Pharmacogenomics of Primary Blood Cancer Cells

Abstract

Current developments of multi-omics technologies greatly accelerate advancements in the field of cancer research. They not only allow us to unravel the complex biology of tumors, but are also a huge step towards precision oncology, where a personalized treatment is proposed based on the tumor’s unique combination of molecular features. Heterogeneity present both within a single tumor and between patients with the same disease pose serious challenges to treatment success and trial design, as many factors influencing drug response in cancer remain unknown. The work presented here investigated the diversity of drug response profiles and their associations with the underlying molecular features of 273 primary cancer samples. The analysis combined data from high-throughput drug profiling of 90 compounds with multi-omics comprising exome sequencing, RNA sequencing and methylation profiling. The analysis of various hematological malignancies uncovered a rich landscape of phenotype genotype relationships. Targeted inhibition of, for example, CHK, SERCA, BCL2 and survivin revealed disease-specific pathway dependencies within B- and T-cell lymphomas. Moreover, the observed similarity of drug response profiles identified unexpected activity of compounds including CHK inhibitors, which implied to act through B-cell receptor (BCR) signaling pathway. By focusing specifically on chronic lymphocytic leukemia, dissection of the molecular fundamentals of known biomarkers was possible. Drug response measured ex vivo confirmed the biological relevance of major predictors of patient clinical outcome: the sensitivity of IGHV unmutated samples to BCR inhibition and the resistance of TP53 mutated samples to chemotherapy and nutlin-3. Susceptibility of trisomy 12 cases to SYK, BTK, PI3K and MEK inhibitors suggested a mode of action through amplification of BCR pathway signaling. Additionally, the study proposed possible targeting options for recurrently mutated genes, such as BRAF, CREBBP and PRPF8. Multivariate analysis estimated the contributions of mutations, RNA expression and DNA methylation to our power to predict drug response. Finally, drug response was shown to be valuable and frequently superior to established biomarkers in predicting patient clinical endpoints. In summary, the combination of viability screening with multi-omics profiling is a powerful tool for studying disregulated signaling pathways in cells. Ex vivo drug profiling performed on primary cancer samples has proved to be a proxy of true biology. Together with faithful reproducibility of measurements this shows great promise in directly advancing precision oncology.