1. Introduction

Lung cancer is known to be responsible for the highest cancer-related mortalities worldwide, and the same is true in Korea^[1]. Histopathologically, non-small cell lung cancer (NSCLC) is the most common type of lung cancer, comprising more than 85% of all lung cancers. More than 70% of patients cannot undergo successful surgical resection because of advanced diseases, such as stage IIIB and IV NSCLC, at initial diagnosis. Although our understanding of NSCLC, and the consequent development of advanced therapeutic strategies for improving the survival of NSCLC patients with further knowledge on the cell-signaling pathway involved, the treatment outcomes are still poor, with a 5-year survival rate of less than 10% in patients with locally advanced or metastatic disease^[2]. Patients with lung cancers frequently experience relapse, with metastasis of the disease, even in patients with complete resection of NSCLC. Recently, there have been several reports suggesting that new chemotherapeutic agents such as molecular-targeted therapy, including epidermal growth factor receptor (EGFR) inhibitors and tyrosine kinase inhibitors (TKIs), as well as immunotherapy including anti-PD1 and anti-PDL therapy, can improve the overall survival of the NSCLC patients. However, most patients ultimately develop drug resistance and relapse, despite dramatic initial responses to such treatments. As a result, a progressive disease usually metastasizes to other organs, including the brain, which can negatively affect patients. With brain metastasis (BM), neurological sequelae commonly prevent patients from performing routine daily activities. This decreased engagement in daily activities prevents patients from pursuing additional cancer treatment following the initial poor treatment response and consequent poor prognosis^[3]. BM is common in NSCLC patients and is present in 10–20% of NSCLC patients at diagnosis^[4]. Approximately 30–50% of NSCLC patients also develop BM during the course of their disease^[5]. Patients with BM commonly have a poor prognosis, and a median survival of just one to two months without treatment, mainly due to impaired performance status^[3].

The spreading and movement of cancer cells are complicated processes. Among these steps, the epithelial-mesenchymal transition (EMT) is the most important step in the progression of cancers to metastasis and invasion. In cancer cells, which are separated from the original mass of cancerous cells, the process of EMT is activated, facilitating intravasation into the bloodstream^[6][7][8][9]. The process of EMT can be aggravated or suppressed at individual steps of genetic or epigenetic expression. The transcriptional drivers and suppressors of the EMT are affected by epigenetic changes such as histone modification and/or the activities of multiple miRNAs^[9][10]. Several EMT-associated transcription factors (EMT-TFs) previously studied focused on epigenetic regulation by histone methylation or demethylation^{[11][12][13][14]}. However, these studies have not been performed using human samples, but in vitro, utilizing a culture of stem cells, or in vivo, using a mouse model^[15][16], and there are few comprehensive analyses of EMT-TFs that can be regulated by histone modification in human cancer samples. Thus, there are frequent debates as to whether the EMT of cancer cells is truly relevant to the human body, as shown in vivo. EMT is one of the major fields of cancer research; however, the association of the EMT in cancer progression with the outcomes of the treatment has not yet been established. There are many limitations associated with the application of modification of the EMT in clinical practice because an understanding of the role of the EMT in the diagnosis and prediction of therapeutic outcome is still incomplete, despite numerous studies showing that the EMT plays a role in cancer progression. There are many discrepancies in vitro data from real patients originating from the intrinsic heterogeneity of cancer cells and their microenvironments.

Epigenetic events are defined as stable changes in gene expression occurring after transcription, without the alteration of DNA sequences. Epigenetic modifications are thought to play a role in many aspects of health and disease, including cancer biology. There are three major types of epigenetic events; histone modification, RNA interference, and DNA methylation. Histone modification is a major part of the mechanism that post-transcriptionally regulates gene expression. This modification can involve the methylation, acetylation, ubiquitylation, or phosphorylation of histone proteins. The methylation of histone proteins usually occurs at the side chains of arginine and lysine residues within the N-terminal region. Several enzymes regulating the methylation status of histone H3 lysine residues have been reported to be of pathological and clinical significance in several cancers^{[17][18][19][20][21]}. However, there are few studies focused on the EMT process and investigating the unique functions of epigenetic alterations at histone H3 lysine residues and their relationship with cancer progression.

2. Epigenetic Role of Histone Lysine Modification Enzymes

Although unique histone lysine modification enzymes did not influence the survival of patients directly, several enzymes such as MLL4 (H3K4 methyltransferase), UTX (H3K27me3 demethylase), and EZH2 (H3K27 methyltransferase) regulate the expression of specific EMT-TFs; these EMT-TFs are associated with survival of patients with lung adenocarcinoma via epigenetic regulation. To date, several studies have demonstrated the epigenetic regulation of histone methylation or demethylation of EMT-TFs. However, these studies have not been performed on human samples, and there are few comprehensive analyses of EMT-TFs that can be regulated by histone modification in human cancer samples. For example, in breast cancer, there are several studies that have shown the epigenetic role of histone modification of EMT-TF expression in the metastasis process. The recruitment of histone lysine-specific demethylase 1 (LSD1) to the E-cadherin promoter for the demethylation of histone H3 lysine 4 (H3K4) is reported to bridge H3K4 demethylation to DNA methylation on the E-cadherin promoter^[11]. In addition, one of the EMT-TFs, Snail, interacted with H3K9 methyltransferase G9a and Suv39H1, the two major methyltransferases responsible for H3K9 methylation that are intimately linked to DNA methylation^[22]. These epigenetic roles of histone lysine modifying enzymes could explain the detailed mechanism underlying E-cadherin silencing in breast cancer. EZH2 has also been suggested to promote metastasis through pleiotropic roles in modifying EMT, such as repression of tumor suppressor genes or miRNA, as well as regulation of cancer stem cells and migration^[23].

The full mechanism of epigenetic regulation of the EMT process via histone lysine modification cannot be understood without a definition of the EMT. Such a description must include the unique genetic alterations that may occur during the transition from an epithelial phenotype to a mesenchymal phenotype, resulting in the capacity for metastasis. There have been many in vitro experiments showing successful results from changes in the expression of epithelial and mesenchymal biomarkers, suggesting that these molecules play a role in the EMT process^[24]. However, these results have not been reproducible or validated in vivo or in clinical practice. Therefore, there is still some confusion and much debate on the role of genetic and epigenetic alterations in the EMT process during cancer metastasis^[25]. To overcome these limitations, cells bearing the hybrid phenotype such as circulating tumor cells (CTCs) have been studied. The present study could not show any results of serial cell lineage from EMT to mesenchymal-to-epithelial transition (MET) using CTC. However, there have been several efforts to define the role of the EMT and the MET process without a comprehensive study using serial cells from primary cancer to metastasis in the clinical fields. Xia et al.^[20]suggested that a quantitative EMT scoring system based on the genetic expression profile of the primary cancer, scoring the state of EMT from −1.0 to +1.0, could define the steps of metastasis. These authors showed that this system could identify the specific characteristics of the steps of the EMT process in the individual pathology of the cancers^[20]. The importance of identifying the individual steps of the EMT process in predicting patient prognosis and the clinical course has also been reported^[20][26]. Comprehensive information about the regulation of the EMT process could provide the advanced ability to predict the clinical outcome of cancer patients. Despite the absence of serial analysis using CTC, the changes in the expression pattern of EMT-associated markers, including epithelial and mesenchymal markers, as well as transcriptional markers, in primary cancer cells and their metastatic counterparts can help investigators define the likelihood of the development of EMT. If there are phenotypes intermediate between primary cancer and metastatic tumors, it is possible to identify the specific cells in the stromal compartment involved due to the tendency of the cancers to maintain the epithelial characteristics of their tissue of origin^[6].

According to the widest study for predicting the prognosis of BM patients, the following factors are associated with the prognosis of patients with BM of lung adenocarcinoma; age, KPS, extracranial metastasis, and number of BMs ^[27][28]. However, the present study showed that the sole factor of KPS was statistically associated with survival in multivariate analysis. Because of the limited selection, our subjects were patients who underwent surgical resection for BM. Therefore, the number of patients was too small to reflect the whole cohort with BM of lung adenocarcinoma, which limits the applications of the data to clinical practice.

Although our study suggests a meaningful role for several histone lysine modification enzymes, such as UTX, MLL4, and EZH2, in regulating the expression of EMT-TFs during the metastasis of lung adenocarcinoma to the brain, it has several limitations. First, our analyses in this study were performed using only immunohistochemical staining and qRT-PCR to define the epigenetic roles of UTX, MLL4, and EZH2 on the regulation of the expression of EMT-TFs. However, UTX is just one of several mechanisms demethylating H3K36m3, MLL4 is one of several mechanisms methylating H3K4, and EZH2 is also one of several mechanisms H3K27. They cannot represent the function of methylation of H3K36me, H3K4m1/2/3 and H3K27m1/2/3, respectively, in cancer biology. In terms of EZH2, mutation or overexpression of EZH2 is known to help cancerous cells divide and proliferate by inhibiting genes responsible for suppressing tumor development. EZH2 is an attractive target for anti-cancer therapy because it is upregulated in multiple cancers, including, but not limited to, breast^[29], prostate^[30], melanoma^[31], bladder cancer^[32], and lymphoma ^[33]. However, the role of EZH2 has not been comprehensively established in lung adenocarcinoma. The major role of EZH2 in oncogenesis is known to engage in the proliferation and development rather than the EMT process of the cancer^[34]. Actually, Abdel et al.^[35] reported that the expression of Twist and EZH2 was significantly higher in colon cancer than that in the normal colonic mucosa and suggested that Twist and EZH2 should serve as prognostic predictors for colon cancer, respectively. However, they did not show the relationship or interaction of Twist with EZH2. In our study, the immunoreactivity of Twist (23.6% vs. 3.3%; p = 0.013) and EZH2 (25.6% vs. 8.2%; p = 0.037) was also significantly higher in lung adenocarcinoma than in epithelial cell of the normal lung. However, the immunoreactivity of Twist (45.9% vs. 2.7%; p = 0.005) was still significantly higher in brain metastasis than in the normal brain, but that of EZH2 (8.6% vs. 7.8%; p = 0.834) was not different between brain metastasis and normal brain. Briefly, the EZH2 expression was not significantly increased in brain metastasis, unlike their primary cancer (lung adenocarcinoma), nor was it associated with the prognosis of lung adenocarcinoma patients independently. This discrepancy may be originated from the different expressions between lung adenocarcinoma and brain metastasis, the different cancer cell type, and the different sample size. As the traditional oncogenic function of EZH2 is focused on inducer of cancer cell proliferation and development, further comprehensive research is necessary for defining the other role of EZH2, such as EMT process, invasion, and angiogenesis. Although the present study suggested that these histone modifications influence the metastasis of lung adenocarcinoma to the brain, more comprehensive scientific evidence, supported by molecular genetic analysis using in vivo as well as in vitro study, is needed to validate the present results. It is also important to identify the unique target genes of these histone lysine modification enzymes to determine their role in the metastasis of lung adenocarcinoma to the brain.

Second, although two different neuropathologists assessed immunoreactivity in the samples, it is not certain whether our assessment of experiments in this study should be absolutely correct or not because the interpretation of the results obtained by immunohistochemical staining may be rather subjective. The optimal assessment of immunohistochemical staining results can differ according to the concentration of the antigen used for staining because of the difficulty in establishing standard conditions. In addition, there is no standard rule for determining the cutoff value between positive and negative findings. Therefore, it is necessary to establish a reasonable cutoff value in order to repeat the experiments for validation and to cooperate and communicate detail regarding the interpretation of the data among the investigators. In order to overcome the flaws in immunohistochemical staining, we used ROC curve analysis to establish the cutoff value in a principled manner. To determine the identity of immunoreactivity for the EMT and MET markers, EMT-TFs, and histone lysine modification enzymes at the cell level, an in vitro study will be helpful. Recently, a deep-learning-based method using artificial intelligence that can automatically localize and quantify the regions expressing biomarker in any selected area on a whole slide image is proposed^[36].

The third limitation is the lack of examination of all EMT markers, MET markers, EMT-TFs, and histone lysine modification enzymes that may be implicated. Importantly, there are many EMT-TF families, such as Snail (zinc finger proteins Snail and Slug), Zeb (zinc finger and homeodomain proteins Zeb1 and Zeb2), and Twist (basic helix-loop-helix proteins E12, E47, Twist1, Twist2, and Id), which have important functions in the normal development of human and EMT processes associated with cancer biology. These EMT-TFs are known to be major drivers of the EMT ^[37]. They can induce the dedifferentiation process of the epithelial component of cancer cells by repressing the transcription of epithelial gene transcription such as E-cadherin, known as a classic epithelial expressing gene; and activating mesenchymal gene such as N-cadherin as a classic mesenchymal expressing gene^[37][38]. However, we did not analyze all EMT-TFs; therefore, our results cannot reflect all the possible mechanisms of epigenetic regulation of EMT in lung adenocarcinomas. It is necessary for the investigators to perform a sequencing analysis to determine the target genes in human samples and validate the results with in vivo and in vitro studies.

Finally, another limitation of this study is the bias originating from the retrospective design of the study. If the sample size is sufficiently large, it can surmount this obstacle. However, our study involved a relatively small number of subjects and may, therefore, not meet the assumptions of the statistical tests used. We did our best to reduce the bias by obtaining the clinical data obtained from computerized data archives using a uniform system and included the candidate patients who were treated with the same protocol in a single center. The multiple researchers involved in this study did not have any clinical information or experimental results to help avoid preconception. We also independently reviewed the pathological slides and radiological images, and we cannot clearly say that no bias originated from this retrospective study. Despite these efforts, however, the conclusions drawn from our study need further validation through prospective and randomized clinical trials.