IBCN 2017: Molecular Subtyping I
Matthias Hoglund from Lund University, Lund, Sweden provided further validation of the Lund taxonomy classification of advanced urothelial carcinomas. Luminal can be subcategorized into UroA-prog, GU and UroC which have unique signatures while basal cells can be subcategorized into Mes-like, Ba/sq and UroB. Lastly, the Sc/NE-like have been identified and taken together these subtypes are characterized by unique infiltration and aggressiveness properties. Using the Lund cohort, they created a RNA-expression based tumor-cell phenotype classifier as well as validation studies using established gene signatures. Taking these additional validation studies into account they noted UroB and Sc/Ne have worse overall survival.
Sarah Dominguez from the Spanish National Cancer Research Centre, Madrid, Spain, discussed specific risk patterns for bladder cancer taxonomy-based subphenotypes. Bladder cancer is a complex disease with differential risk profiles according to subphenoptypes. Using the Spanish Bladder Cancer EPICURO study they identified 1355 cases and 1270 controls. Immunohistochemical expression of KRT5/6, FOXA1 and GATA3 was assessed to define subphenotypes by means of unsupervised hierarchical cluster analysis. Three UBC clusters were identified: BASQ (N=65, 7%), luminal-like (high GATA3-FOXA1, low KRT14-KRT5/6, N=595, 66%) and mixed (N=247, 27%). BASQ-like tumors were mostly MIBC and nonpapillary. They were less strongly associated with tobacco smoking and risk-occupational exposure and more strongly associated with bladder infections. Luminal-like tumors were inversely associated with NSAIDs consumption and high levels of plasma vitamin D. Mixed-like tumors were inversely associated with asthma and BPH. This study provides suggestive evidence that UBC subphenotypes display distinct etiological associations with environmental/life-style factors.
Seth Lerner from Baylor, Houston, TX presented his abstract regarding non-coding RNA using TCGA. This work characterized microRNA (miR) and long non-coding (lnc) RNA expression patterns and their association with survival. They calculated expression profiles for 8167 (Ensembl v82) lncRNAs and extracted lncRNAs that were robustly and variably expressed, and identified subtypes by unsupervised consensus clustering. Four unsupervised lncRNA clusters were associated with purity, EMT score, CIS gene sets, and 5-year survival (p = 0.015). The lncRNA clusters were concordant with mRNA subtypes and further discriminated within them. miRNA subtypes were associated with purity, EMT scores, and 5-year survival (p = 0.002). They were also concordant with subtypes for mRNA, lncRNA, hypomethylation, RPPA, and with histological subtype, TNM stage, and CIS gene sets. miRNA subtype 3 was enriched in lncRNA 3 and showed the best survival among the 4 subtypes, consistent with low EMT scores and high miR-200 levels. Multivariate regression analyses identified lncRNA and miRNA subtypes as independent predictors of survival. In summary, distinct miRNA and lncRNA subtypes for muscle-invasive bladder cancer, and describe a high level of correlation with mRNA subtypes and survival outcomes.
Written by: Stephen B. Williams, M.D., Associate Professor, Division of Urology, The University of Texas Medical Branch, Galveston, TX. and Ashish M. Kamat, M.D. Professor, Department of Urology, Division of Surgery, The University of Texas MD Anderson Cancer Center, Houston, TX., at the International Bladder Cancer Network - October 21, 2017- Lisbon, Portugal