Telomere biology plays a crucial and intricate role in the onset and advancement of cancer [29]. Dysfunctional telomeres can impede cancer development by inducing replicative senescence or apoptotic pathways. However, they can also facilitate tumor initiation and progression by inducing oncogenic chromosomal rearrangements. Recent studies have shown that telomere dysfunction increases the sensitivity of cancer is related to T cell immune dysfunction [30]. It is noteworthy that telomerase reverse transcriptase (TERT), a subunit of telomerase, plays a significant role in angiogenesis, invasion, epithelial-mesenchymal transformation, inflammation, immunosuppression, and other critical gene expression profiles. These TERT-mediated activities can significantly affect the dynamic equilibrium of the TME [31, 32]. Recent investigations have also uncovered the involvement of TERT in the lymphatic and vascular metastasis of GC, which leads to a poor prognosis [33, 34]. GC is a significant contributor to mortality associated with tumors. It is a highly heterogeneous, with significant variations in ethnic, regional, and pathological features. These differences result in distinct treatment options and outcomes for patients with gastric cancer in Eastern and Western countries. However, the overall prognosis of gastric cancer remains suboptimal [35, 36]. To date, the TNM stage is the primary predictor of prognosis in GC patients. However, individuals with the same TNM stage often have different clinical outcomes. This suggests that the current TNM staging guidelines are not adequate for disease risk stratification. Several biomarkers, including PD-L1 combined positive score (CPS), microsatellite instability-high (MSI-H), and TMB, are widely regarded as promising for predicting the immune response in GC. These biomarkers have some limitations, including predictive instability, intra-tumor heterogeneity, and limited predictive benefit for the population [37, 38]. Therefore, the advancement telomere-related new biomarkers for GC that can accurately forecast prognosis and response to immunotherapy would be highly significant. Recently, several studies have constructed similar models, including gene models related to DNA damage repair, senescence, pyroptosis, neutrophil extracellular traps, and immunity [39–45]. These models can predict prognosis and immunotherapy response. Our model has an above-average capacity to predict overall survival in GC patients compared to other models. And all these models, including ours, can distinguish to some extent between GC patients that are sensitive and resistant to immunotherapy. Therefore, our model is a valuable supplement to current biomarkers for forecasting GC patient prognosis and identifying potential responders to ICI therapy.
Our study utilized publicly available databases to identify 100 telomere-related hub genes using differential expression analysis and WGCNA. We then constructed a 9-TRG risk panel based on uni-Cox, LASSO, and multi-Cox regression analyses to predict the prognosis of GC patients and built three molecular subtypes (cluster A, B and C) based on a consensus clustering algorithm. The cluster B and C had higher risk scores than cluster A. We also validated that the risk panel showed good performance in the GEO validation cohort. Finally, we established a nomogram incorporating the TRGscore and clinicopathological factors to predict OS rates of GC patients. A high TRG risk score was correlated with poor survival and pro-tumor TIME of GC. By constructing a TRG-related risk panel, we not only focused on its predictive significance but also better analyzed the immune infiltration and treatment sensitivity differences among patients. Our TRGscore can guide personalized chemotherapy and immunotherapy for patients with GC, thereby improving their prognosis.
In our study, the TRG risk panel was composed of 9 genes (CALML6, CFAP58, CABP2, DST, ELOVL2, HIST1H3G, MYF6, PDE1B and TOP3B) that have been reported in cancer development and progression to some extent. CALML6, (Calmodulin-Like Protein 6) is an oncogene in GC and is involved in mitochondrial reprogramming. Down-regulation of its level has been associated with reduced survival in rectal cancer patients [46, 47]. Aberrant expression of CFAP58 (Cilia- And Flagella-Associated Protein 58) is associated with poor outcome in bladder cancer, endometrial cancer, lung cancer and triple-negative breast cancer [48–51]. CABP2 (Calcium Binding Protein 2) expression is upregulated in the MUT-high subtype of diffuse large B-cell lymphoma and correlates with reduced T-cell immune infiltration in the TME [52]. Aberrant expression of the DST (Dystonin) has been observed in various cancers, including melanoma, lung cancer, metastatic prostate cancer and invasive breast cancer [53–55]. DST may affect the development and progression of breast cancer by influencing the TIME [56]. The downregulation of DST could be indicative of an invasive phenotype and metastasis of cancers [57]. ELOVL2 (Elongation of very long chain Fatty acid elongase 2) is involved in the regulation of autophagy and the activity of the mTOR signaling pathway, and it’s upregulated in kidney cancer and downregulated in breast cancer are associated with tumor growth and poor prognosis [58, 59]. ELOVL2 is involved in lipid metabolic reprogramming and dysregulated immune status, which correlates with adverse outcomes in retroperitoneal liposarcoma [60]. HIST1H3G (Histone Cluster 1 H3 Family Member G) is expressed abnormally in several types of cancer. High expression of HIST1H3G is linked to poor prognosis in glioma and laryngeal squamous cell carcinoma, and to chemosensitivity in ovarian cancer [61–64]. MYF6 (Myogenic Factor 6, a telomerase transcription factor) hypermethylation has been reported to be associated with lung cancer [65]. PDE1B (Phosphodiesterase 1B) has the highest weight coefficient in our panel. Interestingly, studies indicated that PDE1B may have a regulatory function in the differentiation of various immune cell types by reducing intracellular levels of both cAMP and cGMP [66]. Moreover, previous studies found that through cytokines like GM-CSF, the upregulation of PDE1B might facilitate the differentiation of macrophages into M1 subtype macrophages, thus bolstering the anti-tumor immune response and potentially enhancing patient prognosis [67, 68]. TOP3B (DNA Topoisomerase III Beta) is involved in DNA repair/damage response. Based on data from a randomized phase III clinical trial, SAMIT, a machine learning study has identified TOP3B as a predictive marker for paclitaxel sensitivity in GC [69]. We also found that certain genes share similar functions to some extent [70]. For instance, ELOVL2, CALML6, CABP2, and HIST1H3G are involved in material metabolism and cell signalling, while DST and PDE1B are involved in regulating the TIME. Several genes (CFAP58, CABP2, DST, ELOVL2, HIST1H3G, MYF6, and PDE1B) were found to be associated with GC for the first time. Therefore, more research is warranted to uncover the mechanism of action of these genes in GC.
The TME is a critical determinant of tumor development, growth, and migration, and response to therapy in GC, with immune cell infiltration playing a pivotal role [71]. In the present study, patients with higher risk scores had a higher TME score. Our findings indicated that substantial disparities in immune landscape across distinct groups of GC. Particularly, the high-risk group exhibited an increased infiltration of Tregs, M2 macrophages, resting mast cells and resting NK cells. These findings suggest an exhausted immune phenotype of the TME and a poor prognosis for patients with GC. Consistent with our findings, previous research has confirmed that the infiltration of M2 macrophages in the TME is associated with a poor prognosis for GC [72, 73]. Another study showed that tumor-infiltrating mast cells foster immune suppression through TNF-α-PD-L1 pathway and stimulate Treg cells through an IL-33 and IL-2 axis to promote GC progression [74, 75]. Further functional enrichment analyses in different risk groups and clusters showed involvement of signaling pathways associated with cancer cell progression and immune suppression such as TGF-β signaling pathway, ECM receptor pathways in the groups with high riskscores. Previous studies showed that ECM is an important component of the TME and can regulate cancer behaviors. ECM remodeling is crucial in the regulation of GC progression [76, 77]. The TGF signaling pathway can suppress tumors, including cell cycle arrest and apoptosis [78]. TGF-β in the gastric tumor microenvironment is reported to promote the differentiation and expansion of both Tregs and M2 macrophages[79, 80]. These results suggested that our gene panel can distinguish between immune infiltration characteristics of high- and low-risk groups and that the poor prognosis of GC patients in the high-risk group might be associated with the immunodepression microenvironment.
Immunotherapy is a crucial element in treating GC. In this study, we examined the TIME to elucidate the relationship between TIME and TRGs in GC. Given the influence of immune checkpoints in the immunotherapy, we evaluated the differences in immune checkpoints between low and high-risk patients [81]. We found that most of the immune checkpoint genes were down-regulated in the low-risk patients. Furthermore, research has shown that a high TMB is correlated with a positive response to immune and targeted therapies in cancer patients, and is often indicative of favorable survival rates [82, 83]. Consistent with the previous results, our study found TMB were more significant in the low-risk group and patients with high TMB scores showed better the prognosis. Our research also showed that the low-risk group had higher mutation rates for TTN, TP53 and MUC16. Several studies have found that mutations in MUC16 and TP53 are linked to a better prognosis and higher TMB in GC, while TTN mutations are associated with a better response to ICIs therapy in solid tumors [84, 85]. Although TP53 is one of the most mutated gene, its prognostic significance in GC is still controversial [18, 86]. Additionally, significant differences were observed in our study between the various risk groups for several frequently applied immunotherapy biomarkers. The TIDE score is a recently developed method for predicting the effectiveness of anti-PD1 and anti-CTLA4 therapy. It is more accurate than TMB or PD-L1 expression [87]. The TIDE contains two potential tumor immunologic evasion mechanisms: T-cell exclusion and T-cell dysfunction. A higher TIDE score indicates a poorer tumor response to ICI therapy and a worse prognosis [88]. Our study showed that patients with high-risk scores had significantly higher TIDE, T-cell dysfunction, and T-cell exclusion scores than those with low-risk scores. This suggests that patients with low-risk scores may be more responsive to immunotherapies, which was confirmed in the immunotherapy cohort. This suggested that TRGscore is a valid biomarker for predicting response to immunotherapy. The low TRGscore group may have a better response to ICI therapy. Chemotherapy sensitivity analysis also proved the distinct sensitivity of each subtype to certain drugs.
The present study still has some highlights, even though other comparable publications have been published employing built characteristics to predict the prognosis of GC patients. First, we examined the patient’s prognosis for GC for the first-time using telomere-related mRNAs, and our models have a high level of predictive power. Second, we successfully confirmed the signature using in the public validation cohort. Moreover, we examined clinical tissue samples to confirm the mRNA levels of the genes responsible for the composition of the panel. It is admitted that our study has certain limitations. This study relies mainly on public databases without real-world cohort for validation. To further assess this signature, future large multicenter randomized controlled investigations are required. Additionally, further in vivo and vitro research is necessary to investigate the expression, prognostic predictive relevance, and particular mechanisms of these genes in GC.
In conclusion, we have developed a TRG-panel based on a recognized and effective strategy for predicting GC prognosis and immunotherapy efficacy. In addition, by identifying the complex relationship between TRGs and oncogenic pathways, we provided insight into TRGs' role in tumorigenesis and TME reshaping. In combination with immune infiltration, immune checkpoint factors, and other biomarkers, we demonstrated that TRGscore effectively distinguishes responders and non-responders, enabling ICI therapy to be more precisely stratified by benefit. Therefore, this work might facilitate the identification of prognostic biomarkers and provide guidance for developing personalized immunotherapy.