close

Join our Mailing List

Receive the latest news, event invites, funding opportunities and more from the Ontario Institute for Cancer Research.

New treatment for recurrent pediatric brain cancer enters clinical testing

OICR-supported study helps move promising CAR-T cell therapy into a first-in-child clinical trial

Recurrent brain tumours are some of the most difficult cancers to treat, with no approved targeted therapies available and only a few potential therapies in clinical trials. Developing new drug treatments for these tumours is challenging in part because the drugs must overcome the blood-brain barrier and specifically target cancer cells while sparing the surrounding critical regions of the brain. Scientists at The Hospital for Sick Children (SickKids) have discovered a new solution.

In a study, recently published in Nature Medicine, a SickKids-led research team describes a novel treatment approach that delivers chimeric antigen receptor T (CAR-T) cell therapy directly into the cerebrospinal fluid that surrounds the tumour. Their findings show that the approach was effective in treating ependymoma and medulloblastoma, two common types of brain tumours, in experimental mouse models of human disease.

“The vast majority of children with recurrent metastatic medulloblastoma or ependymoma currently have a deadly prognosis, so it is very exciting to think we have identified a novel approach to treat this underserved patient population,” says senior author Dr. Michael Taylor, Neurosurgeon, Senior Scientist in the Developmental and Stem Cell Biology program and Garron Family Chair in Cancer Research at SickKids and Co-lead of OICR’s Brain Cancer Translational Research Initiative.

CAR-T cell therapies, which use genetically engineered immune cells to attack cancer cells, are remarkably effective in treating certain types of lymphomas and leukemias. Whereas CAR-T therapies are typically delivered through the blood stream, the research team discovered that delivering their engineered T cells directly into the cerebrospinal fluid provided a better chance for the therapy to reach and eliminate brain tumours.

The team performed in-depth molecular studies to design CAR-T cells that can recognize specific molecules on the surface of brain tumour cells. They also found that the use of a complementary approved cancer medication, azactyidine, boosts the efficacy of their approach.

Now, building on these findings, collaborators at Texas Children’s Hospital have launched a first-in-child clinical trial to test the safety and anti-tumour efficacy of their new strategy.

“This work was possible thanks to the concerted collaboration of our Pediatric Cancer Dream Team, which brought together scientists studying tumor genomics and tumor immunotherapy around the world to enable the design of more effective therapies for children with incurable and hard to treat cancers,” says corresponding author Dr. Nabil Ahmed, associate professor of pediatrics and immunology, section of hematology-oncology at Baylor and Texas Children’s Hospital.

This research was supported in part by OICR through OICR’s Brain Cancer Translational Research Intitiative and funding provided to the Stand Up to Cancer (SU2C) Canada Cancer Stem Cell Dream Team.

Read SickKids news release.

A message from OICR’s President and Scientific Director, Dr. Laszlo Radvanyi

Dr. Laszlo Radvanyi

I hope that everyone has continued to stay safe and healthy as the world continues to grapple with the risks and challenges presented by COVID-19. The impacts of the pandemic have been felt by individuals and organizations across society, including cancer patients and Ontario’s cancer research community.

While things are obviously not business as usual, I am happy to see OICR’s people rise to the challenge and find solutions to allow us to continue to focus on cancer research while working remotely. My thanks go to OICR’s staff, Board and Scientific Advisory Boards, collaborators and others who have quickly adapted to continue our work as best we can. A big thank you also to our funders at the Ministry of Colleges and Universities for their continued support. We will gradually restore our onsite cancer research activities in a manner that will ensure a safe work environment for all our onsite staff. Our priority remains to improve the lives of those with cancer through research.

OICR’s leadership recognizes that the pandemic has resulted in unprecedented challenges for cancer researchers across Ontario. We have taken steps to ease this burden and are working with OICR-funded researchers and partner organizations to overcome these challenges together. More information about how we are assisting our funded researchers can be found on our website.

Due to our collaborative, cross-disciplinary research strengths, OICR is well-situated to contribute to COVID-19 research. OICR researchers are engaged in numerous projects with others in Ontario and abroad. It has been heartening to see such a swell of collaborative spirit and to see the research community doing what we can to help overcome COVID-19. I invite you to visit our website to learn more about how OICR is doing its part. We are especially cognizant on how these research activities impact cancer patients, as they are an especially vulnerable population at this time.

COVID-19 has disrupted cancer research on a global scale. I look forward to a time when we can resume all of our research activities and once again contribute to the international campaign against cancer at full capacity. During the pandemic, cancer has not and will not cease to be a reality for the thousands of Ontarians living with this disease and their families. Everyone at OICR remains steadfast in our commitment to improve the lives of those facing cancer.

In closing, I offer my deepest appreciation to all those working on the front lines of this crisis and thank all off the members of Ontario’s cancer research community for their continued dedication during this difficult time. All our thoughts also go out to any families that have been affected during this crisis.

Sincerely,
Dr. Laszlo Radvanyi
President and Scientific Director
OICR

Ontario Cancer Research Ethics Board welcomes Natascha Kozlowski as new Executive Director

The province’s oncology-specialized research ethics board honours Janet Manzo’s contributions as she retires and welcomes new Executive Director, Natascha Kozlowski

Since 2006, two years after its inception, Janet Manzo has led the Ontario Cancer Research Ethics Board (OCREB) as Executive Director. Through her tireless commitment to research ethics, she has established OCREB as a leading central research ethics board for the province that is also widely recognized across the country for its innovative model and approach to research ethics. Manzo has recently announced her retirement and joins OCREBs members its Advisory Committee in welcoming Natascha Kozlowski to succeed her as Executive Director.

Kozlowski, who is the former Director of Research at Lakeridge Health, a five-site hospital system serving Durham Region, brings nearly two decades of experience in clinical research to OCREB’s leadership team.

We sat down with Manzo and Kozlowski to discuss changes in research ethics since OCREB was established and OCREB’s next chapter.

How have cancer clinical trials and ethics processes evolved in Ontario since OCREB began?

Janet Manzo, retiring OCREB Executive Director.

Janet Manzo (JM): OCREB was essentially like a start-up. When it was created, there was no model like it in Canada. OCREB has radically changed the research ethics environment for multi-centre cancer trials in Ontario. Today, OCREB enables research by streamlining the review process, minimizing redundancy in hospitals across Ontario, promoting consistency, and saving time and costs by serving as a specialized, consolidated committee for ethics reviews.

Looking back over these years, it’s clear that cancer clinical trials have become more complex. For example, we’ve seen new and innovative study designs, more and more thorough consent forms, more frequent inclusion of quality-of-life assessments, and increases in biologic specimen collection for biomarker development, genetic testing and future research. I’m grateful to have had the opportunity to lead OCREB operations for more than fourteen years, a period of constant growth and change, leading to a successful and well-respected model of ethics review for multi-centre cancer research.

Natascha Kozlowski (NK): From my experience as the Director of Research at a large hospital, I witnessed – on the ground level – how OCREB positively impacts cancer research at hospitals. I can attest that OCREB has helped improve research ethics processes at sites across Ontario while enabling life-saving, ethically-sound clinical research.

How does OCREB adapt to changes in cancer clinical trials?

JM: For research ethics boards, the increasing complexity of clinical trials means it is more important than ever to stay current and to have the right expertise around the table, such as experts in pathology or genetics. OCREB has and will continue to evolve as new cancer technologies and clinical trial designs emerge.

NK: To echo Janet: new clinical trial designs bring new ethical considerations to the table. I think Janet has set up a tremendous organization with a great network of support – OICR included – that can continue evolving to enable innovative research.

JM: I’d like to thank all of the OCREB staff, advisors and members for their unfailing support and for their steadfast dedication to the protection of research participants. I will miss everyone but I am confident that OCREB is in good hands.

What does the future hold for OCREB? What are you looking forward to the most?

NK: OCREB has a strong history of excellence in the clinical trial environment, often being consulted as a respected source of ethics guidance. I look forward to working with our Members and Advisory Committee in the years to come to uphold and strengthen OCREB’s reputation while advancing cancer research.

I would like to express my sincerest thanks to Janet Manzo for many years of dedication and leadership in building an outstanding research ethics board. Over the years, Janet has earned the trust and respect of OCREB’s many members – including clinicians, researchers, ethicists, privacy experts, and community members – while serving Ontario’s cancer research hospitals and centres. I wish Janet all the best as she begins her retirement.

Learn more about OCREB.

University students step up to flatten the COVID curve

Former OICR intern leads the development of a COVID-tracking site used by more than 400,000 people in Canada to date

Flatten is quickly becoming a go-to source of information about how COVID-19 is spreading across Canada.

In less than a month, more than 400,000 people have submitted data on their symptoms, travel history, age and medical conditions, making Flatten the country’s leading crowdsourced COVID data repository.

Behind the project is a team of first- and second-year university students who are determined to help.

Yifei Zhang, Vice President of Flatten.

“We just wanted to put our technical skills to good use during this time,” says Vice President of Flatten, Yifei Zhang, in a University of Waterloo story. “It’s been great working together with everybody trying to build a platform that will be useful for Canadians across the country.”

As a web-based, data-gathering platform, Flatten provides a real-time heat map of self-reported confirmed and potential COVID-19 cases across the country. The platform helps increase awareness and flatten the curve of COVID-19 cases.

Over the last four weeks, Flatten has rapidly evolved from an idea into an incorporated non-profit organization, with support from advisors such as Dr. Geoffrey Hinton and sponsors such as Google Cloud, the Vector Institute and CIFAR.

The team behind Flatten has established collaborations with health authorities across Canada, such as in Montreal, and plans to work with other municipal governments and provinces..

“We work with leading advisors and collaborators to make sure we’re surveying the right questions and providing the right information for Canadians today to help flatten the curve,” says Zhang.

Zhang, who is completing his second year as a software engineering student at the University of Waterloo and leads Flatten’s website development, attributes his website development knowledge to his internship with OICR’s WebDev team.

“My time at OICR reinforced my interest in working in health and biology, giving me the motivation and drive to pursue this initiative,” says Zhang. “At OICR, I gained experience working with a high volume of data using robust techniques and I was able to bring that knowledge into developing Flatten.ca. A lot of the fundamentals we used to build this site came from best practices that I learned from my term at OICR.”

Learn more at flatten.ca.

Measure up: Taking strides to automate cancer pathology practices

Dr. Tamara Jamaspishvili

OICR-supported researchers quantify common prostate cancer outcome predictor

Advances in cancer research have opened the door to new tests to better assess tumours and help recommend the most appropriate course of treatment for a patient. Research pathologists play a critical role in turning scientific knowledge into tests that can be used in an everyday clinical setting.  

“Scientists are constantly advancing our understanding of cancer, but that understanding cannot help patients unless it’s applied in practice,” says Dr. Tamara Jamaspishvili, Research Pathologist at Queen’s Cancer Research Institute. “Our role as research pathologists is to bridge that gap, and transform discoveries into more accurate diagnoses and prognoses for patients that could be implemented and actionable in practice.” Jamaspishvili’s work is supported by the Ontario Molecular Pathology Research Network, an OICR-funded province-wide network that conducts high-quality cancer research focussed on clinical impact.

An example of the challenge of clinical translation is found in PTEN testing. PTEN is a cancer-preventing gene that – when absent in a cell – may lead to uncontrolled tumour growth. Research has shown that the loss of PTEN within a prostate tumour could help predict the severity of a man’s prostate cancer, but PTEN is not routinely tested.

“Simply put, some cells in a tumour sample may have PTEN loss and some cells don’t, but nobody has clearly quantified how the ratio of cells with or without PTEN contribute to a patient’s health,” says Jamaspishvili.

Jamaspishvili teamed up with collaborators to address the subjectivity of PTEN testing. Her collaborators include Drs. David Berman, Palak Patel, Robert Siemens, Paul Peng, and Yi Niu from Queen’s Cancer Research Institute, Drs. Fred Saad and Anne-Marie Mes-Masson from the University of Montreal, Dr. Tamara Lotan from Johns Hopkins University, and Dr. Jeremy Squire and colleagues at the  University of São Paulo.

Their study, recently published in the Journal of the National Cancer Institute, proposes a new quantitative approach to assess PTEN. They clarify how pathologists can predict the severity of a patient’s prostate cancer based on the number of cells with PTEN loss. These findings can help standardize PTEN testing, but their approach can also be applied to other pathology tests that are still highly subjective.

“Quantifying qualitative tests helps us move towards automated pathology techniques,” says Jamaspishvili. “This is the future of pathology.”

Jamaspishvili is now working to automate PTEN digital pathology analysis in collaboration with Dr. Stephanie Harmon and colleagues in Dr. Baris Turkbey’s lab as part of the National Cancer Institute’s Molecular Imaging Program.

“Now, we can apply machine learning image analysis tools to analyze PTEN loss and make better predictions for the benefit of patients. We look forward to using artificial intelligence in digital pathology to help fill the gaps between research and clinical practice.”

Read more about OMPRN.

Teaming up to decode DNA damage repair

Ovarian and pancreatic cancer researchers join forces to debunk which treatments work for which patients

Ovarian and pancreatic cancer are some of the most challenging cancers to treat but their common characteristics have pointed to new treatments for certain subsets of patients. Drs. Stephanie Lheureux and Grainne O’Kane have teamed up to find out which patients can benefit from these new therapies.

Over the next year, with the support of an OICR Translational Research Initiative (TRI) Collaboration Award, Lheureux and O’Kane will be taking a deeper look into patient tumour samples that have a specific DNA damage repair deficiency, called homologous recombination deficiency (HRD). These tumours are thought to be sensitive – meaning, they can be eliminated – with a certain class of drugs called PARP inhibitors, but it is difficult to predict in the clinic whether a patients tumour has HRD or not. Further, it is difficult to determine whether a patient will benefit from using PARP inhibitors.

Lheureux, who is a medical oncologist specializing in ovarian cancers, and O’Kane, who is a medical oncologist specializing in pancreatic cancers, have set out to perform whole-genome analyses on patients with HRD to find a better way to identify which patients may respond to PARP inhibitors. Both researchers are excited to tap into each other’s expertise.

“Dr. Lheureux cares for many patients facing these challenges,” says O’Kane. “She has deep clinical expertise in this area.”

“Dr. O’Kane and her closest collaborators have excellent expertise in whole genome sequencing and bioinformatics,” says Lheureux. “We’re eager to work together.”

Their analyses may help them understand the biological mechanisms driving HRD and how HRD tumours become resistant to treatment. Their findings may also extend beyond ovarian and pancreatic cancers.

“We want to define the biological response to PARP inhibitors and the mechanism of resistance so that we can help these patients make the best treatment decisions for their specific disease,” says O’Kane.

“We’re motivated to redefine HRD and understand it on a deeper level to help us overcome resistance to treatment and extend the lives of those with these cancers,” says Lheureux.

Lheureux and O’Kane’s collaboration is supported by OICR’s TRI Collaboration Award, a pilot funding stream to support the training of young investigators and encourage collaboration amongst OICR’s TRI teams.

Learn more about OICR’s Pancreatic Cancer TRI, Ovarian Cancer TRI or read about the latest TRI News.

Image credit: Background vector created by pikisuperstar – www.freepik.com

The donations behind the discoveries: The Ontarians who made the Pan-Cancer Project possible

A technician at an Ontario Tumour Bank site at Kingston General Hospital works with frozen specimens.

The Pan-Cancer project made international headlines this month, but not without the contributions of thousands of individuals and the teams that preserve their specimens

In an unprecedented, decade-long study of whole cancer genomes, OICR researchers and collaborators have improved our fundamental understanding of the disease, indicating new directions for developing diagnostics and treatments. The Project was powered by 2,800 people with cancer who donated their biologic specimens to research. These contributions were facilitated and protected by groups such as the Ontario Tumour Bank (OTB).

From the operating room to the freezer

Many advances in cancer research, like those made by the Pan-Cancer Project, rely on hundreds – and sometimes thousands – of biospecimens. A patient’s donated blood, tumour and surrounding tissue may hold clues to future innovations in cancer diagnostics and therapies. But without biobanks – the repositories that collect and care for biological samples – the clues within these donations may never be discovered.

“Good science is built on good data and good omics data can only be drawn from well-preserved tissues,” says Monique Albert. “The advancements made by the Pan-Cancer Project would not have been possible without the diligent work of biobanking teams.”

Albert is the Director of OTB – a provincial bioresource operating in partnership with four state-of-the-art hospitals and cancer centres across Ontario. OTB plays a quiet but crucial role between the patient and the researcher, providing the fundamental biologic resources that research is built on.

Lowering the temperature and raising the bar

Day-to-day, biobanking teams – like OTB – work to implement the highest standards of preservation. From the operating room to the freezer and back to the lab, these teams tirelessly strive to maintain the quality of patient samples to inform cancer discoveries. OTB has held and raised leading biobanking standards for over 15 years.

“When The Cancer Genome Atlas started, biobanks around the world promised thousands of samples, but only a fraction of these samples were adequate for research,” says Albert, referring to Libraries of Flesh: The Sorry State of Human Tissue Storage. “This served as a wake-up call for the sector to unite, share best practices and set higher standards together.”

At the launch of The Cancer Genome Atlas (TCGA) in the early 2000s, OTB was up to – and in many ways exceeded – existing biobanking standards. This was thanks to the foresight of Dr. Brent Zanke and Sugy Kodeeswaran, who recognized the importance of stringent biobanking practices nearly a decade before biobanking became popularized.

As the only Canadian repository that was able to contribute to TCGA, OTB allowed hundreds of people from Ontario to contribute to this international initiative and to subsequent studies like the Pan-Cancer Project.

Since its inception, OTB has collected more than 185,000 samples donated by more than 21,000 individuals from across Ontario, enabling these donations to have a greater impact today and for years to come.

“Each sample represents a trace of an individual’s life, and we’re honoured to care for these valuable donations to science,” says Albert. “When they’re preserved properly, they become a lasting resource with infinite value. We’re proud that the donations from Ontario patients are paving the way for better and more targeted cancer treatment.”

OTB plays a critical role in leading the development of Canadian biobanking standards through the Canadian Tissue Repository Network (CTRNet), and biobanking standards around the world through the International Society for Biological and Environmental Repositories (ISBER).

Read more about OTB’s research resources and how OTB is collaborating to improve biobanking around the world by visiting their website at ontariotumourbank.ca.

Tackling brain cancer from all angles

Dr. Jüri Reimand

The Terry Fox Research Institute (TFRI) announced today that Dr. Jüri Reimand, OICR Investigator, has been granted the Terry Fox New Investigator Award to support his research into the evolution of glioblastoma, a deadly brain cancer that often recurs after treatment, with no long-term cure.

“This is a terrible disease with a dismal prognosis. It is usually fatal within a year or two after diagnosis and current therapies mostly fail to halt its recurrence and progress,” says Reimand. “We are taking a data-driven approach to see if we can change the tide on this disease by mapping the evolutionary history of each tumor and identifying genes and pathways that could be targeted through new or existing drugs.”

Backed by TFRI support, Reimand and collaborators are creating a robust multi-omics dataset derived from samples of glioblastoma tumours, including those that have returned after initial treatment. The dataset will incorporate many types of layered data from each sample including whole genome sequencing data, RNA sequencing data and proteomic data.

Reimand, who has expertise in integrating complex datasets, will develop machine learning strategies to identify new potential targets for treatment. The tools and methodologies will be designed to be applicable to other cancer types and will be made freely available for the research community to use.

“We hope that our expertise in computational biology can help shed new light on glioblastoma recurrence by analyzing tens of thousands of genes, proteins and RNAs in complex interaction networks, and ultimately provide a small number of high-confidence targets for further experimental work and therapy development,” said Reimand.

This research is enabled in large part by Reimand’s partnership with Dr. Sheila Singh, a clinician-scientist at McMaster University in Hamilton.

“We are routinely generating large amounts of complementary data utilizing different platforms that are difficult to compare,” says Dr. Singh. “This is why we are so excited to collaborate with Dr. Reimand to decipher GBM recurrence, as he brings invaluable expertise in computational biology, bioinformatics and machine learning. Dr. Reimand’s multi-omics integrative analysis will deliver our PPG with target genes, pathways and drug interactions that will help us to identify new therapies and understand the complex mechanisms of GBM recurrence.”

Read more about Dr. Jüri Reimand’s work.

This post has been adapted from the original announcement made by TFRI.

New clues to cancer in the genome’s other 99 per cent

OICR leads more than 1,300 researchers from around the world in an unprecedented investigation into the dark matter of the human cancer genome.

Adapted from a story in OICR’s 2018-2019 Annual Report.

Three billion letters of code make up our complete genetic blueprint, yet everything we know about cancer to date comes from only one per cent of those letters.

What about the other 99 per cent? Could those regions be holding clues to new cancer solutions and cures? What could we find if we looked into this dark matter? Dr. Lincoln Stein wanted to find out – and he wasn’t alone.

In the fall of 2015, more than 1,300 investigators from the International Cancer Genome Consortium (ICGC) expressed interest in exploring these uncharted regions. Four years and hundreds of terabytes of data analysis later, they’ve found ways to map the evolutionary history of cancer, identified traces of the disease long before it is diagnosed, and elevated the world’s standards for genomics data sharing and research.

A collective goal, a collaborative feat

Jennifer Jennings

“When this project was first announced, we were delighted by the overwhelming interest,” says Jennifer Jennings, Senior Project Manager of the ICGC. She says that was when the scientific leadership of ICGC realized that a concerted effort was needed to address common computational and logistical challenges, leverage the strengths of collaborators and develop shared infrastructure to achieve the ultimate goals of this research.

They named this project PCAWG, the Pan-Cancer Analysis of Whole Genomes Project, also known as the Pan-Cancer Project , which would soon become the largest ever pan-cancer analysis of whole genomes and one of the largest coordinated cancer research endeavors to date.

Stein and a small group of scientific leaders took on the challenge of synchronizing research groups with similar research goals, strategically rearranging expertise and coordinating collaboration on an international scale.

“Organizing and bringing these researchers together was the greatest challenge,” says Stein, who is the Head of Adaptive Oncology at OICR. “Working with others may be slower at first and the benefits aren’t always evident, but the rigour of the resulting science and the progress made is greater than what any of us could do on our own.”

Turning data into discoveries

Dr.Lincoln Stein

PCAWG researchers went on to investigate more than 2,600 cancer whole genomes from ICGC patient donors across more than 20 primary disease sites such as the pancreas and the brain. They created the computational tools and established the necessary infrastructure to process and analyze more than 800 terabytes of genomic data in a standardized, accurate and timely fashion.

Powered by these tools, they were able to order the progression of genetic changes that lead to certain types of cancer and showed that these events may occur decades before diagnosis.

“For exceptional cases like in certain ovarian cancers, we were able to see these early events happening 10 to 20 years before the patient has any symptoms,” says Stein. “This opens up a much larger window of opportunity for earlier detection and treatment than we thought possible.”

Understanding the order of genetic changes that lead to cancer – or the probability that one will occur after another – may allow researchers to outsmart how a tumour evolves. This knowledge could help devise new strategies to treat these changes as they occur or prevent them from occurring in the first place, Stein says.

PCAWG researchers have also discovered common patterns in the distribution of genetic mutations that may point to new causes of cancer. Similar to the common genetic signatures associated with smoking and ultraviolet radiation, these patterns may point to unknown environmental or behavioural causes that, once fully understood, could be used to change course and help prevent cancer.

“The biological insights discovered through PCAWG have tremendously advanced our understanding of cancer genomics and we’re approaching a place where we know all the molecular pathways involved with cancer,” says Stein. “We’ve discovered the causes of two thirds of cancers that were previously unexplained — but this is just the beginning.”

Setting new standards for the future

Last July, PCAWG data were officially made available for the scientific community to use as a resource for future cancer research. The key PCAWG findings were recently published in a collection of more than 20 scholarly papers in Nature and its affiliated journals. An expected 40 additional papers relying on PCAWG data will be published within the next year alone.

PCAWG methodologies are now the world’s gold standard for whole genome data processing and analysis. They will continue to be used for years to come as more patient samples are collected and sequenced around the world. All related computational tools, including the data exploration and discovery tools, have been made publicly available.

“We made both the genomic data, and the computational pipelines to analyze it, free to use for the global cancer research community,” says Stein. “Now, others can analyze these data – or new data – at the same level as we have in the pursuit of new cancer research discoveries.”

Related links

AI algorithm classifies cancer types better than experts

Gurnit Atwal and Wei Jiao

Pan-Cancer Project researchers develop deep learning system that can determine where a cancer originates with better accuracy than human experts

If doctors know where a patient’s cancer started, they can better treat the disease. Unfortunately, this is not always possible, but AI could play a role in solving that.

In a study published today in Nature Communications, a Toronto-based researcher group developed a deep learning system that can accurately classify cancers and identify where they originated based on patterns in their DNA. The system could potentially help clinicians differentiate difficult-to-classify tumours and help recommend the most appropriate treatment option for their patients.

“We reasoned that there was something within the cancer’s DNA that could help us classify these tumours,” says Dr. Quaid Morris, OICR Senior Investigator and co-lead author of the study1. “But I didn’t expect our system to work at well as it does – in some cases, far better than pathologists.”

The team

The initiative began with the dataset: 2,600 whole genomes across 38 tumour types from the Pan-Cancer Analysis of Whole Genomes Project, also known as the Pan-Cancer Project or PCAWG.

Dr. Lincoln Stein, Head, Adaptive Oncology at OICR and member of the Pan-Cancer Project Steering Committee, and his team began to work with these data to identify patterns in a cancer’s genetic material that could help classify these tumours. To them, this was a perfect problem for AI.

When we started to collaborate, We realized we had something amazing.
– Wei Jiao

“Deep learning models excel when they’re trained on large amounts of data,” says Wei Jiao, Research Associate in the Stein Lab and co-first author of the study. “We had an incredibly large dataset to work with, the most comprehensive dataset of whole cancer genomes to date, but we also needed the machine learning expertise.”

The Stein Lab posted their progress on bioRxiv, an open-access repository for biology publications that have not yet been peer-reviewed, which in turn sparked the collaboration between his team and the Morris Lab – a group with deep machine learning expertise.

The system

The development of their deep learning system was not simple. They mined through terabytes of data looking for patterns in the type of mutations, the source of mutations and where mutations occurred in the genome, among other factors.

To their surprise, they found that patterns in driver mutations – the changes in DNA that are thought to ‘drive’ the development of cancer – were not useful in determining where the tumour originated. Instead, they found that patterns in the distribution of mutations and the type of mutation within a patient’s sample could better classify the patient’s disease.

“We knew that we could distinguish between two different types of healthy cells by looking at how the DNA within the cell types are packaged,” says Stein, who is a co-lead author of the study. “We were surprised and gratified that we could do the same using cancer cells.”

“We saw that the tightly-packaged sections – also known as the closed chromatin – would have many more mutations than the loosely wound sections,” says Gurnit Atwal, PhD Candidate in the Morris Lab and co-first author of the study. “It was like the normal cell was casting a shadow on the cancer cell, and we just had to read the shadows.”

To achieve the highest accuracy, the research group developed a deep learning neural network-based system, a type of system that is loosely modeled after the human brain and commonly used to recognize patterns in images, audio and text. Their system achieved an accuracy of 91 per cent – roughly double the accuracy that trained pathologists can achieve using traditional methods when presented with a primary tumour and no clinical information.

Further, they tested their model on an additional 2,000 tumours from patients in the Netherlands who donated their cancer genomic data to the Hartwig Medical Foundation and the system still performed with a remarkably high level of accuracy.

 “As more cancer genomes are sequenced, we can gain the ability to classify rarer cancers,” says Atwal. “Where we are now is great, but there is more work to be done.”

The potential

This study presents a deep learning system that could potentially improve how cancers are classified, enhancing the accuracy of current diagnostic tests and the treatment decisions they inform.

For some patients, this system could tell them where their cancer began, giving them valuable information about which course of treatment to choose. The system also could serve as a tool to help doctors identify whether a tumour in a patient who has been treated for cancer in the past is an entirely new tumour or a recurring tumour that has spread.

“A treatment plan for a cancer that originated in the throat may be very different than one for that originated in the breast, and the treatment for a cancer that has returned is different than for one that has metastasized,” says Atwal. “One day, our tool could help give doctors the power to distinguish these classes of tumours, giving patients valuable information that wouldn’t have been available otherwise.”

The authors of the study suggest that their system could start helping patients soon. They plan to further refine their system for patients with rare cancers before moving towards clinical studies. 

“The potential impact of the system we’ve developed is encouraging,” says Morris. “We look forward to turning this system into a tool that can help clinicians and future cancer patients tackle this disease.”

1Morris is also a Canada CIFAR AI Chair, Faculty Member at the Vector Institute, and Professor at the University of Toronto’s Donnelly Centre for Cellular and Biomolecular Research.

Related links