Long noncoding RNA profiles identify five distinct molecular subtypes of colorectal cancer with clinical relevance

Abstract

Colorectal cancer (CRC) is a heterogeneous disease in terms of clinical behavior and response to therapy. Increasing evidence suggests that long noncoding RNAs (lncRNAs) are frequently aberrantly expressed in cancers, and some of them have been implicated in CRC biogenesis and prognosis. Using an lncRNA-mining approach, we constructed lncRNAs expression profiles in approximately 888 CRC samples. By applying unsupervised consensus clustering to LncRNA expression profiles, we identified five distinct molecular subtypes of CRC with different biological pathways and phenotypically distinct in their clinical outcome in both univariate and multivariate analysis. The prognostic significance of the lncRNA-based classifier was confirmed in independent patient cohorts. Further analysis revealed that most of the signature lncRNAs positively correlated with somatic copy number alterations (SCNAs). This lncRNAs-based classification schema thus provides a molecular classification applicable to individual tumors that has implications to influence treatment decisions.

Keywords: Colorectal cancer; Consensus clustering; Gene expression profiling; Gene set enrichment analysis; Somatic copy number alterations; Survival; lncRNA.

Figure 1

Unsupervised lncRNAs expression profiling analysis of the discovery set of 443 colorectal cancers. (a) NMF consensus clustering analysis and cophenetic coefficient for cluster k = 2 to k = 6 of discovery dataset of GSE39582. Maximum cophenetic coefficient occurred for k = 5. (b) Silhouette plot for the discovery dataset of GSE39582 showing samples from different subtypes and those with positive and negative silhouette score.(c) Heatmap showing 5 subtypes (k = 5) from NMF consensus clustering of the discovery dataset of GSE39582. (d) The heatmap of the 229 lncRNAs ordered by subtype, with annotations associated with each subtype.

Figure 2

GSE39582 (a), GSE39582, GSE33113GSE17536/GSE17537 Gene set enrichment analysis delineates biological pathways and processes that define five distinct subtypes. Cytoscape and Enrichment Map were used for visualization of the GSEA results. Nodes represent enriched gene sets, which are grouped and annotated by their similarity according to related gene sets. Enrichment results were mapped as a network of gene sets (nodes). Node size is proportional to the total number of genes within each gene set. Proportion of shared genes between gene sets is represented as the thickness of the green line between nodes. This network map was manually curated removing general and uninformative sub‐networks, resulting in a simplified network map.

Figure 3

DFS comparison of five CRC subtypes. Kaplan‐Meier Survival curve depicting differential survival for the discovery dataset of GSE39582 (a), validation dataset of GSE39582, GSE33113 and GSE17536/GSE17537 (b) and all the dataset combined (c).

Figure 4

Integrated view of signature lncRNAs regulated by Copy Number. (a) An ideogram of a normal karyotype is shown in the outer ring. The next outermost ring represents the lncRNAs expression levels(Red represents ncRNAs which positively correlates with expression level changes and green represents ncRNAs does not correlate with expression level changes.). The next ring represents copy number as a function of genomic coordinates. (Red represents amplification and green represents deletion).(b) In the center of the figure, the box plot showed two representative lncRNAs, AK022220 and BC017275, positively correlates with SCNA (p < 0.001).