Long non-coding RNA profiling links subgroup classification of endometrioid endometrial carcinomas with trithorax and...

Integrative analysis of endometrioid endometrial carcinoma (EEC) using multiple platforms has distinguished four molecular subgroups. However, the landscape of expressed long non-coding RNAs (lncRNA) and their role in charting EEC subgroups and determining clinical aggressiveness remain largely unknown.

Using stringent criteria, we identified 1,931 expressed lncRNAs and predicted potential drivers through integrative analysis. Unsupervised clustering of lncRNA expression revealed three robust categories: basal-like, luminal-like and CTNNB1-enriched subgroups. Basal-like subgroup was enriched for aggressive tumors with higher pathological grade (p < 0.0001), TNM stage (p = 0.01), and somatic mutations in trithorax-group genes (MLL, MLL2 and MLL3); and it overexpressed polycomb genes EZH2 and CBX2. In contrast to the luminal-like subgroup, progesterone (PGR) and estrogen receptor (ESR1) genes were highly down-regulated in the EEC basal-like subgroup. Consistent with its enrichment for CTNNB1 mutations (69%), lncRNA profile of the CTNNB1-enriched EEC subgroup was highly similar to that of the CTNNB1-enriched liver cancer subgroup.

We performed integrative analysis of lncRNAs in EEC using The Cancer Genome Atlas (TCGA) molecular RNAseq profiles of 191 primary tumors for which genomic data were also available. We established lncRNA subgroup classification, correlated it with chromatin modifying gene expression, and described correlations between our lncRNA classification and clinico-genomic tumor features.

Our results reveal the utility of systematic characterization of clinically annotated EEC in three clinically relevant subgroups. They also highlight the convergence of aberrations in polycomb- and trithorax-group genes in aggressive basal EEC subtypes, providing a rationale for further investigation of epigenetic therapy in this setting.