CN114317756B

CN114317756B - Application of marker

Info

Publication number: CN114317756B
Application number: CN202210018328.0A
Authority: CN
Inventors: 陈文标; 戴勇; 汤冬娥; 候显良; 徐慧旋; 郑凤屏; 张丰
Original assignee: Shenzhen Longhua Peoples Hospital
Current assignee: Shenzhen Longhua Peoples Hospital
Priority date: 2022-01-07
Filing date: 2022-01-07
Publication date: 2024-05-07
Anticipated expiration: 2042-01-07
Also published as: CN114317756A

Abstract

The invention relates to the field of biotechnology, in particular to application of a marker. The invention provides application of gene combination as a marker in preparation of products for predicting tumors, predicting tumor stages and/or predicting tumor prognosis. The gene combination-3-PPI-Mod (3 NM) obtained by the invention is verified in two independent data sets, and the Kaplan-Meier curve shows that the total survival rate time of patients classified as high risk patients is obviously shorter than that of patients classified as low risk patients; and 3-PPI-Mod has a better prognostic significance than the 18-gene, exhibiting better prognostic value than the features produced by gene expression alone. Furthermore, 3-PPI-Mod is significantly associated with intratumoral hypoxia scores. These results indicate that 3-PPI-Mod is highly expressed in tumor exosomes of high clinical grade and can be used as a prognostic and predictive biomarker for hepatocellular carcinoma.

Description

Application of marker

Technical Field

The invention relates to the field of biotechnology, in particular to application of a marker.

Background

Exosomes (exosomes) are membrane particles released by cells, modulating intercellular communication by delivering functional molecules (such as proteins, nucleic acids and lipids) to recipient cells. Cells can secrete exosomes of many different sizes and origins. The outer body is mainly between 50-150nm in diameter, resulting from the polycystic inner body. Other exosomes (100-1000 nm diameter), such as Microbubbles (MVS), exosomes or microparticles, bud directly from the cell membrane.

In recent years, there has been increasing evidence that exosomes play an important role in the development of tumors. Cancer-derived exosomes are involved in a variety of cancerous processes including malignant transformation, angiogenesis, immunosuppression, invasion and therapeutic resistance. Exosomes released by the tumor microenvironment may also affect the properties of cancer cells. One common feature of tumor microenvironments is the induction of tumor exosomes to the receptor cell metastasis invasion and metastasis phenomena. In addition, exosomes released by tumors can alter the distant microenvironment, forming a pre-metastatic niche, promoting the formation of metastases. Suggesting that cancer cell exosomes may play a role both locally and remotely.

Cancerous exosomes detected in various body fluids are considered to be a new non-invasive biomarker. For example, small exosomes (less than 200nm in diameter) in the circulation carry microRNAs (miRNAs) and proteins, which are promising diagnostic and prognostic biomarkers in tumors. However, exosomes identified in body fluids may represent mixed populations of tumor and other tissue release. To date, distinguishing tumor-specific exosomes from exosomes released by other tissues remains a significant challenge due to the lack of specific biomarkers. Protein levels in exosomes released into body fluids are reported to be consistent with protein expression in NSCLC (non-small cell lung cancer) tissues. Thus, comprehensive analysis of primary tumor and secretory exosomes may be a viable method of identifying tumor-specific exosomes-associated biomarkers.

Disclosure of Invention

In view of this, the present invention provides the use of markers that can be prognostic and predictive biomarkers for hepatocellular carcinoma.

In order to achieve the above object, the present invention provides the following technical solutions:

The invention provides application of gene combination as a marker in preparation of products for predicting tumors, predicting tumor stage and/or tumor prognosis;

The gene combinations include one or more of gene combination 1, gene combination 2, or gene combination 3;

The gene combination 1 comprises genes KRT1, HOXC13, KRT34, KRT2, KRT31, HOXC12, HOXC11 and HOXC5;

the gene combination 2 comprises genes PAX2, GBX1, LHX1, WNT7A, SOST, SP7 and FGF23;

The gene combination 3 comprises genes SPAG6, WDR63 and DNAH9.

In some embodiments of the invention, the above gene combination 1 is associated with developmental differentiation of epidermal cells; the gene combination 2 is related to tumor related channels; the above gene combination 3 is involved in the constitution and/or activity of cell membrane components.

In some embodiments of the invention, the above-described gene combination 1 forms an interactive relationship with the above-described gene combination 2 via the gene ISL 1; the gene combination 2 is in an interaction relationship with the gene combination 3 through the genes PVALB, NPY, PF, SPAG 17.

In some embodiments of the invention, expression of the above-described gene combinations is up-regulated in high clinical staging tumors.

In some embodiments of the invention, the above-described combination of genes predicts survival of cancer patients.

In some embodiments of the invention, the above gene combinations are significantly positively correlated with the ratio of immune cells.

In some embodiments of the invention, the ratio of partial genes to immune cells in the above described gene combinations is significantly positively correlated.

In some embodiments of the invention, the above-described gene combinations are associated with a tumor microenvironment.

In some embodiments of the invention, the above-described combination of genes is positively correlated with tumor hypoxia.

In some embodiments of the invention, hypoxia pathway-related genes are significantly enriched in the high-risk tumors predicted by the above-described gene combinations.

In some embodiments of the invention, the above-described score for tumor hypoxia is significantly positively correlated with the expression level of hypoxia inducible factor 1.

In some embodiments of the invention, the above gene combinations are significantly positively correlated with the expression level of hypoxia inducible factor 1 in a high risk subgroup.

The invention also provides a primer combination for amplifying the gene combination.

The invention also provides a reagent, a kit, a system or a device for predicting tumor, predicting tumor stage and/or tumor prognosis, which is characterized by comprising the amplification primer of the gene combination or a product aiming at obtaining the sequence of the gene combination, and acceptable auxiliary materials, auxiliary agents, carriers, modules or components.

In some embodiments of the invention, the steps of manipulating the above-described reagents, kits, systems or devices include extracting exosome genes, sequencing, and/or analyzing the sequencing results using the above-described gene combinations.

The invention also provides a detection method, which comprises the steps of designing and synthesizing primers by using the gene combination as a template and/or analyzing a sample gene by using the gene combination as a reference.

In some embodiments of the invention, the steps of the above detection method comprise extracting exosome genes, sequencing, and/or analyzing the sequencing results using the above gene combinations.

The marker of the invention has the following effects:

1. Expression of 3-PPI-Mod (3 NM) was up-regulated in tumors of the high clinical grade group. The risk of clinical features in patients was predicted by 3-PPI-Mod, with an area under ROC of 0.7368. 79.74% (126/158) of patients were correctly classified when compared to the true clinical grade. The Kaplan-Meier curve demonstrates that 3-PPI-Mod can predict overall survival of patients (p=0.0057), demonstrating that 3-PPI-Mod can be used to predict tumor progression as well as tumor prognosis.

2. In 3-PPI-Mod, mod_5 is involved in the developmental differentiation of epidermal cells; mod_8 is significantly associated with tumor-associated pathways; the gene in mod_15 is widely involved in the composition and activity of cell membrane components.

3. The same algorithm is used for establishing a prediction model-18 genes based on gene expression, and the result 18-genes have poorer prognosis significance than 3-PPI-Mod, which shows that the related characteristics of the exosome show better prognosis value compared with the characteristics of the pure gene expression.

4. All 3 modules in 3-PPI-Mod are positively correlated with hypoxia, and hypoxia pathway related genes are remarkably enriched in high-risk tumors predicted by 3-PPI-Mod. Intratumoral hypoxia scores have a significant correlation with hypoxia inducible factor 1 (HIF 1) expression levels, and up-regulation of HIF1 expression levels in a high risk subgroup identified by 3-PPI-Mod, suggesting that 3-PPI-Mod may be useful for predicting, assessing, and tumor hypoxia microenvironment.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.

FIG. 1 illustrates an algorithm of a modular construction;

FIG. 2 shows a flow chart of the method of the present patent;

FIG. 3 illustrates PPI network construction; wherein blue ellipses represent exosome-related genes; black lines represent interactions between blue ellipses;

FIG. 4a illustrates cluster mapping according to the expression scores of the 16 module configuration file; the cluster map also shows the link between clusters and clinical pathology stage; T-N-M: tumor classification forms represent tumor-lymph node-metastasis;

FIG. 4b illustrates determining an optimal module using an RF algorithm; the abscissa represents the number of modules of the study and the ordinate represents the cross-validation of each predictive model;

FIG. 4c shows expression scoring based on 3 modules; the abscissa represents the modules 5, 8, 15; the ordinate represents the module score; red is a high clinical tumor stage; blue is low clinical tumor stage;

FIG. 4d shows the validation of prognosis prediction for 3NM by ROC curve; the abscissa indicates the specificity; the ordinate represents sensitivity;

FIG. 4e shows a survival analysis of TCGA-LIHC HCC patients based on 3NM features using KM analysis; the abscissa indicates time; the ordinate represents the survival probability;

FIG. 5a shows the interaction of exosome-related genes from 3 modules by overlapping genes;

FIG. 5b shows GO and KEGG analysis of 3 modules;

FIG. 6 shows survival analysis of HCC patients from different data sets using KM analysis based on 3NM exosome-associated genes; wherein a is survival analysis of GEO-GSE76427 dataset; b is ICGC-LIHC-survival analysis of the JP dataset;

FIG. 7 shows survival analysis of HCC patients on TCGA-LIHC dataset using KM analysis based on 3NM exosome-related genes;

FIG. 8a shows a correlation analysis of 3NM with tumor immune cells and tumor microenvironment status;

FIG. 8b shows tumor hypoxia status using GSEA analysis according to 3NM classification for each group;

FIG. 8c shows the analysis of HIF1A correlation with hypoxia using the Pearson correlation algorithm; the abscissa represents hypoxia score; the ordinate indicates HIF1A expression scores;

FIG. 8d shows a comparison of high and low risk groups using tumor microenvironment status scores (hypoxia); the abscissa represents the grouping; the ordinate represents the hypoxia status score;

FIG. 8e shows a comparison of high and low risk groups using tumor microenvironment status score (HIF 1A); the abscissa represents the grouping; the ordinate indicates HIF1A expression scores.

Detailed Description

The invention discloses application of a marker, and a person skilled in the art can properly improve the technological parameters by referring to the content of the marker. It is expressly noted that all such similar substitutions and modifications will be apparent to those skilled in the art, and are deemed to be included in the present invention. While the methods and applications of this invention have been described in terms of preferred embodiments, it will be apparent to those skilled in the relevant art that variations and modifications can be made in the methods and applications described herein, and in the practice and application of the techniques of this invention, without departing from the spirit or scope of the invention.

Exosomes (exosomes) mediate intercellular communication in the tumor microenvironment and are involved in tumor invasion phenomena. Although exosomes in body fluids are considered ideal biomarkers for tumor diagnosis and prognosis, it is difficult to distinguish between exosomes of tumor origin and exosomes released by other tissues. The invention assumes that analysis of exosomes-associated molecules in tumor tissue helps to estimate the prognostic value of tumor-specific exosomes, and analyzes the gene expression profile of exosomes protein-encoding genes in hepatocellular carcinoma within tumors. Based on a protein-protein interaction (PPI) network method, exosomes related gene characteristics are established, and 3 network modules (3-PPI-Mod) are determined as characteristics for predicting clinical grades of hepatocellular carcinoma. This feature is further validated against the predicted values in both retrospective datasets. In addition, the relationship of gene signature to tumor microenvironment, including hypoxia status and stromal cell abundance, was also studied. These results indicate that the 3-PPI-Mod signal is highly expressed in tumor exosomes of high clinical grade and can be used as a prognostic and predictive biomarker for hepatocellular carcinoma.

Hypoxia/anoxia as referred to herein refers to a condition in which the partial pressure of oxygen in a tissue drops below a critical value or oxygen reduction is effectively utilized.

The microenvironment referred to in the present invention refers to the cellular matrix and the body fluid components therein.

Tumor microenvironment refers to the fact that tumor development, growth and metastasis are closely related to the internal and external environment in which tumor cells are located, including not only the structure, function and metabolism of the tissue in which the tumor is located, but also to the internal environment of the tumor cells themselves (nuclear and cytoplasmic).

The invention is further illustrated by the following examples:

Example 1: establishing predictive features for clinical grade risk of cancer

1. Patient and clinical characteristics

The invention uses data from 3 different databases as the subject of investigation. They are LIHC (hepatocellular carcinoma) transcript sequencing data from TCGA (TCGA-LIHC), japanese case transcript sequencing queues from LIHC of ICGC (ICGC-LIHC-JP) and LIHC chip sequencing data from the Singapore bioinformatic center (GEO: GSE 76427), respectively. These 3 different independent queues cover two different transcript quantification methods, different populations in asia and europe. Wherein, the total number of samples of TCGA-LIHC data is 356, and the number of usable samples after OS (overall survival rate) is less than 30 days and clinical data insufficiency cases is 316. Similarly, ICGC-LIHC-JP data total number of samples 212, available number of samples 161, GSE76427 data total number of samples 115, available number of samples 94. The sample information is detailed in table 1.

Table 1 clinical characteristics of patients in different LIHC databases

TCGA data was downloaded from:

https://www.cancer.gov/about-nci/organization/ccg/research/structural-genomics/tcga/

ICGC data were downloaded from:

https://dcc.icgc.org/

GSE76427 data was downloaded from:

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgiacc＝GSE76427/

2. exoRBase exosome database

The liver cell cancer-associated exosome genes are from exosome gene database exoRBase. The download address is http:// www.exorbase.org/exoRBase/browse/tomRNAIndex/. The database of liver cell cancer exosome characteristic genes comprises 17221 genes. Further enrichment by bioinformatics means is required to obtain exocrine expression genes closely related to hepatocellular carcinoma.

3. Gene expression data processing

TCGA-LIHC data is two sets of data of read counts and FPKM, ICGC-LIHC-JP data is read counts, and GSE76427 data extracts probe values (log 2 intensity) and probe notes from an original file downloaded from a gene expression integrated database (GEO). The data normalization was performed by Z-transforming the FPKM data of TCGA-LIHC, ICGC-LIHC-JP and GSE76427 data. Differential expression analysis of tumor and paranormal samples was performed using read counts data from TCGA-LIHC.

In the differential expression analysis, the invention uses EBSeq software to analyze the gene expression difference between the tumor and the paracancerous normal sample in the TCGA-LIHC queue.

4. PPI data processing

Construction of the human protein interaction network PPI was performed using STRING (https:// STRING-db. Org /). The build process uses default parameters.

And identifying the PPI module by using the MCODE. Degree cutoff value is 2,node score cutoff, K-score value is 2, max. Depth value is 100.

5. Identification of clinical grade related PPI modules

To further screen out of the potential PPI modules with significant clinical grading prediction functions. The PPI module obtained by the method is integrated with the expression data of TCGA-LIHC. The present invention first calculates the Expression Score (Expression Score; e) of each module. In a given module M with M genes, the expression score (e) of M in sample j is defined as:

Wherein Z _ij is the Z-transgene expression value of gene i.

Then, the discriminant score S (M) of the M module is defined as the Mutual Information (MI) between class e' and clinical classification class (c):

Where e' represents a discrete form of e, the expression score e is discretized to 9 (log ₂ (N) +1), N being the number of samples. The calculation process is shown in fig. 1. In clinical grading, stage I and II are classified as Low Stage group and III and IV are classified as HIGH STAGE group.

Thereafter, the present invention calculates MI values of randomly selected "modules" by randomly extracting the same genes as the number of M module genes from PPI network genes. Each module was run randomly 1000 times. And (3) carrying out statistical test by using the calculation result of random sampling and the calculation result of an actual module, and selecting a module with significant p value (p < 0.0001) for further analysis.

6. Recursive predictive PPI module

Gene expression scores of candidate modules as described previously were used to construct PPI network-based gene signatures. Feature selection and modeling is performed using R-packets randomForest based on a Random Forest (RF) algorithm, i.e., the predictive importance of each candidate module repetition is estimated using the initial RF of 5000 decision trees. And determining the optimal combination of the recursion prediction candidate modules by adopting a step-by-step backward selection method. In each iteration, 10% of the features were excluded and the remaining features were used to construct an RF containing 3000 decision trees. This procedure is stopped when only two functions remain. Among all the iteration results, the RF model with the smallest feature number is selected. In the final RF model, the clinical grade risk of the patient is determined by oob (out-of-bag data) probabilities. Finally, the invention selects 3 PPI modules meeting the requirements.

7. Other statistical methods

The present invention uses R package clusterProfiler to perform GO and KEGG path enrichment analysis on PPI modules. GSEA was used to compare the gene set of interest to a subgroup of patients classified by 3-PPI-Mod. Correlation of PPI modules with hypoxia, angiogenesis and inflammation scores and stromal cell abundance was calculated by pearson correlation coefficients. Multiple tests were adjusted by the error discovery rate (FDR) using the method of Benjamini-Hochberg. Patient survival rates were compared for the 3-PPI-Mod assigned low-risk and high-risk groups using the Kaplan-Meier curve and the log-rank test. The multivariate cox model was used to evaluate the prognostic and predictive value of 3-PPI-Mod signals. All statistical analyses were performed using R software (version 3.3.1). p <0.05 indicates significant differences.

8. Results and analysis

Fig. 2 depicts an overall flow chart of the present invention.

1134 Specific tumor and paracancerous differential expressed genes (PPDE > 0.99) were obtained by analysis of the present invention. On the basis, the invention uses the differential expression genes and the liver cancer specific exosome expression genes in exoRBase databases to carry out intersecting operation, thus obtaining 437 liver cancer exosome specific expression genes.

Based on the obtained 437 liver cancer exosomes specifically expressed genes, PPI network constructed by STRING was used. The network contains 321 nodes, 938 interactions (fig. 3).

For specific proteins in the exosomes of hepatocellular carcinoma, 321 proteins were mapped onto a reference PPI network. The exogenously related PPI network integrates with the gene expression profile of the training dataset.

And identifying the PPI modules by using the MCODE, and finally obtaining 16 potential PPI modules. The set of genes included in the module are as follows:

Module 1: PYY, PF4, NPY4R, CCR, ADCY2, OPRD1, NMU, GRM3, CNR2, NPY, PPBP, PPY, CHRM4, GALR1, OXGR1, HTR1F, HTR a;

Module 2: MYH7, MYL2, ACTN3, MYBPC3, MYH2, MYH6, MYBPC2, TNNT3

Module 3: GYPA, GYPB, RHAG, GATA1, AHSP, ALAS2, HBA1, HBQ1;

Module 4: HBD, HBE1, SCN2A, CALB, BDNF, CARTPT, SLC17A8, SLC18A3, NTSR1, GHSR, CHRM1, KISS1R, QRFPR;

Module 5: KRT1, HOXC13, KRT34, KRT2, KRT31, HOXC12, HOXC11, HOXC5;

and (6) module 6: ALPPL2, CEACAM5, RAET1L, LYPD2, NTNG1, NLGN1, SLC17A7, OLFM3, GABRG2, RBFOX1;

Module 7: UCP1, HCRT, LEP, NR A1;

Module 8: PAX2, GBX1, LHX1, WNT7A, SOST, SP7, FGF23;

Module 9: CATSPERD, CACNG3, CACCNG 6;

module 10: AZU1, DNTT, DEFA3, MPO, CD1A;

Module 11: chra 2, PCDH19, LGI1;

Module 12: PVALB, NCAN, GRIA2;

module 13: EMX2, HOXD, IRX6;

module 14: CLCA2, KLK5, DSG3;

Module 15: SPAG6, WDR63, DNAH9;

Module 16: FBXO40, RNF182, FBXL13.

A random sampling method is used to estimate the significance of the discrimination score for each module. Finally, by random forest modeling analysis, the results showed that the recurrence score was significantly higher for 16 modules than for contingency (p <0.001, table 2). The details of the 16 candidate blocks are shown in table 2.

Table 2 module MI values and significance

Module	MI value	P value
			Module 8	0.06861101	7.18×10^-28
Module 15	0.06685537	1.55×10^-38
			Module 5	0.06665415	1.37×10^-11
Module 16	0.06464605	5.05×10^-10
			Module 1	0.06455428	1.66×10^-10
Module 6	0.06433018	5.85×10^-10
			Module 4	0.06269076	3.76×10^-7
Module 10	0.06220704	4.32×10^-7
			Module 11	0.06148547	1.07×10^-6
Module 7	0.0605967	1.01×10^-18
			Module 2	0.06023881	1.31×10^-22
Module 12	0.06015081	1.00×10^-14
			Module 9	0.06011182	2.96×10^-17
Module 14	0.05942734	7.16×10^-25
			Module 3	0.05891626	5.58×10^-32
Module 13	0.05820099	2.46×10^-30

The expression score distribution of the 16 PPI modules is shown in fig. 4a, where a cluster map shows the relationship between clusters and clinical pathology stage. Next, the greedy search procedure and random sampling found 13 modules to be able to significantly distinguish clinical classifications (P < 0.001). The cluster map of expression scores for the 16 modules divides the modules into two broad categories (clusters 1, 2). Cluster 1 contains 3 modules (including module 1, module 4 and module 6) genes up-regulated at tumor stage I and II, and cluster 2 contains 13 modules up-regulated at tumor stage iii+iv (p=0.033). Furthermore, clinical pathology classification of HCC, such as T (p=0.043) or grade stage (p=0.046), is associated with these 16 modules.

The present invention then contemplates the use of 13 PPI modules to build an optimal model for predicting patient clinical grade risk. By using a Random Forest (RF) algorithm, the present invention finds that the combination of the three modules (mod_5, mod_8, mod_15) achieves the optimal prediction accuracy by the RF algorithm (fig. 4 b). 3 modules were up-regulated in HIGH STAGE groups of tumors (fig. 4 c). A predictive model, called the 3-PPI-Mod signature, was then constructed using these 3 PPI modules. The risk of clinical profile in patients was predicted by 3-PPI-Mod profile with an Area Under ROC (AUROC) of 0.7368 (FIG. 4 d). Patients were classified into a high clinical grade risk group and a low clinical grade risk group with the median of the predicted risk factors as a threshold. Overall, 79.74% (126/158) patients were correctly classified when compared to the true clinical grade. The Kaplan-Meier curve demonstrates that the 3-PPI-Mod profile predicts overall survival in patients (p=0.0057, fig. 4 e). Multivariate cox regression showed that the 3-PPI-Mod signature was an independent prognostic factor for OS (corrected risk ratio [ HR ] =2.7, 95% ci,1-7.2, p=0.045, table 3).

TABLE 3 multivariable Cox regression of OS in training dataset

In the 3-PPI-Mod marker, 8, 7 and 3 genes are contained in module 5 (Mod_5), module 8 (Mod_8) and module 15 (Mod_15), respectively. Mod_5 is connected to Mod_8 and Mod_15 in the PPI network (FIG. 5 a). GO and route enrichment analysis showed that mod_5 is involved in developmental differentiation of epidermal cells, mod_8 is significantly involved in tumor-related pathways, and genes in mod_15 are widely involved in the composition and activity of cell membrane components (fig. 5 b).

Example 2: verification of 3-PPI-Mod features in independent queues

The statistical method described in example 1 was used to verify the 3-PPI-Mod prognosis using two independent sets of verification data. The prediction of patient relapse risk was performed on each validation dataset using the 3-PPI-Mod features established by the training dataset. The Kaplan-Meier curve shows that the OS time classified as high risk patients is significantly shorter than that classified as low risk patients (log-rank test: GSE76427, p=0.0039; icgc-LIHC-JP, P < 0.0001) (fig. 6a, fig. 6 b).

Example 3: comparison of prognostic value of exosome-related Signal with simple Gene expression Signal

The exosome-related 3-PPI-Mod markers were compared to 18-gene markers (constructed based on gene expression alone) using the statistical method described in example 1 (fig. 7). The 18 gene signature was established using the same algorithm as the 3-PPI-Mod signature. Whether in the training dataset or in the validation set, the 18 gene signature showed similar predictive performance as the 3-PPI-Mod signature in distinguishing clinical grade from OS (fig. 6, fig. 7). However, the prognostic significance of the 18-gene signature was worse than that of the 3-PPI-Mod signature to a significant extent compared to the two, indicating that the exosome-related signature showed better prognostic value than the signature of the pure gene expression.

Example 4:3-PPI-Mod reflects tumor interstitial interactions and hypoxic tumor microenvironment

1. Tumor microenvironment identification

Hypoxia metabolites of different cancer types were obtained from previous studies with core angiogenesis markers of (Buffa F.M.Harris A.L.West C.M.Miller C.J.Large meta-analysis of multiple cancers reveals a common,compact and highly prognostic hypoxia metagene.Br.J.Cancer.2010;102:428-435). primary tumors (Masiero M).F.C.Han H.D.Snell C.Peterkin T.Bridges E.Mangala L.S.Wu S.Y.Pradeep S.Li D.et al.A core human primary tumor angiogenesis signature identifies the endothelial orphan receptor ELTD1 as a key regulator of angiogenesis.Cancer Cell.2013;24:229-241). Inflammatory cytokinins were used to estimate intratumoral inflammation levels (Saloura V.Zuo Z.Koeppen H.Keck M.K.Khattri A.Boe M.Hegde P.S.Xiao Y.Nakamura Y.Vokes E.E.et al.Correlation of T-cell inflamed phenotype with mesenchymal subtype,expression of PD-L1,and other immune checkpoints in head and neck cancer.J.Clin.Oncol.2014;32(6009–6009)). tumor cells hypoxia, angiogenesis and inflammation scores were calculated by averaging the Z-normalized expression values of the corresponding marker genes. The present invention uses cibert to calculate the abundance of immune and non-immune cells in a tumor microenvironment through a gene expression profile for a tissue infiltrating cell population.

2. Results and analysis

Non-cancerous cells within a tumor play an important role in the construction of the tumor microenvironment, particularly the infiltration of immune cells. It is therefore hypothesized that exosome-specific 3-PPI-Mod may be associated with the tumor microenvironment. For this, identification of the proportion of cell subtypes associated with immunization was first performed in the TCGA dataset using CIRBERSORT. Interestingly, the proportion of partial gene expression in 3-PPI-Mod was in a significant positive correlation with the immune cell, but 3-PPI-Mod was not itself significantly correlated.

Further, the present invention analyzes the relationship between the 3-PPI-Mod signal and the tumor microenvironment status. The results showed that all 3 mods were positively correlated with hypoxia, and that all of mod_5 was correlated with angiogenesis and inflammation relative to mod_8 and mod_15 (fig. 8 a). Hypoxia pathway-related genes were significantly enriched in 3-PPI-Mod predicted high-risk tumors (fig. 8 b). Intratumoral hypoxia scores were significantly correlated with hypoxia inducible factor 1 (HIF 1) expression levels, with HIF1 expression levels upregulated in the high risk subgroups identified by the 3-PPI-Mod markers (fig. 8c, 8d, 8 e).

The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims

1. The application of the gene combination as a marker in preparing a liver cancer prognosis product;

the gene combination consists of a gene combination 1, a gene combination 2 and a gene combination 3;

the gene combination 1 is genes KRT1, HOXC13, KRT34, KRT2, KRT31, HOXC12, HOXC11 and HOXC5;

the gene combination 2 is genes PAX2, GBX1, LHX1, WNT7A, SOST, SP7 and FGF23;

the gene combination 3 is genes SPAG6, WDR63 and DNAH9.

2. The use according to claim 1, wherein the gene combination 1 is associated with developmental differentiation of epidermal cells; the gene combination 2 is related to a tumor-related pathway; the gene combination 3 is involved in the composition and/or activity of cell membrane components.

3. The use of claim 2, wherein said gene combination 1, said gene combination 2 and said gene combination 3 are predictive of survival in liver cancer patients.

4. The use of claim 3, wherein said gene combination 1, said gene combination 2 and said gene combination 3 are associated with a tumor microenvironment.

5. The use of claim 4, wherein said gene combination 1, said gene combination 2 and said gene combination 3 are positively correlated with tumor hypoxia.

6. The use of claim 5, wherein the score for tumor hypoxia is positively correlated with the level of hypoxia inducible factor 1 expression.

7. A primer combination for amplifying the gene combination 1, the gene combination 2 and the gene combination 3 according to claim 1.

8. A reagent, kit, system or device for prognosis of liver cancer comprising the amplification primers of gene combination 1, gene combination 2 and gene combination 3 of claim 1, and acceptable adjuvants, modules or components.