Researchers from the Wellcome Sanger Institute in Cambridge, U.K. employing a comprehensive analysis of CRISPR screens, multiomics data, and protein–protein interaction networks have identified 370 candidate priority drug targets across 27 types of cancer.
Their findings, published in Cancer Cell, have the potential to at least double the proportion of patients who may be eligible for gene targeted therapy.
First author Clare Pacini and colleagues explain that at present only 14% of patients with cancers are eligible for treatment with an approved genome-targeted therapy. However, they estimated that 60% of individuals from a cohort of nearly 7000 patients with 20 different types of cancer had a least one marker associated with a priority target from their analysis. For targets that already have an approved or preclinical drug, the proportion was 32%.
In the most comprehensive study of its kind, the Wellcome Sanger Institute, Open targets—a public-private collaboration working on improving drug discovery—and other collaborators carried out CRISPR-Cas9 screens on targeting 17,647 genes in 930 cancer cell lines derived from 27 cancer types.
They then used multiomics data to analyze relationships between molecular biomarkers and cancer dependencies, that is, weaknesses within different cancer types that cancer cells rely on to survive. Finally, they looked at protein–protein interactions to explore how dependency-marker pairs fit into known networks of molecular interactions within cells, providing clues as to how cell biology is disrupted by cancer, and which targets might yield the most effective therapies.
This resulted in 370 unique priority cancer drug candidates for the 27 cancer types, of which 302 were cancer-type specific.
The work provides a clearer understanding of which types of cancer can possibly be treated by existing drug discovery strategies and pinpoint areas where novel and innovative approaches are needed. This is particularly important when drug development is known for its high costs and poor efficiency, with failure rates of around 90%.
Pacini et al write that their study “highlights the value of multiomic datasets and provides new insights into the landscape of cancer dependencies.”
They continue: “Through a rigorous data-driven approach, we integrate an unprecedented wealth of genomic, experimental, and clinical data to nominate candidate oncology targets for further consideration to fulfill the ultimate aim of enabling personalized treatments for patients based on their tumor molecular profile.”
Dr Francesco Iorio, co-lead author of the study from the Computational Biology Research Centre of Human Technopole, said: “Analyzing the largest-ever cancer dependency dataset, we present the most comprehensive map yet of human cancers’ vulnerabilities – their ‘Achilles heel’. We identify a new list of top-priority targets for potential treatments, along with clues about which patients might benefit the most – all made possible through the design and use of innovative computational and machine intelligence methodologies.”
Commenting on the findings, Dr. Marianne Baker, science engagement manager at Cancer Research UK, added: “Two people might have the same type of cancer, but their diseases can behave differently. That is why we need precision medicine. This ambitious work is a compelling example of research informing drug discovery from the start, paving the way for more effective precision cancer therapies. Giving people treatments for their unique cancer can improve the odds of success and help more people affected by cancer live longer, better lives.”