Industry, Supercluster invest CA$17.8M to better diagnose cancer recurrence and relapse

Canada’s federally-funded Digital Technology Supercluster is joining with industry to invest CA$17.8 million (US$13.9 million) in an artificial intelligence (AI)-powered technologies program to advance personalized treatment for patients with cancer. Led by Vancouver, British Columbia-based Imagia Canexia Health Inc., the Candetect project is developing precision oncology software that Imagia CMO David Huntsman said will help monitor resistance to cancer therapy and treatment relapse in individual cancer survivors.

“There’s been a huge international effort to personalize cancer care decisions using information from a patient’s tumor and doing sequencing or analysis to say `This is the way your cancer should be treated,’” Huntsman told BioWorld. “The problem for most patients is precision oncology stops at this point. The Candetect project allows us to use some very sophisticated genomic tools, AI and cloud-based informatics to bring precision approaches into post management care.”

Treatment relapse and resistance to therapy place enormous strain on patients recovering from cancer and on Canada’s notoriously long cancer care wait lists. Huntsman said many patients benefit from precision oncology decision making “that is fantastic, there’s massive progress being made. But longer term follow up of patients becomes very generic after that.”

What’s needed is closer monitoring of treatment response in patients who may relapse or resist treatment post therapy, Huntsman said. As part of the Candetect project his technology will enable clinicians to perform real-time assessment of tumor status, detect earlier signs of patient relapse and recommend alternative treatment options.

The starting points for this work are the traditional “wet lab” procedures at Imagia’s Vancouver location and “dry lab” analytics, i.e., sophisticated, cloud-based informatics with elements of machine learning “where de-identified data is uploaded and analyzed and only the clinically relevant findings come down to populate a patient report. This can then be used to inform care,” said Huntsman.

By “informed care” Huntsman also means more localized care. Medical institutions typically operate very pragmatic schedules that direct patients to return for follow-up care, he explained, “but these schedules aren’t optimized to individual patients,” said Huntsman. That’s particularly so for patients living at remote distances, he noted.

“We believe there’s a great benefit in removing geographic access barriers and just making it easier for patients to get their care by simply making a simple blood test at their local clinic. In effect, the goal “is to make survivorship a precise decision-making process,” Huntsman said. Patients will be monitored “to determine has their cancer stayed away and if it’s come back how has it changed and what treatment decisions should be made next.”

Going deeper
Imagia’s assays rely on AI-driven machine learning to identify cellular mutation in DNA sequences from plasma samples. But what is missing generally across health care, Huntsman said, is “the big clinical decision support space everybody had been anticipating” with the advent of AI.

Here, having enough data derived from AI for a health care system to meaningfully contribute to understanding disease is only one challenge. “The bigger challenge is how to safely share data to treat patients who are already in the system,” said Huntsman. This becomes particularly salient addressing relapse and patient resistance to treatment.

“Candetect is not focused on determining if a healthy person has suddenly got cancer, but to identify features of relapse which can lead to a change in treatment,” said Huntsman. “When you set up a system like this you need what’s called `orthogonal’ data, i.e., data from another source against which you measure your progress.”

Here, that source is represented by many other patients with cancer who have been previously tested. “As more and more data gets processed using AI and machine learning the fidelity of the system and accuracy improves with time,” said Huntsman. Eventually new technology like Imagia’s testing software will outstrip any existing gold standard in analytics validation.

“In other words, the test system is better than the validation system and you have to get cleverer and cleverer in terms of how you get the orthogonal data to validate your findings,” said Huntsman.

What is also required is enormous learning of the human variety, notably the vast amount of data derived from human genome technology. Genomic readouts provide a very broad “almost impressionist view of what’s going on in a solid tumor cancer,” Huntsman explained. You then combine this “with a very high- fidelity view of the particular mutations which you can treat. This is the way forward.”

That way forward is underwritten by CA$12.5 million (US$9.75 million) from Imagia, Toronto-based Dnastack Inc., Microsoft Inc., and CA$5.3 million (US$4.1 million) co-invested through Ottawa’s Supercluster’s Technology Leadership program. Those contributing in-kind resources are BC Cancer Research, Toronto-based Dnastack, Microsoft, Queen’s University and Toronto’s University Health Network.

The project’s immediate goal is to introduce diagnostic tools like precision oncology software at the clinic level “that are at least as good or better than what we would consider the standard of care for monitoring patients,” said Huntsman.

The longer-term efficacy and optimal frequency of the technology’s use “will require prospective clinical studies done with our assays and others to really understand how best to use this technology and positively impact patient care.”

Back