Abstract

   Lung cancer is one of the most invasive cancers affecting over a
   million of the population. Non-small cell lung cancer (NSCLC)
   constitutes up to 85% of all lung cancer cases, and therefore, it is
   essential to identify predictive biomarkers of NSCLC for therapeutic
   purposes. Here we use a network theoretical approach to investigate the
   complex behavior of the NSCLC gene-regulatory interactions. We have
   used eight NSCLC microarray datasets [46]GSE19188, [47]GSE118370,
   [48]GSE10072, [49]GSE101929, [50]GSE7670, [51]GSE33532, [52]GSE31547,
   and [53]GSE31210 and meta-analyzed them to find differentially
   expressed genes (DEGs) and further constructed a protein–protein
   interaction (PPI) network. We analyzed its topological properties and
   identified significant modules of the PPI network using cytoscape
   network analyzer and MCODE plug-in. From the PPI network, top ten genes
   of each of the six topological properties like closeness centrality,
   maximal clique centrality (MCC), Maximum Neighborhood Component (MNC),
   radiality, EPC (Edge Percolated Component) and bottleneck were
   considered for key regulator identification. We further compared them
   with top ten hub genes (those with the highest degrees) to find key
   regulator (KR) genes. We found that two genes, CDK1 and HSP90AA1, were
   common in the analysis suggesting a significant regulatory role of CDK1
   and HSP90AA1 in non-small cell lung cancer. Our study using a network
   theoretical approach, as a summary, suggests CDK1 and HSP90AA1 as key
   regulator genes in complex NSCLC network.

   Keywords: non-small cell lung cancer, key regulator, differentially
   expressed genes, protein–protein interaction

1. Introduction

   Lung Cancer is one of the most invasive cancer types causing more than
   1.38 million deaths worldwide [[54]1]. Non-Small Cell Lung Cancer
   (NSCLC) is a kind of epithelial lung cancer. It is markedly different
   from Small Cell Lung Carcinoma (SCLC), and accounts for about 85% of
   all lung cancer cases [[55]2]. NSCLCs can be further sub-categorized
   into adenocarcinoma (32–40%), squamous cell carcinoma (25–30%), and
   large cell carcinoma (8–16%) based on the type of lung cells undergoing
   uncontrolled proliferation [[56]3]. In adenocarcinomas, cells that
   secrete mucus proliferate uncontrollably. This type of cancer occurs
   mainly among non-smokers [[57]4]. It is more common in women than in
   men and has a higher probability of occurring in younger people, among
   other types of lung cancers. It is usually found in the outer parts of
   the lung; thus, it is likely to be detected before it metastasizes
   [[58]5].

   On the other hand, in squamous cell carcinoma, flat squamous cells that
   line the inside of the airways in the lungs undergo proliferation.
   Here, the tumor is usually found in the central part of the lungs, near
   the bronchi. In large cell carcinoma, the tumor can grow in any part of
   the lung [[59]6]. It grows and spreads very fast, making it harder to
   treat. Other subtypes of NSCLCs, such as adeno squamous carcinoma and
   sarcomatoid carcinoma, are relatively less common.

   Out of various reasons, smoking history is highly correlated to the
   pathogenesis of lung cancer [[60]7]. It is, so far, the most crucial
   risk factor for lung cancer. Cigarette smoke contains more than 6000
   components, most of which result in DNA damage [[61]7,[62]8]. Moreover,
   other sources of lung cancer include subjection to passive smoke,
   radon, exposure to materials such as soot, beryllium, nickel, chromium,
   asbestos, or tar, genetic predisposition to lung cancer, and air
   pollution [[63]9,[64]10].

   DNA damage appears to be the primary reason for pathogenesis in
   multiple different cancers at the genomic level, let alone NSCLC cases.
   Even though the DNA repair system can repair most of the damage,
   frequent exposure to smoke and other causative factors makes it
   vulnerable [[65]11]. Additionally, epigenetic gene silencing of DNA
   repair genes can happen during incompletely finished repair sites
   formed during DNA double-strand breaks or other DNA damage repair
   [[66]12,[67]13]. The epigenetic gene silencing of DNA repair genes
   plays a crucial role in molecular pathogenesis of NSCLC. At minimum,
   nine DNA repair genes that usually function in DNA repair pathways are
   often suppressed by promoter hypermethylation in NSCLC. They are namely
   NEIL1, WRN, MGMT, ATM, MLH1, MSH2, BRCA1, BRCA2, and XRCC5
   [[68]14,[69]15,[70]16,[71]17,[72]18,[73]19,[74]20]. FEN1, a component
   of the DNA repair pathway, is expressed at an increased level due to
   hypomethylation at its promoter region in NSCLC [[75]21]. The complex
   molecular underpinning of NSCLC is yet to be fully understood for
   better treatment approaches and drug target identification.

   There are not many full-proof treatment options available. Current
   treatment options include surgical intervention at the early stages
   like radical mastectomy. Cisplatin-based chemotherapy coupled with
   radiotherapy has been the go-to treatment with the advancement in the
   cancer stages and onset of metastasis. However, the multimodal approach
   is taken in certain cases, where third generation cytotoxic and
   cytostatic agents such as anti-VGFR and anti-EGFR drugs are prescribed
   [[76]22].

   Thus, further research into recognizing new players, drugs, and
   combinatorial therapies is necessary to expand the clinical interest to
   a broader patient population and better outcomes in NSCLC. This study
   utilizes a network theoretical approach to unveil the complexity of
   non-small cell lung cancer.

2. Materials and Methods

2.1. Data Collection

   Non-small cell lung cancer data sets were retrieved from Gene
   Expression Omnibus (GEO), NCBI (GEO,
   [77]https://www.ncbi.nlm.nih.gov/geo/, accessed on 13 September 2020).
   A total of eight datasets, [78]GSE19188 (65 control and 91 cancer
   patients’ samples), [79]GSE118370 (six control and six cancer patients’
   samples), [80]GSE10072 (49 control and 58 cancer patients’ samples),
   [81]GSE101929 (34 control and 32 cancer patients’ samples), [82]GSE7670
   (27 control and 27 cancer patients’ samples), [83]GSE33532 (20 control
   and 80 cancer patients’ samples), [84]GSE31547 (20 control and 30
   cancer patients’ samples), [85]GSE31210 (20 control and 226 cancer
   patients’ samples), were taken into this study. The description of
   datasets and controls and patient numbers are described in [86]Table 1.

Table 1.

   List of datasets used in the meta-analysis.
   Microarray Datasets  Platforms  Control Cases    References