Background
Bladder cancer (BC) is known as the ninth most frequent malignancy around the world. Because of high rate of recurrence, it has been one of the most expensive solid cancer once the patients seed for continued surveillance [
1]. Further, the high recurrence rate of BC is partly owning to the insufficient number of prognosis-related biomarkers. Thus, to find out more favorable biomarkers for early detection and prognosis prediction of bladder cancer is becoming more critical.
The long non-coding RNAs (lncRNAs) are characterized as RNA transcripts > 200 bases, but they cannot translate into proteins [
2,
3]. Currently, plenty of studies have indicated that lncRNAs participate in diverse biological processes, such as cell proliferation [
4], differentiation [
5], chromatin modification, and so on [
6]. The lncRNAs have been revealed playing critical roles in tumorigenesis and progression. Referring to previous studies, a series of lncRNAs, such as MALAT1, TUG1, H19, played a critical role in predicting the prognosis, the risk of metastasis [
7‐
11], as well as drug treatment resistance [
12]. Mechanisms studies found the function of lncRNA is partly depended on its location. For example, the lncRNA-HOTAIR, located in the nuclear, regulates the gene expression through recruiting the chromatin modifiers on the promoter region of transcriptional factors, enhancing their transcriptional activities. For these lncRNAs located in the cytoplasm, regulates the protein expression mostly through post-transcription, such as influencing the protein stability.
Besides, the lncRNAs could also serve as competing endogenous RNAs to influence the expression of targeted genes by competing with the microRNAs (miRNAs). In addition to the functional role of lncRNAs, many other studies have highlighted their role in predicting the prognosis of cancer patients. For example, Yang et al. [
13] identified a six-lncRNA-based signature, which can predict the recurrence risk of ovarian cancer. In Song et al.’s [
14] work, they established a lncRNA-based signature that provides the prognostic value for outcomes of gastric cancer patients. Similar studies were also conducted for thyroid papillary carcinoma [
15], pancreatic cancer [
16] and esophageal squamous cell carcinoma [
17], and revealed satisfied outcomes. Although few studies have tried to construct gene-based signature for bladder cancer patients, there is no study specifically focus on muscle-invasive BC patients, who mostly have poor survival outcome than these patients with in situ tumors.
Here, we performed a systematic screening of lncRNAs that related to the recurrence-free survival (RFS) of muscle-invasive BC patients, and established a fourteen-lncRNA-based signature. Our study provides the tool for clinicians to develop the personalized medicine for muscle-invasive BC patients.
Methods
Bladder cancer datasets and patient information
Clinical information, particularly for RFS (status and follow-up time), and the RNA expression profile for lncRNAs in invasive BC tissues were directly download from TCGA database (
https://cancergenome.nih.gov/). A total of 320 patients were diagnosed as muscle-invasive BC, and then they were divided into the training and validating groups (224 vs. 96), randomly.
Signature construction and risk stratification
The lncRNAs were subjected to the univariable Cox regression analysis to find out lncRNAs that correlated with the RFS of muscle-invasive BC patients. Then, the LASSO Cox regression analysis was conducted to establish the risk signature based on these RFS-related lncRNA candidates. The risk value of each patient was depended on the expression of lncRNAs, and their matched co-efficient. The formula of our signature was presented as below:
$$ Risk\ score={\beta}_{\mathrm{lncRNA}1}\times {expr}_{\mathrm{lncRNA}1}+{\beta}_{\mathrm{lncRNA}2}\times {expr}_{\mathrm{lncRNA}2}+\dots +{\beta}_{\mathrm{lncRNA}\mathrm{n}}\times {\beta}_{\mathrm{lncRNA}\mathrm{n}} $$
The patients in both the training and validation cohorts were ranked by the risk scores, and then they were classified into high- or low-risk group according to the risk score of each sample (high-risk: risk score > 0; low-risk: risk score < 0).
Kaplan-Meier (K-M) and receiving operating curve (ROC)
The K-M analysis was applied to find out the RFS difference between the low- and high-risk groups, and the ROC curve was used to determine the prognostic value of our signature in both the training and validation cohorts. Hence, the multivariate analyses were executed to determine the independent effect of our signature with clinicopathological features (such as sex, age, tumor grade and tumor stage). All these analyses were conducted based on R software (version 3.5.2) with the following R packages: ‘glmnet’, ‘survivalROC’ and ‘ggplot2’. Besides, the P-value less than 0.05 was considered as statistically significant.
Differential expression and functional analysis
Differential expression analysis was performed by DESeq2 [
18], and differentially expressed genes (DEGs) were defined as fold change more than 1 and adjusted
p-value less than 0.05. Functional categories of genes were analyzed using Proteomap (
https://www.proteomaps.net/).
Discussion
The disease progression of muscle-invasive BC patients is dependent on many risk factors including phenotypes and genotypes. However, clinical criteria such as age, gender, pathological TNM stage and tumor grade may not reflect the entire biology of muscle-invasive BC. Here, we investigated the efficacy of the 14-lncRNA-based gene signature to predict the RFS of muscle-invasive BC patients. Despite previously developed gene signature-based prognostic models, it is still valuable to update new models to improve the management of muscle-invasive BC. An effective gene signature could guide patient counseling and help people to identify candidates who need more aggressive management. We demonstrated that this model has more prediction power than independent traditional clinical features. Although it is lack of novelty and function work in our study and our results require further investigation of the efficacy of the 14-lncRNA-based signature panel in patients, this panel could be extremely beneficial to identify patients at elevated risk of recurrence that may require adjuvant therapy.
We identified a set of 14 lncRNAs that showed differential expressions between high-and low-risk cancer patients included in the data sets (Fig. S
2). Such differentiations signified their potential roles in carcinogenesis. Recent researches has found that lncRNA LOC554202 is significantly downregulated in bladder cancer tissues compared with adjacent noncancerous tissues, and lncRNA LOC554202 expression level in bladder cancer patients was negatively associated with advanced TNM stage [
19]. SNHG10 is known to be over-expressed in hepatocellular carcinoma, and we found it facilitates hepatocarcinogenesis and metastasis [
20]. SOX2 overlapping transcript mainly play crucial role in tumor initiation and/or progression as well as regulating pluripotent state of stem cells [
21]. CACNA2D1 is most the most extensively investigated and validated of these markers. A retrospective study showed that positive expression of Cacna2d1 was significantly associated with advanced FIGO stage (
P < 0.001), histological subtype (
P = 0.017) and tumor differentiation (
P = 0.015) [
22]. Data coming from Sui et al.’s research [
23] have confirmed that radio-resistance of non-small cell lung cancer induced by CaCna2D1. The roles of the rest of the lncRNA genes identified in bladder cancer remain unclear.
Conclusion
By applying public TCGA data, we successfully built and validated a RFS prediction model of muscle-invasive BC based on a novel 14-lncRNA signature. Comparing with independent clinical features, this model has more efficiency to predict RFS of muscle-invasive BC. The model may help facilitate doctor-patient consultations, guide muscle-invasive BC treatment strategy and eventually benefit patients.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.