Marker for predicting persistent drug resistance and clinical lifetime of tumor treatment and application of marker

Abstract

The invention discloses a marker for predicting persistent drug resistance and clinical lifetime of tumor treatment and application of the marker. The marker is composed of at least 16 genes selected from the following 23 genes: EDIL3, ARHGEF37, MAP2, TMEM132B, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, LMO1, NOV, GRIK2, GRIA2, SNAI2, LY96, ALPP, IGFL1, BEST3, IL13 or NTF4. The invention also discloses a method for preparing the marker. The marker is remarkably up-regulated when a plurality of cancers are subjected to clinical drug treatment and have persistent drug resistance, and is remarkably related to shortened clinical lifetime and poor prognosis of tumor patients. The occurrence of persistent drug resistance during clinical tumor treatment and the total clinical lifetime of tumor patients can be predicted by detecting the expression quantity of the marker in the tumor patients.

Description

Technical Field

[0001] The present invention relates to the technical field of medical diagnostic markers, and more specifically, to a marker for predicting persistent drug resistance and clinical survival period of tumor treatment and its application. Background Art

[0002] At present, malignant tumors are one of the leading causes of death worldwide. In addition to surgical treatment, drug therapy is the main means to fight cancer. However, tumor resistance and recurrence are the main reasons for the failure of cancer treatment. Lung cancer patients treated with EGFR-TKi gefitinib for about a year will develop irreversible TKI resistance in most patients; liver cancer patients who are sensitive to sorafenib initial treatment will develop acquired resistance after about 6 months of treatment; BRAFV600 MUT Even with the combined treatment of BRAFi and MEKi, less than 20% of melanoma patients survive for more than five years, so tumor resistance is a major challenge that needs to be overcome in the clinic. However, more and more studies have shown that in addition to genetic resistance of tumors, reversible non-genetic resistance plays an important role in the final clinical resistance and recurrence of tumors. This group of non-genetic resistant tumor cells is called "drug tolerance persister cancer cells" (DTP) by scientists. They gain a short-term survival advantage under the lethal pressure of therapeutic drugs, and gradually acquire genomic changes related to drug resistance in a persistent state, and eventually develop into irreversible resistant cells. It can be seen that persistent drug-resistant tumor cells (DTP) are actually a "reservoir" of drug-resistant cells and a major "culprit" of tumor resistance and recurrence. However, clinical treatment methods and effective therapeutic drugs for DTP have not yet been established, and molecular markers and marker genes that can detect this persistent drug resistance during tumor treatment have not yet been established. Summary of the invention

[0003] The purpose of the present invention is to overcome the defects and shortcomings of the prior art, which lack molecular markers and detection methods for predicting persistent drug resistance and clinical survival of tumor treatment, and to provide a group of markers for predicting persistent drug resistance and clinical survival of tumor treatment.

[0004] The second object of the present invention is to provide applications of the marker.

[0005] The above-mentioned object of the present invention is achieved through the following technical solutions:

[0006] The present invention first provides a group of marker combinations for predicting persistent drug resistance and clinical survival in tumor treatment, wherein the marker combination consists of at least 16 of the 23 genes EDIL3, ARHGEF37, MAP2, TMEM132B, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, LMO1, NOV, GRIK2, GRIA2, SNAI2, LY96, ALPP, IGFL1, BEST3, IL13 or NTF4.

[0007] in:

[0008] EDIL3 is the EGF-like repeats and discoidin I-likedomains 3 gene;

[0009] ARHGEF37 is the Rho Guanine Nucleotide Exchange Factor 37 gene;

[0010] MAP2 is the microtubule-associated protein 2 gene;

[0011] TMEM132B is the transmembrane protein 132B gene;

[0012] TSPAN10 is the tetraspanin 10 gene;

[0013] FAM71F2 is the Golgi Associated RAB2Interactor1A gene;

[0014] PSORS1C1 psoriasis susceptibility 1 candidate 1 gene;

[0015] FOXD4 is the forkhead box D4 gene;

[0016] GDAP1L1 is the ganglioside-induced differentiation-associated protein 1-like 1 gene;

[0017] TMEM52B is the transmembrane protein 52B gene;

[0018] SCIN scinderin gene;

[0019] KIF26B is the kinesin family member 26B gene;

[0020] LMO1 is the LIM domain only 1 (rhombotin 1) gene;

[0021] NOV is a nephroblastoma over-expressed gene;

[0022] GRIK2 is the glutamate receptor, ionotropic, kainate 2 gene;

[0023] GRIA2 is the gene for AMPA 2 glutamate receptor, ionotropic, AMPA 2.

[0024] SNAI2 is the snail homolog 2 (Drosophila) nuclear gene;

[0025] LY96 is the lymphocyte antigen 96 gene;

[0026] ALPP is alkaline phosphatase, placental (Regan isozyme) gene;

[0027] IGFL1 is the insulin growth factorlike family member 1 gene;

[0028] BEST3 is the Vitelliform Macular Dystrophy2-Like Protein 3 gene;

[0029] IL13 is the interleukin 13 gene;

[0030] NTF4 is the neurotrophin 4 gene.

[0031] The present invention has discovered a gene set consisting of 23 genes. When persistent drug resistance occurs in various cancers receiving clinical drug treatment, the gene expression levels in the gene set are significantly upregulated, and this abnormally high expression is significantly associated with a shortened clinical survival and poor prognosis of tumor patients. Therefore, this gene set consisting of 23 genes can be used as a molecular detection marker for predicting persistent drug resistance, clinical survival, and poor prognosis in tumor treatment. Since treatment resistance and recurrence often occur in clinical tumor treatment at present, and persistent drug resistance in tumor treatment is a "prelude" to the development of final drug resistance of tumors, molecular markers and detection methods for predicting the occurrence of persistent drug resistance in tumor treatment have not yet been established. The present invention provides a gene set for predicting persistent drug resistance and clinical survival in tumor treatment. By detecting the expression level of this gene set in tumor patients, the probability or risk of persistent drug resistance during clinical tumor treatment and the overall clinical survival of tumor patients can be predicted, thereby helping to timely adjust the clinical treatment plan for tumors, delay the occurrence of final drug resistance in tumor treatment, and improve the overall clinical survival.

[0032] Furthermore, the marker combination consists of 16 genes, namely, ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13. Using this set of 16 genes, a risk model for predicting persistent drug resistance and clinical survival in tumor treatment can be constructed.

[0033] The present invention also provides the use of any of the above-mentioned marker combinations in constructing a risk assessment model or system for predicting persistent drug resistance and clinical survival in tumor treatment.

[0034] The present invention also provides a risk assessment model for predicting persistent drug resistance and clinical survival of tumor treatment, wherein the risk assessment model is based on the expression of markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13 and the corresponding coefficients. The risk value formula composed of the sum of products is used to obtain the risk value score of tumor patients. Risk value = (-0.1150×ARHGEF37)+(0.0470×MAP2)+(-0.0491×TSPAN10)+(-0.0176×FAM71F2)+(0.0240×PSORS1C1)+(-0.0813×FOXD4)+(-0.0058×GDAP1L1)+(-0.0587×TME M52B)+(-0.0122×SCIN)+(-0.0300×KIF26B)+(0.0201×NOV)+(-0.0349×GRIA2)+(0.0734×LY96)+(0.0877×ALPP)+(0.0547×BEST3)+(0.0395×IL13).

[0035] The present invention also provides a risk scoring system for predicting persistent drug resistance and clinical survival in tumor treatment, the system comprising a data input module, a model calculation module and a result output module, the model calculation module calculates the markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 through a scoring formula The risk score of tumor patients was obtained by the risk value formula composed of the sum of the product of the gene expression of IL13 and the corresponding coefficient. Risk value = (-0.1150×ARHGEF37)+(0.0470×MAP2)+(-0.0491×TSPAN10)+(-0.0176×FAM71F2)+(0.0240×PSORS1C1)+(-0.0813×FOXD4)+(-0.0058×GDAP1L1)+(-0.0587×TME M52B)+(-0.0122×SCIN)+(-0.0300×KIF26B)+(0.0201×NOV)+(-0.0349×GRIA2)+(0.0734×LY96)+(0.0877×ALPP)+(0.0547×BEST3)+(0.0395×IL13).

[0036] Specifically, the data input module is used to input the gene expression levels of patient markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13 into the model calculation module.

[0037] The gene expression levels of the above-mentioned markers can be obtained by conventional technical means in the art, such as sequencing, quantitative PCR detection, etc.

[0038] Specifically, the model calculation module also includes comparing the risk value obtained in the model calculation module with a threshold, and the threshold is set to 0. When used to predict persistent drug resistance in tumor treatment, if the score is greater than the threshold, it is a high-risk group, and if the score is less than the threshold, it is a low-risk group; when used to predict poor prognosis of clinical survival, if the score is greater than the threshold, the survival period is shorter, and if the score is less than the threshold, the survival period is longer.

[0039] The present invention also provides the use of any of the above-mentioned markers in the preparation of a kit for predicting persistent drug resistance and clinical survival period of tumor treatment.

[0040] Furthermore, the drugs used in the treatment of persistent drug resistance in tumors include but are not limited to any one of gefitinib, vemurafenib, trametinib or lapatinib.

[0041] Furthermore, the tumor includes but is not limited to any one of non-small cell lung cancer, melanoma or breast cancer.

[0042] The present invention also provides a kit for predicting persistent drug resistance and clinical survival period of tumor treatment, and the kit also contains a reagent for detecting the gene expression amount of any of the above-mentioned marker combinations.

[0043] The present invention also provides the use of the gene set composed of the 23 genes as a therapeutic target and drug development target for persistent drug resistance in anti-tumor treatment and final drug resistance of tumors in the preparation of drugs for persistent drug resistance in anti-tumor treatment. Tumor treatment resistance and recurrence are the biggest obstacles in clinical tumor treatment. Currently, more and more studies are focused on exploring the mechanism of tumor treatment resistance and recurrence, and are committed to developing new targets and drug molecules for anti-tumor treatment resistance. Starting from the study of persistent drug resistance in tumor treatment, the present invention has found a gene set composed of 23 genes, which are abnormally expressed in the occurrence of persistent drug resistance in tumor treatment. This abnormal high expression is significantly correlated with the shortened survival of clinical tumors. Therefore, this gene set composed of 23 genes may be used as a therapeutic target and drug development target for persistent drug resistance and final drug resistance of tumors in anti-tumor treatment.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The present invention provides a marker for predicting persistent drug resistance and clinical survival in tumor treatment. The marker is significantly upregulated when persistent drug resistance occurs in various cancers receiving clinical drug treatment, and is significantly correlated with shortened clinical survival and poor prognosis of tumor patients. The occurrence of persistent drug resistance during clinical tumor treatment and the overall clinical survival of tumor patients can be predicted by detecting the expression level of the marker in tumor patients, thereby helping to adjust the clinical treatment plan of tumors in a timely manner and delay the occurrence of final drug resistance in tumor treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 In order to analyze the transcriptome data of tumor persistent drug resistance cell model and GEO database, a gene set consisting of 23 genes was found. A: Schematic diagram of the occurrence process of tumor persistent drug resistance cells (DTP); B: The intersection of three DTP datasets and TCGA tumor datasets was used to draw a Venn diagram to obtain a gene set consisting of 23 genes; C: The establishment and evaluation process of the risk model for predicting tumor persistent drug resistance and tumor clinical treatment survival for this gene set consisting of 23 genes.

[0047] Figure 2 The expression levels of the 23-gene set in different tumors and the GO analysis of the genes. A: Comparison of the expression levels of the 23-gene set in tumor tissues (T) and normal tissues (N) of breast cancer, liver cancer and lung adenocarcinoma; B: GO analysis of the 23-gene set, and the BP analysis type is selected to draw a table.

[0048] Figure 3 KM survival analysis and survival correlation map of a gene set composed of 23 genes and patients with breast cancer, liver cancer and lung adenocarcinoma. AC: Through KM survival analysis, the 23 genes were screened out and were related to the shortened clinical overall survival (OS) of breast cancer, liver cancer and lung adenocarcinoma; D: The clinical survival effects of the 23 genes on breast cancer, liver cancer and lung adenocarcinoma (survival map). HR stands for Hazard Ratio. The larger the HR, the greater the risk value, and the more likely it is to lead to a shortened clinical survival.

[0049] Figure 4Univariate clinical survival analysis and LASSO regression analysis were performed on the gene set consisting of 23 genes, and a prediction model was established to evaluate the prediction efficiency of the prediction model for clinical overall survival of different tumors. A: Univariate survival analysis was performed on the 23 genes in the gene set, and a forest plot of 17 statistically significant genes and their HR values ​​was obtained; B: LASSO regression analysis was performed on the 17 genes in the previous step, and the coefficients of 16 genes in the LASSO model were obtained; C: The prediction model obtained in the previous step was used to predict the overall survival (OS) of breast cancer (BRCA), liver cancer (LIHC), and lung cancer (LUAD and LUSC) in TCGA; D: AUC curve of the prediction model. DETAILED DESCRIPTION

[0050] The present invention is further described below in conjunction with the accompanying drawings and specific examples, but the examples do not limit the present invention in any form. Unless otherwise specified, the reagents, methods and equipment used in the present invention are conventional reagents, methods and equipment in the art.

[0051] Unless otherwise specified, the reagents and materials used in the following examples are commercially available.

[0052] Example 1

[0053] In this example, a gene set consisting of 23 genes was found through analysis of the transcriptome data of a tumor persistent drug-resistant cell model and the GEO database.

[0054] Method: As Figure 1As shown in A, tumor persistent drug-resistant cells (DTP) refer to tumor cells that are killed by drugs after being treated with drugs for a period of time, but a small number of cells still survive, gaining temporary tolerance to the drugs, and will enter a static, dormant state to resist the lethal pressure of the drugs; when the drugs are withdrawn, that is, the tumor cells are given a "drug holiday", these small numbers of static cells will slowly resume rapid proliferation and regain sensitivity to drug treatment; when the drugs are given again, most of the cells will be killed again. Therefore, based on the reversibility and drug tolerance of this tumor persistent drug-resistant cell (DTP), we established a gefitinib persistent drug-resistant cell model (DTP) of the human non-small cell lung cancer cell line PC9 and a vemurafenib / trametinib persistent drug-resistant cell model (DTP) of the human melanoma A375 cell line in our previous work, and performed transcriptome sequencing on these two DTP cells and their parental control cells (parental), and analyzed the sequencing data. The transcriptome data of the lapatinib persistent drug-resistant cell model (DTP) of the human breast cancer cell line BT474 in the GSE84896 dataset were analyzed. The log2Fc was set to ≥ 1, P ≤ 0.05, and the gene sets significantly upregulated in the three DTP cells were obtained. The intersection of the three datasets and the highly expressed gene sets in tumors from the TCGA database was taken and drawn into a Venn diagram, as shown in the figure below: Figure 1 As shown in Figure B, we finally obtained a gene set consisting of 23 genes. Then, based on this gene set consisting of 23 genes, we established a risk model for predicting persistent drug resistance and poor prognosis of tumor treatment, analyzed the relationship between this gene set consisting of 23 genes and clinical tumor survival, and evaluated the predictive efficiency of the risk model we established. The flowchart is shown in Figure 2. Figure 1 As shown in C.

[0055] Results: The 23-gene gene set we obtained was not only highly expressed in tumors, but also highly expressed in three different types of tumor DTPs.

[0056] Conclusion: This gene panel consisting of 23 genes may become a molecular marker for predicting persistent resistance to tumor treatment and clinical tumor survival.

[0057] Example 2

[0058] This example mainly explores the expression level of the gene set consisting of 23 genes obtained in Example 1 in breast cancer, liver cancer and lung adenocarcinoma and the GO analysis of the genes.

[0059] Methods: We used GEPIA2 online analysis software to analyze and compare the expression (log2TPM+1) of this 23-gene set in tumor tissues and normal tissues of breast cancer, liver cancer, and lung adenocarcinoma, and drew a heat map of gene expression. At the same time, in order to explore the physiological functions of these 23 genes, we used the DAVID online database to perform GO analysis of the genes, focusing on the BP (physiological function) of the genes.

[0060] Results and Conclusions: The heat map results are as follows Figure 2 As shown in A, GO analysis Figure 2 As shown in B, most of the genes in this gene set consisting of 23 genes are significantly upregulated in tumors. At the same time, GO analysis shows that these 23 genes are mainly related to neuronal cell death, apoptosis pathways and programmed cell death.

[0061] Example 3

[0062] This example mainly explores the gene set consisting of 23 genes obtained in Example 1 and the KM survival analysis and survival correlation map of breast cancer, liver cancer and lung adenocarcinoma patients.

[0063] Methods: The relationship between these 23 genes and overall survival (OS) of breast cancer (BRCA), liver cancer (LIHC) and lung adenocarcinoma (LUAD) was analyzed using Kaplan Meier plotter online survival analysis. The GEPIA2 online database was used to analyze and compare the survival effects of these 23 genes in breast cancer (BRCA), liver cancer (LIHC) and lung adenocarcinoma (LUAD) to obtain a heat map.

[0064] Results: Figure 3 As shown in the results, high expression of genes ALPP, EDIL3, GRIK2, LMO1, NTF4 and TMEM132B was significantly associated with shortened overall survival (OS) of breast cancer. Figure 3 A); High expression of genes ARHGEF37, BEST3, GRIA2, PSORS1C1, IL13, LY96 and MAP2 was significantly associated with shortened clinical overall survival (OS) of liver cancer ( Figure 3 B); High expression of genes BEST3, LMO1, NOV, SNAI2 and TMEM52B is significantly associated with shortened clinical overall survival (OS) of lung adenocarcinoma ( Figure 3 C) By Figure 3 As shown in the survival map of D, multiple genes have higher clinical risk values ​​in breast cancer, liver cancer, and lung adenocarcinoma, and are associated with shortened clinical survival and poor prognosis of the three cancers.

[0065] Conclusion: This gene set consisting of 23 genes is significantly associated with clinical survival and poor prognosis of breast cancer, liver cancer and lung adenocarcinoma. A genetic risk assessment model can be established to predict persistent drug resistance and poor clinical survival in tumor treatment.

[0066] Example 4

[0067] This example mainly explores a gene set consisting of 23 genes to establish a gene risk assessment model to predict clinical survival of tumors.

[0068] Methods: We downloaded the survival data of breast cancer (BRCA), liver cancer (LIHC), and lung cancer (LUAD and LUSC) from the TCGA database, and used the R package to perform univariate survival analysis on this gene set consisting of 23 genes. We selected 17 genes with statistical significance at P≤0.05, and drew a forest plot based on the HR value of each gene. Then, in order to evaluate whether these 17 genes can predict the survival analysis of different tumors, we used the survival data of breast cancer (BRCA), liver cancer (LIHC), and lung cancer (LUAD and LUSC) in the TCGA database as the training set, and used the LASSO regression model to identify the genes in the training set to predict clinical overall survival (OS); and used a stepwise multi-Cox regression model to identify the predictive ability of each gene, and calculated the risk coefficient of each gene. Then we used the risk assessment model of 16 genes established in the previous step to predict the survival analysis of breast cancer (BRCA), liver cancer (LIHC), and lung cancer (LUAD and LUSC) in the TCGA database, as well as the ROC curve of the prediction model.

[0069] Results: Figure 4 As shown in A, we used univariate survival analysis to select 17 statistically significant genes from the gene set consisting of 23 genes, and the forest plot of their HR values; Figure 4As shown in B, the 17 genes in the previous step were further subjected to LASSO regression analysis to obtain the coefficients of the 16 genes in the LASSO model; LASSO regression model analysis and multivariate regression model analysis were then performed to finally obtain a risk model consisting of 16 genes, and the risk value was calculated: Risk value = (-0.1150 × ARHGEF37) + (0.0470 × MAP2) + (-0.0491 × TSPAN10) + (-0.0176 × FA M71F2)+(0.0240×PSORS1C1)+(-0.0813×FOXD4)+(-0.0058×GDAP1L1)+(-0.0587×TMEM52B)+(-0.0122×SCIN)+(-0.0300×KIF26B)+(0.0201×NOV)+(-0.0349×GRIA2)+(0.0734×LY96)+(0.0877×ALPP)+(0.0547×BEST3)+(0.0395×IL13), each gene represents the gene expression level. We used the risk prediction model to predict the overall survival (OS) of breast cancer (BRCA), liver cancer (LIHC), and lung cancer (LUAD and LUSC) in TCGA. The results are as follows Figure 4 As shown in C, the higher the risk value, the shorter the overall survival. At the same time, the AUC curve of the prediction model is as follows Figure 4 As shown in D, the areas under the curve (AUC) of the prediction model at 5, 10, 15 and 20 years were 0.655, 0.652, 0.728 and 0.697, respectively, proving that our prediction model has good accuracy in predicting tumor overall survival (OS).

[0070] Conclusion: The gene set consisting of 23 genes obtained and the risk prediction model established based on it can effectively predict the clinical overall survival of different tumors and serve as a predictive indicator of persistent drug resistance in tumor treatment and poor prognosis of clinical survival.

Claims

1. A combination of markers for predicting persistent drug resistance and clinical survival in tumor treatment, characterized in that: The marker combination consists of at least 16 of the 23 genes EDIL3, ARHGEF37, MAP2, TMEM132B, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, LMO1, NOV, GRIK2, GRIA2, SNAI2, LY96, ALPP, IGFL1, BEST3, IL13 or NTF4.

2. The marker combination according to claim 1, characterized in that: The marker combination consists of ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13.

3. Use of the marker combination according to claim 1 or 2 in constructing a risk assessment model or system for predicting persistent drug resistance and clinical survival in tumor treatment.

4. A risk assessment model for predicting persistent drug resistance and clinical survival in tumor treatment, characterized in that: The risk score of tumor patients was obtained by the risk value formula consisting of the sum of the expression levels of markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13 and the product of the corresponding coefficients. Risk value = (-0.1150×ARHGEF37) + (0.0470×MAP2) + (-0.0491×TSPAN10) + (-0.0176×FAM71F2) + (0.0240×PSORS1C1) + (-0.0813×FOXD4) + (-0.0058×GDAP1L1) + (-0.0587×TME M52B)+(-0.0122×SCIN)+(-0.0300×KIF26B)+(0.0201×NOV)+(-0.0349×GRIA2)+(0.0734×LY96)+(0.0877×ALPP)+(0.0547×BEST3)+(0.0395×IL13).

5. A risk scoring system for predicting persistent drug resistance and clinical survival in tumor treatment, characterized in that: The method comprises a data input module, a model calculation module and a result output module. The model calculation module calculates the product of the gene expression amount of markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13 and the corresponding coefficient through a scoring formula. The risk value formula composed of the sum of the above is used to obtain the risk score of tumor patients. Risk value = (-0.1150×ARHGEF37)+(0.0470×MAP2)+(-0.0491×TSPAN10)+(-0.0176×FAM71F2)+(0.0240×PSORS1C1)+(-0.0813×FOXD4)+(-0.0058×GDAP1L1)+(-0.0587×TME M52B)+(-0.0122×SCIN)+(-0.0300×KIF26B)+(0.0201×NOV)+(-0.0349×GRIA2)+(0.0734×LY96)+(0.0877×ALPP)+(0.0547×BEST3)+(0.0395×IL13).

6. The system according to claim 5, characterized in that: The data input module is used to input the gene expression levels of patient markers ARHGEF37, MAP2, TSPAN10, FAM71F2, PSORS1C1, FOXD4, GDAP1L1, TMEM52B, SCIN, KIF26B, NOV, GRIA2, LY96, ALPP, BEST3 and IL13 into the model calculation module.

7. The system according to claim 5, characterized in that: The model calculation module also includes comparing the risk value obtained in the model calculation module with a threshold, where the threshold is set to 0. When used to predict persistent drug resistance in tumor treatment, if the score is greater than the threshold, it is a high-risk group, and if the score is less than the threshold, it is a low-risk group; when used to predict clinical survival, if the score is greater than the threshold, the survival period is shorter, and if the score is less than the threshold, the survival period is longer.

8. Use of the marker according to claim 1 or 2 in the preparation of a kit for predicting persistent drug resistance in tumor treatment and clinical survival or in the preparation of a drug for persistent drug resistance in anti-tumor treatment.

9. The use according to claim 3 or 8, characterized in that: The drug used in the treatment of persistent drug resistance in tumors is any one of gefitinib, vemurafenib, trametinib or lapatinib; and the tumor is any one of non-small cell lung cancer, melanoma or breast cancer.

10. A kit for predicting persistent drug resistance and clinical survival in tumor treatment, characterized in that: Contains a reagent for detecting the expression level of the marker combination gene according to claim 1 or 2.

Images