An approach to the proposal of drug combination for cancer therapy using a pathway data connectivity approach

Within the next twenty years, the number of cancer patients is expected to rise by 70%. Current cancer treatments still face several limitations, such as severe side effects and a high incidence of disease recurrence. Drug combination therapies are a promising strategy to achieve higher therapeutic effects while reducing side effects. This new direction in cancer drug research has led to data-driven medicine. To predict whether certain drugs would have a synergistic effect when combined, the DREAM Challenge coordinators released data for thousands of experimentally tested drug combinations. The DREAM Challenge served as inspiration for selecting drug combinations that have the potential to be synergistic. We here describe an approach using biological pathway knowledge and applying this to the selected combinations with a previously described mathematical model, the Loewe-Additivity approach. The calculated interaction index (II) served to distinguish between synergistic (II < 1), additive (II = 1) and antagonistic (II > 1). Pre-selection of putative drug combinations was performed prior to synergy prediction based on four case scenarios: 1) two drugs share the same target protein, 2) two drugs share the same pathway, 3) drugs are separated by one degree from two targets or 4) drugs are separated by more than one degree from two targets but act upon the same biological pathway. Results: The first method tried was using a drug synergy prediction method called the Loewe-Additivity model, in which two drugs share the same target and form the initial findings for this paper. The Loewe model acts as a baseline estimation to see if more combinations can be identified using the other methods tested. The remaining methods used were able to find additional drug combinations that were not proposed by the standard Loewe model. Although the additional methods did find additional combinations that would be predicted to be synergistic, a prediction is not a guarantee of success, so validation of the new or novel combinations would be needed to verify their effectiveness. This could be done by comparing our results to known data or against biological assays.