Pharma R&D Today
Ideas and Insight supporting all stages of Drug Discovery & Development
Predicting Cancer Drug Combinations That Work
Posted on August 6th, 2020 by Xuanyan Xu in Pharma R&D
Dr. Bart Westerman is a Group Leader and Assistant Professor at the VU University Medical Center in Amsterdam. There he heads research projects in neuro-oncology, as well as in the use of bioinformatics to tackle complex, multidimensional questions to transform cancer therapy decision-making and management. He and his colleagues recently published a cancer drug atlas that predicts synergistic combination therapies for cancer.
Asked what triggered the idea of a cancer drug atlas, Bart starts with a question he’s asked by cancer clinicians: “I have this patient who is not reacting to therapy. What should I do?”
“The problem,” he clarifies, “is that there are a lot of therapy options but only a few will be successful. Determining that success is determined by a myriad of factors that are not only limited to the efficacy of the drug itself.” Off-target effects and the susceptibility of the patient’s specific cancer are just two such cues. The picture is increasingly complex when clinicians contemplate combining two or three drugs.
While dose-response data are available for individual medications, comparable data on specific combinations are rare. “There are roughly 100,000 possible combinations,” says Bart. “So, we asked, how can we predict the effect of a drug combination based on responses to each individual drug?”
Of drug distance and synergistic effects
Bart’s solution lies in simulating the complete landscape of a tumor’s vulnerabilities to different drugs. Bart describes it as a balancing game: “The atlas uses dose-response data to map out key vulnerabilities in tumors and how these are related. The resulting patterns represent an equilibrium of interactions specific to a tumor, and when you interfere with these vulnerabilities simultaneously, strong therapy effects can be observed. The atlas allows you to look at the effect of triggering several vulnerabilities.”
Combination therapies are aimed for synergistic effects, where two or more drugs have a greater effect than the sum of each individually. The cancer drug atlas captures the potential for synergy with the concept of drug distance. Mapped to the atlas, dose-response data cluster by similarity of effect. Two drugs that impact different processes will be found further away on the atlas.
“The greater the distance between two drugs,” explains Bart, “the more likely you are looking at independent pathways. If a tumor is sensitive for both drugs, you can hit these underlying independent processes at the same time and have a stronger effect.” Nearly 500 published cases of successful drug synergies matched interactions mapped in the atlas, validating the model and the drug distance concept.
The nitty gritty of development
The cancer drug atlas draws on data from several sources. Preparing their integration and defining the landscape was a tremendous undertaking. “Should the sensitivity space be on a log scale? Do we use Euclidean distance between clusters? How do we account for differences in scale between drugs and tissues?”
Bart adds: “These were difficult questions, complicated by artifacts in data that can be misleading at the start of any project. The key is to rule those out. Then you build a strategy that offers the most coverage of the available data and integrate as much of the rest by normalization. Finally, you benchmark your work with trusted data to see if you’re going in the right direction. Development work is underestimated, but you take each step to reach an end point,” he says, referring back to the question where it all started.
Where does the journey go next?
Response to the publication of the cancer drug atlas has been very positive. Understandably, hopes are high about benefits for cancer patients. But Bart warns: “We are still a long way from implementation in the clinic.” His team has planned a path toward application, but it requires time and collaboration with industry and healthcare institutions.
In the meantime, Bart is working avidly on a toxicity atlas and the use of artificial intelligence to optimize and integrate it with the drug atlas. Stay tuned for more as we talk to Bart about the next steps in his work and how a community of researchers across sectors and disciplines could catalyze clinical uses of data-based tools.
R&D Solutions for Pharma & Life SciencesWe're happy to discuss your needs and show you how Elsevier's Solution can help.
Sr. Marketing Manager, Life Sciences Audience at Elsevier
- AI Disease Modeling Supports Precision Medicine for Cancer
- Model-based Strategy to Guide the Choice of Clinical Doses for ADC
- Bio-IT World Conference 2020 presentation: A Network Polypharmacology Approach to Diffuse Intrinsic Pontine Glioma
- Rapid Growth and Promising Partnerships
- Finding Support and Success Along the Development Path