Chen et al. Cancer Cell Int (2019) 19:185 https://doi.org/10.1186/s12935-019-0900-4 Cancer Cell International PRIMARY RESEARCH Open Access DNA methylation-based classifcation and identifcation of renal cell carcinoma prognosis-subgroups Wenbiao Chen1, Jia Zhuang3, Peizhong Peter Wang4, Jingjing Jiang1, Chenhong Lin1, Ping Zeng1, Yan Liang1, Xujun Zhang1, Yong Dai2* and Hongyan Diao1* Abstract Background: Renal cell carcinoma (RCC) is the most common kidney cancer and includes several molecular and histological subtypes with diferent clinical characteristics. The combination of DNA methylation and gene expression data can improve the classifcation of tumor heterogeneity, by incorporating diferences at the epigenetic level and clinical features. Methods: In this study, we identifed the prognostic methylation and constructed specifc prognosis-subgroups based on the DNA methylation spectrum of RCC from the TCGA database. Results: Signifcant diferences in DNA methylation profles among the seven subgroups were revealed by consistent clustering using 3389 CpGs that indicated that were signifcant diferences in prognosis. The specifc DNA methylation patterns refected diferentially in the clinical index, including TNM classifcation, pathological grade, clinical stage, and age. In addition, 437 CpGs corresponding to 477 genes of 151 samples were identifed as specifc hyper/hypometh- ylation sites for each specifc subgroup. A total of 277 and 212 genes corresponding to DNA methylation at promoter sites were enriched in transcription factor of GKLF and RREB-1, respectively. Finally, Bayesian network classifer with specifc methylation sites was constructed and was used to verify the test set of prognoses into DNA methylation subgroups, which was found to be consistent with the classifcation results of the train set. DNA methylation-based classifcation can be used to identify the distinct subtypes of renal cell carcinoma. Conclusions: This study shows that DNA methylation-based classifcation is highly relevant for future diagnosis and treatment of renal cell carcinoma as it identifes the prognostic value of each epigenetic subtype. Keywords: Renal cell carcinoma, DNA methylation, Prognosis subgroups, Molecular subtypes Background due to the advances in ultrasound, computed tomogra- Renal cell carcinoma is the most lethal urologic malig- phy and magnetic resonance imaging used for the early nancies and accounts for 3% of all malignant tumors. In detection of renal tumors. Although recent molecular recent decades, the incidence of RCC has been rising targeting therapy has improved the overall survival of RCC patients, long-term prognosis remains precarious [1, 2]. Previous, pathological assessments determined the *Correspondence: [email protected]; [email protected] risk of recurrence and diferentiation in RCC patients, 1 State Key Laboratory for Diagnosis and Treatment of Infectious Diseases, Collaborative Innovation Center for Diagnosis and Treatment of Infectious thus guiding the diagnosis and treatment of recurrent Diseases, The First Afliated Hospital, College of Medicine, Zhejiang diseases. Te classifcation and diagnostic criteria of RCC University, 79 Qing Chun Road, Hangzhou, China are indeterminate and have a high interobserver bias. As 2 Clinical Medical Research Center, Shenzhen People’s Hospital, The Second Clinical Medical College of Jinan University, 1017 Dongmen a result, 30% of patients develop metastatic disease after North Road, Luohu District, Shenzhen, Guangdong, China treatment, with a median survival period of 1 year [3]. Full list of author information is available at the end of the article Tere are growing pieces of evidence that even on the © The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Chen et al. Cancer Cell Int (2019) 19:185 Page 2 of 14 basis of the same RCC histological subtype, there is an island methylation. CpG-rich regions (CpG islands) are inconsistent representation of sets of histologically and found in about 40–50% of promoter regions of human molecularly heterogeneous diseases [4]. With further genes. Abnormal CpG island hypermethylation of understanding of the morphology, immunohistochemis- tumor suppressor genes and hypomethylation of onco- try, molecular and epidemiological features of RCC, the genes are vital in carcinogenesis [12, 13]. Terefore, as World Health Organization (WHO) revised the classif- a promising molecular marker of RCC, abnormal DNA cation of renal cell tumors in 2016. It mainly focuses on methylation appears in early detection, prognosis pre- the molecular pathological features of renal cell tumors diction, molecular classifcation and therapeutic targets [5]. To this end, a more detailed and comprehensive [14]. Previous studies have shown that the increase in understanding of the molecular classifcation character- promoter hypermethylation frequency is associated istics of RCC is required. with higher stage and grade in clear cell RCC, accord- Currently, the analysis of molecular characteristics ing to Te Cancer Genome Atlas (TCGA) Research has shown that the RCC has obvious specifcity. Since Network for cancer genome maps [15]. Ana et al. iden- molecular subtype supplements histopathology, invalua- tifed groups such as OXR, MST1R and HOXA9 pro- ble insights can be drawn from integrated classifcations, moter methylation that positively identifed renal cell which could refne and impact the future application of tumors and diferentiation of subtypes. In addition, the precision-focused, personalized clinical management of groups may improve risk stratifcation in patients with renal cell carcinoma [6]. Ricketts et al. performed a com- small renal masses, helping clinicians to determine the prehensive genomic and phenotypic analysis on RCC best treatment strategy [16]. Gabriel et al. found that in order to identify the common and specifc molecular the DNA methylation profles of RCC subtypes can be characteristics and laid the foundation for the develop- divided into two main epigenetic clusters: one consists ment of subtype-specifc treatment and management of clear-cell RCC, papillary RCC, mucinous and spin- strategies for these cancer patients [7]. Wu et al. used dle cell carcinomas and translocation RCC; the other an unsupervised Consensus Clustering algorithm to includes oncocytoma and chromophobe RCC [17]. identify three distinct molecular subtypes of clear renal Moreover, an epigenetic chart of the 56 genes that iden- clear cell carcinoma (the most common subtype) on the tify the ontogeny of renal cells can predict the results basis of hierarchical clustering. Subtypes with poor prog- of clear cell RCC. However, their classifcation may not nosis were irregularly up-regulated in focal adhesion be detailed enough and has not yet been fully tested for and cytoskeleton-related pathways, and the expression specifc sites related to each category. of core genes, including LIMK1, COL5A1, MMP9 and In this study, our aim is to fully identify the entire CCL26, were negatively associated with the prognosis of DNA methylated profles of RCC from the TCGA data- the patients [8]. To produce a characteristic biomarker base in order to identify the biological and clinical- to determine the high and low risk of clear renal cell car- related subgroups. Our datasets and the accompanying cinoma, Samira et al. developed a 34 genes subtype pre- classifcation schemes may help to identify new mark- dictor that divided 54 patients with metastatic RCC into ers or molecular subtypes of RCC in order to accurately good and poor follow-up groups. A model containing subdivide patients with renal cell carcinoma. Further- subtype-inclusive model was established to analyze the more, the RCC classifcation based on integrated pro- survival outcome of the patients [9, 10]. fling combined with histology and DNA methylation DNA methylation, one of the most common epige- profling analysis provides a more accurate prediction netic modifcations, plays a crucial role in the regu- of clinical behavior compared to the histology alone lation of the structure and expression of genes and is and provides higher risk assessment accuracy for indi- involved in a variety of biological processes. In addi- vidual patients. In addition to the identifcation of clin- tion, changes in DNA methylation are inextricably ical-related groups and a new classifcation basis, our linked to all aspects of cancer genomics and have been datasets can provide an in-depth understanding into shown to be closely related to changes in the gene the pathogenesis of RCC and guide accurate decision- expression, sequence and copy number [11]. In addi- making process during diagnosis and treatment. tion, DNA methylation often occurs on the C (cytosine) of 5′-CpG-3′ and as a result generates 5-methylde- Methods oxycytidine (5mC). DNA methylation has become an Date preprocessing and DNA methylation loci in RCC important research topic in epigenetics and epigenom- A total of 485 RCC samples containing DNA methyla-