Gastric cancer recurrence prediction model

Abstract

The present application provides a gastric cancer recurrence prediction model. First, a set of new mRNAs (AGTR1, DNER, EPHA7 and SUSD5) for postoperative recurrence of gastric cancer patients is determined through systematic and comprehensive transcriptome analysis. Further, it is verified in a large number of public data sets, and then comprehensively verified in tissue samples of six independent clinical cohorts. Finally, the 4-mRNA small group and clinical features are integrated by risk stratification assessment (RSA) model, which has better prediction performance, and also proves that the 4-mRNA has good therapeutic effect on gastric cancer. Continuous monitoring of patients can be achieved, thereby realizing early intervention and personalized treatment adjustment. Not only is it expected to improve the treatment effect of patients by finding recurrence at a more treatable stage, but also to reduce the physical and economic burden related to traditional diagnostic methods.

Description

Technical Field

[0001] This invention belongs to the field of biomedicine, specifically relating to a gastric cancer recurrence prediction model. Background Technology

[0002] Gastric cancer remains a major global health challenge and is the fifth most common cancer. Patients with locally advanced gastric cancer (LAGC) have a particularly poor prognosis, primarily due to high rates of recurrence and metastasis. Despite improvements in treatment strategies, including surgical resection, chemotherapy, and targeted therapy, overall survival for LAGC patients remains unsatisfactory, with approximately 50-60% of LAGC patients experiencing recurrence within two years of initial treatment. Early detection of recurrence is crucial as it significantly impacts patient survival and quality of life. Recurrence typically indicates disease progression to a more advanced stage, making it more difficult to treat effectively.

[0003] Currently, diagnostic methods for detecting LAGC recurrence include imaging techniques such as computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, and positron emission tomography (PET) scans, as well as endoscopy. While these methods are helpful in treating gastric cancer, they also have inherent limitations. First, imaging modalities are often not sensitive enough to detect early or occult metastases, leading to diagnostic delays and reduced treatment outcomes. Second, these techniques are invasive, expensive, and may expose patients to radiation or surgical risks. Furthermore, these examinations are usually performed at predetermined intervals, which may miss the window for early molecular recurrence, thus delaying potential interventions.

[0004] Therefore, developing non-invasive, sensitive, and specific diagnostic tools for the early detection of gastric cancer recurrence is of paramount importance in this field. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, this invention provides a gastric cancer recurrence prediction model.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A first aspect of the invention provides the use of a reagent for detecting the expression level of a biomarker in a sample in the preparation of a product for predicting gastric cancer recurrence / prognosis, said biomarker including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0008] Furthermore, the gastric cancer described is locally advanced gastric cancer.

[0009] A second aspect of the present invention provides a product for predicting gastric cancer recurrence / prognosis, the product comprising a reagent for detecting the expression level of a biomarker in a sample, the biomarker including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0010] Furthermore, the reagent includes oligonucleotide probes that specifically recognize the biomarker gene, primers that specifically amplify the biomarker gene, or binding agents that specifically bind to the protein encoded by the biomarker gene.

[0011] Furthermore, the sequences of the primers for specifically amplifying SUSD5 are shown in SEQ ID NO:1-2.

[0012] Furthermore, the sequences of the primers that specifically amplify EPHA7 are shown in SEQ ID NO:3-4.

[0013] Furthermore, the sequences of the primers that specifically amplify AGTR1 are shown in SEQ ID NO:5-6.

[0014] Furthermore, the sequences of the primers that specifically amplify DNER are shown in SEQ ID NO:7-8.

[0015] Furthermore, the samples include tissues and blood.

[0016] Furthermore, the blood includes peripheral blood and serum.

[0017] Furthermore, the recurrence of gastric cancer includes postoperative recurrence of gastric cancer and preoperative adjuvant therapy.

[0018] Furthermore, the preoperative adjuvant therapy includes one or more of neoadjuvant chemotherapy, concurrent chemoradiotherapy, anti-VEGF-A targeted therapy, anti-HER2 targeted therapy, and immunotherapy.

[0019] Furthermore, the prediction of gastric cancer prognosis includes predicting the survival time and disease-free survival of gastric cancer patients.

[0020] Furthermore, the reagent also includes a detectable marker.

[0021] Furthermore, the products include reagent kits, nucleic acid membrane strips, and test strips.

[0022] Furthermore, the gastric cancer described is locally advanced gastric cancer.

[0023] A third aspect of the present invention provides a model for predicting gastric cancer recurrence / prognosis, wherein the detection indicators of the model include biomarkers, and the biomarkers include one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0024] Furthermore, the detection indicators of the model also include TNM staging, nerve involvement, and postoperative chemotherapy.

[0025] Furthermore, the model is a nodal graph model.

[0026] Furthermore, the calculation formula of the model includes:

[0027] Recurrence probability = [(2.136 × 4-mRNA) + (0.759 × TNM stage) + (0.824 × nerve involvement) + (0.947 × postoperative chemotherapy) + (-5.191)],

[0028] Or recurrence probability = [(3.250×4-mRNA)+(1.908×TNM stage)+(1.798×neural invasion)+(2.512×postoperative chemotherapy)+(-9.625)].

[0029] Furthermore, the gastric cancer described is locally advanced gastric cancer.

[0030] A fourth aspect of the invention provides the application of biomarkers in constructing models for predicting gastric cancer recurrence / prognosis, said biomarkers including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0031] Furthermore, the detection indicators of the model also include TNM staging, nerve involvement, and postoperative chemotherapy.

[0032] Furthermore, the gastric cancer described is locally advanced gastric cancer.

[0033] A fifth aspect of the invention provides the use of inhibitors of biomarkers in the preparation of pharmaceutical compositions for treating gastric cancer / inhibiting gastric cancer metastasis, said biomarkers including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0034] Furthermore, the inhibitors include nucleic acid inhibitors and protein inhibitors.

[0035] Furthermore, the nucleic acid inhibitors include siRNA, shRNA, and ribozymes.

[0036] Furthermore, the nucleic acid inhibitor is selected from siRNA.

[0037] Furthermore, the siRNA of AGTR1 is shown in SEQ ID NO:9 or SEQ ID NO:10.

[0038] Furthermore, the siRNA of DNER is shown in SEQ ID NO:11 or SEQ ID NO:12.

[0039] Furthermore, the siRNA of EPHA7 is shown in SEQ ID NO:13 or SEQ ID NO:14.

[0040] Furthermore, the siRNA of SUSD5 is shown in SEQ ID NO:15 or SEQ ID NO:16.

[0041] Furthermore, the gastric cancer metastasis includes gastric cancer lymph node metastasis and gastric cancer peritoneal metastasis.

[0042] A sixth aspect of the present invention provides a pharmaceutical composition comprising an inhibitor of a biomarker, said biomarker being one or more of AGTR1, DNER, EPHA7, and SUSD5.

[0043] Furthermore, the pharmaceutical composition also includes other pharmaceuticals.

[0044] Furthermore, the pharmaceutical composition also includes a pharmaceutically acceptable carrier.

[0045] A seventh aspect of the present invention provides a method for screening candidate drugs for treating gastric cancer, the method comprising treating a culture system expressing or containing a biomarker gene or its encoded protein with a substance to be screened; and detecting the expression or activity of the biomarker gene or its encoded protein in the system; wherein, when the substance to be screened inhibits the expression level or activity of the biomarker gene or its encoded protein, the substance to be screened is a candidate drug for treating gastric cancer, wherein the biomarker includes one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0046] The eighth aspect of the invention provides the use of biomarkers in screening candidate drugs for the treatment of gastric cancer, said biomarkers including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0047] A ninth aspect of the present invention provides a method for inhibiting the metastasis of gastric cancer cells, the method comprising administering an inhibitor of a biomarker, said biomarker including one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0048] Furthermore, the metastasis includes peritoneal metastasis and lymphatic metastasis.

[0049] The tenth aspect of the present invention provides a system / apparatus for predicting gastric cancer recurrence / prognosis, the system / apparatus comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor executing the model described in the third aspect of the present invention.

[0050] Advantages and beneficial effects of the present invention:

[0051] This application first identified a group of novel mRNAs (AGTR1, DNER, EPHA7, and SUSD5) for postoperative recurrence in gastric cancer patients through systematic and comprehensive transcriptome analysis. These were then validated on a large number of publicly available datasets, followed by comprehensive validation in tissue samples from six independent clinical cohorts. Finally, a risk stratification assessment (RSA) model integrating the 4-mRNA group and clinical characteristics demonstrated superior predictive performance and also proved the therapeutic efficacy of the 4-mRNAs for gastric cancer.

[0052] This application provides a non-invasive, highly sensitive, and specific method for the early detection of recurrence in gastric cancer patients. It enables continuous monitoring of patients, allowing for early intervention and personalized treatment adjustments. This not only promises to improve patient outcomes by detecting recurrence at a more treatable stage but also reduces the physical and economic burden associated with traditional diagnostic methods. Ultimately, this application paves the way for the wider application of transcriptomics-based diagnostics in other cancers, promoting advancements in personalized medicine and improving overall cancer treatment. Attached Figure Description

[0053] Figure 1 This is a flowchart of the research design for predicting postoperative recurrence of gastric cancer patients using 4-mRNA.

[0054] Figure 2 This is a diagram illustrating the discovery process and preliminary validation of candidate biomarkers for postoperative recurrence in gastric cancer patients based on public databases and transcriptome sequencing data.

[0055] Figure 3 This is a training and validation cohort map for identifying gastric cancer recurrence based on 4-mRNA prediction from fresh frozen tissue samples;

[0056] Figure 4 This is a graph that uses endoscopic biopsy samples to validate 4-mRNA to predict gastric cancer recurrence.

[0057] Figure 5 This is a graph showing the prediction and identification of gastric cancer recurrence in training and validation cohorts based on peripheral blood sample 4-mRNA.

[0058] Figure 6 It is to validate the identification and prediction of tumor marker-negative patients based on 4-mRNA from peripheral blood samples and the longitudinal dynamic prediction and identification of different types of recurrence.

[0059] Figure 7 This is a diagram showing how the 4-mRNA gene promotes the proliferation, migration, and invasion of GC cells in vitro.

[0060] Figure 8This is a diagram illustrating how the 4-mRNA gene promotes the in vivo growth and metastasis of GC cell xenograft tumors. Detailed Implementation

[0061] The following provides definitions for some of the terms used in this specification. Unless otherwise stated, all technical and scientific terms used herein generally have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0062] This invention provides the application of reagents for detecting the expression levels of biomarkers in samples in the preparation of products for predicting gastric cancer recurrence / prognosis, wherein the biomarkers include one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0063] In one implementation, AGTR1 includes wild-type, mutant, or fragments thereof. The term encompasses full-length, unprocessed AGTR1, as well as any form of AGTR1 derived from cells and processed. The term encompasses naturally occurring variants of AGTR1 (e.g., splice variants or allelic variants). The term encompasses AGTR1 from, for example, human and any other vertebrate source, including mammalian AGTR1, such as primates and rodents (e.g., mice and rats), gene ID: 185.

[0064] In one implementation, DNER includes wild-type, mutant, or fragments thereof. The term encompasses full-length, unprocessed DNER, as well as any form of DNER derived from cell processing. The term encompasses naturally occurring variants of DNER (e.g., splice variants or allelic variants). The term encompasses DNER from, for example, human and any other vertebrate origin, including mammalian DNERs such as primates and rodents (e.g., mice and rats), gene ID: 92737.

[0065] In one implementation, EPHA7 includes wild-type, mutant, or fragments thereof. The term encompasses full-length, unprocessed EPHA7, as well as any form of EPHA7 derived from cells and processed. The term encompasses naturally occurring variants of EPHA7 (e.g., splice variants or allelic variants). The term encompasses EPHA7 from, for example, human and any other vertebrate sources, including mammalian EPHA7, such as primates and rodents (e.g., mice and rats), gene ID: 2045.

[0066] In one implementation, SUSD5 includes wild-type, mutant, or fragments thereof. The term encompasses full-length, unprocessed SUSD5, as well as any form of SUSD5 derived from cells and processed. The term encompasses naturally occurring variants of SUSD5 (e.g., splice variants or allelic variants). The term encompasses SUSD5 from, for example, human and any other vertebrate source, including mammalian SUSD5, such as primates and rodents (e.g., mice and rats), gene ID: 26032.

[0067] In one embodiment, a detectable marker refers to a composition capable of generating a detectable signal indicating the presence of a target polynucleotide in a sample. Suitable markers include, but are not limited to, radioisotopes, nucleotide chromophores, enzymes, substrates, fluorescent molecules, chemiluminescent components, magnetic particles, and bioluminescent components. Therefore, a marker is any composition detectable by a device or method, including but not limited to spectroscopic, photochemical, biochemical, immunochemical, electrochemical, optical, chemical detection devices, or any other suitable device. In some embodiments, the marker can be visually detected without the aid of a device.

[0068] Among them, radioactive isotopes include but are not limited to 3 H, 14 C 35 S, 125 I, 131 I.

[0069] Enzymes include, but are not limited to, horseradish peroxidase, β-galactosidase, luciferase, alkaline phosphatase, and acetylcholinesterase.

[0070] Fluorescent molecules include, but are not limited to, FITC, rhodamine, and lanthanide phosphors.

[0071] This invention provides the application of biomarkers in constructing models for predicting gastric cancer recurrence / prognosis, wherein the biomarkers include one or more of AGTR1, DNER, EPHA7 and SUSD5;

[0072] The model's detection indicators also include TNM staging, nerve involvement, and postoperative chemotherapy.

[0073] The model is a nodal graph model.

[0074] In one implementation, the nomogram model, also known as the nomogram model, is a clinical prediction model that integrates multiple predictive indicators and then uses scaled line segments to draw them on the same plane at a certain scale. This is used to express the interrelationships between the various variables in the prediction model and to assess disease risk.

[0075] In one implementation, when the model is constructed using a logistic regression formula, the calculation formula is: recurrence probability = [(2.136 × 4 - mRNA) + (0.759 × TNM stage) + (0.824 × nerve invasion) + (0.947 × postoperative chemotherapy) + (-5.191)]. The nomogram model includes seven straight lines arranged from top to bottom and parallel to each other. Each straight line represents a scale with markings.

[0076] The first row contains a fraction scale with values ​​ranging from 0 to 100.

[0077] The second line is the TNM stage scale. When the stage is I, the score is 0; when the stage is II, the score is 30; and when the stage is III, the score is 94.

[0078] The third line is a scale for neural invasion. When no neural invasion occurs (No), the score is 0, and when neural invasion occurs (Yes), the score is 35.

[0079] The fourth line is the Chemotherapy scale. When no postoperative chemotherapy occurred (No), the score is 0, and when postoperative chemotherapy occurred (Yes), the score is 40.

[0080] The fifth row is the 4-mRNA (X4 panel) scale. When the gene expression level is low, the score is 0, and when the gene expression level is high, the score is 100.

[0081] The sixth row is the Total Points scale, with a value range of 0-280.

[0082] The seventh line is a risk of recurrence / metastasis scale, with a value range of 0-0.95.

[0083] In one implementation, when using multivariate logistic regression analysis to construct the model, the calculation formula is: recurrence probability = [(3.250 × 4-mRNA) + (1.908 × TNM stage) + (1.798 × nerve invasion) + (2.512 × postoperative chemotherapy) + (-9.625)]. The nomogram model includes seven straight lines arranged from top to bottom and parallel to each other. Each straight line represents a scale with markings.

[0084] The first row contains a fraction scale with values ​​ranging from 0 to 100.

[0085] The second line is the TNM stage scale. When the stage is I, the score is 0; when the stage is II, the score is 40; and when the stage is III, the score is 64.

[0086] The third line is a scale for neural invasion. When no neural invasion occurs (No), the score is 0, and when neural invasion occurs (Yes), the score is 40.

[0087] The fourth line is the Chemotherapy scale. When no postoperative chemotherapy occurred (No), the score is 0, and when postoperative chemotherapy occurred (Yes), the score is 28.

[0088] The fifth row is the 4-mRNA (X4 panel) scale. When the gene expression level is low, the score is 0, and when the gene expression level is high, the score is 100.

[0089] The sixth row is the Total Points scale, with a value range of 0-260.

[0090] The seventh line is a risk of recurrence / metastasis scale, with a value range of 0-0.8.

[0091] This invention provides the use of biomarker inhibitors in the preparation of pharmaceutical compositions for treating gastric cancer / inhibiting gastric cancer metastasis, wherein the biomarkers include one or more of AGTR1, DNER, EPHA7 and SUSD5.

[0092] In one embodiment, an inhibitor refers to any substance that can reduce the activity of a biomarker protein, decrease the stability of a biomarker gene or protein, downregulate the expression of a biomarker protein, reduce the effective duration of action of a biomarker protein, or inhibit the transcription and translation of a biomarker gene. Such substances can be used in this application as substances useful for downregulating biomarkers, thereby being used for the prevention or treatment of diseases.

[0093] In one embodiment, the inhibitor includes nucleic acid inhibitors and protein inhibitors. The nucleic acid inhibitor is selected from: interfering molecules that target the biomarker or its transcript and are capable of inhibiting the expression or transcription of the biomarker gene, including but not limited to shRNA, siRNA, ribozymes, antisense oligonucleotides, or constructs capable of expressing or forming said shRNA, siRNA, ribozymes, or antisense oligonucleotides. The protein inhibitor is selected from substances that specifically bind to the biomarker protein, such as antibodies or ligands capable of inhibiting the activity of the biomarker protein.

[0094] In a preferred embodiment, the inhibitor is selected from nucleic acid inhibitors.

[0095] In a specific implementation, the nucleic acid inhibitor is selected from siRNA.

[0096] The pharmaceutical composition also includes other drugs.

[0097] In one embodiment, other drugs include other drugs for treating gastric cancer, including but not limited to chemotherapy drugs and targeted therapies. Chemotherapy drugs include oxaliplatin and fluorouracil preparations, such as 5-fluorouracil, or capecitabine, and tegafur. Additionally, targeted therapies for treating gastric cancer include apatinib.

[0098] The invention is further illustrated below with reference to specific embodiments. It should be understood that the specific embodiments described herein are by way of example and are not intended to limit the invention. The main features of the invention can be used in various embodiments without departing from the scope of the invention.

[0099] Example

[0100] 1. Materials and Methods

[0101] 1) Screening for mRNA biomarkers

[0102] The biomarker discovery and validation process used in this study is as follows: Figure 1 As shown in the figure. First, mRNA sequencing data from multiple sources, including the GEO database relapse cohort (GSE62254, 125 relapsed patients vs. 157 non-relapsed patients), the TCGA database cohort (28 relapsed patients vs. 159 non-relapsed patients), and cancer tissue samples from 6 LAGC patients (3 relapsed patients vs. 3 non-relapsed patients), were used to screen for biomarkers.

[0103] To further validate clinical application performance, the expression of selected mRNAs was assessed in 34 pairs of recurrent and non-recurrent cancer tissue samples using quantitative real-time polymerase chain reaction (qRT-PCR), with these samples matched 1:1 propensity score. In addition, peripheral blood samples were collected from 26 pairs of recurrent and non-recurrent patients (matched 1:1 propensity score) and 26 healthy individuals who underwent physical examinations during the same period to detect the expression of candidate mRNAs in peripheral blood. These samples were collected at the Fourth Hospital of Hebei Medical University (FHHMU) from June to December 2023.

[0104] 2) Validate mRNA biomarkers

[0105] A visual nomogram was developed to predict recurrence in patients with latent gastric cancer (LAGC) after radical surgery. The training cohort consisted of 330 patients from the Fourth Hospital of Hebei Medical University (FHHMU) and Shijiazhuang People's Hospital (SJZPH); the validation cohort included 185 patients from Baoding Central Hospital (BDCH), Hengshui People's Hospital (HSCPH), Nanjing University Jinling Hospital (NJJLH), and Wuhan University People's Hospital (WHPH). Fresh frozen samples were collected from January 2017 to December 2019. Exclusion criteria included patients who had received neoadjuvant chemotherapy, targeted therapy, immunotherapy, or had residual gastric cancer after gastrectomy or other concurrent tumors.

[0106] Further analysis was conducted on 126 matched endoscopic biopsy samples from LAGC patients from six institutions to validate the shift from large surgical samples to smaller endoscopic biopsies.

[0107] To facilitate the transition of mRNA sequencing from tissue samples to non-invasive liquid biopsy, a retrospective analysis of serum samples from LAGC patients collected at the six centers mentioned above was conducted. The training cohort comprised 136 patients (June 2017 to December 2020) from FHHMU and SJZPH, while the validation cohort comprised serum samples from 105 patients (January 2016 to December 2019) from the other four institutions. Inclusion and exclusion criteria were the same as for the fresh frozen sample cohort.

[0108] All patients were monitored for recurrence or disease progression via laboratory tests, endoscopy, and abdominal / pelvic CT scans, in accordance with the LAGC treatment guidelines. Tissue samples were immediately frozen in liquid nitrogen and stored at -80°C. Surgical samples were processed according to the Chinese Society of Clinical Oncology guidelines. Tumor and lymph node staging was performed according to AJCC 8th edition. All procedures were in compliance with the Declaration of Helsinki, and written informed consent was obtained from all participants, with approval from the institutional review committees of all participating institutions.

[0109] 3) RNA extraction and gene expression analysis

[0110] Total RNA was isolated from freshly frozen surgical tissue using TRIzol reagent (Invitrogen, Frederick, MA, USA) according to the manufacturer's instructions. For serum samples, total RNA was extracted using the PAXgene Blood RNA Kit (Qiagen, Hilden, Germany). Subsequently, the total RNA was reverse transcribed into cDNA using the GoScript Reverse Transcription System Kit (Promega) according to the manufacturer's instructions. Quantitative reverse transcription PCR (qRT-PCR) analysis was then performed. Using 2^ -ΔΔCTThe relative abundance of the target gene was determined by normalization using GAPDH as an internal reference, where ΔCT represents the difference between the target gene and the GAPDH CT value. The specific PCR primers used are listed in Table 1.

[0111] Table 1 PCR primer sequences

[0112] Name Sequence SUSD5_F GACCGCCTTCACCTTGCTA(SEQ ID NO:1) SUSD5_R TGTCCGGGAATCCTCCATGA(SEQ ID NO:2) EPHA7_F CTAAACGTGGAGCAGCCGAT(SEQ ID NO:3) EPHA7_R CATGGTGCATGAGCAGGTTT(SEQ ID NO:4) AGTR1_F CGGGGCGCGGGTTTG(SEQ ID NO:5) AGTR1_R TCAAATACACCTGGTGCCGA(SEQ ID NO:6) DNER_F GCCGAAAACAGGGCAGAAAG(SEQ ID NO:7) DNER_R CACCCGCAGAGCTGTTAGAA(SEQ ID NO:8)

[0113] 4) Protein-protein interaction (PPI) network analysis

[0114] A human PPI network was constructed using the STRING database (https: / / string-db.org). Enrichr (https: / / maayanlab.cloud / Enrichr / ) was used to perform gene set and pathway analysis on the six-gene candidate list.

[0115] 5) EdU Analysis

[0116] AGS cells transfected with siRNA (Haixing Biosciences, Suzhou, China) were used at a rate of 2 × 10⁻⁶. 5 Cells were seeded at a density of 12-well plates and incubated at 37°C under sterile conditions. Subsequently, Cell-Light was used. TM The EdU DNA Cell Proliferation Kit (EdU: 5′Ethynyl-2′-deoxyuridine; Ribobio, Guangzhou, China) was used to assess AGS cell proliferation according to the manufacturer's protocol. Cell images were acquired using a Zeiss LSM 900 confocal microscope (Zeiss, Germany), and the percentage of positive cells was quantified using ImageJ software.

[0117] 6) CCK-8 analysis

[0118] After siRNA transfection, AGS cells were cultured at a rate of 1×10⁻⁶. 3 Cells were seeded at a density of 10 μL in 96-well plates and cultured aseptically at 37°C overnight. After adhesion, 10 μL of LCK-8 reagent was added to each well and incubated at 37°C for about 2 hours. The absorbance was then measured at 450 nm using a microplate reader (Tecan, USA).

[0119] 7) Colony formation test

[0120] AGS cells transfected with siRNA were seeded into 6-well plates and cultured at 37°C for 1-2 weeks. After the colonies were visible to the naked eye, they were fixed with 4% paraformaldehyde, stained with crystal violet solution, and observed and quantified under a microscope (Leica, Germany).

[0121] 8) Wound healing test

[0122] The confluenced AGS cell monolayer transfected with siRNA was scraped with a 10 μL pipette tip. Migration distances across the scraped areas were measured under a microscope 24–48 hours later. Initial images at 0 hours were used as controls to normalize and calculate relative migration rates.

[0123] 9) Transwell migration and invasion analysis

[0124] Use the Transwell insert with either uncoated (migrating) or Matrigel-coated (invasive) according to the manufacturer's instructions. Transfect AGS cells (4 × 10⁻⁶) 4 Cells were inoculated into the upper chamber and incubated for 24–48 hours. Migrating and invasive cells on the lower surface were quantified in five random microscopic fields.

[0125] 10) Western Blotting

[0126] Transfected AGS cells were lysed in RIPA buffer containing 1% PMSF to extract total protein. Denatured proteins were separated by 10% SDS-PAGE and transferred to PVDF membranes (Seven, Beijing). The membranes were tested with antibodies against AGTR1, DNER, EPHA7, SUSD5, CREB, p-CREB, and GAPDH. Immunoreaction bands were visualized using ECL Plus reagents (Solarbio).

[0127] 11) Immunohistochemistry (IHC)

[0128] Tissue sections were dewaxed at 60°C for 2 hours and treated with xylene. Antigen retrieval was performed using EDTA, followed by quenching of endogenous peroxidase activity with 3% hydrogen peroxide. Sections were then blocked with 5% bovine serum albumin (BSA) and incubated overnight at 4°C with primary antibody. The next day, secondary antibody was used for antigen detection and nuclear counterstaining with diaminobenzidine (DAB) and hematoxylin, respectively. The percentage of positive cells was calculated using ImageJ software, and a scoring system was established based on the percentage of positive cells: ≤25% = 1 point, 26%–50% = 2 points, 51%–75% = 3 points, >75% = 4 points. A scoring system was also established based on staining intensity: no staining = 1 point, light brown = 2 points, distinct brown = 3 points, dark brown = 4 points. The final IHC score was obtained by multiplying the coverage score by the staining intensity score.

[0129] 12) Animal experiments

[0130] In vivo experiments were approved by the Animal Laboratory Animal Management and Use Committee (IACUC) of the Fourth Affiliated Hospital of Hebei Medical University (IACUC Approval No.: 20240255). All BALB / c mice were purchased from Spefair (Beijing) Biotechnology Co., Ltd. To investigate the effects of gene knockdown on tumor formation, lymph node metastasis, and peritoneal metastasis, mice were injected subcutaneously, into the paws, and intraperitoneally with AGS cells transfected with lentivirus encoding shRNA targeting specific genes. Mice were sacrificed when significant differences were observed between the experimental and control groups, and the experimental results were statistically analyzed. Each mouse received only one treatment.

[0131] 13) Data Analysis

[0132] Statistical analyses were performed using IBM SPSS version 23, R version 3.6.3, and GraphPad Prism version 8.0. Univariate and multivariate logistic regression analyses identified important clinicopathological variables and mRNA classifiers as covariates; variables important in univariate analyses were incorporated into multivariate regression. In the discovery phase, differential gene expression between the relapse and non-relapse groups was examined using the Wilcoxon rank-sum test and Bonferroni correction. In the clinical validation phase, a gene-based risk score model was built using logistic regression and backward elimination, and model performance was evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC) values. The AUC was derived from the ROC curves using the pROC package in R, and the DeLong test was used for ROC curve comparisons. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), precision, and accuracy of the 4-mRNA biomarker group were determined using the pROC package, and the results were displayed in a confusion matrix. The optimal cutoff value for the ROC curve was determined using the Youuden index in the pROC package. The Youden index and median risk score were used to categorize individuals into high-risk or low-risk groups to predict recurrence and detect metastasis. Disease-free survival (DFS) was defined as the time from radical surgery to disease recurrence or death due to disease progression, analyzed using the Kaplan-Meier method, with a 5-year retrospective of patients who survived and were relapse-free at this milestone. Patients lost to follow-up within 5 years without signs of recurrence were evaluated at their last visit. Statistical significance was set at P < 0.05.

[0133] 2. Experimental Results

[0134] 1) Identification of candidate mRNAs associated with relapse in LAGC patients

[0135] Transcriptome data from two publicly available LAGC datasets (TCGA and GSE62254) were analyzed along with mRNA sequencing data from three pairs of LAGC tissues that were recurrent or non-recurrent after radical surgery, identifying four differentially expressed genes (AGTR1, DNER, EPHA7, and SUSD5). Figure 2 A). Volcano plots show that these four genes are upregulated in recurrent cancer tissue compared to non-recurrent tissue. Figure 2 B).

[0136] Further analysis of TCGA data showed that the expression of these mRNA genes in the relapse group (Recur-T) of LAGC patients was significantly higher than that in the non-relapse group (Non-Recur-T, P<0.05); Figure 2 C). High expression levels of these genes were also significantly associated with poorer overall survival (OS) and disease-free survival (DFS) in these patients. Figure 2 To validate these findings, a pilot cohort was established at FHHMU, and 34 matched primary cancer tissue samples (recur-T, recur-N; T: tumor, N: normal) and non-recur-T, non-recur-N; T: tumor, N: normal) were analyzed using qRT-PCR. The results confirmed that the expression levels of these mRNAs were higher in recurrent tissues (P<0.05). Figure 2 D). Figure 2 The results showed the relationship between high and low expression profiles of four mRNA genes and clinicopathological features in 34 patients in the pilot cohort, with high expression of each mRNA closely associated with highly aggressive clinical features. Peripheral blood samples also validated these results. Figure 2 E). Western blot analysis showed that these genes were significantly upregulated in recurrent cancer tissues. Figure 2 F, 2S-V; R: recurrence, Non-R: non-recurrence). IHC staining further confirmed that the expression levels of these mRNAs were significantly higher in recurrent (Recur) cancer tissues compared with non-recurrence (Non-Recur) tissues (P<0.05). Figure 2 G).

[0137] Protein-protein interaction network diagrams were constructed using the STRING database and visualized using Cytoscape 3.9.1 to elucidate the potential roles of these four mRNAs in gastric cancer. Figure 2J). Furthermore, analysis using the Timer 2.0 database (http: / / timer.cistrome.org / ) revealed a positive correlation between these four mRNA genes and the metastasis-related genes MMP2, VEGFC, Vimentin, and N-cadherin. Figure 2 I). Mutation analysis results showed that the overall tumor mutation burden and TTN gene mutation frequency were higher in the AGTR1, SUSD5, and EPHA7 low expression groups, while DENR showed the opposite trend. Figure 2 W).

[0138] 2) Validate the effectiveness of 4-mRNA in surgically resected samples from LAGC patients in predicting recurrence.

[0139] First, the expression of four mRNA genes in the training cohort was quantitatively analyzed using qRT-PCR. Multivariate analysis showed that each gene independently affected the risk of recurrence (P<0.05; Table 2). Further multivariate analysis showed that 4-mRNA (OR=8.466, 95% CI: 4.749–15.095, P<0.001), TNM stage (OR=2.136, 95% CI: 1.330–3.430, P=0.002), nerve invasion (OR=2.280, 95% CI: 1.209–4.302, P=0.011), and postoperative chemotherapy (OR=2.577, 95% CI: 1.176–5.648, P=0.018) were independent risk factors for recurrence after LAGC (Table 3). The recurrence probability was estimated using a logistic regression formula: [(2.136 × 4-mRNA) + (0.759 × TNM stage) + (0.824 × neurological involvement) + (0.947 × postoperative chemotherapy) + (-5.191)]. This model, represented in a Noetherian pattern, visually demonstrates the prediction of postoperative recurrence in LAGC patients. Figure 3 A). Patients were categorized into low-risk and high-risk groups based on the cutoff values ​​determined by the Youuden index (clinical model: 0.309; 4-mRNA model: 0.494; RSA model: 0.517). A risk stratification assessment (RSA) model was developed by combining the 4-mRNA group with clinical variables. This model demonstrated excellent recurrence prediction ability, with an AUC of 0.864 (95% CI: 0.825–0.902, P < 0.001). Figure 3 B, 3D, 3F (above). Notably, according to the DeLong test, the RSA model showed a higher AUC value than the clinical model in the training cohort (0.864 vs. 0.745; P = 0.001). The calibration curve of the model further highlights its predictive accuracy ( Figure 3 G). The AUC box plot after 1000 bootstrap iterations shows that the RSA model has better discriminative ability, sensitivity, and specificity than the clinical model and the 4-mRNA model. Figure 3 K).

[0140] Table 2. Multivariate logistic regression analysis of four candidate mRNAs influencing recurrence and metastasis after radical surgery in LAGC patients.

[0141]

[0142] Table 3. Multivariate logistic regression analysis of the influence of recurrence and metastasis after radical surgery in LAGC patients.

[0143]

[0144]

[0145] After calibrating the model using training cohort data, the same statistical parameters were applied to the validation cohort. The model, with consistent statistical parameters, was then applied to an independent external validation cohort of 185 patients with LAGC, demonstrating strong predictive ability (AUC = 0.919, 95% CI: 0.881–0.957, P < 0.001). Figure 3 C, 3E, 3F). Calibration curve analysis further confirmed the enhanced predictive accuracy. Figure 3 H). Similarly, the results of 1000 bootstrap iterations for each model also show that the RSA model has the best predictive performance (H). Figure 3 L). Furthermore, stratified analysis based on different TNM stages, HER2, and PDL1 molecular markers revealed that the RSA model outperformed the clinical characteristic model and the 4-mRNA model in clinical prediction. Figure 3 PS).

[0146] The potential of the RSA model in improving the cost-effectiveness of clinical decision-making was evaluated using DCA curve results. The results showed that both the training and validation cohorts had good clinical benefits. Figure 3 M, N). Double-layer concentric circle plots showed that 56.1% of LAGC patients had a high risk of postoperative recurrence, while 43.9% were at low risk. Follow-up showed that 13.78% (27 out of 196) of high-risk patients and 2.55% (5 out of 196) of low-risk patients experienced recurrence. The RSA model identified more low-risk patients than the clinical characteristic model (50.6% vs. 43.9%), with only 7.9% (26 / 330) of low-risk patients experiencing recurrence, compared to 35.8% (118 / 330) of high-risk patients. Figure 3 On O). Similar results were obtained in the validation set ( Figure 3 (O below).

[0147] Furthermore, survival follow-up of all included LAGC patients showed that the 5-year DFS of high-risk patients in both the training and validation cohorts was significantly worse than that of low-risk patients (training cohort: 30.1% vs. 56.9%, P<0.0001; validation cohort: 32.9% vs. 55.0%, P<0.0001). Figure 3 I, 3J).

[0148] 3) Endoscopic biopsy samples were used to validate the ability of 4-mRNA genomes to predict relapse.

[0149] In addition to surgical resection samples from the training and validation cohorts, this study also obtained 126 matched endoscopic biopsy samples from 6 centers, of which 51 cases showed postoperative recurrence and 75 cases did not. The expression profiles of four genes were highly correlated between the biopsy and surgical samples. Figure 4 AD), there was no significant difference in gene expression among matched samples ( Figure 4 EH). The AUC value and calibration curve confirm the effectiveness and accuracy of the RSA model. Figure 4 IK). In the biopsy cohort, the RSA model demonstrated the highest sensitivity (72.5%) and specificity (89.3%). Figure 4 Furthermore, the RSA model improved the recurrence diagnosis rate in the high-risk group and reduced the recurrence diagnosis rate in the low-risk group. Figure 4 This highlights its potential in enhancing clinical decision-making and minimizing unnecessary interventions.

[0150] Box plots of AUC values ​​after 1000 bootstrap iterations show that the RSA model outperforms clinical and 4-mRNA models in terms of discriminant ability, sensitivity, and specificity. Figure 4 Similar to previous cohort follow-up studies, patients with endoscopic biopsy samples were subjected to a log-rank test and divided into high-risk and low-risk groups based on the nomogram. The results showed that the 5-year DFS in the high-risk group was significantly lower than that in the low-risk group (33.8% vs. 54.1%, P = 0.033). Figure 4 M).

[0151] Furthermore, it was assumed that the RSA model could predict the efficacy and recurrence rate of neoadjuvant therapy in patients with LAGC. Therefore, seven prospective neoadjuvant therapy cohorts conducted by FHHMU were included. Treatment regimens included neoadjuvant chemotherapy (NCT01516944, NCT02555358), concurrent chemoradiotherapy (NCT01962246), anti-VEGF-A targeted therapy (NCT03349866), anti-HER2 targeted therapy (NCT02380131), and immunotherapy (ChiCTR2000030414). The results showed that, except for the neoadjuvant anti-HER2 targeted therapy group, the recurrence rate in the high-risk group was significantly higher than that in the low-risk group (all P<0.05). Figure 4 Q), and the proportion of high-risk patients who responded to neoadjuvant therapy was comparable to that of low-risk patients (both P>0.05); Figure 4 R).

[0152] 4) Validate the 4-mRNA genome in peripheral blood samples to predict relapse in LAGC patients.

[0153] Forty patients were randomly selected from the recruited cohort. The training cohort consisted of patients from the Fourth Hospital of Hebei Medical University (FHHMU) and Shijiazhuang People's Hospital (SJZPH). The validation cohort included patients from Baoding Central Hospital (BDCH), Hengshui People's Hospital (HSCPH), Nanjing University Jinling Hospital (NJJLH), and Wuhan University People's Hospital (WHPH). Quality control analysis was performed on the expression of four mRNAs in their peripheral blood samples. The A260 / 280 and A260 / 230 ratios were within the normal range at different time points. Figure 5 A), gel electrophoresis showed distinct bands for the four mRNAs. Figure 5 The presence of O indicates that the mRNA in peripheral blood has not been degraded and is suitable for qRT-PCR analysis.

[0154] Multivariate logistic regression analysis showed that high expression of four mRNAs was a risk factor for postoperative recurrence in LAGC patients (P<0.05). A nomogram predicting postoperative recurrence was constructed based on multivariate logistic regression analysis. Figure 5 B), the AUC of the four mRNA biomarker groups was 0.824 (95% CI: 0.755-0.893, P<0.001); Figure 5C) indicates a high predictive accuracy. The probability of postoperative recurrence was calculated using the logistic regression coefficient formula: [(3.250×4-mRNA)+(1.908×TNM stage)+(1.798×neural invasion)+(2.512×postoperative chemotherapy)+(-9.625)]. Patients were categorized into low-risk and high-risk groups based on the cutoff value determined by the Youden index (clinical model: 0.469; 4-mRNA model: 0.517; RSA model: 0.603). Notably, the liquid biopsy-based RSA model outperformed the clinical and 4-mRNA models in predicting recurrence. Figure 5 E, 5I), calibration curve analysis confirmed its excellent performance ( Figure 5 G). Comparative clinical benefit analysis showed that the RSA model increased the recurrence detection rate from 33.1% to 39.0% in the high-risk group and decreased the recurrence detection rate from 12.5% ​​to 6.6% in the low-risk group. Figure 5 Patients stratified into low-risk and high-risk groups according to the Youden index showed significantly different five-year DFS rates (35.1% vs. 52.9%, P = 0.023). Figure 5 J).

[0155] The predictive power of the RSA model was further validated in independent cohorts, with an AUC of 0.935 (95% CI: 0.891–0.979, P<0.001). Figure 5 D). Radar chart and confusion matrix analysis confirmed that the RSA model outperformed four mRNA genome and clinical feature models in predicting relapse. Figure 5 (F, 5I below). Calibration curve analysis also confirmed that the predictive performance of the RSA model was improved. Figure 5 H). The above results demonstrate that histological four-genome analysis can be successfully applied to liquid biopsy of peripheral blood samples, and the RSA model combining serum four-genome data and clinical characteristics can effectively predict postoperative recurrence in LAGC patients. Decision curve analysis and bilayer concentric circle plots further confirm the potential of the RSA model in improving clinical decision-making and optimizing patient management. Figure 5 M, 5N). Meanwhile, follow-up revealed that the 5-year DFS in the high-risk group was significantly lower than that in the low-risk group using the RSA model (35.5% vs. 51.4%, P = 0.029). Figure 5 K).

[0156] 5) 4-mRNA genome was used for dynamic monitoring of postoperative recurrence in LAGC patients.

[0157] To evaluate the ability of the RSA model to identify tumor marker-negative patients with a high probability of postoperative recurrence, peripheral blood samples from tumor marker-negative patients in the training and validation cohorts were analyzed. This included 58 patients without postoperative recurrence and 42 patients with recurrence. Training cohort: The Fourth Hospital of Hebei Medical University (FHHMU) and Shijiazhuang People's Hospital (SJZPH); Validation cohort: The validation cohort included patients from Baoding Central Hospital (BDCH), Hengshui People's Hospital (HSCPH), Nanjing University Jinling Hospital (NJJLH), and Wuhan University People's Hospital (WHPH). Nonographs generated in the training cohort were validated. The results showed that the RSA model (AUC = 0.916, 95% CI: 0.862–0.970) was significantly superior to the clinical characteristic model (AUC = 0.769; 95% CI: 0.727–0.893) and the 4-mRNA model (AUC = 0.828, 95% CI: 0.746–0.911) in predicting postoperative recurrence. Figure 6 AB). Confusion matrix ( Figure 6 EG) and calibration curve ( Figure 6 C) Analysis confirmed that the RSA model had higher accuracy than other models in predicting postoperative recurrence. Analysis of the ability of different models to identify recurrence among different risk groups revealed that the RSA model was significantly superior to clinical characteristics and 4-mRNA in detecting postoperative recurrence. Figure 6 HK). Furthermore, in patients with negative tumor markers, the 5-year DFS was significantly lower in the high-risk group than in the low-risk group (28.0% vs. 54.0%, P = 0.0019). Figure 6 D).

[0158] In addition, the diagnostic performance of the 4-mRNA combination in predicting postoperative recurrence was evaluated using serum samples from patients with other gastrointestinal malignancies, including colorectal cancer (n=35), hepatocellular carcinoma (n=38), pancreatic adenocarcinoma (n=29), and esophageal cancer (n=29). Compared with its performance in predicting postoperative recurrence in patients with LAGC, the RSA model showed poor predictive ability for postoperative recurrence in colorectal cancer (AUC=0.606), hepatocellular carcinoma (AUC=0.659), pancreatic adenocarcinoma (AUC=0.714), and esophageal cancer (AUC=0.681). Figure 6 LO). The DeLong test confirmed that the RSA model had higher specificity for LAGC than for other gastrointestinal cancers (P<0.001).

[0159] Furthermore, to further validate the predictive performance of the RSA model for longitudinal dynamic changes in LAGC patients, five patients with recurrence after radical surgery were selected, including peritoneal metastasis, ovarian metastasis, local tumor bed metastasis, lung metastasis, and liver metastasis, which are common metastatic forms after LAGC surgery. Peripheral blood was collected from these five patients during postoperative follow-up, and the difference between the recurrence time predicted by the longitudinal dynamic changes of the 4-mRNA group and the recurrence time detected by traditional CT imaging was analyzed and compared. Red boxes indicate elevated levels of the four mRNAs, suggesting a higher risk of recurrence. Red arrows indicate the time when recurrence was detected by traditional CT. Figure 6 The results of P show that, compared with traditional CT detection, the RSA model constructed by combining four mRNA combinations with clinical features can achieve recurrence early warning through longitudinal dynamic monitoring, especially for common clinical metastatic forms. Figure 6 QU).

[0160] 6) mRNA genes promote GC proliferation and translocation in vivo and in vitro.

[0161] Considering the potential clinical value of the 4-mRNA group in predicting postoperative recurrence in LAGC patients, the biological effects of the four mRNA genes (AGTR1, DNER, EPHA7, and SUSD5) constituting this group were investigated. Specific siRNA sequences targeting these genes were used (Table 4). Scratch healing and transwell assays showed that silencing AGTR1, DNER, EPHA7, and SUSD5 significantly attenuated the migration and invasion abilities of AGS cells. Figure 7 AD, 7O-R). Furthermore, EdU assays showed that DNA replication activity was reduced in AGS cells transfected with these siRNAs compared to control cells. Figure 7 EF, 7S-T). Colony formation and CCK-8 assays further indicated that, after silencing these genes, proliferation ( ) Figure 7 KN) and clonogenic potential ( Figure 7 GJ) decreased significantly.

[0162] Knockdown experiments targeting AGTR1, DNER, EPHA7, and SUSD5 were performed in BALB / c nude mice to investigate the effects of these recurrent mRNA genes on GC tumor growth and metastasis. Compared with the control group, the siRNA group showed reduced subcutaneous tumor growth and weight. Figure 8 AD). IHC analysis of primary subcutaneous xenograft tumors showed that knockdown of these genes decreased the expression of the proliferation marker Ki-67 and mesenchymal markers N-cadherin and vimentin, while increasing the expression of the epithelial marker E-cadherin. Figure 8E, 8O-Q). Compared with the control group, the siRNA group showed a significant reduction in peritoneal metastasis (E, 8O-Q). Figure 8 GH). Consistent with subcutaneous xenograft tumor models, IHC staining showed that silencing these genes reduced the expression of metastatic (MMP9), stromal (N-cadherin, vimentin), and epithelial (E-cadherin) markers. Figure 8 F, 8R-T). Furthermore, analysis of primary foot tumors and popliteal lymph nodes showed a reduced lymph node metastasis burden in the siRNA group (F, 8R-T). Figure 8 IJ). Notably, IHC analysis of the popliteal lymph nodes showed that knockdown of AGTR1, DNER, EPHA7, and SUSD5 decreased the expression of lymphangiogenic marker LYVE1 and stromal markers (N-cadherin and vimentin), while increasing the expression of epithelial marker (E-cadherin). Figure 8 K, 8U-W). In summary, these in vitro and in vivo findings suggest that mRNA panel genes enhance GC migration, invasion, epithelial-mesenchymal transition, lymphangiogenesis, and metastasis.

[0163] Table 4 siRNA sequences

[0164]

[0165] Furthermore, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis showed that all four recurrent mRNA genes were associated with neuroactive ligand-receptor interactions, calcium signaling, and cAMP signaling pathways. Figure 8 L, 8X, 8Z, 8Z2). Western blot analysis showed that phosphorylation of the key cAMP pathway effector CREB was reduced after knockout of the four recurrent mRNA genome genes, suggesting that their potential pro-cancer effects are mediated by cAMP signaling. Figure 8 N, 8Y, 8Z1, 8Z3).

[0166] The above description of the embodiments is only for understanding the method and core ideas of the present invention. It should be noted that those skilled in the art can make various improvements and modifications to the present invention without departing from the principles of the invention, and these improvements and modifications will also fall within the protection scope of the claims of the present invention.

Claims

1. The application of a reagent for detecting the expression level of a biomarker in a sample in the preparation of a product for predicting gastric cancer recurrence / prognosis, wherein the biomarker is mRNA, and the mRNA is composed of AGTR1, DNER, EPHA7 and SUSD5; and the gastric cancer is locally advanced gastric cancer.

2. A product for predicting gastric cancer recurrence / prognosis, characterized in that, The product is a reagent for detecting the expression level of biomarkers in a sample. The biomarker is mRNA, which is composed of AGTR1, DNER, EPHA7, and SUSD5. The gastric cancer is locally advanced gastric cancer.

3. The product according to claim 2, characterized in that, The reagents include oligonucleotide probes that specifically recognize the biomarker gene and primers that specifically amplify the biomarker gene.

4. The product according to claim 3, characterized in that, The sequences of the primers that specifically amplify SUSD5 are shown in SEQ ID NO:1-2.

5. The product according to claim 3, characterized in that, The sequences of the primers that specifically amplify EPHA7 are shown in SEQ ID NO:3-4.

6. The product according to claim 3, characterized in that, The sequences of the primers that specifically amplify AGTR1 are shown in SEQ ID NO:5-6.

7. The product according to claim 3, characterized in that, The sequences of the primers that specifically amplify DNER are shown in SEQ ID NO:7-8.

8. The product according to claim 2, characterized in that, The samples include tissues and blood.

9. The product according to claim 8, characterized in that, The blood includes peripheral blood and serum.

10. The product according to claim 2, characterized in that, The term "gastric cancer recurrence" includes postoperative recurrence of gastric cancer and preoperative adjuvant therapy.

11. The product according to claim 10, characterized in that, The preoperative adjuvant therapy includes one or more of the following: neoadjuvant chemotherapy, concurrent chemoradiotherapy, anti-VEGF-A targeted therapy, anti-HER2 targeted therapy, and immunotherapy.

12. The product according to claim 2, characterized in that, The prediction of gastric cancer prognosis includes the prediction of gastric cancer survival and disease-free survival.

13. The product according to claim 2, characterized in that, The reagent also includes a detectable marker.

14. The product according to claim 2, characterized in that, The products include reagent kits, nucleic acid membrane strips, and test strips.

15. A model for predicting gastric cancer recurrence / prognosis, characterized in that, The detection indicators of the model include biomarkers, the biomarkers being mRNA, which is composed of AGTR1, DNER, EPHA7 and SUSD5; the gastric cancer is locally advanced gastric cancer.

16. The model according to claim 15, characterized in that, The model's detection indicators also include TNM staging, nerve involvement, and postoperative chemotherapy.

17. The model according to claim 15, characterized in that, The model is a nodal graph model.

18. The model according to claim 15, characterized in that, The calculation formula of the model includes: Recurrence probability = [(2.136 × 4-mRNA) + (0.759 × TNM stage) + (0.824 × nerve involvement) + (0.947 × postoperative chemotherapy) + (-5.191)], Or recurrence probability = [(3.250×4-mRNA) + (1.908×TNM stage) + (1.798×neural invasion) + (2.512×postoperative chemotherapy) + (-9.625)].

19. Application of biomarkers in constructing models for predicting gastric cancer recurrence / prognosis, wherein the biomarker is mRNA, and the mRNA is composed of AGTR1, DNER, EPHA7 and SUSD5; and the gastric cancer is locally advanced gastric cancer.

20. The application according to claim 19, characterized in that, The model's detection indicators also include TNM staging, nerve involvement, and postoperative chemotherapy.

21. A system / device for predicting gastric cancer recurrence / prognosis, characterized in that, The system / device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor executing the model according to any one of claims 15-18.

Images