1-1*ANKZF1-1*FAM211A) / 13, where SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A represent the gene expression levels of each gene in the HER2-positive gastric cancer patients; -1 and 1 represent the resistance score coefficients.

2. The construction method according to claim 1, characterized in that, The method for screening taxane resistance gene sets in step S1 includes: screening taxane-sensitive cell lines and taxane-resistant cell lines, and using an algorithm to screen for genes that are sensitive to or resistant to taxanes; taking the intersection of the taxane-resistant genes with taxane resistance-related genes screened through differential gene analysis to obtain the taxane resistance gene set.

3. The construction method according to claim 2, characterized in that, Based on cell lines in the CTRP and PRISM databases, taxane-sensitive and taxane-resistant cell lines were screened according to the AUC values ​​of the taxane drugs.

4. The construction method according to claim 2, characterized in that, Genes associated with binary outcomes of sensitivity or resistance to the taxane drugs were screened using the Blasso algorithm.

5. The construction method according to claim 2, characterized in that, The differential gene analysis between drug-resistant and drug-sensitive cell lines was performed using the limma R package, and cells with log2(fold change) > 0 and corrected for differences were screened. P Genes with a value < 0.05 are designated as taxane drug resistance-related genes.

6. The construction method according to claim 1, characterized in that, Step S1, the method for screening the trastuzumab resistance gene set, includes: performing differential gene analysis between trastuzumab-resistant and sensitive cell lines using the DESeq2 R package, and screening for genes with log2 (fold change) > 1.5 and corrected for... P Genes with a value < 0.05 were used to obtain the trastuzumab resistance gene set.

7. The construction method according to claim 1, characterized in that, The correlation analysis in step S2 is a Spearman correlation analysis. By creating a correlation matrix, the co-expression patterns of the taxane drug resistance gene set and the trastuzumab resistance gene set are explored, thereby screening out the co-resistance genes and the co-sensitivity genes.

8. The construction method according to claim 7, characterized in that, By creating the correlation matrix, the correlation coefficients between the taxane drug resistance gene set and the trastuzumab resistance gene set were analyzed, and genes with a correlation coefficient > 0.5 were screened. P Genes with a value < 0.

05.

9. The construction method according to claim 1, characterized in that, The method for obtaining the HER2-positive gastric cancer patient cohort in step S3 includes: obtaining transcriptome sequencing data and copy number variation data of gastric adenocarcinoma from the TCGA database, and screening patients with a copy number of 2 at the ERBB2 site to obtain the HER2-positive gastric cancer patient cohort.

10. The construction method according to claim 1, characterized in that, The construction method also includes verifying the accuracy of the cross-drug resistance scoring model and the accuracy of using the cross-drug resistance scoring model to assess the prognosis.

11. The construction method according to claim 10, characterized in that, Methods for verifying the accuracy of the cross-resistance scoring model include using transcriptome sequencing, immunohistochemistry, or single-cell sequencing.

12. The construction method according to claim 10, characterized in that, The accuracy of using the cross-resistance scoring model to assess the prognosis was verified using the Kaplan-Meier plotter database.

13. A cross-resistance gene set for assessing prognosis and drug sensitivity in patients with HER2-positive gastric cancer, characterized in that, The cross-resistance gene set consists of SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A.

14. Use of the cross-resistance gene set of claim 13 in constructing a cross-resistance scoring model for assessing prognosis and drug sensitivity in patients with HER2-positive gastric cancer.

15. A cross-resistance scoring model for assessing prognosis and drug sensitivity in patients with HER2-positive gastric cancer, characterized in that, The cross-resistance scoring model is the CR score, which is composed of the sum of the products of the expression levels of each gene in the cross-resistance gene set as described in claim 13 and the corresponding resistance score coefficients, and then divided by the number of genes in the cross-resistance gene set. The CR score is defined as: CR score = (-1*SNTB1+CNTNAP3B+MTMR9+NCEH1+RIMKLB-1*SLCO4A1-1*ETS2-1*MAP2K6-1*NR6A1+ARHGAP29+UPP1-1*ANKZF1-1*FAM211A) / 13, where SNTB1, CNTNAP3B, MTMR9, NCEH1, RIMKLB, SLCO4A1, ETS2, MAP2K6, NR6A1, ARHGAP29, UPP1, ANKZF1, and FAM211A represent the gene expression levels of each gene in the HER2-positive gastric cancer patients, and -1 and 1 represent the resistance score coefficients.

16. A kit for assessing the prognosis and drug sensitivity of patients with HER2-positive gastric cancer, characterized in that, The kit includes primers or probes for detecting the expression levels of each gene in the cross-resistance gene set of claim 13.

Images