Gene model for predicting prognosis of hepatocellular carcinoma patient as well as construction method and application of gene model

Abstract

The invention provides a gene model for predicting prognosis of a hepatocellular carcinoma patient as well as a construction method and application of the gene model. The construction method comprises the following steps: S1, acquiring transcriptional profile expression data and prognosis data of the hepatocellular carcinoma patient; s2, expression data of 26 m6A RNA methylation regulation genes are obtained through screening from transcriptional profile data, 9 prognostic genes related to the lifetime are obtained through LASSO regression analysis screening and are used for constructing a risk scoring model, the risk scoring model is expressed as RiskScore = # imgabs0 #, Coefi is the coefficient of a gene i, xi is the expression quantity of the gene i, and n is the number of the prognostic genes; and S3, calculating the risk score of each subject based on the risk scoring model. The model can effectively evaluate and predict prognosis and chemotherapy sensitivity of hepatocellular carcinoma patients through risk scores.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technology, and particularly relates to a gene model for predicting the prognosis of hepatocellular carcinoma patients, a construction method thereof, and an application thereof. Background Art

[0002] Hepatocellular carcinoma (HCC) is one of the most common malignant tumors worldwide. There are approximately 500,000 new cases of liver cancer each year, and hepatocellular carcinoma accounts for up to 85%. Although with the wide application of tumor marker detection and imaging diagnosis technology, as well as the improvement of surgical level and the development of new treatment methods such as transcatheter arterial chemoembolization, the 5-year survival rate of primary liver cancer has increased, but the overall prognosis is still not ideal. One of the main reasons is the lack of effective prognostic markers, which cannot stratify patients according to risk, so early intervention and treatment cannot be carried out for high-risk patients.

[0003] In recent years, RNA epigenetics has attracted much attention in the field of tumor research. Among them, m 6 A methylation (N6-methyladenosine, m 6 A), as one of the most representative modifications, its abnormal level is closely related to the occurrence and development of various diseases and cancers.

[0004] However, there is no relevant research on the relationship between m 6 A RNA methylation regulatory genes and the prognosis of hepatocellular carcinoma (HCC). Therefore, exploring the relationship between m 6 A RNA methylation regulatory genes and hepatocellular carcinoma and constructing a prediction model or tool for hepatocellular carcinoma based on m 6 A RNA methylation regulatory genes is of great significance for better predicting the progression, prognosis, and treatment effect of hepatocellular carcinoma. Summary of the Invention

[0005] To solve the problems in the background art, the present invention provides a gene model for predicting the prognosis of hepatocellular carcinoma patients, a construction method thereof, and an application thereof. This model can effectively evaluate and predict the prognosis and chemotherapy sensitivity of hepatocellular carcinoma patients through risk scoring. The technical solution of the present invention to solve the above technical problems is as follows: In the first aspect, the present invention provides a construction method of a gene model for predicting the prognosis of hepatocellular carcinoma patients, including the following steps: S1. Obtain the transcriptional profile expression data and prognosis data of hepatocellular carcinoma patients; S2. Screen and obtain the expression data of 26 m 6 A RNA methylation regulatory genes from the transcriptional profile data, and based on the obtained 26 m6 The expression data and prognosis data of genes regulated by RNA methylation were used to screen prognosis genes related to survival through LASSO regression analysis, and they were used to construct a risk score model, which is expressed as: RiskScore= , where Coefi is the coefficient of gene i, xi is the expression level of gene i, and n is the number of prognosis genes; S3. Based on the above risk score model, calculate the risk score of each subject.

[0006] According to the above scheme, the prognosis genes related to survival are: ALKBH5, METTL14, ZC3H13, METTL3, YTHDF2, LRPPRC, KIAA1429, IGF2BP2 and YTHDF1.

[0007] According to the above scheme, the risk score model of hepatocellular carcinoma patients is specifically as follows: Riskscore = (-0.0493 × expression level of METTL14 gene) + (0.0981 × expression level of METTL3 gene) + (0.1132 × expression level of LRPPRC gene) + (0.0718 × expression level of KIAA1429 gene) + (-0.2646 × expression level of ZC3H13 gene) + (0.00845 × expression level of IGF2BP2 gene) + (0.2928 × expression level of YTHDF1 gene) + (0.5313 × expression level of YTHDF2 gene) + (-0.0307 × expression level of ALKBH5 gene).

[0008] According to the above scheme, the patients were divided into high-risk group and low-risk group using the optimal cut-off value of the constructed risk score, and there were significant differences in survival between the two groups.

[0009] According to the above scheme, obtain the clinicopathological factor information of hepatocellular carcinoma patients, and construct a nomogram based on the clinicopathological factor information and the calculated risk score to predict the survival probability of hepatocellular carcinoma patients; the clinicopathological factors include: age, gender, tumor stage, pathological grade and TNM stage.

[0010] According to the above scheme, according to the constructed risk score model, calculate the optimal cut-off value, and divide the patients into high-risk group and low-risk group according to the optimal cut-off value.

[0011] In the second aspect, the present invention provides a gene model for predicting the prognosis of hepatocellular carcinoma patients constructed by the above method.

[0012] In a third aspect, the present invention provides the use of the above gene model for predicting the prognosis of hepatocellular carcinoma patients in evaluating the prognosis of hepatocellular carcinoma patients.

[0013] In a fourth aspect, the present invention provides the use of the above gene model for predicting the prognosis of hepatocellular carcinoma patients in predicting the therapeutic effect of immune checkpoint inhibitors.

[0014] In a sixth aspect, the present invention provides the use of the above gene model for predicting the prognosis of hepatocellular carcinoma patients in evaluating the chemosensitivity of patients.

[0015] The beneficial effects of the present invention are as follows: (1) By exploring the relationship between m 6 RNA methylation regulatory genes and the prognosis of hepatocellular carcinoma patients, the present invention successfully constructed a hepatocellular carcinoma prognosis model evaluation containing 9 m 6 RNA methylation regulatory genes. With the help of ROC curve analysis, during the 1-year, 3-year, and 5-year follow-up processes, the AUC value of this model exceeded 0.72. Compared with traditional clinical prognostic indicators (such as T, N, M staging), this model is more accurate in predicting the prognosis of patients, showing significant superiority; (2) The gene model for predicting the prognosis of hepatocellular carcinoma patients constructed by the present invention can effectively predict chemosensitivity, providing strong support for the personalized treatment of hepatocellular carcinoma patients; (3) The gene model for predicting the prognosis of hepatocellular carcinoma patients constructed by the present invention can effectively predict the therapeutic effect of immune checkpoint inhibitors (ICI). The therapeutic effect of high-risk group patients receiving ICI treatment is relatively poor, while low-risk group patients perform better. This finding provides an important reference for whether patients are suitable for receiving ICI treatment; (4) By calculating the risk score of hepatocellular carcinoma patients through the hepatocellular carcinoma prognosis model of the present invention and combining the nomogram constructed with clinicopathological features, the 1-year, 2-year, and 3-year survival probabilities of patients can be visually presented, which cannot be achieved by the traditional TNM staging prognosis system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is the expression heat map of 26 m 6 RNA methylation regulatory genes between the adjacent cancer tissues and cancer tissues of hepatocellular carcinoma patients in the training dataset and validation dataset of the present invention, where A is the training dataset and B is the validation dataset; Figure 2 It is based on m 6Principal component analysis (PCA) of ARNA methylation-regulated genes and overall survival and disease-free survival graphs of two subtypes. Among them, A is the overall survival (OS) of Cluster1 / 2 subtypes in the TCGA-LIHC cohort, B is the disease-free survival (DFS) of Cluster1 / 2 subtypes in the TCGA-LIHC cohort, C is the overall survival (OS) of Cluster1 / 2 subtypes in the ICGC-LIRI-JP cohort, D is the principal component analysis (PCA) of the entire mRNA expression profile in the TCGA-LIHC cohort, and E is the principal component analysis of the entire mRNA expression profile in the ICGC-LIRI-JP cohort; Figure 3 For the present invention based on m 6 Construction and parameter tuning optimization graph of the LASSO-Cox regression model for the hepatocellular carcinoma model regulated by ARNA methylation genes; Figure 4 For the present invention based on m 6 Diagnostic efficacy of the hepatocellular carcinoma model regulated by ARNA methylation genes. Among them, A and D are Kaplan–Meier survival analyses of high- and low-risk groups in the TCGA-LIHC cohort (A) and the ICGC-LIRI-JP cohort (D) respectively, B and E are the distributions of risk scores and survival statuses of liver cancer patients in the TCGA-LIHC cohort (B) and the ICGC-LIRI-JP cohort (E), and C and F are the ROC curves for predicting 1-year, 3-year, and 5-year OS of liver cancer patients in the TCGA-LIHC cohort (C) and the ICGC-LIRI-JP cohort (F) by the m6ARNA methylation prognostic model; Figure 5 Graph showing the relationship between the expression levels of ICI-related biomarkers and risk scores for the present invention; Figure 6 For the present invention, the predicted IC of chemotherapeutic drugs in high- and low-risk groups 50 Box plot; Figure 7 Construction and evaluation of the nomogram in the TCGA-LIHC cohort (training group) for the present invention, and construction of the nomogram model for predicting the survival rate of liver cancer patients in the 7ATCGA-LIHC cohort; Figures B, C, and D are calibration curves at three time points of 1 year, 2 years, and 3 years in sequence. Detailed implementation manners

[0017] The principles and features of the present invention will be described below in conjunction with the accompanying drawings and specific embodiments. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0018] Example 1 1. Obtaining transcriptional profile expression data of hepatocellular carcinoma patients TCGA-LIHC cohort data (training dataset): Downloaded from The Cancer Genome Atlas (TCGA) database, which contains standardized RNA-seq datasets (mRNA in FPKM format) and clinical follow-up information, including a total of 424 samples from 374 liver cancer patients (50 of which are paired adjacent normal tissues); ICGC-LIRI-JP cohort data (validation dataset): The validation set was downloaded from the International Cancer Genome Consortium (ICGCG) database, which contains standardized RNA-seq datasets (mRNA expression information) and clinical follow-up information, including a total of 243 liver cancer tissues and 202 adjacent normal tissues.

[0019] 2. Based on m 6 It is feasible to construct a prognostic prediction model for hepatocellular carcinoma (HCC) by combining m The expression levels of 26 m 6 RNA methylation regulatory genes in cancer tissues and adjacent tissues were detected, and their expression heatmaps are as Figure 1 shown. In the TCGA-LIHC cohort, 22 out of 26 m 6 RNA methylation regulatory genes showed significant differences between cancer tissues and adjacent tissues, and were highly expressed in liver cancer tissues; a similar phenomenon was also observed in the ICGC-LIRI-JP cohort, with 25 showing significant differences.

[0020] Based on the expression data of m 6 RNA methylation regulators, principal component analysis (PCA) was performed on the TCGA-LIHC and ICGC-LIRI-JP cohorts, and the overall survival and disease-free survival of the two subtypes were analyzed. The results are as Figure 2 shown. In the TCGA-LIHC cohort, it was found that the overall survival (OS, p<0.0001) and disease-free survival (DFS, p = 0.009) of Cluster2 were shorter than those of Cluster1 (2A and 2B). The gene expression patterns between the two subtypes were further analyzed using the principal component analysis (PCA) method, and it was found that the gene expression profiles between the two subtypes were well differentiated ( Figure 2 D); using the same method, consistent results were also found in the ICGC-LIRI-JP cohort. According to k = 2, the patients were divided into two subgroups [Cluster1 (n = 166) and Cluster2 (n = 77)]. The overall survival prognosis of Cluster2 was worse than that of Cluster1 ( Figure 2C), PCA analysis can also separate the gene expression profiles of the two subtypes well (Figure 2E). The above PCA results show that the gene expression profiles of the two subtypes can be well distinguished. Based on m 6 RNA methylation regulators can be used for prognostic classification of liver cancer.

[0021] 3. Construction of prognostic model: LASSO regression analysis The construction and parameter adjustment optimization of the prognostic model are as Figure 3 shown. Perform LASSO regression analysis on 26 m 6 RNA methylation regulatory genes to screen variables. Specifically, use the glmnet function in the R package glmnet, and based on 26 m 6 RNA methylation regulatory genes, carry out machine learning LASSO regression modeling work, and perform cross-validation operations with the cv.glmnet function. First, calculate the candidate gene coefficients ( Figure 3 A), determine the optimal lambda value parameter ( Figure 3 B). After selecting lambda.min as the optimal lambda parameter, finally determine 9 m 6 RNA methylation regulatory genes, and use these 9 genes to construct a multivariate Cox model, that is, a risk score model ( Figure 3 C).

[0022] RiskScore = , where Coefi is the coefficient of gene i, xi is the expression level of gene i, and n is the number of prognostic genes.

[0023] Specifically, the risk score: Riskscore = (-0.0493 × expression level of METTL14 gene) + (0.0981 × expression level of METTL3 gene) + (0.1132 × expression level of LRPPRC gene) + (0.0718 × expression level of KIAA1429 gene) + (-0.2646 × expression level of ZC3H13 gene) + (0.00845 × expression level of IGF2BP2 gene) + (0.2928 × expression level of YTHDF1 gene) + (0.5313 × expression level of YTHDF2 gene) + (-0.0307 × expression level of ALKBH5 gene).

[0024] Calculate the risk score of each patient according to the above formula. Through the R software packages "survival" and "survminer", and use the log-rank test to determine the optimal cut-off value of the risk score. According to this optimal cut-off value, divide the patients into a high-risk group (risk score higher than this optimal cut-off value) and a low-risk group (risk score lower than this optimal cut-off value).

[0025] The diagnostic efficacy data of this risk scoring model are as follows Figure 4 shown. The Kaplan-Meier survival analysis results in the TCGA-LIHC cohort and the ICGC-LIRI-JP cohort were consistent. The overall survival rate of liver cancer patients in the high-risk scoring group was significantly lower than that of patients in the low-risk scoring group. The risk score and survival status distribution plots also showed that the survival period of patients in the high-risk scoring group was lower than that of the low-risk scoring group. m 6 The area under the curve (AUC) of the overall survival rate at 1 year, 3 years, and 5 years of the ARNA methylation model in the TCGA-LIHC cohort (training cohort) was 0.748, 0.779, and 0.746, respectively, m 6 The ARNA methylation model also had good predictive efficacy in the ICGC-LIRI-JP cohort (validation cohort), and the area under the curve (AUC) of the overall survival rate at 1 year, 3 years, and 5 years was 0.744, 0.721, and 0.75, respectively.

[0026] 4. Influence of the risk scoring model on the efficacy of ICI treatment The relationship between immune checkpoint inhibitor (ICI) markers and risk scores was analyzed in the training dataset and the validation dataset. The results are as follows Figure 5 shown. The expression levels of markers such as CTLA4 (p < 0.001), CD276 (p < 0.001), VTCN1 (p < 0.001), CD70 (p < 0.01), HAVCR2 (p < 0.001), CD40 (p < 0.001), and CD47 (p < 0.001) in the high-risk scoring group were significantly higher than those in the low-risk scoring group, and the risk score value was significantly positively correlated with the expression levels of these markers.

[0027] The above results indicate that the efficacy of ICI treatment for CTLA4, CD276, VTCN1, CD70, HAVCR2, CD40, and CD47 in the high-risk scoring group of patients is better than that in the low-risk scoring group.

[0028] 5. Effectively predict the sensitivity to chemotherapy based on the risk score of patients Based on the expression profiles of cell lines and IC 50 (half maximal inhibitory concentration) data in the GDSC database, a ridge regression prediction model was constructed using the pRRophetic package to calculate the IC 50 value of each patient and evaluate the application of the risk scoring model in clinical chemotherapy response. The results are as follows Figure 6As shown, Figures A - F are box plots of the predicted IC of six drugs, namely Cisplatin, Doxorubicin, Gemcitabine, Mitomycin C, Paclitaxel, and Vinblastine in the high - and low - risk groups of the TCGA - LIHC cohort. Figures G - L are box plots of the predicted IC of the six drugs in the high - and low - risk groups of the ICGC - LIRI cohort. 50 In the TCGA - LIHC cohort, for the above six chemotherapeutic drugs, there were significant differences in the predicted IC (Estimated IC 50 ) between the high - and low - risk groups. The high - risk score group was associated with a lower half - maximal inhibitory concentration (IC 50 ), indicating that the high - risk score group was more sensitive to commonly used chemotherapy (Cisplatin, Doxorubicin, Gemcitabine, Mitomycin C, Paclitaxel, and Vinblastine; p - value < 0.001 for all). Similarly, in the ICGC - LIRI cohort, the results were consistent with those in the TCGA - LIHC cohort. There were significant differences in the predicted IC 50 between the high - and low - risk groups, and the half - maximal inhibitory concentration (IC 50 ) of the low - risk score group was significantly higher than that of the high - risk score group, indicating that the high - risk score group was more sensitive to commonly used chemotherapy. These results suggest that the predictive model of m 50 ARNA regulatory factors can be used as a potential predictive indicator for the chemosensitivity of hepatocellular carcinoma patients, better guiding clinical chemotherapy drug use.

[0029] 6. A nomogram model was constructed based on the risk score and clinicopathological factors to accurately predict the death risk of hepatocellular carcinoma patients. 50 (Estimated IC 50 ) 50 Based on the risk score and clinicopathological factor information (age, gender, tumor stage, pathological grade, and TNM stage), a nomogram model was constructed using the Nomogram function to accurately predict the death risk of hepatocellular carcinoma patients. 50 The predictive accuracy was evaluated using the C - index and calibration curves, and finally, the clinical utility was evaluated using decision curve analysis. The results are as shown. In the construction of the nomogram model for predicting the survival rate of liver cancer patients in the TCGA - LIHC cohort; B, C, and D are calibration curves at three time points of 1 year, 2 years, and 3 years respectively. The calibration plots show the overlap of the prediction curve (black solid line) of the nomogram model and the standard prediction curve (45° gray solid line). 50 6

[0030]

[0031]

[0032] Figure 7

[0033] The C-index of the nomogram model for the TCGA-LIHC cohort (training group) was 0.712, and the C-index in the ICGC-LIRI cohort (validation group) was 0.743. The nomogram prediction model showed good concordance and discrimination in predicting the survival rate of liver cancer patients.

[0034] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for constructing a gene model for predicting the prognosis of patients with hepatocellular carcinoma, characterized in that: The following steps are involved: S1. Obtain transcriptional expression profiles and prognostic data for patients with hepatocellular carcinoma; S2. 26 mRNAs were obtained from the transcriptional profile data. 6 A RNA methylation regulated gene expression data, and based on the 26 m 6 A RNA methylation regulatory gene expression data and prognostic data were used to screen out prognostic genes related to survival through LASSO regression analysis, which were used to construct a risk scoring model. The risk scoring model is expressed as: RiskScore= , where Coefi is the coefficient of gene i, xi is the expression level of gene i, and n is the number of prognostic genes; S3. Calculate the risk score for each subject based on the above risk scoring model.

2. The method for constructing a gene model for predicting the prognosis of patients with hepatocellular carcinoma according to claim 1, characterized in that: The prognostic genes associated with survival were: ALKBH5, METTL14, ZC3H13, METTL3, YTHDF2, LRPPRC, KIAA1429, IGF2BP2, and YTHDF1.

3. The method for constructing a gene model for predicting the prognosis of patients with hepatocellular carcinoma according to claim 2, characterized in that: The risk score model for patients with hepatocellular carcinoma is as follows: Riskscore=(-0.0493×METTL14 gene expression level) + (0.0981×METTL3 gene expression level) +(0.1132×LRPPRC gene expression level) + (0.0718×KIAA1429 gene expression level) + (-0.2646×ZC3H13 gene expression level) + (0.00845×IGF2BP2 gene expression level) + (0.2928×YTHDF1 gene expression level) +(0.5313×YTHDF2 gene expression level) + (-0.0307×ALKBH5 gene expression level).

4. The method for constructing a gene model for predicting the prognosis of hepatocellular carcinoma patients according to any one of claims 1 to 3, characterized in that: Obtaining clinicopathological factor information of patients with hepatocellular carcinoma, constructing a nomogram based on the clinicopathological factor information and the calculated risk score, and predicting the survival probability of patients with hepatocellular carcinoma; The clinical pathological factors include: age, gender, tumor stage, pathological grade and TNM stage.

5. The method for constructing a gene model for predicting the prognosis of hepatocellular carcinoma patients according to any one of claims 1 to 3, characterized in that: According to the constructed risk score model, the optimal cutoff value was calculated, and patients were divided into high-risk group and low-risk group according to the optimal cutoff value.

6. A gene model for predicting the prognosis of hepatocellular carcinoma patients constructed according to the method of any one of claims 1 to 5.

7. Use of the gene model for predicting the prognosis of hepatocellular carcinoma patients according to claim 6 in evaluating the prognosis of hepatocellular carcinoma patients.

8. Use of the gene model for predicting the prognosis of patients with hepatocellular carcinoma according to claim 6 in predicting the therapeutic effect of immune checkpoint inhibitors.

9. Use of the gene model for predicting the prognosis of patients with hepatocellular carcinoma according to claim 6 in evaluating the sensitivity of patients to chemotherapy.

Images