Researchers at the University of British Columbia have made a breakthrough that could improve the treatment of endometrial cancer, the most common type of gynecologic cancer. By using artificial intelligence to analyze thousands of cancer cell images, they have identified a specific subset of endometrial cancer that poses a higher risk of recurrence and death, which would not have been detected using traditional methods. This discovery will assist doctors in identifying patients with high-risk cancer who may require more intensive treatment.Columbia University has made a commitment to enhancing the care for patients with endometrial cancer, which is the most common form of gynecologic cancer. By utilizing artificial intelligence (AI) to analyze patterns in thousands of cancer cell images, researchers have identified a specific subset of endometrial cancer that poses a significant risk of recurrence and death. This subset would not be recognized through traditional pathology and molecular diagnostics. The results of this study, which are published in Nature Communications, will assist doctors in identifying patients with high-risk disease who may require more extensive treatment.Dr. Jessica McAlpine, a professor and the Dr. Chew Wei Chair in Gynaecologic Oncology at UBC, along with being a surgeon-scientist at BC Cancer and Vancouver General Hospital, emphasized the importance of recognizing the differences in endometrial cancer patients. She stressed the significance of identifying high-risk patients to prevent cancer from recurring. Dr. McAlpine believes that using an AI-based approach will ensure that all patients have the chance to receive life-saving interventions. This discovery is a step forward in AI-powered precision medicine, building on the groundwork laid by Dr. McAlpine and her colleagues in British Columbia.In 2013, the Gynecologic Cancer Initiative, a collaboration between UBC, BC Cancer, Vancouver Coastal Health, and BC Women’s Hospital, found that endometrial cancer can be categorized into four subtypes based on the molecular characteristics of the cancerous cells. Each subtype presents a different level of risk to patients. Dr. McAlpine and the team then created a molecular diagnostic tool called ProMiSE, which accurately distinguishes between the subtypes. This tool is now used in B.C., parts of Canada, and internationally to help guide treatment decisions.Challenges still exist in understanding and treating endometrial cancers, with approximately 50 per cent falling into a category that lacks clear molecular features. Dr. McAlpine noted that some patients in this category have positive outcomes, while others have unfavorable ones, but there have been no tools to identify those at risk and provide appropriate treatment. To address this, Dr. McAlpine sought the expertise of machine learning expert Dr. Ali Bashashati to develop solutions.Dr. Bashashati and his team at UBC’s department of pathology and laboratory medicine have been using advanced AI methods to further categorize cancer subtypes. They developed a deep learning AI model that can analyze tissue samples and differentiate between different subtypes. After examining over 2,300 cancer tissue images, the AI identified a new subgroup with significantly lower survival rates.
Dr. Bashashati highlighted the power of AI in objectively analyzing large sets of images to identify patterns that may be missed by human pathologists. He described it as finding the “needle in the haystack” and emphasized that the AI’s findings provide valuable insights into cancer subtypes.Cancers with these particular characteristics are the most harmful and pose a greater risk for patients. The research team is now investigating how the AI tool could be used in clinical practice alongside traditional molecular and pathology diagnostics, thanks to a grant from the Terry Fox Research Institute. Dr. McAlpine stated, “The two work together, with AI adding an extra layer to the testing we’re already conducting.” An advantage of the AI-based approach is that it is cost-effective and can be easily implemented across different geographic locations.AI technology examines images that are routinely collected by pathologists and healthcare providers, even at smaller hospital locations in rural and remote areas. These images are shared when seeking second opinions on a diagnosis. By combining molecular and AI-based analysis, it may be possible for many patients to receive less intensive surgery in their local communities, while still ensuring that those who need treatment at a larger cancer center can access it.
Dr. Bashashati highlighted the potential for greater equity and access with this approach. The AI technology does not differentiate between patients in large urban centers or rural communities, offering the opportunity for more widespread and fair access to medical care.We believe that this new technology has the potential to greatly improve the diagnosis and treatment of endometrial cancer for patients worldwide.