A New Prognostic Signature Constructed with Necroptosis-Related lncRNA in Bladder Cancer

Background Bladder cancer (BC) accounts for the most common urologic malignancy, leading to a heavy social burden over the world. We aim to search for a novel prognostic biomarker with necroptosis-related lncRNAs of bladder cancer in this study. Methods We download the RNA-sequencing data and corresponding clinical information of BC patients from TCGA. We performed Pearson correlation analysis to identify necroptosis-related lncRNAs (NRlncRNAs). Then, we used univariate Cox regression, Lasso Cox analysis, and multivariate Cox regression to construct the optimal prognostic model. Next, we used Kaplan–Meier curves, Cox regression, receiver operating characteristic (ROC) curves, nomogram, and stratified survival analysis to evaluate the capacity of the prognostic signature. Furthermore, gene set enrichments in the signature and the correlation between prognostic signature and necroptosis genes, tumor microenvironment, immune infiltration, and immune checkpoints of BC were also explored. Results A 7-NRlncRNAs signature comprising FKBP14-AS1, AL731567.1, LINC02178, AC011503.2, LINC02195, AC068196.1, and AL136084.2 was constructed to predict the prognosis of BC in this research. Cox regression analysis showed that the signature could be an independent prognostic factor for BC patients (P < 0.001). Compared to other clinicopathological characteristics, this signature displayed a better capacity of prediction with the area under the curve (AUC) of 0.745. Stratified analysis using various clinical variables demonstrated that the prognostic signature has good clinical fitness. GSEA showed that focal adhesion and the WNT signaling pathway were enriched in the high-risk group. Immune infiltration analysis indicated that the signature was significantly inversely correlated with infiltration of CD8+ T cells and CD4+ T cells while positively correlated with macrophages and cancer associated fibroblasts. Immune checkpoint analysis revealed that the expressions of protective factors were significantly lower in the high-risk group, while expressions of cancer promotors were significantly higher in this group. The gene expression analysis displayed that necroptosis genes such as FADD, FAS, MYC, STAT3, PLK1, LEF1, EGFR, RIPK3, CASP8, BRAF, ID1, GATA3, MYCN, CD40, and TNFRSF21 were significantly different between the two groups. Conclusions The 7-NRlncRNAs signature can predict the overall survival of BC and may provide help for the individualized treatment of BC patients.


Introduction
As the most common malignance in the urinary system, bladder cancer (BC) leads to a heavy social burden with over 200000 related deaths worldwide annually [1]. Many risk factors such as smoking, chronic infection or irritation, and occupational exposure to carcinogenic chemicals have been found associated with occurrence of BC [2]. However, the pathogenesis of BC is still unclear. BC patients with similar histology or pathological stage may have completely different prognosis. Patients with nonmuscle-invasive bladder [4,5]. Necroptosis has been confrmed to play critical roles in infection diseases and noninfammatory diseases, including oncogenesis, metastasis, and immune escape of cancer [6,7]. Interestingly, necroptosis can hamper tumor progression in some tumors when apoptosis failed to be induced on the one hand and provoke the infammatory responses and promote cancer immunosuppression in some cancers on the other hand [8]. Tese dual efects on cancers hint that targeting necroptosis would be a new strategy for the remedy of cancers, especially immunotherapy of cancers such as BC.
Long noncoding RNA (LncRNA) is a type of nonprotein-coding RNA with a length of more than 200 nucleotides existing in the nucleus or cytoplasm. Although lncRNA lacks meaningful open reading frame, it plays crucial roles in the proliferation, diferentiation, and apoptosis of cells, including the oncogenesis and progression of cancers [9]. For example, LINC01614 promotes the proliferation, migration, and invasion of bladder cancer cells through the miR-217/RUNX2/Wnt/β-catenin axis [10]. LncRNA SLC16A1-AS1 induces metabolic reprogramming of BC cells towards favoring invasiveness by acting as a target and coactivator of E2F1 in bladder cancer [11]. Up to now, there have been few research works of lncRNAs related to necroptosis and necroptosis-related lncRNAs (NRlncRNAs) in BC that have not been studied.
In our research, we identifed NRlncRNAs in BC using transcript data from TCGA and constructed a prognostic signature with diferentially expressed NRlncRNAs between bladder cancer tissues and normal tissues. Ten, we analyzed and evaluated its capacity of predicting overall survival of BC by diferent methods. Furthermore, we explored the relationships between this signature and immune infltration, immune checkpoint, and anticancer drug sensitivity.

Data Acquisition.
Te RNA-sequencing data and relevant clinical information of bladder cancer tissues and normal tissues were downloaded from Te Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) in March 2022. Te RNA-sequencing data were normalized to fragments per kilobase million (FPKM) format and preprocessed with Perl language (Version strawberry-perl-5.32.1.1; https://www. perl.org/) to obtain gene expression matrix of BC and normal tissues. Patients with missing overall survival values (OS) or short survival (OS < 30 days) in this study were excluded to reduce statistic bias.

Identifcation of Necroptosis-Related lncRNAs.
We obtained a list of 67 necroptosis genes from previously mentioned literature in PubMed [12]. Te correlation between necroptosis genes and lncRNAs in BC was analyzed with Pearson correlation analysis. Ten, NRlncRNAs were identifed with the standard of |Coefcient| > 0.5 and P < 0.001. Subsequently, we screened out the diferentially expressed NRlncRNAs in BC and normal tissues with the standard of |Log 2 fold change| > 1 and P < 0.05 using limma R package. Te network of mutual regulation between these lncRNAs and target genes was visualized with "igraph" package in R language.

Construction and Verifcation of the Prognostic Signature.
All enrolled BC patients were randomly divided into the training set and testing set at the ratio of 1 : 1. In the training set, univariate Cox regression analysis was performed to identify NRlncRNA related to prognosis of BC patients (P < 0.05). Lasso Cox analysis was applied to determine the optimal NRlncRNAs associated with BC patients' prognoses via the glmnet R package. To prevent overftting, 10foldcross-validation and P < 0.05 were set in the processing. Subsequently, multivariate Cox regression analysis was used to construct a predictive model with the optimal NRlncRNAs [13]. Ten, the risk score of each BC patients was calculated according to the expression levels of the NRlncRNAs and corresponding regression coefcients with the following formula: risk score-� n k�1 coef(lncRNA k ) × exp(lncRNA k ). Coef (lncRNA) represents the coefcient of each lncRNA in the model, while exp (lncRNA) represents the expression levels of the lncRNA. Finally, all patients were divided into the high-risk group or low-risk group according to the median risk score. Using the "survival" package in R, Kaplan-Meier curves were plotted, and the log-rank test was performed to compare whether there was a diference of survival between the high-risk group and the low-risk group. Te testing set and entire set were used to validate the prognostic model.

Assessment of the Prognostic Signature.
Univariate and multivariate Cox regression analyses were used to examine where the risk score of the model was an independent prognostic factor for BC patients. Receiver operating characteristic (ROC) curves of the prognostic signature and clinical characteristics were plotted, and the area under the curve (AUC) was calculated with "survival," "survminer" and "timerROC" packages in R. A nomogram was generated to predict 1-year, 3-year, and 5-year overall survival of BC patients by combining risk score and age, gender, clinical stage, T stage, N stage, M stage, and grade classifcation. Calibration curves were plotted to evaluate whether the nomogram was consistent with the actual situation. In this process, "survival," "regplot," and "rms" packages in R were used. In addition, stratifed survival analysis according to diferent clinicopathological characteristics was performed to further evaluate the capacity of the prognostic signature.  Journal of Oncology based on algorithms including XCELL, TIMER, QUAN-TISEQ, MCPCOUNTER, EPIC, CIBERSORT-ABS, and CIBERSORT. Te correlations between immune cells and risk scores were evaluated using Spearman correlation analysis and visualized with a bubble chart. In this process, the Wilcoxon signed-rank test and R packages including "limma," "scales," "ggplot2," "ggtext," "tidyverse," and "ggpubr" were used. Te tumor microenvironment (TME), infltrated immune cells, immune checkpoint activation, and necroptosis gene expression between two risk groups were analyzed with the ESTIMATE algorithm, single sample gene set enrichment analysis (ssGSEA), and "ggpubr" package in R language, respectively.

Clinical Signifcance of the Signature in Drug Terapy.
We used the "pRRpphetic" package in R language to assess the diferences of responses to drug therapy between the low-risk group and the high-risk group. Te response to anticancer drugs was determined by half-maximal inhibitory concentration (IC50) of patients in Genomics of Drug Sensitivity in Cancer (GDSC) (https://www.cancerrxgene. org/).

Necroptosis-Related lncRNAs in BC.
We obtained the transcriptome RNA-seq and clinical data of 411 BC tissues and 19 normal tissues from TCGA database. By Pearson correlation analysis and expression analysis, 440 NRlncRNAs were identifed diferentially expressed in BC tissues and normal tissues (|Log 2 fold change| > 1 and P < 0.05), including 59 down-regulated lncRNAs and 381 up-regulated lncRNAs, as shown in Figure 1(a). Te heatmap of the 100 lncRNAs with most signifcance (50 upregulated lncRNAs and 50 down-regulated lncRNAs) is shown in Figure 1(b). Te regulation network between these lncRNAs and target genes is shown in Figure 1(c).

Construction of the Prognostic Signature.
Using univariate Cox regression analysis and Lasso Cox regression analysis, we obtained 15 lncRNAs correlated to the overall survival of BC patients when the frst-rank value of Log(λ) was the minimum likelihood of deviance (Figures 2(a) and 2(b)). Te P values and hazard ratio of these lncRNAs are shown in Figure 2(c). Te Sankey diagram showed positive regulation between necroptosis genes and all lncRNAs obtained ( Figure 2(d)). Ten, in the training set, we constructed the optimal prognostic model with 7 lncRNAs using multivariate Cox regression analysis. Te risk score of each BC patient is calculated with the formula: risk score � (1.91355×FKBP14- Based on the median value of risk score, each patient was sorted into the low-risk group or high-risk group. Te distribution of the risk score and survival status of the training set, test set, and entire set are shown in Figures 3(a)-3(f ), suggesting that deaths increased as the risk score elevated. Heatmaps of Figures 3(g)-3(i) showed that FKBP14-AS1, LINC02178, and AL136084.2 were up-regulated in the high-risk group, while AL731567.1, AC011503.2, LINC02195, and AC068196.1 were up-regulated in the lowrisk group. Kaplan-Meier curves of various sets indicated that patients with a higher risk score had worse overall survival compared to patients with a lower risk score (Figures 3(j)-3(l), P < 0.001).
In addition, a nomogram comprising clinicopathological features and risk scores was set up to predict the 1-year, 3year, and 5-year survival rate of BC patients ( Figure 4(e)). It is interesting that gender showed signifcant diference in calculation of points in this nomogram ( * P < 0.05), while gender was not an independent factor of prognosis. Tis diference may be afected by the signifcant gender diferences of BC patients in TCGA database and diferent incidence of BC between men and women. Te calibration curves showed that the nomogram has good consistence with practical outcomes (Figure 4(f )). To further assess the clinical applicability of the prognostic signature, we carried out stratifed analysis using various clinical variables. As shown in Figures 5(a)-5(n), in the subsets of age >65 and age <65, females and males, high grade, stages III-IV, T3-4, M0, N0, and N1-3, patients with a high-risk score had worse prognosis than those with a low-risk score (P < 0.05), demonstrating that the prognostic signature has good clinical ftness.

Pathway Enrichment Analysis.
We performed KEGG pathway analysis with GSEA software to analyze the pathway enrichment between the low-risk group and the high-risk group. As illustrated in Figure 6(a), the top 10 pathways enriched in the high-risk group were focal adhesion, melanoma, prostate cancer, regulation of actin cytoskeleton, Journal of Oncology 3 WNT signaling pathway, renal cell carcinoma, endometrial cancer, pathways in cancer, gap junction, and cardiomyopathy (P < 0.001, FDR < 0.25, |NES| > 2.0). Te result revealed that pathway enrichment in the high-risk group was signifcantly correlated with tumorigenesis and tumor invasion. In the low-risk group, no signifcant enrichment of pathways was observed (FDR > 0.25).

Exploration of Immune Infltration, TME, Immune
Checkpoints, and Necroptosis Genes. We used Spearman correlation analysis to evaluate the correlation between 22 types of common immune cells and risk scores, and the result is illustrated as a bubble chart in Figure 6(b). Te result revealed that the risk score was signifcantly negatively correlated with CD8 + T cells, plasma B cells, CD4 + T cells, and T cells (|cor| > 0.2, P < 0.001), while it was positively correlated with macrophages, cancer associated fbroblasts, and M0 macrophages endothelial cells (|cor| > 0.2, P < 0.001). By the ESTIMATE algorithm, we found higher stromal infltration in the high-risk group ( Figure 6(c), P � 0.0025). Te result of ssGSEA analysis also revealed higher concentrations of macrophages and mast cells and lower concentration of T2 cells in the high-risk group, indicating that there is a diferent TME between the two groups ( Figure 6(d)).

Diferences of Anticancer Drug Sensitivity between Two
Subsets. Tere were signifcant diferences in the responses to anticancer drugs between the high-risk group and the low-risk group (Figure 6(g)). Patients in the low-risk group were more sensitive to ABT.888 (veliparib), geftinib, methotrexate, and Nutlin-3a, while patients in the high-risk group were more sensitive to bexarotene, CGP.60474, docetaxe, embelin, imatinib, and pazopanib. Tis result may provide a reference for individualized drug therapy for BC patients.

Discussion
A number of research works have explored the signatures of prognosis and classifcation of BC with SUMOylation, ferroptosis, immune infltration-related lncRNA, and toll-like receptor 4 or pyroptosis-associated lncRNA [14,15]. However, there were only a few studies concentrated on the efect of necroptosis in BC at present. For example, researchers have found that the PKM2 (pyruvate kinase M2) inhibitor shikonin could kill the T24 cisplatin resistant cells by inducing necroptosis rather than apoptosis [16]. Another study in vitro revealed that ABT-737, a BCL-2 family inhibitor, could restrain the proliferation and invasion of bladder cancer cells by inducing necroptosis [17]. Up to now, there is no study of the NRlncRNAs in BC. In this study, we are the frst to identify the lncRNAs associated with necroptosis in BC and to construct a prognostic signature of NRlncRNAs. Further analysis demonstrated that the risk   Journal of Oncology 5 score of this signature is an independent prognostic factor and presents a favorable predictive ability of the outcome of the BC patients. Te prognostic signature of BC comprises 7 NRlncRNAs, namely, FKBP14-AS1, LINC02178, AL136084.2, AL731567.1, AC011503.2, LINC02195, and AC068196.1. Among these lncRNAs, LINC02195 has been illustrated as a regulator of MHC I (major histocompatibility complex class I) molecules, which plays a crucial role in the immunosurveillance in head and neck squamous cell carcinoma [18]. High expression of LINC02195 is positively correlated with an increased number of CD8 + and CD4 + T cells in the tumor microenvironment, profering a favorable prognosis in patients with head and neck squamous cell carcinoma. Interestingly, we discovered that LINC02195 was up-regulated in the low-risk group and correlated to a favorable prognosis of BC patients in this research, suggesting that LINC02195 may act as a tumor suppressor and plays an important role in the immunity of BC. Te regulation network revealed that LINC02195 positively regulates FASLG, which together with FAS initiates cell death and prevents tumor progression [19]. LINC02178 was identifed as an autophagy-related lncRNA and negatively correlated to the prognosis of BC in a previous study [20]. AC011503.2 was identifed as a glycolysis-related lncRNA and a protective factor of the prognosis of BC [21]. Our research also revealed that LINC02178 and AC011503.2 were upregulated and down-regulated in the high-risk group, indicating that these lncRNAs may play important roles in the cell death and metabolism of BC. Furthermore, AL731567.1,  Journal of Oncology AC011503.2, and AC068196.1 were found down-regulated in the high-risk group and positively regulated in BARF in this study. Te mutations of BARF have been demonstrated to promote many cancers such as melanoma, lung cancer, and colorectal cancer [22]. Te result also indicated that these lncRNAs may be involved in the regulation network of bladder cancer.
Te result of GSEA showed that the WNT signaling pathway, focal adhesion, and gap junction are enriched in the high-risk group. Numerous research works have demonstrated that aberrant activation of the WNT signaling pathway plays important roles in the pathological process of many cancer types, including BC [23,24]. Target genes of the WNT pathway induce the epithelial-mesenchymal transition       TNFRSF4  ICOSLG  PDCD1LG2  TNFSF9  CD27  TNFRSF25  PDCD1  TMIGD2  TNFRSF14  LGALS9  NRP1  CD276  CD40  BTNL2  TIGIT  CD160  TNFSF15 RIPK3  TRIM11  CASP8  IPMK  MYC  PANX1  CYLD  USP22  STAT3  DIABLO  DNMT1  BRAF  AXL  ID1  HAT1  SIRT2  PLK1  BACH2  GATA3  MYCN  ALK  TERT  RNF31  IDH1  KLF9  HDAC9  HSP90AA1  LEF1  CD40  EGFR  DDX58  TARDBP  APP  TNFRSF21 10.0 (EMT) process of cancers by involving in cell adhesion, leading to conversion of nontumorigenic cells into cancer stem cells. Besides, many genes in the WNT pathway were found associated with interferon signaling, profering that this pathway participates in the interferon-mediated immune responses of cancer cells. Terefore, the KEGG enrichment analysis also confrms that patients in the high-risk group have high malignance and worse prognosis. Immune infltration analysis indicated that the risk score was signifcantly inversely correlated with infltration of CD8 + T cells and CD4 + T cells and positively correlated with macrophages and cancer-associated fbroblasts. Prior research works have proven that CD8 + T cells and CD4 + T cells are the crucial defenders in the antitumor immunity [25][26][27]. Te dysfunction and exhaustion of cytotoxic T cells contribute to the immune related tolerance and immunosuppression during the progression of cancers. On the other hand, high infltration of macrophages and cancerassociated fbroblasts has been evidenced to have been associated with the tumor progression and poor outcomes of patients [28][29][30][31]. Te further analysis of immune checkpoints in this study revealed that the expressions of protective factors such as TNFRSF4, CD27, TNFRSF25, PDCD1, TNFRSF14, CD40, and TIGIT were signifcantly lower in the high-risk group, while expressions of cancer promotors such as PDCD1LG2, TNFSF9, NRP1, CD276, and CD44 were signifcantly higher in this group, also suggesting that the high-risk group has poor overall survival [32][33][34][35]. In addition, we discovered that the expression of necroptosis genes such as RIPK3, FADD, FAS, MYC, STAT3, PLK1, EGFR, CASP8, BRAF, ID1, GATA3, and CD40 were signifcantly diferent in the high-risk and low-risk groups. Abundant research works have confrmed that these genes play crucial roles in the oncogenesis, progression, and drug resistance of cancers [36][37][38][39]. Te expression of PLK1, GATA3, and CD40 were even related to the prognosis of bladder cancer [40][41][42]. Tis result indicated that necroptosis may partly lead to the diferences in the survival of the high-risk and low-risk group.
Te present study also has certain limitations. Firstly, only TCGA database was used for internal validation in this research, and more datasets are needed to further verify the capacity of the signature. Secondly, the prognostic signature was established based on the transcriptome RNA-seq and clinical data of BC patients form TCGA, and expression of lncRNAs in this signature needs more verifcation with clinical specimens from diferent centers. In addition, the functions and regulatory pathways of the NRlncRNAs in BC need further explorations with in vitro or in vivo experiments.

Conclusions
In this study, we constructed a prognostic signature for BC with 7 diferentially expressed NRlncRNAs. Tis signature can predict the overall survival of BC patients and provide a reference for the individualized treatment of BC. Further explorations are needed to defne the functions and pathways of NRlncRNAs in BC.

Data Availability
Te data used in this study are available in TCGA database (https://portal.gdc.cancer.gov/repository). Detailed data of (c) Comparison of TME between the low-risk group and the high-risk group. (d) Immune cells analysis between the two groups by ssGSEA scores. (e) Te comparison of immune checkpoint activation between the two risk groups. (f ) Te expression of necroptosis genes in two groups. (g) Te diferences of 10 common anticancer drug sensitivity between the two groups. * P < 0.05, * * P < 0.01, and * * * P < 0.001. the manuscript are available from the correspondence author.

Conflicts of Interest
Te authors declare that they have no conficts of interest.