A prognostic marker related to gluconeogenesis of hepatocellular carcinoma patients and application thereof, a prognosis prediction model and a prognosis evaluation device

Abstract

The present application relates to the field of biomedical technology, in particular to a prognostic marker of hepatocellular carcinoma patients, application of the prognostic marker, a prognosis prediction model and a prognosis evaluation device. The present application analyzes the transcriptome data of 374 hepatocellular carcinoma patients from the TCGA database, and verifies in samples from the ICGC database. After database screening, 7 hub genes (i.e. the prognostic marker described in the present application) are screened from 117 gluconeogenesis-related genes from gluconeogenesis metabolism, and a prognosis prediction model and a prognosis evaluation device are established using the prognostic marker, which provides a new guide for the molecular heterogeneity of hepatocellular carcinoma. The prognostic marker provided by the present application has important clinical significance, and the prognosis of hepatocellular carcinoma patients can be evaluated using the prognostic marker, and patients with poor prognosis can be identified, which provides a basis for precise treatment of patients.

Description

Technical Field

[0001] This invention relates to the field of biomedical technology, specifically to a prognostic biomarker related to gluconeogenesis in hepatocellular carcinoma patients, its application, a prognostic prediction model, and a prognostic assessment device. Background Technology

[0002] Liver cancer is one of the most common malignant tumors of the digestive system, ranking fifth in incidence and third in mortality among all cancers. Hepatocellular carcinoma (HCC) is a common subtype of liver cancer, accounting for approximately 80% of primary liver cancers. Despite achievements in the treatment of HCC, the prognosis for patients with advanced HCC is very poor, with a 5-year overall survival rate of only 15%, and even with early surgical resection, the 5-year recurrence rate is as high as 50%. Therefore, identifying prognostic biomarkers and therapeutic targets for hepatocellular carcinoma is crucial.

[0003] In recent years, glucose metabolism reprogramming has been identified as an important marker of cancer. The shift from oxidative phosphorylation to gluconeogenesis in hepatocellular carcinoma (HCC) promotes rapid tumor cell proliferation and provides a favorable microenvironment for HCC progression. As the reverse process of glycolysis, gluconeogenesis-related genes have been shown to play a crucial role in the development and progression of HCC; however, there are currently no reports on gluconeogenesis-related genes that serve as prognostic markers for HCC patients. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the purpose of this invention is to provide a prognostic biomarker related to gluconeogenesis in patients with hepatocellular carcinoma, its application, a prognostic prediction model, and a prognostic assessment device, so as to solve the problems mentioned in the background art.

[0005] The present invention solves the technical problem by adopting the following technical solution: This invention provides a prognostic biomarker for hepatocellular carcinoma, wherein the prognostic biomarker is a combination of genes, including the PPARGC1A gene, SLC2A1 gene, SLC2A2 gene, SLC2A5 gene, ADH4 gene, DLAT gene and NR1H3 gene. The NCBI Gene ID of the PPARGC1A gene is 10891, and the NCBI Gene ID of the SLC2A1 gene is 6513. The NCBI Gene ID of the SLC2A2 gene is 6514, the NCBI Gene ID of the SLC2A5 gene is 6518, the NCBI Gene ID of the ADH4 gene is 127, the NCBI Gene ID of the DLAT gene is 1737, and the NCBI Gene ID of the NR1H3 gene is 10062.

[0006] The present invention also provides the use of a reagent for detecting the expression level of the prognostic biomarker of claim 1 in the preparation of products for assessing the prognosis of hepatocellular carcinoma.

[0007] Preferably, the product includes a reagent kit, a gene chip, or a device containing a gene detection module; the gene detection module is used to detect the expression level of the prognostic biomarker of claim 1.

[0008] Preferably, the prognosis includes the survival time of patients with hepatocellular carcinoma.

[0009] The present invention also provides a prognostic assessment device for hepatocellular carcinoma, including a gene detection module, a risk score determination module, and a prognostic assessment module; The gene detection module is used to detect the expression levels of the prognostic markers in samples from hepatocellular carcinoma patients. The risk score determination module is used to calculate the prognostic risk score of hepatocellular carcinoma patients based on the expression level of the prognostic biomarker. The formula for calculating the prognostic risk score is as follows: The prognostic risk score is calculated as follows: (0.205 × SLC2A1 gene expression level) - (0.108 × PPARGC1A gene expression level) - (0.052 × ADH4 gene expression level) + (0.460 × NR1H3 gene expression level) + (0.237 × DLAT gene expression level) - (0.034 × SLC2A2 gene expression level) + (0.039 × SLC2A5 gene expression level). The prognostic assessment module is used to determine the prognosis of hepatocellular carcinoma patients based on this prognostic risk score. When the prognostic risk score is ≥2.489, the patient is considered high-risk and has a poor prognosis; when the prognostic risk score is <2.489, the patient is considered low-risk and has a good prognosis.

[0010] Preferably, the system further includes a calibration module; the calibration module is used to construct a multivariate Cox regression model by combining the prognostic risk score with the patient's tumor stage, visualize the multivariate Cox regression model using a nomogram, and calculate the patient's 1-year, 3-year, or 5-year survival probability.

[0011] The prognostic assessment device also includes a medication assessment module; the workflow of the medication assessment module includes: S1. Obtain the expression data of the seven genes in hepatocellular carcinoma patients from the TCGA database, and divide the patients into high-risk and low-risk groups through the risk score determination module and prognosis assessment module; S2. Download sensitivity data of potential chemotherapy and targeted drugs for liver cancer from anticancer drug sensitivity genomics databases (such as the GDSC database), and use the "oncoPredict" and "ggplot2" packages in R software to predict the sensitivity of high-risk and low-risk patients to different drugs based on gene expression levels; S3. Use the WilCoxon test to compare the drug sensitivity scores of the high-risk group and the low-risk group, and screen for drugs with significant differences in sensitivity: drugs with low sensitivity scores in the low-risk group are suitable for low-risk patients, and drugs with low sensitivity scores in the high-risk group are suitable for high-risk patients.

[0012] This invention also provides a method for constructing a prognostic prediction model for hepatocellular carcinoma, comprising the following steps: S1. Data Acquisition: Transcriptomic data and clinical information (including age, sex, tumor stage, and tumor grade) of 374 hepatocellular carcinoma patients and transcriptomic data of 50 adjacent normal liver tissues were obtained from the TCGA database; transcriptomic data and clinical survival information of 240 hepatocellular carcinoma patients were obtained from the ICGC database; and 117 gluconeogenesis-related genes were obtained from the MSigDB database. S2. Differential Gene Screening: Transcriptome data were standardized using the R software "limma" package (converting FPKM to log2 (TPM+1)). Unsupervised clustering was performed on 374 hepatocellular carcinoma samples from the TCGA database using nonnegative matrix factorization (NMF). Univariate Cox regression analysis was performed on 117 gluconeogenesis-related genes using the R software "survival" package to screen for genes that showed significant differential expression between hepatocellular carcinoma and normal tissues and were associated with overall survival (p < 0.05). S3. Hub gene identification: The differentially expressed genes obtained in step S2 were subjected to LASSO regression analysis using the R software "glmnet" package. The 7 genes with the smallest model error (smallest λ value) were selected as hub genes, i.e., the prognostic biomarkers. S4. Model Construction: Using the R software "survival" package, multivariate Cox regression analysis was performed on 7 key genes to obtain the regression coefficient of each gene. Combined with gene expression levels, a prognostic risk score calculation formula was constructed, namely the prognostic risk score formula, to form a prognostic prediction model for hepatocellular carcinoma. S5. Model Validation: The model constructed in step S4 was applied to 240 hepatocellular carcinoma samples from the ICGC database. The effectiveness of the model was validated by Kaplan-Meier survival analysis, ROC curve analysis (AUC value), PCA analysis, and tSNE analysis.

[0013] The present invention also provides a method for assessing the prognosis of patients with hepatocellular carcinoma, comprising the following steps: S1. Collect tumor tissue samples from patients with hepatocellular carcinoma and use the gene detection module to detect the expression levels of the seven prognostic markers in the samples; S2. The patient's prognostic risk score is calculated using the aforementioned risk score determination module; S3. Determine the patient's risk level and prognosis using the aforementioned prognostic assessment module; S4. Calculate the patient's 1-year, 3-year, or 5-year survival probability through the calibration module, or select appropriate treatment drugs through the medication assessment module.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention relates to prognostic biomarkers for HCC patients and their applications, a prognostic prediction model, and a prognostic assessment device. The prognostic biomarkers provided by this invention are seven genes screened and found to be associated with overall survival (p < 0.05) in HCC patients. These prognostic biomarkers can be used to assess the prognosis of HCC patients, identify liver cancer patients with poor prognoses, guide clinicians to provide more aggressive treatment plans for high-risk patients, and avoid overtreatment of low-risk liver cancer patients.

[0015] This invention provides a prognostic biomarker for hepatocellular carcinoma (HCC), comprising a combination of genes including: PPARGC1A, SLC2A1, SLC2A2, SLC2A5, ADH4, DLAT, and NR1H3. This invention analyzed transcriptomic data from 374 HCC patients obtained from the TCGA database and validated the data in samples from the ICGC database. After database screening, this invention identified seven pivotal genes (i.e., the prognostic biomarker described in this invention) from 117 gluconeogenesis-related genes. Using these prognostic biomarkers, a prognostic prediction model and a prognostic assessment device were established, providing new insights into the molecular heterogeneity of HCC. The prognostic biomarker provided by this invention has significant clinical value. It can be used to assess the prognosis of HCC patients, identify liver cancer patients with poor prognoses, and provide a basis for precision treatment. Attached Figure Description

[0016] Figure 1 Cluster heatmap showing the grouping of HCC patients included in this invention and the overall survival rate of each group; Figure 2 The overall survival rate for each group of patients as shown in the Kaplan-Meier survival analysis; Figure 3 This invention presents the expression levels and clinical characteristics of 24 gluconeogenic differential prognostic genes in each subgroup of HCC patients included in this invention. Figure 4 This figure shows the results of identifying the gluconeogenesis hub gene in HCC patients and constructing a prognostic model for gluconeogenesis-related genes in this invention. Figure 5 This image shows the results of identifying the gluconeogenesis hub gene in HCC patients and constructing a prognostic model for gluconeogenesis-related genes in this invention. Figure 6 Figures showing the results of PCA and tSNE analyses for high-risk and low-risk groups based on prognostic models; Figure 7 Figures showing the results of PCA and tSNE analyses for high-risk and low-risk groups based on prognostic models; Figure 8 To establish a prognostic model based on the seven gluconeogenesis hub genes selected, Kaplan-Meier survival curves (E plots) of patients were drawn according to the high-risk and low-risk groups defined in the gluconeogenesis prognostic model. Figure 9 F-plot showing the ROC curve analysis results of the gluconeogenesis prognostic model for 1-year, 2-year, and 3-year overall survival. Figure 10G to J represent the prognostic value of gluconeogenesis prognostic models validated in an independent cohort (ICGC database), including PCA analysis (G) plots; Figure 11 For tSNE analysis (H) plot; Figure 12 For tSNE analysis (H) plot; Figure 13 Plot of Kaplan-Meier survival curves (I); Figure 14 ROC curve analysis (J) plot; Figure 15 and Figure 16 The results validate the expression and prognostic profile of the seven pivot genes screened in this invention. Figure 17 To obtain sensitivity data of nine targeted and chemotherapeutic drugs for liver cancer in high-risk and low-risk groups from the GDSC database; Figure 18-19 This invention establishes a clinical prediction model for HCC patients based on gluconeogenesis hub gene risk scores and tumor staging.

[0017] Figure 20-21 To visualize the multivariate Cox regression model using a nomogram and to calibrate it using a calibration curve. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to specific examples. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This embodiment provides a prognostic biomarker for hepatocellular carcinoma, comprising a combination of genes, including: PPARGC1A gene, SLC2A1 gene, SLC2A2 gene, SLC2A5 gene, and ADH4 gene. DLAT gene and NR1H3 gene.

[0020] This embodiment describes the application of a reagent for detecting the expression levels of prognostic biomarkers described in the above-described technical solution in the preparation of products for assessing the prognosis of hepatocellular carcinoma.

[0021] The products described in this embodiment include: a reagent kit, a gene chip, or a device containing a gene detection module, wherein the gene detection module is used to detect the expression level of the prognostic biomarker.

[0022] The prognosis in this embodiment includes survival time.

[0023] This embodiment of a prognostic assessment device for hepatocellular carcinoma includes: a gene detection module, a risk score determination module, and a prognostic assessment module; The gene detection module is used to detect the expression levels of prognostic biomarkers in patients with hepatocellular carcinoma; the prognostic biomarkers are those described in the above technical solution. The risk score determination module is used to determine the prognostic risk score of hepatocellular carcinoma patients; the prognostic risk score = (0.205×SLC2A1 gene expression level) - (0.108×PPARGC1A gene expression level) - (0.052×ADH4 gene expression level) + (0.460×NRIH3 gene expression level) + (0.237×DLAT gene expression level) - (0.034×SLC2A2 gene expression level) + (0.039×SLC2A5 gene expression level); The prognostic assessment module is used to determine the prognosis of hepatocellular carcinoma patients based on the prognostic risk score, including: when the prognostic risk score is ≥2.489, the patient is a high-risk patient with a poor prognosis; when the prognostic risk score is <2.489, the patient is a low-risk patient with a good prognosis.

[0024] The prognostic assessment device in this embodiment further includes a calibration module; the calibration module is used to assess the prognostic risk. A multivariate Cox regression model was constructed using the patient's score and tumor stage. The multivariate Cox regression model was visualized using a nomogram to calculate the patient's survival probability.

[0025] The prognostic assessment device in this embodiment also includes a medication assessment module; the medication assessment module is used to select drugs suitable for patients, including: obtaining gene expression data of hepatocellular carcinoma patients from a database, and using a risk score determination module and a prognostic assessment module to divide all patients into high-risk and low-risk groups; the genes include PPARGC1A gene, SLC2A1 gene, SLC2A2 gene, SLC2A5 gene, ADH4 gene, DLAT gene, and NR1H3 gene; the database includes the TCGA database; Sensitivity data of potential chemotherapy and targeted drugs for liver cancer were downloaded from the anticancer drug sensitivity genomics database. The "oncoPredict" and "ggplot2" packages of R software were used to predict the sensitivity of high-risk and low-risk patients to different drugs at the gene expression level, and drugs with significant differences in sensitivity were selected. The WilCoxon test was used to compare the drug sensitivity scores between the high-risk and low-risk groups. Drugs with low sensitivity scores in the low-risk group were suitable for low-risk patients, and drugs with low sensitivity scores in the high-risk group were suitable for high-risk patients.

[0026] In this invention, the PPARGC1A gene preferably includes nucleic acid encoding the PPARGC1A protein, and its NCBI Gene ID is preferably 10891; the DLAT gene preferably includes nucleic acid encoding the DLAT protein, and its NCBI Gene ID is preferably 1737; the SLC2A1 gene preferably includes nucleic acid encoding the SLC2A1 protein, and its NCBI Gene ID is preferably 6513; the SLC2A2 gene preferably includes nucleic acid encoding the SLC2A2 protein, and its NCBI Gene ID is preferably 6514; the SLC2A5 gene preferably includes nucleic acid encoding the SLC2A5 protein, and its NCBI Gene ID is preferably 6518; the ADH4 gene preferably includes nucleic acid encoding the ADH4 protein, and its NCBI Gene ID is preferably 127; the NR1H3 gene preferably includes nucleic acid encoding the NR1H3 protein, and its NCBI Gene ID is preferably 10062.

[0027] This invention analyzed transcriptomic data from 374 HCC patients obtained from the TCGA database and validated the data in samples from the ICGC database. After database screening, this invention identified 7 pivot genes (i.e., the prognostic biomarkers described in this invention) from 117 gluconeogenesis-related genes. Using these prognostic biomarkers, a prognostic prediction model and a prognostic assessment device were established, providing new insights into the molecular heterogeneity of HCC. The prognostic biomarkers provided by this invention have significant clinical value. They can be used to assess the prognosis of HCC patients, identify liver cancer patients with poor prognoses, and provide a basis for precision treatment.

[0028] This invention provides the application of reagents for detecting the expression levels of prognostic biomarkers described in the above-mentioned technical solutions in the preparation of products for assessing the prognosis of hepatocellular carcinoma (HCC). In this invention, the product preferably comprises: a reagent kit, a gene chip, or a device containing a reagent kit or gene chip, wherein the gene detection module is used to detect the expression levels of the prognostic biomarkers; the prognosis preferably includes survival time. By detecting the expression levels of the prognostic biomarkers described in the above-mentioned technical solutions, the prognosis of HCC patients can be assessed, and patients with poor prognoses can be identified, providing a basis for precision treatment of patients.

[0029] Given the current poor prognosis of hepatocellular carcinoma (HCC) patients, who may experience recurrence or metastasis, and the lack of cost-effective prognostic biomarkers and therapeutic targets, this invention addresses this issue by focusing on gluconeogenesis-related genes and constructing a prognostic model based on gene expression and tumor staging. A risk score is used to assess the prognosis of HCC. The assessment results offer possibilities for precision treatment of HCC patients. The prognostic prediction model provided by this invention allows for risk scoring of HCC patients; the prognostic risk score is correlated with survival, with patients having shorter survival and a poorer prognosis. This invention calculated the prognostic risk scores of 374 HCC patients from the TCGA database, obtaining a median of 2.489. Based on the median, patients can be divided into high-risk and low-risk patients. Specifically, when the prognostic risk score is ≥2.489, the patient is considered high-risk with a poor prognosis and short survival; when the prognostic risk score is <2.489, the patient is considered low-risk with a good prognosis and long survival. The model was validated in the ICGC database, demonstrating the feasibility of the prognostic prediction model provided by this invention. This invention provides a prognostic prediction model that can identify liver cancer patients with poor prognoses, guiding clinicians to provide more aggressive treatment plans for high-risk patients while avoiding overtreatment of low-risk liver cancer patients.

[0030] This invention provides a prognostic assessment device for hepatocellular carcinoma, comprising: a gene detection module, a risk score determination module, and a prognostic assessment module; The gene detection module is used to detect the expression levels of prognostic biomarkers in patients with hepatocellular carcinoma; the prognostic biomarkers are those described in the above technical solution. The risk score determination module is used to determine the prognostic risk score of hepatocellular carcinoma patients; the prognostic risk score = (0.205 × SLC2A1 gene expression level) - (0.108 × PPARGC1A gene expression level) - (0.052 × ADH4 gene expression level) + (0.460 × NRIH3 gene expression level) + (0.237 × DLAT gene expression level) - (0.034 × SLC2A2 gene expression level) + (0.039 × SLC2A5 gene expression level). The prognostic assessment module is used to determine the prognosis of patients with hepatitis B virus-related hepatocellular carcinoma based on the prognostic risk score, including: when the prognostic risk score is ≥2.489, the patient is a high-risk patient with a poor prognosis; when the prognostic risk score is <2.489, the patient is a low-risk patient with a good prognosis.

[0031] In this invention, the detection sample for the gene detection module preferably includes tumor tissue from a patient with hepatocellular carcinoma. This invention does not impose special requirements on the method for the gene detection module; any method well-known to those skilled in the art capable of detecting the expression levels of the seven genes responsible for the prognostic markers can be used.

[0032] In this invention, the prognosis preferably includes survival time; when the patient is a high-risk patient, the patient's survival time is short; when the patient is a low-risk patient, the patient's survival time is long.

[0033] In this invention, the prognostic assessment device preferably further includes a calibration module. The calibration module constructs a multivariate Cox regression model using the prognostic risk score and the patient's tumor stage, visualizes the multivariate Cox regression model using a nomogram, and calculates the patient's survival probability. This invention comprehensively considers risk scores and clinical information (including patient age, gender, tumor stage, and tumor grade). Univariate Cox regression analysis found that tumor stage and risk score are associated with a worse prognosis (p < 0.05). These two factors are further used to construct a multivariate Cox regression model, which is visualized using a nomogram to calculate the survival probability of each specific patient through intuitive graphics. The model is calibrated using a calibration curve; the smaller the deviation between the actual curve and the ideal curve, the higher the reliability of the surface curve. Therefore, this invention establishes a clinical model with good discriminative, calibration, and survival prediction capabilities based on the risk score estimated from the stage of seven gluconeogenesis-related genes and HCC tumors.

[0034] In this invention, the prognostic assessment device preferably further includes a medication assessment module; the medication assessment module is used to select drugs suitable for the patient, including: obtaining gene expression data of hepatocellular carcinoma patients from a database, and using a risk score determination module and a prognostic assessment module to divide all patients into high-risk and low-risk groups; the genes include PPARGC1A gene, SLC2A1 gene, SLC2A2 gene, SLC2A5 gene, ADH4 gene, DLAT gene, and NR1H3 gene; the database includes the TCGA database; Drug sensitivity data of cancer cell lines were downloaded from the anticancer drug sensitivity genomics database. The "oncoPredict" and "ggplot2" packages of R software were used to predict the sensitivity of high-risk and low-risk patients to different drugs at the gene expression level, and drugs with significant responses were selected. The WilCoxon test was used to compare drug sensitivity scores between the high-risk and low-risk groups. Drugs with low drug sensitivity scores in the low-risk group were suitable for low-risk patients, and drugs with low drug sensitivity scores in the high-risk group were suitable for high-risk patients.

[0035] In this embodiment of the invention, the medication assessment module utilizes the "oncoPredict" and "ggplot2" packages in R software to predict the sensitivity of high-risk and low-risk patients to different drugs at the gene expression level, thereby selecting drugs with significant responses. It was found that HCC patients showed good responses to nine of these drugs: low-risk HCC patients showed good responses to four drugs, indicating that these four drugs may be suitable for low-risk patients; high-risk patients showed higher sensitivity to five drugs, indicating that these five drugs are suitable for high-risk patients. This invention first classifies patients into high-risk and low-risk groups, and then selects medication based on risk scores, providing a new reference and selection for HCC medication.

[0036] To further illustrate the present invention, the following detailed description, in conjunction with embodiments, of the prognostic biomarkers for HCC patients and their applications, prognostic prediction models, and prognostic assessment devices provided by the present invention, should not be construed as limiting the scope of protection of the present invention.

[0037] Example 1 Data Acquisition for Research: Transcriptome data were downloaded from the TCGA database (https: / / portal.gdc.cancer.gov / ). Transcriptome data from hepatocellular carcinoma (HCC) and normal liver tissue were selected. Ultimately, transcriptome data from 374 HCC patients, along with clinical information (including patient age, sex, tumor stage, and tumor grade) and transcriptome data from 50 paired adjacent liver tissues were downloaded and obtained.

[0038] Transcriptome data were analyzed to identify genes with significantly different expression levels: a set of gluconeogenesis genes was downloaded from the molecular signature database (MSigDB, URL: https: / / www.gsea-msigdb.org / gsea / msigdb), and 117 gluconeogenesis-related genes were identified, as shown in Table 1.

[0039] Table 1. 117 Gluconeogenesis-related genes

[0040]

[0041]

[0042]

[0043] Gene expression data from all patients were standardized using the "limma" package in R software. The FPKM expression data from transcriptomic quantification were converted into log2(TPM+1) form for subsequent screening and analysis. TPM is an optimized RPKM calculation method that can be used for comparisons between different tissues of the same species. Unsupervised clustering was performed using nonnegative matrix factorization (NMF) to analyze 374 HCC samples from TCGA. The optimal number of clusters was determined using cluster heatmaps and genetic values ​​to maximize inter-group variation and minimize intra-group variation. All samples were divided into three clusters (Cluster A, Cluster B, and Cluster C). The results are shown in [Figure number missing]. Figure 1 Overall survival (OS) for different clusters was compared using the "survival" and "survminer" packages in R. Subsequently, the expression of seven key genes in each subgroup of HCC patients was detected using the "pheatmap" package in R, and heatmaps were used to visualize their distribution relative to clinical data. Results are shown below. Figure 2 .

[0044] Screening of pivot genes: Using the R software "survival" package, we first performed a differential expression analysis on 117 gluconeogenesis-related genes between HCC and normal liver tissues, followed by univariate Cox regression analysis. The calculated p-values ​​were used to determine whether these genes were significantly associated with sample survival (p-value < 0.05 indicated a correlation between the gene and sample survival). The results showed that 24 of the 117 gluconeogenesis-related genes in the dataset exhibited significant differences between HCC and normal liver tissues and were associated with overall survival (p-value < 0.05). See [link to relevant documentation]. Figure 3 middle.

[0045] LASSO regression analysis was performed using the "glmnet" package in R software. LASSO regression is a linear regression model that can be used for feature gene selection. The complexity of LASSO regression is directly related to the number of variables in the model; the more variables, the higher the model complexity. The degree of adjustment for LASSO regression complexity is controlled by the parameter Lambda(λ). A smaller λ corresponds to fewer independent variables and better performance. The minimum λ value was calculated using the "glmnet" package in R software, and the 7 genes corresponding to the minimum λ value were selected. The results are shown in [Figure number missing]. Figure 3 ; Construction of prognostic models: Regression analysis was performed using the "survival" package in R software. The regression coefficients of the seven pivot genes obtained from the analysis results, combined with the expression of these seven genes in patients, were used to calculate the risk score, which is the gluconeogenesis prognostic model formula as follows: Where n is the number of genes in the prognostic model, Expi is the expression value of the i-th gene in the model, and βi is the regression coefficient of the i-th gene calculated by multivariate Cox regression analysis.

[0046] The prognostic model obtained in this invention is: Risk score = (0.205×HMMRexp.) - (0.108×B3GAT3exp.) - (0.052×STC2exp.) + (0.460×NRIH3exp.) + (0.237×B4GALT2exp.) - (0.034×HOMER1exp.) + (0.039×SRD5A3exp.) Where exp. refers to the expression level of the gene in the sample; PPARGC1A refers to the nucleic acid encoding the PPARGC1A protein, with NCBI Gene ID: 10891. SLC2A1 refers to the nucleic acid encoding the SLC2A1 protein, with NCBI Gene ID: 6513; SLC2A2 refers to the nucleic acid encoding the SLC2A2 protein, with NCBI Gene ID: 6514; SLC2A5 refers to the nucleic acid encoding the SLC2A5 protein, with NCBI Gene ID: 6518. NR1H3 refers to the nucleic acid encoding the NR1H3 protein, with NCBI Gene ID: 10062; ADH4 refers to the nucleic acid that encodes the ADH4 protein, with NCBI Gene ID: 127.

[0047] Evaluation of prognostic models: The risk score for each HCC patient was calculated based on the risk score, and the median of these scores was taken (the median was 2.489). Patients were then divided into high-risk and low-risk groups based on the median score; those with a risk score equal to or higher than the median were in the high-risk group, and those with a score lower than the median were in the low-risk group. The model was then evaluated based on the high-risk and low-risk groupings.

[0048] First, Kaplan-Meier survival curves were plotted using the "survminer" and "survival" packages in R software. "survival" was used to calculate survival analysis, while "survminer" was used to summarize and visualize the results. In the Kaplan-Meier survival curves, the horizontal axis represents time, and the vertical axis represents overall survival. Different colored lines represent two groups of survival curves, and the curves can be used to determine the prognosis of patients in high-risk and low-risk groups. The results are shown in [Figure 1]. Figure 5 E in the text.

[0049] Receiver operating characteristic (ROC) curves were plotted using the "pROC" package in R software. The ROC curves were plotted with the true positive rate (sensitivity) on the ordinate and the false positive rate (1-specificity) on the x-axis. The area under the curve (AUC) was used to evaluate the predictive value of the gluconeogenesis prognostic model at 1, 3, and 5 years. An AUC > 0.5 and an AUC closer to 1 indicated better diagnostic efficacy. Results are shown in [Figure number missing]. Figure 6 F in the middle.

[0050] Next, this invention uses two methods to evaluate the grouping of the prognostic model: Principal Component Analysis (PCA) was performed on the two groups of samples using the R software package "epPCA". This projected the high-dimensional data into a low-dimensional space, with each point on the two-dimensional coordinate axis representing a different sample. The analysis focused on the dispersion of the scatter points in different groups. The results are shown in [Figure / Reference]. Figure 4 The fact that the samples from different groups can be distinguished well indicates that the method of differentiation is appropriate.

[0051] The tSNE analysis of the two sets of samples was performed using the R software package "Rtsne". tSNE (t-Distributed Stochastic Neighbor Embedding) analysis is also a dimensionality reduction method, representing high-dimensional datasets in two- or three-dimensional space for visualization. The results are shown in [Figure / Reference]. Figure 4 Similarly, each point on the two-dimensional coordinate axis represents a different sample, focusing on whether the samples are dispersed across different groups.

[0052] Both methods used in this invention demonstrate that the high-risk and low-risk groups based on risk scores can be well separated (the scatter points of the two groups rarely overlap), indicating that the risk score of this prognostic model is suitable for distinguishing patient samples.

[0053] Dataset validation: Transcriptomic data and clinical information from 374 HCC patients obtained from the TCGA database were used to obtain transcriptomic data and clinical survival information from 240 HCC patients using the LIRI-JP dataset from the ICGC database (https: / / dcc.icgc.org / projects / LIRI-JP). The prognostic model established in this invention was validated in both the 374 HCC patients from the TCGA database and the 240 HCC patients from the ICGC (LIRI-JP) dataset. The validation method was the same as above, and the results are shown below. Figure 4, Figure 5 , Figure 6 , Figure 7 and Figure 8 .

[0054] First, a risk score was calculated for each patient according to the prognostic model formula of this invention, and patients were ranked according to their risk scores. Then, a graph was plotted with the patient's risk score on the horizontal axis and survival time on the vertical axis. It was observed that as the risk score of HCC patients increased, their survival time shortened, indicating that the risk of hepatocellular carcinoma was related to survival time; patients with high risk scores had shorter survival times and poorer prognoses. The expression of these seven gluconeogenesis-related pivot genes was detected in both high-risk and low-risk groups of two datasets, and the expression levels were presented as heatmaps using the "pheatmap" package. It was found that all seven pivot genes were highly expressed in the high-risk group. Similar conclusions were obtained from both datasets, confirming the potential diagnostic value of the risk score constructed based on these seven genes.

[0055] Potential treatment options for high-risk and low-risk HCC patients: Drug sensitivity data for cancer cell lines were retrieved from the GDSC2 website (https: / / www.cancerrxgene.org / ), which contained sensitivity data for 969 cell lines and 298 chemicals. Drug sensitivity scores for the TCGA-LIHC cohort were calculated using the "oncopredict" R package. Sensitivity to different drugs was predicted at the gene expression level for high-risk and low-risk patients, and box plots were generated. Drug sensitivity analysis was performed using the "oncopredict" package, and box plots were generated using the "ggplot2" package. Results are shown below. Figure 9 and Figure 10 The Wilcoxon test was used to compare drug sensitivity scores between the high- and low-risk groups. Drugs with low drug sensitivity scores in the low-risk group were suitable for low-risk patients, and drugs with low drug sensitivity scores in the high-risk group were suitable for high-risk patients.

[0056] The results showed that HCC patients responded well to 9 of the drugs: low-risk HCC patients showed good responses to 4 drugs (lower drug sensitivity scores), indicating that these 4 drugs are suitable for low-risk patients; high-risk patients showed higher sensitivity to 5 drugs (lower drug sensitivity scores), indicating that these 5 drugs are suitable for high-risk patients. Using the prognostic model constructed in this invention, patients were divided into high-risk and low-risk groups, and medication was selected based on risk scores. This invention provides a new reference and selection for HCC medication.

[0057] Construction of the clinical model: Taking into account risk scores and clinical information (including patient age, gender, tumor stage and tumor grade), univariate Cox regression analysis was used to find that tumor stage and risk scores were associated with worse prognosis (p < 0.05).

[0058] The "survival" package was used to further construct a multivariate Cox regression model using these two factors. A nomogram was then used to visualize this model, allowing for intuitive graphical calculation of the survival probability for each specific patient. The results are shown below. Figure 10 and Figure 11 The calibration curve was plotted using the "timeROC" software package. The model was then calibrated using this calibration curve. The smaller the deviation between the actual curve and the ideal curve, the higher the reliability of the surface curve. Results are shown in [Figure number missing]. Figure 11 Therefore, this invention establishes a clinical model with good discriminative, calibration, and survival prediction capabilities by using a risk score based on the hepatocellular carcinoma tumor staging of seven glycogluconeogenesis hub genes.

[0059] Clinical translation approach: Based on the findings of this invention, a tissue microarray based on gluconeogenesis can be constructed. This microarray contains probes for detecting seven genes (PPARGC1A, SLC2A1, SLC2A2, SLC2A5, ADH4, DLAT, and NR1H3). This gene microarray offers possibilities for precision treatment of hepatocellular carcinoma patients. It enables more targeted and rapid prognostic assessment of postoperative hepatocellular carcinoma patients, facilitating clinical translation.

[0060] In summary, this invention constructs a gene combination model of seven gluconeogenesis-related genes. This gene model can be used to construct a tissue microarray based on gluconeogenesis metabolism to assess the prognosis of hepatocellular carcinoma patients. It can identify liver cancer patients with poor prognosis, guide clinicians to provide more aggressive treatment plans for high-risk patients, and at the same time avoid overtreatment of low-risk liver cancer patients.

[0061] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

[0062] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A prognostic marker for hepatocellular carcinoma, characterized in that, The prognostic markers are a combination of genes, which include the PPARGC1A gene, SLC2A1 gene, SLC2A2 gene, SLC2A5 gene, ADH4 gene, DLAT gene and NR1H3 gene. The NCBI Gene ID of the PPARGC1A gene is 10891, and the NCBI Gene ID of the SLC2A1 gene is 6513. The NCBI Gene ID of the SLC2A2 gene is 6514, the NCBI Gene ID of the SLC2A5 gene is 6518, the NCBI Gene ID of the ADH4 gene is 127, the NCBI Gene ID of the DLAT gene is 1737, and the NCBI Gene ID of the NR1H3 gene is 10062.

2. The use of the reagent for detecting the expression level of the prognostic marker of claim 1 in the preparation of products for assessing the prognosis of hepatocellular carcinoma.

3. The application according to claim 2, characterized in that, The product includes a reagent kit, a gene chip, or a device containing a gene detection module; the gene detection module is used to detect the expression level of the prognostic biomarker of claim 1.

4. The application according to claim 3, characterized in that, The prognosis includes the survival time of patients with hepatocellular carcinoma.

5. A prognostic assessment device for hepatocellular carcinoma, characterized in that, It includes a gene testing module, a risk score determination module, and a prognostic assessment module; The gene detection module is used to detect the expression levels of the prognostic markers in samples from hepatocellular carcinoma patients. The risk score determination module is used to calculate the prognostic risk score of hepatocellular carcinoma patients based on the expression level of the prognostic biomarker. The formula for calculating the prognostic risk score is as follows: The prognostic risk score is calculated as follows: (0.205 × SLC2A1 gene expression level) - (0.108 × PPARGC1A gene expression level) - (0.052 × ADH4 gene expression level) + (0.460 × NR1H3 gene expression level) + (0.237 × DLAT gene expression level) - (0.034 × SLC2A2 gene expression level) + (0.039 × SLC2A5 gene expression level). The prognostic assessment module is used to determine the prognosis of hepatocellular carcinoma patients based on this prognostic risk score. When the prognostic risk score is ≥2.489, the patient is considered high-risk and has a poor prognosis; when the prognostic risk score is <2.489, the patient is considered low-risk and has a good prognosis.

6. The prognostic assessment device according to claim 5, characterized in that, It also includes a calibration module; the calibration module is used to construct a multivariate Cox regression model with the prognostic risk score and the patient's tumor stage, visualize the multivariate Cox regression model using a nomogram, and calculate the patient's 1-year, 3-year, or 5-year survival probability.

7. The prognostic assessment device according to claim 5, characterized in that, It also includes a medication assessment module; the workflow of the medication assessment module includes: S1. Obtain the expression data of the seven genes in hepatocellular carcinoma patients from the TCGA database, and divide the patients into high-risk and low-risk groups through the risk score determination module and prognosis assessment module; S2. Download sensitivity data of potential chemotherapy and targeted drugs for liver cancer from anticancer drug sensitivity genomics databases (such as the GDSC database), and use the "oncoPredict" and "ggplot2" packages in R software to predict the sensitivity of high-risk and low-risk patients to different drugs based on gene expression levels; S3. Use the WilCoxon test to compare the drug sensitivity scores of the high-risk group and the low-risk group, and screen for drugs with significant differences in sensitivity: drugs with low sensitivity scores in the low-risk group are suitable for low-risk patients, and drugs with low sensitivity scores in the high-risk group are suitable for high-risk patients.

8. A method for constructing a prognostic prediction model for hepatocellular carcinoma, characterized in that, Includes the following steps: S1. Data Acquisition: Transcriptomic data and clinical information (including age, sex, tumor stage, and tumor grade) of 374 hepatocellular carcinoma patients and transcriptomic data of 50 adjacent normal liver tissues were obtained from the TCGA database; transcriptomic data and clinical survival information of 240 hepatocellular carcinoma patients were obtained from the ICGC database; and 117 gluconeogenesis-related genes were obtained from the MSigDB database. S2. Differential gene screening: Transcriptome data were standardized using the R software "limma" package (converting FPKM to log2 (TPM+1)). Unsupervised clustering was performed on 374 hepatocellular carcinoma samples from the TCGA database using nonnegative matrix factorization (NMF). Univariate Cox regression analysis was performed on 117 gluconeogenesis-related genes using the R software "survival" package to screen for genes that showed significant differential expression between hepatocellular carcinoma and normal tissues and were associated with overall survival (p < 0.05). S3. Hub gene identification: The differentially expressed genes obtained in step S2 were subjected to LASSO regression analysis using the R software "glmnet" package. The seven genes with the smallest model error (smallest λ value) were selected as hub genes, i.e., the prognostic biomarkers. S4. Model Construction: Using the "survival" package in R software, multivariate Cox regression analysis was performed on 7 key genes to obtain the regression coefficient of each gene. Combined with gene expression levels, a prognostic risk score calculation formula was constructed, namely the prognostic risk score formula, to form a prognostic prediction model for hepatocellular carcinoma. S5. Model Validation: The model constructed in step S4 was applied to 240 hepatocellular carcinoma samples from the ICGC database. The effectiveness of the model was validated by Kaplan-Meier survival analysis, ROC curve analysis (AUC value), PCA analysis, and tSNE analysis.

9. A method for assessing the prognosis of patients with hepatocellular carcinoma, characterized in that, Includes the following steps: S1. Collect tumor tissue samples from patients with hepatocellular carcinoma and use the gene detection module to detect the expression levels of the seven prognostic markers in the samples; S2. The patient's prognostic risk score is calculated using the aforementioned risk score determination module; S3. Determine the patient's risk level and prognosis using the aforementioned prognostic assessment module; S4. Calculate the patient's 1-year, 3-year, or 5-year survival probability through the calibration module, or select appropriate treatment drugs through the medication assessment module.

Images