Abstract 552: Multi-scale omics integration using parallel heatmap clustering for the systemic analysis and biomarker discovery of drug sensitivity in lymphoma cell lines

Abstract

Abstract Background. Pharmacogenomics studies the role of genomics in drug response. By measuring the individual genome, it is hoped that pharmaceutical drug treatments can deviate from a "one-dose-fits-all" approach to a more "personalized" treatment. To achieve this goal, we need to explain which genomic differences cause the lack of response and to be able to predict the response from baseline omics data, that is from data available before starting treatment. Here, we present an integrated omics approach to analyze drug sensitivity based on in vitro experiments. Methods. We measured the drug response in 61 lymphoma cell lines for a number of anti-cancer drugs using a standard MTT cell proliferation assay. The baseline genomics of these cell lines were fully profiled for gene expression, copy number and methylation. Pathway and gene set signatures were computed using GO, KEGG, Biocarta and lymphoma databases to provide gene set level features. Additionally, we computed high level "biological concept" features. These multi-scale features were directly correlated with drug sensitivity but also correlated between the multiple data types. Using a novel parallel clustering approach, we integrated our data for the different omics types and at different scales: gene, gene set and concept level. Finally, a shortest path algorithm extracted the most probable explanation between genotype and response phenotype. Results. As a proof of concept, we first applied our method to the classification of activated B- cell like (ABC) and germinal center B (GCB) subtypes in diffuse large B-cell lymphomas (DLBCL). In accordance to previous knowledge, our method showed the MYD88/CD40/NFKB axis as strongly upregulated in the ABC, while, albeit less pronounced, the PI3K/MTOR and NOTCH signalling pathways were more enriched in the GCB subtype. We then applied our method to the drug sensitivity data of our lymphoma cell line panel. The analysis showed that drug response was largely driven by differential methylation rather than copy number aberrations. We found that B cell receptor signaling (BCR) and CD40 signalling were among the most correlated to drug sensitivity in a drug-specific manner. Genes in these pathways (such as SYK, SPIB and CD79A) appeared as good biomarkers for drug-specific response in lymphoma. Conclusion. Using an integrated analysis of multiple omics data at multiple scales, we successfully identified functional modules related with drug sensitivity in lymphoma cell lines. The same method can be applied to study the sensitivity to other drugs in either hematological and/or solid cancer types. Citation Format: Ivo Kwee, Andrea Rinaldi, Alberto J. Arribas, Eugenio Gaudio, Chiara Tarantelli, Filippo Spriano, Petra Hillmann, Francesco Bertoni. Multi-scale omics integration using parallel heatmap clustering for the systemic analysis and biomarker discovery of drug sensitivity in lymphoma cell lines [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 552. doi:10.1158/1538-7445.AM2017-552