This article is mentioned in:

Introduction

Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, accounting for about 85% of all lung cancers (1). NSCLC is the main cause of lung cancer-related deaths. Most patients are diagnosed with late-stage tumors and only 15% of patients are found in the early stage (2). The survival rate of patients with early-stage NSCLC after complete tumor resection is more favourable. The current therapeutic strategies for NSCLC cases mainly include surgery, chemotherapy and radiotherapy. The combination of two or more treatments is often used for NSCLC cases, which is usually influenced by NSCLC staging. It is reported that the combinational treatment of adjuvant carboplatin plus paclitaxel reduces the mortality rate of patients with stage IB NSCLC (3). The overall survival for stage IB-II NSCLC cases was found to improve after vinorelbine plus cisplatin treatment (4).

Microarray technology is a novel tool to analyze the global gene expression and decipher biological pathways (5). Advances in microarray technology have facilitated the development of novel therapies and the understanding of potential mechanisms of diseases. Based on gene expression arrays, Lu et al found a group of genes related with lung cancer metastasis such as APC, CDH8 and IL8RB, which may be used for predicting the survival of patients with stage I NSCLC and preparing a treatment plan (6). Jiang et al employed tissue lung microarray technology and suggested that aldehyde dehydrogenase 1, a marker for lung tumor stem cells, may be a therapeutic target for lung cancer (7).

The gene expression array of GSE21933 was developed from Asian lung cancer patients. With the microarray data, Lo et al have established that ARHGAP19, FRAT2g, PAFAH1B1 and ZNF322A involved in Rho activity, Wnt signaling, motility control and MAPK signaling are the candidate genes for lung cancer (8). Currently, the integrity analysis of altered gene expression patterns for lung cancer during different stages is rare. In the present study, we further analyzed the expression array data of lung tumor tissues to disclose the difference and similarity in the gene expression profile of lung cancer cases during different stages. We expect that our findings are helpful in understanding the mechanism of lung cancer during different stages and discovering novel candidate genes.

Materials and methods

Microarray data

The gene expression dataset of a previous human NSCLC study was retrieved from the public National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) with accession no. GSE21933. The gene expression data of tumors and matched normal tissues from 21 NSCLC patients were contributed by a previously public study (8). The NSCLC patients included 10 early-stage cases (6 in stage IB, 3 in stage IIB and 1 in stage IA) and 11 late-stage patients (5 in stage IIIA, 5 in stage IV and 1 in stage IIIB). Due to the small number of samples at stage IA and IIIB, we only analyzed the remaining 19 tumor samples compared with the corresponding controls. All the DNA microarray data were analyzed by the platform Phalanx Human OneArray.

Data preprocessing

The raw data (gpr format files) were downloaded and assessed by limma software (9,10). The preprocessing procedure involved a background correction, gene expression normalization and microarray data condensing. Then, the probe IDs were translated into gene symbols by eliminating the probes with no gene symbols. When there were multiple probes for the same gene symbol, the mean value of the probes was calculated as the gene expression value.

DEG screening

The differentially expressed genes (DEGs) were analyzed in NSCLC tumors during different stages. The differential expression patterns were analyzed in stages IB, IIB, IIIA and IV of lung tumors, compared with the normal controls. The significant P-values for DEGs were calculated by a no-paired t-test with the application of a limma package. P<0.05 and log2FC (fold-change) ≥0.4 were defined as the cut-off values. Then the significant DEGs screened were subjected to hierarchical clustering analysis by a gplots package (11).

Function and pathway analysis

The significantly altered biological processes and pathways were identified using Database for Annotation, Visualization and Integrated Discovery (DAVID) (12). The statistical significances of the overrepresented Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were calculated based on hypergeometric distribution. A P-value <0.05 and a count ≥2 were used as the cut-off values.

PPI network

The protein interaction pairs corresponding to DEGs were predicted by integrating the information deposited in STRING [protein-protein interaction (PPI) networks, with increased coverage and integration] (13). The DEG lists in tumors in stages IB, IIB, IIIA and IV were analyzed, respectively. The PPI network was constructed of the protein pairs with a PPI score (medium confidence) ≥0.4 using Cytoscape software (14).

Key nodes in the PPI network

The role of nodes in large complex networks is commonly assessed based on network centralities such as degree centrality (15), betweenness centrality (16) and subgraph centrality (17). The network centralities of nodes were calculated by cytoscape plugin CytoNCA (18). The key nodes were first selected according to a high degree, subgraph and betweenness value. Then the DEGs corresponding to key nodes underwent cluster analysis to evaluate the effect on sample clustering. Finally, DEGs that could distinguish tumor and normal samples were defined as key nodes.

Venn diagram of DEGs in tumors of different stages

VennPlex (19) (https://www.irp.nia.nih.gov/bioinformatics/vennplex.html) has been widely used to separate common DEGs and uniquely altered genes between different gene lists. The Venn diagram of DEGs in the NSCLC tumors of different stages was constructed by VennPlex software. The four DEG lists and log2FC values were uploaded to identify the difference of differential expression patterns among stages IB, IIB, IIIA and IV of lung tumors.

Gene coexpression analysis

The DEGs identified in the four different tumor stages were pooled together. The coexpression coefficient was calculated by Pearson's correlation coefficient. P-values were obtained based on Z-scores and corrected by Benjamini-Hochberg (BH) method to control the false discovery rate (20). A |coexpression coefficient| >0.8 and an adjusted P-value <0.05 were set as the cut-off values for gene coexpression pairs.

Analysis of cancer-related genes

The Comparative Toxicogenomics Database (CTD) (21) is a collection of disease-related genes documented in previously published literature. The genes related with NSCLC were selected among DEGs of tumors in 4 tumor stages based on the information from the CTD.

Cell culture

Human lung epithelial cells (BEAS-2B), NSCLC cell lines such as A549 (adenocarcinoma), NCI-H1299 (large cell carcinoma) and NCI-H1650 (adenocarcinoma) were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). BEAS-2B and A549 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% 100 U/ml penicillin and 0.1 mg/ml streptomycin. NCI-H1299 and NCI-H1650 cells were cultured in RPMI-1640 medium supplemented with 10% FBS and 1% 100 U/ml penicillin, and 0.1 mg/ml streptomycin. All the cells were incubated in an atmosphere of 95% air and 5% CO2 at 37°C.

Reverse transcription real-time quantitative polymerase chain reaction (RT-qPCR)

The total RNA of the BEAS-2B, A549, NCI-H1299 and NCI-H1650 cells was extracted using the RNAiso Plus reagent (Takara Bio) according to the manufacturer's instructions. cDNA was reversely transcribed by PrimeScript (according to Takara Bio) at 37°C for 15 min, followed by 5 sec at 85°C. In order to determine the DNA copy number variations of the candidate natural killer-tumor recognition sequence (NKTR) gene, PCR amplifications were performed with 10 µl SYBR Premix Ex Taq (2X), 1 µl of 10 µM forward primers, 1 µl of 10 µM reverse primers and 8 µl cDNA, in a final volume of 20 µl. The relative quantities of NKTR expression were analyzed using the comparative Ct method and were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression. The primer sequences for NKTR and GAPDH are listed in Table I.

Table I. The primer sequences for NKTR and GAPDH.

Table I.

The primer sequences for NKTR and GAPDH.

Gene	Primer sequence (5′-3′)
GAPDH	F: TGACAACTTTGGTATCGTGGAAGG
	R: AGGCAGGGATGATGTTCTGGAGAG
NKTR	F: ACAGTTACCACCGAGGCAGA
	R: TCGGTCACTTTCACTGTCAGAGC

[i] NKTR, natural killer-tumor recognition sequence; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; F, forward; R, reverse.

Results

DEG identification

Compared with normal tissues, the genes with altered expression in NSCLC tumors were identified (Table II). The total number of DEGs in tumor tissues of stage IB, IIB, IIIA and IV were 499, 602, 592 and 457, respectively. Then, the DEGs for different tumor stages were subjected to hierarchical clustering analysis based on the gene expression patterns. The heat maps illustrated that all the DEGs were able to distinguish tumor and normal tissues clearly, which suggested that the DEGs identified in the present study were significant (Fig. 1).

Figure 1. Hierarchical clustering analysis of DEGs in tumor of stages IB, IIB, IIIA and IV. Red, high expression level; green, low expression level. DEGs, differentially expressed genes.

Table II. The top 10 DEGs identified in tumor tissues of different stages.

Table II.

The top 10 DEGs identified in tumor tissues of different stages.

IB vs. normal (n=499)			IIB vs. normal (n=602)			IIIA vs. normal (n=592)			IV vs. normal (n=457)
Genes	logFC	P-value	Genes	logFC	P-value	Genes	logFC	P-value	Genes	logFC	P-value
AMDHD1	−0.980253	2.03E-07	EVI5L	2.7955747	6.31E-06	LANCL1	−1.315286	4.62E-08	LAPTM4B	−0.755981	5.68E-05
OR5L2	−1.526492	4.07E-06	DPPA2	2.2309604	3.24E-05	RNF121	−3.629547	8.91E-08	CEP135	−1.06584	3.22E-05
POLR2D	−0.808397	5.45E-05	FXYD6	2.1874347	5.23E-05	MKKS	−0.939145	7.80E-06	C19orf56	1.1128544	0.0006198
LRIT3	−0.815737	3.93E-05	EIF2B3	1.5135333	4.10E-05	GPR177	−0.706376	3.25E-05	COX11	−0.582135	0.0006956
SLCO3A1	−0.949987	6.49E-05	PERLD1	−0.989765	4.45E-05	CISH	−0.749059	3.32E-05	DOK4	−0.5833	0.0002439
C3orf1	−1.037834	5.75E-05	CCDC65	−1.024191	4.69E-05	VGF	−0.993612	7.68E-05	CDX4	−0.598101	0.00057
PKM2	−0.785317	0.000128	NUP155	−1.545052	6.07E-05	ENO1	−2.226183	8.59E-05	ANK2	−0.67669	0.0004388
RAB23	−0.674876	0.0001587	ATP6V0A1	−1.035911	0.0001129	PSME1	−0.758268	0.000161	GEMIN7	−0.686458	0.0006848
LANCL2	−0.624205	0.0003193	PCDH18	−1.309749	0.0001096	JMJD2D	−0.797516	0.0001714	CNTN6	−0.784264	0.0005795
CHRNA10	−1.224662	0.0003353	BST2	2.3663933	0.000138	RAMP1	−0.802102	0.0001188	EIF5	−0.895799	0.0007118

[i] The DEGs were ranked according to the P-values. DEGs, differentially expressed genes; n, the total number of DEGs; FC, fold-change.

Significant GO function and pathways

The overrepresented GO terms and pathways for DEGs varied according to tumor stage. The DEGs in tumors of IB tumor stage were closely related with negative regulation of signal transduction and cellular di- and trivalent inorganic cation homeostasis biological process and the apoptosis pathway. The genes altered in stage IIB were significantly enriched in genitalia development, the monosaccharide metabolic process and the p53 signaling pathway. The upregulated DEGs in tumors of stages IIIA and IV were mainly related with transcription and synapse regulation-related biological process (Table III).

Table III. The overrepresented GO terms and pathways for the DEGs.

Table III.

The overrepresented GO terms and pathways for the DEGs.

Stages	Regulated	Category	Term	Count	P-value
IB	Upregulated	BP	GO:0009968~negative regulation of signal transduction	9	1.25E-03
			GO:0010648~negative regulation of cell communication	9	2.58E-03
			GO:0007128~meiotic prophase I	3	4.69E-03
		Pathway	hsa04210:Apoptosis	4	4.57E-02
	Downregulated	BP	GO:0030005~cellular di- and trivalent inorganic cation homeostasis	12	6.93E-04
			GO:0055066~di- and trivalent inorganic cation homeostasis	12	1.05E-03
			GO:0007204~elevation of cytosolic calcium ion concentration	8	1.41E-03
IIB	Upregulated	BP	GO:0048806~genitalia development	3	3.52E-02
			GO:0051054~positive regulation of the DNA metabolic process	4	3.58E-02
		Pathway	hsa04115:p53 signaling pathway	4	4.75E-02
	Downregulated	BP	GO:0005996~monosaccharide metabolic process	11	7.91E-03
			GO:0019318~hexose metabolic process	10	8.99E-03
			GO:0006006~glucose metabolic process	8	2.27E-02
IIIA	Upregulated	BP	GO:0010552~positive regulation of specific transcription from RNA polymerase II promoter	5	4.89E-03
			GO:0006350~transcription	39	5.68E-03
			GO:0016477~cell migration	10	6.27E-03
	Downregulated	BP	GO:0019320~hexose catabolic process	8	3.74E-04
			GO:0010033~response to organic substances	29	3.78E-04
			GO:0006091~generation of precursor metabolites and energy	17	4.20E-04
IV	Upregulated	BP	GO:0050807~regulation of synapse organization	4	1.81E-03
			GO:0050803~regulation of synapse structure and activity	4	2.69E-03
			GO:0010033~response to organic substances	18	4.82E-03
	Downregulated	BP	GO:0050892~intestinal absorption	3	1.41E-02
			GO:0035036~sperm-egg recognition	3	1.58E-02
			GO:0007339~binding of sperm to zona pellucida	3	1.58E-02

[i] DEGs, differentially expressed genes; GO, Gene Ontology; BP, biological process.

PPI network construction

The PPI network for DEGs in tumors of different stages was constructed respectively. The PPI network for stage IB was composed of 276 nodes and 454 pairs. The network for stage IIB included 349 nodes and 615 protein pairs. In addition, there were 366 nodes and 944 pairs for stage IIIA tumors and 252 nodes and 392 pairs for stage IV (Fig. 2).

Figure 2. PPI network for DEGs in tumors of different stages. (A) PPI network for DEGs in tumors of stage IB. (B) PPI network for DEGs in tumors of stage IIB. Red, upregulated genes; green, downregulated genes; blue, key nodes. PPI, protein-protein interaction; DEGs, differentially expressed genes. PPI network for DEGs in tumors of different stages. (C) PPI network for DEGs in tumors of stage IIIA. (D) PPI network for DEGs in tumors of stage IV. Red, upregulated genes; green, downregulated genes; blue, key nodes. PPI, protein-protein interaction; DEGs, differentially expressed genes.

Identification of the key nodes in the PPI network

Key nodes in the PPI network were identified by integration analysis of the scores calculated based on network centrality and the results of the clustering analysis. There were 29 key nodes in the stage IB group, 33 in the IIB group, 35 in the IIIA group and 54 in the IV group (Fig. 2). All the key nodes identified in the present study were differentially expressed and through their expression patterns we were able to distinguish normal and tumor tissues (Fig. 3). The key nodes were significantly enriched in cancer-related pathways, chronic myeloid leukemia, the chemokine signaling pathway, gap junctions and the insulin signaling pathway (Table IV).

Figure 3. Hierarchical clustering analysis of the key nodes in the PPI network. PPI, protein-protein interaction.

Table IV. The significant pathways for key nodes in the PPI network.

Table IV.

The significant pathways for key nodes in the PPI network.

Stages	Term	P-value	Genes
IB	hsa05220:Chronic myeloid leukemia	2.93E-03	STAT5A, NFKBIA, RAF1, MDM2
	hsa05215:Prostate cancer	4.76E-03	PDGFRA, NFKBIA, RAF1, MDM2
	hsa04060:Cytokine-cytokine receptor interaction	1.74E-02	CCR9, PDGFRA, CXCL9, PRL, IL2
	hsa05214:Glioma	2.49E-02	PDGFRA, RAF1, MDM2
	hsa05218:Melanoma	3.11E-02	PDGFRA, RAF1, MDM2
	hsa04062:Chemokine signaling pathway	3.52E-02	CCR9, CXCL9, NFKBIA, RAF1
	hsa05200:Pathways in cancer	3.62E-02	STAT5A, PDGFRA, NFKBIA, RAF1, MDM2
	hsa04540:Gap junction	4.69E-02	CSNK1D, PDGFRA, RAF1
IIB	hsa04270:Vascular smooth muscle contraction	1.50E-02	PPP1CA, MAPK3, ADCY6, PRKCE
	hsa04930:Type II diabetes mellitus	2.03E-02	MAPK3, HK1, PRKCE
	hsa04910:Insulin signaling pathway	2.46E-02	PRKAR2B, PPP1CA, MAPK3, HK1
IV	hsa04540:Gap junction	7.13E-03	EGFR, ADCY2, TUBB2A, PRKG2

[i] PPI, protein-protein interaction.

Venn diagram of DEGs

As shown in Fig. 4, the number of genes that were differentially expressed at each tumor stage was assessed and the diagrams also indicated the common DEGs in tumors of stages IB, IIB, IIIA and IV. Most DEGs were uniquely identified in a single tumor stage and a few genes were commonly differentially expressed in the 4 tumor stages, such as downregulated CNTN6 and LBX2.

Figure 4. Venn diagram for DEGs in tumors of different stages. DEGs, differentially expressed genes.

Identification of the coexpression gene pairs

A total of 1,984 genes were obtained by pooling the DEGs from different tumor stages, among which the coexpression gene pairs were analyzed. The coexpression network was constructed with 102 nodes and 179 gene interaction pairs (Fig. 5). Among the 102 nodes, 16 genes were involved in 65 gene pairs, such as SEMA4C, C19orf56, MYBBP1A, SPTA1, NKTR, WHSC1L1, FARP2, CDH18, C5orf33, ZNF124, ESRRG, ARHGEF10L, RABGAP1, ZMYM1, PIP5K1B and PGBD3.

Figure 5. Coexpression gene network for DEGs in tumors of different stages. Full line, positive correlation; dashed line, negative correlation. DEGs, differentially expressed genes.

NSCLC-related gene identification

According to the CTD database, most of the genes were related with NSCLC during different stages. There were 372 genes (74.55%) among all the DEGs in the IB stage group, 465 (77.24%) for the IIB group, 466 (78.72%) in the IIIA group and 353 (77.24%) for the IV group (Table V).

Table V. NSCLC-related genes in tumors of different stages.

Table V.

NSCLC-related genes in tumors of different stages.

Groups	Total DEG count	Disease related gene count	Percentage
IB	499	372	74.55
IIB	602	465	77.24
IIIA	592	466	78.72
IV	457	353	77.24

[i] NSCLC, non-small cell lung cancer; DEGs, differentially expressed genes.

mRNA expression of NKTR

RT-PCR analysis of NKTR in A549, NCI-H1299 and NCI-H1650 cells was conducted to establish the differential expression in NSCLC, compared with the BEAS-2B cells. As shown in Fig. 6, the mean mRNA expression levels of NKTR in A549, NCI-H1299 and NCI-H1650 cells were significantly higher than those in the BEAS-2B cells (P<0.01), which corresponded to the DEG analysis and further confirmed its significance in NSCLC.

Figure 6. RT-PCR validation of the relative expression of NKTR in NSCLC cell lines. **P<0.01, compared with the control.

Discussion

In the present study, the gene expression profile analysis of NSCLC tissues of different stages revealed the significantly altered biological functions and pathways and proposed some candidate genes for lung cancer treatment.

The data (GSE21933) analyzed in the study were the cDNA microarray data. Both DNA and RNA microarray data can be used to analyze the expression levels of genes by limma software (1,2). The DEG analysis showed that there were 499, 602, 592 and 457 genes differentially expressed in tumors of stages IB, IIB, IIIA and IV, respectively. Clustering analysis revealed that based on the expression level of DEGs, the tumor and normal samples were clearly distinguished. In addition, literature research analysis confirmed that most of the DEGs identified in our study were NSCLC-related genes. These findings suggested that the DEGs established in our study were significant.

Function analysis showed that the significantly altered biological process in the early lung cancer stage mainly included negative regulation of signal transduction, apoptosis, cellular di- and trivalent inorganic cation homeostasis (stage IB) and positive regulation of the DNA metabolic process and the p53 signaling pathway (stage IIB). Previous evidence has shown that lung cancer-related candidate genes are mainly related with apoptosis, signal transduction and DNA repair. The ion transport is a significantly altered biological process for both the Asian and Caucasian population, which may play a key role in tumorigenesis (8). p53 gene is a tumor-suppressor gene and is reported to be frequently mutated in lung cancer patients (22). Non-disruptive p53 mutations may decrease the survival rate in patients with advanced NSCLC (23). Signal transduction induced by focal adhesion kinase is found to inhibit the p53-mediated apoptosis in malignant cells (24).

In the late-stage NSCLC tumors, the DEGs were closely related with cell migration, response to organic substances (stage IIIA) and regulation of synapse organization, structure and activity (stage IV). Cell migration has been reported to be one of the significantly altered biological processes in NSCLC cancer patients (8), which is consistent with our findings. The biological process of response to organic substances was significantly altered in stage IIIA and IV NSCLC tumors, indicating that this biological process may play a dominant role in the late stages of lung cancer. The role of the synapse regulation-related biological process has not been clarified in lung cancer progression. Thus, the altered biological processes warrant further study. Above all, the biological processes and pathways in NSCLC progression may be altered in a stage-dependent manner. The DEG enriched biological processes and pathways were different in early-stage and late-stage NSCLC, which was determined by Venn diagram analysis. The Venn diagram illustrated that a few genes had similar expression patterns in NSCLC at different stages, which revealed that NSCLC at different stages induced differential gene expression patterns leading to the dysfunction of various biological processes.

Additionally, after removing the overlapped genes, a total of 1,984 DEGs were obtained in tumors of stages IB, IIB, IIIA and IV. The coexpression network composed of 16 genes and 65 coexpression pairs was the most significant. The protein interaction pairs corresponding to DEGs can be predicted by integrating the information deposited in STRING (13). Cytoscape is a free software package and widely used in analyzing the results of mRNA expression profiling, functional genomic and proteomic experiments in the context of an interaction network obtained for genes of interest (14). Thus, protein interactions were able to be obtained based on the microarray data in the study. NKTR was one of the genes among the 16 genes that showed multiple coexpression interactions with others. Simultaneously, PPI analysis showed that NKTR was a key node in stage IV tumor tissues according to degree, betweenness and subgraph centrality. In addition, according to expression patterns of NKTR, tumor and normal tissues could be clearly distinguished. RT-PCR of gene validation analysis showed that NKTR was significantly upregulated in NSCLC cells, compared with normal epithelial cells.

All these findings indicated that NKTR is a significant DEG that plays a critical role in NSCLC progression. NKTR has been found to be differentially expressed in human lung squamous cell carcinoma and is located in chromosome 3p23-p21 according to the cDNA library (25). NKTR as a component of a spliceosome, has functions in protein folding. NKTR-102 is considered as a novel PEGylated-irinotecan conjugate, which functions in the inhibition of the growth of human colorectal and lung tumors (26). Although the role of NKTR in NSCLC has not been clarified, NKTR and its coexpressed genes may play a key role in NSCLC progression.

In this study, the microarray data of GSE21933 were developed from NSCLC patients including 10 early-stage patients (6 stage IB, 3 stage IIB and 1 stage IA) and 11 late-stage patients (5 stage IIIA, 5 stage IV and 1 stage IIIB). Due to the small number of samples at stage IA and IIIB, we only analyzed the remaining 19 tumor samples compared with the corresponding controls. In the original study, the microarray data of GSE21276 were produced from tumor and normal tissues of 40 Asian and 20 Caucasian NSLCL patients and was aimed to characterize the gene expression profiles in different racial groups. However, in the present study, we attempted to analyze the changes in gene expression profiles during the different tumor stages. The detailed information of the 19 patients was not provided in the original study, thus we cannot provide information on the type of tumors. In our present study, we performed gene validation experiments in A549 (adenocarcinoma), NCI-H1299 (large-cell carcinoma) and NCI-H1650 (adenocarcinoma) cells, and determined that NKTR was a candidate gene in NSLCL prevention and treatment.

Despite some previous microarray data that were also developed from a small number of samples (27,28), the small sample size used may be a limitation in our study. Further studies with a larger size of samples are warranted in the near future.

In conclusion, the gene expression patterns are different at various stages of NSCLC tumors, from early-stage to late-stage, which lead to the dysfunction of different biological processes. Different mechanisms may be involved in the progression of NSCLC tumors from early-stage to late-stage. An adequate understanding of the pathogenesis underlying different tumor stages may be helpful in new therapy development. NKTR is commonly differentially expressed in stage IB, IIB, IIIA and IV tumors. NKTR and its coexpressed gene pairs may be candidate genes for the prevention and treatment of NSCLC.

Glossary

Abbreviations

Abbreviations:

References