Funded Projects

Funded Projects

The Cancer Pathology Translational Research Grants will complement the education and training program available through the Network to increase molecular pathology and research knowledge amongst the pathologist in Ontario. To foster the development of this research community grantees and their trainees will present there work at the Network’s annual meeting, the Ontario Cancer Pathology Research Meeting.

Interested in getting involved with one of our funded projects?

Projects funded by the OMPRN are listed below.

Image Analysis

Active

Research Team:

Applying Artificial Intelligence For Automated Diagnosis Of Colorectal Polyps

Colorectal cancer (CRC) is 2nd most common cancer in Ontario. It has been shown that individuals at risk of developing CRC form small polyps in the colon that can then progress to become cancerous. Therefore, CRC surveillance is performed by various methods including colonoscopy in people over 50 years of age to identify and remove these polyps. The polyps removed during surveillance colonoscopy forms the bulk of workload of gastrointestinal pathologists, even though the diagnosis in majority of the cases is quite straight-forward. This will continue to increase with increasing age of the population. Hence, we are proposing an automated process to screen these polyps using artificial intelligence (AI) techniques. AI algorithms will be trained to screen and classify these polyps with high accuracy. Any polyps that can not be classified with confidence will be flagged for pathologist review, thus reducing the workload and allowing more time to devote to complex cases. AI will also serve as a ‘second reader’ to alert the pathologist to any discrepancy and as quality indicator.

Image Analysis

Active

Research Team:

Developing a machine learning algorithm for differentiating pseudo- and true invasion in colorectal polyps

Polyps in large bowel are very common, affecting 1 in 4 people in general and 1 in 2 people in those older than 50 years. Most of them are harmless. However, a small proportion of them contain cancer. Colon cancer is the third most common cancer in North America. The diagnosis of malignancy in a polyp is made by pathologists identifying cancer cells in the deep layer of the bowel wall under the microscope: a process known as invasion. However, non-cancer cells can be misplaced in the same location which can mislead the pathologists to make the wrong diagnosis. Although there are a few features pathologists use to tell apart “true” and “pseudo” invasion, this is never an easy task. Often times a panel of expert pathologists cannot reach agreement. With the advanced technology in image analysis and machine learning, the artificial intelligence sometimes outperforms human’s eyes in imaging pattern recognition. We identified 100 cases of colorectal polyps with cancer (true invasion) and 100 cases of colorectal polyps with features mimicking cancer (pseudo-invasion), diagnosed with consensus by a panel of expert pathologists specialized in this field. We will develop a machine learning algorithm that can recognize cancer and its mimics using high resolution whole slide images of those cases. We will compare the diagnostic accuracy of the algorithm with pathologists. After the algorithm is fully validated, we will build a web application and publish it online to allow access by pathologists worldwide. Our ultimate goal is to aid pathologists in making an accurate diagnosis of colon cancer.

Image Analysis

Active

Research Team:

Development of deep learning tools for deciphering intra-tumoral heterogeneity and predicting drug response in High Grade Serous Ovarian Cancer

Ovarian cancer is an aggressive cancer of the female reproductive system that is most commonly diagnosed at advanced stage where only 30% of patients survive more than 5 years from diagnosis12. Despite spirited surgical and medical therapy, this poor outlook has remained fairly unchanged despite many breakthroughs in our understanding of cancer biology13. One recent explanation for these previous failures is that while we often think of a patient’s tumor as being a homogenous mass of identical cancer cells, there are in fact a number of tumor sub-clones within each cancer that respond differently to our conventional and traditional therapies10. This means that existing treatments may not equally target all tumor cells and allow resistant subclones to survive and drive disease recurrence and progression. To address this, this collaborative team plans to harness expertise in artificial intelligence and unique clinical cohorts to explore if this technology can help automate the detection of biologically distinct tumor subregions and understand the biological significance of these computer defined regions. First, they will assess if computer defined regions of variability do indeed show unique biological activity that may be driving partial responses to therapy. Secondly, they will examine if the degree of morphologic variation within these cancers can predict response to existing therapies. Routine detection and characterization of tumor subclones, within each individual patients’ tumor, could help propose personalized and effective drug combinations that together target a larger fraction of the overall tumor biology. This could ultimately provide more durable responses for patients.

Image Analysis

Active

Research Team:

p53 immunohistochemistry interpretation in AML and MDS: using image analysis to better predict mutation status

AML and MDS are rare blood cell cancers that become more common in older individuals, individuals exposed to carcinogens and in individuals with some inherited gene mutations. From many prior research studies, several gene mutations have been identified that contribute to the development and progression of AML and MDS. One important gene, P53, is involved in the repair of DNA damage and maintains the integrity of DNA. When p53 is lost or mutated, genetic “instability” ensues resulting in multiple gene mutations. Genetic instability allows the disease to progress despite our best efforts to treat it. Newer treatments are becoming available that may improve outcomes for such patients therefore knowing the P53 status at the time of diagnosis is critical in deciding on which treatment to choose. To identify a P53 mutation, DNA from bone marrow can be sequenced, however, this process is time-consuming and the result takes 3-4 weeks to be available. Staining for the P53 protein in the bone marrow biopsy may be another way to determine whether a mutation is present, however, the interpretation of protein staining in biopsies is subjective and depends on the pathologist looking at the tissue. Digital imaging is becoming more widespread in pathology and the development of computer algorithms to analyze biopsy samples may be a method that reduces the subjectivity in the interpretation of stains. In this project we propose to use digital imaging analysis of AML and MDS cases to determine whether P53 mutation status can be predicted.

Stream 1

Research Team:

The Role of the Transcriptional Regulator TBX3 in Early Breast Cancer Progression: a possible regulator of the transition from in situ (DCIS) to invasive mammary carcinoma (IMC)

This study is looking at biomarkers in tissue samples of people with Ductal Carcinoma In Situ (DCIS) to help tell which patients will develop invasive mammary carcinoma (an advanced form of breast cancer).

Ductal Carcinoma In Situ (DCIS) means cells inside some of the ducts of the breast have started to turn into cancer cells. These cells are inside the ducts and have not started to spread into the surrounding breast tissue. Doctors diagnose DCIS by looking at whether or not cells within the ducts of a patient’s tissue sample appear benign (non-cancerous) or malignant (cancerous).  However, only some of these patients will go on to develop invasive breast cancer and it is currently difficult for doctors to tell which of these patients are at risk.

Researchers believe that increased activity of the T-box transcription factor (TBX3) gene is linked to more aggressive breast cancer. In this study researchers will assess which other genes are potentially regulated by TBX3, and will use tissue samples from current breast cancer patients to determine whether increased activity of this gene, and/or specific genes that this gene itself activates (such as Slug and Twist1) might be used as an indicator of the potential to break out of the ducts and invade adjacent breast tissue (i.e to become more aggressive).Researchers hope that one or more of these tests will help doctors predict which patients are at risk of progressing to advanced breast cancer.

All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research. Find out more about pathology or why molecular pathology is important in cancer research here [link].

Stream 1

Active

Research Team:

A Minimalist Approach to Cancer Tissue-of-Origin Classification by DNA Methylation

Pathologists are responsible for the accurate classification (or categorization) of human diseases, including cancers (e.g. distinguishing breast cancer metastatic to the lung from primary lung cancer). Only when appropriately classified can patients with cancers receive the optimal site-specific treatments. While the examination of tissue samples under the microscope by pathologists is usually sufficient for achieving this end, there often remains a small percentage of cases that are subject to diagnostic discrepancies and/or may be otherwise difficult to classify. For this, DNA-methylation profiling, by providing tumor tissue-of-origin signatures (e.g. breast versus lung), is a potentially useful adjunct. While recent research studies produced very promising results, the profiling technology itself is technically demanding and may be difficult for clinical laboratories to adopt. Accodingly, the goal of this project to simplify methylation-based cancer classification, in part by reducing the number of markers needed. In a preliminary study, using just 28 selected markers (rather than all 27 thousand available markers), we were able to correctly assign ~90% of >1000 cancer cases to their tissues-of-origins. With funding, we will perform additional analyses and test additional samples with the goal of designing a practical and effective methylation-based test that could be validated for routine clinical use in a follow-up study. All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research.

Stream 1

Research Team:

An integrative DNA sequencing panel to accurately diagnose and appropriately manage patients with ovarian sex cord stromal tumors

In the Canadian health care system, we are often bound by the fiduciary restrictions, and diagnosing rare tumours can be particularly challenging. Research from our group and others have identified that DNA sequencing technology can allow us to accurately diagnose these tumours, allowing the patients to appropriately seek genetic counselling and/or potentially receive new drug that target those mutations. Two barriers prevent such advances: 1) cost of the new tests, and 2) lack of a validating study that confirms the presence of those mutations in a larger group of patients. At the Toronto General Hospital (TGH), we have accrued a large group of cases that would allow us to tackle those two challenges. With a large number of surgeries at TGH, we have enough number of these relatively rare ovarian cancers that would allow us to confirm the presence of those mutations and examine their frequency. Using state-of-the-art DNA sequencing technology available at TGH, we aim to construct a novel sequencing panel that would serve as a one-stop test for a number of these tumours, allowing us to identify these mutations with a single run. We potentially have a solution to the aforementioned barriers that would allow for truly personalized care of these patients. With the TGH. All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research.

Stream 1

Research Team:

Analyzing Renal Papillary adenomas to elucidate the early tumorigenic events of Papillary Renal Cell Carcinoma and their possible correlation with kidney progenitor cells

We are studying a type of kidney cancer called the papillary type. This type of cancer is particularly common in diseased kidney. Findings from our previous work as well as the work of others led us to believe that these cancers arise from kidney regenerative cells.

These cells are normally hidden in the kidney, but when the kidneys start to shows signs of damage, these regenerative cells are increased in number so as to renew/regenerate the lost cells. Though they should be helpful to the kidney, some disruption in their biology happens at that stage which makes them give rise to cancer. It is important to note that there are small lesions termed papillary adenomas that are also very common to find in the damaged kidneys that look identical to the cancer we study (but smaller) and are though to give rise if left to the full blown cancer. So in essence these lesions are the link between the regenerative cells and the cancer. In this project we will study that proposed connection between these regenerative cells and the cancer by tracking the cells all through from the normal kidney, to the damaged kidney to the small adenoma lesions to the cancer. Through that process we will understand a lot more about the biology of how these cancers form.

That knowledge will help us in the future to prevent these kidney cancers. All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research.

Stream 1

Active

Research Team:

Applying deep learning to bone marrow differential counts toward improved diagnosis of hematological neoplasms

When a patient is suspected of having a blood or a bone marrow disorder, a bone marrow biopsy study is performed. This study consists of many different parts that are analyzed by pathologist to make a final diagnosis. This is a lengthy and complicated process that may take days to weeks, depending on the type of bone marrow disorder. However, one part of the bone marrow study, known as the aspirate, provides important information within hours of collection that guides further testing, may support a diagnosis and even lead to early treatment in some cases. As part of the aspirate review, a pathologist classifies and counts many different types of bone marrow cells into categories. Based upon the number of cells in each category, further decisions on testing or treatment are made. However, counting bone marrow cells is time consuming, and different pathologists may disagree on which cells fit into each category. These factors may affect the ability of pathologists to correctly diagnose bone marrow disorders. As a result, there is need for new tools that will help pathologists to better analyze bone marrow aspirate cells. Recently, artificial intelligence has been used to perform a number of diagnostic tasks in pathology by analyzing images of cells. Our team will explore the ability of artificial intelligence to help pathologists analyze and count bone marrow cells. This may eventually lead to artificial intelligence technology that will support the ability of pathologists to diagnose blood and bone marrow disorders, toward better outcomes for patients.

Stream 1

Research Team:

Assessing the clinical relevance of circulating miRNAs expression signature for early colorectal cancer detection

Colorectal cancer (CRC) is a world-wide health problem especially in wealthier countries and the numbers for CRC are going up. An 80% increase in CRC deaths is projected by 2030 unless there are significant improvements in prevention, early disease detection and treatment. CRC usually develops from polyps that over time change to cancer. Therefore early disease detection improves outcomes. Colonoscopy is the gold standard for detection of CRC, but it is costly and invasive for screening purposes. Fecal-based tests are for useful for screening but they are not very sensitive or specific. Therefore, more accurate, non-invasive tumor biomarkers for the detection of early CRC lesions (ECRCL) are required. There is growing evidence about the role played by microRNAs (miRNAs) human cancer. Tumor-specific circulating miRNAs can be detected in the blood of various cancer patients including CRC. The circulating miRNAs are highly stable and can therefore be useful diagnosing and predicting cancer. The main objectives of this project are to: i) investigate circulating miRNA expression profile in blood from patients with ECRCL and compare them with healthy controls  ii) evaluate miRNA expression profile in ECRCL as well as adjacent mucosa and iii) identify disease specific miRNA signatures that are typical of ECRCL.  This research project will advance knowledge on the role of circulating miRNAs biomarkers in CRC diagnosis and their potential predictive role for disease risk assessment. The findings will help in developing a panel of circulating miRNAs markers usable for accurate early detection of CRC, and improve CRC screening strategy. All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research.

Stream 1

Active

Research Team:

Assessment of Progestin Response Potential in Grade 2 Endometrial Endometrioid Adenocarcinoma and Interrogation of Biomarkers Associated with Progestin Resistance

Treatment of endometrial (womb) cancers includes surgery (removal of the womb) or hormonal treatment, depending on the degree of severity and spread of disease. There is a growing interest in hormonal treatment for patients wishing to have children or patients that have other medical conditions which prevent them from having surgery.3 Hormonal treatment includes progesterone based therapy, which is currently reserved for precursor lesions and early low grade endometrial cancer (grade 1). However, not all cases respond to hormonal treatment or some tumors recur after treatment. Moreover, some women with cancers of intermediate grade (grade 2) are also interested in hormonal treatment in order to delay surgery and allow time for a successful pregnancy. A current challenge in management of these cases is to identify patients who would respond to hormonal treatment and for whom surgery can be safely delayed. Our aim is to study the molecular profile of intermediate grade cancers and compare them to that of endometrial precancer and low grade cancer that have a known response to progesterone therapy. We believe that a proportion of intermediate grade cancers resemble the low grade responsive cases, which would suggest that select intermediate grade cases can also be safely treated with progesterone. Knowing which patients would benefit from hormonal treatment before starting therapy would allow a better management of cases in clinical practice.

Stream 1

Active

Research Team:

Deep Learning for Lung Cancer Diagnostics and Biomarker Discovery

Artificial intelligence (AI) research has advanced significantly in recent years, and has given rise to algorithms which are adept at analyzing image data. The application of this technology within pathology is an area of major interest, since these algorithms may be able to help pathologists diagnose cancer with increased accuracy, consistency, and efficiency. Past research has shown that there is considerable variability between the diagnoses of individual pathologists, and the use of diagnostic software promises to bring an added element of objectivity to the diagnostic process. Furthermore, it may become possible to use this software to predict the types of mutations found in a tumour, information which is usually obtained using time-consuming advanced testing techniques. Since many treatments in lung cancer are based on the individual mutations in a patient’s cancer, obtaining this information very rapidly may be allow a patient’s physician to initiate therapies immediately while waiting for definitive test results. Finally, these algorithms may not only be useful for diagnosing known types of cancers, but may help to identify new types of lung cancer which share subtle microscopic features that are not easily visible to the human eye. Since tumours which share microscopic features frequently have similar biological characteristics, identifying new subtypes may help pathologists to classify patients’ tumours, and refine predictions about patients’ expected treatment response and prognosis.

Stream 1

Research Team:

Development of Pediatric OncoScan Genomic and SNP Array for clinical implementation

This study is looking at biomarkers in tissue samples of children with cancer to help diagnose their cancer and to tell which patients are more likely to respond to different available treatments.

Some pediatric (childhood) cancers develop as a result of changes in the genes’ code and/or the number of copies of the genes (gains or losses of the gene). Researchers believe understanding which changes have occurred in a child’s cancer can help to diagnose the cancer and to identify treatments to which the child is more likely to respond. In this study researchers will use a test normally used to for adult cancers and adapt this for children with cancer. Researchers will use tissue samples from current patients with childhood cancer to develop this test so that doctors can figure out which gene changes are present in an individual's cancer. Researchers hope that this test will help doctors to better match children with cancer to treatments.

All studies funded by the OMPRN also have an educational objective for trainee pathologists or early career pathologists. Ensuring new researchers have experience in conducting or leading studies helps the OMPRN to develop the next generation of molecular pathology leaders in cancer research. Find out more about pathology or why molecular pathology is important in cancer research here [link].

Stream 1

Research Team: