Search Tips

A Molecular Taxonomy of Urothelial Carcinoma

UID: 10402

Author(s): Sjödahl, Gottfrid, Lauss, Martin, Lindgren, David, Höglund, Mattias

Summary from the GEO: " Even though urothelial cancer (UC) is the fourth common tumor type among males, progress in treatment development has been deficient. Pathological assessment provides the urologists with only a broad classification, complicated by frequent disagreement among pathologist and the co-existence of different grading systems. Consequently, there is a great need for an objective, reproducible and biologically relevant classification system to make treatment more efficient. In the present investigation we present a molecular taxonomy for UC stratification based on integrated genomics.
We used gene expression profiles from 308 UC to define seven molecular subtypes using step-by-step partitions and a bootstrap approach. Results were validated in three independent and publically available data sets. The subtypes differ significantly with respect to expression of cell cycle genes, of receptor tyrosine kinases particularity FGFR3, ERBB2 (HER2), and EGFR, of an FGFR3 associated gene expression signature, of cytokaratins, and of cell adhesion genes. The subtypes also differ significantly with respect to FGFR3, PIK3CA, and TP53 mutations. The expression of key proteins was validated by IHC on TMA. A further inspection indicated that the subtypes could be reduced to four major types of UC; Urobasal/D-driven, Genomically unstable/E-driven, Evolved urobasal, and Basal/SCC like, with characteristic and highly divergent molecular phenotypes. We show that the molecular subtypes cut across pathological classification and that tumors classified as one subtype maintain their characteristic molecular phenotype irrespective of pathological stage and grade. Available data from the Drugbank database and the Cochrane central registry of controlled trials indicate that susceptibility to specific drugs is more likely to be associated with the molecular stratification than with pathological classification.
The presented molecular taxonomy stratifies UC into subtypes with distinct molecular phenotypes and biological properties. We anticipate that the molecular taxonomy will be a useful tool in future clinical investigations.
Overall Design: Total RNA from fresh-frozen resection samples of 308 urothelial carcinomas was hybridized to the Illumina HumanHT-12 V3.0 expression beadchip arrays (Illumina Inc) at the SCIBLU Genomics Centre at Lund University Sweden (

Supplementary files:
GSE32894_non-normalized_308UCsamples.txt file = Raw intensity values for 308 UC (urothelial tumor) samples subjected only to background correction.
GSE32894_reps_normals_preprocess*.txt files = Descriptive details and non-normalized data for technical replicates and normal samples that were used only in the preprocessing of the data."
Subject of Study
OncoTree Cancer Type(s)
Bladder Urothelial Carcinoma
Access via GEO

TAR and TXT Sequencing data
Accession #: GSE32894

Access via GEO

This dataset partly overlaps with Series GSE32549. Names of the overlapping sample names are the same, but the title of each sample is unique to the hybridization.
Accession #: GSE32549

Access via BioProject

Additional information about overall initiative.
Accession #: PRJNA146635

Access Restrictions
Free to All
Access Instructions
The NCBI Gene Expression Omnibus database provides open access to these files. SRA and/or tar files can be downloaded directly from the site or viewed in the NCBI SRA Run Selector (link at bottom of page).
Associated Publications
Data Type
Equipment Used
Illumina HumanHT-12 V3.0 expression beadchip
Software Used
Dataset Format(s)
Plain Text, SRA, TAR
Dataset Size
6.2 MB (TAR), 62.8 MB (TXT)
Data Catalog Record Updated