Dr. Ekta Khurana, an associate professor of systems and computational biomedicine at Weill Cornell Medicine, has received a two-year, $1 million Challenge Award from the Prostate Cancer Foundation to work with researchers from Weill Cornell Medicine and Memorial Sloan Kettering Cancer Center on an AI-based method for early detection of treatment-resistant prostate tumor subtypes.
Dr. Ekta Khurana
Prostate Cancer Foundation Challenge Awards support cross-disciplinary teams of investigators to conduct bold research that might not receive funding otherwise. Dr. Khurana will be working with Dr. Iman Hajirasouliha, associate professor of systems biology and computational biomedicine at Weill Cornell Medicine; and physician-scientist Dr. Yu Chen and pathologist Dr. Anuradha Gopalan of Memorial Sloan Kettering Cancer Center, on the project. The team combines a broad mix of disciplines, from pathology and genomics to computational and AI expertise.
“In this project, we will develop an AI system that can identify patients whose tumors may be developing these subtypes early on, so they can be enrolled in clinical trials for drugs that might help them,” said Dr. Khurana, who is also a member of the Englander Institute for Precision Medicine and the Sandra and Edward Meyer Cancer Center at Weill Cornell Medicine.
About 300,000 new cases of prostate cancer are diagnosed in the United States each year, many of them at advanced stages. For men, the lifetime risk of getting this cancer is about 12%. Prostate tumor growth is usually driven by excessive signaling through the androgen receptor and is slowed by drugs that reduce testosterone levels or directly inhibit androgen receptor signaling. However, a significant proportion of tumors that are treated this way become treatment-resistant by developing into subtypes that can sustain their growth without androgen receptor signaling. Clinicians currently lack good methods for detecting these treatment-resistant subtypes.
Dr. Khurana and her colleagues have previously discovered key treatment-resistant prostate tumor subtypes.
For the project, the researchers will train AI models using a large set of pathology slides from prostate tumors, as well as tumor sample gene activity patterns and treatment outcomes. The models essentially will learn to predict, given pathology slides for any prostate cancer patient, the likely tumor subtype or mix of subtypes and likely treatment outcomes.
If the best AI model proves to be sufficiently sensitive and selective—few false-negatives and few false-positives—clinicians then could use it to select patients appropriately for tests of experimental new treatments. They also could use it to avoid giving patients standard treatments when those treatments are unlikely to help them.
Once the researchers attain their project goal of developing an accurate treatment-response-predicting model, they plan to validate it with clinical trials in a follow-on project.
# # #
The above article originally appeared on the Weill Cornell Medicine Newsroom website on March 03, 2026.