Identification of prognostic signature in cancer based on DNA methylation interaction network

Background

The identification of prognostic biomarkers for cancer patients is essential for cancer research. These days, DNA methylation has been proved to be associated with cancer prognosis. However, there are few methods which identify the prognostic markers based on DNA methylation data systematically, especially considering the interaction among DNA methylation sites.

Methods

In this paper, we first evaluated the stabilities of microRNA, mRNA, and DNA methylation data in prognosis of cancer. After that, a rank-based method was applied to construct a DNA methylation interaction network. In this network, nodes with the largest degrees (10% of all the nodes) were selected as hubs. Cox regression was applied to select the hubs as prognostic signature. In this prognostic signature, DNA methylation levels of each DNA methylation site are correlated with the outcomes of cancer patients. After obtaining these prognostic genes, we performed the survival analysis in the training group and the test group to verify the reliability of these genes.

Results

We applied our method in three cancers (ovarian cancer, breast cancer and Glioblastoma Multiforme).In all the three cancers, there are more common ones of prognostic genes selected from different samples in DNA methylation data, compared with gene expression data and miRNA expression data, which indicates the DNA methylation data may be more stable in cancer prognosis. Power-law distribution fitting suggests that the DNA methylation interaction networks are scale-free. And the hubs selected from the three networks are all enriched by cancer related pathways. The gene signatures were obtained for the three cancers respectively, and survival analysis shows they can distinguish the outcomes of tumor patients in both the training data sets and test data sets, which outperformed the control signatures.

Conclusions

A computational method was proposed to construct DNA methylation interaction network and this network could be used to select prognostic signatures in cancer.

     

GHSR DNA hypermethylation is a new epigenetic biomarker for gastric adenocarcinoma and beyond

Aberrations of DNA methylation are early events in the development of tumors. In this study, we investigated the DNA methylation status of growth hormone secretagogue receptor (GHSR), a promising pan-cancer biomarker, in gastric cancer (GC). Initially, data sets from DNA methylation and gene expression studies available at Gene Expression Omnibus (GEO) were analyzed. Confirmation was done on primary tumor specimens and adjacent normal stomach tissue samples. Both analyses showed significant hypermethylation of GHSR. For further validation, The Cancer Genome Atlas data on stomach cancer was used. A receiver operating characteristic curve analysis yielded an area under the curve value of 0.85, corroborating its usefulness as a diagnostic marker. A genome-wide comethylation analysis revealed several correlated genes. CREB1 was found to act as an upstream regulator of this gene network. Furthermore, GHSR methylation was found to be a biomarker in several other tumor entities, namely cancers of the bladder, endometrium, esophagus, head and neck, liver, thyroid, kidney, and ovary. Our findings along with previous reports on other types of cancer suggest a high potential of GHSR gene methylation as a pan-cancer biomarker, which could be considered for liquid biopsy applications.     

Clinicopathological and prognostic value of lncRNA PANDAR expression in solid tumors: Evidence from a systematic review and meta-analysis

PANDAR (promoter of CDKN1A antisense DNA damage activated RNA) has been shown to be aberrantly expressed in many types of cancer. Considering conflicting data, the current study was aimed to assess its potential role as a prognostic marker in malignant tumors. A comprehensive literature search of PubMed, Medline, and Web of Science was performed to identify all eligible studies describing the use of PANDAR as a prognostic factor for different types of cancer. Data related to overall survival (OS) and clinicopathologic features were collected and analyzed. The pooled hazard ratio (HR) and odds radio (OR) with a 95% confidence interval (CI) were used to estimate associations. Ten original studies containing 1,231 patients were included. The results showed that in patients with cancer, high PANDAR expression is correlated with lymph node metastasis (LNM; OR = 2.57; 95% CI, 1.76–3.81; p < 0.001), tumor stage (OR = 2.90; 95% CI, 1.25–6.75; p = 0.013), and tumor size (OR = 1.79; 95% CI, 1.11–2.91; p = 0.018). However, sensitivity analysis further demonstrated a significant association between high PANDAR expression and OS, both in multivariate and univariate analysis models (pooled HR 2.01; 95% CI, 1.17–3.44 and pooled HR 2.62; 95% CI, 1.98–3.47, respectively), after omitting one study. These results suggested that PANDAR expression might be indicative of advanced disease and poor prognosis in patients with cancer. Further studies are necessary to determine the value of this risk stratification biomarker in clinical management of patients with cancer.     

LINE1 methylation levels associated with increased bladder cancer risk in pre-diagnostic blood DNA among US (PLCO) and European (ATBC) cohort study participants

Global methylation in blood DNA has been associated with bladder cancer risk in case-control studies, but has not been examined prospectively. We examined the association between LINE1 total percent 5-methylcytosine and bladder cancer risk using pre-diagnostic blood DNA from the United States-based, Prostate, Lung, Colorectal, Ovarian Cancer Screening Trial (PLCO) (299 cases/676 controls), and the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) cohort of Finnish male smokers (391 cases/778 controls). Logistic regression adjusted for age at blood draw, study center, pack-years of smoking, and sex was used to estimate odd ratios (ORs) and 95% confidence intervals (CIs) using study- and sex-specific methylation quartiles. In PLCO, higher, although non-significant, bladder cancer risks were observed for participants in the highest three quartiles (Q2–Q4) compared with the lowest quartile (Q1) (OR = 1.36, 95% CI: 0.96 -1.92). The association was stronger in males (Q2–Q4 vs. Q1 OR = 1.48, 95% CI: 1.00–2.20) and statistically significant among male smokers (Q2–Q4 vs. Q1 OR = 1.83, 95% CI: 1.14–2.95). No association was found among females or female smokers. Findings for male smokers were validated in ATBC (Q2–Q4 vs. Q1: OR = 2.31, 95% CI: 1.62–3.30) and a highly significant trend was observed (P = 8.7 × 10^?7). After determining that study data could be combined, pooled analysis of PLCO and ATBC male smokers (580 cases/1119 controls), ORs were significantly higher in Q2-Q4 compared with Q1 (OR = 2.03, 95% CI: 1.52–2.72), and a trend across quartiles was observed (P = 0.0001). These findings suggest that higher global methylation levels prior to diagnosis may increase bladder cancer risk, particularly among male smokers.     

A two-CpG-based prognostic signature for oral squamous cell carcinoma overall survival

Oral squamous cell carcinoma (OSCC) represents one of the most common head and neck cancer that with dire prognosis due partly to the lack of reliable prognostic biomarker. Here, we aimed to develop a CpG site–based prognostic signature through which we could accurately predict overall survival (OS) of patients with OSCC. We obtained OSCC-related DNA methylation and gene expression data sets from the public accessible Gene Expression Omnibus. Correlations between methylation level of CpG sites and OS of patients with OSCC were assessed by univariate Cox regression analysis followed by robust likelihood-based survival analysis on those CpG sites with permutation P < 0.05 for further screening the optimal CpG sites for OSCC OS prediction based on the risk score formula that composed of the methylation level of optimal CpG sites weighted by their regression coefficients. Besides, differential expression genes (DEGs) and differential methylation genes (DMGs) in OSCC samples compared with normal samples were obtained and shared genes were considered as vital genes in OSCC tumorgenesis and progression. As a result, two CpG sites including cg17892178 and cg17378966 that located in NID2 and IDO1, respectively, were identified as the optimal prognostic signatures for OSCC OS. In addition, 12 overlapping genes between DEGs and DMGs that closely associated with inflammation or blood and tissue development–related biological processes were obtained. In conclusions, this study should provide valuable signatures for OSCC diagnosis and treatment.     

A large-scale transcriptome analysis identified ELANE and PRTN3 as novel methylation prognostic signatures for clear cell renal cell carcinoma

DNA methylation was involved in the progress of many types of cancer including clear cell renal cell carcinomas (ccRCCs). This study aimed to identify the prognostic DNA methylation biomarkers for the ccRCCs by a large-scale RNA-seq analysis. The DNA methylation data and the corresponding clinical information of the patients with ccRCCs were extracted from TCGA database and randomly divided into the training group and the validation group. The differentially expressed CpG sites and the survival-related CpG sites were further identified, which was combined into CpG sites pair and followed by screening the survival-related pairs. The C-index and the forward search algorithms were constructed to identify the prognostic signatures for the patients with ccRCCs. The prognostic signatures were verified by the validation dataset and the protein–protein interactions (PPI) network analysis was performed on the CPG sites of the signature. A total of 9,861 differentially expressed CPG sites were identified and 567 CpG sites were found to relate to the overall survival (OS) of the patients with ccRCCs. Besides, 1,146 CPG sites pairs were found to be related to the OS of the ccRCCs samples and the signature composed of seven CpG sites pairs were obtained to predict the prognosis of patients with ccRCCs and the results were verified in the validation dataset. Besides, the PPI network analysis showed that ELANE and PRTN3 gene may be associated with the invasion and metastasis of ccRCCs and could function as potential prognostic and therapeutic signatures for ccRCCs.     

Systematic analysis of differentially methylated expressed genes and site-specific methylation as potential prognostic markers in head and neck cancer

Head and neck cancer (HNC) remains one of the most malignant tumors with a significantly high mortality. DNA methylation exerts a vital role in the prognosis of HNC. In this study, we try to screen abnormal differential methylation genes (DMGs) and pathways in Head–Neck Squamous Cell Carcinoma via integral bioinformatics analysis. Data of gene expression microarrays and gene methylation microarrays were obtained from the Cancer Genome Atlas database. Aberrant DMGs were identified by the R Limma package. We conducted the Cox regression analysis to select the prognostic aberrant DMGs and site-specific methylation. Five aberrant DMGs were recognized that significantly correlated with overall survival. The prognostic model was constructed based on five DMGs (PAX9, STK33, GPR150, INSM1, and EPHX3). The five DMG models acted as prognostic biomarkers for HNC. The area under the curve based on the five DMGs predicting 5-year survival is 0.665. Moreover, the correlation between the DMGs/site-specific methylation and gene expression was also explored. The findings demonstrated that the five DMGs can be used as independent prognostic biomarkers for predicting the prognosis of patients with HNC. Our study might lay the groundwork for further mechanism exploration in HNC and may help identify diagnostic biomarkers for early stage HNC.     

Prognostic Significance of Promoter DNA Hypermethylation of cysteine dioxygenase 1 (CDO1) Gene in Primary Breast Cancer

Using pharmacological unmasking microarray, we identified promoter DNA methylation of cysteine dioxygenase 1 (CDO1) gene in human cancer. In this study, we assessed the clinicopathological significance of CDO1 methylation in primary breast cancer (BC) with no prior chemotherapy. The CDO1 DNA methylation was quantified by TaqMan methylation specific PCR (Q-MSP) in 7 BC cell lines and 172 primary BC patients with no prior chemotherapy. Promoter DNA of the CDO1 gene was hypermethylated in 6 BC cell lines except SK-BR3, and CDO1 gene expression was all silenced at mRNA level in the 7 BC cell lines. Quantification of CDO1 methylation was developed using Q-MSP, and assessed in primary BC. Among the clinicopathologic factors, CDO1 methylation level was not statistically significantly associated with any prognostic factors. The log-rank plot analysis elucidated that the higher methylation the tumors harbored, the poorer prognosis the patients exhibited. Using the median value of 58.0 as a cut-off one, disease specific survival in BC patients with CDO1 hypermethylation showed significantly poorer prognosis than those with hypomethylation (p = 0.004). Multivariate Cox proportional hazards model identified that CDO1 hypermethylation was prognostic factor as well as Ki-67 and hormone receptor status. The most intriguingly, CDO1 hypermethylation was of robust prognostic relevance in triple negative BC (p = 0.007). Promoter DNA methylation of CDO1 gene was robust prognostic indicator in primary BC patients with no prior chemotherapy. Prognostic relevance of the CDO1 promoter DNA methylation is worthy of being paid attention in triple negative BC cancer.     

Clinical and prognostic effects of CDKN2A,CDKN2B and CDH13 promoter methylation in ovarian cancer: a study using meta-analysis and TCGA data

Abstract

Background: Promoter methylation of tumour suppressor genes (TSGs) CDKN2A, CDKN2B and CDH13 has been reported in ovarian cancer. However, the clinicopathological characteristics and prognostic role of CDKN2A, CDKN2B and CDH13 promoter methylation in ovarian carcinoma remained unclear.

Methods: The pooled odds ratio (OR) or hazard ratios (HRs) with their 95% confidence intervals (95% CIs) were calculated in this meta-analysis. The Cancer Genome Atlas data were obtained to confirm the role of CDKN2A, CDKN2B and CDH13 methylation in ovarian cancer.

Results: CDKN2A, CDKN2B and CDH13 promoter methylation was higher in ovarian cancer than in normal ovarian tissues. CDH13 promoter methylation was correlated with tumour histology (serous vs. non-serous type: OR?=?0.33, p?=?0.031). CDKN2A promoter methylation was not linked to overall survival (OS), but it was correlated with a poor prognosis in progression-free survival (HR?=?1.55, p?=?0.004). TCGA data showed no correlation between CDKN2A, CDKN2B and CDH13 methylation and OS as well as disease-free survival (DFS).

Conclusions: CDKN2A, CDKN2B and CDH13 promoter methylation may correlate with the increased risk of ovarian cancer. CDKN2A promoter methylation may be an independent prognostic biomarker for predicting progression-free survival.     

ABCC9, NKAPL,and TMEM132C are potential diagnostic and prognostic markers in triple‐negative breast cancer

Triple‐negative breast cancer (TNBC) is a highly heterogeneous disease. The aim of this study is to identify the diagnostic and poor prognostic signatures in TNBC by exploring the aberrant DNA methylation and gene expression. Differential expression and methylation analysis of the TNBC and paracancer samples from The Cancer Genome Atlas were performed. Gene set enrichment and protein–protein interaction (PPI) network analysis was used to explore the mechanisms of TNBC. Methylation‐gene expression correlation analysis was performed, and multivariate Cox analysis and receiver operating characteristics analysis were used to further screen the hub genes for TNBC. We identified 1,525 differentially expressed genes and 150 differentially methylated genes between TNBC and paracancer samples. About 96.64% of the methylation sites were located on the CpG island. A total of 17 Gene Ontology biological process terms and 18 signal pathways were significantly enriched. GNG4, GNG11, PENK, MAOA, and AOX1 were identified as the core genes of the PPI network. Methylation‐expression correlations revealed that ABCC9 (cg06951626), NKAPL (cg18675097, cg01031101, and cg17384889), and TMEM132C (cg03530754) showed promise as diagnostic and prognostic markers in TNBC. ABCC9 (cg06951626), NKAPL (cg18675097, cg01031101, and cg17384889), and TMEM132C (cg03530754) were potential diagnostic and prognostic markers in TNBC.     

Identification of DNA methylation signature to predict prognosis in gastric adenocarcinoma

Gastric adenocarcinoma is an important death-related cancer. To find factors related to survival and prognosis, and thus improve recovery prospects, a powerful signature is needed. DNA methylation plays an important role in gastric adenocarcinoma processes and development, and here we report on the search for a significant DNA methylation gene to aid with the earlier diagnosis of gastric adenocarcinoma patients. A Cox proportional risk regression analysis and random survival forest algorithm were used to analyze gastric adenocarcinoma patients’ DNA methylation data from The Cancer Genome Atlas, a public database. DNA methylation gene signature consisting of five genes (SERPINA3, AP000357.4, GZMA, AC004702.2, and GREB1L) were selected. As the most accurate predictor, the area under the curve in the training and test group were 0.72 and 0.61, respectively. The signature was able to sort patients into high- and low-risk groups with meaningful overall survival rates (median: 18.36 vs 72.23 months, log-rank test, P < 0.001) in the training group, which predictive ability was validated in a test data set (median: 25.56 vs 58.80 months, log-rank test, P < 0.016). A multivariate Cox regression analysis showed the significant DNA methylation was an independent prediction prognostic factor for gastric adenocarcinoma patients. Functional analysis suggests that these signature genes may be related to pathways and biological processes associated with tumorigenesis. The significant DNA methylation gene could be a novel prediction and prognostic biomarker that both aids in the treatment and predicts the overall survival likelihoods of gastric adenocarcinoma patients.     

Methylation markers identify high risk patients in IGHV mutated chronic lymphocytic leukemia

Chronic lymphocytic leukemia (CLL) exhibits a very variable clinical course. Altered DNA methylation of genes has shown promise as a source of novel prognostic makers in a number of cancers. Here we have studied the potential utility of a panel of methylation markers (CD38, HOXA4 and BTG4) in 118 CLL patients. Each of the three loci assessed exhibited frequent methylation, as determined by COBRA analysis, and individually correlated with either good (CD38, BTG4 methylation) or poor (HOXA4 methylation) prognosis. Using a combined approach to produce an overall methylation score, we found that methylation score was significantly associated with time to first treatment in CLL patients. Multivariate Cox regression analysis revealed that methylation score was the strongest predictor of time to first treatment, and was independent of IGHV gene mutational status and CD38 expression. This study provides proof of principle that a panel of methylation markers can be used for additional risk stratification of CLL patients.     

Epigenetic changes as prognostic predictors in endometrial carcinomas

Endometrial carcinoma is one of the most frequent gynecological malignancies of the female. The diagnostic and prognostic markers for the high-risk subgroups with unfavorable prognosis are under intense debate worldwide, and, therefore, the aim of this study was to identify new potential DNA methylation markers for the high-risk groups. We used the Illumina Infinium HumanMethylation450 BeadChip to analyze the DNA methylation pattern and investigated its association with clinicopathological features important for defining the high-risk (FIGO-grade 3) and low-risk (FIGO-grade 1) groups of patients with endometrial cancer (n = 31 and n = 39, respectively). We identified specific DNA methylation signature in high-risk endometrial tumors, and potential molecular biomarker genes (TBX2, CHST11, and NID2) associated with unfavorable clinical predictive and prognostic factors.     

DNA methylation biomarkers for head and neck squamous cell carcinoma

DNA methylation plays an important role in the etiology and pathogenesis of head and neck squamous cell carcinoma (HNSCC). The current study aimed to identify aberrantly methylated-differentially expressed genes (DEGs) by a comprehensive bioinformatics analysis. In addition, we screened for DEGs affected by DNA methylation modification and further investigated their prognostic values for HNSCC. We included microarray data of DNA methylation (GSE25093 and GSE33202) and gene expression (GSE23036 and GSE58911) from Gene Expression Omnibus. Aberrantly methylated-DEGs were analyzed with R software. The Cancer Genome Atlas (TCGA) RNA sequencing and DNA methylation (Illumina HumanMethylation450) databases were utilized for validation. In total, 27 aberrantly methylated genes accompanied by altered expression were identified. After confirmation by The Cancer Genome Atlas (TCGA) database, 2 hypermethylated-low-expression genes (FAM135B and ZNF610) and 2 hypomethylated-high-expression genes (HOXA9 and DCC) were identified. A receiver operating characteristic (ROC) curve confirmed the diagnostic value of these four methylated genes for HNSCC. Multivariate Cox proportional hazards analysis showed that FAM135B methylation was a favorable independent prognostic biomarker for overall survival of HNSCC patients.     

Biomarkers for detection and prognosis of breast cancer identified by a functional hypermethylome screen

Breast cancer (BC) is a disease with diverse tumor heterogeneity, which challenges conventional approaches to develop biomarkers for early detection and prognosis. To identify effective biomarkers, we performed a genome-wide screen for functional methylation changes in BC, i.e., genes silenced by promoter hypermethylation, using a functionally proven gene expression approach. A subset of candidate hypermethylated genes were validated in primary BCs and tested as markers for detection and prognosis prediction of BC. We identified 33 cancer specific methylated genes and, among these, two categories of genes: (1) highly frequent methylated genes that detect early stages of BC. Within that category, we have identified the combination of NDRG2 and HOXD1 as the most sensitive (94%) and specific (90%) gene combination for detection of BC; (2) genes that show stage dependent methylation frequency pattern, which are candidates to help delineate BC prognostic signatures. For this category, we found that methylation of CDO1, CKM, CRIP1, KL and TAC1 correlated with clinical prognostic variables and was a significant prognosticator for poor overall survival in BC patients. CKM [Hazard ratio (HR) = 2.68] and TAC1 (HR = 7.73) were the strongest single markers and the combination of both (TAC1 and CKM) was associated with poor overall survival independent of age and stage in our training (HR = 1.92) and validation cohort (HR = 2.87). Our study demonstrates an efficient method to utilize functional methylation changes in BC for the development of effective biomarkers for detection and prognosis prediction of BC.     

Analysis of methylation-driven genes for predicting the prognosis of patients with head and neck squamous cell carcinoma

To help provide evidence for prognosis prediction and personalized targeted therapy for patients with head and neck squamous cell carcinoma (HNSCC), we investigated prognosis-specific methylation-driven genes in HNSCC. Survival time data, RNA sequencing data, and methylation data for HNSCC patients were downloaded from The Cancer Genome Atlas. The MethylMix R package based on the β mixture model was utilized to screen genes with different methylation statuses in tumor tissues and adjacent normal tissues, and a total of 182 HNSCC-related methylation-driven genes were then identified. A survival prediction scoring model based on multivariate Cox analysis was developed to screen the genes related to the prognosis of HNSCC, and a linear risk model of the methylation status of six genes (INA, LINC01354, TSPYL4, MAGEB2, EPHX3, and ZNF134) was constructed. The prognostic values of the six genes were further independently explored by survival analysis combined with methylation and gene expression analyses. The 5-year survival rate in the high-risk group of patients in the test set was 30.4% (95% CI: 22.7%-40.8%) and that in the low-risk group of patients was 65.5% (95% CI: 56.1%-76.5%). The area under the receiver operating characteristic curve for the model was 0.723, which further verified the specificity and sensitivity of the model. In addition, subsequent combined survival analysis revealed that all six genes could be used as independent prognostic markers and thus might be potential drug targets. The innovative method provides new insight into the molecular mechanism and prognosis of HNSCC.