AI-assisted interventions
University of Waterloo tackling cancer with personalized immunotherapy
Immune therapy is a powerful new way to treat cancer, harnessing the body’s natural defences to find and kill cancer cells.
By applying artificial intelligence (AI), scientists at the University of Waterloo are working to strengthen this mechanism, making it possible to develop personalized cancer-fighting drugs.
“Every patient is different, and every cancer is different – so cancer treatment shouldn’t be the same for all,” says Ming Li, a university professor in Waterloo’s Cheriton School of Computer Science. “Treatment should be tailored to the patient, and that’s what our machine-learning model allows us to do.”
Immunotherapy is quickly becoming a fourth option in cancer treatment, alongside surgery, chemotherapy and radiotherapy, says Dr. Li, who holds the Canada Research Chair in Bioinformatics. He says AI tools make it possible to target the disease in a more precise and even personalized way.
Cancer is a disease of uncontrolled cell growth and regulation, caused by mutations to the genes that control cell division, growth and differentiation. The body’s immune system protects against cancer through T-cells. By applying machine learning, Dr. Li’s team identifies specific treatments to enhance those T-cells in individual patients.
“What we are doing is making personalized cancer vaccines, as one type of immunotherapy, affordable to common people.”
Dr. Li was inspired to apply his expertise in AI to cancer treatment when his wife developed breast cancer in 2004 and passed away six years later. “I learned a lot during this period of time, attending conferences and reading up on how to cure cancer,” he recalls. “We tried very hard, but it didn’t work.”
In 2017, he read about cancer immunotherapy in an article in the scientific journal Nature Biotechnology. The article said there was a need to discover “neoantigens,” small pieces of protein on the surface of cancer cells, through a technique called mass spectrometry, which identifies molecules based on their weight, Dr. Li says. “So I worked on this problem for five years and finally we have a solution.”
The key is to determine the amino acid sequence of the antigens in order to make vaccines that “train” the body’s T-cells to target the cancer, Dr. Li says. “Immunotherapy has to be very personalized, because my peptides are very different from your peptides.”
His team studies samples of cancer obtained from biopsies, using mass spectrometry to sequence the peptides. Machine-learning software called DeepNovo is used to rank the peptides and pinpoint the mutated ones present. These “immunogenic” peptides can be used in cancer vaccines that boost the immune system and eliminate the cancer through the body’s T-cells.
Each person can have more than 1,000 peptides, but only a few are immunogenic, Dr. Li points out. Using bioinformatics tools, these special peptides can be found in just a few days at a cost of perhaps $10,000, which is relatively swift and inexpensive, he comments. Dr. Li is part of a company that is already providing this service using cancer biopsies taken in hospitals in different parts of the world. At the same time, his team of researchers is learning more about the mechanism and developing ways to make it more viable.
“I think in the near future, we will not be too worried about many types of cancers; we can just depend on our own defensive system to kill them,” he says.
Dr. Li expects his solution “will be part of the pipeline for personalized immunotherapy.” While his technology is available today, its widespread application in the medical field could require years of clinical trials, he notes. Meanwhile, however, there have already been some uses in late-stage cancer patients.
“What we are doing is making personalized cancer vaccines, as one type of immunotherapy, affordable to common people,” Dr. Li adds.
To view this report on The Globe's website, visit globeandmail.com
To view the full report as it appeared in The Globe's print edition Excellence in
research & innovation