Four methylation‐driven genes may be prognostic biomarkers in clear cell renal carcinoma

Hang Yin, Haiyang Zhang, Xiaoyuan Wang, Qingyong XuClinical 2020 · DOI

TL;DR

This study identifies four genes (LAT, KRT17, MAGEC2, and RNASE2) whose methylation and expression patterns can predict the survival of clear cell renal carcinoma (ccRCC) patients. These genes offer potential as prognostic biomarkers, improving our ability to stratify patients and guide treatment decisions.

What Problem Does This Paper Address?

Clear cell renal carcinoma (ccRCC) is a common cancer with a significant 5-year mortality rate (44-69%). Currently, there's a pressing need for reliable biomarkers to predict patient survival and guide clinical management. Epigenetic modifications, particularly DNA methylation, play a crucial role in cancer development by altering gene expression. While large-scale genomic data is available, pinpointing specific methylation-driven genes that are both aberrantly methylated and differentially expressed, and then validating their prognostic value in ccRCC, remains a challenge.

Methods

The researchers utilized a multi-step approach to identify and validate prognostic biomarkers. They began by analyzing gene expression and methylation data from ccRCC patients, identifying differentially expressed genes using a fold-change threshold and false discovery rate. Aberrant methylation was assessed using the MethylMix package. They then focused on "methylation-driven" genes, which showed both altered methylation and expression. Functional enrichment analysis (DAVID, KEGG pathways) was performed to understand the biological roles of these genes. Univariate and multivariate Cox regression analyses were employed to identify genes significantly associated with overall survival. Finally, a risk score based on the expression of the identified prognostic genes was developed and validated using Kaplan-Meier curves and ROC analysis. To further explore the methylation-expression link, they treated ccRCC cell lines with 5-azacytidine (5-Aza), a DNA methylation inhibitor, and observed changes in gene expression.

Datasets

  • TCGA (The Cancer Genome Atlas) database: Used for obtaining clinical data, mRNA expression data, and methylation data (Human Methylation 450k Illumina Infinium methylation arrays) for ccRCC patients.
  • 524 patients with complete clinical and mRNA data (539 cancer, 72 para-carcinoma tissues).
  • 160 normal and 325 tumor samples for methylation data.
  • ccRCC cell lines (OSRC2 and 7860): Used for experimental validation of the methylation-expression relationship.

Key Insights and Ideas

  • Methylation as a Driver of Prognosis: The study strongly suggests that aberrant DNA methylation, leading to altered gene expression, is a key mechanism influencing ccRCC patient outcomes.
  • Identification of a Four-Gene Signature: The discovery of a specific set of four genes (LAT, KRT17, MAGEC2, and RNASE2) that are consistently associated with survival is a significant finding.
  • Functional Relevance: The identified genes are involved in important cellular processes like cell periphery, platelet activation, immune response, and cell-cell junctions, hinting at their roles in tumor progression and metastasis.
  • Experimental Validation of Methylation Inhibition: The cell line experiments provide direct evidence that inhibiting DNA methylation can restore the expression of these prognostic genes, reinforcing their link to epigenetic regulation.

Key Results

  • 51 Abnormally Methylated and Differentially Expressed Genes: Identified in ccRCC.
  • Eight Genes Correlated with Overall Survival: Found through univariate Cox analysis.
  • Four Independent Prognostic Biomarkers: LAT, KRT17, MAGEC2, and RNASE2 were identified as independent risk factors for survival in multivariate Cox regression (P < .001).
  • 4-Gene Expression-Based Survival Risk Score: Developed with a calculated score of 0.991.
  • Survival Stratification: Patients with high-risk scores had significantly shorter survival times (5-year survival: 50.7% vs. 75.6% in the training set; 52.7% vs. 86.2% in the testing set).
  • Prognostic Accuracy: The 4-gene signature achieved an AUC of 0.708 in the training set and 0.732 in the testing set, indicating good predictive power.
  • Correlation with Methylation: Expression of LAT, MAGEC2, and RNASE2 was associated with methylation levels at multiple sites.
  • 5-Aza Treatment Effect: Treatment with 5-Aza significantly increased the expression of LAT, KRT17, MAGEC2, and RNASE2 in ccRCC cell lines.

Limitations and Future Work

  • Limitations: The study primarily relies on retrospective data from TCGA. While validated in a testing set, further prospective studies are needed to confirm these findings in diverse patient populations. The exact biological mechanisms by which these four genes influence ccRCC progression are not fully elucidated.
  • Future Work: The authors suggest that these four genes hold potential as prognostic biomarkers for ccRCC patients. Future research could focus on elucidating the precise molecular pathways through which these genes exert their influence on tumor development and progression, and explore their potential as therapeutic targets. Further validation in larger, prospective clinical trials is warranted before clinical implementation.

Summary updated

Abstract

Dear Editor, Clear cell renal carcinoma (ccRCC) is one of the most common urinary system carcinomas with a 5-year survival about 44–69%.1 There is an urgent need to explore effective biomarkers to predict the survival of patients with ccRCC. Epigenetic mechanisms as an important part in the development of tumorigenesis mainly include heritable and reversible DNA methylation. DNA methylation could negatively regulate gene expression in the development of tumors, and be also used as biomarkers for prognostic prediction. It is important and urgent to explore the abnormity of DNA methylation-associated genes. Recently, there are more and more studies based on the Cancer Genome Atlas (TCGA) database with large amounts of genetic and clinical profiles.2 In addition, MethylMix is used to identify methylation status according to the β model.3 The present study aims at understanding gene expression and methylation as well as the association with clinical information in patients with ccRCC and screening the methylation gene-based biomarkers for prognostic prediction. We obtained 524 complete clinical data and mRNA data of patients with ccRCC (539 cancer and 72 para-carcinoma tissues). Note that 2332 differential expression genes were considered as discriminatively expressed (the absolute value of fold changes is > 2, and the value of false discovery rate is < 0.01). The methylated data were obtained from TCGA and Human Methylation 450k Illumina Infinium methylation arrays. The methylated data included 160 normal samples and 325 tumor samples. The aberrant methylation and expression of genes in ccRCC were explored by LIMMA package. Then the aberrant genes were further explored by the MethylMix package (|logFC| > 0, P < .05, |Cor| > 0.3). We also explored 168 methylation-driven genes (Figure 1A,B), of which 75 genes (44.6%) were hypermethylated and the other genes were hypomethylated. The expression of genes with significant difference in tumor samples as compared with the normal were investigated, of which 51 abnormally methylated and differentially expressed genes in ccRCC (Figure 1C). We explored the function of these genes by the Discovery 6.8 (DAVID) (https://david.ncifcrf.gov/) and constructed KEGG pathways using ConsensusPathDB. The results showed the function of these genes were cell periphery, platelet activation, immune response, cell-cell junction, and others. The KEGG pathways of these genes were rheumatoid arthritis, platelet activation, hepatitis C, and so on (Figure 1D,E). To analyze the function of abnormally methylated and differentially expressed genes, we calculated the relationship between the 51 genes and overall survival in the training set. The results of univariate cox analysis showed that eight genes were correlated with overall survival (P < .05). Of eight genes in multivariate Cox regression, four genes including Linker for Activation of T Cells (LAT), Keratin 17 (KRT17), MAGE Family Member C2 (MAGEC2), and Ribonuclease A Family Member 2 (RNASE2) were independent risk factors effecting existence (P < .001; Table S1). The 4-gene expression-based survival risk score was calculated as 0.991. We also compared the survival rate between patients with low and high risks according to the risk score and the Kaplan-Meier curves. The patients with high risk had shorter survival time than those with low risk (P < .001). The 5-year survival times of patients with low and high risks were 75.6 (95% CI = 0.67--0.825) and 50.7 (95% CI = 0.4115--0.625), respectively. The area under ROC curve (AUC) was 0.708 (Figure 1F). We further evaluated the value of the four target genes in the prognostic significance. The 5-year survival time of patients with high risk (n = 168; 52.7 (95% CI = 0.446--0.624)) was shorter than those with low risk (n = 97; 86.2 (95% CI = 0.789--0.942), P < .001), respectively. The AUC was 0.732 (Figure 2A). Figure 2B demonstrated the result of the four genes in the entire group. The correlation among gene methylation and expression with patient survival using the survival package of R. KRT17, LAT, MAGEC2, and RNASE2 were significantly correlated with overall survival (P < .05, Figure 2C). The correlations between methylation sites and gene expressions were also analyzed. The absolute value of Cor > 0.2 was the cut off value. The expression of LAT, MAGEC2, and RNASE2 were associated with the methylation level of multiple sites (Figure S10). To explore the correlations between methylation and expression, OSRC2 and 786O cells of ccRCC were treated by 5-Aza (8 µM) to explore the relationship between the blocking of promoter in methylation and expression. The expression of LAT, KRT17, MAGEC2, and RNASE2 genes increased significantly in both cells after the treatment, as compared with P16 and CST6 as controls (Figure 2D). In conclusion, our study explored the methylation-driven and differently expressed genes in ccRCC and found four genes with potentials of prognostic biomarkers for ccRCC patients. This work was supported by Haiyan Foundation of Harbin Medical University Cancer Hospital (No. JJZD2018-04 to Q.Y.X.), Hei Long Jiang Postdoctoral (LBH -Z18224 to X.Y.W.), China Postdoctoral Science Foundation (2018M641861 to X.Y.W.), Fundamental Research Funds for the Provincial Universities in Heilongjiang (2017LCZX94 to X.Y.W.), and National Natural Science Fund of China (No. 81903120 to H.Y. and 81902317 to X.Y.W.). The authors declare no potential conflict of interest. H. Yin and X.Y. Wang performed experiments. H. Yin and H.Y. Zhang wrote the manuscript. Q.Y. Xu designed the study. All authors approved the final manuscript. All the data and materials are available upon request from the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
BibTeX
@inproceedings{yin2020four,
  author = {Hang Yin and Haiyang Zhang and Xiaoyuan Wang and Qingyong Xu},
  title = {{Four methylation‐driven genes may be prognostic biomarkers in clear cell renal carcinoma}},
  booktitle = {Clinical},
  year = {2020},
  doi = {10.1002/ctm2.45},
}
Figure