CN112908470B - Hepatocellular carcinoma prognosis scoring system based on RNA binding protein gene and application thereof - Google Patents

Abstract

The invention provides a hepatocellular carcinoma prognosis scoring system based on RNA-binding protein genes and application thereof, wherein input variables of the hepatocellular carcinoma prognosis scoring system comprise a coefficient of kennel and a score of the hepatocellular carcinoma-related RNA-binding protein genes in a data set, and the score is determined according to the mRNA expression level and risk ratio of the hepatocellular carcinoma-related RNA-binding protein genes in the data set; the output variable of the hepatocellular carcinoma prognosis scoring system is a weighted sum of the coefficient of the kennel and the score of the hepatocellular carcinoma-associated RNA-binding protein gene. The hepatocellular carcinoma prognosis scoring system can effectively evaluate and predict prognosis of patients, has the characteristics of cross-platform and universal applicability, and has wide clinical application prospect, and the score is obviously related to part of clinical characteristics of the patients.

Description

A prognostic scoring system for hepatocellular carcinoma based on RNA-binding protein genes and its application

Technical field

The invention belongs to the technical field of oncology and relates to a hepatocellular carcinoma prognosis scoring system based on RNA binding protein genes and its application.

Background technique

Due to its insidious onset and difficulty in early diagnosis, hepatocellular carcinoma (HCC) has a 5-year survival rate of less than 20%. In recent years, significant progress has been made in surgery, radiotherapy and chemotherapy technology, targeted therapy technology and immunotherapy technology, bringing new hope to HCC patients. However, the curative effect for intermediate and advanced HCC is still very limited. Many molecular mechanisms in the occurrence and development of HCC are still unclear. Therefore, it is necessary to further improve the puzzle of HCC molecular mechanisms and find key molecules to facilitate early diagnosis, personalized treatment and prognosis judgment of HCC.

RNA-binding proteins (RBPs) are important proteins involved in post-transcriptional regulatory events. With their variable RNA-binding regions and flexible structures, RBPs dynamically control many RNA metabolic behaviors, including : RNA splicing, localization, transport and stability maintenance. It is now clear that RBPs can participate in the occurrence and development of tumors through post-transcriptional regulation and other functions. The target genes include oncogenic genes, cell cycle/apoptosis regulatory factor genes, autophagy regulatory factor genes, and inflammatory factor genes; cancer tissues and adjacent cancer tissues There are some differentially expressed RBPs in tissues, and these differentially expressed RBPs are related to the prognosis and clinical characteristics of patients. Recent high-throughput data analysis studies based on TCGA suggest that expression changes of RBPs exist in various cancers, especially HCC. Many studies have identified RBPs related to the occurrence and development of HCC: For example, Zhao et al. found that the RNA-binding protein RPS3 promotes the proliferation of HCC cells through post-transcriptional regulation of SIRT1 (ZhaoL, Cao J, Hu K, Wang P, Li G, He X, et al. al.RNA-binding protein RPS3 contributes tohepatocarcinogenesis by post-transcriptionally up-regulating SIRT1. NucleicAcids Res 2019;47(4):2011-28.); Dong et al. found that RNA-binding protein RBM regulates the production of circular RNASCD-circRNA 2 To promote the growth of HCC cells (Dong W, Dai ZH, Liu FC, Guo XG, Ge CM, DingJ, et al. The RNA-binding protein RBM3 promotes cell proliferation in hepatocellular carcinoma by regulating circular RNA SCD-circRNA2production. EBioMedicine2019; 45: 155-67).

However, most previous studies focus on the function and mechanism of individual RBPs in HCC cells in vitro. The expression patterns of the same gene in different data sets may be significantly different, or even completely opposite results. If only a small number of cohorts are included in the study, the results obtained are often limited, not generally applicable, and cannot reflect the real situation. The existing technology lacks systematic review studies and clinical application studies on RBPs. Although some studies have involved the relationship between RBPs and clinical prognosis, these studies are mostly based on a single data set, so the research results are inconsistent and contradictory.

Therefore, it is necessary to obtain RBPs-related data with consistency and clinical application value based on multiple cohorts and large-scale samples.

Contents of the invention

In view of the shortcomings and actual needs of the existing technology, the present invention provides a hepatocellular carcinoma prognostic scoring system based on RNA binding protein genes and its application. The hepatocellular carcinoma prognostic scoring system can effectively predict the overall survival rate of HCC patients ( overall survival (OS) and disease-free survival (DFS), which have high clinical application value.

To achieve this goal, the present invention adopts the following technical solutions:

In a first aspect, the present invention provides a hepatocellular carcinoma prognostic scoring system based on RNA binding protein genes. The input variables of the hepatocellular carcinoma prognostic scoring system include the Gini coefficient of the hepatocellular carcinoma-related RNA binding protein genes in the data set. and score;

The score is determined based on the mRNA expression level and risk ratio of the hepatocellular carcinoma-associated RNA-binding protein gene in the data set.

In the present invention, based on the results of Cox survival analysis and random forest model, the hepatocellular carcinoma prognostic scoring system RBP-score is constructed using the mRNA expression values of 10 key HCC-related RBPs genes. RBP-score combines the 10 HCC-related The importance of RBPs genes is related to the prognosis and clinical characteristics of HCC patients. In different HCC data sets, patients with different RBP-scores have significant differences in overall survival rate and disease-free survival rate. The higher the RBP-score, the better the overall survival rate. The worse the disease-free survival rate and disease-free survival rate, the hepatocellular carcinoma prognostic scoring system RBP-score is an effective, cross-platform, and universally applicable patient prognosis assessment tool. The prognostic assessment ability is not weaker than the clinical TNM staging method and has practical Clinical value.

Preferably, the hazard ratio (HR) is determined based on the overall survival rate of the hepatocellular carcinoma-associated RNA binding protein gene in the data set based on the Cox proportional hazard model, preferably based on GSE14520, TCGA-LIHC and ICGC-LIRI-JP, The HR is obtained using the univariate COX proportional risk model. When calculating the COX proportional risk model, the software will automatically give the risk ratio, which is the HR value.

Preferably, the score is 0 or 1;

The mRNA expression level of the hepatocellular carcinoma-related RNA-binding protein gene is ≥ the average expression level (median expression level) and the risk ratio is > 1, or the mRNA expression level of the hepatocellular carcinoma-related RNA-binding protein gene is < average Expression level (median expression level) and risk ratio <1, the score is 1, otherwise the score is 0.

Preferably, the hepatocellular carcinoma-related RNA binding protein genes include PRPF3, SLBP, CPEB3, PPARGC1A, IGF2BP3, SF3B4, ILF2, CSTF2, ACO1 and FBL.

Preferably, the data set includes any one or a combination of at least two of the Comprehensive Gene Expression Database of the Hepatocellular Carcinoma Cohort, the Hepatocellular Carcinoma Genome Atlas, or the International Cancer Genome Consortium Japanese Liver Cancer Data.

In the present invention, by integrating multiple large-scale cross-platform HCC cohort mRNA expression data, 30 hepatocellular carcinoma-related RNA-binding protein genes with consistent expression differences in HCC tissues were identified, and further used machine learning algorithm random forest screening 10 key hepatocellular carcinoma-related RNA-binding protein genes were identified, with better accuracy and higher application value.

Preferably, the hepatocellular carcinoma cohort gene expression comprehensive database includes any one or a combination of at least two of GSE14520, GSE22058, GSE25097, GSE36376, GSE45436, GSE64041, GSE76427, GSE54236 or GSE63898.

Preferably, the data set includes GSE14520, Hepatocellular Carcinoma Genome Atlas, and International Cancer Genome Consortium Japanese Liver Cancer Data.

Preferably, the output variable of the hepatocellular carcinoma prognostic scoring system is the weighted sum of the Gini coefficient and the score of the hepatocellular carcinoma-related RNA binding protein gene.

Preferably, the calculation formula of the hepatocellular carcinoma prognostic scoring system is:

RBP-score=∑(Gene_score×Gene_Weight)

Among them, Gene_Weight is the Gini coefficient, and Gene_score is the score, which takes the value 0 or 1.

In the present invention, the hepatocellular carcinoma prognostic scoring system RBP-score constructed based on the random forest algorithm has high potential application value in indicating and predicting the prognosis of patients. RBP-score can effectively predict the OS and DFS of HCC patients, and is comparable to other patients. Clinical features TNM stage, AFP and risk of metastasis are also related to a certain extent, and their clinical value has been verified in data sets from different platforms. Cox analysis suggests that high RBP-score is an independent risk factor for poor prognosis in HCC patients.

Specifically, the Gene_score is determined according to the following table. If the integrated HR of a gene is >1 and the mRNA expression is ≥ the median value, or the integrated HR is <1 and the mRNA expression is < the median value, the Gene_score of the gene is 1; otherwise ,Gene_score is 0. High expression of gene expression levels in the table is defined as expression value ≥ median value, low expression is defined as expression value < median value, the integrated HR value is based on GSE14520, TCGA-LIHC and ICGC-LIRI-JP data using univariate COX proportion score Obtained from the risk model.

In a second aspect, the invention provides a hepatocellular carcinoma-related RNA-binding protein gene marker, which includes PRPF3, SLBP, CPEB3, PPARGC1A, IGF2BP3, SF3B4, ILF2, CSTF2, ACO1 and FBL.

Preferably, the hepatocellular carcinoma-related RNA binding protein gene markers also include POLR2G, MBNL2, PUF60, RALY, LSM4, CASC3, ZFP36, LARP1, SNRPC, TSNAX, RBM34, IGF2BP2, ABCF1, NSUN6, RBMS3, NONO, LSM2, SNRPB, ZGPAT and XPO5.

In the present invention, the overall technical roadmap is shown in Figure 1. First, in the gene chip expression comprehensive database of 9 hepatocellular carcinoma cohorts, a robust ranking integration algorithm (RRA) was used to screen out 30 RBPs with clear functions from 430. For RBPs with consistent expression patterns in the 9 HCC cohorts, the Robust Ranking and Integration Algorithm (RRA) can effectively integrate microarray data from multiple different platforms to obtain effective integration results. The robust ranking and integration algorithm is used to not only screen out the specificity Good HCC correlation RBPs, and achieved data dimensionality reduction, narrowing the research scope from 430 RBPs to 30 RBPs, greatly reducing the research burden. Subsequently, the integrated results were verified in the RNA sequencing data of the two cohorts TCGA-LIHC and ICGC-LIRI-JP. The expression patterns of these 30 HCC-related RBPs in the RNA sequencing data were consistent with those in the 9 mRNA gene chip data. The expression pattern is completely consistent.

In the present invention, the random forest algorithm was further used to calculate the importance of 30 HCC-related RBPs genes in determining the patient's 5-year survival period, and the 10 most important HCC-related RBPs genes were screened out.

In a third aspect, the present invention provides a method for screening hepatocellular carcinoma-related RNA-binding protein gene markers described in the second aspect, and the screening method includes:

RNA-binding protein genes with consistent expression differences were screened out from the comprehensive gene expression database of the hepatocellular carcinoma cohort, and the initial hepatocellular carcinoma-related RNA-binding protein genes were obtained.

Preferably, 30 RNA-binding protein genes with consistent expression differences were screened out from the gene expression comprehensive database of 9 hepatocellular carcinoma cohorts using the Robust rank aggregation algorithm (RRA), and further analyzed in TCGA-LIHC and ICGC-LIRI. - Validation in the JP cohort; 30 initial RNA-binding protein genes with consistent expression differences were analyzed for copy number variation, single nucleotide mutation and promoter region methylation degree. Some RNA-binding protein genes are risk factors for patient prognosis. or protective factors.

Preferably, the screening method further includes:

Using the hepatocellular carcinoma genome map as a training set, the training set samples were divided into 5-year survival patients and 5-year non-survival patients, and a random forest classification model was established;

The random forest classification model is used to classify the initial hepatocellular carcinoma-related RNA-binding protein genes to obtain key hepatocellular carcinoma-related RNA-binding protein genes.

In the fourth aspect, the present invention provides the application of the hepatocellular carcinoma prognosis scoring system described in the first aspect in the preparation of hepatocellular carcinoma prognosis monitoring products.

Preferably, the hepatocellular carcinoma prognosis monitoring product includes a hepatocellular carcinoma prognosis monitoring kit and/or a hepatocellular carcinoma prognosis monitoring medical device.

Compared with the prior art, the present invention has the following beneficial effects:

(1) The present invention integrates and analyzes multiple large-scale cross-platform HCC cohort mRNA expression data to obtain HCC-related RBPs gene expression profiles with highly consistent expression differences, and uses the machine learning model random forest to construct a hepatocellular carcinoma prognosis scoring system. , the hepatocellular carcinoma prognostic scoring system is a simple and powerful prognostic evaluation tool with cross-platform characteristics and is suitable for different subgroups of HCC patients;

(2) The HCC-related RNA-binding protein gene identified in the present invention may be used as a new target for diagnosis and treatment of hepatocellular carcinoma and used for HCC prognosis prediction and assessment.

Description of the drawings

Figure 1 is the overall technology roadmap;

Figure 2A shows the identification of 30 HCC-related RBPs genes with highly consistent expression characteristics in the HCC cohort through the robust ranking integration algorithm. Figure 2B shows 30 HCC-related RBPs genes with highly consistent expression characteristics in 9 different HCC chip data cohorts. Figure 2C shows the expression of 30 HCC-related RBPs genes with highly consistent expression characteristics in TCGA-LIHC RNA sequencing data. Figure 2D shows the expression of 30 HCC-related RBPs genes with highly consistent expression characteristics in ICGC-LIRI-JPRNA. Expression status in sequencing data. Figure 2E shows the expression status of 30 HCC-related RBPs genes with highly consistent expression characteristics in TNM stages of HCC patients in the TCGA-LIHC cohort. Figure 2F shows 30 HCC-related RBPs genes with highly consistent expression characteristics. Expression status in TNM staging of HCC patients in the ICGC-LIRI-JP cohort, Figure 2G shows 30 HCC-related RBPs genes with highly consistent expression characteristics that can effectively distinguish cancer tissue and peritumoral tissue in the TCGA-LIHC cohort, Figure 2H shows 30 HCC-related RBPs genes with highly consistent expression characteristics can effectively distinguish cancer tissue and peritumoral tissue in the ICGC-LIRI-JP HCC cohort;

Figure 3A uses the Cox model to calculate the overall survival hazard ratio of 30 HCC-related RBPs genes in TCGA-LIHC. Figure 3B shows the Kaplan-Meier curve of RBPs genes related to overall survival in TCGA-LIHC. Figure 3C shows TCGA-LIHC. Kaplan-Meier curve of RBPs genes related to disease-free survival in TCGA-LIHC, GSE14520 and ICGC-LIRI-JP. Figure 3D shows the comprehensive survival analysis results of 30 RBPs genes related to HCC in three HCC data sets: TCGA-LIHC, GSE14520 and ICGC-LIRI-JP;

Figure 4A shows the overall survival rate of TCGA-LIHC patients using the Kaplan-Meier curve. Figure 4B shows the overall survival rate of the GSE14520 patients using the Kaplan-Meier curve. Figure 4C uses the Kaplan-Meier curve to calculate the overall survival rate of ICGC-LIRI-JP patients. Overall survival rate, Figure 4D is the use of ROC curve to evaluate the accuracy and specificity of RBP-score in predicting 1-year, 3-year and 5-year overall survival in TCGA-LIHC patients, Figure 4E is the use of ROC curve to evaluate the accuracy and specificity of RBP-score in predicting GSE14520 patients. Accuracy and specificity of 1-year, 3-year and 5-year overall survival rates. Figure 4F shows the use of ROC curve to evaluate the accuracy and specificity of RBP-score in predicting 1-year, 3-year and 5-year overall survival rates of ICGC-LIRI-JP patients. Specificity, Figure 4G shows the use of Kaplan-Meier curve to calculate the disease-free survival rate of TCGA-LIHC patients, Figure 4H shows the use of Kaplan-Meier curve to calculate the disease-free survival rate of GSE14520 patients;

Figure 5A shows the prognostic evaluation results of each subgroup of patients using RBP-score after HCC patients were stratified by gender. Figure 5B shows the prognostic evaluation results of each subgroup of patients using RBP-score after HCC patients were stratified by age. Figure 5C Figure 5D shows the prognostic evaluation results of HCC patients using RBP-score for each subgroup of patients after stratification by TNM stage. Figure 5D shows the prognostic evaluation results of HCC patients using RBP-score for each subgroup of patients after stratification by AFP level. Figure 5E Figure 5F shows the prognostic evaluation results of each subgroup of patients using RBP-score after stratifying HCC patients according to HBV infection status. Figure 5F shows the prognostic evaluation results of each subgroup of patients using RBP-score after stratifying HCC patients according to HCV infection status. Figure 5G shows the prognostic evaluation results of each subgroup of patients using RBP-score after HCC patients were stratified by liver cirrhosis.

Detailed ways

In order to further illustrate the technical means adopted by the present invention and its effects, the present invention will be further described below with reference to the embodiments and drawings. It can be understood that the specific embodiments described here are only used to explain the present invention, but not to limit the present invention.

If specific techniques or conditions are not specified in the examples, the techniques or conditions described in literature in the field shall be followed, or the product instructions shall be followed. If the manufacturer of the reagents or instruments used is not indicated, they are all conventional products that can be purchased through regular channels.

Example 1

This example first obtains the mRNA chip data GSE14520, GSE22058, GSE25097, GSE36376, and 9 HCC cohorts from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) GSE45436, GSE64041, GSE76427, GSE54236 and GSE63898, The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA- LIHC), collecting International Cancer Genome Consortium Japanese liver cancer (ICGC-LIRI-JP) data from ICGC DCC (https://dcc.icgc.org/releases).

The gene identifiers of the above HCC cohort data sets are all the latest HUGO gene names, and the mRNA expression data are normalized using log2.

In the following examples, R software (version 3.6.1) was used for statistical analysis. The difference in normally distributed data between the two groups was analyzed using the independent sample t test. The difference in non-normally distributed data was analyzed using the wilcoxon test; single The relationship between HCC-related RBPs genes and RBP-score and patient OS and DFS was analyzed using Kaplan-Meier survival analysis log-rank test. Single HCC-related RBPs genes, RBP-score and other clinical indicators that affect OS were analyzed using single factor and Multifactor Cox analysis was obtained; the relationship between RBP-score and clinical characteristics such as TNM stage and AFP level of patients was analyzed using chi-square test; P<0.05 was defined as a statistically significant difference.

Example 2

This embodiment uses Robust rank aggregation (RRA) (Kolde R, Laur S, Adler P, Vilo J. Robust rank aggregation for gene list integration and meta-analysis. Bioinformatics 2012; 28(4):573-80 ) Integrated 9 mRNA chip data to obtain mRNA with consistent expression patterns, and the RRA algorithm was executed using R software (version 3.6.1). Select genes with differential expression fold change (fold change) >1.5 or <-1.5 and P<0.05 in cancer tissues and normal tissues to establish the HCC RRA list. The HCC RRA list contains a total of 1326 genes with significant differential expression. These genes are in 9 There was consistent up-regulation or down-regulation in the HCC cohort.

By searching The database of RNA-binding specificities (RBPDB, http://rbpdb.ccbr.utoronto.ca/) and referring to the research work of Gerstberger (Gerstberger S, Hafner M, Tuschl T. A census of human RNA-binding proteins. Nat Rev Genet2014;15(12):829-45), 430 RNA-binding protein genes were obtained from significantly differentially expressed genes, and a human RBPs gene list was constructed. The translation products of the genes in this list are functionally identified RNAs that can exert Functional protein binding.

As shown in Figure 2A, the HCC RRA list and the RBPs gene list were intersected, and 30 RBP mRNAs with consistent differential expression in the chip data sets of the 9 HCC cohorts were identified, which were defined as HCC-associated RBPs genes (HCC-associated RBPs genes). . The expression differences of these 30 RBP mRNAs in 9 HCC cohorts (cancer tissue vs. normal tissue) are shown in Figure 2B. Among them, 8 RBP mRNAs showed low expression in HCC tissues (P<0.05), and 22 RBP mRNAs showed low expression in HCC tissues (P<0.05). It was highly expressed in HCC tissues (P<0.05).

Next, the RNA sequencing data of TCGA-LIHC and ICGC-LIRI-JP were used to verify the expression of HCC-related RBPs genes. As shown in Figure 2C and Figure 2D, 30 HCC-related RBPs genes were expressed in TCGA-LIHC and ICGC- The expression profile in LIRI-JP was highly consistent with that in the 9 HCC cohorts.

Further analysis of the expression of 30 HCC-related RBPs genes in different TNM stage tissues, as shown in Figure 2E and Figure 2F, there are significant differences in the expression of some RNA-binding proteins such as XPO5 and CPEB3 in early tissues and late tissues, and this This difference is consistent with its changing trend in normal tissues and tumor tissues, so these RNA-binding proteins are likely to play a tumor-promoting or tumor-suppressing role. The PCA analysis results of 30 HCC-related RBPs genes in TCGA-LIHC and ICGC-LIRI-JP are shown in Figure 2G and Figure 2H. PCA analysis found that the mRNA expression profiles of 30 HCC-related RBPs genes can effectively distinguish tumors. tissue and normal tissue, which shows that the 30 identified HCC-related RBPs genes are highly related to HCC and have the value of further research.

Example 3

This example further explores the clinical application value of 30 HCC-related RBPs genes and analyzes the association between the mRNA expression data of individual RBP genes and prognosis in three HCC cohorts GSE14520, TCGA-LIHC and ICGC-LIRI-JP.

The survival analysis results of 30 RBP genes based on the Cox proportional hazards model are shown in Figure 3A; using the median value of mRNA expression as cut-off, each HCC-related RBP gene has an impact on overall survival and disease-free survival. The Kaplan-Meier curves of the rate are shown in Figure 3B and Figure 3C (only the results of log-rank test P<0.05 are shown); the integrated analysis results of the three data sets are shown in Figure 3D.

Most of the 30 RBPs genes are related to the survival of HCC patients, suggesting that these RBPs genes may play a tumor-promoting or tumor-suppressing role. After integrating the three data sets of TCGA-LIHC, GSE14520 and ICGC-LIRI-JP, it can be found that the performance of genes such as CSTF2, SF3B4, PPARGCA1 and RALY is relatively consistent in the three data sets. Based on the above evidence, it is shown that some of the 30 HCC-related RBPs genes are closely related to the survival of HCC patients.

Example 4

In order to obtain RBPs gene markers containing a smaller number of genes, this embodiment uses a random forest algorithm to further screen key HCC-related RBPs genes. Proceed as follows:

Using the TCGA-LIHC data set as the training set, all patients were divided into 5-year survival patients and 5-year non-survival patients, and a random forest classification model was established, with the mRNA expression data of HCC-related RBPs genes as input variables (ntree=500); After 10-fold cross validation (CV=10) stratification, 10 key HCC-related RBPs genes were selected based on the mean Gini value (cut-off=5.1) to construct RBPs genes. landmark.

In order to apply HCC-related RBPs genes to clinical applications, this example uses the screened 10 key HCC-related RBPs genes to construct an HCC prognostic score system (RBP-score). The calculation formula of RBP-score is as follows:

RBP-score=∑(Gene_score×Gene_Weight)

Among them, Gene_Weight is the Gini coefficient of the random forest model, Gene_score is calculated and determined based on the mRNA expression levels of 10 key HCC-related RBPs genes and the corresponding integrated hazard ratio (Integrated HR), and Integrated HR is calculated based on 10 key HCC-related RBPs genes. The integrated results of Cox regression analysis of the overall survival rate of HCC-related RBPs genes in the three cohorts of GSE14520, TCGA-LIHC and ICGC-LIRI-JP determined that if the Integrated HR of a gene is >1 and the mRNA expression is >the average expression , or Integrated HR<1 and mRNA expression <average expression, then Gene_score=1 for the gene, otherwise Gene_score=0.

Next, the relationship between RBP-score and patient prognosis was verified in three cohorts: GSE14520, TCGA-LIHC and ICGC-LIRI-JP. The RBP-score of the patient's HCC tissue was calculated according to the above formula, and the lower quartile of the RBP-score was used. , median and upper quartile as cut-off, divide the patients into four groups: Q1, Q2, Q3 and Q4, RBP-scoreQ1<RBP-scoreQ2<RBP-scoreQ3<RBP-scoreQ4, analyze the overall results of each group Overall survival (OS) and disease-free survival (DFS).

As shown in Figure 4A, Figure 4B and Figure 4C, it can be clearly observed that the patient OS decreases with the increase of RBP-score; as shown in Figure 4D, Figure 4E and Figure 4F, ROC analysis RBP-score for 1 year The prediction accuracy of OS, 3-year OS, and 5-year OS was >65% in each dataset; as shown in Figure 4G and Figure 4H, in GSE14520 and TCGA-LIHC, a higher RBP-score implies a poorer DFS.

According to the chi-square analysis results of RBP-score and other clinical characteristics of HCC patients (TCGA-LIHC andGSE14520), it can be found that patients with higher RBP-score have AFP>300ng/mL, TNM stage is advanced (III-IV), and CLIP stage is Advanced stage (>3), tumor size >5cm, and vascular invasion are more likely to occur.

In TCGA-LIHC and GSE14520, the results of the Cox proportional hazards analysis (Coxproportional hazards analysis) combined with other clinical characteristics of patients showed that RBP-score was an independent risk factor for poor overall survival in HCC patients (HRTCGA-LIHC=2.57, HRGSE14520=1.66, P<0.05).

Example 5

This example performs subgroup-survival analysis and divides HCC patients into subgroups based on seven clinical parameters: gender, age, TNM stage, alpha-fetoprotein (AFP) level, HBV status, HCV status, and liver cirrhosis. , patients in each subgroup were further divided into high RBP-score group and low RBP-score group (cut-off = mean RBP-score), and the overall survival (OS) and disease-free survival of each group were analyzed. rate (disease-free survival, DFS).

As shown in Figure 5A, Figure 5B, Figure 5C, Figure 5D, Figure 5E, Figure 5F and Figure 5G, using the median value as cut-off, RBP-score can effectively predict OS in each subgroup of TCGA-LIHC. Similar subgroup-survival analysis was also conducted on GSE14520 and ICGC-LIRI-JP. Except for subgroups with distribution bias, RBP-score can indicate OS in most subgroups. It should be noted that RBP-score can also effectively predict OS in patients of the same clinical stage. It can be seen that the HCC molecular prognostic scoring system based on RBPs genes has universal applicability.

In summary, the hepatocellular carcinoma prognostic scoring system constructed based on the hepatocellular carcinoma-related RNA binding protein gene of the present invention is a simple and powerful prognostic evaluation tool, which is suitable for different subgroups of HCC patients and has cross-platform characteristics. , the identified hepatocellular carcinoma-related RNA-binding protein gene may serve as a new target for HCC diagnosis and treatment.

The applicant declares that the present invention illustrates the detailed methods of the present invention through the above embodiments, but the present invention is not limited to the above detailed methods, that is, it does not mean that the present invention must rely on the above detailed methods to be implemented. Those skilled in the art should understand that any improvements to the present invention, equivalent replacement of raw materials of the product of the present invention, addition of auxiliary ingredients, selection of specific methods, etc., all fall within the protection scope and disclosure scope of the present invention.

Claims (6)

1. A hepatocellular carcinoma prognosis scoring system based on an RNA-binding protein gene, wherein the input variables of the hepatocellular carcinoma prognosis scoring system comprise a coefficient of kurther and a score of the hepatocellular carcinoma-associated RNA-binding protein gene in a dataset;

the score is determined based on mRNA expression levels and risk ratios of the hepatocellular carcinoma-associated RNA-binding protein gene in the dataset;

the hepatocellular carcinoma-associated RNA binding protein gene comprises PRPF3, SLBP, CPEB3, PPARGC1A, IGF BP3, SF3B4, ILF2, CSTF2, ACO1 and FBL;

the data set comprises GSE14520, hepatocellular carcinoma genomic profile, and international cancer genome alliance, japan liver cancer data;

the output variable of the hepatocellular carcinoma prognosis scoring system is a weighted sum of the coefficient of the kennel and the score of the hepatocellular carcinoma-associated RNA-binding protein gene;

the calculation formula of the hepatocellular carcinoma prognosis scoring system is as follows:

RBP-score＝∑(Gene_score×Gene_Weight)

wherein RBP-score is the score in the hepatocellular carcinoma prognosis scoring system, gene_weight is the coefficient of foundation, gene_score is the score, and Gene_score takes a value of 0 or 1.

2. The hepatocellular carcinoma prognostic scoring system according to claim 1, wherein the risk ratio is determined based on overall survival of the hepatocellular carcinoma-associated RNA binding protein gene in the dataset based on a univariate Cox-proportional risk model.

3. The hepatocellular carcinoma prognosis scoring system of claim 1, wherein the score is 0 or 1;

the mRNA expression level of the relevant RNA binding protein gene of the liver cell cancer is more than or equal to the average expression level and the risk ratio is more than 1, or the mRNA expression level of the relevant RNA binding protein gene of the liver cell cancer is less than the average expression level and the risk ratio is less than 1, the score is 1, otherwise, the score is 0.

4. A method of screening for a gene marker for a hepatocellular carcinoma-associated RNA binding protein, the method comprising:

screening out RNA binding protein genes with consistent expression difference from a comprehensive database of liver cell cancer queue gene expression to obtain initial liver cell cancer related RNA binding protein genes;

the screening method further comprises the steps of:

taking a liver cell cancer genome map as a training set, dividing the training set sample into a 5-year survival patient and a 5-year non-survival patient, and establishing a random forest classification model;

classifying the initial hepatocellular carcinoma-related RNA binding protein genes by using the random forest classification model to obtain key hepatocellular carcinoma-related RNA binding protein genes;

the RNA binding protein genes include PRPF3, SLBP, CPEB3, PPARGC1A, IGF2BP3, SF3B4, ILF2, CSTF2, ACO1, and FBL.

5. A method of preparing a hepatocellular carcinoma prognosis monitoring product, characterized in that the method employs the hepatocellular carcinoma prognosis scoring system of any one of claims 1-3.

6. The method of claim 5, wherein the hepatocellular carcinoma prognosis monitoring product comprises a hepatocellular carcinoma prognosis monitoring kit and/or a hepatocellular carcinoma prognosis monitoring medical device.