Predictive approaches for drug combination discovery in cancer

Abstract

Drug combinations have been proposed as a promising therapeutic strategy to overcome drug resistance and improve efficacy of monotherapy regimens in cancer. This strategy aims at targeting multiple components of this complex disease. Despite the increasing number of drug combinations in use, many of them were empirically found in the clinic, and the molecular mechanisms underlying these drug combinations are often unclear. These challenges call for rational, systematic approaches for drug combination discovery. Although high-throughput screening of single-agent therapeutics has been successfully implemented, it is not feasible to test all possible drug combinations, even for a reduced subset of anticancer drugs. Hence, in vitro and in vivo screening of a large number of drug combinations are not practical. Therefore, devising computational methods to efficiently explore the space of drug combinations and to discover efficacious combinations has attracted a lot of attention from the scientific community in the past few years. Nevertheless, in the absence of consensus regarding the computational approaches used to predict efficacious drug combinations, a plethora of methods, techniques and hypotheses have been developed to date, while the research field lacks an elaborate categorization of the existing computational methods and the available data sources. In this manuscript, we review and categorize the state-of-the-art computational approaches for drug combination prediction, and elaborate on the limitations of these methods and the existing challenges. We also discuss about the recent pan-cancer drug combination data sets and their importance in revising the available methods or developing more performant approaches.

Figure 1

Schematic diagram of steps in which mathematical and computational methods can contribute in drug combination development: (A) quantification of experimental drug combination; (B) potential data sources for predictive drug combination approaches; (C) predictive computational approaches.

Figure 2

Schematic representation of steps in quantification methods: (A) median-effect method: m_a, m_b and m_c are slopes and D_ma, D_mb and D_mab are y-intercepts of the log–log diagrams (Hill-type coefficients) for drug A, drug B and their combination, respectively. D_A and D_B also defined as D_A = (Dx)_AB (a/(a + b)), D_B = (Dx)_AB (b/(a + b)) where a/b is the ratio of drugs A and B in the combination. Mutually exclusive drugs and non-exclusive drugs are the ones that act competitively and noncompetitively on the same target, respectively (their difference can be seen in the log–log diagrams (Supp. Fig. 1)). (B) Loewe additivity method: D_A and D_B are the drug doses of drugs A and B in the combination, which can cause the assumed inhibitory effect (such as 50% inhibition). (C) Bliss independence method: E_a, E_b and E_ab are the inhibitory effects of drugs A and B and their combination. E_add is also the reference measure or additive effect of the combination of drugs. E_observed is also the observed inhibitory effect of the combination of drugs A and B with the considered doses a and b [23, 50, 56, 57, 63, 67].

Figure 3

Potential data for predictive in silico pipelines of drug combinations prediction including information about drugs, targets as well as response of the cell lines or tumor cells to drugs as single agent or in combination with other agents.

Figure 4

Accuracy measures for different assessment strategies.