Integrated Computational and Experimental Approaches for Accelerated Drug Combination Discovery and Development : Applications in Cancer Pharmacology

Abstract: Today the norm in modern cancer treatment is to use different forms of drug combinations. Recently anti-cancer treatment using drug combinations has gained increased attention due to the outstanding pharmacotherapeutic opportunities provided by combination therapies. However, the potential of this field is largely unexplored, partly due to the complexities associated with the astronomical number of possible combinations and partly due to the lack of means for quantifying clinically relevant adverse side effects in the early stages of the combination discovery and development process. This has resulted in relatively limited progress in this area. Motivated by this unfortunate state-of-affairs, the research reported in this thesis was aimed at developing and implementing computational and experimental methods to facilitate and accelerate the discovery and development of anti-cancer therapies. In paper I, the largely overlooked concept of therapeutic synergy is re-introduced and demonstrated to be useful already at the level of combination discovery by taking both curative and adverse effects into account. In paper II, a semiautomatic combination discovery platform was developed based on a tailored programming of a pipetting robot system and application of a new in-house developed combination search algorithm, the therapeutic algorithmic combinatorial screen (TACS) algorithm. TACS seems to be the first algorithm of its kind that takes experimental variability into account during the iterative search process. The semiautomatic hardware platform along with TACS can perform de novo or knowledge based combination drug discovery and development without brute force comprehensive search efforts. One promising discovery made using this platform is a combination of the drugs 17-AAG, afungin and trichostatin a for treatment of colorectal cancer carcinoma (CRC). In paper III, an algorithm is developed and applied in order to use single drug induced systemic gene expression profiles for rational drug combination design by assuming additive combination effects. The resulting algorithm, combo-CMap, is applied and validated using a slightly extended version of the freely available Connectivity Map (CMap) database which is currently containing 6190 chemically induced mRNA gene expression signatures. In paper IV, a software (R package) was developed and applied to perform improved synergy/antagonism analysis, in particular joint Loewe and Bliss analyses while taking associated experimental variability into account using non-parametric statistics including bootstrap intervals. Applying this software to the synergy analysis of interaction effects among clinically used and/or relevant drugs in CRC cell lines revealed complex patterns of synergy and antagonism. In conclusion, the work presented here offers important contributions and findings that may accelerate and/or improve different parts of the field of drug combination discovery and development.