A liver cancer prognosis model, a construction method and application thereof

Abstract

The application belongs to the field of gene detection, and relates to a liver cancer prognosis model and a construction method and application thereof. It is found that SLAMF7 high-expression macrophages are mainly enriched in cancer tissues, and it is clear that there is obvious spatial co-localization between the SLAMF7 high-expression macrophages and tumor-reactive CD8 T cells. Further communication analysis shows that the SLAMF7 high-expression macrophages can regulate the cytotoxicity-related genes of the CD8 T cells and play an immune stimulation function. Based on liver cancer transcriptome data, the liver cancer risk assessment model is constructed by using the communication and receptor matching of the SLAMF7 high-expression macrophages and the tumor-reactive CD8 T cells. The results show that the liver cancer patients with high risk scores have a shorter overall survival period than the liver cancer patients with low risk scores, and the liver cancer patients with high risk scores are often resistant to immunotherapy. The model score is used to quantitatively evaluate the immunotherapy response of the patients, and the model can be used for assisting in evaluating the prognosis of the liver cancer patients and individualized guidance.

Description

Technical Field

[0001] This invention belongs to the field of gene detection and relates to a liver cancer prognostic model, its construction method, and its application. Background Technology

[0002] With the discovery of immune checkpoints and the advent of immune checkpoint inhibitors, immunotherapy has been included in the guidelines for the diagnosis and treatment of primary liver cancer. Recently, multiple clinical trials focusing on anti-PD1 monoclonal antibodies have demonstrated their potential in the treatment of advanced liver cancer. However, some patients still do not benefit from it. The advent of single-cell sequencing technology has made it possible to analyze cellular heterogeneity within the tumor microenvironment, while spatial transcriptome sequencing technology can provide spatial information on different cell populations. A recent study integrating single-cell and spatial transcriptome analyses found that SPP1-highly expressing macrophages enhance the function of tumor-associated fibroblasts through cell communication, forming an "immune barrier" at the tumor lesion boundary that blocks the infiltration of anti-tumor immune cells. This is associated with shorter survival rates and immunotherapy resistance in liver cancer patients. The above research provides a potential prognostic assessment model: constructing a risk assessment model based on the communication receptors of key cell populations involved in disease progression. This is of great significance for the development of individualized treatment plans for liver cancer patients.

[0003] The widespread application of next-generation sequencing technology has deepened our understanding of liver cancer, and various research teams have developed numerous prognostic models for patient assessment. However, previous liver cancer prognostic models were typically constructed based on differentially expressed genes screened from tissue RNA sequencing expression profiles. This approach may miss key features of cancer development and incorporate interfering signals unrelated to disease progression, resulting in models with insufficient efficacy and limited application. By integrating single-cell transcriptomics and spatial transcriptomics analysis, key cell populations within the microenvironment influencing disease development can be identified, downstream effects of cell communication can be clarified, and receptor-ligand communication within target cell populations can be utilized to construct more efficient and interpretable liver cancer prognostic models. Summary of the Invention

[0004] To address the shortcomings of existing liver cancer prognostic models mentioned above, this invention provides a method for constructing a liver cancer prognostic model and its application.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] On the one hand, this invention provides a liver cancer prognostic model, constructed based on the communication of receptor genes between SLAMF7-overexpressing macrophages and tumor-reactive CD8 T cells; specifically, the model is defined as Riskscore = ∑ i Expression (mRNA) i *Coefficent (mRNA)i , where i is the key gene for screening.

[0007] Secondly, this invention provides an application of the aforementioned liver cancer prognostic model, which is used to assess the risk of liver cancer patients in each cohort. First, the prognostic model calculates the patient's risk score. Then, based on the `surv_cutpoint` function of the `survival` package, the optimal cut-point value for the risk score is determined. When a patient's risk score is equal to or higher than the optimal cut-point value, the patient is included in the high-risk group; when a patient's risk score is lower than the optimal cut-point value, the patient is included in the low-risk group.

[0008] Furthermore, the optimal critical value for the aforementioned risk score is determined using the R packages survival and survivor.

[0009] Thirdly, this invention provides a method for constructing a prognostic model for liver cancer, comprising the following steps:

[0010] 1) Single-cell data were collected from the same publicly available cohort containing cancer single-cell data to identify tumor-reactive CD8 T cells and macrophages;

[0011] 2) Clarify the spatial location between tumor-reactive CD8 T cells and macrophages obtained in step 1) from the perspective of spatial transcriptome data. It was found that the tumor-reactive CD8 T cells and SLAMF7-highly-expressing macrophages have a high degree of co-localization, thus obtaining SLAMF7-highly-expressing macrophages with high spatial correlation with tumor-reactive CD8 T cells.

[0012] 3) Using Nichenetr, the signal sent by SLAMF7-overexpressing macrophages to tumor-reactive CD8 T cells obtained in step 2) was inferred, and the top-performing receptor-ligand pairs were identified. Then, univariate Cox regression analysis was used to screen the top-performing receptor-ligand pairs identified by Nichenetr to obtain receptor-ligand genes that have a predictive effect on prognosis.

[0013] 4) Construct a liver cancer prognostic model based on the ligand receptor genes obtained in step 3) that have predictive value for prognosis using multivariate Cox regression.

[0014] Furthermore, the specific method for identifying tumor-reactive CD8 T cells in step 1) of this invention is as follows:

[0015] 1.1) Collect single-cell data of various cancers, such as hepatocellular carcinoma, mixed hepatocellular carcinoma, intrahepatic cholangiocarcinoma, and adjacent normal cell data, from a public cohort containing single-cell data of primary liver cancer; to obtain single-cell data of primary liver cancer.

[0016] 1.2) The cell type annotation of the primary liver cancer single-cell data obtained in step 1.1) was performed using the R package seurat, and the CD8T cell population with high expression of CD8A was extracted;

[0017] 1.3) Use the FindClusters function to subdivide the CD8T cells with high CD8A expression obtained in step 1.2); use OR analysis to screen out CD8T cell subpopulations enriched in cancer tissue;

[0018] 1.4) Use the fgsea function to perform GSEA analysis on the CD8T cells enriched in cancer tissue obtained in step 1.3) to screen out the CD8T cell subset that highly expresses the T cell activation pathway, namely the tumor reactive CD8T cell subset.

[0019] Furthermore, the specific method for identifying SLAMF7-overexpressing macrophages in step 2) of this invention is as follows:

[0020] The R package seurat was used to annotate the single-cell data of primary liver cancer obtained in step 1.1), and macrophages with high expression of CD68 were extracted. The FindClusters function was used to further subdivide the obtained macrophages with high expression of CD68, and a subset of SLAMF7 macrophages with high spatial correlation with tumor reactive CD8T cells was identified.

[0021] Furthermore, the specific implementation method of step 3) adopted in this invention is as follows:

[0022] Using Nichenetr, the signals sent by SLAMF7-overexpressing macrophages to tumor-reactive CD8 T cells obtained in step 2) were inferred. The top 30 ligand-receptor pairs with the highest activity fractions and the genes regulated by the ligands were retained, thus obtaining the communication ligand-receptor genes to be screened.

[0023] The GSE76427 cohort with prognostic information was collected from the GEO database. Based on this cohort, univariate Cox regression analysis was used to screen the communication receptor genes obtained above. Genes with p < 0.05 were retained as receptor genes with indicative prognostic value for model construction.

[0024] Furthermore, the specific implementation method of step 4) in this invention is as follows:

[0025] The TCGA cohort with prognostic information was collected from the UCSCXena database, the GSE76427, GSE144269, and GSE14520 cohorts with prognostic information were collected from the GEO database, and the CHCC hepatocellular carcinoma cohort with prognostic information was collected from the NODE database. A multivariate Cox model was constructed using these cohorts based on the ligand receptor genes identified in step 3) that provide prognostic indications. The model calculation formula is as follows:

[0026] Riskscore = ∑ i Expression (mRNA) i *Coefficent (mRNA) i

[0027] Riskscore represents the patient's risk score, i represents each selected gene, Expression represents the expression level of gene i, and Coefficent represents the regression coefficient of gene i.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] This invention provides a prognostic model for liver cancer, its construction method, and its application. After identification, this invention found that SLAMF7-overexpressing macrophages are mainly enriched in cancerous tissue, and clearly demonstrated significant spatial co-localization with tumor-reactive CD8 T cells. Further communication analysis revealed that SLAMF7-overexpressing macrophages can regulate cytotoxicity-related genes of CD8 T cells, exerting an immunostimulatory function. Therefore, based on liver cancer transcriptome data, this invention utilizes the communication receptor-ligand interaction between SLAMF7-overexpressing macrophages and tumor-reactive CD8 T cells to develop a liver cancer risk assessment model. Results show that patients with high risk scores have shorter overall survival than those with low scores, and those with high risk scores are often resistant to immunotherapy. Based on quantified model scores, the model assesses patients' immunotherapy response. This model can be used to assist in assessing the prognosis of liver cancer patients and provide individualized guidance. Attached Figure Description

[0030] Figure 1 is a diagram illustrating the identification and functional analysis of tumor-reactive CD8 T cells used in this invention. Figure 1-A The UMAP diagram shows nine CD8 T cell subsets. Figure 1-B Stacked bar charts show the proportion of each CD8 subgroup in different samples. Figure 1-C The heatmap shows the expression of classic T cell functional genes in different CD8 T cell subsets. Figure 1-D The heatmap shows the expression of T cell activation pathways in different CD8 T cell subsets. Figure 1-E The heatmap shows the correlation between the CD8 T cell subsets in this cohort and the tumor-responsive CD8 T cell subsets in Eberhardt et al. Figure 1-F The UMAP diagram shows the CD8 T cell subsets of my own cohort and those of Liu et al. Figure 1-G Box plots show the amplification fractions of CD8T cell subsets observed by Liu et al.

[0031] Figure 2 is an identification diagram of potential communication partners for tumor-reactive CD8 T cells used in this invention, wherein... Figure 2-A Venn diagrams show the intersection genes of the top 200 genes most closely related to the SLAMF7 gene in the four cohorts. Figure 2-B The violin diagram illustrates the scores of intersecting genes in each cell population. Figure 2-C The scatter plot shows the correlation between the degree of myeloid cell infiltration and the degree of CXCL13-high expression CD8T cell infiltration in the four cohorts;

[0032] Figure 3 is a graph showing the identification and prognostic correlation analysis of SLAMF7-overexpressing macrophages used in this invention. Figure 3-A The UMAP diagram shows eight macrophage subsets. Figure 3-B The stacked bar chart shows the proportion of different macrophage subsets in different samples. Figure 3-C The histogram shows the functional pathways enriched from the Reactome database. Figure 3-D The heatmap shows the expression of classic macrophage functional pathways in different macrophage subsets. Figure 3-E Kaplan-Meier curves illustrate the clinical characteristics of Scissor+ macrophages and Scissor-macrophages. Figure 3-F The histogram shows the proportion of different macrophage subsets in Scissor+ macrophages and Scissor- macrophages;

[0033] Figure 4 is a spatial co-localization analysis diagram of SLAMF7-overexpressing macrophages and tumor-reactive CD8 T cells used in this invention. Figure 4-A The CellTrek algorithm was used to spatially map SLAMF7-highly-expressing macrophages and CXCL13-highly-expressing CD8 T cells. Figure 4-B Box plots show the spatial distances between CXCL13-overexpressing CD8 T cells and various macrophage subsets;

[0034] Figure 5 shows the construction of the risk assessment model based on the exchange-ligand receptor used in this invention and the cohort assessment diagram; wherein, Figure 5-A Nichenetr's hypothesized ligands for macrophages Figure 5-B Nichenetr speculates that CXCL13 high expression is a target gene regulated by CD8 T cells. Figure 5-C Forest plots illustrate allotropic receptor genes associated with prognosis. Figure 5-DKaplan-Meier curves demonstrate the prognostic efficacy of gene selection. Figure 5-E The Kaplan-Meier curve illustrates the prognostic efficacy of the constructed model. Figure 5-F Box plots demonstrate that the constructed model can be used to predict immunotherapy responses. Figure 5-G The Kaplan-Meier curve illustrates the prognostic efficacy of the constructed model in the immunotherapy cohort. Detailed Implementation

[0035] This invention provides a method for constructing a prognostic model for liver cancer, which mainly includes the following steps:

[0036] Step 1: Identification and functional analysis of tumor-reactive CD8 T cells. Specifically, seven samples each of hepatocellular carcinoma, mixed hepatocellular carcinoma, intrahepatic cholangiocarcinoma, and adjacent normal tissue were collected from the public cohort of Xue et al. (National Genome Data Center, Project ID: PRJCA007744). The single-cell data from these 28 primary liver cancer tissues were annotated using the R package seurat. CD8 T cells highly expressing CD8A were extracted from these samples. The extracted CD8 T cells were then further subdivided. The resulting CD8 T cell data are as follows: Figure 1-A As shown, the UMAP plot reveals that CD8T cells are divided into nine subsets. The cell proportion plot shows the distribution ratio of different CD8T cell subsets in different sample types. Figure 1-B As shown, CD8 T cells with high CXCL13 expression mainly infiltrate various types of liver cancer samples, especially CHC samples. The classical T cell gene set scores of the above nine different CD8 T cell subsets were performed using UCELL, as follows: Figure 1-C As shown, CD8 T cells with high CXCL13 expression exhibited higher co-stimulation and cytotoxicity scores. GSEA analysis of five T cell activation pathways was performed on nine different CD8 T cell subsets using the fgsea function, as shown... Figure 1-D As shown, CXCL13-overexpressing CD8T cells upregulate five activation pathways. The CD8T cell (target CD8T cell) data obtained above were integrated with CD8T cell (external CD8T cell) data validated in vitro by Eberhardt et al. (GEO database, project number GSE180268) for correlation analysis, as shown... Figure 1-E As shown, the target CD8T cells showed the highest correlation with external CD8T cells. The CD8T cell data obtained above were integrated with CD8T cell data from Liu et al. with paired TCR sequencing (GEO database, project number GSE179994) and then subjected to UMAP clustering, as shown... Figure 1-FAs shown, Liu et al. also observed a CD8T cell subset highly expressing CXCL13 in their CD8T cells. Analysis of TCR data using the STARTRAC algorithm revealed... Figure 1-G As shown, CD8T cells with high CXCL13 expression have the highest amplification fraction.

[0037] The above results collectively indicate that the CD8 T cell subset that highly expresses CXCL13 is a tumor-reactive CD8 T cell.

[0038] Step 2: Identification of potential communication partners for tumor-reactive CD8 T cells, specifically:

[0039] The TCGA-LIHC cohort was collected from the UCSCXena database, and the GSE25097, GSE14520, and GSE36376 cohorts were collected from the GEO database. Spearman correlation analysis was performed on SLAMF7 and other genes in these four liver cancer tissue sequencing cohorts, each with over 200 samples. Genes were sorted from highest to lowest correlation coefficient, and the intersection of the top 200 genes from each cohort was taken (Figure 2A), resulting in 32 genes. The addmodulescore function was then used to score these 32 genes across different cell types. As shown in Figure 2B, the scores indicate high expression of these 32 genes in myeloid cells such as monocytes, neutrophils, dendritic cells, and macrophages. Finally, Cibersortx deconvolution was performed on the four cohorts to assess cell type abundance, and Spearman correlation analysis was conducted. As shown in Figure 2C, macrophages and tumor-responsive CD8 T cells showed a stable positive correlation. These results suggest that macrophages are potential communication partners for tumor-responsive CD8 T cells.

[0040] Step 3: Identification and prognostic association analysis of SLAMF7-overexpressing macrophages, specifically:

[0041] The R package seurat was used to subdivide macrophages expressing CD68, such as... Figure 3-A As shown, the UMAP plot reveals that macrophages are divided into 8 subpopulations, and the cell proportion plot shows the distribution ratio of different macrophage subpopulations in different types of samples, such as... Figure 3-B As shown, the macrophage subset with high SLAMF7 expression mainly infiltrates various types of liver cancer samples, especially CHC samples. Next, Spearman correlation analysis of SLAMF7 with other genes was performed in the TCGA liver cancer cohort. The top 200 genes, ranked by correlation coefficient from highest to lowest, were then subjected to reactome enrichment analysis, as shown below. Figure 3-C As shown, the SLAMF7 gene is closely associated with the T cell activation pathway. Gene set scoring of different macrophage populations was performed using UCELL, such as... Figure 3-DAs shown, SLAMF7-overexpressing macrophages are M1 macrophages, and their antigen presentation function is upregulated. Finally, the prognostic value of the target macrophages was evaluated using the Scissor algorithm, such as... Figure 3-E As shown, Scissor+ was associated with better survival prognosis, while Scissor- was associated with worse survival prognosis. Finally, the relationship between different macrophage types and prognosis was assessed. Figure 3-F As shown, Scissor+ macrophages are mainly SLAMF7-highly expressed macrophages, indicating that targeted macrophages are associated with better prognosis.

[0042] Step 4: Spatial co-localization analysis of SLAMF7-overexpressing macrophages and tumor-reactive CD8 T cells, specifically:

[0043] The downsample function was used to resample macrophages and CD8T cells, taking 500 cells from each cell subpopulation. Then, the CellTrek algorithm was used to map the single-cell data onto spatial transcriptome data, such as... Figure 4-A As shown, SLAMF7-highly expressing macrophages and tumor-responsive CD8 T cells co-localized on three spatial transcriptome slices of liver cancer. The kdist function was then used to calculate the spatial distance between different macrophage subsets and tumor-responsive CD8 T cells, as shown below. Figure 4-B As shown, SLAMF7-overexpressing macrophages have the closest spatial distance to tumor-responsive CD8T cells compared to other macrophage subsets.

[0044] Step 5: Construction of a risk assessment model based on communication-based recipients and a cohort assessment graph, specifically:

[0045] Using Nichenetr, we inferred the signals sent by SLAMF7-overexpressing macrophages to tumor-responsive CD8 T cells, such as... Figure 5-A and Figure 5-B As shown, the top 30 ligand-receptor pairs and the genes regulated by the ligands are retained.

[0046] The GSE76427 cohort with prognostic information was collected from the GEO database. Based on this cohort, univariate Cox regression analysis was used to screen for Nichenetr-identified match-receptor pairs, resulting in five match-receptor genes, such as... Figure 5-C As shown, the forest plot reveals five receptor genes that are significantly associated with the prognosis of liver cancer patients (Coxp < 0.05). These genes are all protective factors (HR < 1). Figure 5-D As shown, the Kaplan-Meier curve indicates that these genes are associated with better patient prognosis.

[0047] Based on the five selected receptor-transfer genes, a multivariate Cox model was constructed in the GSE76427 cohort. The model formula is as follows:

[0048] Riskscore = ∑ i Expression (mRNA) i *Coefficent (mRNA) i

[0049] Riskscore is the patient's risk score, i is each of the five selected receptor genes, Expression is the expression level of gene i, and Coefficent is the regression coefficient of gene i.

[0050] Patient hazard scores were calculated using a constructed prognostic model. Then, the optimal cutpoint value for each hazard score was determined using the `surv_cutpoint` function of the `survival` package. Patients with hazard scores equal to or higher than the optimal cutpoint were placed in the high-risk group, while those with hazard scores lower than the optimal cutpoint were placed in the low-risk group. The survival prediction performance of the model was evaluated using Kaplan-Meier curves, such as... Figure 5-E As shown, the model demonstrated excellent predictive performance in the GSE76427, TCGA, CHCC, GSE14520, and GSE144269 hepatocellular carcinoma cohorts, with patients having higher risk scores exhibiting shorter overall survival. The model also showed excellent survival predictive performance in the hepatocellular carcinoma cohort receiving immunotherapy, such as... Figure 5-G As shown, those with higher risk scores have worse prognoses. This model can also be used to assess patient response to immunotherapy, such as... Figure 5-F As shown, those who do not respond to immunotherapy exhibit higher risk scores.

[0051] The above results indicate that this model can be used to assist in assessing the prognosis of liver cancer patients and guide individualized treatment.

[0052] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for constructing a prognostic model for liver cancer, characterized in that, Includes the following steps: 1) Single-cell data were collected from the same publicly available cohort containing cancer single-cell data to identify tumor-reactive CD8 T cell subsets and macrophages; 2) Clarify the spatial location between the tumor-reactive CD8 T cell subset and macrophages obtained in step 1) from the perspective of spatial transcriptome data. It was found that the tumor-reactive CD8 T cell subset and the SLAMF7-highly-expressing macrophage subset have a high degree of co-localization, thus obtaining the SLAMF7-highly-expressing macrophage subset with high spatial correlation with the tumor-reactive CD8 T cell subset. 3) Using Nichenetr, the signals sent by SLAMF7-overexpressing macrophages to tumor-reactive CD8 T cells obtained in step 2) were inferred, and the receptor-ligand pairs with the highest activity scores were identified. Then, univariate Cox regression analysis was used to screen the receptor-ligand pairs with the highest scores identified by Nichenetr to obtain receptor-ligand genes that have a predictive effect on prognosis. The receptor-ligand genes are: LTB, IL10, CD69, CD4, and ATP8B4. 4) Construct a liver cancer prognostic model based on the ligand receptor genes obtained in step 3) that have predictive value for prognosis using multivariate Cox regression; the calculation formula for the prognostic model is as follows: Riskscore=∑ i Expression(mRNA) i *Coefficent(mRNA) i Where: i is the key gene for screening; The prognostic model is used to assist in assessing the prognosis of liver cancer patients and to guide individualized treatment.

2. The method for constructing a liver cancer prognostic model according to claim 1, characterized in that, The step of identifying tumor-reactive CD8 T cells in step 1) includes: 1.1) The R package seurat was used to annotate the cell types of the primary liver cancer single-cell data in the publicly available cohort containing primary liver cancer single-cell data, and CD8T cells with high expression of CD8A were extracted; 1.2) Use the FindClusters function to subdivide the CD8T cells obtained in step 1.1); use OR analysis to screen out CD8T cell subpopulations enriched in cancer tissue; 1.3) Use the fgsea function to perform GSEA analysis on the CD8T cells enriched in cancer tissue obtained in step 1.2) to screen out the CD8T cell subset that highly expresses the T cell activation pathway, thus obtaining the tumor reactive CD8T cell subset.

3. The method for constructing a liver cancer prognostic model according to claim 2, characterized in that, The step 2) of identifying SLAMF7-overexpressing macrophages includes: using the R package seurat to annotate the single-cell data of primary liver cancer obtained in step 1.1), extracting macrophages that overexpress CD68; and using the FindClusters function to further subdivide the obtained macrophages that overexpress CD68, identifying a subset of SLAMF7-overexpressing macrophages that have a high spatial correlation with tumor-reactive CD8T cells.

4. The method for constructing a liver cancer prognostic model according to claim 3, characterized in that, Step 3) includes: using Nichenetr to infer the signals sent by SLAMF7-overexpressing macrophages to tumor-reactive CD8T cells, retaining the receptor-ligand pairs with the highest activity scores and the genes regulated by the ligands, i.e., obtaining the communication receptor-ligand genes to be screened; collecting the GSE76427 cohort with prognostic information from the GEO database, and using univariate Cox regression analysis based on this cohort to screen the obtained communication receptor-ligand genes; retaining genes with p<0.05 as receptor-ligand genes that have a predictive role in prognosis, and the receptor-ligand genes that have a predictive role in prognosis are used to construct a liver cancer prognostic model.

5. The method for constructing a liver cancer prognostic model according to claim 4, characterized in that, Step 4) includes: The TCGA-LIHC cohort with prognostic information was collected from the UCSCXena database, the GSE76427, GSE144269, and GSE14520 cohorts with prognostic information were collected from the GEO database, and the CHCC hepatocellular carcinoma cohort with prognostic information was collected from the NODE database. A multivariate Cox model was constructed using these cohorts based on the ligand receptor genes identified in step 3) that provide prognostic indications. The calculation formula for the model is as follows: Riskscore=∑ i Expression(mRNA) i *Coefficent(mRNA) i in: i is the key gene for screening.

6. A liver cancer prognostic model constructed based on the liver cancer prognostic model construction method according to any one of claims 1-5, characterized in that, The expression for the liver cancer prognostic model is as follows: Riskscore=∑ i Expression(mRNA) i *Coefficent(mRNA) i in: Riskscore is the patient's risk score; i represents each gene selected; Expression represents the expression level of gene i; Coefficent is the regression coefficient of gene i.

7. Application of the liver cancer prognostic model according to claim 6 in risk assessment of liver cancer samples.

8. The application according to claim 7, characterized in that, The application is implemented as follows: First, the patient's risk score is calculated using a prognostic model. Then, the optimal cutpoint value of the risk score is determined based on the surv_cutpoint function of the survival package. When the patient's risk score is equal to or higher than the optimal cutpoint value, the patient is included in the high-risk group. When the patient's risk score is lower than the optimal cutpoint value, the patient is included in the low-risk group.

Images