Microbial marker combination for predicting prognosis of gastric cancer and application thereof

Abstract

The invention relates to the field of medical detection, in particular to a microbial marker combination for predicting gastric cancer prognosis and application thereof. According to the invention, obvious correlation between the microbial spectrum and gastric cancer progression is found, and through comprehensive data analysis of microbiome abundance, clinical information and transcriptome data on a patient, specific 20 microbial markers related to gastric cancer prognosis are determined, and a characteristic model of prognosis layering of the patient is established. The microbial model is used for predicting the immunotherapy curative effect and reaction of the gastric cancer patient, and can help a doctor to more accurately judge which patients are more likely to benefit from immunotherapy before treatment, so that a more personalized and effective treatment scheme is formulated for the patient.

Description

Technical Field

[0001] This invention belongs to the field of medical testing technology, and in particular relates to a combination of microbial biomarkers for predicting the prognosis of gastric cancer and its application. Background Technology

[0002] Gastric cancer (GC) is one of the most common malignant tumors, a major global health problem, and a leading cause of cancer death. As part of the tumor microenvironment, the gastric microbiota is receiving increasing attention. Although Helicobacter pylori is widely considered a contributing factor to GC, substantial evidence suggests that other microorganisms present in the gastric mucosa also participate in disease progression. Gastric microbes promote GC by stimulating inflammation, modulating host immune responses, producing harmful metabolites, and inducing DNA damage. Studies have found that the diversity and function of the gastrointestinal microbiota change during cancer development. The abundance of microorganisms in the tumors of GC patients differs significantly from that in healthy individuals; Streptococcus, Fusobacterium, and Prevotella are more abundant in tumor tissue, and these differing microbial communities may play a crucial role in GC progression. Discovering GC-specific microbes holds promise as potential biomarkers and therapeutic targets, allowing for patient stratification based on the abundance of key microbes and predicting drug efficacy. Research on other GC microbial biomarkers besides Helicobacter pylori is scarce, and this field urgently needs further exploration. Summary of the Invention

[0003] To address the aforementioned problems in the existing technology, this invention provides a combination of microbial biomarkers for predicting the prognosis of gastric cancer and its application, which can predict the efficacy of immunotherapy in gastric cancer patients.

[0004] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0005] In a first aspect, this application provides a combination of microbial biomarkers for predicting the prognosis of gastric cancer, the combination of microbial biomarkers comprising: Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter.

[0006] Optionally, the microbial biomarker combination consists of Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter.

[0007] Secondly, this application also provides the use of a reagent for determining the abundance of each microorganism in the microbial biomarker combination described in the first aspect in the preparation of a kit for predicting the prognosis of gastric cancer.

[0008] Optionally, the microorganisms are derived from tumor samples from gastric cancer patients.

[0009] Thirdly, this application also provides a risk assessment method for predicting the prognosis of gastric cancer, including:

[0010] Samples were collected from patients, and the abundance of various microorganisms in the samples was determined. These microorganisms included Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter.

[0011] Based on the abundance data of microorganisms, a risk score is calculated according to a risk scoring model, which is: Risk Score = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola*0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_Thermosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7286)+g_Sulfitobacter*(-0. 3210)+g_Frigoribacterium*(-0.0740)+g_Flectobacillus*(-0.4914)+g_Serinicoccus*0.1775+g_Sphingobacterium*(-0.14 31)+g_Sandarakinorhabdus*0.0207+g_Verminephrobacter*(-0.1341)+g_Klebsiella*0.3162+g_Congregibacter*(-0.3978);

[0012] Among them, Risk Score represents the risk score, and g_microbial Latin name represents the abundance of microorganisms of that genus in the patient sample;

[0013] Patient prognosis is predicted based on risk scores; the lower the risk score, the better the prognosis.

[0014] Optionally, the method of predicting patient prognosis based on risk scores includes: using the median of the risk score as a cutoff value; when the risk score is ≥ the median, the patient is classified as a high-risk group; when the risk score is < the median, the patient is classified as a low-risk group; and the overall survival of the low-risk group is significantly better than that of the high-risk group.

[0015] Fourthly, this application also provides a system for predicting the prognosis of gastric cancer, the system comprising the following modules:

[0016] 1) Data collection module: Collects patient samples and measures the abundance of various microorganisms in the samples, including Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter;

[0017] 2) Model Calculation Module: Based on the abundance data of microorganisms, the risk score is calculated according to the risk scoring model, which is: Risk Score = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola*0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_Thermosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7286)+g_Sulfitobacter*(-0. 3210)+g_Frigoribacterium*(-0.0740)+g_Flectobacillus*(-0.4914)+g_Serinicoccus*0.1775+g_Sphingobacterium*(-0.14 31)+g_Sandarakinorhabdus*0.0207+g_Verminephrobacter*(-0.1341)+g_Klebsiella*0.3162+g_Congregibacter*(-0.3978);

[0018] 3) Output prediction module: Predicts the patient's prognosis based on the risk score. The lower the patient's risk score, the better the prognosis.

[0019] Fifthly, this application also provides a kit comprising reagents for detecting the abundance of each microorganism in the microbial biomarker combination described in the first aspect.

[0020] Compared with the prior art, this application provides at least the following beneficial effects:

[0021] This invention identifies specific microbial biomarkers associated with gastric cancer prognosis and establishes a characteristic model for patient prognostic stratification. Our study included 448 gastric cancer patients, encompassing comprehensive data analysis of microbiome abundance, clinical information, and transcriptome data. We utilized high-throughput sequence analysis and linear discriminant analysis to identify and visualize important microbial biomarkers. We further validated these biomarkers using univariate Cox and LASSO Cox regression methods, resulting in 20 microbial biomarkers used to construct a risk scoring model: Risk Score = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola* 0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_The rmosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7286)+g_Sulfitobacter*(-0. The results were calculated as follows: (3210) + g_Frigoribacterium*(-0.0740) + g_Flectobacillus*(-0.4914) + g_Serinicoccus*0.1775 + g_Sphingobacterium*(-0.1431) + g_Sandarakinorhabdus*0.0207 + g_Verminephrobacter*(-0.1341) + g_Klebsiella*0.3162 + g_Congregibacter*(-0.3978). We found a significant correlation between the microbial profile and gastric cancer progression; for example, specific microorganisms g_Sphingobacterium and g_Achromobacter showed significant prognostic significance. This microbial model can predict the efficacy and response to immunotherapy in gastric cancer patients. This model can help physicians more accurately determine which patients are more likely to benefit from immunotherapy before treatment, thereby developing more personalized and effective treatment plans. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0023] Figure 1 This section describes the distribution of boundaries, phylum, class, order, family, and genus in gastric cancer and adjacent normal samples. Figure 1 Distribution of A-level hierarchy by kingdom, phylum, class, order, family, and genus. Figure 1 The color blocks in B, from right to left, represent: p_Caldiserica, p_Spirochaetes, p_Chlamydiae, p_Tenericutes, p_Crenarchaeota, p_Planctomycetes, p_Proteobacteria, p_Thaumarchaeota, p_Euryarchaeota, p_Synergistetes, p_Deinococcus, p_Bacteroidetes, p_Chlorobi, p_Actinobacteria, p_Firmicutes, p_Nitrospinae, p_Chrysiogenetes, p_Armatimonadetes, p_Thermotogae, and p_Acidobacteria.

[0024] Figure 1 In C++, the color blocks from right to left represent: c_Caldisericia, c_Acidithiobacillia, c_Spirochaetia, c_Chlamydiia, c_Bacteroidia, c_Mollicutes, c_

[0025] Phycisphaerae, c_Thermoprotei, c_Ktedonobacteria, c_Methanobacteria, c_Tissierellia, c_Erysipelotrichia, c_Halobacteria, c_Gammaproteobacteria, c_Thermococci, c_Planctomycetia, c_Opitutae, c_Delta proteobacteria, c_Alphaproteobacteria, c_Coriobacteriia;

[0026] Figure 1The color blocks from right to left in D represent in sequence: o_Mycoplasmatales, o_Streptomycetales, o_Kiloniellales, o_Aeromonadales, o_Spirochaetales, o_Caldisericales, o_Vibrionales, o_Acidithiobacillales, o_Desulfovibrionales, o_Pasteurellales, o_Chlamydiales, o_Geodermatophilales, o_Thermogemmatisporales, o_Alteromonadales, o_Sulfolobales, o_Haloferacales, o_Brachyspirales, o_Bacteroidales, o_Xanthomonadales, o_Campylobacterales;

[0027] Figure 1 The color blocks from right to left in E represent in sequence: f_Bacteroidaceae, f_Listeriaceae, f_Chlamydiaceae, f_Colwelliaceae, f_Rikenellaceae, f_Shewanellaceae, f_Desulfovibrionaceae, f_Waddliaceae, f_Sutterellaceae, f_Succinivibrionaceae, f_Gordoniaceae, f_Polydnaviridae, f_Nocardiaceae, f_Mycoplasmataceae, f_Pseudoalteromonadaceae, f_Nimaviridae, f_Retroviridae, f_Streptomycetaceae, f_Atopobiaceae, f_Kiloniellaceae; Figure 1The color blocks in F, from right to left, represent: g_Terrabacter, g_Bacteroides, g_Proteus, g_Listeria, g_Neisseria, g_Salmonella, g_Bordetella, g_Shigella, g_Aeromonas, g_Campylobacter, g_Streptomyces, g_Chlamydia, g_Nocardia, g_Sanguibacteroides, g_Vibrio, g_Colwellia, g_Enterococcus, g_Luteibacter, g_Desulfovibrio, and g_Leuconostoc.

[0028] Figure 2 Statistical display of LDA scores for characteristic microorganisms in LEfSe analysis results;

[0029] Figure 3 A branching diagram of microbial evolution for LEfSe analysis is presented;

[0030] Figure 4 Forest plot of univariate Cox regression results;

[0031] Figure 5 The results are presented as a univariate Cox regression; A. Volcano plot showing the prognostic effects of the abundance of differentially expressed microbial markers; B-G. KM curves showing the top 6 most significant p-values.

[0032] Figure 6 The KM curves (A, B) and ROC curves (C) for the high and low risk groups of the training set are shown.

[0033] Figure 7 To validate the KM curves (A, B) and ROC curve (C) for high and low risk groups;

[0034] Figure 8 To construct and display the nomograms: AB. Nomograms of the training and validation sets; CD. ROC curves of the nomograms of the training and validation sets; E. Nomograms of the training and validation sets and the C-index of the Risk score.

[0035] Figure 9 This demonstrates the correlation between risk scores and clinical characteristics.

[0036] Figure 10 This demonstrates the correlation between gene expression and the abundance of microbial species in the feature model.

[0037] Figure 11Functional enrichment analysis of significantly related genes;

[0038] Figure 12 The difference in enrichment scores of immune infiltrating cells between high- and low-risk groups;

[0039] Figure 13 Differences in scores related to 13 immune functions between high- and low-risk groups;

[0040] Figure 14 To illustrate the differences in immune checkpoint expression between high- and low-risk groups;

[0041] Figure 15 External validation shows that the risk scoring model has good performance. Detailed Implementation

[0042] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0043] In the following examples, g represents the genus of microorganism.

[0044] Example 1: Collection and Data Analysis of Clinical Information from TCGA Gastric Cancer Samples

[0045] Data from 448 gastric cancer patients were collected from the TCGA database, and the collected data and analyses are summarized in Sections 1-7 below and Table 1:

[0046] 1. Microbiome abundance data:

[0047] https: / / ftp.microbio.me / pub / cancer_microbiome_analysis / TCGA / Kraken / Kraken-TCGA-Voom-SNM-Plate-Center-Filtering-Data.csv;

[0048] 2. Clinical data from the microbiome:

[0049] https: / / ftp.microbio.me / pub / cancer_microbiome_analysis / TCGA / Kraken / Kraken-TCGA-Raw-Data-17625-Samples.csv;

[0050] 3. Presenting data:

[0051] https: / / portal.gdc.cancer.gov / projects / TCGA-STAD;

[0052] 4. Survival data:

[0053] https: / / portal.gdc.cancer.gov / projects / TCGA-STAD;

[0054] 5. Phenotypic data:

[0055] https: / / portal.gdc.cancer.gov / projects / TCGA-STAD;

[0056] 6. Download the human GTF file (Homo_sapiens.GRCh38.104.gtf.gz) from the Ensembl database:

[0057] http: / / ftp.ensembl.org / pub / release-104 / gtf / homo_sapiens / Homo_ sapiens.GRCh38.104.gtf.gz ;

[0058] 7. The set of immune cell-related genes was downloaded from Table S6 of the literature PMID: 28052254, "List of Pan-cancer Immune Metagenes";

[0059] Table 1. Clinical information of gastric cancer samples from the TCGA database

[0060]

[0061] Example 2: Screening and identification of prognostic microbial characteristics of gastric cancer

[0062] Based on the microbial abundance information obtained from each gastric cancer patient sample in Example 1, the average abundance of species was displayed according to the kingdom, phylum, class, order, family, and genus levels. Due to the large number of species within each phylum, class, order, family, and genus, box plots were created using the top 20 abundant microorganisms from each level. The results showed that the proportions of each level were relatively evenly distributed in both gastric cancer and adjacent normal samples. Figure 1 ).

[0063] LEfSe analysis was performed on gastric cancer and adjacent normal samples. The analysis was primarily conducted using the R package microeco (v1.1.0). A significance threshold of alpha < 0.05 was used to identify gastric cancer-related characteristic microbial markers. The top 20 microbial LDA values ​​were then used to create an LDA statistical map of the characteristic microorganisms, followed by an evolutionary clade plot based on the top 150 most abundant microbial communities. The results showed a total of 683 significantly different microbial markers between gastric cancer and adjacent normal samples. Of these, 326 belonged to microorganisms with higher LDA in the gastric cancer group, while the remaining 357 belonged to microorganisms with higher LDA in the control group. Figure 2 The phylogenetic clade diagram shows that most bacteria in the kingdoms k_Archaea and k_Viruses play important roles in the gastric cancer group, while most bacteria in the kingdom k_Bacteria mainly play important roles in the control group. Figure 3 The 683 differential microbial markers here include those at the kingdom, phylum, class, order, family, and genus levels. After integration, a total of 463 differential microorganisms at the genus level were obtained.

[0064] Example 3: Construction and Validation of a Prognostic Feature Model for Gastric Cancer

[0065] 1. Univariate Cox regression:

[0066] Microbial abundance data were pooled according to the TCGA patient ID numbers (median value was used). Significant characteristic microorganisms obtained from LEfSe analysis were used for subsequent analyses. Gastric cancer samples from the TCGA-STAD data were split 2:1, with 67% of the data used for modeling (training set) and 33% for validation (validation set). The training set consisted of 298 samples, and the validation set consisted of 150 samples. Further, based on the abundance of 463 genus-level differentially expressed microbial markers, and combined with overall survival (OS) data, batch Cox univariate regression analysis was performed on the abundance values ​​of characteristic microbial markers in the training set samples using the R packages survival (v3.2-7) and survminer (v0.4.8). After regression analysis, characteristic microbial markers significantly associated with overall survival were screened using a p<0.05 threshold for subsequent analysis. The results showed that, based on the p<0.05 threshold, 20 microbial markers significantly associated with prognosis were identified. Figure 4 Of these, 10 had an HR > 1, indicating that an increase in microbial abundance may lead to a poor prognosis, while the remaining 10 had an HR < 1, indicating that a decrease in microbial abundance may lead to a poor prognosis. Further analysis was conducted using KM curves to analyze the top 6 microorganisms by p-value. Figure 5 ).

[0067] 2. LASSO Cox regression analysis:

[0068] Further dimensionality reduction using lasso regression was performed on prognostic-related microbial markers, reducing the number to 20 for constructing a risk scoring model. This process primarily relied on the R package glmnet (v4.0-2). To construct a more accurate regression model, lambda filtering was first performed using cross-validation. Then, the model corresponding to lambda.min was selected, and the abundance matrix of relevant microbial markers in the model was further extracted. The risk score for each sample was calculated based on the following formula:

[0069]

[0070] Where, abundance represents the abundance of the corresponding microbial marker, β represents the regression coefficient (coef) of the corresponding marker in the lasso regression result, RScore represents the sum of the abundance of the significantly relevant marker in each sample multiplied by the coef of the corresponding marker, i represents the sample, and j represents the microbial marker.

[0071] The final prognostic risk scoring model for microbial markers is: Risk Score = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola*0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_Thermosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7286)+g_Sulfitobacter*(-0.0 ... .3210)+g_Frigoribacterium*(-0.0740)+g_Flectobacillus*(-0.4914)+g_Serinicoccus*0.1775+g_Sphingobacterium*(-0.1 431)+g_Sandarakinorhabdus*0.0207+g_Verminephrobacter*(-0.1341)+g_Klebsiella*0.3162+g_Congregibacter*(-0.3978)

[0072] Using the median risk score obtained from the above formula as the dividing line, patients were divided into a high-risk group and a low-risk group. Survival analysis results showed that, in both the training and validation sets, the low-risk group had significantly better overall survival compared to the high-risk group, with statistically significant differences. This model can predict patient survival at different time points, and the AUC curve also demonstrates that the risk score can predict patient survival with high accuracy, further validating the effectiveness of our constructed prognostic model and highlighting the importance of tumor microbiota in patient prognostic assessment.

[0073] 3. Validate the effectiveness of the microbial marker prognostic risk scoring model using the training set:

[0074] Based on the risk scores of the training set samples, high-risk and low-risk groups were divided using the median as the node. KM curves were plotted using the overall survival and progression-free survival (DFS) data, and p-values ​​were calculated. A p-value < 0.05 was considered statistically significant for the high-risk and low-risk groups. Then, the sample risk scores were used as the model's prediction results, and the AUC value of the model was calculated in conjunction with survival data. ROC curves were then plotted. The curves showed that the AUC values ​​at 1 year, 3 years, and 5 years were all greater than 0.6, indicating good model performance. Figure 6 ).

[0075] 4. Validation set validation of the efficacy of the microbial marker prognostic risk scoring model:

[0076] The model's effectiveness was further validated on the validation set. Sample hazard scores were calculated using the formula, and based on these scores, high-risk and low-risk groups were divided using the median as the dividing line. Combined with overall survival data, KM curves were plotted. The survival curves of the two groups showed significant differences (p < 0.05), and the KM curves based on progression-free survival also showed significant differences. Then, the sample hazard scores were used as the model's prediction results, and the AUCs for 1-year, 3-year, and 5-year survival were calculated, all of which were greater than 0.6, indicating good model effectiveness. Figure 7 ).

[0077] 5. Construct a nomogram:

[0078] Further, nomograms were plotted using the Risk score, age, gender, and stage data from the training and validation sets to guide clinical practice, and ROC curves based on the nomograms were also generated. Nomograms visualize the results of Cox regression. They establish scoring criteria based on the magnitude of the regression coefficients of all independent variables, assigning a score to each value level of each independent variable. A total score is calculated for each patient, and the probability of the outcome time for each patient is calculated using a conversion function between the score and the probability of outcome occurrence. The nomogram plotting primarily utilizes the R packages rms (v6.1-0) and survival (v3.2-7). First, a Cox proportional hazards regression model is constructed using cph() with age, gender, and stage data. Then, the survival probability is calculated using the survival() function. Finally, the nomogram object is constructed using the nomogram() function and plot() is used for visualization. In addition, the C-index of the training and validation set data is calculated and statistical graphs are generated for display. The C-index of the noctilinear plot and Risk score were statistically analyzed in the training and validation sets respectively. The results show that the C-index of both models is greater than 0.5, indicating positive predictive power. Moreover, the C-index of the noctilinear plot is higher than that of the Risk score, indicating that the overall model has better prognostic performance. Figure 8 ).

[0079] Example 4: Clinical Relevance Study of Microbial Marker Prognostic Risk Scoring Model

[0080] 1. Validation of the correlation between the prognostic risk scoring model and patient prognosis and clinical characteristics: By combining validation and training set data, the differences in risk scores among age, gender, pathologic_M, pathologic_N, pathologic_T, and stage groups were statistically analyzed in the total data. The significance of the differences was detected using Wilcox test and Kruskal test, and box plots were drawn. Figure 9 ) will be displayed.

[0081] 2. Correlation between characteristic microbial populations and gene expression

[0082] (1) Correlation between gene expression and characteristic models including microbial species abundance:

[0083] Further, the expression matrix of protein-coding genes in the samples was obtained, and then the correlation between protein-coding gene expression and the abundance of characteristic microorganisms in the model was calculated. Correlation heatmaps were plotted and displayed according to the thresholds |correlation|>0.3 and p<0.01. First, we obtained 19597 protein-coding genes. Then, based on the p<0.01 correlation threshold, we obtained 689 combinations with significant correlations, including 573 protein-coding genes. The correlation heatmap showed that g_Klebsiella, g_Achromobacter, g_Thermosipho, and g_Methylopila had a higher proportion of positive correlations with the relevant genes, while g_Verminephrobacter, g_Uliginosibacterium, and g_Congregibacter had a higher proportion of negative correlations with the relevant genes. Figure 10 ).

[0084] (2) Gene function enrichment analysis

[0085] Further functional enrichment analysis was performed on 573 protein-coding genes significantly associated with microbial markers in the model using the R package clusterProfiler (v4.6.2). GO enrichment analysis was divided into three parts: biological processes, cellular components, and molecular functions. For genes, the enriched pathways in biological processes mainly included cytoplasmic translation, detection of chemical stimulus involved in sensory perception of smell, and sensory perception of smell. The enriched pathways in cellular components mainly included large ribosomal subunit, cytosolic ribosome, and respiratory chain complex. The enriched pathways in molecular functions mainly included structural constituent of ribosome, olfactory receptor activity, and aspartic-type endopeptidase activity. The enriched pathways in KEGG mainly included ribosome, cholesterol metabolism, and olfactory transduction. Figure 11 ).

[0086] 3. Immune cells, immune function, and prospects of immunotherapy related to prognostic risk scoring models.

[0087] (1) Display of differences in immune cell infiltration and immune function: The enrichment scores of 28 immune infiltrating cells in cancer samples were calculated using the R package GSVA (v1.34.0). After obtaining the results, the data were standardized using the scale() function, and then box plots were drawn for display of high and low risk groups. At the same time, the enrichment scores of immune function in the samples were calculated, and the data were standardized using the scale() function and then box plots were drawn for display. The results showed that there were differences in the proportion of 9 immune infiltrating cells in the high and low risk groups of microorganisms. The proportions of Activated CD4 T cells, Central memory CD4 T cells, and Effector memory CD8 T cells in the high-risk group of microorganisms were significantly higher than those in the low-risk group of microorganisms. The proportions of the remaining Activated B cells, Activated dendritic cells, Eosinophils, Immature dendritic cells, Natural killer cells, and T follicular helper cells in the low-risk group of microorganisms were significantly higher than those in the high-risk group of microorganisms. Figure 12 Further analysis revealed differences in scores for 13 immune functions between the high- and low-risk microbial groups. Only two immune function-related scores showed significant differences between the high- and low-risk groups: APC co-inhibition and CCR, both of which had higher enrichment scores in the low-risk group. Figure 13 ).

[0088] (2) Differences in immune checkpoint expression: Immune checkpoint information was obtained from the literature (PMID:32814346), and the differences in immune checkpoint expression between groups were statistically analyzed and plotted using box plots. The results showed that for 64 immune checkpoints, 3 showed significant differences between the high- and low-risk microbial groups, and all were significantly highly expressed in the low-risk microbial group, suggesting that immunotherapy may be more effective in low-risk microbial groups. Figure 14 ).

[0089] Example 5: External Data Validation of the Prognostic Risk Scoring Model

[0090] To further evaluate the predictive performance of the prognostic risk scoring model, we obtained two external validation patient datasets (PRJNA375772 and PRJNA532731) from the NCBI-SRA database. Based on the prognostic risk scoring model obtained in the aforementioned examples, we performed ROC analysis using the microorganisms selected in the model. The results showed that the AUC of the external validation set PRJNA375772 was 0.718, and the AUC of the validation set PRJNA532731 was 0.759. Figure 15Therefore, the results of external validation indicate that the prognostic risk scoring model constructed from 20 microbial markers has good predictive performance.

[0091] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A combination of microbial biomarkers for predicting the prognosis of gastric cancer, characterized in that, The microbial marker combination includes: Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgras sella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella and Congregibacter.

2. The microbial biomarker combination for predicting gastric cancer prognosis according to claim 1, characterized in that, The microbial biomarker combination consists of Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter.

3. The use of the reagent for detecting the abundance of each microorganism in the combination of microbial markers according to claim 1 or 2 in the preparation of a kit for predicting the prognosis of gastric cancer.

4. The application according to claim 3, characterized in that, The microorganisms were derived from tumor samples from gastric cancer patients.

5. A risk assessment method for predicting the prognosis of gastric cancer, characterized in that, include: Samples were collected from patients, and the abundance of various microorganisms in the samples was determined. These microorganisms included Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter. Based on the abundance data of microorganisms, a risk score is calculated according to a risk scoring model, which is: RiskScore = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola*0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_Thermosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7286)+g_Sulfitobacter*(-0. 3210)+g_Frigoribacterium*(-0.0740)+g_Flectobacillus*(-0.4914)+g_Serinicoccus*0.1775+g_Sphingobacterium*(-0.14 31)+g_Sandarakinorhabdus*0.0207+g_Verminephrobacter*(-0.1341)+g_Klebsiella*0.3162+g_Congregibacter*(-0.3978); Among them, Risk Score represents the risk score, and g_microbial Latin name represents the abundance of microorganisms of that genus in the patient sample; Patient prognosis is predicted based on risk scores; the lower the risk score, the better the prognosis.

6. The risk assessment method for predicting the prognosis of gastric cancer according to claim 5, characterized in that, The method of predicting patient prognosis based on risk scores includes: using the median risk score as a cutoff value; when the risk score is ≥ the median, the patient is classified as a high-risk group; when the risk score is < the median, the patient is classified as a low-risk group; compared with the high-risk group, the low-risk group has a significantly better overall survival.

7. A system for predicting the prognosis of gastric cancer, characterized in that, The system includes the following modules: 1) Data collection module: Collects patient samples and measures the abundance of various microorganisms in the samples, including Cobetia, Hyalangium, Oceanicola, Veillonella, Uliginosibacterium, Belliella, Achromobacter, Thermosipho, Methylopila, Sulfobacillus, Snodgrassella, Sulfitobacter, Frigoribacterium, Flectobacillus, Serinicoccus, Sphingobacterium, Sandarakinorhabdus, Verminephrobacter, Klebsiella, and Congregibacter; 2) Model Calculation Module: Based on microbial abundance data, a risk score is calculated according to a risk scoring model. The risk scoring model is: Risk Score = g_Cobetia*0.3622+g_Hyalangium*0.5966+g_Oceanicola*0.1391+g_Veillonella*0.3182+g_Uliginosibacterium*(-0.4405)+g_Belliella*0.3143+g_Achromobacter*(-0.1845)+g_Thermosipho*(-0.2381)+g_Methylopila*0.3131+g_Sulfobacillus*0.6303+g_Snodgrassella*(-0.7 286)+g_Sulfitobacter*(-0.3210)+g_Frigoribacterium*(-0.0740)+g_Flectobacillus*(-0.4914)+g_Serinicoccus*0.1775+g_Sphingobac terium*(-0.1431)+g_Sandarakinorhabdus*0.0207+g_Verminephrobacter*(-0.1341)+g_Klebsiella*0.3162+g_Congregibacter*(-0.3978); 3) Output prediction module: Predicts the patient's prognosis based on the risk score. The lower the patient's risk score, the better the prognosis.

8. A reagent kit, characterized in that, Includes reagents for detecting the abundance of each microorganism in the microbial biomarker combination of claim 1 or 2.

Images