Fatty acid β-oxidation-associated prognostic model in gastric cancer and functional validation of IL-6 as a potential therapeutic target Authors: Chao Qu, Xuetao Yuan, Shuting Yang, Yifan Qiao, Renjianzhi Zhang, Yunhua Wu, Mengke Zhu, Jiayin Du, Gan Li, Rui Zhang, Xuejun Sun, Xuqi Li Published: 2026 ABSTRACT AbstractGastric cancer (GC) poses a significant health threat, and alterations in Fatty acid β-oxidation (FAO) may influence its progression. However, the precise mechanisms underlying this association remain unclear. FAO-related genes were analyzed using transcriptomic datasets from databases of GEO and TCGA. Totally 160 FAO-associated genes were identified, and a risk scoring model was subsequently established to stratify patients into groups of low- and high-risk. Immune characteristics, drug sensitivities, and hub genes, including IL-6, were assessed. Subsequently, immunoblotting and immunohistochemistry were performed on GC cell lines and tissue samples to evaluate IL-6 expression. Analysis of the TCGA and GEO databases revealed a FAO-related gene signature comprising ACADS, ACO2, CPT2, SLC22A5, AOC3, CD36, CIDEA, G0S2, GABARAPL1, and SERINC1. We also examined gene mutations and constructed a prognostic risk scoring model with validation achieved through a nomogram to predict gastric cancer risk. Immune infiltration analysis and drug sensitivity testing (e.g. AG-014699, Axitinib, BX-795, and Cisplatin) were also conducted. IL-6 emerged as a core gene with significant expression difference across cellular and tissue levels. FAO plays a critical role in the prognosis of GC, and IL-6 may serve as a key biomarker for diagnosis and therapeutic strategies.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-50112-1.Keywords: Fatty acid β-oxidation, Gastric cancer, IL-6, Prognosis modelSubject terms: Biomarkers, Cancer, Computational biology and bioinformatics, Oncology Introduction Gastric cancer (GC) remains a major global health challenge, ranking among the foremost contributors to cancer-associated mortality worldwide1,2. Currently, management of GC mainly relies on surgical resection together with perioperative combined chemotherapy and molecular targeted therapy, which, however, brings limited clinical benefits for patients. Thus further improvement in studies on treatment of GC is needed3–7. Fatty acid β-oxidation (FAO) constitutes the central process in cellular lipid metabolism, in which cells gradually break down long-chain fatty acids with energy released, and was known as a key biochemical process that regulates the abundance of ATP and NADPH8,9. Accumulating investigations have shown the pathogenic involvement of high FAO in many kinds of malignant tumors10,11. In GC, promoted proliferation and epithelial-mesenchymal-transition (EMT) are the results of FAO, together with peritoneal metastasis12. Interestingly, interventions targeting the FAO process have exhibited effective inhibition of tumor progression9, which shows potential value in treatment of GC. Interleukin-6 (IL-6), an inflammatory mediator synthesized by heterogeneous cellular populations, has been proved to be an pro-tumor factor during tumor progression13. Significant anti-tumor effects have been observed in IL-6 inhibiting treatment14. Researches have reported that elevated FAO levels are related to increased IL-615,16. Based on this, we speculate a potential correlation between high FAO levels and IL-6 levels, exploring this relationship may help discover new targeted treatment strategies and improve clinical benefits for patients. In this study, a gene set related to FAO through preliminary screening were obtained, based on which a FAO related GC prognosis model (Nomogram) is constructed with bioinformatics analysis of clinical data. Our work shows good predictive performance. We further explore transcriptome data and confirmed that IL-6 may contribute to GC progression and may be linked to FAO-related biological processes. Subsequent molecular biology experiments verify this conclusion. Method Patients and histological samples All GC patients (May 2022-June 2024) in the First Affiliated Hospital of Xi’an Jiaotong University underwent histopathologically confirmed definitive diagnosis, adhering to standardized histopathological criteria established by WHO Tumor Classification: Digestive System Tumors (Fifth Edition, 2020). All participants have signed informed consent forms. The research design has obtained formal ethical approval by the Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University. The ethics authorization number is LLSBPJ-2025-447. The research is in accordance with Helsinki Declaration. Western Blotting When GES-1, HGC27, SNU216, MKN45 and AGS cells are cultured to an appropriate density. Total protein was extracted from the cells or tissues. The protein concentration quantification was conducted via BCA method and was separated under electrophoresis in SDS-PAGE gel. Subsequently, the expression of the target protein was determined through steps such as membrane transfer, blocking, antibody incubation, and signal detection. Further confirm the expression of the target protein by comparing its molecular weight with the standard protein marker. Quantitative analysis of protein expression levels using image analysis software. Immunohistochemical staining Preparation of tissue sections: Cut 4–5 µ m thick slices from tissue samples and fix them on glass slides. Removal of embedded materials and antigen repair: Paraffin embedding is removed through steps such as dewaxing and hydration, and antigen repair steps are used to restore antigen epitopes that may have been lost during the fixation process. Blocking, antibody incubation, and colorimetric reaction: By blocking non-specific binding sites, specific antibodies are incubated, and the expression of the target protein is observed through colorimetric reaction. Analyze the degree of staining under a microscope to evaluate the localization, intensity, and distribution of the target protein. Preparation and processing of FAO data We obtained transcriptome data and clinical data through TCGA and GEO databases. Specifically, 448 cases of data (n = 36, healthy gastric tissue; n = 412 cases of tumor tissue) were acquired from the TCGA database. In addition, clinical data of GC patients, including age, gender, TNM staging, and stage information, are also downloaded. Obtain mutation information in simple nucleotide variation in TCGA. (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). In the GEO database, we searched and selected the dataset GSE84437 to obtain transcriptome and clinical information of 483 GC patients, including at least age, gender, and follow-up data. We integrated information from the GO term annotations and MSigDB-curated pathways (GSEA resource, https://www.gsea-msigdb.org) regarding the genes involved in fatty acid β-oxidation, and performed intersection analysis with the differentially expressed genes obtained from the transcriptome data. Ultimately, we included 160 differentially expressed FAO related genes. (https://www.ncbi.nlm.nih.gov/). Analysis of differential expressed gene related to FAO Cluster analysis of FAO related genes using pheatmap and limma17 packages. Compare the differential expression profiling across normal samples and tumor samples by calculating the mean expression level of the samples and retaining non-zero terms. Intersection the differential gene data from GEO and TCGA databases to obtain the differential gene expression levels in both databases. Perform cluster analysis using cluster Profiler18,19 and perform KEGG 20–22 and GO enrichment analysis using GO plot and enrich plot packages. Use Gene Set Variation Analysis (GSVA) 23 to perform cluster differential analysis on model functionality, while analyzing the differential expressed genes between low and high-risk groups, and obtaining risk differential genes. Set the condition as |log2FC|> 0.585 and P < 0.05. Construction of predictive risk model Univariate Cox regression analysis was utilized to identify differentially expressed genes associated with prognosis, and these genes were then displayed through forest plots. Perform prognostic analysis on clinical data by combining univariate and multivariate COX analysis24, and evaluate the significance of clinical indicators in prognosis. Tumor mutation burden (TMB) was calculated obtain the tumor mutation waterfall plot. The construction of the risk model uses TCGA data as the training group and GEO data as the validation group. Construct a prognostic model utilizing LASSO regression analysis to minimize the risk of overfitting, and calculate the patient’s risk score on the basis of gene expression levels as well as regression coefficients. Specificly, the risk scores were calculated by: Among them, ai is the weight coefficient of each gene, and bi is the expression level of each gene. Draw ROC curve using timeROC package, evaluate model’s the ability of prediction, and calculate AUC value. ROC analysis for evaluation of the model’s sensitivity and specificity were carried out after grouping based on clinical indicators and survival time. Principal component analysis By constructing a risk model and calculating the risk difference between the FAO gene and the model gene, two PCA analyses were conducted separately. Visualize PCA results using ggplot2 package to evaluate the distribution of fatty acid β—oxidation genes and model genes in different samples. Survival analysis By designing a risk model and extracting patient survival data, survival curves were plotted using the Survival and Survminer packages to evaluate the overall survival (OS) and progression free survival (PFS) of patients. Divide patients into high-risk and low-risk groups based on risk scores, and compare the survival differences between the two groups. Immune evaluation By analyzing the immune subtypes in pan cancer using ggpubr package, the immune subtypes were calculated. By utilizing the CIBERSORT package25, the scores of 22 types of immune cells in tumor microenvironments were evaluated, together with and athe differential analysis graphs involving number and functions of immune cell were finally obtained. Immune evasion of tumors was assessed via Tumor Immune Dysfunction and Exclusion score (TIDE)26. Calculate the efficacy score of immune therapy through TNN analysis. Drug sensitivity analysis Perform correlation analysis and sensitivity analysis on drugs in the drug database using limma, ggpubr, pRRophetic (Genomics of Drug Sensitivity in Cancer, available at https://www.cancerrxgene.org/), and ggplot2 packages. By determining the relationship between the IC50 index of drugs and patient risk, a p-value < 0.001 is set as the significance standard27. Statistical analysis Statistical analyses were performed with R software (v4.3.1) in this study. Kaplan Meier method is used for generating survival curves, while COX regression analysis is used for univariate and multivariate analysis. LIMMA package was utilized for differential analysis and ROC analysis in evaluating the predictive ability of scores. The statistical significance level is set to p < 0.05. Results Identification and analysis of FAO related genes in GC We first obtained the FAO-related gene set from the MSigDB database in GSEA and identified differentially expressed genes (DEGs) in GC using transcriptome data from the TCGA database. The intersection of these two datasets yielded 160 overlapping genes associated with FAO and GC (Supplementary Table S1). Subsequently, we retrieved transcriptome data from TCGA (448 samples, including 412 tumors and 36 normal gastric tissues) and GEO (483 tumor samples). A heatmap and volcano plot were constructed to visualize the expression difference of FAO-related genes across normal and tumor tissues (Fig. 1a, b). Cox regression analysis identified 10 genes significantly associated with OS (p < 0.05), including ACADS, ACO2, CPT2, and SLC22A5, which were positively correlated with OS, while AOC3, CD36, CIDEA, G0S2, GABARAPL1, and SERINC1 were negatively correlated with OS (Fig. 1c). Mutation analysis of these genes revealed that 45 out of 431 GC cases (10.44%) had mutations in the prognostic genes, with mutation frequencies of AOC3 (2%), ACO2 (2%), CPT2 (2%), GABARAPL1 (1%), CD36 (1%), SERINC1 (1%), ACADS (1%), CIDEA (1%), G0S2 (0%), and SLC22A5 (0%). No significant co-occurrence or mutual exclusivity was observed among these mutated genes (Fig. 1d, e). (a, b) Heatmap (a) and volcano plot (b) of differential expressed FAO-related genes; (c) Cox regression analysis of 10 prognostic related genes; (d–e). Tumor mutational analysis shows mutation status (d) and correlation (e) of prognostic genes. Validation of a prognostic model on GC Based on the prognostic genes identified through Cox analysis, Lasso regression was conducted to construct a prognostic model containing 7 FAO-related characteristic genes (ACADS, ACO2, CD36, CPT2, G0S2, SERINC1, and SLC22A5) (Fig. 2a), which was used to establish a GC risk score (Supplementary Table 2, Supplementary Fig. S1). The TCGA gastric cancer cohort constituted the training set, whereas the GEO dataset (GSE84437) underwent external validation. principal component analysis (PCA) revealed distinct separation between patients of low-risk and high-risk groups stratification by FAO- Risk Score (Fig. 2b, c). Patients of high risk had significantly worse OS and PFS than low-risk patients (Fig. 2d). Univariate and multivariate Cox regression analyses revealed that both the risk score and clinical stage were independent prognostic factors for OS (p < 0.001), further confirming that the FAO-based risk score is an independent predictor of patient outcomes (Fig. 2e). The FAO-Risk Score demonstrated better predictive performance than clinical factors such as age, with an area under the curve (AUC) of 0.741 (Fig. 2f). Interestingly, within the TCGA cohort, higher FAO-Risk Scores were observed in patients with later tumor grades and more advanced TNM stages, showing an increasing trend (Fig. 2g, h). Overall, the FAO-Risk Score is strongly associated with clinical features and demonstrates excellent predictive ability for the prognosis of GC patients. (a) Construction of a prognostic model containing 7 FAO related characteristic genes by lasso regression analysis; b-c. Principal component analysis for patients divided in low- and high- risk groups based on FAO-risk score, data of training cohort was from TCGA (b), and data of GC(GSE84437) was used as external cohort (c); (d) OS and PFS analyses of low and high risk group; e. Univariate and multivariate Cox regression analyses on the risk score, age, gender and clinical stage; (f) Evaluation of predictive ability on survival rates using Risk Score; g-h. Subgroup analysis based on clinical information stratification in TCGA. Construction of a nomogram for GC prediction Since the FAO-Risk Score was an independent prognostic factor for GC and holds significant clinical relevance, a nomogram integrating the risk score and clinical information was further constructed to more accurately predict the prognosis of GC patients (Fig. 3a). The calibration curve for 1-, 3-, and 5-year survival demonstrated superior performance of the nomogram (Fig. 3b). Compared to clinical variables alone, the nomogram exhibited a higher AUC for predicting patient prognosis (Fig. 3c). Univariate Cox regression analysis revealed the nomogram was significantly associated with overall survival, which were consistent with those from the GEO dataset (Fig. 3d, e). The regression coefficients (β) derived from multivariate Cox analysis were used to construct the nomogram. The corresponding hazard ratios (HR) and regression coefficients are as follows: age (HR = 1.029, β = 0.029), gender (HR = 1.019, β = 0.019), tumor stage (HR = 2.256, β = 0.813), and risk score (HR = 1.285, β = 0.251). Tumor stage exhibited the largest regression coefficient, followed by the FAO-associated risk score, indicating their dominant contributions to survival prediction. These findings highlight that the nomogram, combining the FAO- Risk scores and clinical information, provides high credibility and accuracy in prognostic prediction for GC patients. (a) The nomogram integrating the risk score and clinical information to predict the prognosis of GC patients; (b, c) The calibration curve for 1-, 3-, and 5-year survival (b) and discriminatory performance (c) to evaluate the performance of the nomogram; (d, e) Univariate Cox regression analysis with nomogram included, data of training cohort was from TCGA (d), and data of GC (GSE84437) was used as external cohort (e). Analysis of difference on immune microenvironment based on FAO We performed GSVA based on the KEGG pathway using the GC gene expression matrix, selecting the top 50 most variable pathways with the highest significance (− Log10 P) as shown in Fig. 4a. This analysis uncovered profound transcriptional divergence across FAO- riskscore stratified low- and high-risk GC, which may contribute to the tumor immune microenvironment’s heterogeneity. To further explore this, the immune status of patients in both the low- and high-risk groups was analyzed. Notably, TTN (titin) mutations, which play an important role in immunotherapy for solid tumors and GC28, were found to be significantly associated with higher FAO-Risk Scores in GC patients (p = 0.035 < 0.05) (Fig. 4b). Additionally, we examined the relationship between the FAO-Risk Score and tumor-associated immune subtypes, revealing a significant difference in risk score distribution between the C1 (wound healing) and C2 (IFN-γ dominant) subtypes (Fig. 4c). The TIDE analysis was then used to evaluate the immune escape ability of tumors, showing TIDE scores with significant difference across the low- and high- risk groups, suggesting poorer immune checkpoint inhibitor efficacy in high-risk patients. GSEA analysis (Fig. 4e, f) further demonstrated inconsistent immune infiltration and immune scores between the two groups, underscoring the heterogeneity of the tumor immune microenvironment. Overall, our findings indicate significant differences in tumor immune status between high-risk and low-risk groups based on FAO-Risk Score classification. (a) Heatmap of top 50 highly variable pathways with the highest significance (- Log10 P) by GSVA analysis based on KEGG pathway on the gene expression matrix of GC data; (b) TTN(Titin) mutation analyses on low and high risk group of patients; (c) The relationship between FAO risk score and tumor associated immune subtypes; (d) Prediction of the immune escape ability of GC on low and high risk group of patients using TIDE algorithm; (e,f) Distribution of immune infiltration (e) and immune Fraction (f) by ssGSEA. Drug sensitivity analysis TME plays a complex role in tumorigenesis, progression, and metastasis, which is crucial for selecting effective anticancer therapies. In this study, we utilized the pRophetic algorithm to investigated the relationship between the FAO- riskscore and the responses to various commonly used anticancer drugs (AG-014699, AMG-706, AUY922, Axitinib, BX-795, Camptothecin, CCT007093, CCT018159, and Cisplatin) in GC patients. As shown in Fig. 5, we analyzed the IC50 values of these drugs across patients with various FAO-Risk Scores and found a negative correlation between the drug IC50 values and the risk score. Specifically, compared with the low-risk group, the high-risk group exhibited lower IC50 values for these drugs, indicating increased predicted drug sensitivity. These findings suggest that GC patients with different FAO-risk scores may show distinct responses to chemotherapy, with patients in the high-risk group potentially being more sensitive to these agents. The relationship between the FAO risk score of GC patients and the response to various commonly used anti-tumor drugs in clinical practice (AG-014699, AMG-706, AUY922, Axitinib, BX-795, Camptothecin, CCT007093, CCT018159, and Cisplatin) was explored using the pRRophetic algorithm; (a–i) Correlation between IC50 of drugs and FAO risk score; (j–r) The IC50 values of each drug in low and high risk group of patients. Identification of regulator genes for FAO Given the significant differences of immune infiltration and drug sensitivity between the low- and high- risk groups, we proceeded with pathway enrichment analysis. Compared to normal tissues, the transcriptome gene expression in the two groups showed notable changes. DEGs were further annotated using Gene Ontology (GO) analysis (Fig. 6a, b) and KEGG pathway analysis (Fig. 6c, d). The most statistically significant GO terms (Top8 terms in Biological Process [BP], Cellular Component [CC], and Molecular Function [MF]) and KEGG pathways were selected, and their relationships with FAO-related DEGs were illustrated using a Circos plot (Fig. 6e). To further investigate these interactions, the STRING database was used to construct the protein–protein interaction (PPI) network of FAO-related genes and the DEGs between the high-risk and low-risk groups. The intersection of this analysis led to the identification of 10 hub genes, including IL-6 and THBS1, which were obtained using the CytoHubba plugin (Fig. 6f–h). (a,b) GO analysis on transcriptome gene expression in low (a) and high (b) risk group of patients; c-d. KEGG analysis on transcriptome gene expression in low (c) and high (d) risk group of patients (The KEGG pathway data used in this study were obtained from the KEGG database (www.kegg.jp/kegg/kegg1.html)); (e) The relationship between 8 most significant GO terms and KEGG pathway through Circos plot; (f) Construction of PPI network of genes related to FAO (f) and differentially expressed genes between low and high risk group of patients (g) using the STRING database; h. Use the CytoHubba to take the intersection of 10 hub genes from the former two PPI networks. IL-6 is associated with gastric cancer progression Kaplan–Meier survival analysis was conducted to investigate the prognostic relevance of hub genes in GC. The findings showed that elevated expression of the 10 hub genes were correlated with significantly worse overall survival outcomes in GC patients (p < 0.05) (Fig. 7a–i). Further analysis involving the relationship between IL-6 and clinical characteristics of GC patients revealed that high IL-6 expression was linked to advanced age, later clinical stage, distant metastasis, and higher T stage (Fig. 7k–q). Additionally, high IL-6 expression significantly altered the immune infiltration of tumor tissues, increasing the abundance of M0 and M1 macrophages, as well as neutrophils (Fig. 7r). (a–j) Kaplan Meier analysis to determine the correlation between ten hub genes and the prognosis of GC patients; (k–q) The relationship between IL-6 and clinical characteristics of GC patients; (k) age, (i) gender, (m) pathological grade, (n) stage, (o–q) Tumor Node Metastasis staging; (r,s) Immune infiltration and immune fraction of IL-6 low and high expressed GC patients. As comprehensive bioinformatics analysis showed, IL-6 was identified as a pivotal mediator in gastric cancer progression, and immunohistochemistry further results revealed differential IL-6 expression across tumor grades. Specifically, IL-6 expression increased progressively with tumor grade, as did the immunohistochemical positivity rate (Fig. 8a, b). Western blot analysis demonstrated that IL-6 expression was markedly lower in normal gastric epithelial cell lines compared to gastric tumor cell lines (Fig. 8c). Immunofluorescence analysis further confirmed that IL-6 expression is elevated in gastric tumor cell lines (Fig. 8d). We next silenced IL-6 expression in tumor cell lines using siRNA technology and observed a significant inhibition of the cell cycle (Fig. 8e). Furthermore, both clonogenicity and invasive capacity were markedly reduced following IL-6 knockdown (Figs. 8f and g). A wound healing assay also confirmed that IL-6 suppression impaired the migration ability of tumor cells (Fig. 8h). (a, b) Gastric cancer tissue samples from patients at the First Affiliated Hospital of Xi’an Jiaotong University were analyzed by immunohistochemistry across stages I to IV to assess IL-6 expression levels; Scale bar, 100 μm; (c) Protein expression levels of IL-6 were examined in gastric cancer cell lines (AGS, HGC27, SNU216, MKN45) and normal gastric mucosal cells (GES-1) (original gels are presented in Supplementary Fig. S2); (d) Expression patterns and subcellular localization of IL-6 were analyzed in AGS cells and GES-1 cells. Scale bar, 100 μm; (e) Cell cycle analysis of HGC27 and AGS cells after IL-6 knockdown (siIL-6) compared with the control group (siC). Flow cytometry histograms and quantitative analysis showing the distribution of cells in G1, S, and G2/M phases; (f) Colony formation assays in HGC27 and AGS cells transfected with siIL-6#1 and siIL-6#2 or siC, indicating that IL-6 knockdown significantly suppressed the clonogenic ability of gastric cancer cells; (g) Transwell migration assays in HGC27 and SNU216 cells following IL-6 silencing. Representative images showing reduced migratory capacity in the siIL-6 groups compared with the control group; Scale bar, 50 μm;(h) Wound-healing assays in HGC27 and AGS cells transfected with siIL-6 or siC; Representative images were captured at 0 h, 24 h, and 48 h, demonstrating that IL-6 knockdown significantly inhibited cell migration; Scale bar, 100 μm. Discussion Cancer cells usually utilize lipids as their energy supply and storage, the latter one often participate in synthesis of components for cell membrane as well as messengers for signal transduction29. In order to catch the needs of rapid proliferation and metabolism, malignant tumor cells often reprogram the lipid metabolism to reshape their overall lipid utilization30,31. Exploring the occurrence and development of tumors through lipid metabolism has become a current research focus in various types of cancer32. GC is one of the types of cancer with the highest incidence rate. GC’s global incidence rate and mortality reached 5.6% and 7.7% according to GLOBOCAN 2020 reports, ranking fourth and fifth33. Patients with late stage gastric cancer face grave prognoses, evidenced by a 5-year survival rate of merely 6% in clinical cohorts31. FAO is a key pathway that regulates the breakdown metabolism of fatty acids and the production of ATP and NADPH9. In GC, characteristics such as stemness and chemotherapy drug resistance of tumor cells are positively regulated by FAO levels9, meanwhile, high FAO level is also an important dependent factor for peritoneal metastasis of GC12. Therefore, there is an urgent need to conduct studies on regulatory genes groups related to tumor lipid metabolism, establish a prognostic prediction model based on a new perspective, and look for core regulatory genes with their related pathways, providing new ideas for clinical intervention targets, and contribute to the precise treatment of GC. In this study, a FAO related gene set were established, through which we obtained 10 overall survival related genes including AOC3, ACADS, CD36, etc. with analyzing publicly available transcriptome sequencing data. Combined with MAF tools, we showed that these genes all contributed to the genetic heterogeneity of GC to vary degrees. Subsequently, 7 genes further filtered by LASSO regression and evaluated were used for construction a new risk score, based on which GC patients were divided into high and low risk groups, and a new GC prognostic model were constructed subsequently. Through verification, our prediction model has shown good prognostic ability, indicating that FAO has a positive significance in promoting the occurrence and development of GC, which is consistent with previous research results. Results of GSVA, ssGSEA, CIBERSORT and tumor mutation burden analysis showing significant difference status of tumor microenvironment and heterogeneity between GC patients in low- and high- risk groups. Drug sensitivity analysis using pRophetic algorithm furtherly showed that as the risk score increased, the predicted IC50 values of several anticancer drugs decreased, suggesting increased predicted sensitivity to these agents. This suggests that FAO participates in driving tumor progression and drug resistance. Subsequently conducted transcriptome analysis across low- and high- group identified IL-6 as the main hub gene. Clinic correlation analysis, tumor stage, survival analysis and molecular biology validation were exerted to verify the conclusion that IL-6 was a key regulatory factor affects the development of GC through FAO. IL-6 is an extremely important cytokine in mediating inflammation and immune regulation, it binds to IL-6R, GP130, etc. to initiates signal transduction34, targeted inhibitors have been approved clinically in various diseases35. In the field of tumor, it is reported that chronic sustained activation of IL-6 could promote the progression of tumor; in the tumor microenvironment, IL-6 mainly comes from tumor cells, stromal cells and infiltrating immune cells, sustained IL-6/STAT3 signaling leads to cell cycle progression, proliferation and survival of cancer cell13. Additionally, Rustgi et al. demonstrated that IL-6 is the main factor in dynamic crosstalk between cancer cells and cancer associated fibroblast (CAF)36. In GC, upregulation of IL-6 in tumor tissue is related to poor chemotherapy responsiveness and invasion, metastasis of GC cells.IL-6 has been proved to be an independent marker of poor prognosis in GC37,38. Considering that IL-6 has a relationship with FAO15,39, certain role played by this relationship need to be further clear. Our study set up from FAO related gene set, and ultimately identified IL-6 to be strongly positively correlated gene. This firstly provides the research evidence for IL-6/FAO regulation in GC, and offers new insights for related research on GC. Several limitations should be acknowledged. First, the prognostic model was derived from retrospective public datasets and requires validation in larger prospective cohorts. Second, although additional in vitro experiments were performed in multiple gastric cancer cell lines, in vivo validation remains lacking. Third, the mechanistic link between IL-6 and FAO was inferred from bioinformatics analyses and indirect functional evidence, and therefore requires further dedicated metabolic and mechanistic investigations. In conclusion, this study initiated from a perspective of FAO, has demonstrated the positive driving role of IL-6/FAO in promoting the progression of GC, this further confirms that IL-6 is a clinically correlated marker and prognostic predictor of GC. The result that the mechanistic relationship between IL-6 and FAO is worth exploring further detailed mechanisms in future life science research. Conclusion FAO demonstrates significant involvement in gastric cancer pathogenesis. A FAO-driven risk prediction model for GC was established and exhibited excellent predictive performance. IL-6 was identified as a hub gene associated with the FAO-related prognostic signature in GC. There are promising prospects in targeted therapy of GC based on FAO and IL-6. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Acknowledgements We sincerely thank all investigators of TCGA Research Network for publicly sharing their valuable research data. We also appreciate the Biobank of the First Affiliated Hospital of Xi’an Jiaotong University for providing platform support for this study. Abbreviations Fatty acid β-oxidation Gastric cancer Elial-mesenchymal-transition Tumor mutation burden Interleukin-6 Overall survival Differentially-expressed gene Progression-free survival Area under the curve Gene set variation analysis Principal component analysis Author contributions CQ and XY designed and conducted the experiments and drafted the manuscript. YQ, GL, and RZ assisted with the experiments, literature review, and revision of the manuscript. SY contributed to study design and data analysis. YW and MZ were responsible for data collection. XS designed the study and supervised the project. XL was responsible for manuscript editing. All authors contributed to the work and reviewed and approved the submitted version of the manuscript. Funding This study was supported by the National Natural Science Foundation of China (Grant No. 82470537 and 82200563) and the Shaanxi Provincial Science and Technology Program (Grant No. 2025SF-YBXM-037). We gratefully acknowledge the financial support provided by the National Natural Science Foundation of China and the Shaanxi Provincial Science and Technology Department. Data availability The data supporting the findings of this study are included in the article and its supplementary materials. Please contact the corresponding author for further inquiries. Declarations Competing interests The authors declare that they have no competing interests to disclose. Ethical approval Human pathological samples from gastric cancer patients were obtained with the approval of the Ethics Committee of the First Affiliated Hospital of Xi’an Jiaotong University. Written informed consent was obtained from all participants prior to their inclusion in the study. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Chao Qu and Xuetao Yuan have contributed equally to this work. Contributor Information Xuejun Sun, Email: sunxy@mail.xjtu.edu.cn. Xuqi Li, Email: lixuqi@163.com. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Data Availability Statement The data supporting the findings of this study are included in the article and its supplementary materials. Please contact the corresponding author for further inquiries.