Genome-wide methylation analysis and comprehensive molecular profiling on endometrial cancers

Abstract

Endometrial cancer is the most prevalent gynaecological cancer worldwide, with a rising incidence in Hong Kong. However, to the best of my knowledge, limited studies in Asia and none in Hong Kong have reported molecular subgroup classification of endometrial cancer and its impact on clinical outcomes. Accurate prognosis prediction can improve treatment planning. Hence, my study aimed to identify the local molecular subgroup profiles and their associations with clinicopathological features and prognostic outcomes.

Ninety-one endometrial cancer patients were recruited for molecular classification” to avoid misunderstanding. The tissue samples were analysed for POLE exonuclease domain mutations using bidirectional Sanger sequencing and immunohistochemical evaluations of p53 and MMR proteins. POLE-mutant (POLEmut) (3.3%), MMR-deficient (MMRd) (19.8%), p53 abnormal (p53abn) (13.2%) and p53 wild type/nonspecific molecular profile (p53wt/NSMP) (63.7%) subgroups were identified. My findings confirmed that the p53abn subgroup was associated with the most aggressive phenotypes and the worst prognosis. While POLEmut cases had better clinical outcomes, distinct POLE variants different from those reported in western literature were found.

Accurate staging of synchronous endometrial and ovarian cancer (SEOC) remains a diagnostic challenge (metastatic vs double primaries) and often depends on the evaluation of clinicopathological features. By whole mitochondrial DNA Sanger sequencing, accurate tumour clonality was assessed, confirming five endometrial cancers with metastasis and two without metastasis.

Aberrant DNA methylation is one of the cancer hallmarks. Differential methylation analysis was performed on endometrial cancers with and without metastasis to determine the genome-wide methylation landscapes and differential methylation genes (DMGs). The Infinium Methylation EPIC microarray, which covers over 850,000 CpG sites, was employed. Multiple DMGs, including previously reported biomarkers (APC and MLH1) and novel biomarkers (TBX18, ISL1 and ZNF354C), were identified. In silico analysis of the uterine corpus endometrial carcinoma of the Cancer Genome Atlas (TCGA) dataset with MethSurv showed that low methylation levels in APC, ISL1 and ZNF354C significantly correlated with a poor prognosis in these patients.

Methylation status of APC gene was further evaluated using bisulfite PCR, and pyrosequencing was performed on 121 endometrial cancers. Hypomethylation at APC-cg20311501 was significantly associated with shorter survival. Hence, APC-cg20311501 methylation status may be a potential prognostic biomarker.

Moreover, cell-cell adhesion flipping experiments indicated that overexpression of TBX18 in endometrial cancer cells (Ishikawa) reduced its adherence to normal human endometrial cells (EM-E67/hTRERT2) but not normal fallopian tubal epithelial cells (OE-E6/7) or ovarian surface epithelial cells (HOSE11-12). The reduced cell adhesiveness related to TBX18 overexpression may promote tumour exfoliation, favouring metastases.

This is the first study to present the local molecular subgroup profile of endometrial cancers correlated with clinical outcomes. POLE variants different from those reported in western literature suggested potential geographical heterogeneity in POLE variants. By whole-genome methylome analysis, dysmethylated APC, MLH1, TBX18, ISL1 and ZNF354C may be potential biomarkers to predict metastasis and survival. Low methylation at APC-cg20311501 showed impacts on survival, and overexpression of TBX18 reduced cell adhesion and may favour metastasis.