Combining bioinformatics techniques to explore the molecular mechanisms involved in pancreatic cancer metastasis and prognosis

Abstract This article aims to explore the underlying molecular mechanisms and prognosis‐related genes in pancreatic cancer metastasis. Pancreatic cancer metastasis‐related gene chip data were downloaded from GENE EXPRESSION OMNIBUS(GEO)database. Differentially expressed genes were screened after R‐package pre‐treatment. Functional annotations and related signalling pathways were analysed using DAVID software. GEPIA (Gene Expression Profiling Interactive Analysis) was used to perform prognostic analysis, and differential genes associated with prognosis were screened and validated using data from GEO. We screened 40 healthy patients, 40 primary pancreatic cancer and 40 metastatic pancreatic cancer patients, collected serum, designed primers and used qPCR to test the expression of prognosis‐related genes in each group. 109 differentially expressed genes related with pancreatic cancer metastasis were screened, of which 49 were up‐regulated and 60 were down‐regulated. Functional annotation and pathway analysis revealed differentially expressed genes were mainly concentrated in protein activation cascade, extracellular matrix construction, decomposition, etc In the biological process, it is mainly involved in signalling pathways such as PPAR, PI3K‐Akt and ECM receptor interaction. Prognostic analysis showed the expression levels of four genes were significantly correlated with the overall survival time of patients with pancreatic cancer, namely SCG5, CRYBA2, CPE and CHGB. qPCR experiments showed the expression of these four genes was decreased in both the primary pancreatic cancer group and the metastatic pancreatic cancer group, and the latter was more significantly reduced. Pancreatic cancer metastasis is closely related to the activation of PPAR pathway, PI3K‐Akt pathway and ECM receptor interaction. SCG5, CRYBA2, CPE and CHGB genes are associated with the prognosis of pancreatic cancer, and their low expression suggests a poor prognosis.

PI3K-Akt pathway and ECM receptor interaction. SCG5, CRYBA2, CPE and CHGB genes are associated with the prognosis of pancreatic cancer, and their low expression suggests a poor prognosis.

| INTRODUC TI ON
Pancreatic cancer is a highly invasive malignant tumour of the digestive system with a 5-year survival rate of <8%. 1 Because of the earlier invasion and metastasis, the quality of life and treatment of patients with pancreatic cancer are often unsatisfactory. Therefore, in-depth study of the specific molecular mechanism of pancreatic cancer metastasis is of vital importance for the treatment and prognosis of pancreatic cancer. Gene chip, also known as DNA chip or DNA microarray, with high flux, high integration, miniaturization, automation, etc can detect the expression levels of thousands of gene transcripts in parallel and quickly and has been widely used in mutant gene testing, differential gene screening, gene library mapping, drug targets, tumour typing, polymorphism detection, etc. 2,3 In this study, we analysed the differentially expressed genes of pancreatic cancer metastasis by analysing the data of pancreatic cancer metastasis-related gene chip in the public database of gene chip (GEO) and analysed GO enrichment analysis, KEGG pathway analysis, protein interaction network and prognostic value analysis. And we designed experiments to verify the conclusions, providing ideas for further exploration of the molecular mechanisms of pancreatic cancer metastasis.

| Materials
The three pancreatic cancer metastasis-related gene chip data used in this study (GSE19279, 4 GSE42952, 5 GSE71729 6 ) were derived from the National Bioinformatics Center (NCBI) Gene Chip Public Database (GEO). 7 Among them, gse19279 was included in 9 cases, including 4 cases of primary pancreatic cancer and 5 cases of metastatic pancreatic cancer; gse42952 included 23 cases, including 12 cases of primary pancreatic cancer and 11 cases of metastatic pancreatic cancer; gse71729 included 206 cases, including 145 Primary pancreatic cancer and 61 metastatic pancreatic cancer.

| Method
The above data were pre-treated and screened for differentially expressed genes by Limma method. 8,9 The screening threshold was set to P < 0.05, and the fold change was >2 fold. The differentially expressed genes screened by the above three data sets were taken to reduce the false positive rate; thus, the gene set obtained was used as a differentially expressed gene for pancreatic cancer metastasis for subsequent analysis. First, the DAVID online analysis software was used to perform gene function annotation and pathway analysis on differentially expressed genes of pancreatic cancer metastasis 10,11 to clarify its biological function and the involved cell signal regulation network; then use the STRING online database and Cytoscape software to perform protein interaction network analysis of pancreas Cancer metastasis differentially expressed genes, 12,13 key genes and key modules were calculated, and their possible regulation and mechanism of action were further explored. Finally, the prognostic analysis of differentially expressed genes in pancreatic cancer metastasis was performed by GEPIA online analysis tool. 14 Prognostic analysis of the selected prognosis-related genes was performed with data from patients with pancreatic cancer in the GEO database and KM survival curves were drawn. Further, use GEPIA online analysis tool to analyse the relationship between prognosisrelated genes and clinical stage of pancreatic cancer, so as to provide some clues for clinical prognosis.

| Experimental verification of the selected prognosis-related gene
The study was approved by the medical ethics committee of the First Affiliated Hospital of Nanchang University. All patients involved in the study signed the informed consent voluntarily. All methods were performed in accordance with the relevant guidelines and regulations.  SPSS 19.0 software was used for statistical analysis. All the quantitative data acquired from the experiment fitted normal distribution and could be expressed as (x ± s). The comparison between groups was conducted by single factor difference analysis and HSD-q test. The difference is statistically significant when P < 0.05.

| Screening of differentially expressed genes in pancreatic cancer metastasis
Through the screening of differentially expressed genes, 1070 differentially expressed genes in the gse19279 data set were obtained, of which 492 were up-regulated and 578 were downregulated. There were 909 differentially expressed genes in the gse42952 data set, of which 350 were up-regulated and 559 were down-regulated. There were 455 differentially expressed genes in the gse71729 data set, of which 171 were up-regulated and 284 were down-regulated. The three data sets were taken to obtain 109 differentially expressed genes, of which 49 were up-regulated and 60 were down-regulated ( Figure 1).   (Table 3). Up-regulation of differentially expressed genes is involved in the PPAR signalling pathway (Figure 2).  Table 4).The KEGG pathway analysis found that it mainly focuses on pancreatic secretion, protein digestion and absorption, ECM receptor interaction, local adhesion and PI3K-Akt signalling pathways (Table 5). Down-regulation of differentially expressed genes is involved in the PI3K-Akt signalling pathway (Figure 3).

| Interaction of differentially expressed genes in pancreatic cancer metastasis
The protein interaction network of 109 pancreatic cancer metasta-

| Experimental verification of differential genes
Compared with the control group, the expression levels of the four genes in the primary pancreatic cancer group and the metastatic pancreatic cancer group were significantly decreased. The relative expression levels of the four genes in metastatic pancreatic cancer were lower than those in primary pancreatic cancer, and the difference was significant (P < 0.05) ( Figure 6B).

| D ISCUSS I ON
Metastasis is an important risk factor for poor prognosis in patients with pancreatic cancer. Because patients with metastases often lose their chance of surgery, they can only rely on palliative care. 15 At present, the specific molecular mechanism of pancreatic cancer metastasis has not been fully elucidated, which involves the abnormal expression of many genes and the imbalance of related signalling pathways. Therefore, studying pancreatic cancer metastasis-related F I G U R E 6 Relative expression of four prognostic-related genes in the control group, primary pancreatic cancer group and metastatic pancreatic cancer group(Using the expression value of GAPDH as "1", calculate the relative expression levels of the four genes) Based on the STRING online database, we performed a protein interaction network analysis on differentially expressed genes in pancreatic cancer metastasis. It found that the protein products encoded by the above genes are involved in the construction of protein interaction network maps, which are connected by wires and some genes are used as centres and are closely related to the surrounding gene products. In this study, 10

| CON CLUS ION
In summary, this study used an integrated bioinformatics method to comprehensively analyse the data of pancreatic cancer metastasisrelated gene chips in the GEO database and successfully screened 109 differentially expressed genes, which were analysed by GO enrichment analysis and KEGG pathway analysis and obtained a series of related biological processes and signalling pathways, providing a theoretical basis for laboratory research on pancreatic cancer metastasis. In addition, by constructing a protein interaction network map, two key modules and 10 key genes were obtained, and the study of IGFBP1 suggested its important position in tumour metastasis. In addition, by prognostic analysis of the above differentially expressed genes, four genes with significant prognostic significance were obtained, which is important for evaluating and predicting the prognosis of patients with pancreatic cancer.

ACK N OWLED G EM ENTS
Jiasheng Xu and Kaili Liao contributed equally to this article, they are the co-first authors. This study was supported by the National Natural Science Foundation of China (grant number:81860034)

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest. Jiarui He: revised the manuscript (supporting).

E TH I C A L A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
The study was approved by the medical ethics committee of the First Affiliated Hospital of Nanchang University. All patients involved in the study signed the informed consent voluntarily. All methods were performed in accordance with the relevant guidelines and regulations.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing not applicable to this article as no data sets were generated or analysed during the current study.