N6-methyladenosine-related lncRNAs as a biomarker for predicting the prognosis and immune microenvironment of clear cell renal cell carcinoma

Background: Patients with advanced clear cell renal cell carcinoma (ccRCC) have a poor prognosis and lack effective prognostic biomarkers. This study uses bioinformatics analysis to identify N6-methyladenosine-related lncRNAs (m6A-related lncRNAs) as new prognostic biomarkers for ccRCC. Methods: Gene expression data and related clinical information of ccRCC patients were extracted from the Cancer Genome Atlas Database. m6A-related lncRNAs were obtained by co-expression analysis. Univariate Cox regression analysis was performed on these lncRNAs to nd the prognostic-related m6A-related lncRNAs, and consensus clustering analysis was performed. The prognostic signature was screened by LASSO regression and a prognostic model was constructed. The predictive performance of the prognostic model was evaluated and validated by survival analysis and ROC curve analysis, etc. In addition, we also systematically analyzed the expression of immune checkpoints and immune cell inltration in ccRCC patients. Results: First, 27 m6A-related lncRNAs associated with prognosis were identied, which were signicantly differentially expressed between tumor and normal tissues. Consensus clustering analysis indicated that cluster 2 was associated with poor prognosis, low stromal score, high expression of PD-1, PD-L1, CTLA-4, LAG-3, TIM-3, TIGIT, and immunosuppressive cell inltration. The signaling pathways related to tumor progression, drug resistance, and angiogenesis and biological processes related to protein methylation and phosphatidic acid metabolism are signicantly enriched in cluster 2. Subsequently, LASSO regression analysis was used to construct a prognostic risk model based on 7 m6A-related lncRNAs signature, which can be used as an independent prognostic indicator. After a series of analyses, it was shown that this model had good sensitivity and specicity, can predict the prognosis of patients with different clinical stratications and was associated with the progression of ccRCC. The expression levels of immune checkpoints were signicantly increased in high-risk patients, and there was a certain Genomes; GO, Gene Ontology; TME, tumor microenvironment; Target of Rapamycin; VEGF, vascular endothelial growth factor; NES, normalized enrichment score; PD-1, programmed cell death 1; PD-L1, programmed cell death-Ligand 1; CTLA-4, cytotoxic T lymphocyte-associated antigen 4; LAG-3, lymphocyte activation gene-3; TIM3, T cell immunoglobulin and mucin domain-containing protein 3; TIGIT, T cell immunoglobulin and ITIM domain.


Background
Renal cell carcinoma (RCC) is a common genitourinary tumor, and its incidence has increased year by year, accounting for 3%-5% of adult malignancies, and is higher in men than in women [1]. Clear cell renal cell carcinoma (ccRCC) is the most common pathological type of RCC, accounting for approximately 75% of RCCs [2]. The classical RCC triad is hematuria, ank pain, and abdominal masses. However, most RCC patients are asymptomatic at onset and are already at an advanced stage at the time of diagnosis.
Although the 5-year survival rate of early RCC is 93%, the prognosis of patients with locally advanced or metastatic RCC is generally poor, and the 5-year survival rate is only approximately 12% [3][4]. Therefore, an effective prognostic marker is of great signi cance for early diagnosis and treatment as well as for improving patient prognosis. ccRCC is resistant to traditional radiotherapy and chemotherapy, and the treatment methods for patients with advanced ccRCC are very limited. Targeted drugs such as sorafenib are the standard means of treatment for advanced stage patients. In recent years, some clinical studies have demonstrated that immunotherapy is effective for advanced RCC [5][6]. As people's awareness of tumor immunity continues to rise, novel immunotherapeutic drugs continue to emerge. Searching for novel biomarkers is critical for the better selection of patients who may bene t most from these treatments [7].
Long non-coding RNAs (lncRNAs) are a group of noncoding RNAs with a length greater than 200 nucleotides [8]. They can regulate gene expression through epigenetic regulation, transcriptional regulation and posttranscriptional regulation and are involved in biological processes such as the regulation of tumor cell cycle, cell proliferation, differentiation, apoptosis and migration [9][10][11]. An increasing number of lncRNAs have been shown to be abnormally expressed in cancers and have a signi cant correlation with treatment and clinical prognosis, providing the possibility for lncRNAs to become novel tumor biomarkers and therapeutic targets [12]. For example, the level of lncRNA small nucleolar RNA host gene 14 (SNHG14) is signi cantly increased in ovarian cancer tissues, and the inhibition of SNHG14 signi cantly inhibits the migration and invasion of cells [13]. The upregulation of the lncRNA gastric cancer metastasis associated long noncoding RNA (GMAN) in gastric cancer tissues is associated with metastasis in patients [14]. The lncRNA Urothelial cancer associated 1 (UCA1) is signi cantly upregulated in RCCs and is positively correlated with tumor differentiation and tumor-nodemetastasis (TNM) staging; it promotes the malignant phenotype of RCCs through the regulation of the miR-182-5p/DLL4 axis [15].
N6-methyladenosine (m6A) is methylation that occurs in the N6-position of adenosine [16]. It was rst discovered in eukaryotic messenger RNA (mRNA) in 1970 and is considered to be the most common and abundant RNA modi cation [17]. m6A modi cation is a reversible and dynamic process. It is regulated by 3 types of m6A regulators. m6A is catalyzed by methyltransferases (called "writers", such as METTL3, METTL14, and RBM15), is removed by demethylases (called "erasers", such as FTO and ALKBH5), and is recognized by binding proteins (called "readers", such as YTHDC1-2 and YTHDF1-3), thereby regulating RNA metabolism [16][17]. m6A modi cation has been con rmed to be involved in the regulation of the occurrence and progression of a variety of cancers, including ccRCC and breast cancer [18][19]. Studies have shown that lncRNAs are also widely modi ed by m6A [20] and that the interaction between the two is involved in tumor progression, metastasis, drug resistance and immune responses [21]. Studies have reported that m6A-related lncRNAs can be used as potential biomarkers to predict the prognosis of patients with glioma [22]. However, the biological role of the interaction between lncRNA expression and m6A-related genes in ccRCC has not been explored. Therefore, we used bioinformatics to screen for m6Arelated lncRNAs that are closely related to ccRCC prognosis and constructed a prognostic model. We hope that our results will help to better assess the prognosis of ccRCC patients and guide clinical treatment. Our results also provide a basis for further exploration of the potential mechanism of m6A modi cation of lncRNAs in ccRCC.

Data acquisition and collation
Transcriptomic sequencing (RNA-seq) data, fragments per kilobase of transcript per million mapped reads (FPKM), and related clinical information of ccRCC patients were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer. gov/). The Genome Reference Consortium Human Build 38 (GRCh38) was downloaded from GENCODE (https://www.gencodegenes.org/human), and the expression data of lncRNAs in the transcriptome were extracted based on the gene biotype. These data were obtained from public databases, and therefore, ethics approval was not required.
Biological characteristics of m6A-related lncRNAs To investigate the clinical signi cance of m6A-related lncRNAs associated with prognosis, we used the Consensus Clustering algorithm and the ConsensusClusterPlus package [25] of R, version 4.0.3, to classify ccRCC patients into different subtypes based on the expression levels of prognosis-related lncRNAs. The differences in clinical features and prognosis were analyzed. Consensus clustering is a method that provides quantitative evidence for determining the membership and number of possible clusters in a dataset and is widely used in cancer genomics. Because resampling was used, the obtained clustering results had excellent stability.
To further explore the biological processes mediated by m6A-related lncRNAs, we used GSEA software (version 4.1.0) to determine gene expression enrichment for different subtypes in the Molecular Signatures Database (MSigDB) Collection (c5.go.bp.v7.2.symbols.gmt; c2.cp.kegg. v7.2symbols.gmt) to analyze the difference between GO functional enrichment and KEGG pathway enrichment. A normalized enrichment score (NES) > 1 and nominal p-value (NOM p-val) < 0.05 were used to determine the difference between different genotypes.
Correlation analysis of m6A-related lncRNAs and immune checkpoints To investigate the indication effect of m6A-related lncRNAs in immunotherapy for ccRCC, the differential expression of key genes associated with currently used tumor immune checkpoint inhibitors, including programmed cell death 1 (PD-1), programmed cell death-Ligand 1 (PD-L1), cytotoxic T lymphocyteassociated antigen 4 (CTLA-4), lymphocyte activation gene-3 (LAG-3), T cell immunoglobulin and mucin domain-containing protein 3 (TIM-3) and T cell immunoglobulin and ITIM domain (TIGIT), was compared in tumor and normal samples and among different clusters using the limma package of R software [26]. The corrplot package [27] was used to visualize the coexpression relationship between the above genes and m6A-related lncRNAs associated with prognosis.
Correlation analysis of m6A-related lncRNAs and the immune microenvironment To analyze the effect of m6A-related lncRNAs on the tumor immune microenvironment, we used the CIBERSORT algorithm [28] to evaluate the immune cell in ltration of different clusters. CIBERSORT is a tool for the deconvolution of the expression matrix of human immune cell subtypes based on the principle of linear support vector regression. Immune cell in ltration can be estimated by RNA-seq data.
Leukocyte signature matrix (LM22) contains 547 genes that distinguish 22 human immune cell phenotypes, including seven T cell types, naive and memory B cells, plasma cells, NK cells, and myeloid subsets. We used the CIBERSORT algorithm to analyze the RNA-seq expression pro le based on LM22 classi cation. We evaluated the abundance of immune cell subsets in samples and visualized the differences in immune cell in ltration between different types. In addition, we used the ESTIMATE algorithm in the R estimate package [29] to calculate the ratio of immune cells and stromal cells in the tumor microenvironment (TME) for each sample to evaluate tumor purity.

Construction and validation of the prognostic risk model
First, the Caret package [30] was used to randomly divide the samples with complete survival information in the TCGA database into 2 groups, i.e., a training set and a test set, with each set consisting of approximately 50% of cases. Subsequently, LASSO regression analysis was used to construct the prognostic model. The optimized model was obtained using the penalty parameter estimated by 10-fold cross-validation. The risk score of the prognostic model = , where Coef represents the regression coe cient, and β represents the m6A-related lncRNA expression value. The risk score for each patient was calculated using the formula, and then, the training set and test set were divided into a high-risk group and a low-risk group based on the median score. The R survival package was used for Kaplan-Meier analysis. The log-rank test was used to compare the overall survival rates for the high-and low-risk groups. A time-dependent receiver operating characteristic (ROC) curve was plotted. Area under the curve (AUC) > 0.60 was considered an acceptable prediction. In addition, we also used univariate and multivariate Cox regression analyses to assess whether the risk score could be an independent prognostic factor for ccRCC. To evaluate the predictive ability of the model for different populations, Kaplan-Meier analysis was performed for age, sex, grade, stage, and TNM staging. Finally, we analyzed the correlation between risk scores and clinicopathological features to assess the ability of this model to predict ccRCC progression.

Analysis of immune differences in patients with different prognostic risks
To understand the signi cance of this risk model in the assessment of the immune microenvironment and immunotherapeutic e cacy in ccRCC, we analyzed the differences in immune checkpoint expression between high-and low-risk patients and the correlation between risk scores and tumor-in ltrating immune cells.

Results
Identi cation of m6A-related lncRNAs associated with ccRCC prognosis The RNA-seq data of 611 samples (tumor samples, 539; normal samples, 72) downloaded from TCGA were combined with GRCh38 downloaded from GENCODE to obtain 14086 lncRNAs. Person correlation analysis of the expression matrix of these lncRNAs and m6A-related genes yielded a total of 239 lncRNAs that were signi cantly positively correlated with RBM15 and METTL3 ( Figure 1A, Table S1). Through univariate Cox regression analysis, 27 m6A-related lncRNAs signi cantly correlated with prognosis were obtained (p < 0.001, Figure 1B, Table S2). The expression levels of these lncRNAs were signi cantly different between ccRCC tumor tissue and normal tissue ( Figure 1C Consensus clustering of m6A-related lncRNAs correlates with prognosis and biological functions in ccRCC We adopted the resampling method to sample 80% of the samples and used the K-Means clustering algorithm to select the k value with the highest intracluster correlation, i.e., k = 2, as the optimal number of clusters, based on the results of k typing from 2 to 9 (Figure S1A-C). A total of 530 ccRCC patients were classi ed into 2 subtypes, namely, cluster 1 (n = 315) and cluster 2 (n = 215) (Figure 2A). The prognoses of clusters 1 and 2 were signi cantly different, and the survival rate for cluster 2 was lower than that for cluster 1 (p < 0.001, Figure 2B). Twenty-four high-risk m6A-related lncRNAs were highly expressed in cluster 2, and there were no signi cant differences in age, sex, grade, stage, or TNM staging between the 2 genotypes ( Figure 2C).

Consensus clustering of m6A-related lncRNAs correlates with immune genes
To explore the association between m6A-related lncRNAs and immunotherapy, we evaluated the differential expression of 6 types of immune checkpoints between the 2 subtypes and the correlation between immune checkpoints and m6A-related lncRNAs. We found that the expression levels of PD-1, PD-L1, CTLA-4, LAG-3, TIM-3, and TIGIT in tumor samples were signi cantly higher than those in normal samples (all p < 0.001) and the expression levels of PD-1, PD-L1, and CTLA. -4, LAG-3, and TIGIT in cluster 2 were signi cantly higher than those in cluster 1 ( Figure 3A

Consensus clustering of m6A-related lncRNAs correlates with immune in ltration
In the TME, in addition to tumor cells, there are many nontumor components closely related to the biological processes of tumors, including stromal cells and immune cells. Using the unique properties of the transcription pro le of tumor samples, we can predict the content of stromal cells and immune cells, i.e., the immune score and stromal score; the higher the score, the greater is the proportion of the corresponding components. Finally, the tumor purity of each tumor sample was calculated based on the sum of the 2 (ESTIMATE score); the higher the score, the lower is the tumor purity. The content of stromal cells in cluster 1 was higher than that in cluster 2 (p = 0.045). However, there were no signi cant differences in the percentage of immune cells and tumor purity between the 2 ( Figure 4A-C).
In addition, we analyzed the in ltration of 22 immune cells in 2 subtypes and found that the in ltration of naive B cells and neutrophils in cluster 1 was high, while the in ltration of CD8 T cells, follicular helper T cells and resting NK cells in cluster 2 was more signi cant. In addition, although T regulatory cells (Tregs) were not signi cantly different between the 2 subtypes, the in ltration was more signi cant in cluster 2 ( Figure 4D, E).
Based on the median risk score, patients in the training set were divided into high-and low-risk groups for Kaplan-Meier survival analysis. The results showed that the prognosis of patients in the high-risk group was lower than that of patients in the low-risk group (p < 0.001, Figure 5A). The AUCs at 1 year, 3 years, and 5 years were 0.729, 0.762, and 0.777, respectively, indicating that this model could well predict the prognosis of patients ( Figure 5B). To further verify the accuracy of the prognostic model, we used it to analyze the test set. The results showed that the survival rates for patients in the high-risk and low-risk groups were signi cantly different (p < 0.001, Figure 5C); the AUCs were 0.69, 0.664, and 0.706, respectively, suggesting that the prognostic model had excellent sensitivity and speci city ( Figure 5D).
The risk score curve, survival status, and heat map of gene expression for patients in the high-risk and low-risk groups are shown in Figure 5E-F. As the risk score increased, ccRCC patient deaths increased. In high-risk patients, 5 high-risk lncRNAs (LINC00342, AF117829.1, AC009948.2, SNHG10, and AL133243.3) were upregulated, and 2 low-risk lncRNAs (AC018752.1 and RPL34-AS1) were downregulated; in contrast, the expression levels of these lncRNAs in low-risk patients were opposite.
Next, we performed univariate and multivariate Cox regression analyses to evaluate the prognostic value of risk scores and other clinical features of the prognostic model. Univariate Cox regression analysis indicated that age, grade, stage, and risk score were correlated with the prognosis of ccRCC patients ( Figure 5G-H). Multivariate Cox regression analysis suggested that the risk score could be used as an independent prognostic risk factor (training set, p < 0.001; test set, p = 0.002; Figure 5I-J).

Prognostic risk score correlates with clinicopathological characteristics
We further performed Kaplan-Meier survival analysis of different ages, sex, grades, stages, and TNM stages and compared the survival rates for high-and low-risk patients in all strati cations to explore whether our prognostic model is suitable for patients within different clinical strati cations. The results are shown in Figure 6A-G. In older age (>65 years) or younger age (≤65 years) patients, males or females, with poorly differentiated or undifferentiated tumors (G3-4), different tumor sizes and in ltration depths (T1-2, T3-4), no lymph node (N0) or distant metastasis (M0), and early stage (stage I-II) or advanced stage (stage III-IV) disease, the survival rate of high risk patients signi cantly decreased. These results indicated that the m6A-related lncRNA-based prognostic model is a powerful tool for predicting the prognosis of ccRCC patients within different clinical strati cations.
Finally, we analyzed the correlation of risk scores with different clinicopathological characteristics and found that risk scores were signi cantly different among different clusters, grades, immune scores, TNM staging, and stage. Risk scores were signi cantly higher for patients in cluster 2, G3-4, low immune scores, T3-4, N1, and M1 staging, and stage III-IV ( Figure 7A-J), suggesting that the expression of m6Arelated lncRNAs is related to the occurrence and development of ccRCC and that the model may be able to predict the progression of ccRCC.

Prognostic risk score correlates with immune checkpoint expression and immune cell in ltration
To determine whether our risk model could re ect the conditions of patients' immune microenvironment and to provide guidance for immunotherapeutic e cacy, we analyzed the differential expression of immune checkpoints among patients with different risks and the relationship between the risk score and immune cell in ltration. The expression levels of 6 immune checkpoints, i.e., PD-1, PD-L1, CTLA-4, LAG-3, TIM-3, and TIGIT, were signi cantly increased in the high-risk group ( Figure 8A). The contents of memory B cells, CD8 T cells, follicular helper T cells, and Tregs were positively correlated with risk scores. The contents of these cells increased in ccRCC tissue with increased risk score. However, the other immune cell types, i.e., naive B cells, activated dendritic cells, resting dendritic cells, resting mast cells, monocytes, and resting memory CD4 T cells, were negatively correlated with risk score ( Figure 8B). Discussion lncRNAs have been con rmed to be abnormally expressed in a variety of malignant tumors and to participate in tumor occurrence, development, invasion, and metastasis. lncRNAs play important mediating roles in cancer signal transduction pathways by interacting with proteins, RNAs and lipids [31]. Because lncRNAs have high organ and cell speci city, they can be found in many tissues and body uids of patients. Some speci c lncRNAs can be used as novel tumor biomarkers for tumor diagnosis, prognostic evaluation, therapeutic targets, and drug sensitivity prediction [32][33]. m6A modi cation is the most abundant epigenetic methylation modi cation in mammalian mRNAs and lncRNAs [34], affecting almost every process of RNA metabolism, including mRNA processing, translation and the biological occurrence of lncRNAs [35]. The m6A modi cation of lncRNAs plays an important regulatory role in the occurrence and development of a variety of cancers. For example, the level of m6A of the lncRNA NEAT1-1 can effectively predict the risk of death of prostate cancer patients, and high levels of m6A of NEAT1-1 is related to prostate cancer bone metastasis [36]. METTL3-mediated and METTL14-mediated m6A modi cation enhances the stability of LNCAROD in head and neck squamous cell carcinoma (HNSCC) cells, an effect that is associated with the high expression of LNCAROD in HNSCC [37]. Here, we identi ed 7 m6A-related lncRNAs that were signi cantly correlated with the prognosis of ccRCC patients and used those lncRNAs to construct a reliable risk model to predict the prognosis of patients and the TME, information that has certain guiding signi cance for tumor immunotherapy.
In this study, 27 prognostic m6A-related lncRNAs were identi ed by analyzing gene expression data of 530 ccRCC patients in the TCGA database. Among those 27 lncRNAs, 24 high-risk lncRNAs were highly expressed in tumor tissue, suggesting that these lncRNAs might have carcinogenic effects in the occurrence and development of ccRCC. The expression of these lncRNAs was signi cantly positively correlated with the m6A writers RBM15 and METTL3. RBM15 binds the m6A complex and recruits it to a special RNA site [38]. METTL3 is the rst m6A methyltransferase that has been extensively studied in tumors and plays a major catalytic role in the m6A addition process [39]. METTL3 can promote tumorigenesis and malignant progression, is highly expressed in a variety of malignant tumors, such as bladder cancer, breast cancer, and lung cancer, and is an indicator of poor patient prognosis [40][41][42]. It was found that the expression of METTL3-induced lncRNAs ABHD11-AS1 and LINC00958 was upregulated in non-small cell lung cancer (NSCLC) and liver cancer tissues and cells and closely associated with the poor prognosis of patients [43][44]. To investigate the biological characteristics of these m6A-related lncRNAs associated with prognosis, we classi ed, via consensus clustering, ccRCC patients into 2 subtypes based on the expression of lncRNAs. We found that patients in cluster 2 with high expression of high-risk lncRNAs had a poor prognosis, suggesting that m6A-related lncRNAs can be used as biomarkers to predict the prognostic risk of ccRCC. In addition, cluster typing was closely related to immune checkpoints, stromal scores, and immune cell in ltration. In cluster 2 patients, the stromal score was low, the contents of stromal cells were low, and immune-inhibitory cell in ltration was more obvious; PD-1, PD-L1, CTLA-4, LAG-3, and TIGIT were highly expressed and positively correlated with m6Arelated lncRNAs. These results suggest that prognostic m6A-related lncRNAs might become a signature for the assessment of the TME and immunotherapeutic effects.
To understand the regulatory mechanism of m6A-related lncRNAs in ccRCC, we performed pathway and functional enrichment analysis for clusters 1 and 2. The results showed that signaling pathways associated with tumor occurrence and development, drug resistance, and angiogenesis, such as the mTOR signaling pathway, Notch signaling pathway, VEGF signaling pathway, and ABC transporters, were enriched to varying degrees in cluster 2 patients with a poor prognosis. mTOR is regulated by a variety of cell signals, mainly through the phosphatidylinositol 3-kinase (PI3K)/Akt/mTOR pathway, to achieve cell proliferation, autophagy and apoptosis, among other regulatory functions. In many cancer types, the mTOR signaling pathway is abnormally activated and is involved in tumor formation, the regulation of immune cell differentiation and tumor metabolism [45]. Notch is widely expressed in many species and is highly evolutionarily conserved. It affects cell differentiation, proliferation and apoptosis and is associated with the occurrence and development of cancers. lncRNAs can not only participate in the NOTCH signaling pathway as regulatory factors of target genes but also affect the transcription of downstream genes in the nucleus [46]. ABC transporters are a family of energy-dependent transport proteins located on the cell membrane. ABC transporters can mediate the unidirectional e ux of antitumor drugs and cause multidrug resistance in tumors [47]. GO analysis indicated that m6A-related lncRNAs were involved in biological processes such as protein methylation, microtubule-based movement regulation, and phosphatidic acid metabolism.
Based on the above results, we can con rm that m6A-related lncRNAs are closely related to the prognosis and biological process of ccRCC patients. Therefore, based on univariate and multivariate Cox regression analyses, we ultimately obtained 7 m6A-related lncRNAs that were signi cantly correlated with prognosis (LINC00342, AC018752.1, RPL34-AS1, AF117829.1, AC009948.2, SNHG10, and AL133243.3) and constructed a prognosis risk model. Studies have found that long intergenic non-protein coding RNA 00342 (LINC00342) can regulate the growth, invasion and metastasis of colorectal cancer and NSCLC cells and is closely associated with the poor prognosis of patients [48][49]. The lncRNA ribosomal protein L34 antisense RNA 1 (RPL34-AS1) is localized on human chromosome 4q25 and has antitumor effects in esophageal carcinoma and papillary thyroid carcinoma. RPL34-AS1 overexpression can inhibit tumor cell proliferation and invasion and promote apoptosis [50][51]. Small nucleolar RNA host gene 10 (SNHG10) has been reported to be an oncogenic lncRNA of gastric cancer, liver cancer, osteosarcoma and other malignancies [52][53][54]. It is highly expressed in a variety of malignant tumors and is involved in the proliferation, invasion, and metastasis of tumor cells. However, some studies also reported that the high expression of SNHG10 predicts a good prognosis for NSCLC patients [55].
Survival analysis, ROC curves and risk curves were used to analyze the accuracy and stability of the model; the results indicated that the model can accurately distinguish high-and low-risk patients and accurately predict the prognostic risk of ccRCC. The strati cation of different clinical traits showed that this model could effectively predict the prognosis of patients of different ages, sex, and stages. Univariate and multivariate Cox regression analyses showed that this risk model could be used as an independent prognostic indicator for ccRCC patients. In addition, we also analyzed the relationship between the expression of prognostic m6A-related lncRNAs in the model and different clinicopathological characteristics. We found that the risk score was associated with the progression of ccRCC. In summary, our prognostic model is reliable and can be used to identify the risk and prognosis of ccRCC patients, information that is conducive to early intervention and treatment.
In recent years, immunotherapy targeting immune checkpoints has made substantial breakthroughs, bringing new hope to ccRCC patients. However, the complex microenvironment of tumors can mediate immune escape, leading to the failure of immunotherapy [56]. lncRNAs are overexpressed during the development, differentiation and activation of immune cells, such as macrophages, dendritic cells, neutrophils, T cells, B cells and bone marrow mesenchymal stem cells [57]. Recent studies have found that lncRNAs are involved in various processes of the immune response in the TME and the promotion of tumor immunosuppression [58] and play a role in the evaluation and measurement of the immunotherapeutic response in various cancers, such as endometrial cancer and liver cancer [59][60]. Huang et al. [61] found that the lncRNA NKILA alters the balance between immune activating and immunosuppressive T cell subsets in the TME by regulating the apoptosis sensitivity of T cell subsets, resulting in tumor immune escape. Wang et al. [62] showed that lncRNA-MALAT1 promotes the immune escape of diffuse large B-cell lymphoma by targeting miR-195. MALAT1 gene knockout promotes the proliferation of CD8+ T cells and inhibits the epithelial-mesenchymal transition (EMT)-like signal transduction process through Ras-extracellular signal-regulated kinase (RAS/ERK) signaling pathways. The lncRNA SNHG15 promotes PD-L1 expression through the inhibition of miR-141 and participates in the immune escape of gastric cancer [63]. Currently, there are limited studies that have investigated the effects of m6A-regulated lncRNAs on the immune microenvironment of ccRCC. In this study, we found that the expression levels of PD-1, PD-L1, CTLA-4, LAG-3, TIM-3, and TIGIT were higher in high-risk patients, who are more likely to bene t from immunotherapy. The risk score was positively correlated with the in ltration level of memory B cells, CD8 T cells, follicular helper T cells, and Tregs. The risk score was negatively correlated with the in ltration level of naive B cells, activated dendritic cells, resting dendritic cells, resting mast cells, monocytes, and resting memory CD4 T cells. These results suggest that m6Arelated lncRNAs are involved in the regulation of the immune microenvironment to a certain extent. Pan et al. [64] found that dendritic cells resting, dendritic cells activated, mast cells resting, mast cells activated and eosinophils are associated with favorable prognosis in patients with ccRCC, whereas B cells memory, T cells follicular helper and Tregs are correlated with poorer outcome.
This study has some limitations. First, the risk model constructed in this study was based on a public clinical database: TCGA. The results were con rmed in a TCGA cohort but lacked external validation. Therefore, further testing in larger multicenter clinical patient cohorts is needed in the future. In addition, the speci c regulatory mechanism of the screened m6A-related lncRNAs in ccRCC requires further study.

Conclusion
This study systematically analyzed the prognostic value of m6A-related lncRNAs in ccRCC patients, as well as their signi cance in the assessment of the immune microenvironment and immunotherapeutic e cacy. A prognostic risk model based on 7 m6A-related lncRNAs was constructed and validated. The

Declarations
Ethics approval and consent to participate The data of the patients in this study were obtained from the public database datasets, and written informed consent was obtained from these patients.