Making drug combination testing a reality

Researchers from the University of Helsinki have developed a novel experimental and computational pipeline that enables more cost-effective and simultaneous large-scale drug combination testing in cancer.

The study was conducted by research groups lead by Jing Tang at the Institute for Molecular Medicine Finland (FIMM), Helsinki Institute of Life Science (HiLIFE) and the Faculty of Medicine and Caroline Heckman at FIMM. The results were recently published in PLOS Computational Biology.

Why drug combinations?

Cancer remains one of the leading causes of death worldwide. Although new, targeted treatments have been developed, these only benefit patients with an actionable mutation. In addition, cancer is extremely heterogeneous and drug resistance can eventually develop. There is an urgent need for novel, multi-targeted drug combinations that can effectively inhibit the cancer cells while incurring minimal damage to normal cells.

 "It is now widely acknowledged that effective cancer treatments need to go beyond the traditional ‘one disease, one drug, one target’ paradigm. In contrast, ‘polypharmacology‘ focuses on developing multi-targeted drugs or drug combinations, showing great promise to reach effective and sustained clinical responses", explains Assistant Professor Jing Tang, the lead principle investigator of the study.

Combining synergistically acting drugs has many benefits since it can improve the efficacy of the treatment and minimize the doses at which the desirable efficacy is achieved.

How to assess drug combinations?

High-throughput drug screening allows researchers to quickly perform a substantial amount of drug combination tests using a large variety of cancer cell lines and patient-derived cancer samples. This way the most promising drug combinations can be identified.

Many high-throughput drug combination screens utilize a dose response matrix design, where every dose of one drug is tested with every dose of another drug. By applying computational methods on the dose response matrix data, two important drug combination properties, sensitivity and synergy, can be evaluated.

There are, however, both experimental and computational challenges.

“Patient-derived cancer samples are often restricted in volume, limiting the number of drug concentrations tested, which will make it challenging for computational approaches to identify promising drug combinations for an individual patient”, explains PhD student Alina Malyutina, first author of the study.

 “Many existing computational tools for drug combination analysis focus on the synergistic effect, but not the sensitivity of drug combinations. A drug combination may act synergistically, but if its overall sensitivity is not big enough to kill most of the cancer cells then we should not consider it as a positive hit”, adds Muntasir Mamun Majumder, post-doctoral researcher at FIMM.

What can be done better?

To overcome these methodological challenges, researchers from the Tang and Heckman groups proposed a novel cross design.

In the new design, the researchers allow one of the tested drugs to span over multiple doses while the dose of the other drug is fixed at its IC50 concentration, i.e. the drug dose that is needed to kill half of the treated cells. The resulting drug combination dose-response curves are then summarized as the drug combination sensitivity score (CSS), from which a CSS-based synergy score, called S score, can be determined.

“By fixing the drugs at their IC50 concentrations, we can avoid a drug combination that achieves its efficacy only at higher doses, which are usually too toxic for the patients”, says Caroline Heckman, Group Leader at FIMM.

Using a publicly available high-throughput dataset, the team was able to demonstrate that the cross design coupled with the novel scoring methods provide robust and accurate characterization of both sensitivity and synergy. Importantly, the results also showed that CSS and S scores can detect the true positive drug combinations at an accuracy level comparable to that of the full matrix design, however, requiring much less cell material.

The proposed experimental and computational techniques are expected to be widely applicable in the field of personalized drug combination discovery.

“With our approach the drug combination testing may become more amenable for patient-derived cancer samples, which are usually difficult to obtain and limited in the amount of cells available for testing”, Jing Tang concludes.

Read more:

Malyutina, A., Majumder, M. M., Wang, W., Pessia, A., Heckman, C. A., Tang, J. Drug combination sensitivity scoring facilitates the discovery of synergistic and efficacious drug combinations in cancer. PLOS Computational Biology (2019). https://doi.org/10.1371/journal.pcbi.1006752

Tang, J., 2017. Informatics approaches for predicting, understanding, and testing cancer drug combinations. In Kinase signaling networks (pp. 485-506). Humana Press, New York, NY. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6322649/