Login to Access Video or Poster Abstract: MP88-18
Sources of Funding: none

Introduction

The goal of precision medicine is to predict the best treatment strategy from available genomic information, on a patient-by-patient basis. Bladder cancer genomics has emerged as a new area of research, whereby molecular subtypes of bladder cancer based on gene expression models may have selective therapeutic targets. Here we implement a novel bioinformatics approach integrating gene expression and drug sensitivity analyses to determine molecular subtype-specific therapeutic vulnerabilities in bladder cancer.

Methods

Gene expression profiles for 26 bladder cancer cell lines were obtained from the Cancer Cell Line Encyclopedia (CCLE) and analyzed by unsupervised hierarchical clustering using Morpheus software (Broad Institute, Cambridge, MA). Cell line clusters were classified according to validated genomic classification systems. Mutational status and drug sensitivity data for 19 bladder cancer cell lines treated with 224 anti-cancer drugs was obtained from the Genomics of Drug Sensitivity in Cancer (GDSC) database (Sanger Institute, Cambridge, UK). Differential sensitivity analyses were performed using Graphpad Prism.

Results

Unsupervised hierarchical clustering of gene expression data revealed major subgroups that clustered according to classified molecular subtypes: Squamous cell carcinoma-like (SCC-like), Urobasal A, Urobasal B, and Urobasal A/B (Fig 1A). Differential sensitivity analyses revealed that certain subtypes are preferentially sensitive to specific drugs. The most significant drug/subtype combination was the sensitivity of SCC-like cell lines to Phosphatidylinositide 3-kinase beta (PI3Kβ) inhibitor AZD6482, compared to Urobasal A or B lines (Fig 1B; P<0.05 by Kruskal-Wallis test). This unique sensitivity is associated with PTEN loss of function, which was found to be more commonly altered in SCC-like vs Urobasal cell lines (p<0.05).

Conclusions

Using cell line gene expression profiling and drug sensitivity data, we developed a novel bioinformatics approach and demonstrated sensitivity of SCC-like bladder cancers to the PI3Kβ inhibitor AZD6482, which may represent a novel therapeutic target. Furthermore, PTEN mutational status may represent a biomarker for sensitivity to these agents.

Funding

none