|Year : 2018 | Volume
| Issue : 10 | Page : 675-679
Identification of genes correlated with oral squamous cell carcinoma
Ting Lin1, Bin Zhang2, Hong He3
1 The State Key Laboratory Breeding Base of Basic Science of Stomatology (Hubei-MOST) and Key Laboratory of Oral Biomedicine Ministry of Education (KLOBM), School & Hospital of Stomatology, Wuhan University, Wuhan 430079, Hubei; Key Laboratory of Oral Medicine, Guangzhou Institute of Oral Disease, Stomatology Hospital of Guangzhou Medical University, Guangzhou 510140, P.R. China
2 Key Laboratory of Oral Medicine, Guangzhou Institute of Oral Disease, Stomatology Hospital of Guangzhou Medical University, Guangzhou 510140, P.R. China
3 The State Key Laboratory Breeding Base of Basic Science of Stomatology (Hubei-MOST) and Key Laboratory of Oral Biomedicine Ministry of Education (KLOBM), School & Hospital of Stomatology, Wuhan University, Wuhan 430079, Hubei, P.R. China
|Date of Web Publication||24-Sep-2018|
Department of Orthodontics, School and Hospital of Stomatology, Wuhan University, No. 237, Luoyu Road, Hongshan District, Wuhan 430079
Source of Support: None, Conflict of Interest: None
Objectives: The objective was to study the mechanisms of oral squamous cell carcinoma (OSCC).
Materials and Methods: We analyzed microarrays of GSE23558 and GSE25103. GSE23558 and GSE25103 were downloaded from Gene Expression Omnibus. GSE23558 included 27 OSCC samples, 4 independent and 1 pooled normal samples. GSE25103 included 112 OSCC samples and ten normal samples. The differentially expressed genes (DEGs) and the risk single nucleotide polymorphisms (SNPs) separately were obtained by limma package and plink software. Then, candidate disease genes were screened from the common genes of the genes carrying SNPs and the DEGs using Fisher's combination method. Using TargetMine online tool, potential functions of the candidate disease genes were analyzed by functional and pathway enrichment analyses. Besides, protein–protein interaction (PPI) network of these genes was constructed by STRING and Cytoscape software. Furthermore, modules of PPI network were screened by the ClusterONE.
Results: We screened 2353 DEGs and 35635 risk SNPs in OSCC samples compared with normal samples. Moreover, CA9 was the most significant upregulated genes. There were 754 candidate disease genes, including 299 upregulated (e.g., VEGFC and FAT1) and 455 downregulated genes. For the candidate disease genes, the enriched functions were mainly in biological process categories. Importantly, FN1 (degree = 42) and CCNA2 (degree = 38) had high degrees in the PPI network. Furthermore, FN1 and CCNA2 were separately involved in module 1 and module 2 of the PPI network. FN1, CCNA2, CA9, VEGFC, and FAT1 might affect OSCC.
Conclusion: In general, our study obtained important genes implicated in OSCC.
Keywords: Differentially expressed genes, enrichment analysis, oral squamous cell carcinoma, protein–protein interaction network, risk single nucleotide polymorphisms
|How to cite this article:|
Lin T, Zhang B, He H. Identification of genes correlated with oral squamous cell carcinoma. J Can Res Ther 2018;14, Suppl S3:675-9
| > Introduction|| |
Oral squamous cell carcinoma (OSCC) originates from the mucosa of the oropharynx and oral cavity., Especially, tobacco smoking and alcohol drinking are the two main risk factors for OSCC. As the sixth most prevalent cancer in the world, OSCC affects up to 405,000 people each year. Thus, it is necessary to identify genes correlated with OSCC.
Recently, several studies have focused on investigating the molecular mechanisms of OSCC. For example, as a key member of the phosphatidylinositol 3 kinase (PI3K) signaling pathway, RAS may be involved in the tumorigenesis of OSCC. With genetic changes, p53 gene mutation is thus playing a role in the genesis of OSCC and may be useful in the diagnosis of this neoplasm., Expression of SNAIL and E-cadherin have a reverse correlation in OSCC cells, and the expression of SNAIL contributes to the invasion and metastasis of OSCC. Being induced by hepatocyte growth factor, expression of the Ets-related E1AF transcription factor gene can activate matrix metallopeptidase 9 genes and cause OSCC cell invasion., As an inhibitor of apoptosis protein, survivin expression may be used to identify more aggressive and invasive OSCC, thus, can help to make an appropriate therapeutic scheme.,,, High-mobility group AT-hook 2 can promote the aggressiveness of carcinoma and may be used as a valuable marker in the prognosis of oral carcinomas.
In 2012, Ambatipudi et al. used a gene expression analysis to analyze the differentially expressed genes (DEGs) between gingivobuccal complex (GBC) cancer samples and normal GBC samples and obtained 315 DEGs. In 2011, Peng et al. used Affymetrix GeneChip Human Gene 1.0 ST and Affymetrix single nucleotide polymorphism (SNP) 6.0 arrays to investigate the genetic variations in gene expression related to poor prognosis in OSCC and identified 85 copy number variations-associated transcripts. Using the same data, we aimed to separately screen the DEGs and the risk SNPs between OSCC samples and normal samples. Then, the genes carrying SNPs were compared to the DEGs to obtain their common genes, and candidate disease genes were screened from the common genes. Besides, potential functions of the candidate disease genes were analyzed by Gene Ontology (GO) and pathway enrichment analyses. In addition, the interaction relationships between these candidate disease genes were investigated by protein–protein interaction (PPI) network and modules of PPI network.
| > Materials and Methods|| |
Gene expression profile of GSE23558 deposited by Ambatipudi et al. and SNP array data GSE25103 deposited by Peng et al. were downloaded from Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/). GSE23558, which was based on the platform of GPL6480 Agilent-014850 Whole Human Genome Microarray 4 × 44K G4112F (Probe Name version), included a collective of 27 OSCC samples, 4 independent normal samples, and 1 pooled normal samples. GSE25103, which was based on the platform of GPL6801 (GenomeWideSNP_6) Affymetrix Genome-Wide Human SNP 6.0 Array, included a collective of 112 OSCC samples and 10 normal samples.
Differentially expressed genes and risk single nucleotide polymorphisms screening
After GSE23558 was downloaded, microarray data were preprocessed. Combining with the annotations in the platform of GPL6480, probe IDs were converted into gene symbols. Then, the mean value of probes corresponding to one gene was obtained as gene expression value. The linear models for microarray data (limma) package in Bioconductor (http://www.bioconductor.org/packages/release/bioc/html/limma.html) were used to screen the DEGs between OSCC samples and normal samples. The adjusted P < 0.05 and | log2 fold-change (FC)| ≥1 were used as the cut off criteria.
To SNP array data, probes with typing rates <90% were excluded. The plink software (http://www.pngu.mgh.harvard.edu/purcell/plink/) was applied in calculating risk SNPs between OSCC samples and normal samples. All the parameters were set as default values, and associated function was used in the correlation analysis. P ≤ 0.05 was used as the cut off criterion.
Candidate disease genes obtaining
After risk SNPs had been obtained, genes carrying SNPs were identified based on the annotations in the platform GPL6801. To one gene carried several SNPs, the SNP with the minimum P value was considered as its representative SNP. Then, the genes carrying SNPs were compared to the DEGs, and their common genes were obtained. For screening candidate disease genes from the common genes, the Fisher's combination method was used to integrate the P values of risk SNP and DEG corresponding to one common gene. P < 0.05 was used as the cut off criterion.
Functional and pathway enrichment analysis
GO terms are useful in describing the subcellular location and molecular function of genes, as well as the biological processes (BPs) involving genes. Kyoto encyclopedia of genes and genomes (KEGG) is a database that provides function mechanisms of molecules or genes. Using TargetMine online tool (http://www.targetmine.nibio.go.jp), GO and KEGG pathway analyses were conducted for the candidate disease genes between OSCC samples and normal samples. P ≤ 0.05 was used as the cut off criterion.
Protein–protein interaction network and module construction
Interaction relationships of the proteins encoded by the candidate disease genes were searched by the STRING online software (http://www.string-db.org). All the parameters were set as default values, and the combined score ≥0.4 was used as the cut off criterion. Then, the PPI network was visualized by the Cytoscape software (http://www.cytoscape.org). The ClusterONE (http://www.paccanarolab.org/cluster-one/index.html) was used to screen modules from the PPI network. Minimum size was set as 10, and Edge weights were same as combined score, as well as the other parameters were set to the default values. P ≤ 0.05 was used as the cut off criterion.
| > Results|| |
Differentially expressed genes, risk single nucleotide polymorphisms, and candidate disease genes analysis
Compared with normal samples, there were 2353 DEGs (including 1076 upregulated and 1277 downregulated genes) and 35635 risk SNPs in OSCC samples. Moreover, carbonic anhydrase IX (CA9) was the most significant up regulated genes. Meanwhile, a total of 754 candidate disease genes were screened, including 299 upregulated (e.g., vascular endothelial growth factor C, VEGFC and FAT tumor suppressor homolog 1, FAT1) and 455 downregulated genes.
Functional and pathway enrichment analysis
The enriched GO functions for the upregulated genes in candidate disease genes were mainly in BP categories, including mitotic cell cycle (P = 4.41882E-06), anatomical structure morphogenesis (P = 0.000400403), and cell cycle process (P = 0.00056096) [Table 1]. The enriched KEGG pathways for the upregulated genes in candidate disease genes included extracellular matrix (ECM)-receptor interaction (P = 0.000159375), focal adhesion (P = 0.004293289), PI3K-Akt signaling pathway (P = 0.023100961), and DNA replication (P = 0.036752839) [Table 1]. The enriched GO functions for the downregulated genes in candidate disease genes were also mainly in BP categories, including multicellular organismal development (P = 0.000813621), single-multicellular organism process (P = 0.000813621), and urogenital system development (P = 0.001972897) [Table 1]. There was no enriched pathway for the downregulated genes in candidate disease genes.
|Table 1: The enriched Gene Ontology functions and Kyoto Encyclopedia of Genes and Genomes pathways separately for the up- and down-regulated candidate disease genes|
Click here to view
Protein–protein interaction network and module analysis
The PPI network of the candidate disease genes had 489 nodes and 1425 interactions [Figure 1]. Especially, fibronectin 1 (FN1, degree = 42), minichromosome maintenance 8 (MCM8, degree = 42), and cyclin A2 (CCNA2, degree = 38) were with high degrees in the PPI network. Besides, total 6 modules (module 1, module 2, module 3, module 4, module 5, and module 6) were obtained from the PPI network [Figure 2]. Meanwhile, FN1 and CCNA2 were separately involved in module 1 and module 2 of the PPI network. Moreover, GO enrichment analyses were performed for the modules and the results are listed in [Table 2].
|Figure 1: The protein–protein interaction network of the candidate disease genes|
Click here to view
|Figure 2: The modules (module 1, module 2, module 3, module 4, module 5, and module 6) obtained from the protein–protein interaction network of the candidate disease genes|
Click here to view
|Table 2: The enriched most significant function for the differentially expressed genes involved in the 6 modules of the protein–protein interaction network|
Click here to view
| > Discussion|| |
In this study, we screened 2353 DEGs (including 1076 upregulated and 1277 downregulated genes) and 35635 risk SNPs in OSCC samples compared with normal samples. There were a total of 754 candidate disease genes, including 299 upregulated and 455 downregulated genes. The enriched GO functions for the up- and down-regulated genes in candidate disease genes were mainly in BP categories. Importantly, FN1 (degree = 42) and CCNA2 (degree = 38) had high degrees in the PPI network of the candidate disease genes. Furthermore, FN1 and CCNA2 were separately involved in module 1 and module 2 of the PPI network.
It is reported that the ECM molecule FN can contribute to the metastatic phenotype of OSCC cells and is a promising prognostic indicator in OSCC. By inducing FN and hypoxia inducible factor-1α-dependent α5 integrin, hypoxia can promote OSCC cell invasion. These declared that FN1 might have a close correlation with OSCC. As members of cyclins which function in mediating the passage of dividing cells, cyclin A and cyclin B1 proteins overexpress in oral carcinoma. Cyclin A expresses in the germinal center cells of the lymphoid follicles under the SCCs of the esophagus and may be an immunological signal of the progression and proliferation of the carcinoma. The defect in the signaling component that regulates downregulation of CCNA2 and cyclin-dependent kinase 2 expression induced by IFNgamma can lead to the resistance of OSCC to IFNgamma. Thus, the expression levels of CCNA2 might have relation to OSCC.
The CA9 expression is a tumor-specific event in OSCC and it is related to disease recurrence, poor clinical outcome, and nodal metastasis, as well as may be a potential prognostic factor for OSCC.,, As a glycoprotein located on the surface of cell membranes, CA9 expresses in most OSCC patients and patients with higher CA9 expression have a worse outcome. VEGFC expression may induce lymphatic angiogenesis and serve as a reliable predictor of regional lymph node metastasis in early OSCC., As a member of the cadherin superfamily, intrinsic membrane protein FAT1 may be implicated in the migration and invasion of OSCCs; thus, it may serve as a key target for developing new therapeutic strategies. These might indicate that the expression levels of CA9, VEGFC, and FAT1 had relation to OSCC.
| > Conclusion|| |
In general, our study obtained important genes implicated in OSCC. We screened 2353 DEGs and 35635 risk SNPs in OSCC samples compared with normal samples. Besides, a total of 754 candidate disease genes were identified. Moreover, some genes might be related to OSCC, such as FN1, CCNA2, CA9, VEGFC, and FAT1. Nevertheless, further experimental researches are still needed to validate the functions of these genes in OSCC.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| > References|| |
Gillison ML. Current topics in the epidemiology of oral cavity and oropharyngeal cancers. Head Neck 2007;29:779-92.
Moore SR, Johnson NW, Pierce AM, Wilson DF. The epidemiology of mouth cancer: A review of global incidence. Oral Dis 2000;6:65-74.
Bundgaard T, Bentzen SM, Wildt J. The prognostic effect of tobacco and alcohol consumption in intra-oral squamous cell carcinoma. Eur J Cancer B Oral Oncol 1994;30B: 323-8.
Peng CH, Liao CT, Peng SC, Chen YJ, Cheng AJ, Juang JL, et al.
A novel molecular signature identified by systems genetics approach predicts prognosis in oral squamous cell carcinoma. PLoS One 2011;6:e23452.
Koumaki D, Kostakis G, Koumaki V, Papadogeorgakis N, Makris M, Katoulis A, et al.
Novel mutations of the HRAS gene and absence of hotspot mutations of the BRAF genes in oral squamous cell carcinoma in a Greek population. Oncol Rep 2012;27:1555-60.
Sakai E, Rikimaru K, Ueda M, Matsumoto Y, Ishii N, Enomoto S, et al.
The p53 tumor-suppressor gene and ras oncogene mutations in oral squamous-cell carcinoma. Int J Cancer 1992;52:867-72.
Hsieh LL, Wang PF, Chen IH, Liao CT, Wang HM, Chen MC, et al.
Characteristics of mutations in the p53 gene in oral squamous cell carcinoma associated with betel quid chewing and cigarette smoking in Taiwanese. Carcinogenesis 2001;22:1497-503.
Yokoyama K, Kamata N, Hayashi E, Hoteiya T, Ueda N, Fujimoto R, et al.
Reverse correlation of E-cadherin and snail expression in oral squamous cell carcinoma cells in vitro
. Oral Oncol 2001;37:65-71.
Hanzawa M, Shindoh M, Higashino F, Yasuda M, Inoue N, Hida K, et al.
Hepatocyte growth factor upregulates E1AF that induces oral squamous cell carcinoma cell invasion by activating matrix metalloproteinase genes. Carcinogenesis 2000;21:1079-85.
Uchida D, Kawamata H, Omotehara F, Nakashiro Ki, Kimura-Yanagawa T, Hino S, et al.
Role of HGF/c-met system in invasion and metastasis of oral squamous cell carcinoma cells in vitro
and its clinical significance. Int J Cancer 2001;93:489-96.
Lo Muzio L, Pannone G, Staibano S, Mignogna MD, Rubini C, Mariggiò MA, et al.
Survivin expression in oral squamous cell carcinoma. Br J Cancer 2003;89:2244-8.
Lo Muzio L, Staibano S, Pannone G, Mignogna MD, Mariggiò A, Salvatore G, et al.
Expression of the apoptosis inhibitor survivin in aggressive squamous cell carcinoma. Exp Mol Pathol 2001;70:249-54.
Engels K, Knauer SK, Metzler D, Simf C, Struschka O, Bier C, et al.
Dynamic intracellular survivin in oral squamous cell carcinoma: Underlying molecular mechanism and potential as an early prognostic marker. J Pathol 2007;211:532-40.
Lo Muzio L, Farina A, Rubini C, Pezzetti F, Stabellini G, Laino G, et al.
Survivin as prognostic factor in squamous cell carcinoma of the oral cavity. Cancer Lett 2005;225:27-33.
Miyazawa J, Mitoro A, Kawashiri S, Chada KK, Imai K. Expression of mesenchyme-specific gene HMGA2 in squamous cell carcinomas of the oral cavity. Cancer Res 2004;64:2024-9.
Ambatipudi S, Gerstung M, Pandey M, Samant T, Patil A, Kane S, et al.
Genome-wide expression and copy number analysis identifies driver genes in gingivobuccal cancers. Genes Chromosomes Cancer 2012;51:161-73.
Smyth GK. Limma: Linear models for microarray data. Bioinformatics and Computational Biology Solutions Using R and Bioconductor. New York: Springer; 2005. p. 397-420.
Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al.
PLINK: A tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81:559-75.
Banik N, Köhne K, Bauer P. On the power of fisher's combination test for two stage sampling in the presence of nuisance parameters. Biom J 1996;38:25-37.
Tweedie S, Ashburner M, Falls K, Leyland P, McQuilton P, Marygold S, et al.
FlyBase: Enhancing drosophila gene ontology annotations. Nucleic Acids Res 2009;37:D555-9.
Kanehisa M, Goto S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 2000;28:27-30.
Chen YA, Tripathi LP, Mizuguchi K. TargetMine, an integrated data warehouse for candidate gene prioritisation and target discovery. PLoS One 2011;6:e17844.
Franceschini A, Szklarczyk D, Frankild S, Kuhn M, Simonovic M, Roth A, et al.
STRING v9.1: Protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res 2013;41:D808-15.
Kohl M, Wiese S, Warscheid B. Cytoscape: Software for visualization and analysis of biological networks. Methods Mol Biol 2011;696:291-303.
Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al.
Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res 2003;13:2498-504.
Zhang Y, Lu H, Dazin P, Kapila Y. Squamous cell carcinoma cell aggregates escape suspension-induced, p53-mediated anoikis: Fibronectin and integrin alphav mediate survival signals through focal adhesion kinase. J Biol Chem 2004;279:48342-9.
Lyons AJ, Bateman AC, Spedding A, Primrose JN, Mandel U. Oncofetal fibronectin and oral squamous cell carcinoma. Br J Oral Maxillofac Surg 2001;39:471-7.
Ryu MH, Park HM, Chung J, Lee CH, Park HR. Hypoxia-inducible factor-1alpha mediates oral squamous cell carcinoma invasion via upregulation of alpha5 integrin and fibronectin. Biochem Biophys Res Commun 2010;393:11-5.
Kushner J, Bradley G, Young B, Jordan RC. Aberrant expression of cyclin A and cyclin B1 proteins in oral carcinoma. J Oral Pathol Med 1999;28:77-81.
Nozoe T, Korenaga D, Futatsugi M, Saeki H, Ohga T, Sugimachi K. Cyclin A expression in superficial squamous cell carcinoma of the esophagus and coexisting infiltrated lymphocyte follicle. Cancer Lett 2002;188:221-9.
Hiroi M, Mori K, Sekine K, Sakaeda Y, Shimada J, Ohmori Y. Mechanisms of resistance to interferon-gamma-mediated cell growth arrest in human oral squamous carcinoma cells. J Biol Chem 2009;284:24869-80.
Choi SW, Kim JY, Park JY, Cha IH, Kim J, Lee S. Expression of carbonic anhydrase IX is associated with postoperative recurrence and poor prognosis in surgically treated oral squamous cell carcinoma. Hum Pathol 2008;39:1317-22.
Kim SJ, Shin HJ, Jung KY, Baek SK, Shin BK, Choi J, et al.
Prognostic value of carbonic anhydrase IX and Ki-67 expression in squamous cell carcinoma of the tongue. Jpn J Clin Oncol 2007;37:812-9.
Brockton NT, Klimowicz AC, Bose P, Petrillo SK, Konno M, Rudmik L, et al.
High stromal carbonic anhydrase IX expression is associated with nodal metastasis and decreased survival in patients with surgically-treated oral cavity squamous cell carcinoma. Oral Oncol 2012;48:615-22.
Kondo Y, Yoshikawa K, Omura Y, Shinohara A, Kazaoka Y, Sano J, et al.
Clinicopathological significance of carbonic anhydrase 9, glucose transporter-1, Ki-67 and p53 expression in oral squamous cell carcinoma. Oncol Rep 2011;25:1227-33.
Kishimoto K, Sasaki A, Yoshihama Y, Mese H, Tsukamoto G, Matsumura T. Expression of vascular endothelial growth factor-C predicts regional lymph node metastasis in early oral squamous cell carcinoma. Oral Oncol 2003;39:391-6.
Sedivy R, Beck-Mannagetta J, Haverkampf C, Battistutti W, Hönigschnabl S. Expression of vascular endothelial growth factor-C correlates with the lymphatic microvessel density and the nodal status in oral squamous cell cancer. J Oral Pathol Med 2003;32:455-60.
Nishikawa Y, Miyazaki T, Nakashiro K, Yamagata H, Isokane M, Goda H, et al.
Human FAT1 cadherin controls cell migration and invasion of oral squamous cell carcinoma through the localization of ß-catenin. Oncol Rep 2011;26:587-92.
[Figure 1], [Figure 2]
[Table 1], [Table 2]
|This article has been cited by|
||Combined impacts of histamine receptor H1 gene polymorphisms and an environmental carcinogen on the susceptibility to and progression of oral squamous cell carcinoma
| ||Yi-Fang Ding, Yung-Wei Lin, Wen-Kuan Chiu, Chiao-Wen Lin, Yi-Chieh Yang, Lun-Ching Chang, Jungshan Chang, Shun-Fa Yang, Ming-Hsien Chien |
| ||Aging. 2022; |
|[Pubmed] | [DOI]|
||Construction of Prognostic Risk Prediction Model of Oral Squamous Cell Carcinoma Based on Nine Survival-Associated Metabolic Genes
| ||Zhen-Dong Huang, Yang-Yang Yao, Ting-Yu Chen, Yi-Fan Zhao, Chao Zhang, Yu-Ming Niu |
| ||Frontiers in Physiology. 2021; 12 |
|[Pubmed] | [DOI]|
||Integrated Analysis of lncRNA-Associated ceRNA Network Identifies Two lncRNA Signatures as a Prognostic Biomarker in Gastric Cancer
| ||Shuyan Zhang, Shanshan Li, Jian-Lin Guo, Ningyi Li, Cai-Ning Zhang, Jie Liu, Paul Ashwood |
| ||Disease Markers. 2021; 2021: 1 |
|[Pubmed] | [DOI]|
||Epstein–Barr Virus—Oral Bacterial Link in the Development of Oral Squamous Cell Carcinoma
| ||Daniela Núñez-Acurio, Denisse Bravo, Francisco Aguayo |
| ||Pathogens. 2020; 9(12): 1059 |
|[Pubmed] | [DOI]|
||Network-based identification of signature genes KLF6 and SPOCK1 associated with oral submucous fibrosis
| ||Prithvi Singh, Arpita Rai, Ravins Dohare, Shweta Arora, Sher Ali, Shama Parveen, Mansoor Syed |
| ||Molecular and Clinical Oncology. 2020; |
|[Pubmed] | [DOI]|
||Bioinformatics analysis to screen key genes in papillary thyroid carcinoma
| ||Yuanhu Liu, Shuwei Gao, Yaqiong Jin, Yeran Yang, Jun Tai, Shengcai Wang, Hui Yang, Ping Chu, Shujing Han, Jie Lu, Xin Ni, Yongbo Yu, Yongli Guo |
| ||Oncology Letters. 2019; |
|[Pubmed] | [DOI]|