Warning: Undefined variable $skip_to_content in /home/www/hdrn/wp-content/themes/hdrn-canada/header.php on line 92

Endometrial cancer is preventable & multi-regional data can help identify risk

A smiling woman with long, dark hair. Text reads: DASH. Researcher Spotlight. Dr. Aline Talhouk. Logo for Health Data Research Network Canada is at top.
thumbnail

Endometrial cancer is the most common gynecologic cancer in Canada. If detected early, before the cancer mestastasizes, the survival rates for endometrial cancer are excellent. The problem, according to Dr. Aline Talhouk, an assistant professor in the department of Obstetrics & Gynecology at the University of British Columbia, is that even though prevention interventions are effective, they may not be cost effective at the population level. There are presently no recommendations for prevention or screening in endometrial cancer.

“Endometrial cancer is pretty much curable when detected early, but detection and prevention interventions are costly and invasive, and delays in diagnosis significantly impact survival rates,” said Dr. Talhouk, who also serves as Director of Data Science and Informatics at OVCARE, BC’s ovarian and gynecological cancer research program. That’s why her Endometrial Cancer Risks Study aims to evaluate and improve risk prediction models to better identify people at high risk, to motivate access to preventive care.

It was important to use multi-regional data because we wanted the risk models to be generalizable beyond the population they were originally built on. ~ Aline Talhouk

A machine learning scientist and statistician by training, Dr. Talhouk believes that risk models are important because they enable the identification of populations that could benefit from prevention or early screening interventions. Interventions for early detection include things like endometrial biopsies, which are currently only recommended for people with hereditary cancers due to their higher risk. Risk prediction models can provide a more comprehensive way to identify people who are at an elevated risk based on a number of additional factors.

The first phase of Dr. Talhouk’s project focused on evaluating existing risk prediction models for Canada’s population, which has never been previously done. With this baseline, the research team aims to build on these models by incorporating additional factors such as genetic risk, gender identity, environmental implications and socioeconomic status. In particular, they want to build a model that better accounts for how risk of endometrial cancer can change over time. “Existing models only look at specific points in time, but a lot can happen in between that could increase or decrease risk, such as a person using a hormonal intrauterine device (IUD),” said Dr. Talhouk.

The research team opted to use linked data from the Canadian Partnership for Tomorrow’s Health (CanPath), Canada’s largest population study that collects data from over 300,000 Canadians about health, lifestyle, environment and behaviour. Participants answer questions that measure many of the risk factors for endometrial cancer, making CanPath data a rich source of information for Dr. Talhouk’s research. To fully evaluate the risk prediction models, however, it was critical to link these data to cancer registries across the country.

The data linkage was facilitated by HDRN Canada’s Data Access Support Hub (DASH), a one-stop data access service portal for researchers requiring multi-regional health administrative data in Canada. Dr. Talhouk’s Endometrial Cancer Risks Study was one of the first projects that benefitted from DASH’s coordination support in linking CanPath data with multi-regional administrative data sets in three provinces. “It was important to use multi-regional data because we wanted the risk models to be generalizable beyond the population they were originally built on,” she explained.

The DASH Team helped determine what types of data access would be feasible by convening representatives from data centres in different provinces, coordinating a single  data access request, and helping to move the project along. “It was really nice to have everyone there at the same time explaining exactly what was available in each province. Some provinces only had a few cases of endometrial cancer on record, so it wasn’t worth it to go through the process of accessing those data,” explained Dr. Talhouk.

Because Canada is such a large country with varied demographics and characteristics, studies using multi-regional data are important for developing models that are generalized and usable across health care systems. According to Dr. Talhouk, multi-regional data also allows researchers to access larger sample sizes than they would find in one province, and data from various demographics which can improve health equity. “Ultimately we want to make sure that everyone can benefit from our findings, not just people in one region, so we want to try to use data from a broad range of people. Having your data included in research is an equity issue.”

Once the study finishes evaluating existing endometrial cancer risk prediction models, Dr. Talhouk will submit a new access request for genotype data. The next phase of analysis will incorporate factors that current models haven’t included like sex and gender, environmental variables and socioeconomic status. “These are things no one has looked at before in the context of risk prediction modeling. What we’re really hoping to do is work towards a more dynamic model that accounts for how risk of endometrial cancer may change over the course of someone’s life.”

Latest Posts

Icons showing a shield with medical cross, a stethoscope, a medical across on a phone screen, a pair of test tubes, hard drives besides data chart. Health Data Research Canada and Cantrain logos are at bottom.

Statistical charts, a heart containing ECG graph, cloud storage, syringe and pills, and medical records on a computer screen. Logos of Health Data Research Network Canada and Canada's Drug Agency are at bottom.