Long non-coding RNAs are differentially expressed between bladder cancer subtypes

Background: The recent identification of molecular bladder cancer subtypes by whole transcriptome studies showed similarities to molecular breast cancer phenotypes. We here validate these subtypes with a sensitive 36 gene nCounter screening and analyse relevant lncRNA for their differential expression. Methods: RNA has been extracted from chemotherapy-naïve muscle-invasive bladder cancer (MIBC) after radical cystectomy (follow-up: 12 years, n = 48). A multiple marker gene panel has been quantified with the nCounter technology. In silico validation of the classifier geneset on 170 MIBCs has been performed. All squamous carcinoma were excluded. LncRNAs were analyzed in a clustering-independent assessment. Multivariate analyses were performed by a Cox proportional hazards model. Results: 36 consensus genes were generated by Venn diagrams based on the Mannheim, Lund, Chungbuk and MDA cohorts. This minimal set of genes generated 3 stable clusters: basal, luminal and infiltrated. The subtype specific assessment of 14 lncRNAs relevant in bladder cancer showed a highly subtype specific expression for 9 lncRNAs. The infiltrated subtype, characterized by an activated p53 downstream signature, showed an overexpression of SRA1 and MEG3 (p ≤ 0.003) - the latter is known for promoting the expression of TP53. The lncRNAs H19, GAS5, TUG1 and CBR3-AS1 showed a significant upregulation in the luminal subtype (p < 0.05) whereas SNHG16 showed an exclusive suppression. MALAT1 was suppressed in the basal subtype. A distinct cutoff of the lncRNA H19 allowed a risk stratification into high- and low-risk patients. The luminal subtype and H19 were the only independent risk factors in multivariate analysis adjusted for TNM and were predictive for a 3- to 4-fold higher risk of death (p < 0.03). Conclusions: In this study, MIBC subtypes have been validated by a sensitive quantification method. Molecular subtypes and H19 prove to be independent risk factors superior to TNM. This study demonstrates for the first time a differential expression of lncRNA between MIBC subtypes. The potential impact of lncRNA on phenotype determination has to be investigated in vivo. Legal entity responsible for the study: BRIDGE Consortium Funding: None Disclosure: R. Sébastien: Novartis Research Fund. All other authors have declared no conflicts of interest.