Drug sensitivity

Objectives

We will advance the bioinformatic and statistical analysis of systematic phenotypic screens of drug response of primary cancer cells. This work package (Genomics of Drug Sensitivity of Primary Cancer Cells) will apply and test in practice the tools developed in WP1, and we will do this on the data generated by SMART, a large experimental and clinical research programme of Partner UKL-HD, in which NCT systematically measures the response to comprehensive panels of inhibitors (up to 3000 drugs at different concentrations) in large, representative cohorts (currently n=300) of patient samples (leukaemia and lymphoma). To date, our high-throughput phenotypic assays have generated a characteristic sensitivity profile for each sample across diverse substances and inhibitors with in total over 250,000 data points. All data acquisition including RNA sequencing (n=200), WES (n=200) and methylation arrays (n=200) will be completed in October 2014 and is entirely funded outside SOUND.

We associate the variable drug response (sensitivity and resistance) with molecular findings from multiple layers of ’omics technologies that were used to characterize the molecular make-up of the tumours: DNA somatic SNVs and structural variants, RNA expression, DNA methylome, metabolome, constitutive SNPs. The overall aim of the NCT/SMART programme is three-fold: to increase our understanding of the molecular principles that underlie variable treatment response and disease outcome, to discover actionable tumour-specific vulnerabilities (e.g., specific pathway inhibition) at individual patient resolution, and to enable rational individualized treatment choice [Tyner, Cancer Res. 2013; Pemovska, Cancer Discov. 2013; Sellner, ASH 2013]. Altogether, we aim to develop the tools for precision medicine for patients with blood cancer.

Specifically, we aim in this work package to further develop and practically validate the bioinformatic tools needed to:

1) Identify all molecular determinants associated with variable drug response.

2) Construct multivariate predictors of drug response that optimally integrate all relevant layers of molecular information, including RNA expression, DNA variants and methylation profiles.

3) Understand the underlying hierarchies and relationships between the molecular determinants of response.

4) Understand the mechanisms underlying drug response heterogeneity as a means of furthering the overall aims stated above.

With its large scale and variety of ’omics data types, the SMART programme serves as a challenging and relevant testing ground to validate the tools developed by SOUND in practical application.

Participating partners

  • German Cancer Research Center (Lead Partner)
  • European Molecular Biology Laboratory, Heidelberg
  • ETH Zurich
  • BeDataDriven, The Hague