During the last few years, combination therapies have become increasingly popular and promising for complex diseases like cancer, due to their potential to overcome the resistance mechanisms that often hinder the long-term efficacy of conventional cytotoxic chemotherapies or targeted agents inhibiting single pathways. However, systematic testing of all possible drug combinations is infeasible, due to the massive combinatorial search spaces.
To enable systematic tailoring of personalized drug combination for each patient individually, researchers from the University of Helsinki (FIMM and Hematology Research Unit Helsinki) and RWTH Aachen University joined their efforts and developed a computational-experimental drug combination prediction and testing (DCPT) platform. The platform combines information from genomic and drug sensitivity profiles and prioritizes the most potential combinations for further pre-clinical testing.
An example of 2D synergy matrix (left) and 3D synergy landscape (right) of a patient-specific combination between venetoclax and tacrolimus. Credit: American Association for Cancer Research
We are living the era of big data and artificial intelligence. Machine learning enables us to solve many challenging problems in life science, and DCPT platform is a fine example of that. I am currently continuing the development of the platform to further improve its predictions.
- PhD student Liye He from FIMM, the lead author of the study
As the first case study, DCPT platform was applied to predict synergistic combinations for patients with T-cell prolymphocytic leukemia a rare and aggressive blood cancer of mature T-cells with median overall survival of less than one year. The platform successfully predicted distinct synergistic combinations for each of the tested patients, each presenting with different resistance patterns and synergy mechanisms.
The platform is the outcome of a successful collaboration between computational, biological and translational researchers. It was great to see that the DCPT platform worked successfully in the real leukemia cases.
- Satu Mustjoki, Professor of translational hematology, Hematology Research Unit Helsinki, University of Helsinki
The platform makes use of drug-target interaction networks from Drug Target Commons, and combines exome-sequencing and RNA-sequencing profiles with drug sensitivity score to predict patient-specific drug combination effects using a network pharmacology-based machine learning model. Importantly, DCPT makes use of healthy control cells to prioritize cancer-selective synergies to avoid combinations with overlapping toxicities.
To facilitate mechanistic interpretation and clinical translation, the DCPT platform also implements patient-customized network visualizations to suggest potential biomarkers for the observed combination synergies. Visualization combines data from the patient-specific genomic aberrations with information on the targets of the compounds having synergy in the patient sample.
This pilot case study successfully showed the power of DCPT platform for predicting synergistic and selective drug combinations for cancer patients, and we are currently applying the DCPT platform to other types of leukemia patients and later to solid tumors.
- Professor Tero Aittokallio from FIMM
The developers have made the source-code and web-applications implementing the DCPT platform freely available, so that the cancer researchers can apply and further develop the platform for their own applications. Please see the links below.
Liye He, Jing Tang, Emma I Andersson, Sanna Timonen, Steffen Koschmieder, Krister Wennerberg, Satu Mustjoki, Tero Aittokallio. Patient-customized Drug Combination Prediction and Testing for T-cell Prolymphocytic Leukemia Patients. Cancer Research 2018.