Abstract
The epigenetic regulation of N6-methyladenosine (m6A) has attracted considerable interest in tumor research, but the potential roles of m6A regulator-related genes, remain largely unknown within the context of gastric cancer (GC) and tumor microenvironment (TME). Here, a comprehensive strategy of data mining and computational biology utilizing multiple datasets based on 28 m6A regulators (including novel anti-readers) was employed to identify m6A regulator-related genes and patterns and elucidate their underlying mechanisms in GC. Subsequently, a scoring system was constructed to evaluate individual prognosis and immunotherapy response. Three distinct m6A regulator-related patterns were identified through the unsupervised clustering of 56 m6A regulator-related genes (all significantly associated with GC prognosis). TME characterization revealed that these patterns highly corresponded to immune-inflamed, immune-excluded, and immune-desert phenotypes, and their TME characteristics were highly consistent with different clinical outcomes and biological processes. Additionally, an m6A-related scoring system was developed to quantify the m6A modification pattern of individual samples. Low scores indicated high survival rates and high levels of immune activation, whereas high scores indicated stromal activation and tumor malignancy. Furthermore, the m6A-related scores were correlated with tumor mutation loads and various clinical traits, including molecular or histological subtypes and clinical stage or grade, and the score had predictive values across all digestive system tumors and even in all tumor types. Notably, a low score was linked to improved responses to anti-PD-1/L1 and anti-CTLA4 immunotherapy in three independent cohorts. This study has expanded the important role of m6A regulator-related genes in shaping TME diversity and clinical/biological traits of GC. The developed scoring system could help develop more effective immunotherapy strategies and personalized treatment guidance.
Similar content being viewed by others
Introduction
Gastric cancer (GC) is a highly prevalent tumor, ranking fifth with an incidence rate of 5.6% among all cancers1. Most patients with GC are diagnosed at an advanced stage, due to its asymptomatic nature2. Current treatment options for GC mainly include surgery, chemotherapy, radiotherapy, and immunotherapy3,4. Immunotherapy, especially immune checkpoint inhibitors (ICIs), is emerging as the first-line treatment of advanced or metastatic GC5. The immunological landscape within the tumor microenvironment (TME) plays a critical role in selecting appropriate immunotherapy strategies and predicting treatment outcomes. Tumors can be immunologically categorized as immune-inflamed, immune-desert, or immune-excluded phenotypes according to the extent of immune cell infiltration and activation status6. An immune-inflamed phenotype, characterized by robust immune cell infiltration and activated immunity, is associated with a favorable response to immunotherapy. On the other hand, poor infiltration is indicative of an immune-desert phenotype while a high density of immune cells at the tumor margin without infiltration corresponds to an immune-excluded phenotype6. Both immune-desert and immune-excluded phenotypes are associated with low response to immunotherapy and poor clinical outcomes7,8. These three categories are designed as clinically relevant classifiers, with “hot” representing the immune-inflamed phenotype and “cold” corresponding to both immune-desert phenotype and immune-excluded phenotype9,10. In summary, the TME is closely related to the selection of immunotherapy strategies and immunotherapeutic efficacy, and accurately assessing the exact immune phenotype within the microenvironment can offer clinical benefits.
N6-methyladenosine (m6A) modification is the most prevalent internal RNA modification in eukaryotes and plays a pivotal role in regulating post-transcriptional gene modification11,12,13. m6A modification is regulated by a group of m6A regulators, including “writers,” “erasers,” “readers,” and “anti-readers.” The writers, which are methyltransferases, include METTL3, RBM15, METTL14, ZC3H13, WTAP, and CBLL1. Conversely, the erasers, which function as demethyltransferases, comprise ALKBH5 and FTO. Together, they maintain the dynamic balance of m6A modification14. A diverse array of RNA-binding proteins named “readers” include HNRNPA2B, YTHDC1/2, YTHDF1/2/3, LRPPRC, and FMR1 and play a crucial role in the recognition of m6A motifs15. Additionally, novel regulators, such as EWSR1, G3BP1, and LIN28A, act as anti-readers during m6A modification on RNA, consequently diminishing the efficiency of protein-RNA binding16,17. m6A regulators are involved in various crucial biological processes, such as tissue development, spermatogenesis, cell proliferation, and stem cell differentiation18,19,20. The aberrant expression of m6A regulators and/or genetic changes are closely associated with the occurrence, metastasis, and chemo- and radio-resistance of multiple malignant tumors, including GC, lung cancer, and hepatocellular carcinoma21,22. Moreover, m6A regulators play a crucial role in TME regulation. For instance, FTO, the first known m6A demethyltransferase, is closely related to TME remodeling and tumor escape23,24, and FTO inhibitors exhibit strong antitumor effects in various types of cancer25. A combination of treatment with FTO inhibitors and anti-PD-1 therapy exhibits good clinical efficacy in patients with melanoma26.
The pivotal roles of m6A regulators in TME regulation have attracted extensive attention27,28. Additionally, m6A regulator-related noncoding RNAs (ncRNAs), especially long noncoding RNAs (lncRNAs), have attracted considerable research. Emerging lines of studies have uncovered the clinical relevance of m6A regulator-related lncRNAs in various tumors, such as GC, lung adenocarcinoma, colon adenocarcinoma, breast cancer, and bladder cancer, and these lncRNA signatures exhibit excellent clinical applicability in tumor diagnosis and prognosis29,30,31,32. Furthermore, investigations demonstrate a close correlation between m6A regulator-related lncRNA signatures and TME, potentially predicting the efficacy of tumor immunotherapy31,33,34. However, the role of m6A regulator-related coding genes in tumors is rarely reported, despite their fundamental contributions to nearly all biological processes. In this investigation, we aim to comprehensively evaluate the m6A regulator-related patterns, derived from both coding genes and ncRNA genes, as well as their correlation with the immune cell infiltrating landscape of the TME. We have identified three distinct m6A regulator-related patterns, and found that the TME traits of these patterns were highly consistent with the immune-inflamed, immune-excluded, and immune-desert phenotypes, indicating that m6A regulator-related genes play crucial roles in TME modulation. To quantify the m6A regulator-related patterns in individual patients, we have developed a corresponding scoring system, which can be utilized for clinical treatment decisions and prognostic assessment for patients with GC.
Results
Identification of m6A regulator-related genes in the Meta-cohort and TCGA dataset of GC
To evaluate the potential role of m6A regulator-related genes, four public GC datasets from GEO were enrolled into one Meta-cohort, which consisted of 697 samples after batch effects and PCA outliers were removed (Additional files 1 and 2: Table S1 and Fig. S1A). Then, the expression data of 28 m6A regulators from the Meta-cohort were gathered, and their expression levels in normal and GC tissues were explored. The results showed that most of the m6A regulators exhibited differential expression profiles between the normal and GC samples in the Meta-cohort (Fig. 1A), and similar results were acquired when we analyzed the TCGA-STAD cohort (Fig. 1B). Subsequently, the protein expression levels of all 28 m6A regulators were further investigated using proteomics data from the gastric cancer (PDC000214) cohort within CPTAC project. Our findings indicated that 27 of these regulators were detected at the protein expression level. Notably, the majority of these m6A regulators exhibited different protein expression levels when compared to the control group (Fig. S1B and Table S2). These results showed the differential expression spectrum of m6A regulators, suggesting the aberrant expression of m6A regulators played an important role in the tumorigenesis and progression of GC.
To investigate m6A regulator-related genes, firstly, a total of 1025 candidate genes, whose expression levels closely correlated with m6A regulators, were selected based on the criterion of Pearson R > 0.5; P-value < 0.001 (Additional file 1: Tables S3 and S4). The results demonstrated that 27 of the 28 m6A regulators had correlated genes, and FTO exhibited the largest number of related genes (505 genes), followed by HNRNPC (160 genes). A Sankey diagram was used to visualize the connection between m6A regulators and their related genes (Fig. 2A). Subsequently, univariate Cox regression analysis was performed. 56 of 1025 candidate genes found to be significantly related to the prognosis of GC in both Meta-cohort and TCGA-STAD cohort, and these were designated as m6A regulator-related genes (Additional files 1 and 2: Fig. S2A, Tables S5 and S6). The detailed correlation between the 56 m6A regulator-related genes and 28 m6A regulators in the Meta-cohort was illustrated in Fig. 2B. Based on the expression levels of these 56 m6A regulator-related genes, the GC samples and normal samples could be completely distinguished (Fig. 2C). Further validation analysis confirmed that these m6A regulator-related genes exhibited significantly expression profiles at both the transcriptional (GSE54129) and protein levels (PDC000214) between normal and GC samples (Fig. S2B and C and Table S7). Collectively, our study identified 56 m6A regulator-related genes that were closely associated with GC, and further investigating their roles in GC could provide novel insight into the pathogenesis, clinical diagnosis and prognosis of GC.
Distinct m6A regulator-related patterns mediated by m6A regulator-related genes and hallmark gene analysis
To investigate the functions of m6A regulator-related genes in GC, unsupervised clustering with the R package ConsensusClusterPlus was performed to categorize patients into distinct patterns based on the expression levels of 56 m6A regulator-related genes. Three distinct m6A regulator-related patterns were eventually identified, and were abbreviated as m6A-related clusters 1–3 (or m6A-related patterns 1–3) in the following manuscript. There was a significantly different expression profiles among these three m6A-related clusters (Fig. 3A). Notably, m6A-related cluster 1 was characterized by the expression of genes, such as CHAF1A, EID1, and MID2, which are involved in immune activation, cell cycle regulation, cell proliferation, and DNA repair (Additional file 2: Fig. S3A). By contrast, m6A-related cluster 3 exhibited high expression levels of genes, such as LRCH2, SLIT2, and DDR2, which are primarily associated with cell-extracellular matrix interaction, angiogenesis, cytoskeleton, and cell migration (Additional file 2: Fig. S3A). Intriguingly, the majority of m6A regulator-related genes correlated with angiogenesis, EMT, signal transduction, and immune activation were highly expressed in m6A-related cluster 2 (Additional file 2: Fig. S3A). Subsequently survival analysis revealed that patients in m6A-related cluster 1 had a significantly better prognosis than those in m6A-related clusters 2 and 3 (HR, 1.48 [1.11–1.99]; Fig. 3B), which may be attributed to the distinct profiles of m6A regulator-related genes. These results indicated that m6A regulator-related genes were closely related to immune response, and investigating their roles may provide new ideas for understanding the TME modulation in GC.
Biological annotation and TME immune infiltration in diverse m6A-related patterns
To comprehensively evaluate the biological features of m6A-related clusters 1–3 in TME, GSVA enrichment analysis was performed on m6A-related patterns of individual patients with GC. The results showed that m6A-related cluster 1 was significantly enriched in immune and DNA repair pathways, including natural killer cell-mediated cytotoxicity, mismatch repair, and base excision repair (Fig. 3C–D and Additional file 1: Table S8), whereas m6A-related cluster 3 was enriched in stromal and carcinogenic activation pathways, such as the TGF-β signal pathway (Fig. 3D and Additional file 1: Table S8). Meanwhile, m6A-related cluster 2 exhibited significant enrichment in stromal and immune activation pathways, including the TGF-β signal pathway, ECM receptor interaction, focal adhesion, cell adhesion molecules, and cytokine-cytokine receptor signaling pathways (Fig. 3C and Additional file 1: Table S8).
To further confirm the TME landscape, we conducted immune cell infiltration analysis in these three patterns (Additional file 1: Tables S9 and S10). We found that these different m6A-related patterns exhibited distinct characteristics of TME immune cell infiltration (Fig. 4A). m6A-related cluster 1 displayed a dramatic advantage in immune activation. Most types of immune cells, including CD8+ T and CD4+ T cells, which represent immune activation, were significantly enriched in m6A-related cluster 1 (Fig. 4A and Additional file 1: Table S10). m6A-related cluster 3 exhibited the worst performance in T cell infiltration and was notably enriched in immunosuppressive cells, such as Tregs and MDSCs (Fig. 4A). Additionally, m6A-related cluster 2 was enriched not only in T cells but also in immunosuppressive cells (Fig. 4A). These results were consistent with the above showing that patients in m6A-related clusters 2 and 3 had worse survival outcomes than those in m6A-related cluster 1 (Fig. 3B). In addition, pathway enrichment analysis demonstrated that m6A-related cluster 1 was significantly enriched in the immune activation pathway, especially various pathways related to DNA repair, such as mismatch repair, the roots of microsatellite instability (MSI; Fig. 4B). Previous studies demonstrated that patients with MSI-H showed a high level of immune activation and enhanced response to immunotherapy35,36. These results indicated that patients in m6A-related cluster 1 may benefit from immunotherapy. By contrast, m6A-related cluster 3 was markedly enriched in the stromal activation pathways, including EMT, the TGF-β pathway, and angiogenesis (Fig. 4B), which were related to the immunosuppression of TME37. Notably, m6A-related cluster 2 was enriched in pathways related to stromal and immune activation (Fig. 4B). Based on above results, we found three m6A-related patterns exhibited distinct immune-infiltration profiles. m6A-related cluster 1 was categorized as an immune-inflamed phenotype, featured by a high level of immune cell infiltration. Similarly, m6A-related cluster 2 was categorized as an immune-excluded phenotype, featured by immune cell enrichment and stromal activation. Additionally, m6A-related cluster 3 was categorized as an immune-desert phenotype, featured by suppressed immunity. To further validate the biological characteristics in distinct m6A-related patterns, DEGs between these m6A-related patterns were identified using the “limma” R package. Then, GO enrichment analysis for 401 DEGs was conducted by the ClusterProfiler package and the enriched biological processes were listed in Additional file 1: Table S11. Results revealed that these genes were significantly enriched in biological processes related to immunity and EMT, such as extracellular matrix organization, regulation of angiogenesis, regulation of T cell activation and T cell proliferation, which reaffirmed that m6A -related patterns were closely related to immune modulation in TME (Fig. 4C). The above analyses demonstrated that various m6A-related patterns had a distinct TME cell infiltration landscape, and evaluating immune characteristics of m6A-related patterns will provide important insights into the understanding of TME.
Identification of m6A-related patterns and integrated analysis of clinical traits in the ACRG cohort
To further investigate the biological behavior and clinical traits of these patterns, the ACRG cohort was selected as it contained the most comprehensive clinical information. Similar to the Meta-cohort clustering, unsupervised analysis also revealed three distinct m6A-related patterns (Fig. 5A), and significant differences in transcriptional profile were observed among these patterns (Fig. 5B, Additional file 2: Fig. S4A). m6A-related cluster 1 exhibited high expression of CHAF1A and a decline in the expression levels of other m6A regulator-related genes (Additional file 2: Fig. S4A). Meanwhile, m6A-related cluster 2 dominantly expressed TIMP2 and RAB31, while m6A-related cluster 3 showed high expression of RECK, FERMT2 and SLIT2, etc. (Additional file 2: Fig. S4A). Then, the correlation of the clinical characteristics with m6A-related patterns were further explored (Fig. 5C–F and Additional file1: Table S12). Patients with MSI types were characterized by m6A-related cluster 1, whereas patients with EMT types were characterized by m6A-related cluster 3 (Fig. 5C). We also noted that patients in m6A-related clusters 2 and 3 exhibited higher pathological stages (stages III/IV), and were enriched in the mesenchymal phenotype (MP) and diffuse histological type (Fig. 5D–F). m6A-related cluster 1, which was mostly enriched in the intestinal histological type and epithelial phenotype (EP), exhibited an earlier stage (stages I/II) (Fig. 5D–F). In GC, EMT type, higher pathological stages, diffuse histological type and MP were significantly related to poorer prognosis, while patients with MSI type, earlier stage, intestinal histological type and EP were associated with better clinical outcomes38,39,40. Consistent with the above results, survival analysis confirmed that m6A-related cluster 1 was significantly linked to prolonged survival (HR, 1.63 [1.10–2.44]), while m6A-related clusters 2 and 3 were related to poor survival, especially m6A-related cluster 3 (Additional file 2: Fig. S4B). These results revealed distinct m6A-related patterns exhibited different clinical characteristics and clinical prognosis. Patients with EMT types were mostly enriched in m6A-related cluster 3, while few were clustered into m6A-related cluster 1, which again reinforced m6A-related pattern 1 was related to immune activation and m6A-related pattern 3 was linked to stromal activation.
Landscape of the immune/stromal activation-related index of m6A-related patterns
To elucidate the mechanisms of distinct m6A-related patterns on TME modulation, we performed a comprehensive analysis of cytokine/chemokine and immune checkpoint expression in the three m6A-related patterns. This approach enabled us to comprehensively evaluate the intricate interplay between m6A-related patterns and the immune landscape within the TME. The factors were selected according to previously published literature, of which TGFB1, TGFB2, TGFB3, IL10, CCL22, PRF1, GZMA, GZMB, IL2, IL23A, TNF, IL1B, and CCL28 were considered to be involved in immune regulation. CTLA4, IDO1, LAG3, PDCD1, CD274, PDCD1LG2, TIGIT, TNFRSF9, LMTK3, and HAVCR2 were considered as immune checkpoints. ACTA2, COL4A1, PDGFRA, SMAD9, TGFB2, TGFBR2, CLDN3, ZEB1-AS1, VIM, and TWIST1 were considered to be associated with stromal activation41,42,43. Our analysis revealed that the transcripts facilitating immune activation, such as IL23A, TNF, and IL1, were notably upregulated in m6A-related clusters 1 and 2 (Additional file 2: Fig. S5A). These transcripts enhance the activation and expansion of T cells44. In addition, m6A-related cluster 2 also showed significantly evaluated expressions of transcripts related to immunosuppressive molecules, immune checkpoints, and stromal activation, especially TGFB1 and IL10 (Additional file 2: Fig. S5A-C). m6A-related cluster 3 upregulated the expression of mRNAs correlated with immunosuppressive molecules and stromal activation (Additional file 2: Fig. S5A and C), but had the lowest expression levels of transcripts related to immune activation (Additional file 2: Fig. S5A). Previous studies have demonstrated that TGF-β, IL10, and some immune checkpoints were closely correlated with the differentiation of Tregs45,46. Therefore, these results reaffirm our results in Fig. 4B, showing that Tregs were mainly enriched in m6A-related clusters 2 and 3. The aforementioned findings indicate that m6A-related cluster 1 is characterized by T cell activation, corresponding to the immune-inflamed phenotype; m6A-related cluster 2 can recruit T cells but expresses various immunosuppressive molecules and stromal activation molecules, preventing T cells from entering the tumor parenchyma, corresponding to the immune-excluded phenotype; m6A-related cluster 3 is featured by immune suppression, corresponding to the immune-desert phenotype.
Next, we conducted a profound analysis of representative GC tissue slides derived from the TCGA-STAD dataset (n = 325), with the objective of meticulously examining the phenotypic traits associated with distinct m6A-related patterns histologically. Through semi-quantitative pathological evaluation, we observed a significant preponderance of lymphocyte score in m6A-related cluster 1, demonstrating a notably higher degree of lymphocyte infiltration compared to other clusters (Fig. 6A). Subsequently, m6A-related cluster 2 exhibited a certain degree of lymphocyte infiltration, albeit slightly inferior to m6A-related cluster 1 in terms of lymphocyte score (Fig. 6A). Conversely, m6A-related cluster 3 displayed a significantly lower level of lymphocyte infiltration compared to m6A-related cluster 1 (Fig. 6A). Additionally, m6A-related cluster 1 exhibited the lowest stroma score (Fig. 6B). Furthermore, our analysis revealed significant disparities in the histologic characteristics of lymphocytes and stromal spatial distribution among the various m6A-related patterns (Fig. 6C–E). Notably, m6A-related cluster 1 displayed a substantial number of lymphocytes tightly infiltrating the tumor cells, accompanied by relatively low stromal infiltration (Fig. 6C). This distribution pattern mirrors the immune-inflamed phenotype, suggesting a more robust immune response47,48. In contrast, m6A-related cluster 2 exhibited a coexistence of lymphocytes and stromal infiltration, with lymphocytes primarily localized within the stroma and a comparably lower number surrounding the tumor cells (Fig. 6D). This distribution pattern is reminiscent of the immune-excluded phenotype, indicating immune activation capacity that is nevertheless constrained by stromal infiltration47,48. In contrast, m6A-related cluster 3 exhibited an extremely low lymphocyte content accompanied by stromal infiltration, which aligns with an immune-desert phenotype47,48 (Fig. 6E). This observation suggests a limited immune response in this cluster. These findings further corroborate the close correlation between distinct m6A-related patterns and varying immune phenotypes. Consequently, identification of m6A-related patterns offers a valuable approach for analyzing the TME immune phenotypes of GC patients, providing significant insights into the prognosis and immunotherapy responsiveness of these individuals.
Construction of m6A-related signatures and characterization of clinical and biological traits
The m6A-related patterns play a crucial role in shaping distinct TME immune-infiltration phenotypes. However, we cannot use m6A-related patterns to predict TME characteristics in individual patients, as the aforementioned analyses primarily focused on patient populations. Given the complexity of m6A-related patterns and individual heterogeneity, we devised a scoring system to quantify these patterns in individual patients according to the 56 m6A regulator-related genes. We designated this scoring system as the m6A-related score. First, the properties of individual patients were investigated and visualized using an alluvial diagram (Fig. 7A), which showed that patients in m6A-related cluster 1 had a lower m6A-related score than those in m6A-related cluster 3. To further evaluate the significance of m6A-related score, we examined the relationship between m6A-related scores and some well-known signatures. The m6A-related score was positively correlated with stromal activation-related signatures, but negatively with DNA repair-related signatures (Fig. 7B and Additional file 1: Table S13). The analyses for immune activation and stromal activation showed low scores were significantly associated with DNA damage repair and mismatch repair, whereas high scores were related to EMT, TGF-β pathway, and angiogenesis (Fig. 7C). Similarly, Kruskal–Wallis tests found a significant difference of m6A-related score between different m6A-related patterns. m6A-related cluster 1 exhibited the lowest median m6A-related score, whereas m6A-related cluster 3 had the highest score (Fig. 7D). In addition, patients with EMT types exhibited the highest m6A-related score compared to other ACRG types (Fig. 7D). These results demonstrate that a high m6A-related score is linked to stromal activation, while a low m6A-related score is associated with immune activation.
Subsequently, we further evaluated the value of the m6A-related score in predicting prognosis for patients with GC. At the cutoff value 0.34 according to the survminer package, patients were categorized into high or low m6A-related score groups. Survival analysis demonstrated that the high m6A-related score group had poor prognosis (Fig. 7E). Whether the m6A-related scores could serve as independent prognostic biomarkers for patients with GC was also investigated. A multivariate Cox regression model, considering age, gender, stage, and TNM status of patients, confirmed m6A-related scores were robust and independent prognostic biomarkers for patients with GC (HR, 1.78 [1.44–2.20] in the Meta-cohort; Additional file 2: Fig. S6A; HR, 1.38 [1.15–1.66] in the TCGA-STAD cohort; Additional file 2: Fig. S6B). We tested the performance of the m6A-related score in predicting the efficacy of adjuvant chemotherapy in patients with GC. The results demonstrated that patients with low m6A-related scores and receiving adjuvant chemotherapy had significantly high survival rates (HR, 2.70 [1.26–5.80]; Fig. 7F). Furthermore, patients with low m6A-related scores exhibited high survival rates with or without receiving adjuvant chemotherapy (Fig. 7F). Finally, we demonstrated that patients with diffuse histological subtype, elder age, advanced stage, and EMT type are notably related to a higher m6A-related score, which suggested these patients were corresponding to m6A-related cluster 3 and immune-desert phenotype, with poor prognosis (Additional file 2: Fig. S6C). These results demonstrated the m6A-related score was an independent and robust prognosis biomarker for patients with GC and can be used to evaluate clinical characteristics.
Characteristics of m6A-related scores in TCGA-STAD cohort and tumor somatic mutation analysis
A comprehensive molecular phenotype has been built for GC by TCGA project, which categorized patients with GC into five molecular subtypes, including chromosomal instability, genome stability (GS), EBV infection, MSI, and hypermutated single nucleotide variation (HM-SNV)49,50. We evaluated the distinction m6A-related score between these molecular subtypes. Our results revealed that the high m6A-related score group was significantly enriched in the GS subtype and exhibited a poor prognosis with a median survival time of approximately 5.08 years, whereas the low m6A-related score group was focused on the molecular subtypes of HM-SNV, MSI, and EBV, which were linked to survival benefit (median survival time of approximately 6.92 years; Fig. 8A and B). Multivariate Cox regression analysis verified that the m6A-related score is an independent prognostic biomarker for patients with GC (Additional file 2: Fig. S6B). Moreover, elevated m6A-related scores were observed predominantly in the MSS molecular subtype, advanced stage patients, and G3 patients, which were associated with poor prognosis (Fig. 8C). Previous studies have showed that the molecular subtypes of GC are closely related to clinical response to immunotherapy. GC patients with the EBV-positive, MSI and HM-SNV molecular subtypes exhibited improved efficacy after immunotherapy, as the increasing expression of PD-L1 or a high tumor mutational burden (TMB)49,51,52. In our study, patients with MSI and EBV molecular subtypes were significantly related to m6A-related cluster 1 with low m6A-related score, whereas patients with GS molecular subtype were related to m6A-related cluster 3 with a high m6A-related score (Additional file 2: Fig. S7A). Similar analysis revealed that patients with high differentiation (G III), early stage (stage I), and MSI molecular subtype were concentrated in m6A-related cluster 1 (Additional file 2: Figs. S7A–D).
Clinical trials have shown that tumor somatic mutations are closely linked to immunotherapy. Higher somatic mutation rates led to an efficient response to ICIs, such as anti-PD-1/PD-L1 immunotherapy53. Then, the differences in somatic mutations between high and low m6A-related score groups were investigated in the TCGA-STAD cohort. We found high m6A-related score group exhibited lower TMB status than low m6A-related score group, with the rate of the 40th highest mutated genes being 12% versus 15% (Fig. 8D–E). This observation was verified using TMB quantification analysis, as the high m6A-related score group had a lower TMB score (Additional file 2: Fig. S7E). Similarly, correlation analysis demonstrated that the m6A-related score was negatively correlated with the TMB status (Additional file 2: Fig. S7F). Accumulated results have reported that patients with high TMB status showed a better response to immunotherapy54. Therefore, the above results indicated that the distinct m6A-related patterns may be an important factor for the response to immunotherapy, and evaluating the m6A-related score of individual patients may be beneficial for predicting efficacy of immunotherapy.
Previous investigations have demonstrated that patients with higher TMB status exhibited a better response to ICIs. The mutation of some crucial genes may have a close relationship with sensitivity or resistance to immunotherapy. For genes in TCGA-STAD cohort such as TP53 and PIK3CA, wild type has remarkably higher m6A-related score compared to mutant type, whereas there was no difference in RHOA and ARID1A (Fig. 8C). The results above suggested exploring the mechanism of m6A-related patterns in TMB, which may provide new perspective for TME modulation and ICIs.
Clinical value of m6A-related scores for predicting prognosis and immunotherapy with anti-PD-1/L1 and anti-CTLA4
Our m6A-related scoring system exhibited predictive value in the TCGA-STAD cohort, confirming its potential as a prognostic biomarker for GC. To further explore its prognostic efficiency, we verified it in other independent GC cohorts, including GSE15459, HR = 2.14 (1.43–3.22); GSE34942, HR = 1.52 (0.67–3.46); GSE57303, HR = 1.57 (0.80–3.09); and GSE26253, HR = 2.0 (1.43–2.82; Fig. S8A). Furthermore, we extended the m6A-related scoring system to all digestive system tumors, including esophageal carcinoma, colorectal adenocarcinoma, cholangiocarcinoma, pancreatic adenocarcinoma, and hepatocarcinoma in the TCGA database (HR, 1.76 [1.42–2.17]; Additional file 2: Fig. S8E) and even the combined set of all TCGA tumor types (HR, 2.58 [2.38–2.80]; Additional file 2: Fig. S8F). The above results demonstrated that m6A-related score exhibited an excellent prognostic value. ROC curves further confirmed the predictive performance of m6A-related scoring system, especially in elderly patients (Additional file 2: Figs. S8G and H).
Immunotherapies including PD-1, PD-L1 and CTLA4 blockade have been recognized as a major breakthrough in GC therapy. Thus, we explored the effect of m6A-related score in predicting the response to ICIs in patients from three immunotherapy cohorts: the anti-PD-L1 cohort (IMvigor210), the anti-PD-1 cohort (GSE78220), and the anti-PD-1/anti-CTLA4 cohort (GSE91061). Among these three cohorts, patients with low m6A-related score had prolonged survival and improved response to ICIs (Fig. 9A–I; IMvigor210, HR = 1.62 [1.20–2.19], Fig. 9A; GSE78220, HR = 10.55 [3.40–32.72], Fig. 9D; GSE91061, HR = 3.02 [1.52–6.01], Fig. 9G). The response to immunotherapy and therapeutic advantage in patients with a low m6A-related score were verified (Fig. 9B and C; E and F and H and I). In addition, further investigations revealed Angiogenesis, WNT target and EMT were remarkable activated in patients with high m6A-related score, which mediated immune suppression of TME (Fig. 9J). Meanwhile, pathways related to DNA repair, such as DNA damage repair, Nucleotide excision repair, Mismatch repair and homologous recombination, were significantly activated in patients with a low m6A-related score, which was associated with improved response to immunotherapy. Moreover, another indicator closely related to immunotherapy, tumor neoantigen burden, was also evaluated. The results showed that patients with a low m6A-related score exhibited remarkable therapeutic advantages among patients with a high tumor neoantigen burden (Fig. 9K). Consequently, the above results suggested that m6A-related patterns had a close relationship with TME phenotypes and response to immunotherapy, and evaluating the m6A-related pattern/score in individual patients would improve the clinical efficacy prediction of immunotherapy.
Discussion
Accumulating evidence has proved the m6A modification plays an essential role in regulating immune cell infiltration and angiogenesis in TME by interaction with m6A regulators55,56,57. Meanwhile, m6A regulator-related ncRNAs were also reported to play an indispensable role in diagnosis, prognosis and even modulating TME in various tumors58,59,60. Nevertheless, most investigators mainly focused on m6A regulator-related ncRNAs. The comprehensive roles of m6A regulator-related genes, both coding genes and ncRNAs, in TME infiltration characterization are not well recognized. An integrated analysis of the functions of m6A-related patterns in GC TME modulation will enhance our understanding of TME antitumor immunity, and guide immunotherapy strategies.
In this study, 56 m6A regulator-related genes, comprising 54 coding genes and 2 ncRNA genes, were eventually identified. Based on the expression patterns of these related genes, three distinct m6A-related patterns were also established. Our findings revealed these three patterns have distinct TME infiltration landscape, respectively. m6A-related cluster 1 was featured by immune activation, corresponding to an immune-inflamed phenotype; while m6A-related cluster 2 was featured by stromal and immune activation, corresponding to an immune-excluded phenotype, m6A-related cluster 3 was featured by immune suppressing, corresponding to an immune-desert phenotype. The immune-inflamed phenotype is characterized by abundant infiltration of immune cells, clinically referred to as “hot” tumors, which exhibit favorable responsiveness to immunotherapy61. In contrast, the immune-excluded phenotype and the immune-desert phenotype, known as “cold” tumors, are characterized by little immune cell infiltration in TME61,62,63. Although the immune-excluded phenotype possesses abundant of immune cells, the immune cells mainly reside in the periphery of tumors parenchyma, rather than infiltrating the TME62,63. In line with our investigations, m6A-related cluster 1 we identified exhibits significant infiltration of T cells, monocytes, and antigen-presenting cells in TME (Fig. 4A) and infiltration of lymphocytes surrounding the tumor cells, resembling an immune-inflamed phenotype (Fig. 6C). This is also accompanied by a higher level of pro-inflammatory factors, such as IL-23, TNF-α and IL-1β, which are considered to be associated with immune activation44. m6A-related cluster 2 exhibited a notable infiltration of not only a substantial number of Tregs and MDSCs but also a considerable population of CD4+ T cells and CD8+ T cells (Fig. 4A). Furthermore, this cluster expressed a high abundance of transcripts linked to stromal activation (Fig. S5C). Pathological evaluation revealed that these lymphocytes were predominantly infiltrated in the stroma (Fig. 6D), suggesting a further suppression of immune activation. Notably, these features closely resembled an immune-excluded phenotype, indicating a potential mechanism underlying the immune evasion within the TME. m6A-related cluster 3, characterized by high expression levels of TGF-β, IL-10, IDO1, and EMT, exhibits a state of extensive stromal activation (Fig. S5A and B), which are all regarded as immune suppression64,65,66,67. Notably, this cluster shows an absence of infiltrating immune activation-associated cells (Figs. 4A and 6E), indicating a suppressive effect on immune activation within the TME, corresponding to an immune-desert phenotype. Additionally, an intriguing result was found when investigating the cytokines/chemokines profiles among three distinct clusters. The m6A-related cluster 1 exhibits a higher proportion of CD8+ T cells, but a lower level of GZMB (Fig. S5A and B), which seems to be contradictory to the traditional theory that CD8+ T cells are the primary source of GZMB secretion68. Recent studies have reported that Tregs, besides CD8+ T cells, are the primary source of GZMB secretion in TME69. More importantly, Tregs could modulate the TME to immune suppressive status through releasing GZMB, which concurrently induces apoptosis of effector T cells. In our research, both Tregs and GZMB are significantly enriched in m6A-related clusters 2 and 3, which suggesting Tregs may be the main source of GZMB release, and also may be the crucial manipulator for modulating the TME.
With the development of immunotherapy, the efficacy of immunotherapy has been clinically proven. However, only a small percentage of patients with cancer can benefit from immunotherapy70,71. Currently, predictive biomarkers for immunotherapy include not only the expression of immune checkpoints but also TMB, MSI, deficient mismatch repair genes (dMMR), and infiltration of tumor-specific T cells, all of which can predict the sensitivity of patients to immunotherapy72,73,74. dMMR impairs the ability to rectify mutated genes, resulting in elevated microsatellite instability (MSI-H) and subsequently increased TMB, eventually facilitating the recruitment of immune cells and enhancing the responsiveness to immunotherapy75,76. In our study, mismatch repair, base repair and other related pathways were enriched in the m6A-related cluster 1 (Fig. 3C and D), subsequently leading to higher MSI and TMB levels (Figs. 5C and S7A). Further survival analysis also revealed that individuals in the m6A-related cluster 1 have a better clinical prognosis and clinical efficacy after immunotherapy (Fig. 3B). Moreover, the expression levels of some immune checkpoint genes, which were the primary targets of immunotherapy, were also analyzed. Some star genes related to immunotherapy, such as PD-1, PD-L1 and CTLA4, exhibited relatively lower levels in m6A-related cluster 1 (Fig. S5B), despite the fact that patients in cluster 1 obtained better clinical benefits from immunotherapy. These results suggest that the predicting the clinical benefits from immunotherapy is extremely complicated, and various factors, such as TMB, MSI, immune checkpoint genes and maybe some unknown factors, may play an important role in improving the efficacy of immunotherapy. Combining multiple factors building efficient models for immunotherapy efficacy prediction is also urgently needed. Therefore, distinct m6A-related patterns exhibit different responsiveness to immunotherapy, which may serve as potential predictive biomarkers for immunotherapy.
Further, considering the heterogeneity among individual patients, we developed an m6A-related scoring system to assess the m6A-related patterns of individual GC patients. The m6A-related cluster 1 featuring an immune-inflamed phenotype exhibited lower m6A-related score, while the m6A-related clusters 2 and 3, featuring immune-exclude and immune-desert phenotypes, showed higher m6A-related scores. These results indicated that m6A-related score could be used to comprehensively evaluate individual patients with GC, confirming the TME immune infiltration patterns which are determining factors for immunotherapy. In addition, integrated analyses of the correlation of m6A-related score with various clinical traits also showed that patients with EMT and GS subtypes, closely linked to immune suppressive TME and a low response to immunotherapy77,78, were associated with higher m6A-related scores. However, patients with MSI, MSS and EBV subtypes, who had an improved response to immunotherapy79,80,81, had lower m6A-related scores. Meanwhile, the immunotherapy cohort results confirmed lower immunotherapy responsiveness in the high m6A-related score group, whereas the low m6A-related score group primarily consisted of patients from m6A-related cluster 1 (Fig. 7D), showing significantly higher clinical benefits from immunotherapy (Fig. 9). GSVA analysis also revealed that m6A-related scores were closely associated with pathways of EMT signal and stromal activation (Fig. 9), underlining the core role of stromal activation in predicting response to immunotherapy8,82,83. Subsequently, the predictive value of m6A-related score was appraised across various cohorts and tumor types. Our findings revealed that the m6A-related signature exhibited exceptional efficacy in other GC cohorts, as well as in the combined cohort of all TCGA tumors, thus underscoring the potential diagnostic value for other tumors beyond GC. Nonetheless, more research is needed to completely address this issue. Consequently, the m6A-related score could serve as a robust potential biomarker to predict the prognosis and clinical benefits of immunotherapy for GC patients. The integrated of m6A-related score with other biomarkers including TMB, neoantigen and tumor subtypes, could provide more effective immunotherapeutic strategies for individual patients with GC.
In summary, our study depicts a comprehensive landscape of TME infiltration characteristics mediated by m6A regulator-related genes in GC. First, we explored the characteristics of distinct m6A-related patterns in TME, expanding the significant role of m6A regulator-related genes in shaping TME diversity and the clinical or biological traits of GC. Second, we constructed an m6A-related scoring system for individual patients with GC. This scoring system provides an independent prognostic biomarker for prognosis and clinical benefits of immunotherapy and may facilitate the development of effective immunotherapy strategies and personalized treatment plans.
Materials and methods
Acquisition of GC datasets and preprocessing
In this study, we acquired six publicly available GC datasets (GSE15459, GSE34942, GSE57303, GSE26253, GSE66229, GSE54129, and TCGA-STAD) from the Gene Expression Omnibus (GEO) and the Cancer Genome Atlas (TCGA) databases to conduct further investigations (Additional file 1: Table S1)84,85,86,87,88,89. For microarray data obtained from the Affymetrix Human Genome U133 Plus platform, we retrieved the raw “CEL” files and utilized a powerful multiarray averaging approach, employing the affy and simpleaffy packages, to perform background adjustment and quantile normalization. In the case of microarray data from alternative platforms, the normalized matrix files were obtained directly. With regard to datasets in TCGA, RNA sequencing data, measured in transcripts per million (TPM) values, were procured from the Genomic Data Commons (GDC) (https://portal.gdc.cancer.gov/). Four GEO datasets (GSE15459, GSE34942, GSE57303, GSE66229), totaling n samples = 726, from Affymetrix were enrolled into one Meta-cohort. Batch effects in our Meta-cohort from non-biological technical biases were adjusted by the “ComBat” algorithm of the sva package. Of the 726 samples in the Meta-cohort, 29 samples did not pass quality control due to being identified as PCA outliers, resulting in the final Meta-cohort of 697 samples. These datasets were processed and then harmonized as illustrated in Additional file 1 and Additional file 2: Figure S1 and Table S1. Proteomics data of m6A regulators and m6A regulator-related genes were obtained from PDC00024 (https://pdc.cancer.gov/pdc/study/PDC000214) for subsequent research90 and the “limma” R package was used to compare the differences in proteomics expression between normal and GC samples. The somatic mutation data utilized in this study was obtained from the TCGA database. Analysis of the data was conducted using R software (version 4.3.0) and R Bioconductor packages.
Acquisition of GC samples
The GC samples utilized in this study were sourced from the GEO database and the TCGA database. A total of 726 GC samples were obtained from the GEO database, among which 28 samples were subsequently excluded due to the removal of PCA outliers and adjustment for batch effects. This screening process yielded 697 GC samples for further analysis. All GC samples retrieved from the TCGA database were included in our investigation. Notably, GC samples lacking survival data were excluded from survival analyses.
Identification of m6A regulator-related gene signature
Initially, we identified 28 m6A methylation regulators from previously published literature, including 8 writers (CBLL1, WTAP, VIRMA, RBM15B, RBM15, METTL3, METTL14, and METTL16), 2 erasers (ALKBH5, FTO), 15 readers (IGF2BP1, IGF2BP2, IGF2BP3, ELAVL1, FMR1, HNRNPA2B1, HNRNPC, LRPPRC, RBMX, ZC3H13, YTHDC1, YTHDC2, YTHDF1, YTHDF2, YTHDF3), and 3 anti-readers (EWSR1, G3BP1, LIN28A). Subsequently, Pearson’s correlation analysis was used to screen m6A regulator-related genes, resulting in the acquisition of 1025 genes satisfying the criteria of |Pearson R|> 0.5 and P < 0.001 in both the Meta-cohort and the TCGA-STAD cohort. We used the “ggalluvial” R package to map a Sankey diagram in order to visualize the relationships between various m6A regulators and their related genes91. These genes were selected as potential candidates for further investigation. Following this, Univariate Cox regression analysis was performed on the candidate genes, and 56 genes with P-values less than 0.05 in both the Meta-cohort and the TCGA-STAD cohort were designated as the m6A regulator-related gene signature.
Unsupervised clustering for 56 m6A regulator-related genes
We aimed to identify distinct m6A regulator-related patterns in GC by analyzing 56 m6A regulator-related genes obtained from four GEO datasets. These genes comprised LEPROT, CHAF1A, JAZF1, NR3C1, MID2, PKNOX2, PTPRM, QKI, FSTL1, PLXDC2, PJA2, BICC1, LAMA4, TIMP2, PDGFRB, EID1, AKT3, CYP2U1, RNF217, MMRN2, PRKD1, EDNRA, CC2D2A, DLC1, ARMCX2, SVEP1, RAB31, MCC, DOK6, ZNF385D, PHLDB2, PCDHB4, LAMA2, RBMS3, FAM229B, RECK, FBXL7, JAM3, CNRIP1, NR2F1-AS1, KCNT2, PDE2A, ZNF521, ROR2, MAGI2-AS3, MPDZ, LRCH2, CLIP4, FERMT2, FBN1, SLIT2, DDR2, ZFPM2, GEM, ADAMTS1, and RERG. We performed Univariate Cox regression analysis to evaluate the correlation between these genes and the prognosis of GC. Furthermore, we used unsupervised clustering analysis to classify the distinct m6A regulator-related patterns, based on the expression profiles of the 56 m6A regulator-related genes. The ConsensusClusterPlus package was utilized for this analysis34, and we conducted 1000 repetitions to ensure the stability of the classification.
Functional annotation and Gene set variation analysis (GSVA)
To gain insights into the functional roles of m6A regulator-related genes, we performed functional annotation using the clusterProfiler R package34, applying a cutoff value of FDR less than 0.05. We also conducted Gene Set Variation Analysis (GSVA) enrichment analysis to investigate the differences in biological processes among various m6A regulator-related patterns. The GSVA enrichment analysis was performed using the “GSVA” R package92, and the analysis was run using gene sets of “c2.cp.kegg.v6.2.symbols” downloaded from MSigDB. We deemed adjusted P-values less than 0.05 to be statistically significant.
Exploration of TME cell infiltration
To quantify the relative abundance of each cell infiltration in the TME of GC, we adopted the single-sample gene-set enrichment analysis (ssGSEA) algorithm. We obtained gene sets for various human immune cell subsets, including activated dendritic cells, activated B cells, helper T cells, activated CD4 T cells, activated CD8 T cells, regulatory T cells (Tregs), natural killer T cells, macrophages, MDSCs and others, from previous studies93,94. We employed these gene sets to calculate the enrichment scores, which represent the relative abundance of each infiltrating cell in TME of each sample.
Identification of differentially expression genes (DEGs) among distinct m6A regulator-related patterns
In our research, patients were categorized into three distinct m6A regulator-related patterns, according to the expression levels of 56 m6A regulator-related genes. Differential expression analysis of genes across these three distinct m6A regulator-related patterns was performed using the “limma” R package95. The significant threshold for identifying DEGs was set at |logFC|> 1 and an adjusted P-value < 0.05.
Histological examination of the STAD TCGA samples
We acquired pathology slides of 375 GC samples through the TCGA data portal (https://portal.gdc.cancer.gov/), comprising 202 GC samples from m6A-related cluster 1, 120 GC samples from m6A-related cluster 2 and 53 GC samples from m6A-related cluster 3. 50 poor-quality histological sections were excluded from the analysis, leaving a total of 325 sections for further investigation (Table S14). All of these tissue slides underwent blind re-examination of diagnostic formalin-fixed paraffin-embedded slides of the tissue by a pathologist. Furthermore, semi-quantitative pathological assessment was performed on tissue slides. Pathological analysis was focused on describing the tumor immune-inflammatory response through calculating the lymphocyte scores and the stroma scores, which were representative by the percentage of fibroblasts. A 5-grade system (0–4) was applied to determine the semi-quantitative scores. Tumor purity was evaluated through histopathology by visually estimating the percentage of tumor cells among all cells on the tissue slides. Representative images of each tumor microenvironment subtype were captured at 200 × magnification using Aperio ImageScope, a pathology side-viewing software, from slides of suitable quality.
Construction of m6A gene signature
In order to quantify the m6A regulator-related patterns of tumors individually, we developed a scoring system to assess the m6A regulator-related gene signature of GC patients, which we refer to as the m6A-related score. The steps to establish the m6A-related score are outlined below:
Initially, the m6A regulator-related genes (with P < 0.001 and cor > 0.5) were normalized across all GC samples. Subsequently, we conducted prognostic analysis for each gene, utilizing the Univariate Cox regression model. Genes that exhibited significant prognostic value in both Meta-cohort and TCGA-STAD cohort were chosen for further analysis. The m6A-related score of each sample was calculated using the following formula:
where Coefi represents the coefficients obtained from the Univariate Cox regression model, and βi indicates the TPM value of the corresponding m6A regulator-related genes.
Relation between m6A gene signature and biological processes
We investigated the relationship between the m6A gene signature and various biological processes. To do so, we analyzed gene sets associated with different biological processes, including immune checkpoint, CD8 T-effector signature, antigen processing machinery and presentation, DNA damage repair mechanisms such as mismatch repair and nucleotide excision repair, DNA replication, EMT markers (EMT1, EMT2 and EMT3), angiogenesis signature, pan-fibroblast TGF beta (TGF-β) response signatures, and WNT targets8,96,97. Then, a correlation analysis was performed to establish the relationships between the m6A gene signatures and these biological processes.
Gathering immune checkpoint blockade genomic and clinical benefit
To investigate the potential relationship between m6A-related scores and immunotherapy, we analyzed three immunotherapeutic cohorts. We finally selected three immunotherapeutic cohorts: advanced urothelial cancer treated with the anti-PD-L1 antibody (IMvigor210 cohort, obtained from http://research-pub.Gene.com/imvigor210corebiologies), metastatic melanoma treated with the anti-PD-1 antibody (GSE78220 cohort downloaded from GEO) and advanced malignancies treated with the anti-PD-1 and anti-CTLA4 (GSE91061 cohort downloaded from GEO) (Additional file 1: Table S1)98,99.
Statistical analysis
To compare the differences between multiple groups, we conducted one-way ANOVA and Kruskal–Wallis tests. For proteomics data, GC samples were categorized into the high expression group and the low expression group based on the protein expression disparity with normal samples and the significant expression differences between normal and GC samples were analyzed by the χ2 test. We used the pheatmap R package to create a heatmap visualizing the relationships between m6A regulators and m6A regulator-related genes91. To divide the patients into high and low m6A-related score groups, we used the “surv-cutpoint” function from the survminer R package, based on the maximal choice of the log-rank statistics100. The Kaplan–Meier method was used for prognosis and survival analysis. We conducted Univariate Cox regression analysis to calculate the hazard ratio (HR) and Multivariable Cox regression analysis to select independent prognostic factors. The data from the Multivariable Cox regression analysis were visualized using the forestplot R package101. To investigate the mutation landscape in patients with high and low m6A-related score groups in TCGA-STAD cohort, we used the waterfall function of the maftools R package91. The result exhibited that patients with high m6A-related scores had a significantly lower mutation burden than patients with low m6A-related scores. We set the significance level at P < 0.005 for all statistical analyses. All processes of data manipulation were conducted by the R software (version 4.3.0) and R Bioconductor packages.
Data availability
The datasets in this study can be found in the GEO databases (https://www.ncbi.nlm.nih.gov/geo/), TCGA databases (https://www.cancer.gov/ccg/research/genome-sequencing/tcga), Imvigor210 cohort (http://research-pub.Gene.com/imvigor210corebiologies) and PDC00024 (https://pdc.cancer.gov/pdc/study/PDC000214).
References
Sung, H. et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 Countries. CA Cancer J. Clin. 71(3), 209–249 (2021).
Jelski, W. & Mroczko, B. Molecular and circulating biomarkers of gastric cancer. Int. J. Mol. Sci. 23(14), 7588 (2022).
Conti, C. B. et al. Early gastric cancer: Update on prevention, diagnosis and treatment. Int. J. Environ. Res. Public Health 20(3), 2149 (2023).
Alsina, M., Arrazubi, V., Diez, M. & Tabernero, J. Current developments in gastric cancer: From molecular profiling to treatment strategy. Nat. Rev. Gastroenterol. Hepatol. 20(3), 155–170 (2023).
Hogner, A. & Moehler, M. Immunotherapy in gastric cancer. Curr. Oncol. 29(3), 1559–1574 (2022).
Chen, D. S. & Mellman, I. Elements of cancer immunity and the cancer-immune set point. Nature 541(7637), 321–330 (2017).
Ortiz-Munoz, G. et al. In situ tumour arrays reveal early environmental control of cancer immunity. Nature 618(7966), 827–833 (2023).
Mariathasan, S. et al. TGFbeta attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells. Nature 554(7693), 544–548 (2018).
Qi, Z. et al. Overcoming resistance to immune checkpoint therapy in PTEN-null prostate cancer by intermittent anti-PI3Kalpha/beta/delta treatment. Nat. Commun. 13(1), 182 (2022).
Xiao, Y. et al. Multi-omics profiling reveals distinct microenvironment characterization and suggests immune escape mechanisms of triple-negative breast cancer. Clin. Cancer Res. 25(16), 5002–5014 (2019).
Oerum, S., Meynier, V., Catala, M. & Tisne, C. A comprehensive review of m6A/m6Am RNA methyltransferase structures. Nucleic Acids Res. 49(13), 7239–7255 (2021).
Mi, S., Shi, Y., Dari, G. & Yu, Y. Function of m6A and its regulation of domesticated animals’ complex traits. J. Anim. Sci. https://doi.org/10.1093/jas/skac034 (2022).
Kumari, R. et al. mRNA modifications in cardiovascular biology and disease: With a focus on m6A modification. Cardiovasc. Res. 118(7), 1680–1692 (2022).
Chen, L. et al. The Efficacy of transarterial chemoembolization plus apatinib or sorafenib in the treatment of advanced hepatocellular carcinoma. J. Oncol. 2021, 8169012 (2021).
Jiang, X. et al. The role of m6A modification in the biological functions and diseases. Signal Transduct. Target Ther. 6(1), 74 (2021).
Knight, H. M., Demirbugen Oz, M. & PerezGrovas-Saltijeral, A. Dysregulation of RNA modification systems in clinical populations with neurocognitive disorders. Neural Regen. Res. 19(6), 1256–1261 (2024).
Somasekharan, S. P. et al. Regulation of AR mRNA translation in response to acute AR pathway inhibition. Nucleic Acids Res. 50(2), 1069–1091 (2022).
Geula, S. et al. Stem cells. m6A mRNA methylation facilitates resolution of naive pluripotency toward differentiation. Science 347(6225), 1002–1006 (2015).
Lin, Z. et al. Mettl3-/Mettl14-mediated mRNA N(6)-methyladenosine modulates murine spermatogenesis. Cell Res. 27(10), 1216–1230 (2017).
Cui, Q. et al. m(6)A RNA methylation regulates the self-renewal and tumorigenesis of glioblastoma stem cells. Cell Rep. 18(11), 2622–2634 (2017).
Wu, P. et al. N6-methyladenosine modification of circCUX1 confers radioresistance of hypopharyngeal squamous cell carcinoma through caspase1 pathway. Cell Death Dis. 12(4), 298 (2021).
Chen, X. Y., Zhang, J. & Zhu, J. S. The role of m(6)A RNA methylation in human cancer. Mol. Cancer 18(1), 103 (2019).
Tsuruta, N. et al. RNA N6-methyladenosine demethylase FTO regulates PD-L1 expression in colon cancer cells. Biochem. Biophys. Res. Commun. 530(1), 235–239 (2020).
Liu, D. et al. ONECUT2 facilitates hepatocellular carcinoma metastasis by transcriptionally upregulating FGF2 and ACLY. Cell Death Dis. 12(12), 1113 (2021).
Su, R. et al. Targeting FTO suppresses cancer stem cell maintenance and immune evasion. Cancer Cell 38(1), 79-96 e11 (2020).
Yang, S. et al. m(6)A mRNA demethylase FTO regulates melanoma tumorigenicity and response to anti-PD-1 blockade. Nat. Commun. 10(1), 2782 (2019).
Zhang, W. et al. m6A regulator-mediated tumour infiltration and methylation modification in cervical cancer microenvironment. Front. Immunol. 13, 888650 (2022).
Zhang, F. et al. Crosstalk among m(6)A RNA methylation, hypoxia and metabolic reprogramming in TME: From immunosuppressive microenvironment to clinical application. J. Hematol. Oncol. 15(1), 84 (2022).
Gao, C. et al. Development and validation of the potential biomarkers based on m6A-related lncRNAs for the predictions of overall survival in the lung adenocarcinoma and differential analysis with cuproptosis. BMC Bioinform. 23(1), 327 (2022).
Lv, W. et al. Identification and validation of m6A-related lncRNA signature as potential predictive biomarkers in breast cancer. Front. Oncol. 11, 745719 (2021).
Li, Z., Li, Y., Zhong, W. & Huang, P. m6A-related lncRNA to develop prognostic signature and predict the immune landscape in bladder cancer. J. Oncol. 2021, 7488188 (2021).
Yu, J. et al. Identification of an m6A-related lncRNA signature for predicting the prognosis in patients with kidney renal clear cell carcinoma. Front. Oncol. 11, 663263 (2021).
Xu, F., Huang, X., Li, Y., Chen, Y. & Lin, L. m(6)A-related lncRNAs are potential biomarkers for predicting prognoses and immune responses in patients with LUAD. Mol. Ther. Nucleic Acids 24, 780–791 (2021).
Wang, Y. et al. Comprehensive analysis of tumor immune microenvironment and prognosis of m6A-related lncRNAs in gastric cancer. BMC Cancer 22(1), 316 (2022).
Marabelle, A. et al. Efficacy of pembrolizumab in patients with noncolorectal high microsatellite instability/mismatch repair-deficient cancer: Results from the phase II KEYNOTE-158 study. J. Clin. Oncol. 38(1), 1–10 (2020).
Hua, H. et al. Genomic and transcriptomic analysis of MSI-H colorectal cancer patients with targetable alterations identifies clinical implications for immunotherapy. Front. Immunol. 13, 974793 (2022).
Taki, M. et al. Tumor immune microenvironment during epithelial-mesenchymal transition. Clin. Cancer Res. 27(17), 4669–4679 (2021).
Ge, S. et al. A proteomic landscape of diffuse-type gastric cancer. Nat. Commun. 9(1), 1012 (2018).
Oh, S. C. et al. Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype. Nat. Commun. 9(1), 1777 (2018).
Song, J., Wei, R., Huo, S., Gao, J. & Liu, X. Metastasis related epithelial-mesenchymal transition signature predicts prognosis and response to immunotherapy in gastric cancer. Front. Immunol. 13, 920512 (2022).
Zhu, J., Xiao, J., Wang, M. & Hu, D. Pan-cancer molecular characterization of m(6)A regulators and immunogenomic perspective on the tumor microenvironment. Front. Oncol. 10, 618374 (2020).
Lv, J., Ren, J., Zheng, J., Zhang, F. & Han, M. Prognosis of tumor microenvironment in luminal B-type breast cancer. Dis. Markers 2022, 5621441 (2022).
Liu, Y. J. et al. USP51/ZEB1/ACTA2 axis promotes mesenchymal phenotype in gastric cancer and is associated with low cohesion characteristics. Pharmacol. Res. 188, 106644 (2023).
Gerard, C. L. et al. Turning tumors from cold to inflamed to improve immunotherapy response. Cancer Treat. Rev. 101, 102227 (2021).
Palomares, O. et al. Regulatory T cells and immune regulation of allergic diseases: roles of IL-10 and TGF-beta. Genes Immun. 15(8), 511–520 (2014).
Kanamori, M., Nakatsukasa, H., Okada, M., Lu, Q. & Yoshimura, A. Induced regulatory T cells: Their development, stability, and applications. Trends Immunol. 37(11), 803–811 (2016).
Tiwari, A. et al. Towards a consensus definition of immune exclusion in cancer. Front. Immunol. 14, 1084887 (2023).
Mashukov, A. et al. Histological differentiation impacts the tumor immune microenvironment in gastric carcinoma: Relation to the immune cycle. World J. Gastroenterol. 27(31), 5259–5271 (2021).
Liu, Y. et al. Comparative molecular analysis of gastrointestinal adenocarcinomas. Cancer Cell 33(4), 721–35 e8 (2018).
Chia, N. Y. & Tan, P. Molecular classification of gastric cancer. Ann. Oncol. 27(5), 763–769 (2016).
Wang, J. et al. EBV miRNAs BART11 and BART17-3p promote immune escape through the enhancer-mediated transcription of PD-L1. Nat. Commun. 13(1), 866 (2022).
Puliga, E., Corso, S., Pietrantonio, F. & Giordano, S. Microsatellite instability in gastric cancer: Between lights and shadows. Cancer Treat. Rev. 95, 102175 (2021).
Sahin, U. & Tureci, O. Personalized vaccines for cancer immunotherapy. Science 359(6382), 1355–1360 (2018).
Palmeri, M. et al. Real-world application of tumor mutational burden-high (TMB-high) and microsatellite instability (MSI) confirms their utility as immunotherapy biomarkers. ESMO Open 7(1), 100336 (2022).
Zhang, N., Zuo, Y., Peng, Y. & Zuo, L. Function of N6-methyladenosine modification in tumors. J. Oncol. 2021, 6461552 (2021).
Liu, X. et al. The m6A methyltransferase METTL14 inhibits the proliferation, migration, and invasion of gastric cancer by regulating the PI3K/AKT/mTOR signaling pathway. J. Clin. Lab. Anal. 35(3), e23655 (2021).
Hu, Y. et al. Demethylase ALKBH5 suppresses invasion of gastric cancer via PKMYT1 m6A modification. Mol. Cancer 21(1), 34 (2022).
Chen, B. et al. M6A-related lncRNAs predict clinical outcome and regulate the tumor immune microenvironment in hepatocellular carcinoma. BMC Cancer 22(1), 867 (2022).
Guo, T. et al. Prognostic signature of hepatocellular carcinoma and analysis of immune infiltration based on m6A-related lncRNAs. Front. Oncol. 11, 691372 (2021).
Xie, P. et al. Construction of m6A-related lncRNA prognostic signature model and immunomodulatory effect in glioblastoma multiforme. Front. Oncol. 12, 920926 (2022).
Han, X., Lu, H., Tang, X., Zhao, Y. & Liu, H. Immunogenomic characterization in gastric cancer identifies microenvironmental and immunotherapeutically relevant gene signatures. Immun. Inflamm. Dis. 10(1), 43–59 (2022).
Turley, S. J., Cremasco, V. & Astarita, J. L. Immunological hallmarks of stromal cells in the tumour microenvironment. Nat. Rev. Immunol. 15(11), 669–682 (2015).
Hornburg, M. et al. Single-cell dissection of cellular components and interactions shaping the tumor immune phenotypes in ovarian cancer. Cancer Cell 39(7), 928–44 e6 (2021).
Mittal, S. K. & Roche, P. A. Suppression of antigen presentation by IL-10. Curr. Opin. Immunol. 34, 22–27 (2015).
Yang, L., Pang, Y. & Moses, H. L. TGF-beta and immune cells: An important regulatory axis in the tumor microenvironment and progression. Trends Immunol. 31(6), 220–227 (2010).
Liang, X. et al. Abrine, an IDO1 inhibitor, suppresses the immune escape and enhances the immunotherapy of anti-PD-1 antibody in hepatocellular carcinoma. Front. Immunol. 14, 1185985 (2023).
Jiang, Y. & Zhan, H. Communication between EMT and PD-L1 signaling: New insights into tumor immune evasion. Cancer Lett. 468, 72–81 (2020).
Kato, Y. et al. Lenvatinib plus anti-PD-1 antibody combination treatment activates CD8+ T cells through reduction of tumor-associated macrophage and activation of the interferon pathway. PLoS One 14(2), e0212513 (2019).
Tibbs, E., Kandy, R. R. K., Jiao, D., Wu, L. & Cao, X. Murine regulatory T cells utilize granzyme B to promote tumor metastasis. Cancer Immunol. Immunother. 72(9), 2927–2937 (2023).
Wang, D. R., Wu, X. L. & Sun, Y. L. Therapeutic targets and biomarkers of tumor immunotherapy: Response versus non-response. Signal Transduct. Target Ther. 7(1), 331 (2022).
Wang, Y., Johnson, K. C. C., Gatti-Mays, M. E. & Li, Z. Emerging strategies in targeting tumor-resident myeloid cells for cancer immunotherapy. J. Hematol. Oncol. 15(1), 118 (2022).
Le, D. T. et al. Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade. Science 357(6349), 409–413 (2017).
Guan, J. et al. MLH1 Deficiency-Triggered DNA Hyperexcision by Exonuclease 1 Activates the cGAS-STING Pathway. Cancer Cell 39(1), 109–21 e5 (2021).
Samstein, R. M. et al. Tumor mutational load predicts survival after immunotherapy across multiple cancer types. Nat. Genet. 51(2), 202–206 (2019).
Lee, K. W. et al. Association of tumor mutational burden with efficacy of pembrolizumab+/-chemotherapy as first-line therapy for gastric cancer in the phase III KEYNOTE-062 study. Clin. Cancer Res. 28(16), 3489–3498 (2022).
Lizardo, D. Y. et al. Immunotherapy efficacy on mismatch repair-deficient colorectal cancer: From bench to bedside. Biochim. Biophys. Acta Rev. Cancer 1874(2), 188447 (2020).
Gu, Y., Zhang, Z. & Ten Dijke, P. Harnessing epithelial-mesenchymal plasticity to boost cancer immunotherapy. Cell Mol. Immunol. 20(4), 318–340 (2023).
Guo, S. et al. Genomic instability drives tumorigenesis and metastasis and its implications for cancer therapy. Biomed. Pharmacother. 157, 114036 (2023).
Panda, A. et al. Immune activation and benefit from avelumab in EBV-positive gastric cancer. J. Natl. Cancer Inst. 110(3), 316–320 (2018).
Kwak, Y., Seo, A. N., Lee, H. E. & Lee, H. S. Tumor immune response and immunotherapy in gastric cancer. J. Pathol. Transl. Med. 54(1), 20–33 (2020).
Xu, Y. Y., Shen, A. & Zeng, Z. L. A potential EBV-related classifier is associated with the efficacy of immunotherapy in gastric cancer. Transl. Cancer Res. 11(7), 2084–2096 (2022).
Tauriello, D. V. F. et al. TGFbeta drives immune evasion in genetically reconstituted colon cancer metastasis. Nature 554(7693), 538–543 (2018).
Egan, H. et al. Targeting stromal cell sialylation reverses T cell-mediated immunosuppression in the tumor microenvironment. Cell Rep. 42(5), 112475 (2023).
Hu, X. et al. Molecular classification reveals the diverse genetic and prognostic features of gastric cancer: A multi-omics consensus ensemble clustering. Biomed. Pharmacother. 144, 112222 (2021).
Pan, S. et al. Integrative analysis-based identification and validation of a prognostic immune cell infiltration-based model for patients with advanced gastric cancer. Int. Immunopharmacol. 101(Pt B), 108258 (2021).
Zeng, D. et al. Tumor microenvironment evaluation promotes precise checkpoint immunotherapy of advanced gastric cancer. J. Immunother. Cancer 9(8), e002467 (2021).
Huo, J., Fan, X., Sun, W. & Sun, P. Immunotherapy efficacy-related risk classifier differentiate prognostic characteristics of gastric cancer-a large-scale retrospective study. J. Immunother. 46(8), 323–332 (2023).
Pang, T. et al. Construction of a novel immune-related prognostic-predicting model of gastric cancer. Gene 852, 147032 (2023).
Szeto, C. Y. et al. Integrated mRNA and microRNA transcriptome sequencing characterizes sequence variants and mRNA-microRNA regulatory network in nasopharyngeal carcinoma model systems. FEBS Open Bio 4, 128–140 (2014).
Mun, D. G. et al. Proteogenomic characterization of human early-onset gastric cancer. Cancer Cell 35(1), 111–24 e10 (2019).
Zhang, C. et al. Implications of m6A-associated snRNAs in the prognosis and immunotherapeutic responses of hepatocellular carcinoma. Front. Immunol. 13, 1001506 (2022).
Meijing, Z., Tianhang, L. & Biao, Y. N6-methyladenosine modification patterns and tumor microenvironment immune characteristics associated with clinical prognosis analysis in stomach adenocarcinoma. Front. Cell. Dev. Biol. 10, 913307 (2022).
Wang, A., Nie, S., Lv, Z., Wen, J. & Yuan, Y. Infiltration of immunoinflammatory cells and related chemokine/interleukin expression in different gastric immune microenvironments. J. Immunol. Res. 2020, 2450569 (2020).
Charoentong, P. et al. Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep. 18(1), 248–262 (2017).
Wang, H., Meng, Q. & Ma, B. Characterization of the prognostic m6A-related lncRNA signature in gastric cancer. Front. Oncol. 11, 630260 (2021).
Qin, S. et al. The functional roles, cross-talk and clinical implications of m6A modification and circRNA in hepatocellular carcinoma. Int. J. Biol. Sci. 17(12), 3059–3079 (2021).
Zhang, B. et al. m(6)A regulator-mediated methylation modification patterns and tumor microenvironment infiltration characterization in gastric cancer. Mol. Cancer 19(1), 53 (2020).
Li, J. et al. A novel hypoxia- and lactate metabolism-related signature to predict prognosis and immunotherapy responses for breast cancer by integrating machine learning and bioinformatic analyses. Front. Immunol. 13, 998140 (2022).
Chen, X. et al. A T-cell-related signature for prognostic stratification and immunotherapy response in hepatocellular carcinoma based on transcriptomics and single-cell sequencing. BMC Bioinform. 24(1), 216 (2023).
Lei, L., Li, N., Yuan, P. & Liu, D. A new risk model based on a 11-m(6)A-related lncRNA signature for predicting prognosis and monitoring immunotherapy for gastric cancer. BMC Cancer 22(1), 365 (2022).
Zhao, Y., Yan, X., Wang, Y., Zhou, J. & Yu, Y. N6-methyladenosine regulators promote malignant progression of gastric adenocarcinoma. Front. Oncol. 11, 726018 (2021).
Acknowledgements
We thank the Gene-Expression Omnibus (GEO), The Cancer Genome Atlas (TCGA) Database and IMvigor210 for sharing a large amount of data.
Funding
This study was supported by the National Natural Science Foundation of China (32270153), Key Laboratory of Guangdong Higher Education Institutes (2021KSYS009), the Science and Technology Program of Guangzhou (202102010128, 2024A04J8217 and 2024A03J0077), Guangzhou Key Laboratory of Biological Targeting Diagnosis and Therapy (202201020379), and Open Project of State Key Laboratory of Respiratory Disease (SKLRD-OP-202323).
Author information
Authors and Affiliations
Contributions
Z.L., X.Z., W.Y., G.Y., M.L. and B.P. contributed to conception and design of the study; Z.L., Y.L. and M.L. collected the data; Z.L., B.P., Y.L. and Y.X. organized the data; M.L., Z.L., B.P. and Y.Q. performed the statistical analysis; B.P., W.Y. and Z.L. wrote the main manuscript text; Y.L., G.Y. and M.L. prepared figures. All authors reviewed the manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Peng, B., Lin, Y., Yi, G. et al. Comprehensive landscape of m6A regulator-related gene patterns and tumor microenvironment infiltration characterization in gastric cancer. Sci Rep 14, 16404 (2024). https://doi.org/10.1038/s41598-024-66744-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-66744-0
- Springer Nature Limited