Background
Lung cancer is the leading cause of cancer-related mortality worldwide. There are two clinical subtypes for lung cancer: non-small cell lung cancer (NSCLC) (approximately 85% occurrence), and small cell lung cancer (SCLC) (approximately 15% occurrence) [
1]. Based on pathological and molecular features, NSCLC is divided into the following major subtypes: lung squamous cell carcinoma (LUSC), lung adenocarcinoma (LUAD), and large cell lung cancer [
2]. Recent advances in targeted treatments, such as epidermal growth factor receptor (EGFR) kinase inhibitors, have increased the overall survival (OS) of patients with LUAD [
3]. However, no specific biomarkers or relatively optimal targeted therapies have been identified for LUSC patients, and the 5-year survival rate of LUSC is less than 20% [
4]. Therefore, it is necessary to explore specific diagnostic and prognostic biomarkers for LUSC.
Energy metabolism reprogramming is a process that promotes cancer cell growth and proliferation via adjustment of energy metabolism, and it has been regarded as an emerging hallmark of cancer [
5]. Under aerobic conditions, normal cells obtain energy through mitochondrial oxidative phosphorylation. Under anaerobic conditions, the cells obtain energy via glycolysis instead of oxygen-consuming mitochondrial metabolism [
6]. Glycolysis, also known as the Warburg effect, is often observed in human cancer cells, in which the cancer cells favor glucose metabolism via glycolysis even in the presence of oxygen [
7]. This phenomenon is a unique energy metabolism that exists in cancer cells. In recent years, many biomarkers for LUSC have been discovered, including glycolysis-associated genes such as kininogen 1 (KNG1) [
8] and tripartite motif-containing protein 59 (TRIM59) [
9].
With the development of high-throughput sequencing, various patient genome databases have been constructed, which enables us to acquire a deep understanding of genomic changes. Based on database mining, an increasing number of biomarkers have been identified that are related to the survival of patients with cancer [
10,
11]. However, a single gene cannot be used to obtain satisfactory predictive effects. A multigene prognostic model from an original tumor biopsy can guide clinicians to choose more effective treatment strategies. Thus, a signature based on multigene expression associated with glycolysis should be established to predict the prognosis of LUSC patients.
In the present study, we used a genome-wide analysis of LUSC patient mRNA expression profiles from The Cancer Genome Atlas (TCGA) to construct a glycolysis-related gene signature that could effectively predict the prognosis in LUSC patients.
Discussion
Recently, numerous genes have been considered as biomarkers for cancer prognosis, and the clinical significance of the biomarkers has been explored. For example, a study by Tang and his colleagues found that the overexpression of dipeptidyl peptidase 9 (DPP9) was a significant independent factor for poor prognosis in patients with NSCLC [
14]. Similarly, Feng et al. [
15] reported that high expression of forkhead box Q1 (FoxQ1) was associated with poor prognosis in patients with NSCLC. However, the expression level of single genes can be influenced by multiple factors, and thus, these biomarkers can be unreliable for independent prognosis indications. Therefore, a statistical model based on a combination of multiple genes was used to improve the prediction of prognosis in cancer patients. Studies have shown that a pool of multiple genes was more accurate than a single gene in predicting the prognosis of patients with cancer [
16,
17].
In the present study, we obtained mRNA expression profiles for 501 LUSC patients from the TCGA database. We found that 5 glycolysis-related gene sets were significantly enriched in LUSC samples using GSEA. Univariate and multivariate Cox regression analyses were carried out to identify the risk score for the three-gene signature with prognostic value for patients with LUSC. Kaplan-Meier curve analysis indicated that patients with a high risk score had a poor prognosis when compared to patients with a low risk score. Additionally, in the stratified analysis, the risk score for the three-gene signature effectively predicted the prognosis of LUSC patients in all subgroups except for the subgroup of patients with remote tumor metastasis. The reason for this discrepancy might be that the number of patients with remote tumor metastasis was too small (n = 7). These results demonstrated that the risk score for the three-gene signature could be used as an independent prognostic indicator for LUSC patients. Moreover, measuring the patient risk score might assist clinicians in choosing optimal therapy methods.
The metabolism of tumor cells is more active than that of normal cells, and therefore, tumor cells require a greater amount of energy to maintain their higher proliferation [
18]. Glycolysis and oxidative phosphorylation are the two important metabolic pathways related to energy supply. Glycolysis is a relatively low-energy-providing pathway compared with oxidative phosphorylation. In the 1920s, Warburg found that cancer cells are very active in glycolysis and require a large amount of glucose to obtain ATP for metabolic activities [
19]. This aberrant phenomenon of glucose metabolism was called aerobic glycolysis, and is also known as the Warburg effect [
19,
20]. There was further research on the main genes and enzymes related to glycolysis to gain understanding of their functions in the metabolism of tumor cells.
In recent years, studies have shown that aerobic glycolysis plays a significant role in tumorigenesis, tumor progression, and metastasis. For example, enolase 1 (ENO1) was proved to promote cell glycolysis, growth, migration, and invasion in NSCLC [
21]. Glucose transporter 1 (GLUT1) facilitated increased transport of glucose into cancer cells to maintain an elevated rate of glycolysis under aerobic conditions [
22]. A high expression of GLUT1 was significantly associated with a poor prognosis in lung cancer patients [
23]. However, no set of glycolysis-related genes for predicting LUSC prognosis has been established.
HKDC1, a recently identified fifth hexokinase, plays an important role in cellular glucose metabolism [
24]. Aberrational expression of HKDC1 is associated with various cancers, including colorectal cancer [
25], liver cancer [
26], and breast cancer [
27]. Additionally, Wang and his colleagues reported that HKDC1 was overexpressed in LUAD tissues, and high expression of HKDC1 promoted proliferation, migration, and invasion in LUAD [
28]. Thus, HKDC1 can serve as a prognostic biomarker for LUAD patients [
28]. The aldehyde dehydrogenase (ALDH) superfamily comprises 19 enzymes that play a vital role in maintaining epithelial homeostasis. ALDH activity has been implicated in detoxification, cell proliferation, differentiation, drug resistance, and response to oxidative stress [
29,
30]. Thus, deregulation of these enzymes could result in various cancers, including esophageal squamous cell carcinoma [
31] and breast cancer [
32]. Giacalone et al. [
33] reported that ALDH7A1, one of the ALDH superfamily members, was correlated with OS and recurrence in patients with surgically resected stage I NSCLC.
MDH1, an NAD(H)-dependent enzyme, is an important part in the malate/aspartate shuttle (MAS) [
34]. This metabolic cycle contributes to maintaining intracellular NAD(H) redox homeostasis as it transfers the reducing equivalent NAD(H) across the mitochondrial membrane [
34]. It has been reported that abnormal expression of MDH1 is related to tumor occurrence and progression [
35]. For example, MDH1 promoted pancreatic ductal adenocarcinoma cell proliferation and metabolism through NAD production to support glycolysis [
35,
36]. Zhang et al. [
37] reported that MDH1 expression was elevated in NSCLC tissue compared with normal lung tissue. However, there were no combinations of these three glycolysis-related genes (HKDC1, ALDH7A1, and MDH1) to predict the prognosis of LUSC.
This study is the first to report that a glycolysis-based three-gene signature can serve as a prognostic indicator for patients with LUSC. A higher risk score indicates a worse prognosis. Of course, some study limitations remain. First, the risk score model was constructed using the TCGA database and should be verified in other cohorts in future studies. Second, studies on the three predicted genes should be performed to explore concrete mechanisms in the occurrence and development of LUSC.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.