Bart Westerman

DR.

20012021

Research activity per year

If you made any changes in Pure these will be visible here soon.

Personal profile

Research interests

Identifying the relation between therapy combination efficacy and tumor heterogeneity for brain cancer

Cancers are commonly treated with unspecific and rather toxic drugs such as the alkylating drug temozolomide. These chemotherapies affect healthy and diseased cells thereby resulting in many adverse effects. More recently, new therapies have been developed that target specific proteins that are directly involved in tumor formation and tumor growth. While these new drugs are better tolerated by the patients, they tend to become less effective because mutations can occur in the target protein or because patients have a genetic make-up that causes therapy resistance (Saleem and Westerman et al, 2019). Therapy using synergistic combinations of drugs is attractive because more effective therapies can be applied. Selection of drug combinations is however not a trivial task since hundred thousand combinations can be made with clinically approved drugs. By using a bioinformatics strategy we have predicted and subsequently validated drug-drug synergies in vitro based on a Vonoroi mapping strategy that we call the drug atlas. By correctly predicting synergistic, multi-drug combinations, we have a powerful methodology to enable multi-drug therapy and this enables identification of biomarkers that predict synergy-sensitivity.

 

Ongoing projects are:

AI-IMPACT (NWO funded): Decision support for polypharmacology Most tumor patients have more than one kinase mutation that drives their tumor. We have shown that relative high sensitivity of poly-pharmacological drugs for cell lines that harbour two matching activating mutations. To optimally adapt therapies to mutations, we have generated a target predictor for each kinase inhibitor. Based on the availability of 3000 drug-kinase structures present in the KLIFS (www.klifs.vu-compmedchem.nl), we developed a convolutional neural network prediction model to predict the target fingerprint of 150,000 kinase inhibitors. This will enable to match bioactivity (target) fingerprints to personalized (mutation) fingerprints. 

THE TOXICITY ATLAS (NWO funded): Decision support for undesired combination therapies Therapy combinations with desirable efficacies might not be easily translated for clinical use given the potential toxic effect of drug combinations. We provide a rationale for selecting therapy combinations aimed to provide the best possible quality of life for the patient. This is expected to enable further implementation of personalized combination therapies in the clinic. Our approach based on the toxicology atlas, forms a global representation of different responses of the human body to FDA approved drugs. This representation will guide us to which vulnerabilities such as additive toxicity have to be avoided.

GENE-ATLAS: Decision support for tumor evolution Intratumoral heterogeneity plays a dominant role in tumor evolution and is considered the major cause therapy resistance. We performed a comprehensive analysis of 16 different tumor types of 10,000 patients of whole-exome sequencing and copy number variation (CNV), obtained from cBioPortal for Cancer Genomics. This showed that directional evolution in combination with onset mutations are commonly found in 5% of the tumors. Genes that are co-mutated with converging mutations show overlap with genes that are commonly co-mutated within cancers. Based on this information, we have developed a prediction model for convergent evolution which is expected to guide future progression-prevention therapies.

 

External positions

VU/UvA Lecturer Bioinformatics, Amsterdam University College

1 Jan 2020 → …

Network

Recent external collaboration on country/territory level. Dive into details by clicking on the dots or