Cancer treatment is one of the most intricate challenges of contemporary medicine. One complication that often arises is the trial and error prescription of drugs that are often ineffective against a given type of tumour or for a particular patient. Moreover, these treatments often produce exhausting side effects.
The ability to identify the type of tumour and develop targeted treatment unique to each patient can dramatically increase both their survival rate and quality of life. This approach to treating patients based on individual characteristics is called precision medicine.
A recent initiative by the Research Institute of the McGill University Health Centre (RI-MUHC) and MEDTEQ, a major Canadian medical technology organization, aims to integrate current treatment methods like immunotherapy and chemotherapy, precision medicine principles, and artificial intelligence to achieve a personalized approach to cancer treatment.
“Cancers of all types are heterogeneous,” Dr. Peter Metrakos, head of the Cancer Research Program at the RI-MUHC, said in an interview with The McGill Tribune. “Every tumour [has] a unique set of mutations and a unique set of drivers. If we want to be successful, we are going to have to be able to stratify them and uniquely target them.”
To achieve this, researchers are looking for biomarkers in patients’ blood that are linked to specific types of tumours. Cancer cells release extracellular vesicles into the bloodstream that contain proteins and genetic material such as DNA and RNA. Examining the content of these vesicles can indicate tumour identity and help doctors develop targeted treatment plans.
However, once the components of these vesicles are extracted from the blood sample and their protein and genetic content is sequenced, a significant challenge arises. Protein, DNA, and RNA sequences, in addition to a patient’s medical history, constitute a tremendous amount of data to be analyzed. No scientist presented with this amount of information could detect patterns, but a computer can. This is where developments in artificial intelligence come into play.
“The algorithm sees trends and is able to call them out,” Dr. Anthoula Lazaris, a scientist at the RI-MUHC who co-leads the project, said in an interview with The McGill Tribune.
My Intelligent Machines (MIMs), a Montreal-based leader in artificial intelligence, plans to use machine learning algorithms to perform a high-level analysis of protein and genetic sequences combined with clinical information. This method may uncover links between specific tumour types and biochemical signatures in the blood. Then, CellCarta, a company specializing in biomarker development, will develop tools to identify the presence of these signatures from a simple liquid biopsy taken from the patient.
This collaborative research initiative could transform cancer care by reducing precious diagnostic time and ensuring more targeted treatment.
“The patient walks into the clinic [where] we take a blood sample, run an assay, find a signature and identify the unique features of the tumour,” said Lazaris. “Combined with the patient’s clinical profile, [we use this information to] tailor the treatment accordingly,” Lazaris said.
It would seem the current process of trying different drugs and readjusting prescriptions based on the outcomes will soon become obsolete. Doctors will be able to better identify the most effective treatments based on signatures identified in the patient’s blood, increasing the chance of successful treatment and making cancer care less debilitating for the patient.
Metrakos explained that the findings of their study represent an important milestone towards a major shift in cancer treatment strategy.
“What we should go towards is a tumour-agnostic approach, which means that you don’t care where the tumour comes from,” Metrakos said. “You look at its mutations, you look at its drivers, you look at its protein makeup, and you target that, rather than where it comes from.”