Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Abstract

Objective

This study aimed to identify steroid metabolism–related molecular subtypes, investigate the gene expression patterns of these subtypes, and construct a prognostic risk model as well as predict therapeutic response in gastric cancer.

Methods

We analyzed 410 TCGA-STAD and 483 GSE84437 gastric cancer samples. Unsupervised consensus clustering based on steroid metabolism–related genes identified molecular subtypes. Differential expression and functional enrichment analyses (GO, KEGG, GSEA) were performed. Prognostic genes were intersected with survival-associated genes, and a Lasso-Cox regression model was used to build a seven-gene risk score. Immune infiltration, tumor mutational burden (TMB), immune checkpoint expression, and drug sensitivity were evaluated.

Results

Two steroid metabolism–related gastric cancer subtypes were identified, with steroid-metabolism–poor prognosis subtype showing poorer overall survival. Differential expression analysis revealed 1,709 genes enriched in immune regulation, calcium signaling, cell adhesion, and extracellular matrix remodeling. Seven key genes (PRICKLE1, SERPINE1, APOD, RIMS1, GLP2R, CDH19, GRP) were used to construct a risk score, which correlated with advanced stage, steroid-metabolism–poor prognosis subtype subtype, and worse survival, and was an independent prognostic factor. High-risk and steroid-metabolism–poor prognosis subtype tumors displayed higher immune infiltration and immune scores, lower TMB, and upregulated immune checkpoint genes, indicating an immunosuppressive microenvironment. Drug sensitivity differed across subtypes and risk groups, suggesting potential implications for personalized therapy.

Conclusions

Steroid metabolism defines molecular heterogeneity, immune features, and prognosis in gastric cancer. The seven-gene risk model provides a reliable tool for survival prediction and may guide personalized therapeutic strategies.

Keywords

gastric cancer steroid metabolism prognosis immune checkpoint therapeutic response

1. Introduction

Gastric cancer remains one of the most prevalent malignancies and a leading cause of cancer-related mortality worldwide.¹ Despite advances in surgical techniques, chemotherapy, targeted therapy, and immunotherapy, the overall prognosis of patients with advanced gastric cancer remains unsatisfactory.^2,3 Tumor heterogeneity at the molecular, metabolic, and immune levels contributes substantially to differences in clinical outcomes and therapeutic responses among patients.^4-7 Therefore, identifying novel molecular classifications and prognostic biomarkers is essential for improving risk stratification and guiding personalized treatment strategies in gastric cancer.

Metabolic reprogramming is recognized as a hallmark of cancer and plays a crucial role in tumor initiation, progression, and therapeutic resistance.^8-10 Among various metabolic pathways, steroid metabolism has attracted increasing attention due to its involvement in hormone synthesis, lipid homeostasis, immune regulation, and signal transduction.^11,12 Steroid metabolites and steroidogenic enzymes have been shown to influence tumor cell proliferation, epithelial–mesenchymal transition (EMT), angiogenesis, and interactions with the tumor microenvironment.^13-17 Dysregulation of steroid metabolism has been reported in several malignancies, including breast, prostate, and gastrointestinal cancers,^18-20 suggesting that steroid metabolism may contribute to tumor aggressiveness and clinical heterogeneity.

In gastric cancer, accumulating evidence indicates that alterations in steroid metabolic genes are associated with tumor progression, immune infiltration, and patient prognosis.^21,22 However, current research on steroid-related mechanisms in gastric cancer has predominantly focused on individual steroid-associated genes. Systematic investigations of steroid metabolism at the pathway or gene set level, particularly regarding steroid metabolism–related molecular subtypes and their comprehensive associations with prognosis, immune microenvironment, tumor mutational burden, and therapeutic response, remain largely unexplored.

With the rapid development of high-throughput sequencing technologies and public cancer databases, integrative bioinformatics approaches provide powerful tools for dissecting tumor heterogeneity and identifying clinically relevant molecular signatures. Unsupervised clustering based on pathway-related gene expression profiles enables the identification of novel tumor subtypes with distinct biological characteristics and clinical outcomes. Moreover, the construction of robust prognostic models based on differentially expressed genes can facilitate individualized risk assessment and improve prognostic prediction beyond traditional clinicopathological factors.

In the present study, we systematically explored the expression patterns of steroid metabolism–related genes in gastric cancer using the TCGA-STAD cohort. We identified distinct steroid metabolism–associated molecular subtypes and comprehensively characterized their biological behaviors, functional pathways, immune infiltration patterns, and clinical outcomes. Based on differentially expressed genes between these subtypes, we constructed and validated a steroid metabolism–related prognostic risk score model in both the TCGA and GSE84437 cohorts. Furthermore, we developed nomograms incorporating this risk score to enhance prognostic prediction accuracy. Finally, we investigated the associations between steroid metabolism patterns, immune checkpoint expression, tumor mutational burden, and drug sensitivity, aiming to provide insights into potential therapeutic implications.

Collectively, this study elucidates the critical role of steroid metabolism in shaping the molecular heterogeneity, immune microenvironment, and prognosis of gastric cancer. Our findings suggest that steroid metabolism–based classification and the derived risk score may serve as valuable tools for prognostic evaluation and personalized treatment decision-making in gastric cancer.

2. Materials and Methods

2.1 Data Sources

Two publicly available gastric cancer cohorts were incorporated into this study for subsequent analyses. Transcriptomic data and clinical characteristics of stomach adenocarcinoma were obtained from The Cancer Genome Atlas (TCGA), including 410 tumor samples and 36 corresponding adjacent normal tissues. Clinical information available for this cohort encompassed overall survival time and status, patient age, sex, and tumor stage. An independent validation cohort, GSE84437, was downloaded from the Gene Expression Omnibus (GEO) database and consisted of 483 gastric cancer cases analyzed on the GPL6947 platform (Illumina HumanHT-12 V3.0 expression beadchip), with detailed annotations covering survival outcomes, age, sex, and tumor stage. To depict the mutational landscape of STAD samples, waterfall plots were constructed using the ComplexHeatmap package in R.²³ Prior to downstream analyses, raw expression data from both TCGA and GEO datasets were processed and normalized using the R package limma.²⁴

2.2 Bulk RNA-Seq DE and Intersection

Differentially expressed genes (DEGs) between steroid metabolism subtypes were identified using the DESeq2²⁵ package to identify DEGs (adjusted p < 0.05, |log2FC| > 1).

2.3 Subtype Clustering by Steroid Metabolism

Steroid metabolism–related genes were obtained from the MSigDB gene set C5_GOBP_STEROID_CATABOLIC_PROCESS. Unsupervised consensus clustering was performed using the “ConsensusClusterPlus” R package²⁶ based on the expression profiles of 35 steroid metabolism–related genes. The number of clusters was evaluated by varying the clustering parameter K from 2 to 9, and the optimal cluster number was determined according to the cumulative distribution function (CDF) curves and the relative change in the area under the CDF curve (delta area).

2.4 Functional Enrichment Analysis

To characterize functional discrepancies between the two groups, enrichment analyses were carried out using Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and gene set enrichment analysis (GSEA). All analyses were implemented in R with the clusterProfiler package²⁷ and GSVA package.²⁸ Pathways were defined as significantly enriched when the normalized enrichment score exceeded an absolute value of 1, accompanied by a nominal P value < 0.05 and a false discovery rate (FDR)–corrected q value < 0.05.

2.5 Construction of Steroid Metabolism–Related Prognostic Risk Model

Prognostic DEGs were identified by intersecting subtype-specific DEGs with genes significantly associated with overall survival in TCGA and GSE84437 cohorts. Lasso-Cox regression analysis was applied to select key prognostic genes. A risk score model was established using the expression levels of seven key genes weighted by their Cox regression coefficients: Risk score = PRICKLE1 × 0.0039 + SERPINE1 × 0.1249 + APOD × 0.0668 + RIMS1 × 0.1565 + GLP2R × 0.0012 + CDH19 × 0.0038 + GRP × 0.0457. Patients were divided into high- and low-risk groups according to the median risk score. Kaplan–Meier survival analysis, risk curves, and receiver operating characteristic (ROC) curves were used to evaluate model performance.

2.6 Nomogram Construction and Validation

A nomogram integrating risk score and clinical variables was constructed using the “rms” R package. Calibration curves were plotted to compare predicted and observed survival probabilities. The nomogram and risk model were further validated in the independent GSE84437 cohort.

2.7 Immune Infiltration and Tumor Mutational Burden Analysis

Immune cell infiltration was estimated using ESTIMATE,²⁹ CIBERSORT.³⁰ Single-sample gene set enrichment analysis (ssGSEA) method from the R package GSVA²⁸ was used to calculate the infiltration levels of 28 immune cell types based on the expression profiles of genes from 28 published immune cell gene sets.³¹ Differences in immune cell populations and immune scores between steroid metabolism subtypes and risk groups were analyzed. Expression of immune checkpoint–related genes was compared to evaluate potential responsiveness to immunotherapy. Tumor mutational burden (TMB) was calculated using somatic mutation data from the TCGA cohort and visualized using the “Maftools” R package.³²

2.8 Drug Sensitivity Analysis

The “oncoPredict” R package³³ was used to predict drug response based on gene expression profiles. Differences in predicted sensitivity to multiple anticancer agents were compared across steroid metabolism subtypes and between high- and low-risk score groups.

2.9 Statistical Analysis

Data processing and statistical evaluations were performed using R and SPSS. All graphical illustrations were generated with Adobe Illustrator. Differences between groups shown in box plots were analyzed using the non-parametric Wilcoxon rank-sum test. Relationships among variables were examined through Spearman correlation analysis. Prognostic effects of relevant factors were assessed by applying both univariate and multivariate Cox proportional hazards regression models. Patient survival outcomes were estimated using the Kaplan–Meier approach, and group comparisons were conducted with the log-rank test. All statistical analyses were two-tailed, and results with a p value below 0.05 were regarded as statistically significant.

3. Results

3.1 Gastric Cancer Classification Pattern Based on Steroid Metabolism

To systematically investigate the expression characteristics of steroid metabolism–related genes in gastric cancer, we performed an unsupervised clustering analysis based on 410 samples from the TCGA-STAD cohort. Consensus clustering was conducted using the expression profiles of 35 steroid metabolism–related genes, and gastric cancer patients were classified into two distinct subtypes, designated as group 1 and group 2 (Figure 1A, Supplementary Fig. S1A-B).

Further differential expression analysis demonstrated that the two subtypes exhibited significantly distinct expression patterns of steroid metabolism–related genes (Figure 1B and C). Survival analysis revealed significant differences in clinical outcomes between the two groups, with patients in group 2 showing significantly worse overall survival than those in group 1 (Figure 1D). Therefore, group 1 was defined as the steroid-metabolism–favorable prognosis subtype, and group 2 was defined as the steroid-metabolism–poor prognosis subtype. In the subsequent analyses, the terms “steroid-metabolism–favorable prognosis subtype” and “steroid-metabolism–poor prognosis subtype” were used to replace steroid-metabolism–favorable prognosis subtype and steroid-metabolism–poor prognosis subtype, respectively.

3.2 Biological Behaviors and Pathways Associated With Steroid Metabolism

Differential expression analysis based on steroid metabolism–related subtypes identified a total of 1,709 differentially expressed genes (DEGs) (Figure 2A and B). KEGG enrichment analysis revealed that these genes were mainly involved in multiple pathways, including neuroactive ligand–receptor interaction, complement and coagulation cascades, cytokine–cytokine receptor interaction, calcium signaling pathway, cell adhesion molecule (CAM) interactions, integrin signaling, cAMP and cGMP–PKG signaling pathways, hematopoietic cell lineage, vascular smooth muscle contraction, and protein and fat digestion and absorption. In addition, GO enrichment analysis indicated that these genes were predominantly associated with biological processes related to humoral immune response, leukocyte migration and chemotaxis, cell killing, extracellular matrix and extracellular structure organization, cell–matrix adhesion, muscle contraction, regulation of cytosolic calcium ion concentration, calcium-mediated signaling, regulation of monoatomic ion transport, and blood circulation (Figure 2C and D). Collectively, these enrichment results suggest that steroid metabolism–related prognostic genes are closely associated with immune regulation, calcium signaling, cell adhesion, and extracellular matrix remodeling. GSEA enrichment analysis indicated that these DEGs were predominantly involved in key hallmark biological processes (Figure 2E). Overall, these results suggest the existence of two distinct regulatory patterns related to steroid metabolism in gastric cancer. The functional enrichment differences between the subtypes likely represent an important biological basis for the observed variations in gene expression profiles among patients.

3.3 Construction of a Prognostic Model Based on Differentially Expressed Steroid Metabolism–Related Subtypes

To further investigate the association between steroid metabolism–associated molecular subtypes and gastric cancer prognosis, we constructed a prognostic model based on differentially expressed genes (DEGs) identified among distinct steroid metabolism–related subtypes. By intersecting subtype-associated prognostic DEGs with survival-related genes derived from the TCGA and GSE84437 cohorts, a total of 260 candidate prognostic genes were obtained.

Subsequently, Lasso-Cox regression analysis was performed to reduce dimensionality and select the most robust prognostic features. Seven genes (PRICKLE1, SERPINE1, APOD, RIMS1, GLP2R, CDH19, and GRP) were ultimately identified and used to construct a multigene risk score model (Figure 3A and B). The risk score was calculated using the following formula: PRICKLE1 (0.0039) + SERPINE1 (0.1249) + APOD (0.0668) + RIMS1 (0.1565) + GLP2R (0.0012) + CDH19 (0.0038) + GRP (0.0457).

According to the median risk score, gastric cancer patients were stratified into high-risk and low-risk groups. Survival analysis demonstrated that patients in the high-risk group had significantly worse overall survival compared with those in the low-risk group (Figure 3C and D). Time-dependent risk distribution plots further indicated that increasing risk scores were associated with a higher probability of death (Figure 3E and F).

The prognostic performance of the model was further evaluated using ROC curve analysis. The overall AUC was 0.656, while the AUCs for 1-, 3-, and 5-year survival predictions were 0.659, 0.697, and 0.722, respectively (Figure 3G and H), indicating moderate but consistent predictive ability. In addition, we integrated clinical parameters, including age, sex, and AJCC T and N stages, with the gene-based risk score to construct a combined prognostic model, and evaluated its predictive performance using time-dependent ROC analysis. The combined model demonstrated improved predictive accuracy, with 1-, 3-, and 5-year AUCs of 0.722, 0.774, and 0.769, respectively, in the TCGA training cohort. In the independent GSE84437 validation cohort, the corresponding AUCs were 0.735, 0.747, and 0.746, all of which were superior to those of the gene-based risk score model alone (Supplementary Fig. S3A-B). These findings suggest that integrating clinicopathological parameters with molecular features may further enhance the predictive accuracy and potential clinical utility of the prognostic model.

We next assessed the expression patterns of the seven signature genes across steroid metabolism–related subtypes and risk groups. All seven genes showed higher expression in steroid-metabolism–poor prognosis subtype and in the high-risk group (Figure 3I and L), suggesting that this subtype-associated transcriptional program is linked to poor prognosis. Correlation analysis further revealed positive co-expression relationships among most of these genes (Supplementary Fig. S2).

Collectively, we developed a seven-gene prognostic signature derived from steroid metabolism–associated subtype stratification, which effectively stratifies gastric cancer patients by survival outcome. The high-risk group corresponded to the steroid metabolism–associated steroid-metabolism–poor prognosis subtype, indicating that this transcriptional program may reflect a more aggressive tumor state and could serve as a potential prognostic biomarker set for gastric cancer.

3.4 A Nomogram Based on Steroid Metabolism–Associated Gene Risk Score for Predicting Gastric Cancer Prognosis

Univariate and multivariate Cox regression analyses were conducted based on the risk score derived from differentially expressed genes associated with steroid metabolism subtypes. The results indicated that the risk score remained significantly associated with prognosis in gastric cancer in both analyses (Figure 4A and B). To enhance the accuracy of prognostic prediction, a nomogram was constructed based on the risk score of steroid metabolism–related differentially expressed genes (Figure 4C). Calibration curves demonstrated that the survival probabilities predicted by the nomogram closely matched the observed outcomes (Figure 4D and F). In conclusion, the risk score derived from steroid metabolism–related differentially expressed genes serves as an independent prognostic factor in gastric cancer. The constructed nomogram demonstrated accurate survival prediction, highlighting the potential utility of this risk score for individualized prognostic evaluation and clinical decision-making.

3.5 Validation of a Nomogram Based on Steroid Metabolism–Related Gene Risk Score for Prognostic Prediction in Gastric Cancer

In the validation cohort (GSE84437), we developed a nomogram based on the risk score of steroid metabolism–related differentially expressed genes (Figure 5A). Calibration curves showed good agreement between the survival probabilities predicted by the nomogram and the observed outcomes (Figure 5B–D). Gastric cancer samples were divided into high- and low-risk groups according to their risk score, and the model was further validated by examining differences in risk score between survival statuses and through Kaplan–Meier survival analysis. Risk curve analysis revealed that patient mortality risk increased markedly with higher risk score (Figure 5E and F). The overall predictive performance of the steroid metabolism–related gene risk score, as measured by the area under the curve (AUC), was 0.605, demonstrating good accuracy for predicting 1-, 3-, and 5-year survival; the AUCs for the entire cohort at 1, 3, and 5 years were 0.584, 0.617, and 0.623, respectively (Figure 5G and H). Additionally, in the forest plot model, the hazard ratios (HRs) of these seven core genes and the risk score were all greater than 1 in both the TCGA and GSE84437 datasets (Supplementary Fig. S4A-B). These findings suggest the steroid metabolism–related gene risk score was successfully validated in the GSE84437 cohort, effectively stratifying patients into high- and low-risk groups with significant differences in survival. The nomogram demonstrated good predictive accuracy, and the hazard ratios of the seven core genes and the risk score consistently exceeded 1 across both TCGA and GSE84437 cohorts, highlighting the robustness and prognostic value of this risk score in gastric cancer.

3.6 Expression Differences of Risk Score Across Clinical Subtypes and Steroid Metabolism Subtypes in Gastric Cancer

We investigated the differences in risk score expression across different stages, T, N, and M classifications in gastric cancer. The results showed that risk score expression was significantly elevated in advanced stages, as well as in higher T and N classifications (Figure 6A–D). Moreover, among different steroid metabolism subtypes, the risk score of steroid-metabolism–poor prognosis subtype was higher (Figure 6E). In summary, risk score were positively associated with tumor progression, showing higher expression in advanced stages and in tumors with elevated T and N classifications. Additionally, steroid metabolism steroid-metabolism–poor prognosis subtype exhibited higher risk score compared with steroid-metabolism–favorable prognosis subtype, indicating a potential link between steroid metabolism patterns and prognostic risk in gastric cancer.

3.7 Immune Infiltration and Tumor Mutational Burden of Risk Score and Steroid Metabolism Subtypes

In light of the enrichment of differentially expressed genes in immune- and matrix-related pathways (Figure 2C–E), we next examined the differences in immune functions between steroid metabolism subtypes and risk score–defined groups. We performed CIBERSORT, ssGSEA and ESTIMATE analyses. Analysis of immune cell infiltration across different steroid metabolism–related subtypes and risk score groups revealed that multiple immune cell populations differed and immune score between the groups. Among them, immune cell infiltration and immune scores were higher in steroid metabolism steroid-metabolism–poor prognosis subtype and the high-risk score group than in steroid-metabolism–favorable prognosis subtype and the low-risk score group (Figure 7A–D) (Supplementary Fig. S5). However, the composition of immune infiltrates revealed important functional heterogeneity. Specifically, ssGSEA and CIBERSORT analysis indicated that the high-risk and steroid-metabolism–poor prognosis subtype tumors were characterized by increased proportions of immunosuppressive cell populations, including M2 macrophages, MDSC and regulatory T cells (Figure 7A–D). To explore the potential responsiveness to immunotherapy in gastric cancer with different steroid metabolism–related subtypes and risk score groups, we compared the expression of immune checkpoint–related genes between groups. Our results indicated that most immune checkpoint–related genes were significantly upregulated in steroid-metabolism–poor prognosis subtype and the high-risk score group (Figure 7E and F).

Higher tumor mutational burden (TMB) can lead to increased production of self-neoantigens and stronger immune recognition, thereby enhancing the probability of T cell-mediated tumor cell recognition and response, and indicating potentially improved outcomes with immune checkpoint inhibitor (ICI) therapy.³⁴ The distribution of somatic mutations in the TCGA cohort was analyzed using the “Maftools” package (Figure 8A–C). TMB was found to be higher in steroid-metabolism–favorable prognosis subtype and the low-risk group, while it was lower in steroid-metabolism–poor prognosis subtype and the high-risk group (Figure 8D–I).

These results indicate that although steroid-metabolism–poor prognosis subtype and the high-risk score group exhibit higher levels of immune cell infiltration and elevated immune scores, they also show high expression of immune checkpoint genes and lower tumor mutational burden, suggesting that their tumor microenvironment may be characterized by a state of immune cell functional suppression, which may limit warrant combination treatment strategies.

3.8 Drug Sensitivity Analysis

Finally, we used “oncoPredict” to perform drug sensitivity analysis across different steroid metabolism subtypes and risk score groups.

The results showed that sensitivity to Lapatinib_1558, Sapitinib_1549, Dihydrorotenone_1827, TAF1_5496_1732, Acetalax_1804, and BI.2536_1086 differed among steroid metabolism subtypes, with steroid-metabolism–poor prognosis subtype showing higher sensitivity to these agents (Figure 9A).

Meanwhile, differences in drug sensitivity were also observed between the risk score groups. The high-risk score group showed higher sensitivity to multiple agents, including Erlotinib_1168, Gefitinib_1010, Lapatinib_1558, Dihydrorotenone_1827, Oxaliplatin_1089, TAF1_5496_1732, Sapitinib_1549, VX.11e_2096, Afatinib_1032, Ulixertinib_1908, Acetalax_1804, PD0325901_1060, 5-Fluorouracil_1073, VE.822_1613, Oxaliplatin_1806, Crizotinib_1083, Cytarabine_1006, MK.1775_1179, Docetaxel_1007, Paclitaxel_1080, Trametinib_1372, Pevonedistat_1529, AZD6738_1917, and Gemcitabine_1190 (Figure 9B).

4. Discussion

In this study, we systematically investigated the role of steroid metabolism in gastric cancer by integrating transcriptomic data, clinical outcomes, immune characteristics, tumor mutational burden, and drug sensitivity. We identified two distinct steroid metabolism–related molecular subtypes with significantly different prognoses and biological behaviors. Furthermore, we constructed and validated a robust steroid metabolism–related prognostic risk score model based on seven key genes, which demonstrated stable predictive performance across independent cohorts. Our findings emphasize that steroid metabolism may represent a key driving factor underlying molecular heterogeneity, prognosis, immune regulation, and therapeutic response in gastric cancer.

Metabolic reprogramming is increasingly recognized as a driving force of tumor progression, and steroid-related metabolic pathways play pivotal roles in shaping tumor behavior and the tumor microenvironment.^13-17 In our study, unsupervised clustering based on steroid metabolism–related genes revealed two gastric cancer subtypes with markedly different survival outcomes. Patients in steroid metabolism steroid-metabolism–poor prognosis subtype exhibited significantly poorer overall survival, indicating that dysregulated steroid metabolism may promote malignant progression in gastric cancer.

Functional enrichment analyses provided further insights into the biological mechanisms underlying these subtype differences. Differentially expressed genes between the two steroid metabolism subtypes were predominantly enriched in immune-related processes, cell adhesion, extracellular matrix organization, and hallmark oncogenic pathways such as epithelial–mesenchymal transition and KRAS signaling.^35-39 These pathways are known to facilitate tumor invasion, metastasis, and immune evasion, supporting the notion that aberrant steroid metabolism may influence gastric cancer progression through complex interactions with signaling networks and the tumor microenvironment.

Based on subtype-associated prognostic genes, we established a steroid metabolism–related risk score model consisting of seven genes (PRICKLE1, SERPINE1, APOD, RIMS1, GLP2R, CDH19, and GRP) using Lasso-Cox regression. This model effectively stratified patients into high- and low-risk groups, which exhibited significantly different survival outcomes in both the TCGA and GSE84437 cohorts. Importantly, after adjusting for conventional clinicopathological variables, the risk score remained an independent prognostic factor, further demonstrating the stability and reliability of the model. Incorporation of the risk score into a nomogram further improved individualized survival prediction. Although the model demonstrated a certain predictive performance in the TCGA cohort (AUC ≈ 0.66–0.72), its performance decreased in the external validation cohort GSE84437, with AUC values dropping to 0.58–0.62, suggesting limited discriminative ability. These findings indicate that the current model is primarily suitable for exploratory risk stratification analysis, while its clinical predictive utility still requires further optimization and validation.

Biologically, several genes in this signature have been implicated in gastric cancer progression and tumor microenvironment regulation. SERPINE1 (PAI-1) is a well-established oncogenic factor in gastric cancer, promoting epithelial–mesenchymal transition (EMT), and cell proliferation, and its overexpression is associated with poor prognosis.^40,41 APOD (Apolipoprotein D) is involved in lipid transport and steroid-like lipid binding. Its dysregulated expression is associated with lipid metabolic disturbances in gastric cancer, potentially affecting the tumor’s metabolic adaptability and correlating with poor prognosis.^42,43 PRICKLE1 is significantly upregulated in lymph node metastatic lesions of gastric cancer and promotes gastric cancer cell migration and invasion by activating the mTOR signaling pathway, thereby driving metastatic progression.⁴⁴ GLP2R (Glucagon-like peptide-2 receptor) is a G protein–coupled receptor involved in gut mucosal growth⁴⁵; In addition, there is evidence that silencing GLP2R can significantly inhibit the proliferation of gastric cancer cells.⁴⁶ GRP (Gastrin-releasing peptide) functions as a neuropeptide involved in autocrine growth signaling, and GRP/GRPR axis activation has been implicated in promoting proliferation and survival in gastrointestinal tumors.⁴⁷ In contrast, CDH19 is a member of the cadherin family involved in cell–cell adhesion and epithelial integrity. Studies have shown that knockdown of CDH19 expression can inhibit the proliferation and migration of gastric cancer cells.⁴⁸ RIMS1, primarily known for synaptic vesicle exocytosis regulation,⁴⁹ has limited direct evidence in gastric cancer but may reflect neuroendocrine-like signaling components within the tumor microenvironment. Collectively, these genes suggest that steroid metabolism–related transcriptional programs may intersect with lipid handling, neuroendocrine signaling, adhesion dynamics, and EMT processes in gastric cancer progression.

From a mechanistic perspective, this risk model reflects the reprogramming of a lipid–hormone regulatory network closely associated with steroid metabolism. For example, among these genes, APOD (Apolipoprotein D) is one of the genes most closely associated with steroid metabolism. As a member of the lipid transport protein family, APOD can bind cholesterol and steroid-like hydrophobic molecules, participating in intracellular cholesterol transport and the maintenance of lipid homeostasis.⁵⁰

In addition, GLP-2 enhances lipid absorption in the intestine by promoting enterocyte growth, lipid transport, and chylomicron production, thereby increasing the availability of circulating lipids and cholesterol.⁵¹

Although SERPINE1 is not a classical steroid metabolism gene, its expression is regulated by hormone-related signaling pathways (such as glucocorticoid and estrogen signaling).⁵² This suggests that SERPINE1 may function as a downstream responsive gene within the steroid metabolism network.

In summary, although these genes are not classical steroid biosynthetic enzymes, they collectively form a regulatory network connecting lipid transport, cholesterol metabolism, and hormonal responses, thereby reflecting the systemic reprogramming of steroid metabolism–related biological processes in gastric cancer. Future studies may consider integrating clinical variables (such as TNM stage and age) with molecular features to construct a combined model, thereby further improving model stability and predictive performance.

Tumor mutational burden has been proposed as a biomarker for predicting responsiveness to immune checkpoint inhibitor therapy.³⁴ The immune landscape analysis revealed notable differences between steroid metabolism subtypes and risk score groups. The steroid-metabolism–poor prognosis subtype and the high-risk group exhibited higher immune cell infiltration and elevated immune scores, accompanied by increased expression of immune checkpoint–related genes. However, these groups also showed lower tumor mutational burden, suggesting that despite increased immune infiltration, the tumor microenvironment may be functionally immunosuppressed rather than immunologically effective.

Importantly, this apparent paradox can be explained by the complex regulatory network within the tumor microenvironment (TME). Recent studies have demonstrated that the functional state of immune infiltration is highly dependent on cellular composition and intercellular signaling within the TME, rather than the absolute level of immune cell abundance. Specifically, enrichment of immunosuppressive cell populations, such as regulatory T cells and MDSC, can contribute to immune evasion and attenuate anti-tumor immune responses, even in the context of high immune infiltration.⁵³ In addition, broader regulatory mechanisms within the tumor microenvironment (TME)—including interactions among immune cells, signaling pathways, and cytokines—play a critical role in determining the efficacy of immune checkpoint blockade therapy.⁵⁴

Taken together, the combination of low TMB, elevated immune checkpoint expression, and a potentially immunosuppressive TME suggests that steroid-metabolism–poor prognosis subtype and high-risk patients may represent an “immune-excluded” or “immune-dysfunctional” phenotype. These findings highlight that immune infiltration alone is insufficient to predict immunotherapy response, and functional characterization of the TME is essential. Therefore, patients in these subgroups may benefit from combination therapeutic strategies aimed at reprogramming the immune microenvironment to enhance anti-tumor immunity.

Drug sensitivity analysis further explored the potential clinical implications of steroid metabolism patterns. We observed differences in the predicted sensitivity to multiple targeted agents and chemotherapeutic drugs across steroid metabolism subtypes and risk score groups. These findings suggest that steroid metabolism–related molecular features may be associated with variations in therapeutic response; however, their potential value in treatment stratification requires further validation through experimental studies.

Several limitations of this study should be acknowledged. First, our analyses were primarily based on retrospective public datasets, which may introduce inherent selection bias and limit the generalizability of the findings. In addition, potential batch effects and heterogeneity across different platforms (RNA-seq and microarray) may also influence model robustness, despite our efforts to perform normalization and validation procedures. Therefore, prospective validation in large, multicenter, and well-annotated cohorts is still required to confirm the clinical utility of the proposed model. Second, although comprehensive bioinformatics analyses revealed significant associations between steroid metabolism, immune regulation, tumor mutational burden, and patient prognosis, the underlying molecular mechanisms remain largely speculative. The current findings are based on transcriptomic correlations and therefore cannot establish causal relationships. Functional experiments, such as in vitro and in vivo studies, are necessary to validate the biological roles of the identified genes and to elucidate the mechanistic links between steroid metabolism reprogramming and gastric cancer progression. Third, the predictive performance of the risk score, particularly for immunotherapy response and drug sensitivity, was derived from computational inference rather than direct clinical evidence. While these algorithms provide useful hypotheses, they cannot fully recapitulate the complexity of real-world treatment response. Consequently, validation in prospective clinical trials or immunotherapy-treated cohorts is essential before clinical translation. Finally, although an external cohort (GSE84437) was used for validation, the number of available independent datasets remains limited, and differences in population characteristics and sequencing platforms may still affect model stability. Future studies incorporating more diverse populations and integrating multi-omics data may further improve the robustness and applicability of the model.

5. Conclusion

In conclusion, our study provides a comprehensive overview of steroid metabolism–associated molecular heterogeneity in gastric cancer. The identified steroid metabolism subtypes and the derived prognostic risk score model offer valuable tools for risk stratification and prognostic evaluation. Moreover, the close association between steroid metabolism, immune suppression, tumor mutational burden, and drug sensitivity highlights its potential as a therapeutic target and a basis for personalized treatment strategies in gastric cancer.

Supplemental Material

Supplemental Material - Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Supplemental Material for Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer by Linen Li, Huiling Zhu, Guang Gao, Hao Chen in Cancer Informatics

Supplemental Material

Supplemental Material - Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Supplemental Material

Supplemental Material - Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Supplemental Material

Supplemental Material - Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Supplemental Material

Supplemental Material - Integrative Analysis of Steroid Metabolism–Based Molecular Subtypes Reveals Prognostic Subtypes and Therapeutic Implications in Gastric Cancer

Footnotes

ORCID iD

Hao Chen

Author Contributions

L.L. collected data, analyzed relevant data and drafted the manuscript; H.Z. collected data and validated results; G.G. collected data and revised the manuscript; H.C. revised and edited the manuscript. All authors read and approved the final manuscript. They warrant that the article is the authors’ original work, hasn’t received prior publication and isn’t under consideration for publication elsewhere.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data available on request from the authors.*

Supplemental Material

Supplemental material for this article is available online.

References

Thrift

Wenker

El-Serag

. Global burden of gastric cancer: epidemiological trends, risk factors, screening and prevention. Nature Reviews Clinical Oncology. 2023;20(5):338-349. doi:10.1038/s41571-023-00747-0.

Zhang

Liu

Sun

. Research Progress of Immunotherapy for Gastric Cancer. Technol Cancer Res Treat. 2023;22:15330338221150555. doi:10.1177/15330338221150555. Epub 2023/04/13PubMed PMID: 37042029; PubMed Central PMCID: PMCPMC10102952.

Arai

Nakajima

. Recent Developments of Systemic Chemotherapy for Gastric Cancer. Cancers (Basel). 2020;12(5):1000. doi:10.3390/cancers12051100. Epub 2020/05/02PubMed PMID: 32354119; PubMed Central PMCID: PMCPMC7281322.

Jia

Wang

Yuan

Zhu

Long

. Heterogeneity of the tumor immune microenvironment and its clinical relevance. Exp Hematol Oncol. 2022;11(1):24. doi:10.1186/s40164-022-00277-y. Epub 2022/04/25PubMed PMID: 35461288; PubMed Central PMCID: PMCPMC9034473.

Chen

Han

Qing

. Metabolic heterogeneity in tumor microenvironment - A novel landmark for immunotherapy. Autoimmun Rev. 2024;23(6):103579. doi:10.1016/j.autrev.2024.103579. Epub 2024/07/15PubMed PMID: 39004158.

Wang

Sun

Kunzke

, et al. Metabolic heterogeneity affects trastuzumab response and survival in HER2-positive advanced gastric cancer. Br J Cancer. 2024;130(6):1036-1045. doi:10.1038/s41416-023-02559-6. Epub 2024/01/25PubMed PMID: 38267634; PubMed Central PMCID: PMCPMC10951255.

MacDonald

Purcell

Pinho-Schwermann

Stubbs

Srinivasan

El-Deiry

. Heterogeneity in Cancer. Cancers (Basel). 2025;17(3):441. doi:10.3390/cancers17030441. Epub 2025/02/13PubMed PMID: 39941808; PubMed Central PMCID: PMCPMC11816170.

Nong

Han

Xiang

, et al. Metabolic reprogramming in cancer: Mechanisms and therapeutics. MedComm. 2020;4(2):e218. doi:10.1002/mco2.218. Epub 2023/03/31PubMed PMID: 36994237; PubMed Central PMCID: PMCPMC10041388.

Zhu

Yan

Tong

, et al. Metabolic Reprogramming: A Crucial Contributor to Anticancer Drug Resistance. MedComm. 2025;6(9):e70358. doi:10.1002/mco2.70358. Epub 2025/09/08PubMed PMID: 40919131; PubMed Central PMCID: PMCPMC12413565.

10.

Huang

Han

Chen

Shen

Zeng

Cai

. Tumor metabolic regulators: key drivers of metabolic reprogramming and the promising targets in cancer therapy. Mol Cancer. 2025;24(1):7. doi:10.1186/s12943-024-02205-6. Epub 2025/01/10PubMed PMID: 39789606; PubMed Central PMCID: PMCPMC11716519.

11.

Smith

Ramar

Riley

3rd Mathias

Lee

. Steroid hormone regulation of immunometabolism and inflammation. Front Immunol. 2025;16:1654034. doi:10.3389/fimmu.2025.1654034. Epub 2025/10/03PubMed PMID: 41041336; PubMed Central PMCID: PMCPMC12483911.

12.

Chakraborty

Pramanik

Mahata

. Revisiting steroidogenesis and its role in immune regulation with the advanced tools and technologies. Genes Immun. 2021;22(3):125-140. doi:10.1038/s41435-021-00139-3. Epub 2021/06/16PubMed PMID: 34127827; PubMed Central PMCID: PMCPMC8277576.

13.

Shin

H-J

Kwon

Y-J

, et al. Human steroid sulfatase induces Wnt/β-catenin signaling and epithelial-mesenchymal transition by upregulating Twist1 and HIF-1α in human prostate and cervical cancer cells. Oncotarget. 2017;8(37):61604-61617. doi:10.18632/oncotarget.18645. PubMed Central PMCID: PMCPMC5617449.

14.

Liu

Goodwin

. The Effect of Glucocorticoids on Angiogenesis in the Treatment of Solid Tumors. J Cell Signal. 2020;1(3):42-49. PubMed Central PMCID: PMCPMC7388649.

15.

Swatler

Anderson

Lugli

. Tumors recycle glucocorticoids to drive Treg-mediated immunosuppression. J Clin Invest. 2023;133(18):e173141. doi:10.1172/JCI173141. Epub 2023/09/15PubMed PMID: 37712416; PubMed Central PMCID: PMCPMC10503790 serves as a consultant for BD Biosciences. ACA is a member of the scientific advisory board for Tizona Therapeutics, Trishula Therapeutics, Compass Therapeutics, Zumutor Biologics, ImmuneOncia, and Excepgen, which have interests in cancer immunotherapy. ACA is also a paid consultant for iTeos Therapeutics and Larkspur Biosciences.

16.

Hou

Wang

Kucharzewska

Guo

Zhang

. 27-Hydroxycholesterol in cancer development and drug resistance. J Enzyme Inhib Med Chem. 2025;40(1):2507670. doi:10.1080/14756366.2025.2507670. Epub 2025/05/22PubMed PMID: 40401382; PubMed Central PMCID: PMCPMC12100970.

17.

Raza

Meyer

Goodyear

Hammer

KDP

Guo

Ghribi

. The cholesterol metabolite 27-hydroxycholesterol stimulates cell proliferation via ERbeta in prostate cancer cells. Cancer Cell Int. 2017;17:52. doi:10.1186/s12935-017-0422-x. Epub 2017/05/16PubMed PMID: 28503095; PubMed Central PMCID: PMCPMC5425984.

18.

Capper

Rae

Auchus

. The Metabolism, Analysis, and Targeting of Steroid Hormones in Breast and Prostate Cancer. Horm Cancer. 2016;7(3):149-164. doi:10.1007/s12672-016-0259-0. Epub 2016/03/13PubMed PMID: 26969590; PubMed Central PMCID: PMCPMC4860032.

19.

Wenxuan

Liu

Zhang

Qiu

Deng

. Role of gonadally synthesized steroid hormones in the colorectal cancer microenvironment. Front Oncol. 2023;13:1323826. doi:10.3389/fonc.2023.1323826. Epub 2023/12/20PubMed PMID: 38115900; PubMed Central PMCID: PMCPMC10728810.

20.

Gan

Dai

Liu

. Intricate roles of estrogen and estrogen receptors in digestive system cancers: a systematic review. Cancer Biol Med. 2024;21(10):898-915. doi:10.20892/j.issn.2095-3941.2024.0224. Epub 2024/10/30PubMed PMID: 39475214; PubMed Central PMCID: PMCPMC11523274.

21.

Zhou

Meng

Xiao

Shen

. CYP19A1 promotes gastric cancer as part of a lipid metabolism-related gene signature related to the response of immunotherapy and prognosis. BMC Med Genomics. 2023;16(1):228. doi:10.1186/s12920-023-01664-y. Epub 2023/10/03PubMed PMID: 37784135; PubMed Central PMCID: PMCPMC10544546.

22.

Chang

W-C

Huang

S-F

Lee

Y-M

, et al. Cholesterol import and steroidogenesis are biosignatures for gastric cancer patient survival. Oncotarget. 2017;8(1):692-704.

23.

Eils

Schlesner

. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics. 2016;32(18):2847-2849. doi:10.1093/bioinformatics/btw313. Epub 2016/05/22PubMed PMID: 27207943.

24.

Smyth

Michaud

Scott

. Use of within-array replicate spots for assessing differential expression in microarray experiments. Bioinformatics. 2005;21(9):2067-2075. doi:10.1093/bioinformatics/bti270. Epub 2005/01/20PubMed PMID: 15657102.

25.

Love

Huber

Anders

. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. doi:10.1186/s13059-014-0550-8. Epub 2014/12/18PubMed PMID: 25516281; PubMed Central PMCID: PMCPMC4302049.

26.

Wilkerson

Hayes

. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26(12):1572-1573. doi:10.1093/bioinformatics/btq170. Epub 2010/04/30PubMed PMID: 20427518; PubMed Central PMCID: PMCPMC2881355.

27.

Wang

Han

. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284-287. doi:10.1089/omi.2011.0118. Epub 2012/03/30PubMed PMID: 22455463; PubMed Central PMCID: PMCPMC3339379.

28.

Hänzelmann

Castelo

Guinney

. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:14-17.

29.

Yoshihara

Shahmoradgoli

Martinez

, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4:2612. doi:10.1038/ncomms3612. Epub 2013/10/12PubMed PMID: 24113773; PubMed Central PMCID: PMCPMC3826632.

30.

Newman

Liu

Green

, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods. 2015;12(5):453-457. doi:10.1038/nmeth.3337. Epub 2015/03/31PubMed PMID: 25822800; PubMed Central PMCID: PMCPMC4739640.

31.

Bindea

Mlecnik

Tosolini

, et al. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity. 2013;39(4):782-795. doi: 10.1016/j.immuni.2013.10.003. Epub 2013/10/22PubMed PMID: 24138885.

32.

Mayakonda

Lin

Assenov

Plass

Koeffler

. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28(11):1747-1756. doi:10.1101/gr.239244.118. Epub 2018/10/21PubMed PMID: 30341162; PubMed Central PMCID: PMCPMC6211645.

33.

Maeser

Gruener

Huang

. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform. 2021;22(6):bbab260. doi:10.1093/bib/bbab260. Epub 2021/07/15PubMed PMID: 34260682; PubMed Central PMCID: PMCPMC8574972.

34.

Jardim

Goodman

de Melo Gagliato

Kurzrock

. The Challenges of Tumor Mutational Burden as an Immunotherapy Biomarker. Cancer Cell. 2021;39(2):154-173. doi:10.1016/j.ccell.2020.10.001. Epub 2020/10/31PubMed PMID: 33125859; PubMed Central PMCID: PMCPMC7878292.

35.

Yuan

Zhang

, et al. Extracellular matrix remodeling in tumor progression and immune escape: from mechanisms to treatments. Molecular Cancer. 2023;22(1):48. doi:10.1186/s12943-023-01744-8.

36.

Liu

Yang

Zhang

. Tumor metastasis: Mechanistic insights and therapeutic interventions. MedComm. 2021;2(4):587-617. doi:10.1002/mco2.100.

37.

Hamarsheh

Gross

Brummer

Zeiser

. Immune modulatory effects of oncogenic KRAS in cancer. Nat Commun. 2020;11(1):5439. doi:10.1038/s41467-020-19288-6. Epub 2020/10/30PubMed PMID: 33116132; PubMed Central PMCID: PMCPMC7595113.

38.

Valdembri

Serini

. The roles of integrins in cancer. Fac Rev. 2021;10:45. doi:10.12703/r/10-45. Epub 2021/06/17PubMed PMID: 34131655; PubMed Central PMCID: PMCPMC8170687.

39.

Huang

Guo

Wang

. KRAS mutation: from undruggable to druggable in cancer. Signal Transduct Target Ther. 2021;6(1):386. doi:10.1038/s41392-021-00780-4. Epub 2021/11/16PubMed PMID: 34776511; PubMed Central PMCID: PMCPMC8591115.

40.

Yang

Zhu

. SERPINE1 as a cancer-promoting gene in gastric adenocarcinoma: facilitates tumour cell proliferation, migration, and invasion by regulating EMT. J Chemother. 2019;31(7-8):408-418. doi:10.1080/1120009X.2019.1687996. Epub 2019/11/15PubMed PMID: 31724495.

41.

Wang

Gao

, et al. SERPINE1 Facilitates Metastasis in Gastric Cancer Through Anoikis Resistance and Tumor Microenvironment Remodeling. Small. 2025;21(19):e2500136. doi:10.1002/smll.202500136.

42.

Wang

Chen

Sun

, et al. Uncovering the potential of APOD as a biomarker in gastric cancer: A retrospective and multi-center study. Comput Struct Biotechnol J. 2024;23:1051-1064. doi:10.1016/j.csbj.2024.02.015.

43.

Ren

, et al. Apolipoproteins and cancer. Cancer Med. 2019;8(16):7032-7043. doi:10.1002/cam4.2587. Epub 2019/10/02PubMed PMID: 31573738; PubMed Central PMCID: PMCPMC6853823.

44.

<PRICKLE1 promotes gastric cancer (2).pdf>.

45.

Brubaker

. Glucagon-like Peptide-2 and the Regulation of Intestinal Growth and Function. Compr Physiol. 2018;8(3):1185-1210. doi:10.1002/cphy.c170055. Epub 2018/07/07PubMed PMID: 29978894.

46.

Huang

Peng

, et al. Development of Tumor Mutation Burden-Related Prognostic Model and Novel Biomarker Identification in Stomach Adenocarcinoma. Front Cell Dev Biol. 2022;10:790920. doi:10.3389/fcell.2022.790920. Epub 2022/04/12PubMed PMID: 35399509; PubMed Central PMCID: PMCPMC8983817.

47.

Yuan

. Gastrin-releasing peptide links stressor to cancer progression. J Cancer Res Clin Oncol. 2010;136(4):483-491. doi:10.1007/s00432-010-0766-2. Epub 2010/02/09PubMed PMID: 20140628; PubMed Central PMCID: PMCPMC11827830.

48.

Wang

Zhang

, et al. Development of oxidative stress- and ferroptosis-related prognostic signature in gastric cancer and identification of CDH19 as a novel biomarker. Hum Genomics. 2024;18(1):121. doi:10.1186/s40246-024-00682-w. Epub 2024/11/06PubMed PMID: 39501397; PubMed Central PMCID: PMCPMC11536560.

49.

Betz

Thakur

Junge

, et al. Functional interaction of the active zone proteins Munc13-1 and RIM1 in synaptic vesicle priming. Neuron. 2001;30(1):183-196.

50.

Rassart

Desmarais

Najyb

Bergeron

Mounier

Apolipoprotein

. Gene. 2020;756:144874. doi:10.1016/j.gene.2020.144874. Epub 2020/06/20PubMed PMID: 32554047; PubMed Central PMCID: PMCPMC8011330.

51.

Mukherjee

Xiao

. GLP-2 regulation of intestinal lipid handling. Front Physiol. 2024;15:1358625. doi:10.3389/fphys.2024.1358625. Epub 2024/03/01PubMed PMID: 38426205; PubMed Central PMCID: PMCPMC10902918.

52.

Vrzalova

Vrzal

. Orchestra of ligand-activated transcription factors in the molecular symphony of SERPINE 1 / PAI-1 gene regulation. Biochimie. 2025;228:138-157. doi:10.1016/j.biochi.2024.09.010. Epub 2024/09/26PubMed PMID: 39321911.

53.

Zavros

Merchant

. The immune microenvironment in gastric adenocarcinoma. Nat Rev Gastroenterol Hepatol. 2022;19(7):451-467. doi:10.1038/s41575-022-00591-0. Epub 2022/03/16PubMed PMID: 35288702; PubMed Central PMCID: PMCPMC9809534.

54.

Liu

Chu

, et al. JAK inhibition in PD-1 immunotherapy and tumor microenvironment. Front Immunol. 2026;17:1790936. doi:10.3389/fimmu.2026.1790936. Epub 2026/04/20PubMed PMID: 42004951; PubMed Central PMCID: PMCPMC13083078.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.13 MB

0.05 MB

0.07 MB

0.10 MB

0.30 MB