Analysis of immune status and prognostic model incorporating lactic acid metabolism-associated genes

Abstract

Background

Cancer development is intricately linked with metabolic dysregulation, including lactic acid metabolism, which plays a pivotal role in tumor progression and immune evasion. However, its specific implications in gastric adenocarcinoma (STAD) remain unclear. This study introduces a novel methodology to evaluate lactic acid metabolism comprehensively in STAD, aiming to elucidate its prognostic significance and impact on immunotherapy efficacy. Targeted therapies directed at key lactic acid metabolism genes (LMGs) identified within the tumor microenvironment (TME) hold promise for personalized treatment strategies.

Methods

Lactic acid metabolism patterns were assessed in 415 STAD patients using a panel of 21 LMGs. Cox regression and Lasso regression analyses were employed to develop a predictive risk model based on differentially expressed genes (DEGs). Validation of the model was conducted using independent cohorts from the GEO and TCGA databases, as well as additional datasets focused on immunotherapy responses. Further investigations into TME dynamics of lactic acid metabolism included functional assays targeting SLC16A3, a pivotal gene identified through our analyses.

Results

Patients were stratified into distinct risk groups based on their lactic acid metabolism profiles. Low-risk patients exhibited attenuated lactic acid metabolism, correlating with favorable clinical outcomes characterized by prolonged survival and enhanced responsiveness to immunotherapy. Notably, tumor cells within the TME demonstrated heightened levels of active lactic acid metabolism, particularly impacting tumor-infiltrating lymphocytes such as CD8 + T cells and regulatory T cells. Mechanistically, SLC16A3 emerged as a critical regulator promoting STAD cell proliferation, invasion, and migration while modulating the metabolic landscape.

Conclusion

This study underscores the prognostic value of a lactic acid metabolism-based model in STAD, providing insights into its potential as a predictive biomarker for patient stratification and therapeutic targeting. The findings highlight SLC16A3 as a promising candidate for therapeutic intervention aimed at modulating lactic acid metabolism in the TME, thereby advancing personalized treatment strategies in gastric cancer management.

Introduction

Stomach adenocarcinoma (STAD) constitutes the predominant histological subtype of gastric cancer (GC), a global health challenge with significant regional disparities in incidence and mortality. East Asia notably exhibits higher rates compared to Western countries [1]. Global Cancer Statistics 2020 reported approximately 1.2 million new cases annually worldwide, highlighting the complex etiology involving factors such as Helicobacter pylori infection, dietary habits, and alcohol consumption [2, 3]. Recent therapeutic advancements, including regimens like 5-FU combined with platinum compounds, PD-1/PD-L1 blockade, and targeted therapies, have improved outcomes for select patient populations [4,5,6]. However, the intricate molecular underpinnings of GC remain elusive, necessitating novel prognostic approaches aligned with clinical needs [7,8,9].

Gastric cancer cells exhibit a metabolic preference termed the “Warburg Effect,” characterized by enhanced glucose metabolism and lactic acid production even in the presence of oxygen, a feature central to their malignant phenotype [10, 11]. Accumulated lactic acid within the tumor microenvironment (TME) fuels tumor progression by promoting angiogenesis, facilitating invasion, metastasis, and contributing to therapeutic resistance through TME acidification [12, 13].

Moreover, lactic acid in the TME exerts profound immunosuppressive effects by inducing VEGF and Arg 1 expression, promoting M2 macrophage polarization, and impairing dendritic cell function, thus compromising effective antitumor immune responses [14,15,16]. Suppressed cytotoxic T cell activity further hampers immune surveillance, underscoring lactic acid's critical role in shaping the immunosuppressive TME [17,18,19].

Understanding the specific roles of lactic acid metabolism-related genes (LMGs) in GC progression and their interaction with tumor immunity holds promise for developing targeted therapies. This knowledge may pave the way for integrated therapeutic strategies aimed at enhancing immunotherapy efficacy in gastric cancer [20]. Thus, elucidating the distinct characteristics of lactic acid metabolism in the TME is essential for advancing personalized treatment approaches and improving clinical outcomes in GC management.

Materials and methods

Data collection

RNA-seq transcriptome data (FPKM values), somatic mutation data, copy number variant (CNV) files, overall survival (OS) data, and corresponding clinicopathological information for stomach adenocarcinoma (STAD) were retrieved from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov). Clinical parameters and normalized gene expression data were collected from the Gene Expression Omnibus (GEO) repository, specifically dataset GSE84337 (https://www.ncbi.nlm.nih.gov/geo/). Human gene profiles were used to annotate transcriptome data, excluding samples lacking essential clinical or survival information. The final cohort included 50 normal and 365 STAD samples. Additionally, 21 lactic acid metabolism-related genes (LMGs) were obtained from the MSigDB database (Hallmark Gene set) [21].

Consensus clustering analysis of LMGs

K-means algorithms in the ConsensusClusterPlus package were used to classify distinct lactic acid metabolism-related patterns. The optimal number of clusters was determined through 1000 iterations for stability and consistency. Gene set variation analysis (GSVA) with KEGG gene sets (c2.cp.kegg.v7.4) was employed using the GSVA package to identify biological functional differences in LMGs. Visualization of clustering results utilized the Factoextra and FactoMineR packages.

Relationship between molecular patterns of lactic acid metabolism and clinical characteristics

The clinical significance of identified molecular clusters was assessed by analyzing their association with age, gender, TNM stage, and survival outcomes of STAD patients. Kaplan–Meier curves were generated using the Survival and Survminer packages to compare overall survival between different molecular patterns.

Correlation between STAD molecular patterns and tumor microenvironment

The ESTIMATE algorithm estimated immune, stromal, and tumor cell proportions in STAD tissues based on transcriptome data. Immune and stromal scores were calculated using single-sample gene set enrichment analysis (ssGSEA), and the CIBERSORT algorithm quantified immune cell subtype abundance. The correlation between molecular patterns and immune checkpoint expression (PD-1, PD-L1, CTLA-4) was investigated [22].

Identification of DEGs and functional enrichment analysis

Differential expression analysis (|log2-fold change (FC)|≥ 1 and p-value < 0.05) using the limma package identified DEGs between lactic acid metabolism subgroups. Functional enrichment analysis utilized Gene Ontology (GO) and KEGG pathway analysis from the clusterProfiler package.

Development of prognostic lactic acid metabolism associated genes score (LMG_Score)

DEGs from distinct lactic acid metabolism clusters were analyzed using DESeq2 (|log2-fold change (FC)|≥ 1 and p-value < 0.05). Univariate Cox regression identified survival-associated genes, and PCA generated the LMG_Score using the psych package. The formula used was: LMG_Score = Σ(gene expression [i] × corresponding coefficient [i]) [23].

Clinical significance and classification analysis of prognostic LMG_Score

UniCox and multiCox regression analyses assessed whether LMG_Score was an independent prognostic predictor in STAD patients. Subgroup analysis evaluated predictive performance across clinical variables. Immune infiltration and immune checkpoint expression were compared across LMG_Score subgroups using CIBERSORT. Correlations with tumor mutation burden (TMB), microsatellite instability (MSI), and cancer stem cell (CSC) scores were examined.

Development of predictive nomogram

A nomogram was constructed using the survival and rms packages to predict clinical outcomes (1-, 3-, and 5 year overall survival) based on LMG_Score and clinicopathological characteristics. Calibration curve analysis and decision curve analysis (DCA) validated its clinical utility [24].

Mutation profiles and chemotherapy response in STAD

Mutation profiles and chemotherapy responses in different lactic acid metabolism risk groups were analyzed using the maftools package. TIDE and IPS assessed immune-related features. Drug sensitivity (IC50) predictions were made using pRRophetic.

Experimental validation

Human gastric cancer cell lines (SNU1 and HGC 27) were cultured in RPMI 1640 supplemented with 10% FBS, penicillin, and streptomycin. Real-time PCR and Western blot confirmed SLC16A3 knockdown effects. 8 µm pore size Transwell filters (6.5 mm diameter; Corning Inc., Corning, NY, USA) was applied to evaluate cell invasive capacity. Cells suspension (1 × 105/well) was carefully placed into the upper chamber with 100 µL FBS-free RPMI 1640 medium. Concurrently, the lower chamber was supplemented with 600 µL RPMI 1640 medium containing 20% FBS. After 24 h of incubation in 5% CO2 at 37 °C, the invasive cells were fixed and stained with crystal violet and then were photographed by bright field microscopy and analysed with ImageJ software. The average cell count was ascertained from five distinct random fields of view.

Statistical analysis

R software (version 4.1.2) conducted statistical analyses. Student's t-test (for normally distributed variables) and Wilcoxon rank sum test (for non-normally distributed variables) compared groups. Kaplan–Meier survival analysis (Survminer package) assessed survival outcomes. Results were significant at p < 0.05.

Results

Landscape of genetic mutations in lactic acid metabolism genes (LMGs) in gastric cancer (GC)

We initially analyzed the gene expression profiles of 21 LMGs sourced from MSigDB in tumor and corresponding normal tissue samples from the TCGA-STAD dataset. Among these, 15 differentially expressed genes (DEGs) were identified, with all except LDHD and SLC16A7 showing significant overexpression in tumor tissues (Fig. 1A). Protein–protein interaction (PPI) analysis using String identified LDHA, PFKFB2, HIF1A, SLC16A3, SLC16A1, and EMB as hub genes (Fig. 1B). A waterfall plot depicted somatic mutations across these 21 DEGs, with 54.06% of GC samples showing genetic mutations, notably in TP53 and PER2 (Fig. 1C). Additionally, analysis of copy number variations (CNVs) in these LMGs indicated substantial alterations (Fig. 1D), suggesting a regulatory role in LMG expression through CNVs. A prognostic network diagram illustrated significant correlations between 18 survival-related LMGs (Fig. 1E). Kaplan–Meier and Cox analyses identified predictive values for these LMGs in GC patients (Fig. 2A, B). Our study of 804 GC patients from TCGA-STAD and GSE84437 highlighted significant genomic and expression differences in LMGs between cancerous and normal tissues, implicating their role in GC tumorigenesis.

Fig. 1

Landscape in Genetic Mutation of LMGs in STAD. A Expression distributions of DEGs between GC and normal tissues. B The PPI network of DEGs from the STRING database. C Genetic alteration regarding LMGs query. D Frequencies of CNV gain, deletion, and non-CNV in LMGs. E The prognostic network of survival-related LMGs

Fig. 2

Prognostic value of LMGs in GC patients. A Kaplan–Meier analysis of 21 LMGs in GC patients. B Univariate Cox analysis of 21 LMGs in GC patients

Construction of lactic acid metabolism clusters in GC

To explore associations between survival-related LMG expression patterns and GC subtypes, we performed consensus clustering, dividing 513 STAD patients into optimal clusters A (n = 430) and B (n = 378) based on LMG expression profiles (Fig. 3A). Principal component analysis confirmed distinct intergroup distributions (Fig. 3B). Cluster A exhibited improved prognosis, while Cluster B showed poorer overall survival (Fig. 3C, P = 0.027). Notable differences were observed in LMG scores between clusters (Fig. 3D), with minor variations in clinicopathological characteristics (Fig. 3E). Most LMGs, except for SLC16A3 and SLC16A7, showed higher expression levels in Cluster A (Fig. 3F). Functional enrichment analysis revealed significant involvement of these clusters in pathways related to carcinogenesis, metabolism, and immune modulation (Fig. 4A–C).

Fig. 3

Subgroups and clinicopathological and biological characteristics of two different subtypes of LMGs by consistent clustering. A Heat map defining the consensus matrix for two clusters (k = 2) and their correlation regions. B PCA analysis of the transcriptome of two subgroups. C Kaplan–Meier curve of two subgroups. D Differences in LMGs-score between the two clusters. E Differences in clinicopathological characteristics and expression levels of LMGs between two different subgroups. F Comparison of LMGs-related gene expression levels between the two subgroups

Fig. 4

Identification of gene subgroups based on DEGs. A, B KEGG and GO enrichment analysis of DEGs in two lactate metabolism subgroups. C GSVA of biological pathways between two different subgroups

Development and validation of prognostic LMG score

To establish a prognostic link, we integrated clinical parameters into a nomogram predicting 1-, 3-, and 5 year OS for STAD patients (Fig. 5A, B). PER2, SLC16A3, and SLC16A7 emerged as top prognostic genes (Fig. 5B). Patients were stratified based on their LMG scores, revealing a correlation between higher scores and poorer outcomes (Fig. 5C, D). Validation in training and testing cohorts underscored the robustness of LMG scores in risk stratification (Figs. 5F–K, 6A–F). High AUC values validated their efficacy in predicting survival probabilities (Fig. 6D).

Fig. 5

Development and Validation of Prognostic Scores for LMGs. A Constructing the nomogram for predicting patient prognosis. B Accuracy of predicting 1-, 3-, and 5 year ROC curves for the entire cohort. C–K Ranked dot plots and scatter plots of the distribution of LMGs scores and patient survival and the expression patterns of the 3 selected genes in all patients (C–E), training set (F–H) and test set (I–K)

Fig. 6

Assessment of LMGs-score effectiveness. A–C LMGs-score for predicting prognostic accuracy for all patient groups, test group, and train group patients. D–F Applying ROC curves to predict predictive efficacy in the three subgroups mentioned above

Tumor microenvironment (TME) characteristics in different LMG clusters

CIBERSORT analysis unveiled correlations between LMGs and immune cell profiles, with M0 and M2 macrophages positively associated with LMG scores (Fig. 7A). Conversely, T cell subsets and B cells showed negative associations (Fig. 7A). Stromal and immune scores correlated positively with LMG scores (Fig. 7B). Cluster A displayed higher levels of activated T cells compared to Cluster B (Fig. 7C). Genes in the prognostic signature correlated closely with immune cell enrichments (Fig. 7D).

Fig. 7

Evaluating TME characteristics of different clusters. A Correlation between LMGs scores and immune cell types. B Correlations between LMGs-score and both immune and stromal scores. C Enrichment of immune cells in different LMGs clusters. D Association between the selected genes and the enrichment of immune cells

Microsatellite instability (MSI), cancer stem cell (CSC) score, and drug sensitivity analysis

High TMB and MSI correlated with improved immune responses and potential benefits from immune checkpoint inhibitors (Fig. 8A, B). Low LMG scores associated with MSI-H, suggesting responsiveness to immunotherapy (Fig. 8B). Positive correlations between LMG and CSC scores highlighted stem cell characteristics (Fig. 8C). Somatic mutation analysis revealed distinct patterns between LMG score groups (Fig. 8D, E). Drug sensitivity analysis indicated associations between LMGs and chemotherapy responses (Fig. 8F–H).

Fig. 8

Exploring the potential usage of LMGs score for gastric cancer treatment. A Relationships between LMGs score and TMB. B Relationships between LMGs score and MSI. C Relationships between LMGs score and CSC index. D, E The waterfall plot of somatic mutation features established with high and low LMGs scores. F–H Analysis of the relationship between LMGs scores and sensitivity to the drugs

Functional role of SLC16A3 in gastric cancer

Further investigating the functional role, we focused on SLC16A3. We knocked down SLC16A3 in SNU1 and HGC 27 cell lines. The interference efficiency of three small interfering RNAs (SLC16A3-si1, si2, and si3) was measured by qRT-PCR and western blotting (Fig. 9A and Fig. 9B). Next, we evaluated the proliferation and migration of SNU1 and HGC 27 cells after SLC16A3 knockdown. A CCK-8 assay revealed downregulation of SLC16A3 significantly inhibited proliferation ability of both SNU1 and HGC 27 cells (p < 0.05; Fig. 9C). The knockdown of SLC16A3 significantly suppressed SNU1 and HGC 27 cell migration in the transwell migration assay (Fig. 9D and Fig. 9E). Meanwhile, in the invasion experiment, knocking down SLC16A3 also effectively inhibited cell invasion of SNU1 and HGC 27 cells (Fig. 9F and Fig. 9G). Quantitative and statistical analyses uncovered significant differences (p < 0.05) between the siRNA control group and siSLC16A3 groups after the experiments were repeated three times.

Fig. 9

Verify the Biological functions of gene SLC16A3. A mRNA Expression level of SLC16A3 in gastric cancer cell lines SNU1 and HGC 27. B Validation of SLC16A3 knockdown efficiency. C The CCK8 experiment demonstrated that knocking down SLC16A3 inhibited cell proliferation. D, E The migration experiment demonstrated that knocking down SLC16A3 inhibited cell migration. Figure E showed corresponding statistical analysis. F, G The invasion assay demonstrated knocking down SLC16A3 inhibited cell Invasion. Figure G implied corresponding statistical analysis

Analysis of SLC16A3 expression revealed associations with immune checkpoints and chemokines (Fig. 10A–I), highlighting its potential as a therapeutic target in GC.

Fig. 10

Regulation of cellular immunity and inflammation by SLC16A3. A Relationship between SLC16A3 expression and tumor immune checkpoints. B-I Expression of SLC16A3 in association with chemokines

Moreover, our exploration of drug sensitivity profiles highlighted significant disparities in IC50 values between cohorts stratified by SLC16A3 expression levels (Fig. 11A–I). This suggests that SLC16A3 expression could serve as a predictive biomarker for therapeutic response variability in GC patients, guiding personalized treatment strategies.

Fig. 11

Relationships between SLC16A3 expression level and medicine sensitivity

These insights into the complex interactions of SLC16A3 in GC underscore its clinical relevance as a potential biomarker for immune modulation and therapeutic targeting, emphasizing the need for further investigation into its precise mechanistic roles in tumor progression and treatment resistance.

Discussion

Metabolic reprogramming stands as a pivotal hallmark of malignancies, introducing significant heterogeneities in tumor tissues compared to normal counterparts, particularly within lactic acid metabolism [25, 26]. Lactic acid, a prominent metabolite, plays crucial roles in shaping the immunosuppressive tumor microenvironment and promoting malignant progression, rendering it a promising target for therapeutic intervention in cancer immunoregulation [27, 28].

Tumor cells strategically export lactic acid into the microenvironment, acidifying surroundings and establishing metabolic interdependencies that sustain rapid tumor growth [29, 30]. Moreover, the metabolic by-products from lactic acid metabolism serve not only as crucial energy sources but also modulate protein expression and epigenetic modifications, thereby influencing cellular signaling pathways and gene expression [31, 32].

Extensive evidence underscores the intricate relationship between lactic acid metabolism and tumor immunosuppression [33]. On one hand, it impedes immune surveillance by recruiting excessive immunosuppressive cells to the tumor microenvironment niche, while on the other hand, it promotes malignant progression by inducing and recruiting cells and molecules that contribute to immunosuppression. Consequently, targeting lactate metabolism and its transporters presents a promising avenue for cancer therapy [34].

Despite existing research, most studies tend to focus on individual lactic acid metabolism genes (LMGs) or specific immune cell subtypes. Therefore, there is a pressing need to comprehensively elucidate how different combinations of LMGs collectively regulate the tumor microenvironment and its immune filtration characteristics [35].

In this study, we utilized transcriptomic data from TCGA and GEO databases to analyze STAD patients, identifying 21 differentially expressed LMGs related to lactate metabolism. Our findings revealed prevalent gene mutations among these LMGs in more than 50% of cases, with significant upregulation correlating with poorer prognosis in STAD patients [36]. Using unsupervised clustering, we categorized STAD patients into two distinct lactic acid metabolism subgroups (clusters A and B), revealing unique clinicopathological characteristics and immune activities.

Further analysis incorporated Cox regression and LASSO regression models to develop an LMG score that quantified subgroup attributes related to lactic acid metabolism. The high LMG score subgroup exhibited heightened genomic instability and worse clinical outcomes, including lower overall survival rates. Functional enrichment analysis highlighted pathways associated with mitochondrial activity and metastasis, corroborating the clinical significance of the LMG score in predicting patient outcomes.

Moreover, our study demonstrated that high LMG scores correlated with a repressed immune microenvironment characterized by increased infiltration of inhibitory immune cells such as regulatory T cells, follicular helper T cells, γδ T cells, CD4 + T cells, and CD8 + T cells [27, 35, 37]. This immunosuppressive milieu underscored the potential of lactate metabolism as a key regulator of tumor immunoregulation in STAD [38].

Furthermore, drug sensitivity analysis identified potentially effective treatments tailored to different LMG score subgroups, offering new strategies to mitigate therapeutic resistance. Importantly, our findings suggest that LMG score may serve as a predictive biomarker for immunotherapy response, facilitating personalized treatment decisions in clinical practice [39, 40].

However, our study has limitations, including its retrospective nature and inherent biases associated with bioinformatics analyses. Future studies should incorporate additional clinical variables and prospective designs to validate the robustness and clinical utility of the LMG score. Comprehensive in vivo and in vitro experiments are warranted to further elucidate the mechanistic roles of lactic acid metabolism in tumor progression and its interaction with the tumor microenvironment.

In conclusion, our research underscores the clinical relevance of LMG score as a powerful prognostic indicator in STAD, offering insights into its potential as a therapeutic target for tumor immunoregulation and personalized cancer treatment strategies.