FOLFOX scheme drug effect prediction method and system for digestive tract tumor EBV negative patient

Abstract

The invention relates to the technical field of digestive tract tumor drug treatment and drug effect prediction, and discloses an FOLFOX scheme drug effect prediction method and system for digestive tract tumor EBV negative patients, and the method comprises the steps: obtaining samples of a plurality of PDTX models, and carrying out WES sequencing and RNA sequencing on each sample to obtain corresponding DNA sequencing data and RNA sequencing data; wherein the PDTX model is a PDTX model of mice of the same strain; carrying out EBV detection on the samples of the plurality of PDTX models, dividing the samples of the plurality of PDTX models into EBV infection negative samples and EBV infection positive samples, forming an EBV infection negative sample group by using the EBV infection negative samples, and taking the EBV infection negative sample group as a sample basis for establishment of an FOLFOX scheme efficacy prediction model and FOLFOX scheme efficacy prediction; performing differential gene analysis and gene screening based on the EBV infection negative sample group, and then determining a modeling sample set for establishing an FOLFOX scheme efficacy prediction model; establishing an FOLFOX scheme efficacy prediction model based on the modeling sample set; and the prediction quality of the FOLFOX scheme drug effect prediction model is verified based on a clinical sample.

Description

Technical Field

[0001] The present invention relates to the technical field of drug treatment and drug efficacy prediction for digestive tract tumors, and in particular to a drug efficacy prediction method and system for a FOLFOX regimen for EBV-negative patients with digestive tract tumors. Background Art

[0002] According to the latest global cancer data in 2020, there were 19.29 million new cancer cases in the world in 2020, of which 4.568 million were new cancer cases in China, accounting for 23.7% of the world. The incidence and mortality rates of digestive tract tumors in my country are both high. Among the top 10 high-incidence tumors, digestive tract tumors account for 5, namely colorectal cancer, gastric cancer, liver cancer, esophageal cancer, and pancreatic cancer. It is worth noting that among all cancers diagnosed in China in 2020, 41.5% were digestive system cancers; in terms of mortality, digestive tract tumors in the top 10 included liver cancer, gastric cancer, esophageal cancer, colorectal cancer, and pancreatic cancer, accounting for nearly half. Among the new cancer cases in China in 2020, there were 560,000 colorectal cancer, 480,000 gastric cancer, 410,000 liver cancer, 320,000 esophageal cancer, 120,000 pancreatic cancer, and 30,000 bile duct / cyst cancer. It accounts for 42% of the new cases, and the above tumors can all be treated with the FOLFOX regimen.

[0003] The FOLFOX regimen is a classic tumor treatment regimen that includes 5-fluorouracil (5-FU), leucovorin, and oxaliplatin. Currently, the FOLFOX regimen is still widely used, especially in postoperative adjuvant therapy and initial treatment of advanced cancer. Studies have shown that this regimen can effectively prolong the patient's survival and is relatively well tolerated. However, some patients may experience side effects such as neurotoxicity, requiring individualized precision medication. New research is also exploring the possibility of combining with immunotherapy and targeted therapy to further improve efficacy.

[0004] The FOLFOX regimen is mainly used to treat colorectal cancer, especially in postoperative adjuvant therapy and metastatic disease. In addition, it is also used in the treatment of tumors such as esophageal cancer, gastric cancer, certain types of pancreatic cancer, and bile duct / cystic cancer. The combination of drugs in this regimen helps to inhibit the growth and spread of cancer cells. The FOLFOX regimen has shown good efficacy in the treatment of colorectal cancer and can significantly prolong disease-free survival and overall survival. In the treatment of esophageal cancer and gastric cancer, although the efficacy is relatively weak, it has also shown certain effects, especially when combined with other treatments. Overall, its efficacy varies depending on individual patient differences, cancer type and stage, so a treatment plan needs to be formulated according to specific circumstances.

[0005] Common side effects of the FOLFOX regimen include:

[0006] (1) Gastrointestinal reactions: nausea, vomiting, diarrhea or constipation.

[0007] (2) Hematological effects: reduction in white blood cells, platelets and red blood cells, which may lead to an increased risk of infection and anemia.

[0008] (3) Neurotoxicity: hand-foot syndrome (such as numbness and tingling in hands and feet), usually associated with oxaliplatin.

[0009] (4) Fatigue: Patients may feel significant fatigue.

[0010] (5) Oral problems: oral ulcers or dry mouth.

[0011] The severity of side effects varies from person to person, and doctors will monitor and manage according to the specific situation of the patient.

[0012] The efficacy and side effects of the FOLFOX regimen are indeed affected by differences in the physical constitution of patients, including the following aspects:

[0013] (1) Age: Elderly patients may be more prone to side effects, especially neurotoxicity.

[0014] (2) Baseline health status: Patients with other diseases or comorbidities may have poorer tolerance.

[0015] (3) Genetic factors: Certain gene mutations may affect drug metabolism, resulting in different efficacy and side effects.

[0016] (4) Nutritional status: Malnourished patients may experience more severe side effects during treatment.

[0017] Therefore, doctors usually adjust the treatment plan according to the specific situation of the patient to ensure the best effect and the least side effects.

[0018] The FOLFOX regimen is widely used in the treatment of tumors such as colorectal cancer, esophageal cancer, gastric cancer, liver cancer, pancreatic cancer and bile duct / cystic cancer. However, the efficacy of the FOLFOX regimen is affected by differences in the physical constitution of patients. Due to the differences among tumor patients themselves, the efficacy of the FOLFOX regimen treatment plan also varies from person to person. In order to reduce the side effects of patients with poor efficacy using the FOLFOX regimen treatment plan, affect the treatment of other plans and the loss of money. Therefore, in the prior art, there have emerged technical solutions for predicting the efficacy of the FOLFOX regimen for patients through algorithms such as neural networks and random forests using clinical indicators, gene mutations, gene expression and methylation indicators in a single cancer type.

[0019] For example:

[0020] (1) The literature "Construction of an Artificial Neural Network Efficacy Prediction Model for First-line Treatment of Metastatic Colorectal Cancer with the FOLFOX Regimen" does not consider EBV virus infection in its technical solution. After EBV virus infection, it will affect the gene expression situation and the accuracy of the model. Neural networks have high requirements for the number of training samples. In addition, this solution is only applicable to one type of cancer and cannot predict other cancers.

[0021] (2) Ahn SJ, Kim JH, Park SJ, Han JK. Prediction of the therapeutic response after FOLFOX and FOLFIRI treatment for patients with liver metastasis from colorectal cancer using computerized CT texture analysis. Eur J Radiol. 2016 Oct;85(10):1867-1874. doi:10.1016 / j.ejrad.2016.08.014 IF:3.2 Q1. Epub 2016 Aug 23. PMID:27666629. The technical solution uses clinical feature CT textures for colorectal cancer samples and uses logistic regression modeling to predict the efficacy of the FOLFOX regimen. However, this technical solution is only suitable for colorectal cancer and cannot predict other cancers.

[0022] However, the FOLFOX chemotherapy regimen is a multi-target and multi-signaling pathway chemotherapeutic drug. Changes in the functions of multiple genes in the signaling pathway will directly affect the change in the efficacy of the FOLFOX regimen. Existing technologies find locus changes through DNA, but locus changes do not necessarily change gene functions. Whether each pathway in tumor cells is affected still requires attention to the true gene expression situation. Therefore, the prediction accuracy is relatively low. Summary of the Invention

[0023] To solve the problems existing in the prior art, the present invention provides the following technical solution: A method and system for predicting the efficacy of the FOLFOX regimen for EBV (Epstein-Barr Virus)-negative patients with digestive tract tumors, using the PDTX model (Patient-derived tumor xenografts) of the same strain of mice to obtain pure FOLFOX regimen efficacy data for modeling, and finally using the data of clinical staff to verify the model. Among them, the data of digestive tract tumor samples such as gastric cancer, colorectal cancer, liver cancer, pancreatic cancer, esophageal cancer, and bile duct / cystic cancer are used for modeling, so that the model is not limited to a single type of cancer, expanding the applicability of the method and system.

[0024] In one aspect, the present invention provides a method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors, comprising:

[0025] S1, obtaining samples of multiple PDTX models and performing RNA sequencing on the samples of each PDTX model to obtain RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX models are PDTX models of the same strain of mice;

[0026] S2, performing EBV detection on samples of multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV infection-negative samples and EBV infection-positive samples, and after removing the EBV infection-positive samples, the EBV infection-negative samples constitute an EBV infection-negative sample group as the sample basis for establishing a FOLFOX regimen efficacy prediction model and FOLFOX regimen efficacy prediction;

[0027] S3, determining a modeling sample set for establishing a FOLFOX regimen efficacy prediction model after differential gene analysis and gene screening based on the EBV infection negative sample group;

[0028] S4, establishing a drug efficacy prediction model for the FOLFOX regimen based on the modeling sample set;

[0029] S5. Verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

[0030] Preferably, samples of multiple PDTX models are obtained and WES sequencing is performed on each sample of the PDTX model to obtain DNA sequencing data corresponding to the samples of the multiple PDTX models; wherein: S1 includes:

[0031] S11, performing WES sequencing on each sample of the PDTX model; the WES sequencing uses the lllumina NovaSeqV6000 sequencer, uses whole exome sequencing: Agilent V6 is used as the sequencing reagent, and uses the capture sequencing sequencing method; the sequencing data output is 25G, and the sequencing data requirements are: base quality Q30>90%, GC content is 50±5%, the average depth of the sequencing capture area is 200X-250X, and the coverage requirements are that the capture area coverage is above 95%, and the 10X coverage is above 80%;

[0032] S12, RNA sequencing was performed on each sample of the PDTX model; the RNA sequencing used an Illumina NovaSeq V6000 sequencer, VAHTS mRNA Capture Beads + mRNA VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina as sequencing reagents, and a capture sequencing method was used; the sequencing data output was 10G, and the sequencing data requirements were: base quality Q30> 90%, and GC content was 50±5%.

[0033] Preferably, S2 includes:

[0034] S21, obtaining the drug efficacy data and cancer type information data of the samples of the multiple PDTX models;

[0035] S22, based on the cluster analysis of the drug efficacy data and the cancer information data, the samples of the multiple PDTX models are regionally classified; the regional classification includes enriched gene regions and non-enriched gene regions;

[0036] S23, performing gene enrichment analysis on all genes in the enriched gene region to obtain the EBV pathway;

[0037] S24, performing compatibility verification on all genes in the enriched gene region and the EBV pathway, wherein the EBV genes that pass the compatibility verification are EBV genes that match the EBV pathway;

[0038] S25, drawing a heat map for the samples of the multiple PDTX models based on the EBV gene, and distinguishing the EBV-infected positive samples from the EBV-infected negative samples by using the heat map to obtain a first distinction result;

[0039] S26, performing statistical analysis on the DNA sequencing data and RNA sequencing data of the samples of the multiple PDTX models to obtain the sequence number of EBV virus; determining whether each sample in the samples of the multiple PDTX models is infected with EBV virus one by one based on the sequence number of EBV virus, thereby obtaining a second distinction result for distinguishing EBV-infected positive samples from EBV-infected negative samples;

[0040] S27, determining whether the first distinction result and the second distinction result are consistent; if they are consistent, determining the final EBV infection positive sample and EBV infection negative sample; if they are inconsistent, repeating S21-S26 and performing different regional classifications until a consistent distinction result is obtained.

[0041] Preferably, S3 includes:

[0042] S31, dividing the samples in the EBV infection negative sample group into FOLFOX regimen efficacy positive samples and FOLFOX regimen efficacy negative samples;

[0043] S32, based on determining that the sample expression levels of the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples do not conform to a normal distribution, performing a Wilcoxon rank sum test on the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples to obtain multiple differentially expressed genes;

[0044] S33, performing pathway enrichment analysis on the multiple differentially expressed genes, and obtaining the most significant pathway with the highest correlation with the efficacy of the FOLFOX regimen based on the analysis results;

[0045] S34, performing differential gene analysis on the multiple differential genes, drawing a heat map for the expression amounts of the top M most significant differential genes in the differential gene analysis results, performing PCA sample clustering, and then screening N efficacy classification genes, wherein the efficacy classification genes have a clear classification effect on the FOLFOX regimen efficacy of the multiple FOLFOX regimen efficacy-negative samples;

[0046] S35, removing the collinear genes among the top M most significantly differentially expressed genes, verifying and confirming that the N pharmacodynamic classification genes are partially or completely on the most significant pathway, and determining a modeling sample set for establishing the pharmacodynamic prediction model of the FOLFOX regimen based on the most significantly differentially expressed genes after removing the collinear genes, the N pharmacodynamic classification genes and combining gene functions.

[0047] Preferably, S4 includes:

[0048] S41, respectively using random forest, SVM, Lasso regression and logistic regression modeling methods to establish the FOLFOX regimen efficacy prediction model, thereby obtaining a first FOLFOX regimen efficacy prediction model, a second FOLFOX regimen efficacy prediction model, a third FOLFOX regimen efficacy prediction model and a fourth FOLFOX regimen efficacy prediction model;

[0049] S42, testing the performance of the first FOLFOX regimen pharmacodynamic prediction model, the second FOLFOX regimen pharmacodynamic prediction model, the third FOLFOX regimen pharmacodynamic prediction model and the fourth FOLFOX regimen pharmacodynamic prediction model based on multiple test indicators; selecting the modeling method corresponding to the FOLFOX regimen pharmacodynamic prediction model with the best performance to establish the FOLFOX regimen pharmacodynamic prediction model.

[0050] Preferably, the multiple test indicators include:

[0051] (1) Sensitivity or true positive rate (TPR) = TP / (TP+FN);

[0052] (2) precision or positive predictive value (PPV) = TP / (TP + FP);

[0053] (3) True negative rate (TNR) = TN / (FP+TN);

[0054] (4) Negative predictive value (NPV) = TN / (FN + TN);

[0055] (5) Accuracy = (TP + TN) / (TP + TN + FP + FN).

[0056] Preferably, S5 includes:

[0057] The clinical samples were input into the FOLFOX regimen efficacy prediction model for prediction, and the prediction quality was determined based on the prediction accuracy, true negative rate, true positive rate, negative predictive value, positive predictive value and AUC values.

[0058] The second aspect of the present invention is to provide a FOLFOX regimen efficacy prediction system for EBV-negative patients with digestive tract tumors, which is used to implement the method of the first aspect, comprising:

[0059] A sample and sequencing data acquisition module, used to acquire samples of multiple PDTX models and perform WES sequencing and RNA sequencing on the samples of each PDTX model to acquire DNA sequencing data and RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX models are PDTX models of the same strain of mice;

[0060] An EBV monitoring module, used to perform EBV detection on samples of multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV infection-negative samples and EBV infection-positive samples, and after removing the EBV infection-positive samples, the EBV infection-negative samples constitute an EBV infection-negative sample group as the sample basis for establishing the FOLFOX regimen efficacy prediction model and the FOLFOX regimen efficacy prediction;

[0061] A differential gene analysis and gene screening module, used to determine a modeling sample set for establishing a FOLFOX regimen efficacy prediction model after differential gene analysis and gene screening based on the EBV infection negative sample group;

[0062] A drug efficacy prediction model building module, used to build the drug efficacy prediction model of the FOLFOX regimen based on the modeling sample set;

[0063] The prediction quality verification module is used to verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

[0064] A third aspect of the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores a plurality of instructions, and the processor is configured to read the instructions and execute the method described in the first aspect.

[0065] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be read by a processor and execute the method described in the first aspect.

[0066] The method, system and electronic device provided by the present invention have the following beneficial effects:

[0067] (1) The efficacy of the FOLFOX regimen using PDTX was verified using clinical samples to reduce the impact of human differences and living environment on efficacy.

[0068] (2) Detect EBV before use, and model and predict EBV-negative samples to avoid the interference of EBV on drug efficacy.

[0069] (3) The model is applicable to more types of cancer and is not limited to one type of cancer. The model covers gastric cancer, intestinal cancer, liver cancer, pancreatic cancer, esophageal cancer, and bile duct / cystic cancer.

[0070] (4) It can screen and identify patients for whom treatment is ineffective, thereby reducing the side effects and money wasted on ineffective patients caused by drugs.

[0071] (5) Only 10 gene parameters are required, the number of genes tested is small, and the testing workload is reduced.

[0072] (6) Use four model algorithms to build models simultaneously and select the random forest algorithm with the best performance for modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a flow chart of the method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors according to an embodiment of the present invention;

[0074] Figure 2 This is a flow chart of step S1 of the method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors according to an embodiment of the present invention;

[0075] Figure 3 This is a flow chart of step S2 of the method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors according to an embodiment of the present invention;

[0076] Figure 4 This is a schematic diagram of the results of cluster analysis of samples of the PDTX model provided by the present invention according to the efficacy of the FOLFOX regimen and the type of cancer;

[0077] Figure 5 A heat map drawn for the EBV gene pair samples provided by the present invention;

[0078] Figure 6 Flow chart of step S3 of the method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors according to an embodiment of the present invention;

[0079] Figure 7 After performing pathway enrichment analysis on the multiple differentially expressed genes provided by the present invention, based on the analysis results;

[0080] Fig. 8A and Figure 8B All gene heat maps and PCA maps for obtaining drug efficacy classification genes provided by the present invention;

[0081] Fig. 9 The expression profiles of the 10 genes provided by the present invention in negative and positive samples;

[0082] Fig. 10A and Fig. 10B They are respectively the gene heat map and PCA map of the determined drug efficacy classification genes provided by the present invention;

[0083] Fig.11 This is a flow chart of step S4 of the method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors according to an embodiment of the present invention;

[0084] Fig.12 This is a diagram of the architecture of the FOLFOX regimen efficacy prediction system for EBV-negative patients with digestive tract tumors provided by the present invention;

[0085] Fig.13 A schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0086] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0087] In addition to biomarkers, specific molecular features, such as gene mutations, can also be incorporated into the prognostic model for patients with early-stage liver cancer after surgery. Gene mutations, especially driver gene mutations, are an important cause of cancer development. Gene mutations are also important targets for cancer treatment and serve as important markers for prognosis.

[0088] Embodiment 1

[0089] like Figure 1 As shown, this embodiment provides a method for predicting the efficacy of the FOLFOX regimen for EBV-negative patients with digestive tract tumors, comprising:

[0090] S1, obtain samples of multiple PDTX models and perform RNA sequencing on the samples of each PDTX model to obtain RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX model is a PDTX model of the same strain of mice; of course, step S1 can also be implemented as follows: obtain samples of multiple PDTX models and perform WES sequencing and RNA sequencing on the samples of each PDTX model to obtain DNA sequencing data and RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX model is a PDTX model of the same strain of mice; that is, step S1 is implemented by considering both DNA sequencing data and RNA sequencing data at the same time, wherein the DNA sequencing data is used as verification data for the RNA sequencing data, thereby ensuring the accuracy of the basic data for analysis.

[0091] Using the FOLFOX regimen efficacy obtained from the PDTX model as the basis for analysis can reduce the defects of unstable efficacy caused by differences in human bodies and living environments, which affects the accuracy and reliability of predictions. WES (whole exome sequencing) uses sequence capture technology to capture and enrich DNA in the exon region of the whole genome for high-throughput sequencing.

[0092] like Figure 2 As shown, as a preferred embodiment, the S1 includes:

[0093] S11, performing WES sequencing on each sample of the PDTX model; the WES sequencing uses an Illumina NovaSeqV6000 sequencer, uses whole exome sequencing: Agilent V6 is used as a sequencing reagent, and uses a capture sequencing sequencing method; the sequencing data output is 25G, and the sequencing data requirements are: base quality Q30>90%, GC content is 50±5%, the average depth of the sequencing capture region is 200X-250X, and the coverage requirements are that the capture region coverage is above 95%, and the 10X coverage is above 80%;

[0094] S12, RNA sequencing was performed on each sample of the PDTX model; the RNA sequencing used an Illumina NovaSeq V6000 sequencer, VAHTS mRNA Capture Beads + mRNA VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina as sequencing reagents, and a capture sequencing method was used; the sequencing data output was 10G, and the sequencing data requirements were: base quality Q30> 90%, and GC content was 50±5%.

[0095] In this embodiment, S1 involves RNA data and DNA data of samples of 60 PDTX models, and a total of 59 clinical RNA data are obtained.

[0096] S2, performing EBV detection on samples of multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV infection-negative samples and EBV infection-positive samples, and after removing the EBV infection-positive samples, the EBV infection-negative samples constitute an EBV infection-negative sample group as the sample basis for establishing a FOLFOX regimen efficacy prediction model and FOLFOX regimen efficacy prediction;

[0097] Implementation principle of S2: Samples infected with EBV are not suitable for modeling together with negative samples. This is because after infection with EBV, the expression of genes in the sample will be affected. Therefore, if the efficacy of the FOLFOX regimen is studied based on the transcript level, it is necessary to consider whether the patient is infected with EBV. The subsequent modeling of the present invention based on the EBV-infected negative samples obtained after EBV detection belongs to this situation, so the EBV-infected positive samples are eliminated.

[0098] like Figure 3 As shown, as a preferred embodiment, the S2 includes:

[0099] S21, obtaining the drug efficacy data and cancer type information data of the samples of the multiple PDTX models;

[0100] S22, based on the cluster analysis of the drug efficacy data and the cancer information data, the samples of the multiple PDTX models are regionally classified; the regional classification includes enriched gene regions and non-enriched gene regions;

[0101] S23, performing gene enrichment analysis on all genes in the enriched gene region to obtain the EBV pathway;

[0102] S24, performing compatibility verification on all genes in the enriched gene region and the EBV pathway, wherein the EBV genes that pass the compatibility verification are EBV genes that match the EBV pathway;

[0103] S25, drawing a heat map for the samples of the multiple PDTX models based on the EBV gene, and distinguishing the EBV-infected positive samples from the EBV-infected negative samples by using the heat map to obtain a first distinction result;

[0104] S26, performing statistical analysis on the DNA sequencing data and RNA sequencing data of the samples of the multiple PDTX models to obtain the sequence number of EBV virus; determining whether each sample in the samples of the multiple PDTX models is infected with EBV virus one by one based on the sequence number of EBV virus, thereby obtaining a second distinction result for distinguishing EBV-infected positive samples from EBV-infected negative samples;

[0105] S27, determining whether the first distinction result and the second distinction result are consistent; if they are consistent, determining the final EBV infection positive sample and EBV infection negative sample; if they are inconsistent, repeating S21-S26 and performing different regional classifications until a consistent distinction result is obtained.

[0106] In this embodiment, specifically, 60 samples of the PDTX model were clustered according to the efficacy of the FOLFOX regimen and the type of cancer. The 60 samples can be divided into two categories, namely, the enriched gene region (upper left corner) and the non-enriched gene region (other regions except the upper left corner). Figure 4 shown.

[0107] right Figure 4 Gene enrichment analysis was continued for the genes in the enriched gene region in the upper left corner, and Epstein-Barrvirus infection was obtained as the most significant pathway, namely the EBV pathway. A total of 27 genes in the enriched gene region matched the EBV pathway. The 27 genes are as follows: CDKN1A, CD40, TNFAIP3, PIK3CD, ICAM1, HLA-DMA, HLA-DMB, CCND2, BLNK, PLCG2, E2F2, JAK3, LYN, ENTPD1, ​​SYK, HLA-B, TAP2, TAP1, TYK2, HLA-A, RUNX3, NFKB2, HLA-E, TRAF3, IRF7, FAS, NFKBIE. A heat map of 60 samples was made based on the expression of 27 EBV genes, as shown in the figure below. Figure 5 As shown, EBV infection-positive samples and EBV infection-negative samples are clearly distinguished.

[0108] By analyzing the DNA and RNA sequencing data of 60 samples, the number of EBV virus sequences was counted as shown in Table 1 to determine whether the samples were infected with EBV virus. The results were consistent with the cluster analysis samples, and finally the EBV infection-negative samples were determined.

[0109] Table 1

[0110]

[0111]

[0112] In this embodiment, 18 samples were detected to be infected by EBV virus, which was not used for modeling later. After EBV infection, the expression of genes will be affected. To study the efficacy of the FOLFOX regimen from the transcript level, it is necessary to consider whether the patient is infected by EBV. This embodiment is based on EBV detection negative samples for subsequent steps.

[0113] S3, determining a modeling sample set for establishing a FOLFOX regimen efficacy prediction model after differential gene analysis and gene screening based on the EBV infection negative sample group;

[0114] like Figure 6 As shown, as a preferred implementation, S3 includes:

[0115] S31, dividing the samples in the EBV infection negative sample group into FOLFOX regimen efficacy positive samples and FOLFOX regimen efficacy negative samples;

[0116] In this example, after EBV testing, there were 42 EBV-negative samples, of which 27 samples were positive for FOLFOX efficacy and 15 were negative for FOLFOX efficacy, covering gastric cancer, intestinal cancer, liver cancer, pancreatic cancer, esophageal cancer and bile duct / cystic cancer.

[0117] S32, based on determining that the sample expression levels of the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples do not conform to the normal distribution, perform a Wilcoxon rank sum test on the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples to obtain multiple differentially expressed genes.

[0118] In this example, the two groups of samples were subjected to Wilcoxon rank sum test analysis, and a total of 1376 differentially expressed genes were screened (p<0.05), as shown in Table 2 below.

[0119] Table 2

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133] S33, performing pathway enrichment analysis on the multiple differentially expressed genes, and obtaining the most significant pathway with the highest correlation with the efficacy of the FOLFOX regimen based on the analysis results;

[0134] In this example, pathway enrichment analysis was performed on 1376 genes, and the analysis results are as follows: Figure 7 shown.

[0135] The most significant pathway is the MAPK signaling pathway, which is directly related to the efficacy of the FOLFOX regimen.

[0136] MAPK signaling pathway: The mitogen-activated protein kinase (MAPK) signaling pathway is one of the important pathways in the signal transduction network of eukaryotic organisms. It is a key signaling pathway for cell proliferation, differentiation, apoptosis, and stress response under normal and pathological conditions.

[0137] S34, performing differential gene analysis on the multiple differential genes, drawing a heat map for the expression levels of the top M most significant differential genes in the differential gene analysis results, and performing PCA sample clustering to screen N efficacy classification genes, wherein the efficacy classification genes have a clear classification effect on the FOLFOX regimen efficacy of the multiple FOLFOX regimen efficacy-negative samples.

[0138] like Fig. 8A and Figure 8B As shown in the figure, M is 100 and N is 10, that is, there are 10 genes that have a good classification effect on the efficacy of the FOLFOX regimen of the sample. The 10 genes are: SLC39A13, GADD45G, SPSB3, SULT2B1, DUSP3, JUN, CD59, ARHGEF10L, VHL, and HERC3.

[0139] S35, removing the collinear genes among the top M most significantly differentially expressed genes, verifying and confirming that the N pharmacodynamic classification genes are partially or completely on the most significant pathway, and determining a modeling sample set for establishing the pharmacodynamic prediction model of the FOLFOX regimen based on the most significantly differentially expressed genes after removing the collinear genes, the N pharmacodynamic classification genes and combining gene functions.

[0140] Functions of the 10 genes:

[0141] SLC39A13: This gene encodes a member of the LIV-1 subfamily of the ZIP transporter family. The encoded transmembrane protein functions as a zinc transporter.

[0142] GADD45G: This gene is a member of a gene group whose transcript levels increase after exposure to stressful growth arrest conditions and treatment with DNA damaging agents. The protein encoded by this gene mediates activation of the p38 / JNK pathway via the MTK1 / MEKK4 kinases in response to environmental stress.

[0143] SPSB3: This protein is predicted to be involved in proteasome-mediated ubiquitin-dependent protein degradation. It is predicted to be located in the cytoplasm. It is predicted to be part of the SCF ubiquitin ligase complex.

[0144] SULT2B1: Sulfotransferases catalyze the sulfate conjugation of many hormones, neurotransmitters, drugs, and xenobiotics. These cytoplasmic enzymes differ in tissue distribution and substrate specificity. The structure of this gene (number and length of exons) is similar between family members. This gene is able to sulfate dehydroepiandrosterone but not 4-nitrophenol, a typical substrate of the phenolic and estrogen sulfotransferase subfamily.

[0145] DUSP3: The protein encoded by this gene belongs to the dual-specificity protein phosphatase subfamily. These phosphatases inactivate their target enzymes by dephosphorylating phosphoserine / threonine and phosphotyrosine residues. They negatively regulate the mitosis-activated protein (MAP) kinase superfamily associated with cell proliferation and differentiation (such as MAPK / ERK, SAPK / JNK, p38). Different dual-specificity phosphatase family members show different substrate specificity, tissue distribution and subcellular localization for various MAP kinases, as well as differences in the expression induction mode in response to exogenous stimuli.

[0146] JUN: This gene is considered to be a potential transforming gene of avian sarcoma virus 17. It encodes a protein that is highly similar to viral proteins and can directly interact with specific target DNA sequences to regulate gene expression. The gene is intronless and localized to the 1p32-p31 chromosomal region, which is associated with translocations and deletions in human malignancies.

[0147] CD59: This gene encodes a cell surface glycoprotein that regulates complement-mediated cell lysis and is involved in lymphocyte signaling. This protein is a potent inhibitor of the complement membrane attack complex by binding complement C8 and / or C9 during the assembly of this complex, thereby inhibiting the integration of multiple C9s, which is required for the formation of permeability pores. This protein also plays a role in the signal transduction pathway of T cell activation.

[0148] ARHGEF10L: This gene belongs to the RhoGTPase of the RhoGEF subfamily. Members of this subfamily are activated by specific guanylate exchange factors (GEFs) and participate in signal transduction. The encoded protein is distributed in the cytoplasm. Alternative splicing leads to the generation of multiple transcript variants.

[0149] VHL: This gene encodes a component of a ubiquitination complex. The encoded protein is involved in the ubiquitination and degradation of hypoxia-inducible factor (HIF), a transcription factor that regulates oxygen-dependent gene expression. In addition to oxygen-dependent gene expression, this protein is involved in many other cellular processes, including cilia formation, cytokine signaling, senescence regulation, and extracellular matrix formation.

[0150] HERC3: This gene encodes a member of the HERC ubiquitin ligase family. The encoded protein is located in the cytoplasm and binds ubiquitin through the HECT domain.

[0151] In this embodiment, the gene sample set is concentrated on the MAPK signaling pathway, wherein 3 of the 10 genes, namely DUSP3, GADD45G and JUN, are on the MAPK signaling pathway.

[0152] like Fig. 9 As shown, the expression profiles of 10 genes in negative and positive samples, and the expression of individual genes in 42 samples showed a differential trend between positive and negative FOLFOX regimen efficacy.

[0153] Ten genes had a certain role in the FOLFOX regimen efficacy classification in 42 samples. Fig. 10A Gene heatmap and Fig. 10B The PCA graph shows that there are a few samples inaccurate in the efficacy classification of the FOLFOX regimen for the 10 genes, because the expression of individual cancer genes in multi-cancer samples is inconsistent among these genes, which are then corrected using the model algorithm.

[0154] In this embodiment, 10 genes were finally selected from the EBV negative sample group as the modeling sample set for establishing the FOLFOX regimen efficacy prediction model, including 5 up-regulated genes in negative samples and 5 down-regulated genes in negative samples. The up-regulated genes were GADD45G, SPSB3, SULT2B1, JUN, and ARHGEF10L, and the down-regulated genes were SLC39A13, DUSP3, CD59, VHL, and HERC3.

[0155] S4, establishing a drug efficacy prediction model for the FOLFOX regimen based on the modeling sample set;

[0156] like Fig.11 As shown, as a preferred embodiment, the S4 includes:

[0157] S41, respectively using random forest, SVM, Lasso regression and logistic regression modeling methods to establish the FOLFOX regimen efficacy prediction model, thereby obtaining a first FOLFOX regimen efficacy prediction model, a second FOLFOX regimen efficacy prediction model, a third FOLFOX regimen efficacy prediction model and a fourth FOLFOX regimen efficacy prediction model;

[0158] S42, testing the performance of the first FOLFOX regimen pharmacodynamic prediction model, the second FOLFOX regimen pharmacodynamic prediction model, the third FOLFOX regimen pharmacodynamic prediction model and the fourth FOLFOX regimen pharmacodynamic prediction model based on multiple test indicators; selecting the modeling method corresponding to the FOLFOX regimen pharmacodynamic prediction model with the best performance to establish the FOLFOX regimen pharmacodynamic prediction model.

[0159] As a preferred implementation, the multiple test indicators include:

[0160] (1) Sensitivity or true positive rate (TPR) = TP / (TP+FN);

[0161] (2) precision or positive predictive value (PPV) = TP / (TP + FP);

[0162] (3) True negative rate (TNR) = TN / (FP+TN);

[0163] (4) Negative predictive value (NPV) = TN / (FN + TN);

[0164] (5) Accuracy = (TP + TN) / (TP + TN + FP + FN).

[0165] In this embodiment, the four modeling methods were comprehensively compared, and the Random Forest model performed best in various aspects. Table 3 shows the performance evaluation results of the FOLFOX regimen efficacy prediction model established by the four modeling methods.

[0166] Table 3

[0167] algorithm Accuracy TNR TPR NPV PPV AUC Random Forest 0.80 0.75 0.85 0.83 0.82 0.80 SVM 0.73 0.65 0.81 0.73 0.78 0.73 Lasso 0.75 0.69 0.81 0.75 0.78 0.75 Logistic 0.64 0.62 0.67 0.62 0.69 0.64

[0168] Random forest is a classifier that contains many decision trees. It can be used to handle classification and regression problems, as well as dimensionality reduction problems. It is also very tolerant to outliers and noise, and has better prediction and classification performance than decision trees.

[0169] The establishment process of Random Forest:

[0170] Each decision tree is built according to the following algorithm:

[0171] 1. Use N to represent the number of training cases (samples), and M to represent the number of features.

[0172] 2. Input the number of features m, which is used to determine the decision result of a node on the decision tree: m should be much smaller than M.

[0173] 3. Take samples N times from N training cases (samples) with replacement to form a training set (bootstrap sampling), and use the unsampled cases (samples) for prediction to evaluate their errors.

[0174] 4. For each node, randomly select m features. The decision of each node in the decision tree is based on these features. Based on these m features, calculate the best split method.

[0175] 5. Each tree will grow completely without pruning, which may be adopted after building a normal tree classifier).

[0176] S5. Verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

[0177] In this example, 59 clinical samples were used to verify the prediction quality of the FOLFOX regimen efficacy prediction model, as shown in Table 4.

[0178] Table 4

[0179]

[0180]

[0181]

[0182] Thus, the values ​​of accuracy, true negative rate, true positive rate, negative predictive value, positive predictive value and AUC are obtained, as shown in Table 5. AUC (Area Under Curve) is a commonly used performance indicator used to evaluate the performance of classification models. In machine learning, AUC is usually used to evaluate the prediction quality of binary classification models (such as logistic regression, support vector machine, etc.), and the prediction quality is evaluated here.

[0183] Table 5

[0184]

[0185] Embodiment 2

[0186] like Fig.12 As shown, this embodiment provides a FOLFOX regimen efficacy prediction system for EBV-negative patients with digestive tract tumors, which is used to implement the method of the first aspect, including: a sample and sequencing data acquisition module 101, which is used to obtain samples of multiple PDTX models and perform WES sequencing and RNA sequencing on the samples of each PDTX model to obtain DNA sequencing data and RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX model is a PDTX model of the same strain of mice; an EBV monitoring module 102, which is used to perform EBV detection on the samples of the multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV-negative samples and EBV-positive samples. The EBV-infected positive samples are removed and the EBV-infected negative samples are used to form an EBV-infected negative sample group as the sample basis for establishing the FOLFOX regimen efficacy prediction model and the FOLFOX regimen efficacy prediction; a differential gene analysis and gene screening module 103 is used to determine the modeling sample set for establishing the FOLFOX regimen efficacy prediction model after performing differential gene analysis and gene screening based on the EBV-infected negative sample group; an efficacy prediction model establishment module 104 is used to establish the FOLFOX regimen efficacy prediction model based on the modeling sample set; a prediction quality verification module 105 is used to verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

[0187] The present invention also provides a memory storing a plurality of instructions, wherein the instructions are used to implement the method as in the first embodiment.

[0188] like Fig.13 As shown, the present invention further provides an electronic device, including a processor 301 and a memory 302 connected to the processor 301. The memory 302 stores a plurality of instructions, which can be loaded and executed by the processor, so that the processor can execute the method as in the first embodiment.

[0189] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors, characterized in that: include: S1, obtaining samples of multiple PDTX models and performing RNA sequencing on the samples of each PDTX model to obtain RNA sequencing data corresponding to the samples of multiple PDTX models; wherein the PDTX models are PDTX models of the same strain of mice; S2, performing EBV detection on samples of multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV infection-negative samples and EBV infection-positive samples, and after removing the EBV infection-positive samples, the EBV infection-negative samples constitute an EBV infection-negative sample group as the sample basis for establishing a FOLFOX regimen efficacy prediction model and FOLFOX regimen efficacy prediction; S3, determining a modeling sample set for establishing a FOLFOX regimen efficacy prediction model after differential gene analysis and gene screening based on the EBV infection negative sample group; S4, establishing the FOLF0X regimen efficacy prediction model based on the modeling sample set; S5. Verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

2. A method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 1, characterized in that: The S1 further comprises: Obtain samples of multiple PDTX models and perform WES sequencing on each sample of the PDTX model to obtain DNA sequencing data corresponding to the samples of the multiple PDTX models; wherein: S1 includes: S11, performing WES sequencing on each sample of the PDTX model; the WES sequencing uses the lllumina NovaSeqV6000 sequencer, uses whole exome sequencing: Agilent V6 is used as the sequencing reagent, and uses the capture sequencing sequencing method; the sequencing data output is 25G, and the sequencing data requirements are: base quality Q30>90%, GC content is 50±5%, the average depth of the sequencing capture area is 200X-250X, and the coverage requirements are that the capture area coverage is above 95%, and the 10X coverage is above 80%; S12, RNA sequencing was performed on each sample of the PDTX model; the RNA sequencing used an Illumina NovaSeq V6000 sequencer, VAHTS mRNA Capture Beads + mRNA VAHTS Universal V8 RNA-seq Library Prep Kit for IIlumina as sequencing reagents, and a capture sequencing method was used; the sequencing data output was 10G, and the sequencing data requirements were: base quality Q30> 90%, and GC content was 50±5%.

3. A method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 2, characterized in that: The S2 includes: S21, obtaining the drug efficacy data and cancer type information data of the samples of the multiple PDTX models; S22, based on the cluster analysis of the drug efficacy data and the cancer information data, the samples of the multiple PDTX models are regionally classified; the regional classification includes enriched gene regions and non-enriched gene regions; S23, performing gene enrichment analysis on all genes in the enriched gene region to obtain the EBV pathway; S24, performing compatibility verification on all genes in the enriched gene region and the EBV pathway, wherein the EBV genes that pass the compatibility verification are EBV genes that match the EBV pathway; S25, drawing a heat map for the samples of the multiple PDTX models based on the EBV gene, and distinguishing the EBV-infected positive samples from the EBV-infected negative samples by using the heat map to obtain a first distinction result; S26, performing statistical analysis on the DNA sequencing data and RNA sequencing data of the samples of the multiple PDTX models to obtain the sequence number of EBV virus; determining whether each sample in the samples of the multiple PDTX models is infected with EBV virus one by one based on the sequence number of EBV virus, thereby obtaining a second distinction result for distinguishing EBV-infected positive samples from EBV-infected negative samples; S27, determining whether the first distinction result and the second distinction result are consistent; if they are consistent, determining the final EBV infection positive sample and EBV infection negative sample; if they are inconsistent, repeating S21-S26 and performing different regional classifications until a consistent distinction result is obtained.

4. The method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 3, characterized in that: The S3 includes: S31, dividing the samples in the EBV infection negative sample group into FOLFOX regimen efficacy positive samples and FOLFOX regimen efficacy negative samples; S32, based on determining that the sample expression levels of the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples do not conform to a normal distribution, performing a Wilcoxon rank sum test on the FOLFOX regimen-positive samples and the FOLFOX regimen-negative samples to obtain multiple differentially expressed genes; S33, performing pathway enrichment analysis on the multiple differentially expressed genes, and obtaining the most significant pathway with the highest correlation with the efficacy of the FOLFOX regimen based on the analysis results; S34, performing differential gene analysis on the multiple differential genes, drawing a heat map for the expression amounts of the top M most significant differential genes in the differential gene analysis results, performing PCA sample clustering, and then screening N drug efficacy classification genes, wherein the drug efficacy classification genes have a clear classification effect on the FOLFOX regimen drug efficacy of multiple FOLFOX regimen drug efficacy-negative samples; S35, removing the collinear genes among the top M most significantly differentially expressed genes, verifying and confirming that the N pharmacodynamic classification genes are partially or completely on the most significant pathway, and determining a modeling sample set for establishing the pharmacodynamic prediction model of the FOLFOX regimen based on the most significantly differentially expressed genes after removing the collinear genes, the N pharmacodynamic classification genes and combining gene functions.

5. The method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 4, characterized in that: The S4 includes: S41, respectively using random forest, SVM, Lasso regression and logistic regression modeling methods to establish the FOLFOX regimen efficacy prediction model, thereby obtaining a first FOLFOX regimen efficacy prediction model, a second FOLFOX regimen efficacy prediction model, a third FOLFOX regimen efficacy prediction model and a fourth FOLFOX regimen efficacy prediction model; S42, testing the performance of the first FOLFOX regimen pharmacodynamic prediction model, the second FOLFOX regimen pharmacodynamic prediction model, the third FOLFOX regimen pharmacodynamic prediction model and the fourth FOLFOX regimen pharmacodynamic prediction model based on multiple test indicators; selecting the modeling method corresponding to the FOLFOX regimen pharmacodynamic prediction model with the best performance to establish the FOLFOX regimen pharmacodynamic prediction model.

6. A method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 5, characterized in that: The multiple test indicators include: (1) Sensitivity or true positive rate (TPR) = TP / (TP+FN); (2) precision or positive predictive value (PPV) = TP / (TP + FP); (3) True negative rate (TNR) = TN / (FP+TN); (4) Negative predictive value (NPV) = TN / (FN + TN); (5) Accuracy = (TP + TN) / (TP + TN + FP + FN).

7. A method for predicting the efficacy of FOLFOX regimen for EBV-negative patients with digestive tract tumors according to claim 6, characterized in that: The S5 includes: The clinical samples were input into the FOLFOX regimen efficacy prediction model for prediction, and the prediction quality was determined based on the prediction accuracy, true negative rate, true positive rate, negative predictive value, positive predictive value and AUC values.

8. A FOLFOX regimen efficacy prediction system for EBV-negative patients with digestive tract tumors, used to implement the prediction method according to any one of claims 1 to 7, characterized in that: include: A sample and sequencing data acquisition module (101), used to acquire samples of multiple PDTX models and perform WES sequencing and RNA sequencing on the samples of each PDTX model to acquire DNA sequencing data and RNA sequencing data corresponding to the samples of the multiple PDTX models; wherein the PDTX models are PDTX models of the same strain of mice; An EBV monitoring module (102) is used to perform EBV detection on samples of multiple PDTX models, thereby dividing the samples of the multiple PDTX models into EBV infection-negative samples and EBV infection-positive samples, and after removing the EBV infection-positive samples, the EBV infection-negative samples are used to form an EBV infection-negative sample group as a sample basis for establishing a FOLFOX regimen efficacy prediction model and a FOLFOX regimen efficacy prediction; A differential gene analysis and gene screening module (103), used to determine a modeling sample set for establishing a FOLFOX regimen efficacy prediction model after differential gene analysis and gene screening based on the EBV infection negative sample group; A drug efficacy prediction model building module (104), used to build the drug efficacy prediction model of the FOLFOX regimen based on the modeling sample set; The prediction quality verification module (105) is used to verify the prediction quality of the FOLFOX regimen efficacy prediction model based on clinical samples.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the memory stores a plurality of instructions, and the processor is used to read the instructions and execute the prediction method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be read by a processor to execute the prediction method as described in any one of claims 1-7.

Images