Identification Six Metabolic Genes as Potential Biomarkers for Lung Adenocarcinoma

Abstract

Metabolic genes have been reported to act as crucial roles in tumor progression. Lung adenocarcinoma (LUAD) is one of the most common cancers worldwide. This study aimed to predict the potential mechanism and novel markers of metabolic signature in LUAD. The gene expression profiles and the clinical parameters were obtained from The Cancer Genome Atlas-Lung adenocarcinoma (TCGA-LUAD) and Gene Expression Omnibus data set (GSE72094). A total of 105 differentially expressed metabolic genes of intersect expression in TCGA-LUAD and GSE72094 were screened by R language. Univariate Cox regression model found 18 survival-related genes and then the least absolute shrinkage and selection operator model was successfully constructed. Six significant genes prognostic model was validated though independent prognosis analysis. The model revealed high values for prognostic biomarkers by time-dependent receiver operating characteristic (ROC) analysis, risk score, Heatmap, and nomogram. In addition, Gene Set Enrichment Analysis showed that multiplex metabolism pathways correlated with LUAD. Furthermore, we found the six genes aberrantly expressed in LUAD samples. Our study showed that metabolism pathways play important roles in LUAD progression. The six metabolic genes could predict potential prognostic and diagnostic biomarkers in LUAD.

1. INTRODUCTION

Lung cancer is a very common malignant tumor and one of the most serious threats to human health. Its mortality ranks first in global malignant tumors (Li et al., 2017; Brainard and Farver, 2019). Among them, nonsmall cell lung cancer (NSCLC) accounts for >80% of all lung cancer cases, and lung adenocarcinoma (LUAD) is the major subtype of NSCLC (Shi et al., 2016; Chen et al., 2019b). The epidemiological characteristics of lung cancer include high incidence, rapid growth, high mortality, and poor prognosis (Lee et al., 2018). Despite the continuous improvement of diagnosis and therapy methods in recent years, the prognosis of LUAD patients has not obviously improved, the overall survival is still worse (Pinto et al., 2019). Therefore, it is imperative to research the mechanism and identify powerful prognostic and diagnostic biomarkers of LUAD.

Metabolic pathways are necessary to maintain the homeostasis of normal cells, including glycolysis, fatty acid metabolism, tyrosine metabolism, nucleotide synthesis metabolism, etc. (Kreuzaler et al., 2019; Woolbright et al., 2019). Increased metabolism is a hallmark of tumor cells (El Hassouni et al., 2019). In the procedure of high metabolism, diverse metabolites are abnormally expressed, such as ROS, FAS, TKTL1, and GLUT1. (Furuta et al., 2010). These abnormal metabolites drive a series of physiological cascade reactions in tumor cells and promote the tumor progression (Drewa et al., 2006; Koumarianou et al., 2014; da Costa et al., 2018; Zhu et al., 2018; Meng et al., 2019). Because these could be detected in various stages of tumor tissues, metabolic genes were considered as suitable prognostic and diagnostic biomarkers of cancers (Marin-Rubio et al., 2019). Whereas the role of metabolic genes in LUAD is still an uncertainty.

The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) provide lots of tumor-related information, such as gene expression, methylation, mutation, and clinical parameters. (Bezzecchi et al., 2019). In this study, we initially screened differentially expressed metabolic genes, and then using least absolute shrinkage and selection operator (LASSO), we constructed a significant genes prognostic model. Furthermore, we verified that the model might predict poor overall survival by univariate Cox regression, receiver operating characteristic (ROC) curves, and risk score. In addition, Kyoto Encyclopedia of Genes and Genomes (KEGG) enriched pathways were shown by Gene Set Enrichment Analysis (GSEA). Finally, we compared the expression of the six genes in tumor tissue and normal tissue and validated whether the six genes could be independent factors in predicting prognosis. We aim to expound the relationship of metabolism pathways with LUAD and identified a novel prognostic biomarker for LUAD.

2. METHODS

2.1. TCGA and GEO data collection

We downloaded mRNA expression and clinical data from TCGA-LUAD (https://portal.gdc.cancer.gov/). A total of 535 lung adenocarcinoma (ADC) and 59 normal LUAD samples were obtained from TCGA for gene expression analysis and prognosis analysis. GEO data sets (GSE72094) including 442 LUAD samples were selected for the study. The data of TCGA or GEO samples were extracted, annotated, and normalized by “Strawberry Perl 5.30.”

2.2. Identification of differentially expressed metabolic genes

The file containing metabolic gene annotation codes was download in the GSEA platform (http://software.broadinstitute.org/gsea/downloads). A total of 946 metabolic genes were found from TCGA-LUAD samples. Using Limma package in R, 715 genes of intersect expression in TCGA and GSE72094 were selected. And then the batch effect was eliminated by “sva” packages in 3.6.1 version R language. Differentially expressed genes (DEGs) were screened through R package Limma. p Value <0.01 and |log2-fold change| > 1 were defined for DEGs. Heatmap and volcano were constructed through pheatmap package in R.

2.3. Construction and validation of the prognostic model

Using univariate Cox regression analysis, we acquired 18 genes that correlate between mRNA expression and patients' survival data in TCGA. The significant prognostic genes of p value <0.01 were filtered in LASSO penalized Cox regression analysis. After 1000 resamples by R glmnet package, the prognostic model was built. Univariate Cox regression and multiple Cox regression analyses were performed to validate whether the model could be an independent prognostic factor in GSE72094. Kaplan–Meier analyses were performed to demonstrate the association of significant genes with overall survival (OS) by R “Survival” and “survminer.” Time-dependent ROC analyses were performed in R “survivalROC” to validate gene signatures in TCGA and GSE72094 data sets. If not specifically mentioned, p value <0.05 was considered significant for survival analysis.

2.4. The nomogram prediction model

GES72094 was used for primary cohort and the TCGA-LUAD was used for validation cohort. According to the results of multiple Cox regression, nomogram was constructed by rms package and prognostic factors in R. The calibration plot and concordance index (C-index) analysis were performed to validate the nomogram.

2.5. Gene set enrichment analysis

We used GSEA_4.0.2 to identify the enriched KEGG pathways of the model genes in TCGA-LUAD and GSE72094. Representative pathways were revealed in the graph by R language.

2.6. Gene expression in TCGA-LUAD

Scatter plots were built to show gene levels between tumor and adjacent normal tissue by R “beeswarm.” Paired difference analyses and box-plot were utilized for further verifying the expression relationship. External data from Oncomine (https://www.oncomine.org/resource/main.html) and Human Protein Atlas (www.proteinatlas.org) database were used to examine our results.

3. RESULTS

3.1. Identification of survival-related differentially expressed metabolic genes

A total 946 metabolic genes were found from 535 ADC and 59 normal LUAD samples in the TCGA-LUAD cohort. And then, 715 genes of intersect expression in TCGA-LUAD and GSE72094 were selected. p Value <0.01 and |log2-fold change| > 1 were defined for screening DEGs. Volcano plot was constructed to reveal the 105 significant differentially expressed genes (Fig. 1A) and Heatmap was performed to show the hierarchical clustering analysis of the gene signatures (Fig. 1B). In total, 522 cases from TCGA-LUAD and 442 cases from GSE72094 were extracted for prognosis analysis. Parameters of missing data and survival time <30 days were removed from the cases. Univariate Cox regression analysis dug out 18 survival-related genes in TCGA-LUAD (Fig. 1C, p value <0.01).

FIG. 1.

Survival-related differentially expressed metabolic genes. (A) Heatmap displaying the hierarchical clustering analysis of the differentially expressed metabolic genes in TCGA-LUAD. (B) Volcano plot of gene expression data. Grey dots: DEGs in TCGA-LUAD. Black dots: non-DEGs. p Value <0.01 and |log2-fold change| > 1 were considered as significant. (C) Forest plot of the univariate Cox regression analysis showing the survival-related DEGs in TCGA-LUAD. DEGs, differentially expressed genes; TCGA-LUAD, The Cancer Genome Atlas-Lung Adenocarcinoma.

3.2. Construction and validation of the prognostic model

The 18 metabolic gene signatures were filtered into LASSO penalized Cox regression analysis, after 1000 resamples, a six-gene prognostic model was constructed. The model contained six genes as follows: lactate dehydrogenase A (LDHA), lysophosphatidylglycerol acyltransferase 1 (LPGAT1), glucosamine-phosphate N-acetyltransferase 1 (GNPNAT1), prostaglandin E synthase (PTGES), thymidylate synthetase (TYMS), and aldehyde dehydrogenase 2 family member (ALDH2). The risk score = (0.00312612684709971 × the expression level of LDHA) + (0.00476085842822368 × the expression level of LPGAT1) + (0.0205522625269787 × the expression level of GNPNAT1) + (0.000101832048720509 × the expression level of PTGES) + (0.00831672554817241 × the expression level of TYMS) + (−4.27270512109639e-05 × the expression level of ALDH2). We classified the TCGA-LUAD and GSE72094 samples into high-risk and low-risk groups according to the median risk score.

Kaplan–Meier analyses were performed to demonstrate the high-risk group predicted poor overall survival (Fig. 2A, E). Univariate Cox analysis and multiple Cox regression analysis verified whether the model could be independent progress factors (Table 1). Time-dependent ROC analyses were performed to validate the gene signatures. The area under the ROC curve (AUC) for the model was 0.749 and 0.655 in TCGA and GEO data sets, respectively. Compared with other parameters such as age, sex, etc., the model of the gene signatures showed a better predictive value (Fig. 2B, F). Risk score distribution showed higher mortality in the high-risk group than in the low-risk group (Fig. 2C, G). The expression of the six gene signatures was presented by Heatmaps. LDHA, LPGAT1, GNPNAT1, PTGES, and TYMS showed positive correlation with high-risk set and ALDH2 revealed an opposite role (Fig. 2D, H). p Value <0.05 was considered statistically significant.

FIG. 2.

Kaplan–Meier analysis, time-dependent ROC analysis and risk score analysis for prognostic model in TCGA-LUAD and GSE72094. (A, E) Kaplan–Meier analysis of the prognostic model in TCGA or GEO data sets. (B, F) Time-dependent ROC analysis showing the optimal AUC of gene signature in the two cohorts. (C, G) Risk score distribution of the six gene signatures. (D, H) The hierarchical clustering analysis of six genes. AUC, area under the ROC curve; GEO, Gene Expression Omnibus; ROC, receiver operating characteristic.

Table 1.

Contribution of Various Potential Prognostic Factors to Survival by Univariate Cox Regression and Multiple Cox Regression

	Univariate Cox regression			Multiple Cox regression
	HR	Confidence interval	p	HR	Confidence interval	p
TCGA-LUAD
Age	0.996	0.978–1.015	0.718	1.006	0.986–1.025	0.565
Gender	1.000	0.694–1.441	1.000	0.848	0.584–1.231	0.385
Stage	1.648	1.396–1.946	<0.001^*	1.988	1.227–3.222	0.005^*
T	1.600	1.285–1.994	<0.001^*	1.091	0.855–1.391	0.483
M	1.748	0.959–3.187	0.068	0.392	0.110–1.401	0.150
N	1.787	1.455–2.195	<0.001^*	0.931	0.613–1.416	0.739
Risk score	5.119	3.151–8.316	<0.001^*	4.073	2.424–6.844	<0.001^*
GSE72094
Age	1.001	0.990–1.030	0.344	1.009	0.989–1.030	0.377
Gender	1.501	1.027–2.194	0.036^*	1.664	1.127–2.459	0.011^*
Stage	1.628	1.358–1.951	<0.001^*	1.675	1.390–2.018	<0.001^*
Smoking	1.375	0.603–3.138	0.449	1.066	0.463–2.454	0.880
Risk score	9.208	2.820–30.060	<0.001^*	11.114	3.226–38.288	<0.001^*

p < 0.05 was considered significant.

HR, hazard ratio; TCGA-LUAD, The Cancer Genome Atlas-Lung Adenocarcinoma.

3.3. The nomogram prediction model

Multivariate analysis showed that gender, stage, and the prognostic model were independent factors of OS in GSE72094. According to the Cox regression, nomogram was built to predict prognostics in LUAD (Fig. 3A). The C-index was 0.689 in the GSE72094 cohort, and 0.706 in the validation cohort (TCGA-LUAD). The calibration plots showed the nomogram could be well predicted for OS, especially in 1- and 2-year overall survival (Fig. 3B, C).

FIG. 3.

Construction and validation of the nomogram prediction model. (A) The nomogram plot was built in GES72094. (B, C) The calibration plots revealing the nomogram could be well predicted for OS, especially in 1- and 2-year overall survival. OS, overall survival.

3.4. Gene set enrichment analysis

We identified the enriched KEGG pathways of the model genes in TCGA-LUAD and GSE72094. Most of the enriched pathways focused on metabolism-related pathways, such as, arachidonic acid metabolism, pyrimidine metabolism, cysteine and methionine metabolism, glyoxylate and dicarboxylate metabolism, and riboflavin metabolism. The results also contained several famous cancer-related pathways, such as cell cycle, p53 signal pathway, and basal transcription factors. Figure 4 showed some representative pathways. Related parameters for the pathways are given in Table 2. The results further elucidated the role of metabolism mechanism in LUAD.

FIG. 4.

Ten representative enriched KEGG pathways by GESA. Each group contains five KEGG pathways. Related parameters for the pathways are given in Table 2. GESA, gene set enrichment analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Table 2.

Related Parameters of the Representative Kyoto Encyclopedia of Genes and Genomes Pathways in Figure 4

	Size	ES	NES	p	FDR-q
High risk
KEGG_CELL_CYCLE	124	0.77	2.48	0.000	0.000
KEGG_P53_SIGNALING_PATHWAY	68	0.57	2.17	0.000	0.001
KEGG_AMINOACYL_TRNA_BIOSYNTHESIS	41	0.75	2.12	0.000	0.002
KEGG_CYSTEINE_AND_METHIONINE_METABOLISM	34	0.63	2.02	0.002	0.005
KEGG_RIBOFLAVIN_METABOLISM	16	0.6	1.81	0.006	0.027
Low risk
KEGG_ARACHIDONIC_ACID_METABOLISM	58	−0.59	−2.08	0.000	0.037
KEGG_LINOLEIC_ACID_METABOLISM	29	−0.64	−1.99	0.000	0.052
KEGG_PRIMARY_BILE_ACID_BIOSYNTHESIS	16	−0.67	−1.92	0.000	0.059
KEGG_GLYCEROPHOSPHOLIPID_METABOLISM	77	−0.48	−1.91	0.002	0.054
KEGG_ALPHA_LINOLENIC_ACID_METABOLISM	19	−0.63	−1.86	0.006	0.066

ES, enrichment score; FDR-q, false discovery rate-q value; NES, normalized enrichment score.

3.5. Gene expression in TCGA-LUAD

As in the scatter plots and paired difference plots shown, the levels of LDHA, LPGAT1, GNPNAT1, PTGES, and TYMS were upregulated in TCGA-LUAD tissues compared with the adjacent normal tissues. ALDH2 expression level was downregulated in tumor tissues (Fig. 5A, B). According to tumor stage, we divided the tumor tissues into two groups. LDHA, LPGAT1, GNPNAT1, and TYMS were overexpressed in stage III + IV groups and ALDH2 was upregulated in stage I + II groups (Fig. 5C). External data from Oncomine database showed a consistent result with the TCGA cohort (Fig. 6A). Using the human protein profiles database, we observed the tumor samples could be deeply staining by LDHA, LPGAT1, GNPNAT1, PTGES, or TYMS antibodies and lightly stained by ALDH2 antibody. Unfortunately, in the database, we did not find any standard normal lung tissue stained by the six genes.

FIG. 5.

Gene expression in TCGA-LUAD. (A) The scatter plots showing the levels of genes in tumor tissues and normal tissues. (B) The paired difference plots revealing genes expression in tumor tissues comparing with the adjacent normal tissues. Both of the genes have significant differential expression. (C) LDHA, LPGAT1, GNPNAT1, and TYMS upregulate in stage III + IV groups and ALDH2 downregulate in stage I + II groups. Nondifferential expression only appears in PTGES. p Value <0.05 was considered as significant. ALDH2, aldehyde dehydrogenase 2 family member; LDHA, lactate dehydrogenase A; LPGAT1, lysophosphatidylglycerol acyltransferase 1; GNPNAT1, glucosamine-phosphate N-acetyltransferase 1; PTGES, prostaglandin E synthase; TYMS, thymidylate synthetase.

FIG. 6.

External data from Oncomine database showing a consistent result with TCGA cohort. (A) The chart showing the expression of the six genes in diverse tumors and the data of LUAD tissues versus normal tissues are listed in the chart. Owing to only one set meeting the screening criteria, the contrast of GNPNAT1 expression has not been built. p Value <0.01 and fold change >2 were utilized for screening in Oncomine database. (B) The expression of six genes in tumor specimens from the human protein profiles database.

4. DISCUSSION

LUAD progression is associated with alterations in assorted oncogenes and tumor suppressors (Samy et al., 2016; Kleczko et al., 2019). Despite diagnosis and treatment of LUAD significantly improved, the prognosis was still not optimistic (Abe et al., 2019). Recent researches showed that metabolic pathways play an important role in regulating tumor progression (Lacroix et al., 2019; Phan et al., 2019). The survival and proliferation of cancer cells are dependent on metabolic reprogram (Hoxhaj and Manning, 2019). Quantity studies reported the possibility of metabolic pathways as targeted therapy for tumors. Abdel-Wahab et al. reported that targeting glucose metabolism might be a novel approach to suppress cancer progression (Abdel-Wahab et al., 2019; Agrawal, 2019). However, the underlying mechanism of metabolism in LUAD has not been completely elucidated, hampering targeted therapy of LUAD (Relli et al., 2019). Thus, the priorities of LUAD research are very urgent to discover the novel molecular markers associated with the prognosis of LUAD.

In this study, using LASSO analysis, we delineated six metabolic genes prognostic model in TCGA-LUAD and GSE72094 data sets. Kaplan–Meier analyses demonstrated the model could predict poor overall survival. Univariate Cox analysis and multiple Cox regression analysis verified whether the model could be independent progress factors. The AUC dimension and risk score distribution of gene signature showed a high prognostic value in predict OS. The nomogram C-index was 0.689 in the GSE72094 cohort, and 0.706 in the TCGA-LUAD cohort. As shown in calibration plots, the nomogram could be well predicted for OS, especially in 1- and 2-year survival. Many of the enriched pathways focused on metabolism-related pathways. In addition to metabolic pathways, the high-risk groups showed several cancer-related pathways, such as cell cycle and p53 signal. The results showed the gene signature closely correlated with metabolic pathways and prompted the potential role of metabolic pathways in LUAD. In addition, LDHA, LPGAT1, GNPNAT1, PTGES, and TYMS upregulated in tumor tissues compared with adjacent normal tissues, whereas ALDH2 revealed an opposite role.

Previous studies showed that the great majority of the six genes closely related to tumor progression. LDHA has been reported to correlate with the metabolism pathway of pyruvate and citric acid (TCA) cycle (Mishra and Banerjee, 2019). Numerous researches revealed the aberrantly upregulated LDHA in multiple malignant tumors (Miao et al., 2013). Hou et al. (2019) reported upregulated LDHA promoted glycolysis and accelerated proliferation and invasion in LUAD cells. In light of the harmful effect in tumor progression, LDHA inhibitors and target drugs are intensively developed (Feng et al., 2018). PTGES was shown to associate with the pathways of prostaglandin 2 biosynthesis and metabolism and arachidonic acid metabolism (Ramanan and Doble, 2017). As a cancer promoting factor, PTGES involved in the regulation of multiple tumor procession, such as lung cancer, colon cancer, and breast cancer (Filipenko et al., 2016; Tafakh et al., 2018; Wang et al., 2019). Wang et al. (2019) reported that PTGES was prevented from proteasome-directed degradation through interaction with USP9X, and the dysregulated PTGES promoted cell migration and metastasis in NSCLC. TYMS could catalyze the methylation of deoxyuridylate to deoxythymidylate and regulate DNA replication and repair (Varghese et al., 2019). Recent studies showed the potential role of TYMS in cancer chemotherapy, for example, modulating 5-FU (5-fluorouracil) resistance in colon cancer cells (Chen et al., 2019c).

Chamizo et al. (2015) suggested that upregulated TYMS associated with reduced response to pemetrexed-containing chemotherapy in advanced NSCLC patients. As a non-cytochrome P450 mitochondrial aldehyde oxidizing enzyme, ALDH2 participated in regulating alcohol metabolism and pyruvate metabolism (Chen et al., 2019a). As a tumor suppressor, ALDH2 depletion predicted poor prognosis in diverse tumors, such as lung cancer, liver cancer, gastric cancer, and breast cancer (Li et al., 2016). One research reported depletion of ALDH2 enhanced the sensitivity of NSCLC cells to Taxol (Wang et al., 2018). Compared with the mentioned four genes, there are relatively few reports of LPGAT1 and GNPNAT1, especially in cancer research. LPGAT1 is an activator in reacylation of lysophosphatidylglycerol and modulates the biosynthesis of glycerophospholipid (Yang et al., 2004). MicroRNA-30c could protect the liver from steatosis through repressing LPGAT1 (Vickers and Moore, 2013). So far, there have been no particular reports about LPGAT1 on pulmonary tumors. GNPNAT1, also named GNA1, correlated the pathway of synthesis of substrates in N-glycan biosythesis (Boehmelt et al., 2000). Decreased GNPNAT1 promoted cells proliferation and invasion in castration-resistant prostate cancer (Kaushik et al., 2016). Zhao et al. (2017) reported downregulation of GNPNAT1-induced A549 cells proliferative delay and cell adhesion defect. The gene signatures are important members of the metabolic pathways, and could act as therapeutic target in cancers.

5. CONCLUSIONS

Metabolism pathways play important roles in LUAD progression. The six metabolic genes could be potential biomarkers to predict prognosis and diagnosis of LUAD. However, the precise role of metabolism pathways and the six genes in LUAD progression remain poorly understood. Further researches must be conducted to expound the explicit role of the gene signature, specifically the two less studied genes, LPGAT1 and GNPNAT1.

Footnotes

AUTHOR DISCLOSURE STATEMENT

The authors declare they have no competing financial interests.

FUNDING INFORMATION

The authors received no funding for this article.

References

Abdel-Wahab

A.F.

, Mahmoud

, and Al-Harizy

R.M.

2019. Targeting glucose metabolism to suppress cancer progression: Prospective of anti-glycolytic cancer therapy. Pharmacol Res. 150, 104511.

Abe

, Watanabe

, Shinozaki-Ushiku

, et al. 2019. Identification of a metastatic lung adenocarcinoma of the palate mucosa through genetic and histopathological analysis: A rare case report and literature review. BMC Cancer. 19, 52.

Agrawal

2019. New therapeutic targets for cancer: The interplay between immune and metabolic checkpoints and gut microbiota. Clin. Transl. Med. 8, 23.

Bezzecchi

, Ronzio

, Dolfini

, et al. 2019. NF-YA overexpression in lung cancer: LUSC. Genes (Basel), 10, pii: E937.

Boehmelt

, Fialka

, Brothers

, et al. 2000. Cloning and characterization of the murine glucosamine-6-phosphate acetyltransferase EMeg32. Differential expression and intracellular membrane association. J. Biol. Chem. 275, 12821–12832.

Brainard

, and Farver

2019. The diagnosis of non-small cell lung cancer in the molecular era. Mod Pathol. 32(Suppl 1), 16–26.

Chamizo

, Zazo

, Domine

, et al. 2015. Thymidylate synthase expression as a predictive biomarker of pemetrexed sensitivity in advanced non-small cell lung cancer. BMC Pulm. Med. 15, 132.

Chen

C.H.

, Ferreira

J.C.B.

, and Mochly-Rosen

2019a. ALDH2 and cardiovascular disease. Adv. Exp. Med. Biol. 1193, 53–67.

Chen

, Xu

, Qian

, et al. 2019b. The biological functions and clinical applications of exosomes in lung cancer. Cell Mol. Life Sci. 76, 4613–4633.

10.

Chen

, Yue

, Huang

, et al. 2019c. Inhibition of hydrogen sulfide synthesis reverses acquired resistance to 5-FU through miR-215-5p-EREG/TYMS axis in colon cancer cells. Cancer Lett. 4661, 49–60.

11.

da Costa

I.A.

, Hennenlotter

, Stuhler

, et al. 2018. Transketolase like 1 (TKTL1) expression alterations in prostate cancer tumorigenesis. Urol. Oncol. 36, 472.e21–e472.e27.

12.

Drewa

, Wolski

, Skok

, et al. 2006. The FAS-related apoptosis signaling pathway in the prostate intraepithelial neoplasia and cancer lesions. Acta. Pol. Pharm. 63, 311–315.

13.

El Hassouni

, Granchi

, Valles-Marti

, et al. 2019. The dichotomous role of the glycolytic metabolism pathway in cancer metastasis: Interplay with the complex tumor microenvironment and novel therapeutic strategies. Semin. Cancer Biol pii: S1044-579X(19)30180-4.

14.

Feng

, Xiong

, Qiao

, et al. 2018. Lactate dehydrogenase A: A key player in carcinogenesis and potential target in cancer therapy. Cancer Med. 7, 6124–6136.

15.

Filipenko

, Schwalm

, Reali

, et al. 2016. Upregulation of the S1P3 receptor in metastatic breast cancer cells increases migration and invasion by induction of PGE2 and EP2/EP4 activation. Biochim. Biophys. Acta. 1861, 1840–1851.

16.

Furuta

, Okuda

, Kobayashi

, et al. 2010. Metabolic genes in cancer: Their roles in tumor progression and clinical implications. Biochim. Biophys. Acta. 1805, 141–152.

17.

Hou

X.M.

, Yuan

S.Q.

, Zhao

, et al. 2019. LDH-A promotes malignant behavior via activation of epithelial-to-mesenchymal transition in lung adenocarcinoma. Biosci. Rep. 39, pii: BSR20181476.

18.

Hoxhaj

, and Manning

B.D.

2019. The PI3K-AKT network at the interface of oncogenic signalling and cancer metabolism. Nat. Rev. Cancer. 20:74–88.

19.

Kaushik

A.K.

, Shojaie

, Panzitt

, et al. 2016. Inhibition of the hexosamine biosynthetic pathway promotes castration-resistant prostate cancer. Nat. Commun. 7, 11612.

20.

Kleczko

E.K.

, Kwak

J.W.

, Schenk

E.L.

, et al. 2019. Targeting the complement pathway as a therapeutic strategy in lung cancer. Front. Immunol. 10, 954.

21.

Koumarianou

, Tzeveleki

, Mekras

, et al. 2014. Prognostic markers in early-stage colorectal cancer: Significance of TYMS mRNA expression. Anticancer Res. 34, 4949–4962.

22.

Kreuzaler

, Panina

, Segal

, et al. 2019. Adapt and conquer: Metabolic flexibility in cancer growth, invasion and evasion. Mol. Metab. pii: S2212-8778(19)30906-8.

23.

Lacroix

, Riscal

, Arena

, et al. 2019. Metabolic functions of the tumor suppressor p53: Implications in normal physiology, metabolic disorders, and cancer. Mol. Metab. pii: S2212-8778(19)30921-4.

24.

Lee

W.H.

, Loo

C.Y.

, Ghadiri

, et al. 2018. The potential to treat lung cancer via inhalation of repurposed drugs. Adv. Drug Deliv. Rev. 133, 107–130.

25.

, Jiang

, Leng

, et al. 2017. A prediction model for distinguishing lung squamous cell carcinoma from adenocarcinoma. Oncotarget, 8, 50704–50714.

26.

, Zhao

, Sun

, et al. 2016. ALDH2 gene polymorphism in different types of cancers and its clinical significance. Life Sci. 147, 59–66.

27.

Marin-Rubio

J.L.

, Vela-Martin

, Fernandez-Piqueras

, et al. 2019. FADD in cancer: Mechanisms of altered expression and function, and clinical implications. Cancers (Basel), 11, pii: E1462.

28.

Meng

, Xu

, Luan

, et al. 2019. The progress and development of GLUT1 inhibitors targeting cancer energy metabolism. Future Med. Chem. 11, 2333–2352.

29.

Miao

, Sheng

, Sun

, et al. 2013. Lactate dehydrogenase A in cancer: A promising target for diagnosis and therapy. IUBMB Life, 65, 904–910.

30.

Mishra

, and Banerjee

2019. Lactate dehydrogenases as metabolic links between tumor and stroma in the tumor microenvironment. Cancers (Basel), 11, pii: E750.

31.

Phan

T.K.

, Bindra

G.K.

, Williams

S.A.

, et al. 2019. Combating Human Pathogens and Cancer by Targeting Phosphoinositides and Their Metabolism. Trends Pharmacol. Sci. 40, 866–882.

32.

Pinto

, Petriella

, Lacalamita

, et al. 2019. KRAS-driven lung adenocarcinoma and B cell infiltration: Novel insights for immunotherapy. Cancers (Basel), 11, pii: E1145.

33.

Ramanan

, and Doble

2017. Transcriptional regulation of mPGES1 in cancer: An alternative approach to drug discovery?. Curr. Drug. Targets. 18, 119–131.

34.

Relli

, Trerotola

, Guerra

, et al. 2019. Abandoning the notion of non-small cell lung cancer. Trends Mol. Med. 25, 585–594.

35.

Samy

M.D.

, Yavorski

J.M.

, Mauro

J.A.

, et al. 2016. Impact of SNPs on CpG islands in the MYC and HRAS oncogenes and in a wide variety of tumor suppressor genes: A multi-cancer approach. Cell Cycle, 15, 1572–1578.

36.

Shi

, Hua

, Zhu

, et al. 2016. Somatic genomics and clinical features of lung adenocarcinoma: A retrospective study. PLoS Med, 13, e1002162.

37.

Tafakh

M.S.

, Saidijam

, Ranjbarnejad

, et al. 2018. Sulforaphane, a chemopreventive compound, inhibits cyclooxygenase-2 and microsomal prostaglandin E synthase-1 expression in human HT-29 colon cancer cells. Cells Tissues Organs, 206, 46–53.

38.

Varghese

, Magnani

, Harada-Shoji

, et al. 2019. FOXM1 modulates 5-FU resistance in colorectal cancer through regulating TYMS expression. Sci. Rep. 9, 1505.

39.

Vickers

K.C.

, and Moore

K.J.

2013. Small RNA overcomes the challenges of therapeutic targeting of microsomal triglyceride transfer protein. Circ. Res. 113, 1189–1191.

40.

Wang

N.N.

, Wang

L.H.

, Li

, et al. 2018. Targeting ALDH2 with disulfiram/copper reverses the resistance of cancer cells to microtubule inhibitors. Exp. Cell Res. 362, 72–82.

41.

Wang

, Jing

, Sun

, et al. 2019. Stabilization of PTGES by deubiquitinase USP9X promotes metastatic features of lung cancer via PGE2 signaling. Am. J. Cancer Res. 9, 1145–1160.

42.

Woolbright

B.L.

, Rajendran

, Harris

R.A.

, et al. 2019. Metabolic flexibility in cancer: Targeting the pyruvate dehydrogenase kinase:pyruvate dehydrogenase axis. Mol. Cancer Ther. 18, 1673–1681.

43.

Yang

, Cao

, and Shi

2004. Identification and characterization of a gene encoding human LPGAT1, an endoplasmic reticulum-associated lysophosphatidylglycerol acyltransferase. J. Biol. Chem. 279, 55866–55874.

44.

Zhao

, Li

, Ma

, et al. 2017. Nanoparticle abraxane possesses impaired proliferation in A549 cells due to the underexpression of glucosamine 6-phosphate N-acetyltransferase 1 (GNPNAT1/GNA1). Int. J. Nanomedicine, 12, 1685–1697.

45.

Zhu

, Ma

, Qian

, et al. 2018. The molecular mechanism and clinical significance of LDHA in HER2-mediated progression of gastric cancer. Am. J. Transl. Res. 10, 2055–2067.