Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Alimentary tract cancers (ATCs) are the most malignant cancers in the world. Numerous studies have revealed the tumorigenesis, diagnosis and treatment of ATCs, but many mechanisms remain to be explored.

METHODS:

To identify the key genes of ATCs, microarray datasets of oesophageal cancer, gastric cancer and colorectal cancer were obtained from the Gene Expression Omnibus (GEO) database. In total, 207 differentially expressed genes (DEGs) were screened. KEGG and GO function enrichment analyses were conducted, and a protein-protein interaction (PPI) network was generated and gene modules analysis was performed using STRING and Cytoscape.

RESULTS:

Five hub genes were screened, and the associated biological processes indicated that these genes were mainly enriched in cellular processes, protein binding and metabolic processes. Clinical survival analysis showed that COL10A1 and KIF14 may be significantly associated with the tumorigenesis or pathology grade of ATCs. In addition, relative human ATC cell lines along with blood samples and tumour tissues of ATC patients were obtained. The data proved that high expression of COL10A1 and KIF14 was associated with tumorigenesis and could be detected in blood.

CONCLUSION:

In conclusion, the identification of hub genes in the present study helped us to elucidate the molecular mechanisms of tumorigenesis and identify potential diagnostic indicators and targeted treatment for ATCs.

Keywords

Alimentary tract cancers differentially expressed genes microarray datasets enrichment analysis survival analysis

1. Introduction

Alimentary tract cancers (ATCs) encompass a group of gastrointestinal cancers (GICs), including oesopha-geal cancer (ESC), gastric cancer (GC) and colorectal cancer (CRC), which have high malignancy and poor prognoses worldwide. According to Cancer Statistics, 2019 [1], among three ATCs, CRC is in the top three in terms of both new cases and deaths. Additionally, data from The Global Burden of Cancer 2013 [2] show that oesophageal malignancy ranks sixth among cancer-related deaths worldwide, while GC ranks as the fifth most common cancer and the third leading cause of cancer-related deaths worldwide based on GLOBOCAN data [3].

The complexity and heterogeneity of ATCs emerge from multiple interactions of genetic, environmental, immune and host factors. Smoking, alcohol consumption and obesity are common cancer-related factors [4]. Moreover, previous studies have demonstrated that aberrant expression of genes is involved in the tumorigenesis of ATCs [5]. Several genes and pathways, specifically HER2, E-cadherin, FGFR (fibroblast growth factor receptor)/EGFR (human epidermal growth factor receptor), mTOR (mammalian target of rapamycin), PD-L1, and MMPs (matrix metalloproteinases), and a large number of noncoding RNAs have been identified as biomarkers of ESC, GC and CRC [5, 6, 7]. However, due to the ambiguous symptoms and the lack of specialized diagnostic indicators at the early stage of the disease, the mortality rate of ATCs is still rising. Thus, it is essential to understand the precise mechanisms related to the carcinogenesis, progression and recurrence of ATCs and explore effective predictive biomarkers for diagnosis and therapy.

Due to the success of the Human Genome Project (HGP), technology for high-throughput gene sequencing has been rapidly developed and has been extensively utilized to screen differentially expressed genes (DEGs) at the genome level. In the present study, we collected several of the latest microarray datasets from an online database and utilized bioinformatic analysis to identify the DEGs and biological processes and pathways related to the tumorigenesis, invasion and metastasis of ATCs. However, statistical deviation in a single microarray series obviously contributes to unreliable results. Thus, 6 mRNA microarray datasets (2 each for ESC, GC and CRC) from GEO were obtained to screen DEGs between tumour tissues and paracancerous tissues. Subsequently, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) enrichment analyses were performed. Protein-protein interaction (PPI) network analyses were performed to explore further mechanisms. By determining the correlations between patient characteristics, prognosis, and pathological grade and the expression of key genes with Gene Expression Profiling Interactive Analysis (GEPIA), University of California Santa Cruz (UCSC) Genome Browser and cBioPortal, we identified 2 hub genes as specialized biomarkers for ATCs.

2. Materials and methods

2.1 Microarray acquirement

Microarray datasets of ESC (GSE75241 [8]) and GSE100942 [9]), GC (GSE118916 [10] and GSE26 942 [11]) and CRC (GSE113513 and GSE128449) were downloaded from GEO (http://www.ncbi.nlm.nih.gov/ geo) [12], which is a genomics database of highly differential gene expression and microarray data. By referring to the annotation information, the probe IDs were converted into the corresponding gene symbols. Basic information on the microarray datasets is displayed in Table 1.

Table 1
Basic information of microarray datasets of ATCs

Term	Platform	Species	Sample types		Sample numbers	Public time
GSE75241	GPL5175	Homo sapiens	Esophageal tissues	Tumor	15	Jun 26, 2019
				Normal	15
GSE100942	GPL570	Homo sapiens	Esophageal tissues	Tumor	4	Jul 15, 2017
				Normal	4
GSE118916	GPL15207	Homo sapiens	Gastric tissues	Tumor	15	Jun 10, 2019
				Normal	15
GSE26942	GPL6947	Homo sapiens	Gastric tissues	Tumor	97	Dec 31, 2016
				Normal	12
GSE113513	GPL15207	Homo sapiens	Colorectal tissues	Tumor	14	Apr 24, 2018
				Normal	14
GSE128449	GPL4133	Homo sapiens	Colorectal tissues	Tumor	31	Mar 19, 2019
				Normal	5

2.2 Identification of DEGs

The DEGs between tumour and normal tissues were identified using GEO2R (http://www.ncbi.nlm.nih.gov/ geo/geo2r), which is a web tool able to identify DEGs by comparing microarray datasets in a GEO series. Specific P-value and false discovery rate (FDR) cut-offs were applied to evaluate statistical significance and false positives. In the preanalysis, probes without coincident gene symbols or genes with redundant probe sets were eliminated or processed, respectively. $|$ logFC (fold change) $|$ $>$ 1 and $P$ -value $<$ 0.01 were considered to be statistically significant.

2.3 KEGG and GO enrichment analyses

DAVID (Database for Annotation, Visualization and Integrated Discovery; http://david.ncifcrf.gov; version 6.8) [13] is an online database providing free bioinformatics data and analysis tools and an extensive set of functional annotation information for DEGs for extraction of biological information. KEGG is a well-known resource for illustrating gene functions and biological pathways based on high-throughput analysis datasets [14]. GO is a general bioinformatic annotation tool for gene functions and biological analysis [15]. KEGG and GO enrichment analyses in the present study were performed with DAVID. $P<$ 0.01 was considered to be statistically significant.

2.4 PPI network construction

The PPI network was constructed using STRING (http://string-db.org version 11.0), which is an online database analysing functional protein association networks and providing insights into the progression or development of diseases [16]. In this work, a PPI network of DEGs was constructed using the STRING webtool, and an interaction with a combined score $>$ 0.4 was considered statistically significant.

2.5 Module analysis

Cytoscape (version 3.6.0) is an open-sourcing platform of bioinformatics software that provides visual interactive molecular networks [17]. Molecular Complex Detection (MCODE, version 1.5.1), a plugin of Cytoscape, is a specialized tool for clustering primary networks based on topology to find close linked regions and module genes [18]. The related networks were plotted by Cytoscape based on the STRING database, and the most significant module genes were identified by MCODE (screening principles: MCODE score $>$ 5, node score cut-off $=$ 0.2, k-score $=$ 2, degree cut-off $=$ 2 and max depth $=$ 100). Subsequently, the KEGG and GO enrichment analyses of the module genes were performed by DAVID.

2.6 Identification and analysis of hub genes

The hub genes were extracted from the networks plotted by the MCODE plugin of Cytoscape (degree $\geqslant$ 10). The interactive biological processes of hub genes were visualized in maps with the Biological Networks Gene Oncology tool (BiNGO, version 3.0.3) plugin of Cytoscape. A network of the genes and their coexpressed genes was analysed via the cBioPortal (http://www.cbioportal.org) [19] online platform. A ranked collection of the hub genes was constructed using the UCSC Cancer Genomics Browser (http://genome-cancer.ucsc.edu) [20]. The expression profiles of COL10A1 and KIF14 as well as their relationship with overall survival (OS) and disease-free survival (DFS) were analysed and displayed using the online database Gene Expression Profiling Interactive Analysis (GEPIA; http://gepia.cancer-pku.cn). The relationship between expression patterns and tumour grade was analysed with several tools using the online database Oncomine (http://www.oncomine.com) [21]. Data on COL10A1 expression were obtained from the following sources: ESCA: 1. Hu Esophagus, BMC Genomics, 2010; 2. Su Esophagus 2, Clin Cancer Res, 2011; STAD: 1. Cho Gastric, Clin Cancer Res, 2011; 2. Cui Gastric, Nucleic Acids Res, 2011; 3. DErrico Gastric, Eur J Cancer, 2009; and COAD: 1. Kaiser Colon, Genome Biol, 2007; 2. Ki Colon, Int J Cancer, 2007; 3. Skrzypczak Colorectal, PLoS One, 2010. Data on KIF14 expression were obtained from the following sources: ESCA: 1. Hu Esophagus, BMC Genomics, 2010; 2. Su Esophagus 2, Clin Cancer Res, 2011; 3. Yu Multi-cancer, PLoS Genet, 2008; STAD: 1. Cho Gastric, Clin Cancer Res, 2011; 2. Cui Gastric, Nucleic Acids Res, 2011; 3. Ooi Gastric, PLoS Genet, 2009; and COAD: 1. Gaspar Colon, Am J Pathol, 2008; 2. Hong Colorectal, Clin Exp Metastasis, 2010; 3. Skrzypczak Colorectal, PLoS One, 2010.

Figure 1.

Venn diagram, PPI network and the most significant module of DEGs. (A) DEGs are selected with a fold change $>$ 2 and $P$ -value $<$ 0.01 among the mRNA expression profiling sets of ESC (GSE75241 and GSE100942), GC (GSE118916 and GSE26942) and CRC (GSE113513 and GSE128449). An overlap of 207 genes are generated from the 6 datasets. (B) The PPI network of DEGs is constructed using STRING and Cytoscape. (C) The most significant modules are obtained from PPI network with 5 nodes and 272 edges.

2.7 Cell lines and clinical specimens

Human ATC cell lines (oesophageal cancer: EC109; gastric cancer: SGC7901; and colorectal cancer: SW480) were purchased from the Cell Resource Center of Shanghai Institute of Biochemistry and Cell Biology (Shanghai, China). Cells were cultured in RPMI-1640 supplemented with 10% foetal bovine serum (both from Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Thirty patients diagnosed with early-stage ATCs (10 each for ESC, GC and CRC) were selected for this study and underwent operations in the affiliated Changzhou No. 2 People’s Hospital of Nanjing Medical University and Haimen People’s Hospital from March 2018 to May 2019. Blood samples were obtained from patients on the first day to avoid the interference of drug treatment. Plasma of all blood samples was extracted within 20 minutes and stored at $-$ 80 ${}^{\circ}$ C. In addition, tumour tissues and adjacent tissues were collected during the surgery and stored at $-$ 80 ${}^{\circ}$ C immediately.

2.8 RT-qPCR measurement

Cell lines, blood and tissues were preprocessed to extract total RNA. cDNA was synthesized using a Prime Script RT reagent kit (TaKaRa, Dalian, China). Quantitative PCR was carried out with a 7500 real-time PCR system (ABI, Waltham, MA, USA). PCR primers were synthesized by and purchased from Sangon Biotech (Shanghai, China). COL10A: forward: TTCTGGGATGCCGCTTGTC, reverse: TCGTAGGCGTGCCGTTCTT; KIF14: forward: TTCAGAACACCTCTGCAGGA, reverse: ACTCATGAAGACTACCTGGG. Actin served as an internal control, and fold changes were calculated with the 2 ${}^{-\Delta\Delta Cq}$ method.

Figure 2.

Interaction network and biological process analysis of the hub genes. (A) The biological process analysis of hub genes is constructed using BiNGO. The nodes in yellow refer to the main biological processes. (B) Hub genes and their co-expression genes are formed using cBioPortal. Nodes with bold black outline represent hub genes. Nodes with thin black outline represent the co-expression genes. 5 hub genes are sorted out and listed. The colour depth of nodes refers to the corrected P-value of ontologies. The size of nodes refers to the numbers of genes that are involved in the ontologies. (C) (D) 5 hub genes are enriched using KEEG and GO function analysis. Numbers of each enriched terms are displayed with length of bars.

Figure 3.

Expression of hub genes in TCGA and meta-analysis. (A) (B) (C) Expression of hub genes are obtained from The Cancer Genome Atlas (TCGA) and shown in boxplots. Overall gene expressions of COL10A1 and KIF14 in human body are displayed in body maps (D) (F) and charts (E) (G). (H) (J) Overall COL10A1 and KIF14 gene expression from projects are shown in boxplots. (I)(K) Meta-analysis with projects of COL10A1 and KIF14 gene expression are made. High-expression in tumour tissues compared with normal tissues are displayed in boxes in red. On the contrary, boxes are marked in blue. $P<$ 0.05 is considered statistically significant and labelled as ‘*’.

Figure 4.

Association of prognostic and pathological grades with two genes and experimental verification. Overall survival and disease-free survival analyses of COL10A1 (A) and KIF14 (B) are performed using cBioPortal online platform. (C) (D) Association between the expression of COL10A1 and KIF14 and tumor grade is displayed. mRNA levels of COL10A1 (E) and KIF14 (F) in cell lines, blood samples and tumor tissues are detected and demonstrated in scatter diagrams. P<0.05 is considered statistically significant.

Table 2

Full names and functional roles of 5 hub genes with degree $\geqslant$ 10

No.	Gene symbol	Full names	Functions
1	NR3C2	Nuclear receptor subfamily 3 group C member 2	NR3C2 functions as a ligand-dependent transcription factor that binds to mineralocorticoid response elements in order to transactivate target genes.
2	SLIT2	Slit guidance ligand 2	Both full length and cleaved proteins are secreted extracellularly and can function in axon repulsion as well as other specific processes. Alternative splicing results in multiple transcript variants.
3	KIF14	Kinesin family member 14	KIF14 is localizes to the central spindle during late phase mitosis and its inhibition in tumor cells results in cell cycle arrest and the formation of binucleated cells. KIF14 was also found to regulate adhesive components on the tumor cell surface influencing migratory and invasive properties that promote cell motility during metastasis.
4	COL10A1	Collagen type X alpha 1 chain	Type X collagen is a product of hypertrophic chondrocytes and has been localized to presumptive mineralization zones of hyaline cartilage.
5	GAL	Galanin and GMAP prepropeptide	This gene encodes a neuroendocrine peptide that is widely expressed in the central and peripheral nervous systems and also the gastrointestinal tract, pancreas, adrenal gland and urogenital tract.

2.9 Statistical analysis

In the present study, statistical analyses were conducted using one-way analysis of variance. Data are expressed and displayed as the mean $\pm$ standard error. GraphPad Prism software (version 7, GraphPad Software, Inc., La Jolla, CA, USA) was used to calculate and analyse the research data.

3. Results

3.1 Identification of DEGs in ATCs

After standardization and processing, DEGs (2,391 in ESC datasets, 2,070 in GC datasets and 3,875 in CRC datasets) were selected from microarrays. There were 207 overlapping genes among the 3 cancers, as shown in the Venn diagram (Fig. 1A).

3.2 PPI network and module analyses

To illustrate the mutual effects of overlapping genes, a PPI network of DEGs was constructed (Fig. 1B). Cytoscape was used to plot the PPI network, and the most essential module was obtained from the MCODE plugin with 5 nodes and 272 edges (Fig. 1C). The biological process interaction analyses of the hub genes were performed using the cBioPortal online database, and the results are shown in Fig. 2A. Then, networks of the hub genes and coexpressed genes were analysed, and the results are shown in Fig. 2B. In addition, 5 hub genes (NR3C2, SLT2, KIF14, COL10A1, and GAL) were identified. The functional and pathway enrichment analyses of hub genes was performed with DAVID to analyse biological classification. The compiled analysis of overlapping genes is shown. The KEGG analysis results showed that metabolic pathways and the PI3K-Akt pathway were significantly enriched (Fig. 2C). Moreover, GO analysis results showed that the DEGs were significantly enriched in cellular processes related to protein binding, metabolic processes and others (Fig. 2D).

3.3 Hub gene analysis and biological enrichment analyses of DEGs

A total of 5 genes were selected as hub genes with MCODE (degrees $\geqslant$ 10), which corresponded with the results in Fig. 2B. The abbreviations, full names and functions for these hub genes are shown in Table 2. Subsequently, the expression data of 5 hub genes in the TCGA database was extracted and displayed, in which COL10A1 and KIF14 were significantly highly expressed in cancer compared with normal tissues among the ESC (Fig. 3A), GC (Fig. 3B) and CRC (Fig. 3C) datasets. Thus, according to the differential expression, COL10A1 and KIF14 were selected as key factors in the subsequent research. For further analysis, the expression profiles of COL10A1 and KIF14 in human tissue and their associations with patients clinical characteristics were determined using the GEPIA webtool. The human body maps illustrated that COL10A1 (Fig. 3D) and KIF14 (Fig. 3F) were highly expressed in ESC, GC and CRC tissues. The same quantitative analysis is shown in Fig. 3E–G. In addition, Oncomine-based analysis of cancer vs. normal tissue indicated that COL10A1 (Fig. 3H and I) and KIF14 (Fig. 3J and K) were significantly overexpressed in all three cancer tissues in the different datasets. Moreover, OS and DFS analyses of the hub genes in ESC, GC and CRC were performed using Kaplan-Meier curves via GEPIA (Fig. 4A and B). Related pathology grade analyses were also performed, in which both COL10A1 and KIF14 showed linear increases with pathology grade (Fig. 4C and D). The results above indicate that high expression of COL10A1 and KIF14 is theoretically related to poor prognosis and advanced pathology stage. To confirm the bioinformatics results and verify the expression of key factors, mRNA detection was performed in our laboratory. Ten pairs of tumour and adjacent tissues were collected from ESC, GC and CRC patients in our hospital from 2018 to 2019. Figure 4E and F show that the overall mRNA expression of COL10A1 and KIF14 was upregulated in cell lines, patient blood and tumour tissues to a certain degree.

4. Conclusion

The digestive tract (DT) consists of the oesophagus, stomach and colon and is unique in its direct contact with food. DT is one of most essential organs in the digestive system and is associated with diverse diseases. ATCs, including oesophageal cancer, gastric cancer, and colorectal cancer, are a series of malignant diseases with high incidence and morbidity rates. Among ATCs, oesophageal cancer ranks eighth in prevalence worldwide, and colorectal cancer ranks as the third most prevalent cancer worldwide. In addition, the incidence of gastric cancer is much higher in Asia than in other regions. The prognosis of ATCs is favourable only in the very early stages. Squamous cell carcinoma antigen (SCC Ag), carcinoembryonic antigen (CEA), cancer antigen 199 (CA199), cancer antigen 72-4 (CA72-4), and cancer antigen 50 (CA50) are generally used to predict ATCs in clinical practice [22, 23, 24]. Meanwhile, endoscopic examination is the gold standard for identifying tumours. Due to their low specificity and precision, they are not reliable indictors for early diagnosis. Generally, ATCs in early stages are operable by surgery, while patients in the middle stages are recommended to receive surgical intervention by neoadjuvant chemotherapy. For more advanced stages, chemotherapy or chemoradiotherapy is selected. In terms of chemotherapy, 5-FU, platinum and paclitaxel are usually used for broad ATC treatment [25]. However, the high incidences of side effects and chemoresistance greatly reduce their value. Although molecular targeted drugs have been approved to treat patients with ATCs in advanced stages in recent decades, more effective diagnostic indictors and therapeutic targets are needed. Prominent advances in microarray technology and bioinformatic analysis have conferred further advantages.

In the present study, 6 microarray datasets were downloaded to identify DEGs between cancer tissues and noncancerous tissues of ATCs, among which a total of 207 DEGs were identified. To explore the interactions of DEGs, GO and KEGG enrichment analyses were performed. The DEGs were mainly enriched in cellular processes, protein binding, and metabolic processes, and metabolic pathways and the PI3K-Akt pathway were significantly enriched. Previous studies have reported that dysregulation of the cell cycle and DNA damage plays an essential role in the carcinogenesis and progression of ATCs [26, 27, 28, 29, 30]. In addition, recent studies have proposed a tumour-promoting role for autophagy activation [31]. Moreover, diabetic inflammatory activity often plays a major role and is frequently altered in tumours [32, 33]. In sum, all these theories are consistent with our results (Supplementary Fig. 1).

Next, we selected 5 DEGs (COL10A1, KIF14, NR3C2, GAL, and SLIT2) as hub genes with degrees $\geqslant$ 10. Of these hub genes, COL10A and KIF14 revealed the highest node degrees [29]. COL10A1, a type X collagen, is a product of hypertrophic chondrocytes and has been shown to localize to presumptive mineralization zones of hyaline cartilage. In previous studies, COL10A1 has been discussed in orthopaedic disorders. However, in recent years, an increasing number of reports have shown that COL10A1 is associated with poor prognosis in colorectal cancer and promotes invasion and metastasis in gastric cancer [34, 35, 36, 37, 38, 39]. Meanwhile, KIF14 is prevalent in the central spindle during late phase mitosis, and its inhibition in tumour cells is related to cell cycle arrest. Moreover, KIF14 was reported to regulate adhesive components on the tumour cell surface, affecting migratory and invasive properties that promote cell motility during metastasis. Although relationships between KIF14 and ATC have rarely been reported, a number of studies have illustrated the tumorigenic effects of KIF14 in cancers such as retinoblastoma, ovarian cancer, prostate cancer, and so on [40, 41, 42, 43, 44, 45, 46]. For clinical assessment, we analysed the association of the expression of COL10A1 and KIF14 with overall and disease-free survival. Gene alteration of COL10A1 was related to reductions in disease-free survival. However, in the present study, those reductions in overall survival were not statistically significant. In addition, COL10A1 and KIF14 levels were shown to be increased with pathology grade. We speculate that the reason for this discrepancy may be that the survival analyses in cBioPortal are based on the relationship between prognosis and gene mutation and do not consider gene overexpression specifically. Thus, overexpression of COL10A1 and KIF14 in ATCs may result from modification rather than mutation, and further research is needed to verify our hypothesis. The meta-analysis from the Oncomine database showed that high mRNA levels of COL10A1 and KIF14 were associated with tumorigenesis [47, 48, 49, 50, 51, 52, 53, 54, 55].

Thirty patients diagnosed with early-stage ATCs (10 each for ESC, GC and CRC) were selected for our prospective study, and blood samples and tumour tissues were obtained from these patients at the affiliated Changzhou No. 2. People’s Hospital of Nanjing Medical University and Haimen People’s Hospital. The mRNA levels of COL10A1 and KIF14 were measured with PCR. The data showed that high expression of both genes was found in cell lines and tumour tissues among ATCs, which proved the relationship between gene expression and tumorigenesis. However, although in blood samples, the expression of COL10A1 and KIF14 was upregulated in GC and CRC samples, it was not significantly changed in ESC samples. Regardless, overall, high expression of COL10A1 and KIF14 can be prospectively considered a potential indicator of early-stage ATC, especially GC and CRC.

In conclusion, the present study identified significant DEGs that may be involved in the carcinogenesis or progression of ATCs, among which 2 hub genes (COL10A1 and KIF14) may be regarded as diagnostic and prognostic biomarkers for ATCs. However, more evidence is needed to elaborate the translational function of these 2 genes in ATC treatment.

Ethics approval and consent to participate

The authors declare that this study is approved by the Ethics Committee of the Affiliated Changzhou No. 2 People’s Hospital of Nanjing Medical University and Haimen People’s Hospital. In addition, the authors declare that all blood samples and tissues involved in this study are permitted by patients. However, no other clinical trials were involved in this study, we didn’t provide the committee’s reference number according to national regulations.

Availability of data and material

The datasets generated and analyzed during the curr- ent study are available in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo). ESC: GSE75241 and GSE100942; GC: GSE118916 and GSE26942; CRC: GSE113513 and GSE128449. The clinical correlation analyses are based on TCGA database (https://www.cancer.gov/about-nci/organizati on/ccg/research/structural-genomics/tcga). The projects involved in meta-analyses are provided in supplementary information files.

Funding

The expenditure of this whole study was funded by research grants from the Jiangsu Natural Science Foundation (grant no. BK20181155), the Nanjing Medical University (grant no. 2017NJMU043), the Changzhou Department of Health (grant no. QN201711) and the Changzhou NO. 2 People’s Hospital (grant no. 2018K003).

Footnotes

Conflict of interest

The authors declare no competing interests.

Abbreviation

Supplementary data

Supplementary Fig. S1.

Prisma flow of data selection and analysis.

References

Ain

N.U.

Makitie

and Naz

, Autosomal recessive chondrodysplasia with severe short stature caused by a biallelic COL10A1 variant, J Med Genet 55 (2018), 403–407.

Ashburner

Ball

C.A.

Blake

J.A.

Botstein

Butler

Cherry

J.M.

Davis

A.P.

Dolinski

Dwight

S.S.

Eppig

J.T.

Harris

M.A.

Hill

D.P.

Issel-Tarver

Kasarskis

Lewis

Matese

J.C.

Richardson

J.E.

Ringwald

Rubin

G.M.

and Sherlock

, Gene ontology: tool for the unification of biology. The Gene Ontology Consortium, Nat Genet 25 (2000), 25–29.

Bandettini

W.P.

Kellman

Mancini

Booker

O.J.

Vasu

Leung

S.W.

Wilson

J.R.

Shanbhag

S.M.

Chen

M.Y.

and Arai

A.E.

, MultiContrast Delayed Enhancement (MCODE) improves detection of subendocardial myocardial infarction by late gadolinium enhancement cardiovascular magnetic resonance: a clinical validation study, J Cardiovasc Magn Reson 14 (2012), 83.

Bijlsma

M.F.

Sadanandam

Tan

and Vermeulen

, Molecular subtypes in cancers of the gastrointestinal tract, Nat Rev Gastroenterol Hepatol 14 (2017), 333–342.

Bilgin

Sendur

M.A.

Bulent Akinci

Sener Dede

and Yalcin

, Targeting the PD-1 pathway: a new hope for gastrointestinal cancers, Curr Med Res Opin 33 (2017), 749–759.

Bloom

G.C.

Eschrich

Zhou

J.X.

Coppola

and Yeatman

T.J.

, Elucidation of a protein signature discriminating six common types of adenocarcinoma, Int J Cancer 120 (2007), 769–775.

Bowen

N.J.

Walker

L.D.

Matyunina

L.V.

Logani

Totten

K.A.

Benigno

B.B.

and McDonald

J.F.

, Gene expression profiling supports the hypothesis that human ovarian surface epithelia are multipotent and capable of serving as ovarian cancer initiating cells, BMC Med Genomics 2 (2009), 71.

Chen

Cheung

S.T.

Fan

S.T.

Barry

Higgins

Lai

K.M.

Dudoit

I.O.

Van De Rijn

Botstein

and Brown

P.O.

, Gene expression patterns in human liver cancers, Mol Biol Cell 13 (2002), 1929–1939.

Cho

J.Y.

Lim

J.Y.

Cheong

J.H.

Park

Y.Y.

Yoon

S.L.

Kim

S.M.

Kim

S.B.

Kim

Hong

S.W.

Park

Y.N.

Noh

S.H.

Park

E.S.

Chu

I.S.

Hong

W.K.

Ajani

J.A.

and Lee

J.S.

, Gene expression signature-based prognostic risk score in gastric cancer, Clin Cancer Res 17 (2011), 1850–1857.

10.

Cui

Chen

Chou

W.C.

Sun

Chen

Suo

Zhang

Kong

Hoffman

L.L.

Kang

Olman

Johnson

Tench

D.W.

Amster

I.J.

Orlando

Puett

and Xu

, An integrated transcriptomic and computational analysis for biomarker identification in gastric cancer, Nucleic Acids Res 39 (2011), 1197–1207.

11.

D’Errico

de Rinaldis

Blasi

M.F.

Viti

Falchetti

Calcagnile

Sera

Saieva

Ottini

Palli

Palombo

Giuliani

and Dogliotti

, Genome-wide expression profile of sporadic gastric cancers with microsatellite instability, Eur J Cancer 45 (2009), 461–469.

12.

Dasyani

Tan

W.H.

Sundaram

Imangali

Centanin

Wittbrodt

and Winkler

, Lineage tracing of col10a1 cells identifies distinct progenitor populations for osteoblasts and joint cells in the regenerating fin of medaka (Oryzias latipes), Dev Biol (2019).

13.

Deng

Bragelmann

Schultze

J.L.

and Perner

, Web-TCGA: an online platform for integrated analysis of molecular cancer data sets, BMC Bioinformatics 17 (2016), 72.

14.

Duffy

M.J.

, Use of Biomarkers in Screening for Cancer, Adv Exp Med Biol 867 (2015), 27–39.

15.

Fitzmaurice

Dicker

Pain

Hamavid

Moradi-Lakeh

MacIntyre

M.F.

Allen

Hansen

Woodbrook

Wolfe

Hamadeh

R.R.

Moore

Werdecker

Gessner

B.D.

Te Ao

McMahon

Karimkhani

Cooke

G.S.

Schwebel

D.C.

Carpenter

D.O.

Pereira

D.M.

Nash

Kazi

D.S.

De Leo

Plass

Ukwaja

K.N.

Thurston

G.D.

Yun Jin

Simard

E.P.

Mills

Park

E.K.

Catala-Lopez

deVeber

Gotay

Khan

Hosgood

H.D.

, 3rd Santos

I.S.

Leasher

J.L.

Singh

Leigh

Jonas

J.B.

Sanabria

Beardsley

Jacobsen

K.H.

Takahashi

Franklin

R.C.

Ronfani

Montico

Naldi

Tonelli

Geleijnse

Petzold

Shrime

M.G.

Younis

Yonemoto

Breitborde

Yip

Pourmalek

Lotufo

P.A.

Esteghamati

Hankey

G.J.

Ali

Lunevicius

Malekzadeh

Dellavalle

Weintraub

Lucas

Hay

Rojas-Rueda

Westerman

Sepanlou

S.G.

Nolte

Patten

Weichenthal

Abera

S.F.

Fereshtehnejad

S.M.

Shiue

Driscoll

Vasankari

Alsharif

Rahimi-Movaghar

Vlassov

V.V.

Marcenes

W.S.

Mekonnen

Melaku

Y.A.

Yano

Artaman

Campos

MacLachlan

Mueller

Kim

Trillini

Eshrati

Williams

H.C.

Shibuya

Dandona

Murthy

Cowie

Amare

A.T.

Antonio

C.A.

Castaneda-Orjuela

van Gool

C.H.

Violante

I.H.

Deribe

Soreide

Knibbs

Kereselidze

Green

Cardenas

Roy

Tillmann

Krueger

Monasta

Dey

Sheikhbahaei

Hafezi-Nejad

Kumar

G.A.

Sreeramareddy

C.T.

Dandona

Wang

Vollset

S.E.

Mokdad

Salomon

J.A.

Lozano

Vos

Forouzanfar

Lopez

Murray

and Naghavi

, The Global Burden of Cancer 2013, JAMA Oncol 1 (2015), 505–527.

16.

Franceschini

Szklarczyk

Frankild

Kuhn

Simonovic

Roth

Lin

Minguez

Bork

von Mering

and Jensen

L.J.

, STRING v9.1: protein-protein interaction networks, with increased coverage and integration, Nucleic Acids Res 41 (2013), D808–815.

17.

Hong

Downey

K.W.

Koh

P.K.

and Cheah

P.Y.

, A ‘metastasis-prone’ signature for early-stage mismatch-repair proficient sporadic colorectal cancer patients and its implications for possible therapeutics, Clin Exp Metastasis 27 (2010), 83–90.

18.

Clifford

R.J.

Yang

H.H.

Wang

Goldstein

A.M.

Ding

Taylor

P.R.

and Lee

M.P.

, Genome wide analysis of DNA copy number neutral loss of heterozygosity (CNNLOH) and its relation to gene expression in esophageal squamous cell carcinoma, BMC Genomics 11 (2010), 576.

19.

Zhang

Lou

and Jiang

, A novel COL10A1 mutation in a Chinese pedigree with Schmid type metaphyseal chondrodysplasia, Clin Lab 61 (2015), 227–233.

20.

Huang

D.W.

Sherman

B.T.

Tan

Collins

J.R.

Alvord

W.G.

Roayaei

Stephens

Baseler

M.W.

Lane

H.C.

and Lempicki

R.A.

, The DAVID Gene Functional Classification Tool: a novel biological module-centric algorithm to functionally analyze large gene lists, Genome Biol 8 (2007), R183.

21.

Huang

Zhao

Zhang

Wang

Zhang

Deng

and Liu

, High expression of COL10A1 is associated with poor prognosis in colorectal cancer, Onco Targets Ther 11 (2018), 1571–1581.

22.

Ilson

D.H.

, Advances in the treatment of gastric cancer, Curr Opin Gastroenterol 33 (2017), 473–476.

23.

Jemal

Bray

Center

M.M.

Ferlay

Ward

and Forman

, Global cancer statistics, CA Cancer J Clin 61 (2011), 69–90.

24.

Kaiser

Park

Y.K.

Franklin

J.L.

Halberg

R.B.

Jessen

W.J.

Freudenberg

Chen

Haigis

Jegga

A.G.

Kong

Sakthivel

Reichling

Azhar

Boivin

G.P.

Roberts

R.B.

Bissahoyo

A.C.

Gonzales

Bloom

G.C.

Eschrich

Carter

S.L.

Aronow

J.E.

Kleimeyer

Ramaswamy

Settle

S.H.

Boone

Levy

Graff

J.M.

Doetschman

Groden

Dove

W.F.

Threadgill

D.W.

Yeatman

T.J.

Coffey

R.J.

, Jr., and Aronow

B.J.

, Transcriptional recapitulation and subversion of embryonic colon development by mouse colon tumor models and human colon cancer, Genome Biol 8 (2007), R131.

25.

Kanehisa

, The KEGG database, Novartis Found Symp 247 (2002), 91–101; discussion 101–103, 119–128, 244–152.

26.

Kent

W.J.

Sugnet

C.W.

Furey

T.S.

Roskin

K.M.

Pringle

T.H.

Zahler

A.M.

and Haussler

, The human genome browser at UCSC, Genome Res 12 (2002), 996–1006.

27.

Kim

I.H.

Kwon

M.J.

and Nam

T.J.

, Differences in cell death and cell cycle following fucoidan treatment in high-density HT-29 colon cancer cells, Mol Med Rep 15 (2017), 4116–4122.

28.

K.K.

Xia

Chan

A.K.

Zhang

Z.Y.

Aibaidula

Zhang

Zhou

Yao

and Ng

H.K.

, The kinesin KIF14 is overexpressed in medulloblastoma and downregulation of KIF14 suppressed tumor proliferation and induced apoptosis, Lab Invest 97 (2017), 946–961.

29.

Zhu

Zhao

Zhang

Miao

Cao

and Fei

, FN1, SPARC, and SERPINE1 are highly expressed and significantly related to a poor prognosis of gastric adenocarcinoma revealed by microarray and bioinformatics, Sci Rep 9 (2019).

30.

Huang

Shi

Zhao

Zhang

Liu

and Li

, TGF-beta1-SOX9 axis-inducible COL10A1 promotes invasion and metastasis in gastric cancer via epithelial-to-mesenchymal transition, Cell Death Dis 9 (2018), 849.

31.

Lote

Cafferkey

and Chau

, PD-1 and PD-L1 blockade in gastrointestinal malignancies, Cancer Treat Rev 41 (2015), 893–903.

32.

Lucas

Barnich

and Nguyen

H.T.T.

, Microbiota, Inflammation and Colorectal Cancer, Int J Mol Sci 18 (2017).

33.

Mendonca

F.M.

de Sousa

F.R.

Barbosa

A.L.

Martins

S.C.

Araujo

R.L.

Soares

and Abreu

, Metabolic syndrome and risk of cancer: which link? Metabolism 64 (2015), 182–189.

34.

Ming

X.Y.

Zhang

Cao

T.T.

Zhang

L.Y.

J.L.

Kam

N.W.

Tang

X.M.

Cui

Y.Z.

Zhang

B.Z.

Qin

Y.R.

and Guan

X.Y.

, RHCG Suppresses Tumorigenicity and Metastasis in Esophageal Squamous Cell Carcinoma via Inhibiting NF-κB Signaling and MMP1 Expression, Theranostics 8 (2018), 185–198.

35.

Mirza-Aghazadeh-Attari

Darband

S.G.

Kaviani

Mihanfar

Aghazadeh Attari

Yousefi

and Majidinia

, DNA damage response and repair in colorectal cancer: Defects, regulation and therapeutic implications, DNA Repair (Amst) 69 (2018), 34–52.

36.

Moawia

Shaheen

Rasool

Waseem

S.S.

Ewida

Budde

Kawalia

Motameny

Khan

Fatima

Jameel

Ullah

Akram

Ali

Abdullah

Irshad

Hohne

Noegel

A.A.

Al-Owain

Hortnagel

Stobe

Baig

S.M.

Nurnberg

Alkuraya

F.S.

Hahn

and Hussain

M.S.

, Mutations of KIF14 cause primary microcephaly by impairing cytokinesis, Ann Neurol 82 (2017), 562–577.

37.

Nakamura

and Nishimura

, History, molecular features, and clinical importance of conventional serum biomarkers in lung cancer, Surg Today 47 (2017), 1037–1059.

38.

Nicolau-Neto

Da Costa

N.M.

de Souza Santos

P.T.

Gonzaga

I.M.

Ferreira

M.A.

Guaraldi

Moreira

M.A.

Seuanez

H.N.

Brewer

Bergmann

Boroni

Mencalha

A.L.

Kruel

C.D.P.

Lima

S.C.S.

Esposito

Simao

T.A.

and Pinto

L.F.R.

, Esophageal squamous cell carcinoma transcriptome reveals the effect of FOXM1 on patient outcome through novel PIK3R3 mediated activation of PI3K signaling pathway, Oncotarget 9 (2018), 16634–16647.

39.

O’Hare

Shadmand

Sulaiman

R.S.

Sishtla

Sakisaka

and Corson

T.W.

, Kif14 overexpression accelerates murine retinoblastoma development, Int J Cancer 139 (2016), 1752–1758.

40.

S.C.

Sohn

B.H.

Cheong

J.H.

Kim

S.B.

Lee

J.E.

Park

K.C.

Lee

S.H.

Park

J.L.

Park

Y.Y.

Lee

H.S.

Jang

H.J.

Park

E.S.

Kim

S.C.

Heo

Chu

I.S.

Jang

Y.J.

Mok

Y.J.

Jung

Kim

B.H.

Kim

Cho

J.Y.

Lim

J.Y.

Hayashi

Song

Elimova

Estralla

J.S.

Lee

J.H.

Bhutani

M.S.

Liu

Lee

Kang

W.K.

Kim

Noh

S.H.

Mills

G.B.

Kim

S.Y.

Ajani

J.A.

and Lee

J.S.

, Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype, Nat Commun 9 (2018).

41.

Park

Morley

T.S.

Kim

Clegg

D.J.

and Scherer

P.E.

, Obesity and cancer – mechanisms underlying tumour progression and recurrence, Nat Rev Endocrinol 10 (2014), 455–465.

42.

Qiu

H.L.

Deng

S.Z.

Tian

Z.N.

Song

X.Q.

Yao

G.D.

and Geng

J.S.

, High expression of KIF14 is associated with poor prognosis in patients with epithelial ovarian cancer, Eur Rev Med Pharmacol Sci 21 (2017), 239–245.

43.

Rahmutulla

Matsushita

and Nomura

, Alternative splicing of DNA damage response genes and gastrointestinal cancers, World J Gastroenterol 20 (2014), 17305–17313.

44.

Ronchetti

Melucci

De Nicola

Goeman

Casini

Sperati

Pallocca

Terrenato

Pizzuti

Vici

Sergi

Di Lauro

Amoreo

C.A.

Gallo

Diodoro

M.G.

Pescarmona

Vitale

Barba

Buglioni

Mottolese

Fanciulli

De Maria

and Maugeri-Sacca

, DNA damage repair and survival outcomes in advanced gastric cancer patients treated with first-line chemotherapy, Int J Cancer 140 (2017), 2587–2595.

45.

Sauzay

Petit

Bourgeois

A.M.

Barbare

J.C.

Chauffert

Galmiche

and Houessinon

, Alpha-foetoprotein (AFP): A multi-purpose marker in hepatocellular carcinoma, Clin Chim Acta 463 (2016), 39–44.

46.

Siegel

R.L.

Miller

K.D.

and Jemal

, Cancer statistics, 2019, CA Cancer J Clin 69 (2019), 7–34.

47.

Singel

S.M.

Cornelius

Zaganjor

Batten

Sarode

V.R.

Buckley

D.L.

Peng

John

G.B.

H.C.

Sadeghi

Wright

W.E.

Lum

Corson

T.W.

and Shay

J.W.

, KIF14 promotes AKT phosphorylation and contributes to chemoresistance in triple-negative breast cancer, Neoplasia 16 (2014), 247–256, 256e242..

48.

Skrzypczak

Goryca

Rubel

Paziewska

Mikula

Jarosz

Pachlewski

Oledzki

and Ostrowski

, Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data directed for maximization of analytical reliability, PLoS One 5 (2010).

49.

Smoot

M.E.

Ono

Ruscheinski

Wang

P.L.

and Ideker

, Cytoscape 2.8: new features for data integration and network visualization, Bioinformatics 27 (2011), 431–432.

50.

Sonnenberg

W.R.

, Gastrointestinal Malignancies, Prim Care 44 (2017), 721–732.

51.

Wang

Shi

Cui

and Yang

, Up-regulation of KIF14 is a predictor of poor survival and a novel prognostic biomarker of chemoresistance to paclitaxel treatment in cervical cancer, Biosci Rep 36 (2016).

52.

Zhou

Huang

Liu

Zhan

Zhang

and Zhao

, Role of Smac in determining the chemotherapeutic response of esophageal squamous cell carcinoma, Clin Cancer Res 17 (2011), 5412–5422.

53.

Guo

Sun

Han

Qian

Kryczek

Sun

Nagarsheth

Chen

Hong

Zou

and Fang

J.Y.

, Fusobacterium nucleatum Promotes Chemoresistance to Colorectal Cancer by Modulating Autophagy, Cell 170 (2017), 548–563e516.

54.

Zhang

Liu

Iqbal

Mustafa

Bukhari

Lou

Yang

Bukhari

Aslam

and Xu

, A missense point mutation in COL10A1 identified with whole-genome deep sequencing in a 7-generation Pakistan dwarf family, Heredity (Edinb) 120 (2018), 83–89.

55.

Zhang

Yuan

Liang

Zhang

Guo

Xia

Zhao

Shu

X.S.

Sun

Ying

and Cheng

, Overexpression of a novel candidate oncogene KIF14 correlates with tumor progression and poor prognosis in prostate cancer, Oncotarget 8 (2017), 45459–45469.

Screening and identification of key biomarkers in alimentary tract cancers: A bioinformatic analysis

Abstract

BACKGROUND:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Materials and methods

2.1 Microarray acquirement

Table 1 Basic information of microarray datasets of ATCs

2.3 KEGG and GO enrichment analyses

2.4 PPI network construction

2.5 Module analysis

2.6 Identification and analysis of hub genes

2.8 RT-qPCR measurement

3. Results

3.1 Identification of DEGs in ATCs

3.2 PPI network and module analyses

3.3 Hub gene analysis and biological enrichment analyses of DEGs

4. Conclusion

Ethics approval and consent to participate

Availability of data and material

Funding

Footnotes

Conflict of interest

Abbreviation

Supplementary data

References

Table 1
Basic information of microarray datasets of ATCs