WITHDRAWN: Developing a Risk Model of Lipid Metabolism-associated Genes: Implications for Clinical Prognosis and Tumor Immune Microenvironment in Gastric Cancer

doi:10.21203/rs.3.rs-4657182/v1

Download PDF

Research Article

WITHDRAWN: Developing a Risk Model of Lipid Metabolism-associated Genes: Implications for Clinical Prognosis and Tumor Immune Microenvironment in Gastric Cancer

https://doi.org/10.21203/rs.3.rs-4657182/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

The full text of this preprint has been withdrawn by the authors in order to comply with an institutional policy on preprints. Therefore, the authors do not wish this work to be cited as a reference.

Editorial notes are used to provide important context regarding the topic of a preprint or to alert readers to potential issues concerning that preprint or a downstream publication associated with it. For more information on editorial notes, see our Editorial Policies.

Background

Gastric cancer (GC) remains a significant clinical challenge due to its high mortality rate and complex pathophysiology. Recent advances have suggested that lipid metabolism plays a crucial role in the progression and prognosis of gastric cancer. This study aims to assess the implications of lipid metabolism-associated genes (LMAGs) in the molecular clinical prognosis and tumor immune microenvironment of GC.

Methods

We analyzed RNA-seq sequencing data from 371 GC transcriptomes in The Cancer Genome Atlas (TCGA) and 433 clinic datasets from the Gene Expression Omnibus (GEO). We selected the top 100 LMAGs for analysis. We applied univariate Cox regression to identify survival-related genes and used consensus clustering to explore molecular subtypes. We also performed copy number variation (CNV) and functional enrichment analysis. Using the LASSO and Cox regression analyses to construct and validate the GC-related prognostic risk model.

Results

A total of 3911 differentially expressed genes (DEGs) delineated distinct transcriptomic profiles between GC and normal gastric tissues were identified. Among these, 43 LMAGs were significantly associated with survival, most of which were poor prognostic factors. CNV analysis indicated prevalent gains in LMAGs in tumor samples. Consensus clustering revealed two distinct molecular subtypes (ARGcluster A and ARGcluster B) with significant differences in survival outcomes and gene expression profiles. Additionally, immune profiling revealed that ARGcluster A, associated with poorer outcomes, exhibited higher levels of pro-tumorigenic immune cells. Finally, we validated the expression, prognosis, and immune infiltration of fourteen key LMAGs in GC.

Conclusion

This study underscores the association between LMAGs with the prognosis and tumor immune microenvironment in GC. Further research is needed to validate these findings and explore potential therapeutic targets within the lipid metabolism pathway.

Gastric cancer (GC), a leading cause of cancer-related deaths worldwide, presents significant treatment challenges^[1]. Despite advancements in medical research and treatment strategies, the prognosis for many GC patients remains poor, largely due to the heterogeneity of the disease and its complex molecular underpinnings ^[2].

Recent advancements in molecular biology have highlighted the pivotal role of lipid metabolism in the progression and prognosis of various cancers, including GC ^[3]. Lipid metabolism encompasses a wide array of biochemical processes essential for the growth, proliferation, and survival of cells^[4]. In cancer cells, these processes are often dysregulated, leading to metabolic reprogramming that supports their rapid proliferation and resistance to cell death^[5]. Lipid metabolism not only supplies energy and building blocks for membrane synthesis but also modulates various signaling pathways that promote tumor growth and survival^[6].

The influence of lipid metabolism-associated genes (LMAGs) extends beyond metabolic reprogramming to the modulation of the tumor immune microenvironment (TIME)^[7]. Lipid metabolites can create an immunosuppressive environment that enables cancer cells to evade immune surveillance^[8]. For example, high levels of certain lipids can inhibit the function of T cells and natural killer (NK) cells while promoting the recruitment and polarization of tumor-associated macrophages (TAMs) towards a pro-tumorigenic phenotype^[9]. The infiltration of immune cells such as macrophages, T cells, and dendritic cells into the tumor microenvironment is regulated by lipid metabolism^[10].

In pancreatic cancer, LMAGs like fatty acid synthase (FASN) and acetyl-CoA carboxylase (ACC) are often overexpressed, driving de novo lipid synthesis. This process provides essential lipids for membrane biosynthesis and energy storage, supporting the high metabolic demands of proliferating cancer cells^[11]. Studies have shown that targeting FASN can reduce tumor growth and sensitize cancer cells to chemotherapy^[12]. Furthermore, lipid metabolism pathways such as the PI3K-AKT-mTOR axis are frequently activated in pancreatic cancer, contributing to tumor cell survival and resistance to apoptosis^[13]. Similarly, genes such as sterol regulatory element-binding protein 1 (SREBP1) are involved in lipid biosynthesis and are upregulated in colorectal cancer (CRC)^[14].

However, the LMAGs in GC patients have rarely been studied. In this study, we constructed a risk model incorporating clinical data to stratify patients based on their prognosis. Additionally, we analyzed the tumor immune microenvironment to understand the relationship between gene expression and immune cell infiltration. This integrative approach aims to uncover novel biomarkers and therapeutic targets that can improve personalized treatment strategies for GC.

2.1 Data collection and processing and acquisition of lipid metabolism gene list

We retrieved RNA-seq sequencing data for 371 GC transcriptomes and corresponding clinical data from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/). Additionally, we obtained data from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/ ) for the GSE84437 dataset, which includes survival and gene expression data for 433 patients^[15]. Detailed clinical characteristics of these patients are summarized in Table S. To enhance the accuracy of our study, we normalized the expression matrix using the 'normalize between arrays' function in the R package 'limma' (version 3.54.2)^[16] and excluded patients with a survival time of less than 30 days. Missing values in the clinical data were interpolated using the R package 'mice' (version 3.15.0). We utilized the GENECARDS database (https://www.genecards.org/ ) to identify genes associated with lipid metabolism. By searching for "lipid metabolism", we retrieved all relevant genes and extracted the top 100 most relevant genes for further investigation.

2.2 Consensus clustering analysis and principal component analysis

Consensus clustering was performed using the 'ConsensusClusterPlus' package (version 1.60.0), employing the k-means clustering algorithm with Euclidean distances and 50 replicates for 80% of the samples^[17]. PCA analysis and visualization were conducted using the R package 'FactoMineR' (version 2.7) to demonstrate the distribution between groups. We examined copy number variation (CNV) data from samples within the TCGA database to identify significant CNV amplifications or deletions within the cohort. Both amplifications and deletions were considered to enhance the detection of alterations in each gene. High-frequency CNVs were defined as those with a frequency exceeding 5%.

2.3 Functional enrichment analysis

Gene Ontology (GO) and KEGG enrichment analyses of the differentially expressed genes (DEGs) were performed using the R package 'clusterProfiler' (version 4.7.1.3^)[18]. Terms with a p-value < 0.05 were considered significantly enriched.

2.4 Identification of prognostic lipid metabolism-associated genes and the prognostic model

To investigate the prognostic significance of LMAGs in GC, univariate Cox regression analysis was performed using R software (version 4.2.0) with the 'survival' package. To construct a prognostic model for lipid metabolism, we applied the Least absolute shrinkage and selection operator (LASSO) regression after identifying LMAGs with prognostic value. Each gene in the model was assigned a coefficient based on the optimal lambda value, allowing for the calculation of risk scores for individual patients using the following formula:

Risk score=\(\:{\sum\:}_{i=1}^{n}({\beta\:}_{i}\times\:\text{e}\text{x}\text{p}\text{r}\text{e}\text{s}\text{s}\text{i}\text{o}\text{n}\:\text{o}\text{f}\:\text{l}\text{i}\text{p}\text{i}\text{d}\:\text{m}\text{e}\text{t}\text{a}\text{b}\text{o}\text{l}\text{i}\text{s}\text{m}-\text{a}\text{s}\text{s}\text{o}\text{c}\text{i}\text{a}\text{t}\text{e}\text{d}\:\text{g}\text{e}\text{n}\text{e}\:\text{i})\)

Patients were then categorized into high-risk and low-risk groups based on the median risk value. The model's prognostic utility was evaluated using survival analysis to compare the outcomes between the two groups. Additionally, the model's accuracy and robustness were assessed by generating ROC curves for 3-, 5-, and 8-year survival rates.

2.5 Clinical prediction value of the established prognostic model

To avoid bias and further evaluate the model’s prognostic efficacy, univariate and multivariate Cox regression analyses were performed. Risk scores, along with other clinical characteristics, were included in the analysis to identify independent prognostic markers.

2.6 Immune microenvironment analysis and construction of the nomogram

The tumor microenvironment and gastric carcinoma-infiltrating immune cells were assessed in silico. Using bulk RNA-seq data, the ESTIMATE algorithm was employed to predict the presence of infiltrating stromal and immune cells in the tumor microenvironment. ESTIMATE provided three scores based on single-sample Gene Set Enrichment Analysis (ssGSEA): stromal cell scores, immune cell scores, and ESTIMATE scores. Additionally, CIBERSORT, a deconvolution technique, was used to quantify the proportions of distinct cell types by predicting the cellular composition of complex tissues based on gene expression data. The relationship between the risk score and tumor immune infiltration score was investigated using a total of seven methods. The Tumor Immunology Single Cell Center (TISCH) database (http://tisch.comp-genomics.org/) serves as the expression of LMAGs prognostic model genes in tumor microenvironment (TME) cell types of GC patients. Based on the results of multivariate Cox regression, a nomogram was constructed as a versatile approach for combining multiple risk factors into a single predictive plot. This tool allows for the visual prediction of a patient’s survival probability. The nomogram was created using the R package 'DynNom'.

2.7 Statistical analysis

Statistical analyses were conducted using R software (version 4.2.1). DEGs were identified using linear models combined with the empirical Bayes method. Correlation analysis was performed using the Spearman correlation coefficient. For differential analysis of continuous variables, the student’s t-test was used for normally distributed data, and the Wilcoxon rank-sum test was employed for non-normally distributed data. All p-values in this study are two-tailed, with statistical significance set at p < 0.05.

3.1 Differences in Transcriptomic Profiles of Gastric Tissues in Individuals with and without GC and identification of survival-related genes

According to the cut-off criteria (fold change > 2 and adjusted P value < 0.05), a total of 3911 differentially expressed genes (2021 up-regulated and 1901 down-regulated) were identified in the gastric tissues of the training cohort (Fig. 1A, Supplementary Table S1). Unsupervised clustering based on these differentially expressed gene profiles indicated that the classification of samples was largely consistent with their respective groups (Fig. 1B), suggesting that the gastric tissues of cancer patients exhibited distinct transcriptomic profiles compared to the normal group. Using univariate Cox regression analysis, we identified 43 LMAGs that were significantly associated with survival in the training cohort (p < 0.05). Most of these genes were found to be poor prognostic factors in GC (Fig. 1C). The correlation of their expression levels is shown in Fig. 1D, with the majority displaying a significant positive correlation. The chromosomal distribution of these LMAGs is shown in Fig. 1E, with most located on chromosomes 1 to 12. CNV refers to alterations in the number of copies of a specific gene or genomic region, frequently occurring in tumors due to the various forms of genomic variations, including chromosomal CNVs, within tumor cells. CNV gain indicates an increase in the number of copies of a region, usually due to repeated DNA replication in that area. CNV loss indicates a decrease in the number of copies of a region, often due to the loss of DNA in that area^[19]. In tumors, CNV gains or losses are often associated with tumor development and treatment response. For example, CNV gain may lead to the overexpression of oncogenes, promoting tumor growth and spread, whereas CNV loss may result in decreased expression of tumor suppressor genes, facilitating tumor progression. Therefore, detecting CNV gains or losses is crucial for diagnosis, prognosis evaluation, and treatment selection^[20]. We found that in GC patients, most LMAGs exhibited stronger CNV gains than losses, suggesting their involvement in tumor growth and spread, contributing to poor prognosis (Fig. 1F).

3.2 Gene clustering analysis

Consensus clustering analysis was conducted on the TCGA GC dataset consisting of 335 samples. The CDF curve and the area under the curve changes indicated that the optimal number of clusters is 2 (Fig. 2A). All GC patients were divided into ARGcluster A and ARGcluster B. PCA significantly distinguished between these two molecular subtypes, indicating the reliability of the clustering (Fig. 2B). Survival analysis of the training cohort showed that the prognosis of the ARGcluster A group was significantly worse than that of the ARGcluster B group (p < 0.001) (Fig. 2C). Furthermore, there were significant differences in the expression of LMAGs between ARGcluster A and ARGcluster B groups (p < 0.05) (Fig. 2D). The heatmap of unsupervised clustering of the top 100 differentially expressed genes in GC patients also indicated distinct transcriptional profiles between ARGcluster A and ARGcluster B. Moreover, the analysis of the correlation between the two clusters and clinical characteristics indicated that the relationship of these top 100 DEGs with TNM stage, age, and gender was not significant (p > 0.05) (Fig. 3E). Next, we used "CIBERSORT" to analyze the relative proportions of 22 types of immune cells. As shown in Fig. 2F, 18 types of immune cells had different infiltration levels between ARGcluster A and ARGcluster B. There was a higher infiltration of immature B cells, natural killer T cells, natural killer cells, plasmacytoid dendritic cells, regulatory T cells, follicular helper T cells, and type 1 helper T cells in the ARGcluster A group, suggesting that immune infiltration may play an important role in poor prognosis.

3.3 Enrichment analysis

Functional annotation of differentially expressed genes was performed. The results showed that gene sets related to signaling, cardiovascular diseases, cellular outcomes, and metabolism were primarily enriched in ARGcluster A, while ARGcluster B was enriched not only in metabolism-related pathways but also in DNA replication and repair-related pathways (Fig. 2G). Gene set enrichment analysis indicated that metabolism-related gene sets such as "Calcium signaling pathway", "Dilated cardiomyopathy", "DNA replication", "ECM-receptor interaction" and "Focal adhesion" were predominantly enriched in ARGcluster B (Supplementary Fig. 1).

To avoid potential overfitting, we conducted LASSO Cox regression analysis (Fig. 3A). Ultimately, we selected 14 essential prognostic LMAGs for modeling based on the following formula:

Risk Score=[Expression (SERPINE1)×(0.159821454227009)] +[Expression (MLXIPL) ×(− 0.100918248708407)]+ [Expression (PDK4)×(0.11526650006906)]+

[Expression (BPI)×(− 0.176600664517854)]+[Expression (PROC) ×(0.146951174843476)] +[Expression (TNFRSF9) ×(− 0.429986834334003)]+[Expression (TYRP1) ×(0.132383812828114)] +[Expression (CALCR) ×(0.487315951765447)]+[Expression (TUBB3) ×(0.222312584421539)]+[Expression (CORIN) +[Expression (RTN4RL2) ×(0.284232367720243)]+[Expression (MMP11) ×(− 0.355369600424733)] ×(0.153006941051159)]+[Expression (PAEP) ×(0.22333853930601)] +[Expression (FOXD4) ×(− 0.299546648377551)].

Using the optimal cut-off values determined by the R package 'survminer' (version 0.4.9), each dataset was automatically divided into low-risk and high-risk groups. Kaplan-Meier analysis demonstrated that the prognosis of patients in the high-risk group was significantly worse than that of the low-risk group in the training set, test set, and entire dataset (p < 0.001) (Fig. 3B). The risk score prediction for 1-year, 3-year, and 5-year mortality rates were also evaluated. The area under the receiver operating characteristic curve (AUC) values for the training set were 0.702, 0.761, and 0.757, respectively. For the test set, the AUC values were 0.621, 0.638, and 0.634, respectively. For the entire dataset, the AUC values were 0.702, 0.761, and 0.757, respectively (Fig. 3C). Multivariate Cox regression analysis was used to validate the impact of clinical characteristics and risk scores on patient survival. The risk score was identified as an independent risk factor (Fig. 3D). As shown in Fig. 4E, patient age, N stage, and risk score were associated with the prognosis of GC patients. Age (HR = 1.03, 95% CI = 1.017–1.04, P < 0.001), N2 stage (HR = 1.88, 95% CI = 1.344–2.62, P < 0.001), N3 stage (HR = 2.59, 95% CI = 1.770–3.78, P < 0.001), and risk score (HR = 1.12, 95% CI = 1.085–1.16, P < 0.001) were all independent predictors of prognosis for GC patients. Unsupervised clustering analysis of lipid metabolism-related gene expression profiles based on risk grouping also showed differences between the two groups (Fig. 3E).

Additionally, a nomogram was developed using the identified independent predictors to forecast the 1-year, 3-year, and 5-year survival probabilities of GC patients in the training cohort. The cumulative risk curves and calibration curves indicated that the nomogram had a strong predictive capability for the prognosis of GC patients (Fig. 4A, B, C).

The results of the consensus clustering analysis were consistent with the risk scores, with ARGcluster A having significantly higher risk scores than ARGcluster B (p < 0.001) (Fig. 4D). The Sankey diagram showed that the majority of patients in ARGcluster A belonged to the high-risk group, while most patients in ARGcluster B were classified into the low-risk group (Fig. 4E). To predict the prognosis of GC patients, we developed a comprehensive evaluation model that included age, staging, TMB, and the nomogram. The nomogram demonstrated optimal predictive performance for 1-year, 3-year, and 5-year survival outcomes (Fig. 4F).

3.4 Immune infiltration

The tumor microenvironment is primarily composed of immune cells, extracellular matrix, various growth factors, inflammatory factors, and specific physicochemical characteristics, all of which significantly influence disease diagnosis, survival outcomes, and clinical treatment sensitivity. By analyzing the relationship between key genes and immune infiltration in the GC cohort, we further explored the potential molecular mechanisms by which these genes influence GC progression. Figures 5A and 5B showed the proportions of immune cells in each patient and the correlations between these immune cells, respectively. T cells CD4 memory activated and T cells CD8 had the highest positive correlation (r = 0.42), while T cells CD4 memory resting and T cells CD8 had the highest negative correlation (r=-0.51). Furthermore, the study identified seven significantly different immune cell types between the two groups, including five lymphoid cells (CD8 T cells, CD4 memory resting T cells, CD4 memory activated T cells, and follicular helper T cells) and two myeloid cells (M1 macrophages, and resting mast cells). This indicates that immune cells may play an important role in poor prognosis. Correlation network analysis of infiltrated immune cells within the two groups was conducted to further explore the immune cell types most associated with lipid metabolism. Most types of immune cells were negatively correlated with CD8 T cells (Fig. 5B, C).

The prognostic model can directly affect the content of various immune infiltrating cells in the tissue, such as B cells, M1 macrophages, M2 macrophages, monocytes, CD8 + T cells, and others (Fig. 5D, E). Comparative results of the tumor microenvironment between the two groups indicated that the stromal score of the low-risk group was significantly lower than that of the high-risk group (Fig. 5F), while the immune cell infiltration score was significantly higher in the low-risk group. There was no significant difference in tumor purity between the two groups, indicating that the differences in the tumor microenvironment between the two groups were mainly reflected in stromal cells and immune cells. Using the TISCH database for single-cell analysis of GC from the GSE134520 dataset (Fig. 6A). Expression of 13 LMAGs including SERPINE1, MLXIPL, PDK4, BPI, PROC, TNFRSF9, TYRP1, TUBB3, CORIN, RTN4RL2, MMP11, PAEP, FOXD4, except CALCR (Fig. 6B).

This study emphasizes the crucial role of LMAGs in the prognosis and tumor immune microenvironment of GC. By analyzing RNA-seq data from 335 GC transcriptomes, we identified 43 LMAGs significantly associated with survival, most of which were poor prognostic factors. Finally, we validated the expression, prognosis, and immune infiltration of fourteen key LMAGs in GC.

Lipid metabolism is a critical cellular process encompassing the synthesis, breakdown, and regulation of lipids, essential for maintaining cell membrane integrity, energy storage, and signaling pathways^[4]. In cancer cells, lipid metabolism often becomes dysregulated, leading to metabolic reprogramming that supports tumor cell rapid proliferation, survival, and metastasis^{[4, 5, 21]}. CNVs are structural changes in the genome that can increase the copy number of genes, our analysis indicated CNV prevalent gains in these genes within tumor samples, suggesting their amplification might contribute to oncogenic processes. Understanding these CNVs may help in the diagnosis, prognosis, and targeting of GC^[15].

Consensus clustering analysis further delineated GC into two distinct molecular subtypes, ARGcluster A and ARGcluster B, with significant differences in survival outcomes and gene expression profiles. ARGcluster A was associated with a poorer prognosis and exhibited higher stromal scores and increased infiltration of M1 macrophages, but lower infiltration of CD4 memory resting T cells, resting mast cells, and M2 macrophages. In this context, they may contribute to a chronic inflammatory environment that supports tumor progression and fails to adequately modulate the tumor microenvironment. Our findings align with and extend previous studies^{[4, 22, 23]}. The study by Dai et al. confirmed that the risk scores derived from six LMAGs through univariate and multivariate Cox regression analysis showed that the HRisk group had higher immune and stromal scores and lower tumor purity scores. ADH4, CYP4A11, and ST6GALNAC3 were found to be negatively correlated with M1 macrophages and positively correlated with M2 macrophages. These LMAG characteristics can effectively assess the prognosis of GC patients and reflect metabolic and immune status^[23]. Chen and his colleagues identified three molecular subtypes by analyzing the correlation between survival rates and LMAGs. Their immunological assessment revealed that the subtype with high expression of lipid metabolism genes exhibited a relatively elevated immune status. They constructed a predictive risk model using the lipid metabolism genes AKR1B1, MTF1, PLA2R1, GGPS1, and ETNPPL, which accurately predicted the prognosis of GC patients^[24].Our findings dovetail with these insights, demonstrating that LMAGs act as poor prognostic factors in GC, which is consistent with their roles in enhancing tumor aggressiveness and modulating metabolic pathways.

In our study, 14 essential prognostic LMAGs—SERPINE1, MLXIPL, PDK4, BPI, PROC, TNFRSF9, TYRP1, CALCR, TUBB3, CORIN, RTN4RL2, MMP11, PAEP, and FOXD4—are significantly correlated with GC prognosis. SERPINE1 is a key regulator of the plasminogen activation system, which is crucial for the degradation of the extracellular matrix (ECM). By inhibiting tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA), SERPINE1 can influence ECM remodeling, a process vital for tumor invasion and metastasis^[25]. The researchers found that SERPINE1 expression was significantly higher in GC tissues and was associated with advanced tumor stage and lymph node metastasis^[26]. MLXIPL regulates genes involved in glycolysis and lipogenesis, contributing to the metabolic flexibility of cancer cells^[27]. MLXIPL was identified as a gene expressed in response to CD8 + T cell infiltration, and its expression has been found to correlate with the prognosis of prostate cancer (PCa) ^[28]. PDK4, or Pyruvate Dehydrogenase Kinase 4, is an enzyme that plays a crucial role in regulating glucose metabolism by inhibiting the pyruvate dehydrogenase complex^[29]. This inhibition prevents the conversion of pyruvate to acetyl-CoA, thereby shifting cellular metabolism from aerobic respiration to anaerobic glycolysis^[30]. In the context of GC, PDK4 has been implicated in several key processes that support tumor development and progression^[31]. The related research indicated that PDK4 expression might regulate cell adhesion, metal ion transport, synaptic activity, and cancer cell metabolism in GC. Moreover, PDK4 was associated with the enriched expression of various immune cells^[32]. TNFRSF9 (Tumor Necrosis Factor Receptor Superfamily Member 9) is involved in immune regulation. Its expression can influence the tumor microenvironment by modulating the activity of T-cells and other immune cells^{[33, 34]}.TUBB3 is associated with microtubule dynamics and resistance to chemotherapy. Overexpression in GC is often linked to poor prognosis and drug resistance^{[35, 36]}. Tyrosinase-Related Protein 1 (TYRP1) is a significant gene associated with melanin biosynthesis and has been implicated in various cancers^[37]. Matrix Metallopeptidase 11 (MMP11), also known as stromelysin-3, is a member of the matrix metalloproteinase (MMP) family, which plays a critical role in the degradation of the extracellular matrix (ECM)^{[38, 39]}. This process is essential for normal physiological functions such as tissue remodeling and wound healing. The research by Hua et al. demonstrated that the overexpression of MMP11 in GC cells significantly contributes to the pathogenesis of the disease^[40]. Similarly, Wasenius's team found analogous results in papillary thyroid carcinoma, where the expressions of FN, MMP-11, and TIMP-1 were shown to play crucial roles in the early stages of tumor development, facilitating the survival of cancer cells within the stromal environment. These findings underscore the critical role of MMP11 and related proteins in tumor progression and the importance of the stromal environment in supporting cancer cell survival and proliferation^[41].

Despite these insights, our study has several limitations. The retrospective nature of the data from TCGA and GEO may introduce selection biases, limiting the generalizability of the findings across different populations. Additionally, the study focuses primarily on transcriptomic data without integrating other layers of genomic and epigenomic information. Future research should integrate genomic, epigenomic, and proteomic data to provide a more comprehensive understanding of the role of LMAGs in GC.

We analyzed the expression pattern of LMAGs in GC, leading to the construction of novel risk models and molecular subtypes. Notably, these risk models and molecular subtypes demonstrated high accuracy in predicting the prognosis of GC patients. Our findings highlight the crucial role of lipid metabolism pathways and the tumor immune microenvironment in the prognosis of GC. This study provides a robust foundation for developing future therapeutic strategies targeting these pathways.

Gastric cancer

LMAGs

Lipid metabolism-associated genes

TCGA

The Cancer Genome Atlas

GEO

Gene Expression Omnibus

CNV

Copy number variation

DEGs

Differentially expressed genes

TIME

Tumor immune microenvironment

Natural killer cells

TAM

Tumor-associated macrophage

FASN

Fatty acid synthase

ACC

Acetyl-CoA carboxylase

SREBP1

Sterol regulatory element-binding protein 1

CRC

Colorectal cancer

LASSO

Least absolute shrinkage and selection operator

ssGSEA

Single-sample Gene Set Enrichment Analysis

TME

Tumor microenvironment

tPA

Plasminogen activator

ECM

Extracellular matrix

TNFRSF9

Tumor Necrosis Factor Receptor Superfamily Member 9

TYRP1

Tyrosinase-Related Protein 1

MMP11

Matrix Metallopeptidase 11

Acknowledgements

None declared.

Funding

Not applicable.

Contributions

YNX: designed the study, performed the statistical analyses, and drafted the manuscript. QY: interpreted the data, and critically revised the paper. ZWZ: interpreted the data, and critically revised the paper. SZ: critically revised the paper, and prepared figures. ZW: edited the paper, and prepared figures. XL: edited the paper, prepared figures. HP: performed the statistical analyses, drafted the manuscript. ZQZ: conceived the paper. All authors reviewed the manuscript.

Corresponding authors

Correspondence to Zhiqiang Zhan or Yanan Xiao.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Van Cutsem E, et al. Gastric Cancer. 2016;388(10060):2654–64.
Joshi SS. and B.D.J.C.a.c.j.f.c. Badgwell. Curr Treat recent progress gastric cancer. 2021;71(3):264–79.
Luo X, et al. Emerg roles lipid metabolism cancer metastasis. 2017;16(1):76.
Tracey TJ et al. Neuronal lipid metabolism: multiple pathways driving functional outcomes in health and disease. 2018. 11: p. 10.
Broadfield LA et al. Lipid metabolism in cancer: New perspectives and emerging mechanisms. 2021. 56(10): pp. 1363–1393.
Beloribi-Djefaflia S, Vasseur S, Guillaumond FJO. Lipid metabolic reprogramming cancer cells. 2016;5(1):e189–189.
Gu X et al. Identification of lipid metabolism-associated genes as prognostic biomarkers based on the immune microenvironment in hepatocellular carcinoma. 2022. 10: p. 883059.
Wu H et al. Lipid droplet-dependent fatty acid metabolism controls the immune suppressive phenotype of tumor‐associated macrophages. 2019. 11(11): p. e10698.
Vasseur S, Guillaumond FJO. Lipids in cancer: A global view of the contribution of lipid pathways to metastatic formation and treatment resistance. 2022. 11(1): p. 46.
Pan Y et al. Tumor-associated macrophages in tumor immunity. 2020. 11: p. 583084.
Menendez JA, Lupu RJNRC. Fat acid synthase lipogenic phenotype cancer pathogenesis. 2007;7(10):763–77.
Flavin R et al. Fatty acid synthase as a potential therapeutic target in cancer. 2010. 6(4): p. 551–62.
Manning BD, Cantley LCJC. AKT/PKB signaling: navigating Downstr. 2007;129(7):1261–74.
Wen Y-A et al. Downregulation of SREBP inhibits tumor growth and initiation by altering cellular metabolism in colon cancer. 2018. 9(3): p. 265.
Hu B, Yang XB, Sang XTJCm. Construction of a lipid metabolism-related and immune‐associated prognostic signature for hepatocellular carcinoma. 2020. 9(20): p. 7646–62.
Ritchie ME et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. 2015. 43(7): pp. e47-e47.
Wilkerson MD, Hayes DNJB. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. 2010. 26(12): p. 1572–3.
Wu T et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. 2021. 2(3).
Zarrei M et al. A copy number variation map of the human genome. 2015. 16(3): p. 172–83.
Zhang F et al. Copy number variation in human health, disease, and evolution. 2009. 10: pp. 451–481.
Santos CR. A.J.T.F.j. Schulze. Lipid metabolism cancer. 2012;279(15):2610–23.
Li Z et al. Construction of a novel signature and prediction of the immune landscape in gastric cancer based on necroptosis-related genes. 2022. 12(1): p. 13290.
Dai J et al. Construction of a lipid metabolism-related and immune-associated prognostic score for gastric cancer. 2023. 16(1): p. 93.
Chen Y, et al. Bioinf Anal Struct gastric cancer prognosis model based lipid metabolism immune microenvironment. 2022;13(9):1581.
Yang J-D, Ma L, Zhu ZJJoC. SERPINE1 as a cancer-promoting gene in gastric adenocarcinoma: facilitates tumour cell proliferation, migration, and invasion by regulating EMT. 2019. 31(7–8): p. 408–18.
Chen S et al. SERPINE1 overexpression promotes malignant progression and poor prognosis of gastric cancer. 2022. 2022(1): p. 2647825.
Chang X et al. MLXIPL promotes the migration, invasion, and glycolysis of hepatocellular carcinoma cells by phosphorylation of mTOR. 2023. 23(1): p. 176.
Fan Y et al. MLXIPL associated with tumor-infiltrating CD8 + T cells is involved in poor prostate cancer prognosis. 2024. 15: p. 1364329.
Dou X et al. PDK4-dependent hypercatabolism and lactate production of senescent cells promotes cancer malignancy. 2023. 5(11): p. 1887–910.
Zhang S et al. The pivotal role of pyruvate dehydrogenase kinases in metabolic flexibility. 2014. 11: pp. 1–9.
Anwar S, et al. Target pyruvate dehydrogenase kinase Signal Dev effective cancer therapy. 2021;1876(1):188568.
Zhang Z et al. PDK4 constitutes a novel prognostic biomarker and therapeutic target in gastric cancer. 2022. 12(5): p. 1101.
Fröhlich A et al. Comprehensive analysis of tumor necrosis factor receptor TNFRSF9 (4-1BB) DNA methylation with regard to molecular and clinicopathological features, immune infiltrates, and response prediction to immunotherapy in melanoma. 2020. 52.
Wu J. and Y.J.E.j.o.p. Wang, Role of TNFSF9 bidirectional signal transduction in antitumor immunotherapy. 2022. 928: p. 175097.
Yu J et al. Serum levels of TUBB3 correlate with clinical outcome in Chinese patients with advanced gastric cancer receiving first-line paclitaxel plus capecitabine. 2012. 29: p. 3029–34.
Li M et al. Establishment and analysis of an individualized immune-related gene signature for the prognosis of gastric cancer. 2022. 9: p. 829237.
Ghanem G, Fabrice JJMo. Tyrosinase related protein 1 (TYRP1/gp75) in human cutaneous melanoma. 2011. 5(2): pp. 150–155.
Zhang X, et al. Insights into the distinct roles of MMP-11. tumor biology future Ther. 2016;48(5):1783–93.
Stamenkovic I, Ireland. Extracell matrix remodelling: role matrix metalloproteinases. 2003;200(4):448–64.
Deng H, et al. Matrix metalloproteinase 11 depletion inhibits cell proliferation. gastric cancer cells. 2005;326(2):274–81.
Wasenius V-M et al. Hepatocyte growth factor receptor, matrix metalloproteinase-11, tissue inhibitor of metalloproteinase-1, and fibronectin are up-regulated in papillary thyroid carcinoma: a cDNA and tissue microarray study. 2003. 9(1): p. 68–75.

No competing interests reported.

SupplementaryMaterialsGC6.19.docx

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

WITHDRAWN: Developing a Risk Model of Lipid Metabolism-associated Genes: Implications for Clinical Prognosis and Tumor Immune Microenvironment in Gastric Cancer

Status:

Version 1

Editorial Note

Abstract

Background

Methods

Results

Conclusion

Figures

1. Introduction

2. Methods

2.1 Data collection and processing and acquisition of lipid metabolism gene list

2.2 Consensus clustering analysis and principal component analysis

2.3 Functional enrichment analysis

2.4 Identification of prognostic lipid metabolism-associated genes and the prognostic model

2.5 Clinical prediction value of the established prognostic model

2.6 Immune microenvironment analysis and construction of the nomogram

2.7 Statistical analysis

3 Results

3.2 Gene clustering analysis

3.3 Enrichment analysis

3.4 Immune infiltration

4 Discussion

5 Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1