Microbial marker composition for predicting prognosis of hepatocellular carcinoma and application thereof

Abstract

The invention relates to the field of medical detection, in particular to a microbial marker composition for predicting prognosis of hepatocellular carcinoma and application of the microbial marker composition. Advanced bioinformatics analysis, including linear discriminant analysis effect size and Cox proportional risk regression, is utilized to determine the correlation between specific microorganisms and clinical results of HCC patients, and further explore the relationship between the spectrum of microorganisms in tumor tissues and gene expression so as to clarify the potential approach that the microorganisms may affect HCC progress; the microbe prognosis characteristic model is developed by utilizing the analysis so as to improve the prediction on the survival of the patient, the influence of the related microbiota in the HCC tumor on the immune condition is also explored, and information can be provided for formulating an immunotherapy strategy. According to the invention, by comprehensively analyzing the attributes and prognosis of the microorganisms in the HCC, the understanding of the tumor microenvironment of the HCC is enhanced, so that the formulation of a more accurate and personalized cancer management strategy is promoted.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine technology, and in particular relates to a microbial marker composition for predicting the prognosis of hepatocellular carcinoma and an application thereof. Background Art

[0002] Hepatocellular carcinoma (HCC) is one of the major malignant tumors in the world and the leading cause of cancer-related death. As the third leading cause of cancer death in the world, HCC has a low 5-year relative survival rate of about 18%. Although significant progress has been made in treatments such as surgery, radiotherapy, chemotherapy, immunotherapy and targeted therapy, the long-term prognosis of HCC patients is still not optimistic, mainly due to the high metastasis and recurrence rates of HCC. Recent studies have emphasized the superiority of bacteria as the main microorganisms in the human gastrointestinal tract and clarified their key role in regulating the liver microenvironment, which plays a vital role in the prevention, diagnosis and treatment of liver cancer. In addition, clinical trials have also provided convincing evidence that short-term probiotic supplementation can effectively improve liver injury indicators by restoring the intestinal microbiota. Recent evidence also shows that the microbiota in the liver cancer microenvironment can be used as an independent biomarker to predict the postoperative outcome of liver cancer patients. However, there are few reports on the microorganisms in hepatocellular carcinoma, and this field urgently needs to be explored. Summary of the invention

[0003] Purpose of the invention: In order to solve the problems existing in the above-mentioned prior art, the present invention provides a microbial marker composition for predicting the prognosis of hepatocellular carcinoma and its application.

[0004] Technical solution: The present invention provides a microbial marker composition for predicting the prognosis of hepatocellular carcinoma, wherein the microbial marker composition comprises: Candidatus_Glomeribacter, Sicinivirus, Shuttleworthia, Leucothrix, Schleiferia, Dichelobacter, Salinimicrobium, T4likevirus, Ictalurivirus, Sphaerotilus, Kitasatospora, and Succinimonas.

[0005] The present invention also provides a detection reagent, which takes the abundance of the above microbial markers as the detection purpose.

[0006] The present invention also provides a detection product, comprising the above detection reagent and an acceptable auxiliary agent, carrier or device.

[0007] The present invention also provides an evaluation method for predicting the prognosis of hepatocellular carcinoma, wherein the abundance of the above-mentioned microorganisms is weighted and summed to construct a risk model to obtain a risk score; then the median of the risk model is used as the critical value, and if the risk score is greater than or equal to the critical value, it is considered to be high risk, otherwise it is considered to be low risk.

[0008] Furthermore, the risk model is expressed as:

[0009] RiskScore=Candidatus_Glomeribacter*(-0.1772)+Sicinivirus*(-0.1386)+Shuttleworthia*(-0.1212)+Leucothrix*(-0.0817)+Schleiferia*(-0.0791)+Dichelobac ter*(-0.0460)+Salinimicrobium*(-0.0334)+T4likevirus*(-0.0332)+Ictalurivirus*(-0.0035)+Sphaerotilus*0.0003+Kitasatospora*0.0269+Succinimonas*0.0412

[0010] Among them, RiskScore is the risk score.

[0011] The present invention also provides a system for evaluating a method for predicting the prognosis of hepatocellular carcinoma, comprising a data collection module, a risk assessment module and an output module;

[0012] The data collection module is used to collect samples from patients and measure the abundance of each microorganism in the samples, wherein the microorganism is a microorganism in the aforementioned microbial marker combination;

[0013] The risk assessment module is used to calculate the risk score according to the risk model in claim 5;

[0014] The output module is used to output risk scores.

[0015] Beneficial effects: By utilizing advanced bioinformatics analyses, including linear discriminant analysis effect size (LEfSe) and Cox proportional hazard regression, the present invention determines the correlation between specific microorganisms and clinical outcomes of HCC patients, and further explores the relationship between microbial profiles and gene expression within tumor tissues to elucidate potential pathways by which microorganisms may affect HCC progression. In addition, the development of a microbial prognostic signature model based on these analyses aims to improve the prediction of patient survival and divide patients into different prognostic categories. This model combines microbial data with traditional clinical parameters to provide a more comprehensive prognostic assessment, which can more effectively formulate treatment strategies. In addition, the present invention also explores the impact of HCC tumor-associated microbiota on immune status, which can further provide information for the formulation of immunotherapy strategies. By comprehensively analyzing the microbial properties in HCC and its prognosis, the understanding of the HCC tumor microenvironment is enhanced, thereby facilitating the formulation of more precise and personalized cancer management strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Figure 1 is a graph showing the distribution of the kingdom, phylum, class, order, family, and genus levels in HCC cancer samples and adjacent paracancerous samples; Figure A shows the distribution of the samples at the kingdom level, Figure B shows the distribution of the samples at the phylum level, Figure C shows the distribution of the samples at the class level, Figure D shows the distribution of the samples at the order level, Figure E shows the distribution of the samples at the family level, and Figure F shows the distribution of the samples at the genus level;

[0017] Figure 2 The results of LEfSe analysis, Figure A is the statistical display of LDAscore of characteristic microorganisms in LEfSe analysis results, and Figure B is the evolutionary branch diagram of microorganisms analyzed by LEfSe;

[0018] Figure 3 Figure A is a volcano plot of the prognostic abundance of differential microbial markers and a forest plot of the univariate cox regression results; Figure A is a volcano plot of the prognostic abundance of differential microbial markers, and Figure B is a forest plot of the univariate cox regression results;

[0019] Figure 4 The figures are survival curves of hepatocellular carcinoma microbial markers with significant P values, wherein Figure A is the survival curve of microbial marker g_Dichelobacter; Figure B is the survival curve of microbial marker g_Sicinivirus; Figure C is the survival curve of microbial marker g_Microvirgula; Figure D is the survival curve of microbial marker g_Trabulsiella; Figure E is the survival curve of microbial marker g_Methylophaga; Figure F is the survival curve of microbial marker g_Candidatus_Glomeribacter;

[0020] Figure 5 The process diagram of the model for constructing a risk scoring model using LASSO regression; Figure A is a dynamic process diagram of LASSO screening variables, and Figure B is a process diagram of selecting different combinations of the cross-validation parameter λ for the included variables;

[0021] Figure 6 Survival curves and ROC curves of high and low risk groups in the training set; Figure A is the survival curve of high and low risk groups in the training set, and Figure B is the ROC curve of high and low risk groups in the training set;

[0022] Figure 7 Survival curves and ROC curves of the high and low risk groups in the validation set; Figure A is the survival curve of the high and low risk groups in the validation set, and Figure B is the ROC curve of the high and low risk groups in the validation set;

[0023] Figure 8 Figure A is the clinical nomogram of the training set, Figure B is the clinical nomogram of the validation set, Figure C is the ROC curve of the clinical nomogram of the training set; Figure D is the ROC curve of the clinical nomogram of the validation set;

[0024] Fig. 9 The C-Index of clinical nomogram and Risk score in training set and validation set is shown in the figure;

[0025] Fig.10 Performance prediction graphs for validating the risk scoring model using an external validation dataset;

[0026] Fig.11 Graphs are presented for the correlation between risk scores and clinical characteristics;

[0027] Fig.12 A graph showing the correlation between hepatocellular carcinoma gene expression and the abundance of microbial species included in the signature model;

[0028] Fig.13 Figure 1 is a functional enrichment analysis diagram of significantly related genes; Figure A is a biological process diagram of GO enrichment analysis; Figure B is a cell component diagram of GO enrichment analysis; Figure C is a molecular function diagram of GO enrichment analysis; Figure D is a KEGG enrichment analysis diagram;

[0029] Fig.14 This is a diagram showing the difference in the enrichment scores of immune infiltrating cells between high-risk and low-risk groups;

[0030] Fig.15 This is a graph showing the differences in scores related to 13 immune functions between high-risk and low-risk groups;

[0031] Fig.16 This is a graph showing the difference in TIDE between high-risk and low-risk groups;

[0032] Fig.17The figure shows the difference in immune checkpoint expression between high-risk and low-risk groups. DETAILED DESCRIPTION

[0033] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0034] Studies have shown that the microbiota in the liver cancer microenvironment can serve as an independent biomarker to predict the postoperative outcome of liver cancer patients. Notably, specific bacteria with known anti-tumor properties were significantly reduced in tumor tissues compared with adjacent non-tumor samples, and these differences were positively correlated with prognostic outcomes. The HCC intratumoral microbiota has become a key factor in regulating HCC tumor behavior and host immune responses, providing new opportunities for prognostic and therapeutic interventions.

[0035] This example uses a large amount of data obtained from the Cancer Genome Atlas-Hepatocellular Carcinoma (TCGA-LIHC) project, which includes microbial abundance, clinical parameters, gene expression, survival indicators, and other phenotypic data. By using advanced bioinformatics analysis, including linear discriminant analysis effect size (LEfSe) and Cox proportional hazard regression, this example determines the correlation between specific microorganisms and clinical outcomes of HCC patients, and further explores the relationship between microbial profiles and gene expression in tumor tissues to elucidate potential pathways by which microorganisms may affect HCC progression. In addition, the development of a microbial prognostic signature model based on these analyses aims to improve the prediction of patient survival and classify patients into different prognostic categories. The model combines microbial data with traditional clinical parameters to provide a more comprehensive prognostic assessment, which can more effectively formulate treatment strategies. This example also explores the impact of HCC intratumor-associated microbiota on immune status, which can further provide information for the formulation of immunotherapy strategies. By comprehensively analyzing the microbial properties and their prognosis in HCC, the understanding of the HCC tumor microenvironment is enhanced, thereby facilitating the formulation of more precise and personalized cancer management strategies.

[0036] This embodiment specifically includes the following steps:

[0037] 1. Obtain relevant samples:

[0038] Download TCGA hepatocellular carcinoma microbiome abundance data:

[0039] http: / / ftp.microbio.me / pub / cancer_microbiome_analysis / TCGA / Kraken / Krake n-TCGA-Voom-SNM-Plate-Center-Filtering-Data.csv;

[0040] Download TCGA hepatocellular carcinoma microbiome clinical data:

[0041] http: / / ftp.microbio.me / pub / cancer_microbiome_analysis / TCGA / Kraken / Metad ata-TCGA-Kraken-17625-Samples.csv;

[0042] The clinical data of TCGA hepatocellular carcinoma samples are shown in Table 1:

[0043] Table 1

[0044]

[0045]

[0046] Download TCGA hepatocellular carcinoma patient survival data:

[0047] https: / / tcga-xena-hub.s3.us-east-1.amazonaws.com / download / survival%2FLI HC_survival.txt;

[0048] Download TCGA hepatocellular carcinoma sample phenotype data:

[0049] https: / / tcga-xena-hub.s3.us-east-1.amazonaws.com / download / TCGA.LIHC.sampleMap%2FLIHC_clinicalMatrix.

[0050] Microbial community map display: obtain information related to the microbial abundance of each sample from the microbiome sample, and then display the average abundance of the population based on the kingdom, phylum, class, order, family, and genus levels. Since there are too many species in the phylum, class, order, family, and genus, the top 20 abundances of each level are taken to draw a box plot for display.

[0051] In this example, a flora map was displayed based on the sample microbiome information, and 420 HCC cancer samples and 104 paracancerous samples were obtained. Then, based on the kingdom, phylum, class, order, family, and genus levels, the average values ​​were taken for merging, and the top 20 were selected for display. Figure 1 As shown in the figure, the proportions of each grade in HCC cancer samples and adjacent cancer samples are evenly distributed.

[0052] Identification of LEfSe specific microbial makers: LEfSe analysis was performed based on cancer samples and adjacent paracancerous samples. The R package microeco was used to obtain cancer-related characteristic microbial markers with alpha = 0.05 (significance threshold) as the threshold. Then, the LDA statistical graph of characteristic microbial markers was drawn based on the top 30 LDA values, and an evolutionary branch diagram was drawn based on the top 100 bacterial communities with the highest richness for display.

[0053] This example performs LEfSe analysis based on cancer samples and paracancerous samples. The results show that there are 581 microbial markers with significant differences between cancer samples and paracancerous samples, such as Figure 2 As shown in Figure A, 218 of them were microbial markers with higher LDA in the hepatocellular carcinoma group, and the remaining 363 were microbial markers with higher LDA in the para-cancer group. Figure 2 As shown in Figure B, most bacteria in the Bacteroidetes, Firmicutes, and Euryarchaeota classes play an important role in the tumor group, while most bacteria in the Spirochaetes and Acidobacteria classes mainly play an important role in the para-cancer group. The 581 differential microbial markers here include kingdom, phylum, class, order, family, and genus levels. After integration, a total of 381 differential microorganisms at the genus level were obtained.

[0054] Construction and validation of microbial prognostic feature model: The TCGA hepatocellular carcinoma samples were divided into 5:5, 50% of the data were used for modeling (training set) and 50% of the data were used for validation (validation set). The survival period data were further combined, and the abundance values ​​of the characteristic microbial markers of the training set samples were subjected to batch Cox univariate regression analysis using the R packages survival (v3.2-7) and survminer (v0.4.8). After regression analysis, the characteristic microbial markers significantly correlated with the survival period were screened with a threshold of P < 0.05 for subsequent analysis.

[0055] This example is based on the abundance of 381 genus-level differential microorganisms, combined with the data of total survival, to perform univariate Cox regression analysis to screen microbial markers that are significantly correlated with survival; Figure 3 As shown in Figure A, a total of 62 microbial markers significantly associated with prognosis were obtained, such as Figure 3 As shown in Figure B, HR greater than 1.0 indicates that its abundance increase may cause poor clinical prognosis in patients with hepatocellular carcinoma, and HR less than 1.0 indicates that its abundance increase may improve the clinical prognosis of patients with hepatocellular carcinoma. The survival curve of the microbial markers is further displayed, as shown in Figure 4 As shown, the significant prognostic differences between high and low abundance of microbial markers suggest a unique clinical value for the prognosis of patients with hepatocellular carcinoma.

[0056] LASSOCox regression analysis: lasso regression dimensionality reduction was further performed on the prognosis-related characteristic microbial markers, and a risk score model was constructed. This process mainly relied on the R package glmnet (v4.0-2). In order to build a more accurate regression model, lambda screening was first performed using cross-validation, and then the model corresponding to lambda.min was selected. The abundance matrix of the relevant characteristic microbial markers in the model was further extracted, and the risk score of each sample was calculated based on the following formula:

[0057]

[0058] 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 correlated marker in each sample multiplied by the coef of the corresponding marker, i represents the sample, and j represents the microbial marker.

[0059] In this example, microbial markers are incorporated into LASSO Cox regression to establish a prognostic model to calculate the risk score. Specifically, LASSO regression is used to reduce the dimension of the univariate Cox regression results to obtain 12 microbial markers, such as Figure 5 The risk scoring model was constructed as shown, namely: Risk Score = Candidatus_Glomeribacter abundance*(-0.1772)+Sicinivirus*(-0.1386)+Shuttleworthia*(-0.1212)+Leucothrix*(-0.0817)+Schleiferia*(-0.0791)+Dichelobacter*(-0.0460)+Salinimicrobium*(-0.0334)+T4likevirus*(-0.0332)+Ictalurivirus*(-0.0035)+Sphaerotilus*0.0003+Kitasatospora*0.0269+Succinimonas*0.0412.

[0060] In order to verify the effectiveness of the model, based on the risk scores of the training set samples, the high-risk and low-risk groups were divided with the median as the node. The survival curve was drawn in combination with the survival period data, and the P value was calculated. If the P-value < 0.05, it was judged that the difference between the high-risk and low-risk groups was significant. Then the sample risk score was further used as the model prediction result, and the AUC value of the model was calculated in combination with the survival data, and then the ROC curve was drawn. Figure 6The survival curves of the two groups shown in Figure A were significantly different (P<0.0001). Figure 6 As shown in Figure B, the sample risk score is used as the prediction result of the risk prognosis model, and the AUC values ​​of the model calculated in combination with the survival data for 3, 4, and 5 years are all greater than 0.6, indicating that the model has good performance.

[0061] The model performance was further verified in the validation set. The sample risk score was calculated according to the formula. Based on the sample risk score, the high and low risk groups were divided with the median as the node, such as Figure 7 As shown in A in Figure 1, the survival curve was drawn by combining the overall survival data. The survival curves of the two groups were significantly different (P = 0.031). Figure 7 As shown in Figure B, the sample risk score is used as the model prediction result, and the AUC of the model calculated in combination with the survival data for 3, 4, and 5 years is greater than 0.6, indicating that the model has good performance.

[0062] A clinical nomogram was constructed to guide clinical decision-making. The probability of patient outcomes was calculated based on the scoring indicators. The nomogram was drawn mainly using the R packages rms (v6.1-0) and survival. First, the cox proportional hazard regression model was constructed using the cph() function in combination with age, grade, and clinical_stage. Then the survival() function was used to calculate the survival probability. Finally, the nomogram object was constructed using the nomogram() function and displayed using plot(). In addition, the C-index of the training and validation set data was calculated, and statistical graphs were drawn for display.

[0063] Risk score, age, gender, grade, and stage in the training and validation data were used to draw a nomogram to guide clinical decision-making, and a ROC curve based on the nomogram was also drawn. Figure 8 As shown in the figure, the AUCs of 3, 4, and 5 years are all greater than 0.65 in both the training set and the validation set, indicating that the model is effective. Finally, the C-Index (C index) of the nomogram and Risk score in the training set and the validation set were counted respectively, as shown in the figure below: Fig. 9 As shown in the figure, the C-Index of the two types of models are both greater than 0.5, indicating positive predictive efficacy. However, the C-Index of the nomogram in the training set is slightly lower than the Risk score, while the C-Index of the nomogram in the validation set is significantly higher than the Risk score. The prognostic efficacy of the validation set nomogram is significantly better than the single Riskscore, indicating that the prediction of the multifactor clinical nomogram has certain clinical guiding significance for the screening and treatment of high-risk populations for hepatocellular carcinoma.

[0064] In order to evaluate the predictive performance of the risk scoring model, this example obtained an external validation dataset (PRJNA714196) from the NCBI-SRA database and performed ROC analysis using selected overlapping microorganisms, such as Fig.10 As shown, the AUC of the external validation set PRJNA714196 was 0.68, so the results of external validation showed that the risk score model constructed by 12 microbial markers had good performance.

[0065] The feature model of this embodiment is related to the patient's prognosis and clinical characteristics: This embodiment combines the validation set and training set data, and counts the differences in risk scores in the age, gender, grade, pathologic_M, pathologic_N, pathologic_T, and stage groups in the overall data. Fig.11 As shown, there were significant differences in risk scores among age, gender, pathologic_N, pathologic_T, and stage groups.

[0066] Correlation between hepatocellular carcinoma gene expression and the abundance of microbial species included in the characteristic model: This example statistically analyzed the correlation between the abundance of microbial markers in the model and the expression of protein-coding genes in the overall sample. First, 19,597 protein-coding genes were obtained, and then based on the correlation threshold, 50 groups of significant correlations were obtained, including 49 protein-coding genes. Fig.12 A correlation heat map is given. In the model, Salinimicrobium, Candidatus_Glomeribacter, and Succinimonas have a higher proportion of positive correlation with related genes, while the remaining other microorganisms have a higher proportion of negative correlation with related genes.

[0067] Functional enrichment analysis of hepatocellular carcinoma genes: This example performs functional enrichment analysis on 49 protein-coding genes. GO enrichment analysis is divided into three parts, namely: biological process (BP), cellular component (CC), and molecular function (MF). Fig.13As shown in the figure, for genes, the BP-enriched pathways are mainly nucleotide metabolic process, nucleosidephosphate metabolic process, purine ribonucleotide metabolic process, etc., the CC-enriched pathways are mainly nucleoid, mitochondrial nucleoid, etc., the MF-enriched pathways are mainly lyase activity, carbon-oxygen lyase activity, etc., and the KEGG-enriched pathways are mainly HIF-1signalingpathway, Glycolysis / Gluconeogenesis, Carbon metabolism, etc.

[0068] Immune cell infiltration and immune function difference display: This example shows the GSVA calculation results of immune cells in high-risk and low-risk groups. Fig.14 As shown ( Fig.14 * indicates P<0.05; ** indicates P<0.01). There were differences in the proportions of the four immune infiltrating cells between the high-risk and low-risk groups. The proportion of Plasmacytoid dendritic cells in the high-risk group was significantly higher than that in the low-risk group. The remaining activated CD8 T cells, Eosinophil, and Gamma delta T cells were all significantly higher in the low-risk group than in the high-risk group. The differences in the scores of 13 immune functions between the high-risk and low-risk groups were further statistically analyzed, such as Fig.15 As shown ( Fig.15 * indicates P<0.05; ** indicates P<0.01), among which 3 immune function-related scores showed significant differences between high-risk and low-risk groups, including Cytolytic activity, Inflammation-promoting, and Type II IFN Reponse, all of which had higher enrichment scores in the low-risk group.

[0069] Differences in the expression of immune indicators and checkpoints: Statistical differences in the expression of immune indicators and immune checkpoints between high and low risk groups. Fig.16 As shown ( Fig.16 * indicates P<0.05). The tumor immune dysfunction and exclusion (TIDE) score of the low-risk group was significantly higher than that of the high-risk group, indicating that the effect of immune checkpoint inhibitor treatment in the low-risk group was worse than that in the high-risk group, and was associated with the poor clinical prognosis of the low-risk group samples. Fig.17 As shown ( Fig.17In the figure, * indicates P<0.05; ** indicates P<0.01; *** indicates P<0.001). For the 64 immune checkpoints, 8 showed significant differences between the high-risk and low-risk components, of which 1 was significantly overexpressed in the high-risk group, and the remaining 7 were all significantly overexpressed in the low-risk group.

[0070] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not further describe various possible combinations.

Claims

1. A microbial marker composition for predicting the prognosis of hepatocellular carcinoma, characterized in that: The microbial marker composition includes: Candidatus_Glomeribacter, Sicinivirus, Shuttleworthia, Leucothrix, Schleiferia, Dichelobacter, Salinimicrobium, T4likevirus, Ictalurivirus, Sphaerotilus, Kitasatospora, and Succinimonas.

2. A detection reagent, characterized in that: An object for detecting the abundance of the microbial marker as described in claim 1.

3. Testing product, characterized in that, It comprises the detection reagent as claimed in claim 2 and an acceptable auxiliary agent, carrier or device.

4. A method for predicting the prognosis of hepatocellular carcinoma, characterized in that: Performing weighted summation on the abundance of the microorganisms described in claim 1 to construct a risk model and obtain a risk score; Then the median of the risk model is used as the critical value. If the risk score is greater than or equal to the critical value, it is considered as high risk, otherwise it is considered as low risk.

5. The method for predicting the prognosis of hepatocellular carcinoma according to claim 4, characterized in that: The risk model is expressed as: RiskScore=Candidatus_Glomeribacter*(-0.1772)+Sicinivirus*(-0.1386)+Shuttleworthia*(-0.1212)+Leucothrix*(-0.0817)+Schleiferia*(-0.0791)+Dichelobac ter*(-0.0460)+Salinimicrobium*(-0.0334)+T4likevirus*(-0.0332)+Ictalurivirus*(-0.0035)+Sphaerotilus*0.0003+Kitasatospora*0.0269+Succinimonas*0.0412 Among them, RiskScore is the risk score.

6. A system for use in a method for evaluating the prognosis of hepatocellular carcinoma as claimed in claim 5, characterized in that: It includes data collection module, risk assessment module and output module; The data collection module is used to collect samples from patients and measure the abundance of each microorganism in the samples, wherein the microorganism is a microorganism in the aforementioned microbial marker combination; The risk assessment module is used to calculate the risk score according to the risk model in claim 5; The output module is used to output risk scores.

Images