Construction of prognostic markers for gastric cancer and comprehensive analysis of pyroptosis-related long non-coding RNAs

Abstract

BACKGROUND

China's most frequent malignancy is gastric cancer (GC), which has a very poor survival rate, and the survival rate for patients with advanced GC is dismal. Pyroptosis has been connected to the genesis and development of cancer. The function of pyroptosis-related long non-coding RNAs (PRLs) in GC, on the other hand, remains uncertain.

AIM

To explore the construction and comprehensive analysis of the prognostic characteristics of long non-coding RNA (lncRNA) related to pyroptosis in GC patients.

METHODS

The TCGA database provided us with 352 stomach adenocarcinoma samples, and we obtained 28 pyroptotic genes from the Reactome database. We examined the correlation between lncRNAs and pyroptosis using the Pearson correlation coefficient. Prognosis-related PRLs were identified through univariate Cox analysis. A predictive signature was constructed using stepwise Cox regression analysis, and its reliability and independence were assessed. To facilitate clinical application, a nomogram was created based on this signature. we analyzed differences in immune cell infiltration, immune function, and checkpoints between the high-risk group (HRG) and low-risk group (LRG).

RESULTS

Five hundred and twenty-three PRLs were screened from all lncRNAs (absolute correlation coefficient > 0.4, P < 0.05). Nine PRLs were included in the risk prediction signature that was created through stepwise Cox regression analysis. We determined the risk score for GC patients and employed the median value as the dividing line between HRG and LRG. The ability of the risk signature to predict the overall survival (OS) of GC is demonstrated by the Kaplan-Meier analysis, risk curve, receiver operating characteristic curve, and decision curve analysis curve. The risk signature was shown to be an independent prognostic factor for OS in both univariate and multivariate Cox regression analyses. HRG showed a more efficient local immune response or modulation compared to LRG, as indicated by the predicted signal pathway analysis and examination of immune cell infiltration, function, and checkpoints (P < 0.05).

CONCLUSION

In general, we have created a brand-new prognostic signature using PRLs, which may provide ideas for immunotherapy in patients with GC.

Key Words: Gastric cancer; Pyroptosis; Prognosis; Immune checkpoint; Long non-coding RNA; Immune cell infiltrating

Core Tip: In this study, China's most common malignancy is gastric cancer (GC), which has a poor survival rate. Pyroptosis is connected to cancer development. The function of pyroptosis-related long non-coding RNAs (PRLs) in GC is uncertain. In our study, we took advantage of the 352 gastric adenocarcinoma samples included in the TCGA database and obtained data on 28 cases related to the pyroptotic genes from the Reactome database. We used the Pearson correlation coefficient method to screen out the long non-coding RNAs associated with pyroptosis. Using univariate Cox analysis, we identified the PRL associated with prognosis. By using PRLs, we created a completely new prognostic marker, which may provide ideas for immunotherapy in GC patients.

INTRODUCTION

In 2020, there were more than 1000000 new instances of gastric cancer (GC) and approximately 770000 deaths from GC, based on the Global Cancer Statistics[1]. But GC is hard to detect early, as usually asymptomatic in the early stage. As a result, about 70% of GC patients receive an advanced diagnosis[2]. Approximately 95% of GC are adenocarcinomas[3]. In recent years, Despite the rapid development of radiotherapy, chemotherapy, and immunotherapy, surgical resection of the malignant lesion is the mainstay therapy for GC[4,5]. With such a multidisciplinary treatment approach, the survival rate for advanced GC is still less than 10%[6]. Consequently, there is still a pressing need to investigate new biomarkers for GC patient prognosis prediction.

Cell death mainly includes programmed and non-programmed death, and pyroptosis is pro-inflammatory programmed cell death. The Greek words "pyro" and "ptosis," which denote fever and recession, respectively, are found in the word "pyroptosis"[7]. The ballooning of the cell and multiple processes on the surface of the cell membrane before rupture set pyroptosis apart from apoptosis. The canonical inflammasome pathway and the noncanonical inflammasome pathway are the two primary pathway branches that usually makeup pyroptosis signaling[8]. When cells are menaced by a variety of pathogens, such as bacteria, viruses, fungi, and protozoa, cytosolic pattern recognition receptors identify pathogen-associated molecular patterns and danger-associated molecular patterns. Consequently, it causes proteins farther down the chain to organize into inflammatory bodies and activate caspase-1 or caspase-11/4/5. Gasdermin D (GSDMD) is cleaved by caspase, and the N-terminal domain of GSDMD oligomerizes to generate membrane holes, releasing IL-1 and IL-18 and causing cell pyroptotic death[9,10]. In the noncanonical inflammasome pathway, lipopolysaccharide actives caspase-4/-5/-11 to induce cell pyroptotic death[11].

Pyroptosis is crucial for the onset and progression of many illnesses, including cancers, metabolic disorders, infectious diseases, nerve-related diseases, and atherosclerosis[12]. According to earlier research, pyroptosis, and cancer formation and progression are tightly related. For instance, Rogers et al[12] discovered that in melanoma, knocking down Gasdermin E in the B16-Ova cell line greatly decreases the release of cytochrome c and the activation of caspase-3 and speeds up tumor development[12]. By triggering the pyroptosis pathway at the gene and protein levels, 4-hydroxybenzoic acid was able to stop the growth of a non-small cell lung cancer cell line, according to research by Sannino et al[13]. In colon cancer, silencing Gasdermin C was revealed to suppress colon cancer cell line proliferation and metastasis, while overexpression of Gasdermin C can enhance cell proliferation[14,15]. The high expression of cytokine IL-1β leads to the recruitment of Myeloid-derived suppressor cells in the stomach, which increases the risk of spontaneous gastritis and GC[16]. In addition, a lot of studies on leukemia, uterine cancer, and liver cancer have shown that pyroptosis is closely related to the incidence and progression of carcinoma[17-21].

A transcript with a length of more than 200 nucleotides known as a long non-coding RNA (lncRNA) cannot produce a protein yet has a distinct expression in numerous tissues, including malignancies[22,23]. More than 28000 LncRNA are thought to be encoded in the human genome, and there is still much to be uncovered[24]. LncRNA not only participates in chromatin interaction, transcriptional regulation, and RNA processing but also regulates the stability or translation process of mRNA and affects cell signal cascades[25]. In recent years, more and more research has revealed that lncRNA can regulate the biological functions of tumor proliferation, apoptosis, invasion, and metastasis in multiple signal pathways[26,27]. For instance, PCGEM1 is a lncRNA that is strongly androgen-regulated and tissue-specific to the prostate[28]. In prostate cancer, overexpression of PCGEM1 may increase cellular resistance to chemotherapy by promoting cell growth and clone creation[29,30]. The capacity of human breast cancer cells to metastasize to the lungs is improved by lncRNA MALAT1 knockout while restoring MALAT1 expression weakens the metastatic capability[31]. GC has a greater degree of HOTAIR expression than the nearby normal tissues. Patients with diffuse GC who express more HOTAIR would have increased venous invasion, local lymph node metastases, and a decreased chance of survival. Overexpression of HOTAIR could promote liver metastasis in GC[32]. In addition to the above, publications on the function and mechanism of lncRNA in different malignancies are growing. The pattern of lncRNA expression and its role in GC, however, have not been thoroughly examined.

Although some researchers have shown that lncRNA has a special function in the pyroptosis of GC, little research has been done on this topic, and the precise mechanism is not yet fully understood[33]. Using the pyroptosis-related lncRNA (PRL) collected from The Cancer Genome Atlas (TCGA) database, we developed a signature to predict the prognosis of GC patients. The findings might serve as a guide for doctors treating GC patients and evaluating their prognosis.

MATERIALS AND METHODS

Acquisition and processing of data

We concurrently downloaded the clinical data corresponding to the gastric adenocarcinoma expression data from the TCGA database (GDC, cancer.gov). This research comprised 352 patients in total for a follow-up study.

Identification of prognosis-related PRLs

According to the gene annotation in the expression data of 352 GC cases, 13824 LncRNA were obtained. Twenty-eight genes associated with pyroptosis were collected from the Reactome Pathway Database (Home - Reactome Pathway Database). The Pearson correlation coefficient assessed correlations between pyroptosis-related genes and lncRNA. PRLs were characterized as lncRNAs with an absolute correlation coefficient > 0.4 and P < 0.05, and a total of 532 PRLs were chosen. To identify PRLs that were associated with prognosis, we used univariate Cox regression analysis. Twenty-eight LncRNA with prognostic values related to pyroptosis were further screened (P < 0.05).

Prognostic signature and nomogram construction based on PRLs

For the above 28 PRLs, based on stepwise Cox regression analysis, a further prediction signature was established, including nine PRLs. Risk scores were calculated for each patient based on the normalized expression levels of lncRNAs and their related regression coefficients. This is the method used to determine the risk score: Risk score = sum (lncRNA expression level corresponding coefficient). We created and displayed the co-expression network between nine PRLs and associated pyroptotic genes using the Cytoscape program. In parallel, we developed a Nomogram for clinical use to forecast GC patients' 1, 3, and 5-year survival rates.

Examining prognostic signature clinical benefits

We determined a risk score for each GC patient using the prognostic signature described above and using the median risk score as a cutoff, we separated patients into a low-risk Group (LRG) and a high-risk group (HRG). We used decision curve analysis (DCA) and evaluated the prognostic signature's areas under the receiver operating characteristic (ROC) curve at 1, 2, and 3 years to assess the precision and dependability of the risk signature. To demonstrate the risk score's ability to predict outcomes, the Kaplan-Meier (K-M) survival curves were shown using the R software. Simultaneously, we plotted risk curves and heat maps of lncRNA expression for HRG and LRG. To determine if risk score was an independent prognostic predictor, univariate and multivariate Cox regression analyses were carried out, and a forest plot was shown.

Analysis of differential expression and functional enrichment

We examined the biological functions of differential genes between HRG and LRG. Possible underlying mechanisms were explored by KEGG and GO enrichment analysis. We also looked at the association between risk scores and clinical variables.

Immune infiltration and immune function analysis

We utilized six methods (TIMER[34], CIBERSORT[35], quanTIseq[36], MCP-counter[37], xCell[38], and EPIC[39]) to figure out the number of infiltrating immune cells across HRG and LRG. Finally, we examined immune checkpoint-related gene expression levels and immune function variations in HRG and LRG, offering therapeutic therapy alternatives.

RESULTS

lncRNA identification of PRLs in GC

This research comprised 352 GC patients in total. We performed Pearson correlation analysis on 28 pyroptosis genes obtained from the database and lncRNAs in GC patients and received 532 PRLs (absolute correlation coefficient > 0.4, P < 0.05). Twenty-eight lncRNAs were substantially related to survival, according to further univariate regression analysis (P < 0.05; Figure 1A).

Figure 1 Establishing a risk signature. A: Univariate cox regression analysis was used to identify 28 pyroptosis-related long non-coding RNAs (PRLs) associated with prognosis; B: Patients were divided into high-risk group (HRG) and low-risk group, and a Kaplan-Meier analysis showed that the HRG's prognosis was worse (P < 0.001); C: Cytoscape visualized the coexpression network between nine PRLs and pyroptosis genes.

Construction of GC prognosis PRLs signature

Using stepwise Cox regression analysis, we further filtered nine PRLs from the 28 lncRNAs based on the aforementioned 28 PRLs. These lncRNAs are the most effective risk marker for GC patients' prognosis (Table 1). This is the procedure for calculating the risk score: Risk score = sum (lncRNA expression level a corresponding coefficient). We determined each GC patient's risk score using the algorithm above and used a median value to split the population into HRG and LRG. According to the K-M survival analysis, HRG patients had significantly lower survival times than LRG patients, demonstrating that these nine lncRNAs had prognostic significance (Figure 1B). The co-expression network between the nine PRLs and the related pyroptosis genes was viewed using the Cytoscape program (Figure 1C).

Table 1 Signature-included pyroptosis-related long non-coding RNAs.

LncRNA	Coefficient	HR	HR, 95%CI (lower)	HR, 95%CI (upper)
LINCO1094	0.779872703	2.181194588	1.36366574	3.48883871
Inc-NETO1-1	0.23660583	1.266941629	1.005091889	1.597009298
Inc-FZD4-1	0.680754207	1.975367005	1.12519373	3.467913749
MAGI2-AS3	-0.674642651	0.5093384	0.272837933	0.95084141
Inc-C2CD4A-2	-1.001548394	0.36731026	0.164990512	0.81772476
Inc-ZBTB10-1	0.441911848	1.555678598	1.078798355	2.243362615
Inc-PPP2R5B-5	-0.40825843	0.66480705	0.389416283	1.134951037
Inc-TDRP-4	0.900439987	2.460685544	1.248752759	4.848816788
Inc-ARMCX3-1	0.616084855	1.851664298	0.955899247	3.586843157

LncRNA: Long non-coding RNA; HR: Hazard ratio.

Evaluate the reliability and accuracy of risk signature

We drew the risk curves and expression profiles of nine PRLs using the R software program. From the distribution of risk scores and the patients' survival status, the longer the survival time, the lower the risk score (Figure 2A and B). The expression levels of LINC01094, lnc-NETO1-1, lnc-FZD4-1, MAGI2-AS3, lnc-ZBTB10-1, and-TDRP-4, and lnc-ARMCX3-1 were greater in the HRG than the LRG among these nine lncRNA expression profiles, whereas lnc-C2CD4A-2 and lnc-PPP2R5B-5 were the opposite (Figure 2C). The risk signature was considerably superior to clinical features for predicting the prognosis of GC patients, according to ROC curve analyses (Figure 2D). The survival prognosis areas under the ROC curves (AUCs) at 1, 2, and 3 years were, respectively, 0.678, 0.692, and 0.702 (Figure 2E). The risk signature was considerably superior to clinical features for predicting the prognosis of GC patients, DCA curve analyses (Figure 2F).

Figure 2 Evaluate reliability and accuracy of risk signature. A and B: The dispersion of patients' survival conditions and risk scores revealed that as the risk score rose, more people died; C: Nine pyroptosis-related long non-coding RNAs' expression heatmap between high-risk group and low-risk group; D: Age, sex, clinical stage, distant metastasis, lymph node metastasis, and tumor size had lower reliability than the risk signature; E: By doing a 1-, 2-, and 3-year receiver operating characteristic analysis, the prognostic utility of the risk signature was confirmed; F: Decision curve analysis was used to assess the dependability of the risk signature.

Value of risk scores as an independent predictor

We conducted univariate and multivariate Cox analyses on risk score and several clinical characteristics [age, sex, clinical stage, distant metastasis (M stage), lymph node metastasis (N stage), and tumor size (T stage)] to ascertain if risk score may be a standalone prognostic factor impacting GC patients' survival. The results demonstrate that the risk score was a statistically meaningful prognostic factor and might be utilized independently [P < 0.001, HR 1.923, 95%CI: (1.614-2.292), P < 0.001, HR 1.954, 95%CI: (1.617-2.362); Figure 3A and B]. Clinicians may not be able to effectively use the risk score calculation due to its complexity. As a result, we create a visual nomogram for ease of use based on the risk calculation mentioned above and clinical variables (age and clinical stage; Figure 3C).

Figure 3 The risk signature's independent prognostic value and nomogram development. A: Age, clinical stage, distant metastasis (M), lymph node metastasis (N), tumor size (T), and risk score all exhibited prognostic significance, according to a univariate Cox regression analysis; B: Age and risk score may be independent predictors of outcome, according to multivariate Cox regression analysis; C: The clinical stage and the risk score were combined to create a nomogram. ^bP < 0.01, ^cP < 0.001.

Analysis of differentially expressed genes in high- and low-risk groups for functional enrichment

To investigate the possible biological roles of genes that are expressed differently in HRG and LRG of GC patients. We searched for mRNAs that differed in expression across HRG and LRG using an R software program. Differential expression mRNAs were mostly enriched in the PI3K-Akt signaling pathway and invasion and metastasis-related pathways, including focal adhesion, ECM-receptor interaction, and regulation of the actin cytoskeleton, in the KEGG enrichment analysis (Figure 4A). The above mRNAs were discovered to be enriched in extracellular matrix-related pathways, such as extracellular matrix structural constituent, collagen binding, extracellular matrix organization, and extracellular structure organization. They were also linked to immune responses, such as antigen binding, immunoglobulin complex, and humoral immune response mediated by circulating immunoglobulin, according to GO enrichment analysis (Figure 4B). These findings point in the right direction for deducing the underlying mechanism between PRL and GC and show substantial molecular functional variations between HRG and LRG of GC. We also investigated any variations in the clinical characteristics of patients across HRG and LRG. For this purpose, we compared sex, TNM stage, and age. T and N showed substantial differences (P < 0.05). The higher the risk, the higher the T and N stages of GC, and the worse the prognosis (Figure 4C).

Figure 4 The functional enrichment analysis of differentially expressed genes in high- and low-risk groups was conducted. A and B: The enrichment of differentially expressed genes between high-risk group and low-risk group was examined using KEGG and GO; C: The heatmap illustrates the relationship between risk score and clinical features (^aP < 0.05).

Detailed examination of immune cell infiltration in HRG and LRG

We employed six methods to compute and map the immune cell infiltration Heatmap to assess if there are variations in immune cell infiltration across HRG and LRG. (Figure 5A). The majority of immune function scores were statistically significantly higher in the HRG, as shown by an analysis of immune function differences (Figure 5B). Only LGALS9 and TNFRSF14 had greater expressions in the LRG, according to our additional analysis of the expression levels of immune checkpoints between HRG and LRG. In the HRG, several immune checkpoints, like PDCD1LG2 (PD-L2), TIGIT, BTLA, CD40, and CD80, were strongly expressed (Figure 5C).

Figure 5 Analyses of immune cell infiltration, immunological function, and immune checkpoints with risk groups. A: Six algorithms to analyze the immune cell infiltration between different risk populations; B and C: Scores for immune function and the expression of immune checkpoints differentiate between those at high-risk group and low-risk group. ^aP < 0.05, ^bP < 0.01, ^cP < 0.001; ns: Not significant; HLA: Human leukocyte antigen.

DISCUSSION

Globally, GC is one of the tumors with the highest rates of morbidity and death. However, the vast majority of patients are identified at an advanced stage, and the median overall survival (OS) for advanced GC is only approximately eight months[40,41]. Therefore, it is critical to investigate new accurate biomarkers for GC diagnosis, therapy, and prognosis. Pyroptosis is a kind of programmed cell death that is crucial to the onset and progression of many illnesses, including cancer. There are few studies on PRLs in GC, so it will be of great value to construct a prognostic signature composed of PRLs.

We initially identified 28 PRLs from the TCGA database of gastric adenocarcinoma patients using univariate Cox analysis to select hallmark PRLs linked to GC prognosis (Figure 1A). The nine most useful PRLs were then collected by a stepwise Cox regression analysis, and a risk signature was created. These nine PRLs were LINC01094, lnc-NETO1-1, lnc-FZD4-1, MAGI2-AS3, lnc- ZBTB10-1, lnc-TDRP-4, lnc-ARMCX3-1, lnc-C2CD4A-2 and lnc-PPP2R5B-5 (Table 1). To evaluate the accuracy of the signature, patients were separated into HRG and LRG based on the median risk score (Figure 2A). The findings indicated that the prognosis was worse for the HRG than for the LRG (Figures 1B and 2B). The risk score AUC was 0.678 when combined with six regularly used clinical features, which was considerably higher than clinical characteristics (Figure 2D). This suggested that the risk score might more correctly predict GC patient survival. The AUC was 0.678, 0.692, and 0.702 for predicting the 1-, 2-, and 3-year survival rates, respectively (Figure 2E). DCA evaluation of the clinical value of risk signature revealed that the risk score outperformed clinical factors (Figure 2F).

Of these nine PRLs, only LINC01094 and MAGI2-AS3 were previously reported[42,43]. In glioblastoma and renal clear cell carcinoma, LINC01094 has been widely explored. Poor prognosis is linked to high expression of LINC01094 in renal clear cell carcinoma and glioblastoma. Inhibiting LINC01094 causes these two cancer cells to proliferate, migrate, and invade less often while also increasing apoptosis[44,45]. It has been shown that LINC01094 is a predictor in assessing a patient's prognosis of colon cancer[46]. Its expression level is greater than that of surrounding tissues in colon cancer tumor tissues, and it is connected to vascular invasion, positive lymph node metastases, and TNM stage. The proliferation, invasion, and migration of colon cancer cells can be inhibited by LINC01094 knockdown[47]. Within our research, LINC01094 expression was greater in the HRG, with more advanced T and N stages, compared to the LRG (Figure 4C). Corresponding to the results of autophagy-related lncRNAs in GC, the HR of LINC01094 in our investigation was 1.798, indicating that LINC01094 is an effector that increases the risk of GC[48]. Recent research has shown that LINC01094 is associated with the EMT pathway, macrophage infiltration, and weak prognosis in GC. These findings are in agreement with the findings of our gene enrichment study[49].

MAGI2 antisense RNA 3 (MAGI2-AS3) is abundant in human malignancies, interacts with up to 32 RNAs, and its expression levels correlate with cancer progression and prognosis[50]. However, most existing research focuses on breast, bladder, and lung cancer, and there are only a handful of studies on GC. MAGI2-AS3 expression is reduced in prostate cancer, and MAGI2-AS3 overexpression decreases cell biological processes such as proliferation, migration, and invasion. MAGI2-AS3 also reduces prostate cancer growth via regulating miR-142-3p, according to the researchers[51]. MAGI2-AS3 expression is reduced in non-small cell lung cancer tissues. MAGI2-AS3 overexpression decreases NSCLC cell proliferation and invasion, while MAGI2-AS3 inhibition has the reverse effect[52]. Li et al[52] found that MAGI2-AS3 is a significant predictor factor for OS and disease-free survival of GC and influences cell migration and invasion processes by favorably influencing the expression of ZEB1[52,53]. In a different research, patients with GC who had low MAGI2-AS3 expression survived noticeably longer than those who had high expression[54]. In our research, In the HRG, MAGI2-AS3 expression was considerably elevated. There are few studies on MAGI2-AS3 in GC, and it is yet unknown how it specifically contributes to the formation and progression of the disease. Following a thorough search, there is no information about lnc-NETO1-1, lnc-FZD4-1, lnc-ZBTB10-1, lnc-TDRP-4, lnc-ARMCX3-1, lnc-C2CD4A-2, lnc-PPP2R5B- 5, and more research will be conducted in the future.

We found that the risk signature was an independent prognostic factor using univariate and multivariate Cox analysis of risk scores and other clinical features (Figure 3A and B). We created a nomogram using tumor stage and risk score to predict patients' 1-, 3-, and 5-year OS (Figure 3C). In former times, many scholars constructed their unique nomogram. Gou et al[54] analyzed the operation history, treatment line, lung immune prognostic index, and platelet-to-lymphocyte ratio of immunotherapy-receiving patients with GC metastases to develop a nomogram for assessing the prognosis of these patients[54,55]. Yang et al[56] reported a nomogram based on pathological differentiation degree, Lauren type, infiltration depth, lymph node metastasis, tumor deposit, and family history of malignant tumor to predict the prognosis of GC patients[56]. We hope that the nomogram of this study, along with that existing nomogram, will shed fresh light on the prognostic prediction of GC.

The differentially expressed mRNAs in HRG and LRG were primarily enriched in the PI3K-Akt signaling pathway, invasion, and metastasis-related pathways, extracellular matrix-related pathways, and immune response-related pathways, according to KEGG and GO functional enrichment analyses (Figure 4A and B). One of the most essential signaling routes in the cell is the PI3K-Akt signaling system. It is a crucial cancer regulator, and PI3K inhibitors significantly suppress tumor development[57]. The PI3K-Akt pathway is critical in HER2-positive GC[58]. Therefore, we speculate that patients in the HRG may benefit from targeted therapy, which is worthy of further exploration. These results suggest significant molecular functional differences between HRG and LRG in GC. They may provide information for inferring the veiled mechanism of PRLs affecting the prognosis of GC. As a result, we predict that patients in the HRG may profit from targeted treatment, which merits additional investigation. These findings point to substantial molecular functional variations in GC between HRG and LRG. They may give information for deducing the veiled mechanism of PRLs influencing GC prognosis. Therefore, we hypothesize that targeted treatment may be advantageous for patients in the HRG, which merits further investigation. These findings imply important molecular functional variations between differential GC risk groups. They may provide details to deduce how PRLs impact the prognosis of GC in a covert manner.

The immune response is an essential effect of GC occurrence and treatment. The variations in immune cell infiltration between HRG and LRG were calculated using six methods. The findings revealed that the HRG had much larger levels of immune cell infiltration than the LRG (Figure 5A). Further examination of the variations in immune function between HRG and LRG revealed eight statistically significant differences, with the HRG having better immune function (Figure 5B). For instance, the APC co-stimulation function is greater in the HRG; it is the first signal of initial T-cell activation. Under the co-stimulation of two major signals, T cells can fully activate and acquire the capability to proliferate, survive, and produce related cytokines[59]. Tumor antigen presentation by human leukocyte antigen class I (HLA-I) molecules is required for effective CD8+ T cell-dependent tumor cell killing. Higher HLA scores were discovered in the HRG (Figure 5B). Chowell et al[60] studied 1535 cancer patients who had received immunotherapy and discovered that the greatest heterozygosity at the HLA-I locus (A, B, C) enhanced OS in patients with advanced cancer following immunotherapy[60]. Concerning checkpoints, the HRG had high levels of most immunological checkpoints, including PDCD1LG2 (PD-L2), TIGIT, BTLA, CD40, and CD80 (Figure 5C). Such findings imply that HRG patients have much more active local immune regulation or immune responses. The risk score may influence the effectiveness of immunotherapy and assist doctors in selecting the individuals who will benefit from it.

Our study has several restrictions. First, the mechanism of PRLs in GC has not been thoroughly investigated in any further molecular biology investigations. Second, no external database was used to verify the signature. The accuracy and dependability of the signature must thus be further validated in a larger sample.

CONCLUSION

Overall, nine PRLs were combined to create a novel signature that may be used to forecast GC patients' prognoses. The signature serves as a clinical predicting tool for immunotherapy in GC patients and may be separated from other clinical variables as an independent prognostic factor.