Genetic Features of Lung Adenocarcinoma with Ground- Glass Opacity: What Causes the Invasiveness of Lung Adenocarcinoma?

BACKGROUND: Lung adenocarcinoma (LUAD) with ground-glass opacity (GGO) can become aggravated, but the reasons for this aggravation are not fully understood. The goal of this study was to analyze the genetic features and causes of progression of GGO LUAD.
METHODS: LUAD tumor samples and normal tissues were analyzed using an Illumina HiSeq 4000 system. After the tumor mutational burden (TMB) was calculated, the identified mutations were classified as those found only in GGO LUAD, those present only in non- GGO LUAD, and those common to both tissue types. Ten high-frequency genes were selected from each domain, after which protein interaction network analysis was conducted.
RESULTS: Overall, 227 mutations in GGO LUAD, 212 in non-GGO LUAD, and 48 that were common to both tumor types were found. The TMB was 8.8 in GGO and 7.8 in non-GGO samples. In GGO LUAD, mutations of FCGBP and SFTPA1 were identified. FOXQ1, IRF5, and MAGEC1 mutations were common to both types, and CDC27 and NOTCH4 mutations were identified in the non-GGO LUAD. Protein interaction network analysis indicated that IRF5 (common to both tissue types) and CDC27 (found in the non-GGO LUAD) had significant biological functions related to the cell cycle and proliferation.
CONCLUSION: In conclusion, GGO LUAD exhibited a higher TMB than non-GGO LUAD. No clinically meaningful mutations were found to be specific to GGO LUAD, but mutations involved in the epithelial-mesenchymal transition or cell cycle were found in both tumor types and in non-GGO tissue alone. These findings could explain the non-invasiveness of GGO-type LUAD.

Introduction

Advances in imaging technology have increased the probability of the early detection of lung cancer, especially lung adenocarcinoma (LUAD) with ground-glass opacity (GGO) [1,2]. This type of adenocarcinoma has an excellent survival rate compared to non-GGO adenocarcinomas [3,4]. GGO-type LUAD can be divided into pure and mixed GGO. Pure GGO (which lacks regions of consolidation) is likely to be adenocarcinoma in situ (AIS) or minimally invasive adenocarcinoma (MIA), while mixed GGO (which exhibits areas of consolidation) is likely to be MIA, lepidic predominant invasive adenocarcinoma, or a more invasive type of adenocarcinoma. Therefore, pure GGO LUAD can be managed through sublobar resection, but mixed GGO requires lobectomy, as the solid portion indicates invasiveness [3,5]. In other words, GGO is an important indicator of the extent of resection required and of the patient’s prognosis. Moreover, better knowledge of this phenomenon could facilitate a deeper understanding of tumorigenesis in LUAD, as GGO adenocarcinoma is thought to be the precursor to invasive LUAD. However, because LUAD follows a stepwise progression, GGO-type LUAD cannot be analyzed in isolation [6].

LUAD is known to develop from atypical adenomatous hyperplasia, AIS, MIA, and invasive adenocarcinoma [6,7]. However, the causes of the development of AIS or MIA into an invasive form have not been determined. Many genetic factors, including p16 gene inactivation and mutations of epidermal growth factor receptor (EGFR), Kirsten rat sarcoma 2 viral oncogene homolog (KRAS), MNNG HOS transforming gene (MET), v-Raf murine sarcoma viral oncogene homolog B (BRAF), and tumor protein p53 (TP53), are known to be involved in the development of LUAD, but the detailed molecular events involved in each step have not been identified. Moreover, so-called driver mutations, including those of EGFR, KRAS, and TP53, have been identified frequently in the study of LUAD, but a significant number of LUAD patients have been found to lack these mutations [8,9].

Therefore, the purpose of this study was to investigate the genetic features of pure and mixed GGO LUAD in comparison to those of normal tissue through whole-exome sequencing (WES) of paired tissue (tumor and normal) DNA and to compare the data with non-GGO LUAD.

Methods

Ground-glass opacity and patients

Both pure GGO (which lacks a consolidation component) and mixed GGO (which contains both pure GGO and a consolidated region) were included in the study. Preoperative computed tomography images were reviewed by a single radiologist who was blinded to the pathological results and who classified the areas regarding whether they constituted GGO. All specimens were evaluated microscopically by a single pathologist who was blinded to the clinical and radiological data. Comprehensive histological subtyping was performed according to the eighth edition of the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification system [10].

Patients with clinical T1N0 LUAD with GGO were included in the study. The control groups were patients of the same clinical stage (T1N0) who lacked GGO and exhibited consolidation only. Patients were excluded if they had clinical T2–4 cancer, N1–3 cancer, or any intrathoracic or extrathoracic metastases. Patients receiving neoadjuvant treatment and those with non-adenocarcinoma lung cancer were also excluded. The Institutional Review Board of Chungbuk National University Hospital approved this study (IRB approval no., 2017-04-018-004).

Whole-exome sequencing

Tumor and normal tissues were obtained by the surgeon in the operating room to guarantee the freshness of the tissue. The normal tissue samples were collected as far as possible from the tumor. Then, the tissues were confirmed by the pathologist to be tumor or normal cells. DNA was extracted using a Blood & Cell Culture DNA Mini Kit according the manufacturer’s instructions (Qiagen, Hilden, Germany). It was quantitatively analyzed using a NanoDrop apparatus (Thermo Fisher Scientific, Waltham, MA, USA), and the integrity was assessed using 2% agarose electrophoresis with SYBR Gold (Thermo Fisher Scientific) staining. The genomic DNA of the tumors and matched normal tissues from all patients was captured using Agilent’s in-solution enrichment methodology (SureSelect XT V5; Agilent, Santa Clara, CA, USA) with a biotinylated oligonucleotide probe library (SureSelect XT V5, 51 Mb; Agilent) followed by paired-end 101-base massively parallel sequencing on an Illumina HiSeq 4000 machine (Illumina, San Diego, CA, USA) [11].

Bioinformatics

The raw sequencing data (in the BCL file format) were converted to FASTQ files using Casava software (Illumina). Then, the Burrows-Wheeler alignment tool was utilized to map short reads to the human genome reference sequence (GRCh37). Duplicate read removal was performed using Picard (Broad Institute, Cambridge, MA, USA) and Samtools (Genome Research Limited, London, UK), and recalibration and variant calling were then performed using the Genome Analysis Toolkit (https://gatk.broadinstitute.org/hc/en-us). Variant annotations and effect predictions were subsequently performed with the SnpEff tool (https://pcingola.github.io/SnpEff/index.html).

The variations were identified by comparing the exome sequencing data of the tumor sample to the reference genome sequence derived from the paired normal tissue. Next, a Venn diagram summary of the unique variations was created to reveal any discrepancies in variations between GGO LUAD and the normal tissue and between non-GGO LUAD and the normal tissue in the paired samples [12]. The variations were filtered with regard to the consequence to the protein: start/stop gain/loss, splice, in-frame insertion/deletion, and frameshift variants. Using WES, the tumor mutational burden (TMB) was calculated based on the identified mutations in both tumor and germline DNA [13]. We focused on nonsynonymous mutations among the tumor-specific functional mutations in The Cancer Genome Atlas cohort, since synonymous mutations could not be directly involved in creating neoantigens. Each mutation was classified into one of 3 domains: GGO-specific, non-GGO–specific, and common to both types. After the top 10 genes in each domain were selected, protein interaction network analysis was conducted with the Search Tool for the Retrieval of Interacting Genes/Proteins ver. 10 database (PMID: 25352553) using the Molecular Complex Detection plugin in Cytoscape 3.7.0 (The Cytoscape Consortium, San Diego, CA, USA), which is a visualization tool for integrating many molecular states, such as the expression level and interaction information, into a unified conceptual framework (PMID: 14597658). Finally, we examined the frequency of EGFR, KRAS, and TP53 in each domain. Based on the results, we sequenced each mutation in the stepwise progression of LUAD to estimate its clinical significance.

Results

Patients and tumor mutational burden

In total, 20 samples were analyzed in this study. Of them, GGO (n=2 for the pure type and n=3 for the mixed type) and non-GGO LUAD were each present in 5 patients. These included 7 female patients, 4 of whom had GGO LUAD. No smokers were included in the current study, but 3 patients were ex-smokers (1 of whom had GGO LUAD). The mean size of the mass among the GGO cases was 10 mm, while that among the non-GGO cases was 15 mm. The GGO group consisted of 4 AIS and 1 MIA, while the non-GGO group consisted of 4 well-differentiated and 1 poorly-differentiated LUAD.

Tumors (n=10) and paired normal tissue (n=10) were used for analysis. In this study, 27,466 variations were identified in GGO and 27,376 in non-GGO. Of them, 227 functional variations were found in GGO, 212 in non-GGO, and 48 in common (Fig. 1). The Venn diagram summary showed that many discrepancies were present between the GGO and normal variations and between the non-GGO and normal variations in the paired samples. The TMB was 8.8 in the GGO group and 7.8 in the non-GGO group, with TMB defined as the number of mutations per megabase of DNA using the exome spans of coding sequences, which comprise less than 2% of the genome.

Fig. 1

Venn diagram of altered functional genes derived from whole-exome sequences. Many discrepancies were found among the GGO, non-GGO, and normal variations. The number of GGO-specific mutations was 227, compared to 212 in the non-GGO lung adenocarcinoma and 48 that were common to both tumor types, excluding normal variations and HLA-related genes. GGO, ground-glass opacity; HLA, human leukocyte antigen.

Gene alterations in each group

In the GGO domain, genes were selected as significant if they were found to have at least 2 mutations that were present in at least 2 samples. The top 10 most significant genes were FCGBP, ANKRD36, C90RF50, CENPV, IGSF3, KIR2DS4, LGALS9B, PRR21, SFTPA1, and YBX2 (Fig. 2A). Among them, FCGBP was the most frequently mutated (with 4 mutations and 3 affected samples). Protein interaction network analysis was performed, but no clinically important relationships were found in the GGO domain.

Fig. 2

Landscape of functional mutations in each domain. The top 10 mutations related to cellular functions were identified in each group. (A) FCGBP and ANKRD36 were the most frequent gene alterations in the GGO group. (B) OR2T34 and TCP10L2 were the most frequent gene alterations common to both groups, but FOXQ1, MAGEC1, and IFR5 have clinical implications in lung neoplasms. (C) CDC27 and NOTCH4 were the most frequent mutations in the non-GGO LUAD. GGO, ground-glass opacity; LUAD, lung adenocarcinoma; SNV, single nucleotide variant.

Among the mutant genes commonly expressed in GGO and non-GGO LUAD, the top 10 most significant genes were OR2T34, TCP10L2, forkhead box Q1 (FOXQ1), GBP4, interferon regulatory factor 5 (IRF5), melanoma-associated antigen C1 (MAGEC1), NUTM2F, PRR21, PRSS1, and PTPRT (Fig. 2B). Among them, OR2T34 was the most frequently-appearing gene (n=5, 50%), and TCP10L2 (n=4, 40%) was the next most frequent. Protein interaction network analysis showed that IRF5 and GBP4 were related to the cellular response to cytokine stimuli (GO: 0071345) and the type I interferon signaling pathway (GO: 0060337) (Fig. 3A).

Fig. 3

Protein interaction network analysis. (A) IRF5 (the most frequent mutation in common) and GBP4 showed variable interactions with cytokine stimuli. (B) CDC27 (the most frequent mutation in the non-GGO group) and ZNF717 were determined to be associated with anaphase-promoting complex-dependent catabolism, cell division, and the cell cycle. GGO, ground-glass opacity.

In the non-GGO domain, the top 10 most significant genes were cell division cycle 27 (CDC27), NOTCH4, CES1, FAM58A, FRG2C, IGSF3, MIR4273, NUTM2G, OR2T27, and ZNF717 (Fig. 2C). Among them, CDC27 and NOTCH4 were the most frequently identified genes (with 4 mutations and 2 affected samples). Based on protein interaction network analysis, CDC27 and ZNF717 were determined to be involved in anaphase-promoting complex-dependent catabolism (GO:0031145), cell division (GO: 0051301), and the cell cycle (GO:0007049) (Fig. 3B).

Although mutations of EGFR, TP53, and KRAS are clinically important, these genes were not present among the top 10 most common genes in the GGO or the non-GGO groups. Mutations of EGFR (n=2 in GGO and non-GGO; the mutation was identical) and TP53 (n=2 in GGO and n=1 in non-GGO; all mutations were different) were observed, but no mutation of KRAS was found (Fig. 4).

Fig. 4

Identified mutations and stepwise progression of lung adenocarcinoma. Mutations with clinical implications were identified according to the progression of lung adenocarcinoma. In the GGO step, mutations of FCGBP, SFTPA1, and TP53 were identified, but these mutations lacked clinical implications. The identified TP53 mutations were not identical between the GGO and non-GGO samples. IRF5, FOXQ1, MAGEC1, and EGFR were found to be mutated in both GGO and non-GGO samples; moreover, the mutations of EGFR were identical. In the non-GGO step, mutations of CDC27, NOTCH4, and TP53 were detected. GGO, ground-glass opacity; EMT, epithelial- mesenchymal transition.

Lung cancer progression according to genetic mutations

The identified mutations were arranged in the order of the stepwise progression of LUAD (Fig. 4). FCGBP and STFPA1 were positioned in the GGO step. The protein product of FCGBP (Fc fragment of immunoglobulin G binding protein) is known to be a key regulator of the transforming growth factor-1–induced epithelial-mesenchymal transition (EMT), and it has been found to be associated with cancer progression, biological behaviors, and prognosis in gallbladder cancer [14]. The product of SFTPA1 (surfactant, pulmonary-associated protein A1) plays a key role in innate lung host defense and gene regulation [15]. The expression levels of SFTPA1 have been found to be much higher in the lung than in other tissues; moreover, this gene was observed to be strikingly downregulated in lung tumor tissues relative to the adjacent non-tumor tissues [16].

Among the commonly mutated genes, IRF5, FOXQ1, and MAGEC1 were identified. IRF5 encodes a member of the interferon regulatory factor family. This is a group of transcription factors with diverse roles, including in cell growth, differentiation, apoptosis, and immune system activity. FOXQ1 is a member of the FOX gene family involved in embryonic development, cell cycle regulation, tissue- specific gene expression, cell signaling, and tumorigenesis. MAGEC1 is a member of the melanoma antigen gene family. The proteins in this family are tumor-specific antigens that can be recognized by autologous cytolytic T lymphocytes.

Among the non-GGO–specific mutations, CDC27 and NOTCH4 were investigated. CDC27 encodes a protein that is a component of the anaphase-promoting complex and has tetratricopeptide repeats, which are important for protein-protein interactions. This protein has been shown to interact with mitotic checkpoint proteins, and it may thus be involved in controlling the timing of mitosis. The Notch signaling pathway is an important form of cell-to-cell communication that plays a key role in regulating stem cell proliferation, differentiation, and apoptosis during embryonic development [17].

Discussion

This work produced 3 distinctive findings. First, the TMB was higher in the GGO than in the non-GGO tumor samples. A high TMB was found to be significantly associated with favorable outcomes in a previous study of 908 resected lung cancer tumor specimens, which were sequenced using a targeted panel of 1,538 genes [13]. It remains unclear why a high TMB is associated with better outcomes, but TMB has already been used in the patient selection process for a clinical trial of immune checkpoint inhibitors [18]. Further investigation of the TMB could elucidate the cause of the non-invasiveness of GGO. Second, driver mutations of TP53 and EGFR were found in the current study, but they did not predominate, and no regularity was observed according to the increase in malignancy from GGO to non-GGO. As stated in many previous studies, it is clear that the above genes play an important role in lung cancer, but their role in the tumorigenesis of GGO or in pathologic T1 LUAD seems to be limited. Third, as the degree of malignancy increased, the mutation frequency of genes involved in the cell cycle and apoptosis also increased. In the GGO-specific stage, FCGBP and SFTPA1 were identified, but they were not determined to be significant in lung cancer. However, among the mutations that were common to both tissue types or specific to the non-GGO stage, EMT- and tumor invasion-related genes were detected (Fig. 4).

All but the GGO-specific mutations had clinical significance. FCGBP and SFTPA1 were identified in the GGO domain, but no meaningful research was available regarding lung cancer and protein interaction network analysis. IRF5, FOXQ1, and MAGEC1, which were common to both the GGO and non-GGO tumors, were clinically notable. IRF5 was identified and validated as an EMT-regulating transcription factor in a study of the H358 non–small-cell lung cancer cell line [19]. In a study of FOXQ1 and 4 common EMT indicator proteins (E-cad, MUC1, VIM, and S100A4), increases in FOXQ1 expression were found to be correlated with changes in the expression of EMT indicators [20]. MAGEC1, which is normally expressed in male germ cells, is often expressed in tumors, including non–small-cell lung cancer; thus, it has been considered an attractive target for a cancer vaccine [21]. In the non-GGO samples, CDC27 and NOTCH4 were notable. CDC27 is also involved in the EMT and has been found to promote factors that induce metastasis and invasion, but these findings have been made in the contexts of colorectal and stomach cancer [22,23]. Regarding NOTCH4, a study of its expression found that it was positively associated with tumor size, lymph node metastasis, and distant metastasis, and it was an independent prognostic factor for overall survival in lung cancer patients [24].

Mutations in the EGFR and TP53 genes were found in the present study, but they were not predominant mutations in either the GGO or non-GGO groups. However, given that EGFR is expressed in both the GGO and non-GGO domains at the same time and that their mutations coincide completely, it is clear that the effect of EGFR on tumor progression is apparent in some GGO LUADs [25]. In addition, it should not be concluded that TP53 is meaningless simply because its mutation was not found to be predominant in this study. Despite this lack of predominance and the fact that TP53 does not fit into the stepwise progression schema in the study, it is reasonable to assume that this gene plays a specific role in both the GGO and non-GGO domains [26].

This study is unique in its use of WES in the investigation of GGO and counterpart tumors. Due to cost limitations, most previous studies have been panel studies of limited number of genes. WES could provide additional data regarding unbiased pathogenic sequence mutations, including single-nucleotide variations and small insertion-deletions. Moreover, the landscape of mutations associated with LUAD exhibits racial and ethnic differences. Therefore, it is urgent to build up a database of GGO LUAD from different populations. Although this study was done with a limited sample size, its results could be a stepping stone to identifying the tumorigenesis of LUAD in Asian people.

In conclusion, GGO LUAD exhibited a higher TMB than non-GGO LUAD. No clinically meaningful mutations were found to be specific to GGO LUAD, but frequent mutations involving the EMT or the cell cycle were found in common and in the non-GGO domain. EGFR and TP53 were found to have limited roles in this study, but further research is required. Overall, these findings could explain the non-invasiveness of GGO and may be useful in the search for targets for new medications.