Intratumor microbial marker group for gastrointestinal tumor prognosis, prognosis model and application

Abstract

The invention discloses an intratumor microbial marker group for gastrointestinal tumor prognosis, a prognosis model and application. The biomarker group comprises the following components: Granulicella, Thermodesuftator, Gaetbus, Formosa, Kaistobacter, Candidate Saccharimonas, Dactyloccopsis, Schwartzia, Candidate Kuenia, Sphingosinusus, Microvirgula, Simple, Doria, Halothiobacillus, Rhodobacter, and Oceaniterimus, and is characterized in that the biomarker group comprises the following components: the Granulicella, the Thermodesuftator, the Gaetbus, the Formosa, the Kaistobacter, the Candidate Saccharimonas, the Dactyloccopsis, the Schwartzia, the Candidate Saccharimonas, the The model constructed by the method can accurately predict prognosis, drug response and immunotherapy.

Description

Technical Field

[0001] The present invention relates to the field of tumor prognostic markers, and particularly to an intratumoral microbial marker group, a prognostic model and an application for the prognosis of gastrointestinal tumors. Background Art

[0002] The survival rate of patients with gastrointestinal tumors is closely related to early diagnosis, but the existing diagnostic methods lack sufficient efficiency and accuracy. Especially for advanced gastrointestinal cancers, as a highly heterogeneous cancer, traditional chemotherapy usually has poor effects.

[0003] Microbial communities are widely present in multiple niches of the human body, including the gut, skin and even tumor tissues, and play a key role in regulating the host immune system and influencing treatment responses. With the progress of detection technologies and the in-depth study of the tumor microbiome, it has been found that there are significant differences in the composition and abundance of microbiota between tumor tissues and normal tissues. For example, Chlamydia, Mycoplasma, Acinetobacter and Brucella in ovarian cancer; an unknown species in the genus Porphyromonas in endometrial cancer, and the abundance of these specific microbiota is higher than that of normal tissues. These tumor-specific microbiome features have the potential to serve as diagnostic and prognostic indicators to complement the deficiencies of traditional gene prediction models.

[0004] The gastrointestinal tract provides an ideal habitat for microorganisms due to its extensive mucosal surface area, and its abundance in gastrointestinal tumors is higher than that in other types of cancers. This high microbial abundance makes it possible to develop a microbial-based prognostic model. However, the development of gastrointestinal tumor-specific biomarkers (including microbiome, metabolites, etc.) lags behind, and the existing biomarkers are difficult to achieve early diagnosis, precise stratification and dynamic efficacy monitoring. Therefore, it is of great significance to develop new microbiome-based diagnostic tools, prognostic models and adjuvant treatment means. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides an intratumoral microbial marker group, a prognostic model and an application for the prognosis of gastrointestinal tumors. The present invention can accurately predict tumor metastasis, drug response and immunotherapy, providing a new method for precision cancer treatment and improving the ability of diagnosis, patient stratification and prediction of treatment response.

[0006] In a first aspect, the present invention provides an intratumoral microbial marker group for evaluating the prognosis of gastrointestinal tumors, which is achieved by the following technical solutions.

[0007] An intratumoral microbial biomarker panel for evaluating the prognosis of gastrointestinal tumors, which consists of the following bacterial genera: Granulicella, Thermodesulfatator, Gaetbulibacter, Formosa, Kaistobacter, Candidatus Saccharimonas, Dactylococcopsis, Schwartzia, Candidatus Kuenenia, Sphingosinicella, Microvirgula, Simplicispira, Dorea, Halothiobacillus, Rhodobacter, Oceanithermus.

[0008] Furthermore, the calculation formula for the microbial biomarker panel to evaluate the prognosis of gastrointestinal tumors is: Risk score = (-0.121) * Abundance Dorea + 0.057 * Abundance Granulicella + 0.033 * Abundance T h ermo d esu lf atator + 0.573 * Abundance G aet b u l i b acter + 0.376 * Abundance Formosa + 0.224 * Abundance Kaistobacter + 0.060 * Abundance Candidatus Saccharimonas + 0.170 * Abundance Dacty l ococcopsis + 0.460 * Abundance S c h wartzia + 0.128 * Abundance C an d i d atus Kuenenia + 0.200 * Abundance Sphingosinicella + 0.181 * Abundance Microvirgula + (-0.069) * Abundance Ha l ot h io b aci ll us + 0.186 * Abundance S imp l icispira+(-0.053)*Abundance R h o d o b acter +(-0.053)*Abundance Oceanithermus ; wherein, Abundance represents the abundance of each genus of microorganisms in the tumor sample.

[0009] In a second aspect, the present invention provides a use of an intratumoral microbial biomarker panel for evaluating the prognosis of gastrointestinal tumors, which is achieved through the following technical solutions.

[0010] Application of a reagent for detecting the abundance of each genus of microorganisms in the above-mentioned microbial biomarker panel in the preparation of a product for evaluating the prognosis of gastrointestinal tumors.

[0011] In a third aspect, the present invention provides a model for evaluating the prognosis of gastrointestinal tumors, which is achieved through the following technical solutions.

[0012] A model for evaluating the prognosis of gastrointestinal tumors, the prognosis model is constructed based on the above-mentioned microbial biomarker panel, and the abundance of each genus of microorganisms in the tumor sample is used as an input variable, and the calculation formula is: Risk score

[0013] = (-0.121)*Abundance Dorea +0.057*Abundance Granulicella +0.033*Abundance Thermodesulfatator +0.573*Abundance Gaetbulibacter +0.376*Abundance Formosa +0.224*

[0014] Abundance Kaistobacter +0.060*Abundance Candidatus Saccharimonas +0.170*Abundance Dactylococcopsis +0.460*Abundance Schwartzia +0.128*Abundance Candidatus Kuenenia +0.200*Abundance S p h ingosinice ll a +0.181*Abundance Microvirgu l a +(-0.069)*Abundance Halothiobacillus +0.186*Abundance Simplicispira+(-0.053)*Abundance Rhodobacter +(-0.053)*Abundance Oceanithermus ; wherein, Abundance represents the abundance of each genus of microorganisms in tumor samples.

[0015] Fourthly, the first use of a model for evaluating the prognosis of gastrointestinal tumors provided by the present invention is achieved through the following technical solutions.

[0016] The application of the above model in the preparation of a product for predicting the risk of metastasis of gastrointestinal tumors.

[0017] Fifthly, the second use of a model for evaluating the prognosis of gastrointestinal tumors provided by the present invention is achieved through the following technical solutions.

[0018] The application of the above model in the preparation of a product for predicting the efficacy of drugs for gastrointestinal tumors.

[0019] Sixthly, the third use of a model for evaluating the prognosis of gastrointestinal tumors provided by the present invention is achieved through the following technical solutions.

[0020] The application of the above model in the preparation of a product for predicting the efficacy of immunotherapy for gastrointestinal tumors.

[0021] Seventhly, the present invention provides a computer-readable storage medium, which is achieved through the following technical solutions.

[0022] A computer-readable storage medium stores a computer program, and when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the above model.

[0023] Eighthly, the present invention provides a system for evaluating the prognosis of gastrointestinal tumors, which is achieved through the following technical solutions.

[0024] A system for evaluating the prognosis of gastrointestinal tumors, the system includes a calculation unit, and the calculation unit calculates a score using the above model.

[0025] This application has the following beneficial effects.

[0026] Based on the intratumoral microbiome data of the BIC database, the present invention comprehensively analyzes the intratumoral microbiome of gastrointestinal tumors, identifies 16 core bacterial genera as prognostic factors, and develops a prognostic microbial model for predicting gastrointestinal cancers. At the same time, through a combined analysis with the transcriptome data of the TCGA database, it is determined that the microorganisms in this model can predict the prognosis and metastasis risk of tumors, as well as the drug response of tumor treatment and the immune response of immune cells. This model can assist in diagnosis, drug selection, and immunotherapy, and formulate more effective treatment plans for patients. Brief Description of the Drawings

[0027] Figure 1 It is a diagram of the collection and data analysis of the clinical information of the gastrointestinal cancer samples of the present invention;

[0028] Figure 2 It is a diagram of the LASSO regression analysis (a and b) and the multivariate Cox regression analysis (c) of the present invention;

[0029] Figure 3 It is a prediction effect diagram of the prognostic model verified by the training set of the present invention (wherein, a. Risk score curve, survival status diagram; b. Kaplan-Meier curve; c. ROC curve shows the prediction of the 1-year, 3-year, and 5-year survival rates of gastrointestinal tumor patients by the prognostic model);

[0030] Figure 4 It is an ROC curve diagram of the validation set of the present invention (wherein, a. CHOL cholangiocarcinoma; b. COAD colon adenocarcinoma; c. ESCA esophageal cancer; d. LIHC hepatocellular carcinoma; e. PAAD pancreatic adenocarcinoma; f. STAD gastric adenocarcinoma);

[0031] Figure 5 It is a gene set enrichment analysis of the present invention: using the Hallmark gene set to identify differentially expressed genes between high- and low-risk patients;

[0032] Figure 6 It is a gene set enrichment analysis of the present invention: using the Drug Gene Interaction Database (DGIdb) to identify differential genes between high- and low-risk patients;

[0033] Figure 7 It is a molecular docking analysis diagram of SNAI1 with two drugs, XL999 (a) and Tandutinib (b), of the present invention;

[0034] Figure 8 It is a diagram of the abundance of immune cells in tumor tissues of the high-risk group and the low-risk group of the present invention;

[0035] Figure 9It is a diagram for analyzing the TIDE score difference between the high-risk group and the low-risk group scores of the present invention (where a. TIDE value; b. dysfunction value; c. exclusion value;

[0036] Figure 10 It is a correlation diagram between the microbial abundance and the infiltration scores of 28 immune cells in the heatmap display model of the present invention;

[0037] Figure 11 It is a correlation diagram between the microbial abundance and the enrichment score of the immunotherapy prediction pathway in the heatmap display model of the present invention;

[0038] Figure 12 It is a correlation analysis diagram between the abundances of Dorea (a) and Granulicella (b) and the risk score of the present invention. Detailed implementation manners

[0039] To identify the microbiome features associated with gastrointestinal tumorigenesis, the present application analyzed the microbiome data of six gastrointestinal tumor tissues (cholangiocarcinoma, colon cancer, esophageal cancer, hepatocellular carcinoma, pancreatic cancer, and gastric adenocarcinoma) and their adjacent normal tissues. At the genus level, the abundances of 1,163 bacteria were recorded in total. Through univariate Cox regression analysis, 131 bacterial genera significantly associated with patient prognosis were identified (p < 0.05). Subsequently, LASSO regression analysis was used to reduce the number of candidate bacterial genera, and 57 genera were screened out. Further, a core microbiome consisting of 16 bacterial genera was determined through multivariate Cox regression analysis, including 12 high-risk genera and 4 low-risk genera, and these bacterial genera can serve as independent prognostic factors. Based on the risk coefficients of these 16 bacterial genera, a prognostic model based on the intratumoral microbiome was constructed, and its equation is as follows:

[0040] Risk score = (-0.121) * Abundance Dorea + 0.057 * Abundance Granulicella + 0.033 *

[0041] Abundance T h ermo d esu lf atator + 0.573 * Abundance G aet b u l i b acter + 0.376 * Abundance Formosa + 0.224 * Abundance Kaistobacter + 0.060 * Abundance Candidatus Saccharimonas+0.170*Abundance Dacty l ococcopsis +0.460*Abundance S c h wartzia +0.128*Abundance C an d i d atus Kuenenia +0.200*Abundance S p h ingosinice ll a +0.181*Abundance Microvirgu l a +(-0.069)*Abundance Halothiobacillus +0.186*Abundance Simplicispira +(-0.053)*Abundance Rhodobacter +(-0.053)*Abundance Oceanithermus

[0042] Among them, Risk Score represents the risk score, Abundance represents the abundance of each genus of microorganisms in tumor samples, and the median of the risk score is used as the division threshold to judge whether the patient belongs to the high-risk group (High-risk) or the low-risk group (Low-risk). The overall survival of patients in the low-risk group is significantly better and the prognosis is better. The prediction process is as follows:

[0043] 1) Data collection: Collect samples from patients and measure the abundance of the above 16 bacterial genera in their samples;

[0044] 2) Model calculation: Calculate the total risk score according to the risk score model;

[0045] 3) Output prediction: Predict the prognosis of patients based on the risk score data of gastrointestinal tumor patients. The lower the risk score of the patient, the better the prognosis.

[0046] The invention will be further described below in conjunction with the accompanying drawings and embodiments. Unless otherwise specified, the experimental methods used in the present invention are conventional methods, and the experimental equipment, materials, reagents, etc. used can be purchased from relevant material sales companies.

[0047] In the following experiments of this application, all statistical analyses were completed using R software (version 4.2.3). A two-tailed test was used to determine statistical significance, **** represents p < 0.0001, *** represents p < 0.001, ** represents p < 0.01, * represents p < 0.05, and ns represents no significant difference. The Wilcoxon test was used to compare the ssGSEA scores between the high and low groups for the differences between two groups. The relationship between non-normally distributed quantitative variables was evaluated by Spearman correlation analysis. Survival curve analysis was performed using the Kaplan-Meier method and the Cox proportional hazards regression model, and the differences were evaluated by the log-rank test. Survival and regression analyses were performed using the R software packages Survival (version 3.5.5) and survminer (version 0.4.9), and the ggplot2 R package (version 3.5.1) was used to visualize the survival curves.

[0048] Example 1: Construction and detection of a microbiome-based prognostic model for gastrointestinal tumors

[0049] 1. Collection of clinical information and data analysis of gastrointestinal cancer samples

[0050] In this application, the microbiota composition data (at the genus level) of six types of gastrointestinal cancers (CHOL, cholangiocarcinoma; COAD, colon adenocarcinoma; ESCA, esophageal cancer; LIHC, hepatocellular carcinoma; PAAD, pancreatic adenocarcinoma; STAD, gastric adenocarcinoma) were retrieved from the BIC database (http: / / bic.jhlab.tw / ), including 1,602 tumor tissue samples and 116 adjacent normal tissue samples ( Figure 1)。The corresponding transcriptome data and patient clinical information were obtained from The Cancer Genome Atlas (https: / / portal.gdc.cancer.gov / ). First, bacterial genera were evaluated by univariate Cox regression analysis, and 131 bacterial genera significantly associated with patient prognosis were identified (p<0.05). Subsequently, 74 genera were excluded based on correlation using LASSO regression analysis, and 57 bacterial genera were screened out. Finally, 16 bacterial genera were further identified as independent prognostic factors using multivariate Cox regression analysis, including 12 high-risk genera, namely Granulicella, Thermodesulfatator, Gaetbulibacter, Formosa, Kaistobacter, Candidatus Saccharimonas, Dactylococcopsis, Schwartzia, Candidatus Kuenenia, Sphingosinicella, Microvirgula, Simplicispira; and 4 low-risk genera, namely Dorea, Halothiobacillus, Rhodobacter, Oceanithermus( Figure 2 )In addition, age was included as a covariate in the prognostic model in this application to construct a microbiota-based prognostic model. The risk score for each patient in the TCGA database was calculated using the following formula:

[0051] Risk score=Σ(coefficient genusn ×Abundance level genusn )

[0052] where Risk score represents the risk score, coefficient represents the risk coefficient of each bacterial genus, and Abundance level represents the abundance of the microbiota.

[0053] The finally obtained prognostic risk score model for the microbial marker is:

[0054] Risk score=(-0.121)*Abundance Dorea +0.057*Abundance Granulicella +0.033*

[0055] Abundance T h ermo d esu lf atator +0.573*Abundance G aet b u l i b acter +0.376*Abundance Formosa +0.224*Abundance Kaistobacter +0.060*Abundance CandidatusSaccharimonas +0.170*Abundance Dacty l ococcopsis +0.460*Abundance S c h wartzia +0.128*Abundance C an d i d atus Kuenenia +0.200*Abundance Sphingosinicella +0.181*Abundance Microvirgula +(-0.069)*Abundance Ha l ot h io b aci ll us +0.186*Abundance S imp l icispira +(-0.053)*Abundance Rhodobacter +(-0.053)*Abundance Oceanithermus .

[0056] Among them, Risk Score represents the risk score, Abundance represents the abundance of each genus of microorganisms in tumor samples, and the median of the risk score is used as the division threshold to determine whether the patient belongs to the high-risk group (High-risk) or the low-risk group (Low-risk). The overall survival of patients in the low-risk group is significantly better and the prognosis is better.

[0057] 2. Validation and evaluation of the prediction efficiency of the prognostic model of the microbiome

[0058] To evaluate the ability of the microbiome prognostic signature to predict patient survival at different time points, this application uses the Survival ROC R package to plot the time-dependent ROC (Receiver Operating Characteristic) curve and calculate the AUC (area under the curve). First, the risk score, patient survival time, and survival status (0 = alive, 1 = dead) are extracted from the sample data, and the AUC is calculated at 1 year, 3 years, and 5 years respectively to judge the prediction accuracy of the model at different stages. The closer the AUC value is to 1.0, the stronger the prediction ability; if the AUC > 0.7, it indicates that the microbiome prognostic model has good discriminatory ability at this time point. Kaplan-Meier survival analysis and the two-sided Log-rank test are used for statistical testing of the survival differences between high-risk and low-risk groups of patients. The patients are divided into high-risk and low-risk groups according to the median of the risk score, the Kaplan-Meier survival curve is plotted, and the change trend of the survival probability of the two groups of patients over time is compared. The two-sided Log-rank test is used to calculate the p-value. If p < 0.05, it indicates that there is a significant difference in the survival time between the high- and low-risk groups. The pheatmap is used to plot the risk curve to observe the change trend of the individual risk scores of the patients. The horizontal axis is the order of all patients sorted from low to high according to the risk score, and the vertical axis is the risk score value, showing a trend of increasing high-risk patients. If the curve shows an obvious upward trend, it indicates that the risk score can effectively distinguish between high- and low-risk patients and is related to the survival outcome.

[0059] Using the gastrointestinal cancer cohort (RNA sequencing data selected from the TCGA database) as the training dataset (n = 1718), the risk score of each patient was calculated and plotted. The results showed that the death risk of patients with high risk scores (n = 368) was significantly higher than that of patients with low risk scores (n = 1350), and the survival analysis further confirmed the poor prognosis of high-risk patients. At the same time, the AOC value of the ROC curve was approximately 0.7, indicating that the risk score could predict the patient survival rate with relatively high accuracy ( Figure 3 ).

[0060] To further verify the prediction performance of the model, it was tested in six independent validation cohorts, including CHOL (n = 45), COAD (n = 441), ESCA (n = 194), LIHC (n = 416), PAAD (n = 182), and STAD (n = 440). The results showed that the high-risk groups in each validation cohort showed significantly higher death risks and shorter overall survival times, in sharp contrast to the low-risk groups ( Figure 4 ), which was consistent with the results of the training cohort. The above results indicate that significant microbiome characteristics are closely related to patient prognosis, and the model developed in the present invention can accurately predict the survival and prognosis of gastrointestinal tumor patients.

[0061] Example 2: Predicting the risk of tumor metastasis based on the microbiome prognosis model

[0062] 1. Differential gene functional enrichment analysis

[0063] A total of 1,718 gastrointestinal cancer samples (including 45 cholangiocarcinoma samples, 441 colon adenocarcinoma samples, 194 esophageal cancer samples, 416 hepatocellular carcinoma samples, 182 pancreatic adenocarcinoma samples, and 440 gastric adenocarcinoma samples) were retrieved from the transcriptome sequencing data of the TCGA database. The risk scores were calculated, and the samples were divided into high-risk and low-risk groups according to the median score. The DESeq2 R package (version 1.36.0) was used to standardize the RNA-Seq reads, correct for sequencing depth and differences between samples, and make the data comparable between different samples. DESeq2 was used to calculate the differentially expressed genes (DEGs) between the high- and low-risk groups, and the screening criteria were: adjusted p-value < 0.05 (controlling the false discovery rate); |log2FoldChange| > 1 (gene expression fold change ≥ 2-fold), obtaining lists of genes that were significantly upregulated (log2FC > 1) and significantly downregulated (log2FC < -1), representing genes that were more active in the high-risk or low-risk groups, respectively. Compared with the low-risk group, a total of 459 differentially expressed genes were screened in the high-risk group, including 105 upregulated genes and 354 downregulated genes. Next, the clusterProfiler R package (version 4.4.4) was used for GSEA (Gene Set Enrichment Analysis), and functional enrichment analysis was performed based on the Hallmark gene set (MSigDB) to identify significantly enriched tumorigenesis signaling pathways in the high-risk group.

[0064] 2. Relationship between risk score and risk of tumor metastasis

[0065] GSEA analysis of differentially expressed genes between the high- and low-risk score groups showed that key tumorigenesis signaling pathways such as KRAS and TGF-β in the high-risk score group were significantly activated, and significant activation of the epithelial mesenchymal transition pathway, angiogenesis pathway, and coagulation pathway ( Figure 5 ), indicating that the microbiota of the prognosis model was involved in tumorigenesis and metastasis, and tumors in high-risk individuals had a higher risk of metastasis.

[0066] Example 3: Predicting the therapeutic response of anticancer drugs based on the microbiome prognosis model

[0067] 1. Drug response prediction

[0068] Drug-gene interaction data were retrieved from the Drug-Gene Interaction Database (DGIdb, https: / / dgidb.org / ). A total of 4,870 genes targeted by 81,948 tumor-suppressive drugs were screened, and a drug response gene set was constructed based on this. The gene set was used for Gene Set Enrichment Analysis (GSEA) of differentially expressed genes between the high- and low-risk groups to predict the therapeutic effects of potential drugs. A total of 20 drugs targeting gastrointestinal tumors were screened, including 2 drugs targeting high-risk populations and 18 drugs targeting low-risk populations.

[0069] 2. Interaction analysis of anti-cancer drugs and ligand molecules

[0070] The three-dimensional structure of the receptor protein was predicted by the AlphaFold server (https: / / alphafoldserver.com / ), and the three-dimensional structure of the ligand drug was downloaded from the PubChem database (https: / / pubchem.ncbi.nlm.nih.gov / ). The receptor and ligand files were preprocessed using PyMol software (version 3.0) and converted into PDB format. The Diffdock tool (https: / / huggingface.co / spaces / reginabarzilaygroup / DiffDock-Web) was used for molecular docking of the PDB files of the protein and the drug, and the molecular docking results were visualized using PyMol. The molecular binding energy was calculated using the PRODIGY tool (https: / / rascar.science.uu.nl / prodigy / ).

[0071] 3. Relationship between prognostic risk score and anti-cancer drugs

[0072] The results of GSEA analysis of differentially expressed drug target genes between the high- and low-risk score groups showed that the target genes of some anti-cancer drugs (such as XL999 and tandutinib) were highly expressed in the high-risk score group, indicating that they may be potential treatment options for high-risk patients with gastrointestinal tumors; in contrast, the target genes of common anti-cancer drugs (such as apigenin and lapatinib) were significantly low expressed in the high-risk group ( Figure 6 ), suggesting that these drugs may have poor efficacy in patients with high-risk scores. In addition, molecular docking analysis showed that XL999 and tandutinib were able to bind to SNAI1 (SNAI1 is a key transcription factor regulating epithelial-mesenchymal transition), and the binding affinity of both drugs to the SNAI1 protein was less than 5 kcal / mol, supporting the feasibility of their potential treatment options ( Figure 7) The above results indicate that the microbiome-based prognostic model has the potential to guide clinical drug selection and can provide more precise treatment regimens for patients.

[0073] Example 4: Prediction of the response to immunotherapy by a microbiome-based prognostic model

[0074] 1. Analysis of immune cell infiltration in the tumor microenvironment (TME)

[0075] The Gene Set Variation Analysis (GSVA) R package (version 1.46.0) was used to convert the gene expression data matrix into gene set enrichment scores to evaluate pathway activity. The single-sample gene set enrichment analysis (ssGSEA) algorithm was used to calculate the enrichment scores of immune cells in the TME. Based on the immune-related gene expression profiles of 28 immune cell types (Reference: Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype 540 Relationships and Predictors of Response to Checkpoint Blockade. Cell Rep. 2019, 18: 248-262.), the relative proportions of tumor-infiltrating lymphocytes (TILs) in each cancer type were inferred. The association between immune cell abundance and 16 bacterial genera was evaluated by Spearman correlation test.

[0076] 2. Tumor immune dysfunction and exclusion (TIDE)

[0077] TIDE is a computational framework that predicts tumor immune escape (including T cell dysfunction and T cell exclusion) through the gene expression profiles of cancer samples, thereby judging the response to immune checkpoint blockade (ICB) therapy. This software can be freely obtained through http: / / tide.dfci.harvard.edu. Higher TIDE, Dysfunction, and Exclusion values indicate a higher likelihood of tumor immune escape and a poorer response of gastrointestinal cancer patients to ICB therapy. In addition, the TIDE algorithm can also calculate the genomic microsatellite instability (MSI) value, and higher MSI values are usually associated with better prognosis.

[0078] 3. Immune cells, immune functions, and immunotherapy prospects related to the prognostic risk score model

[0079] The abundances of 28 immune cell types in tumor tissues of the high- and low-risk groups were analyzed by ssGSEA. The results showed that central memory CD4 in the high-risk group +T cell, effector memory CD8 + The abundances of T cell, natural killer cell, natural killer T cell, plasmacytoid dendritic cell, and type 2 T helper cell were significantly higher than those in the low-risk group; the abundances of activated CD8 T + cell, CD56bright natural killer cell, γδ T cell, memory B cell, and monocyte in the low-risk group were significantly higher than those in the high-risk group (p < 0.0001).( Figure 8 ) The above results indicate that the microbiota affects the tumor immune microenvironment and the anti-tumor immune response.

[0080] The results of TIDE analysis showed that the TIDE value, dysfunction value, and exclusion value in the low-risk score group were significantly lower than those in the high-risk score group( Figure 9 ), indicating that patients in the high-risk score group had a poor prognosis for immunotherapy. The above results indicate that the microbiome-based model can predict the response to immunotherapy, and high-risk patients have a poor effect on immunotherapy.

[0081] The results of the correlation analysis between 16 bacterial genera and the abundances of 28 tumor-infiltrating immune cell types and 50 typical biological process signaling pathways showed that Dorea and Granulicella were significantly correlated with immune cells and biological signaling pathways( Figure 10 ). Among them, Dorea was significantly positively correlated with activated CD8+ T cells, while Granulicella was negatively correlated with activated CD8+ T cells. In addition, Granulicella was significantly positively correlated with the tumor metastasis signaling pathway, while Dorea showed the opposite trend( Figure 11 ), indicating that Granulicella plays a more important role in tumor metastasis. The risk score was positively correlated with the abundance of Granulicella and negatively correlated with the abundance of Dorea( Figure 12 ). These results indicate that Granulicella may promote the poor immunotherapy response in high-risk patients because its abundance is closely related to the reduced infiltration of CD8+ T cells in tumor tissues and a higher risk of metastasis.

[0082] The embodiments of this specific implementation manner are all preferred embodiments of the present invention, and do not limit the protection scope of the present invention accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of the present invention should be covered within the protection scope of the present invention.

Claims

1. A panel of intratumoral microbial markers for evaluating the prognosis of gastrointestinal tumors, characterized in that: Consists of the following bacterial genera: Granulicella, Thermodesulfatator, Gaetbulibacter, Formosa, Kaistobacter, Candidatus Saccharimonas, Dactylococcopsis, Schwartzia, Candidatus Kuenenia, Sphingosinicella, Microvirgula, Simplicispira, Dorea, Halothiobacillus, Rhodobacter, Oceanithermus.

2. The intratumoral microbial marker panel for evaluating the prognosis of gastrointestinal tumors according to claim 1, characterized in that: The calculation formula for the microbial marker group to evaluate the prognosis of gastrointestinal tumors is: Risk score = (-0.121) * Abundance Dorea +0.057*Abundance Granulicella +0.033*Abundance Thermodesulfatator +0.573*Abundance Gaetbulibacter +0.376*Abundance Formosa +0.224*Abundance Kaistobacter +0.060*Abundance Candidatus Saccharimonas +0.170*Abundance Dactylococcopsis +0.460*Abundance Schwartzia +0.128*Abundance Candidatus Kuenenia +0.200*Abundance Sphingosinicella +0.181*Abundance Microvirgula +(-0.069)*Abundance Halothiobacillus +0.186*Abundance Simplicispira +(-0.053)*Abundance Rhodobacter +(-0.053)*Abundance Oceanithermus ; Among them, Abundance represents the abundance of each genus of microorganisms in tumor samples.

3. Use of a reagent for detecting the abundance of each genus of microorganisms in the microbial marker group described in claim 1 or 2 in a tumor sample in the preparation of a gastrointestinal tumor prognosis assessment product.

4. A model for evaluating the prognosis of gastrointestinal tumors, characterized in that: The prognostic model is constructed based on the microbial marker group described in claim 1, and the abundance of each genus of microorganisms in the tumor sample is used as an input variable. The calculation formula is: Riskscore = (-0.121) * Abundance Dorea +0.057*Abundance Granulicella +0.033*Abundance Thermodesulfatator +0.573*Abundance Gaetbulibacter +0.376*Abundance Formosa +0.224*Abundance Kaistobacter +0.060*Abundance Candidatus Saccharimonas +0.170*Abundance Dactylococcopsis +0.460*Abundance Schwartzia +0.128*Abundance Candidatus Kuenenia +0.200*Abundance Sphingosinicella +0.181*Abundance Microvirgula +(-0.069)*Abundance Halothiobacillus +0.186*Abundance Simplicispira +(-0.053)*Abundance Rhodobacter +(-0.053)*Abundance Oceanithermus ; Among them, Abundance represents the abundance of each genus of microorganisms in tumor samples.

5. Use of the model according to claim 4 in preparing a product for predicting the risk of gastrointestinal tumor metastasis.

6. Use of the model according to claim 4 in preparing a product for predicting the efficacy of a gastrointestinal tumor drug.

7. Use of the model according to claim 4 in preparing a product for predicting the efficacy of immunotherapy for gastrointestinal tumors.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the model described in claim 4.

9. A system for assessing the prognosis of gastrointestinal tumors, characterized in that: The system comprises a calculation unit, which calculates the score using the model of claim 4.

Images