TY - JOUR
T1 - Specific breast cancer prognosis-subtype distinctions based on DNA methylation patterns
AU - Zhang, Shumei
AU - Wang, Yihan
AU - Gu, Yue
AU - Zhu, Jiang
AU - Ci, Ce
AU - Guo, Zhongfu
AU - Chen, Chuangeng
AU - Wei, Yanjun
AU - Lv, Wenhua
AU - Liu, Hongbo
AU - Zhang, Dongwei
AU - Zhang, Yan
N1 - Publisher Copyright:
© 2018 The Authors. Published by FEBS Press and John Wiley & Sons Ltd.
PY - 2018/6
Y1 - 2018/6
N2 - Tumour heterogeneity is an obstacle to effective breast cancer diagnosis and therapy. DNA methylation is an important regulator of gene expression, thus characterizing tumour heterogeneity by epigenetic features can be clinically informative. In this study, we explored specific prognosis-subtypes based on DNA methylation status using 669 breast cancers from the TCGA database. Nine subgroups were distinguished by consensus clustering using 3869 CpGs that significantly influenced survival. The specific DNA methylation patterns were reflected by different races, ages, tumour stages, receptor status, histological types, metastasis status and prognosis. Compared with the PAM50 subtypes, which use gene expression clustering, DNA methylation subtypes were more elaborate and classified the Basal-like subtype into two different prognosis-subgroups. Additionally, 1252 CpGs (corresponding to 888 genes) were identified as specific hyper/hypomethylation sites for each specific subgroup. Finally, a prognosis model based on Bayesian network classification was constructed and used to classify the test set into DNA methylation subgroups, which corresponded to the classification results of the train set. These specific classifications by DNA methylation can explain the heterogeneity of previous molecular subgroups in breast cancer and will help in the development of personalized treatments for the new specific subtypes.
AB - Tumour heterogeneity is an obstacle to effective breast cancer diagnosis and therapy. DNA methylation is an important regulator of gene expression, thus characterizing tumour heterogeneity by epigenetic features can be clinically informative. In this study, we explored specific prognosis-subtypes based on DNA methylation status using 669 breast cancers from the TCGA database. Nine subgroups were distinguished by consensus clustering using 3869 CpGs that significantly influenced survival. The specific DNA methylation patterns were reflected by different races, ages, tumour stages, receptor status, histological types, metastasis status and prognosis. Compared with the PAM50 subtypes, which use gene expression clustering, DNA methylation subtypes were more elaborate and classified the Basal-like subtype into two different prognosis-subgroups. Additionally, 1252 CpGs (corresponding to 888 genes) were identified as specific hyper/hypomethylation sites for each specific subgroup. Finally, a prognosis model based on Bayesian network classification was constructed and used to classify the test set into DNA methylation subgroups, which corresponded to the classification results of the train set. These specific classifications by DNA methylation can explain the heterogeneity of previous molecular subgroups in breast cancer and will help in the development of personalized treatments for the new specific subtypes.
KW - breast cancer
KW - consensus clustering
KW - DNA methylation
KW - molecular subtypes
UR - http://www.scopus.com/inward/record.url?scp=85049240506&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85049240506&partnerID=8YFLogxK
U2 - 10.1002/1878-0261.12309
DO - 10.1002/1878-0261.12309
M3 - Article
C2 - 29675884
AN - SCOPUS:85049240506
SN - 1574-7891
VL - 12
SP - 1047
EP - 1060
JO - Molecular oncology
JF - Molecular oncology
IS - 7
ER -