NIH Grant Supports Research to Improve Prediction of Advanced Breast Cancer Risk
A new National Institutes of Health (NIH) grant will support a multi-institutional effort to develop advanced imaging biomarkers that could improve breast cancer risk prediction and help guide more personalized screening and prevention strategies.
The five-year study, led by researchers at Columbia University Irving Medical Center (CUIMC), Mayo Clinic, University of Pennsylvania, and University of California San Francisco, will investigate how radiomic features extracted from digital breast tomosynthesis (3D mammography) can identify women at increased risk of developing invasive and advanced breast cancer.
Despina Kontos, PhD, the Herbert and Florence Irving Professor of Radiological Sciences and vice chair for AI and data science research in the Department of Radiology at CUIMC, serves as a multiple principal investigator on the project. Columbia investigators Katherine Crew, MD, MS, Parisa Tehranifar, DrPH, Sachin Jambawalikar, PhD, and Dorothy Sippo, MD, MPH, are also part of the research team.
Advancing Breast Cancer Risk Assessment
While breast density is one of the strongest imaging-based risk factors for breast cancer, researchers increasingly recognize that additional patterns and textures within breast tissue may provide important information about future cancer risk. These imaging characteristics, known as radiomic features, can be extracted using advanced computational methods and may reveal subtle tissue characteristics not visible to the human eye.
The research team previously identified six reproducible radiomic phenotypes from conventional digital mammography that were associated with future invasive breast cancer risk. The new study will expand this work using digital breast tomosynthesis, which provides more detailed, three-dimensional visualization of breast tissue and has rapidly become the standard technology for breast cancer screening in the United States.
Breasts with the same density can have different texture patterns (seen above). In a recent study, patterns of breast tissue apparent on mammograms and quantified by computer analysis were a better predicter of breast cancer risk than breast density. Images provided by Despina Kontos.
Studying Breast Cancer Risk Across Diverse Populations
The project will analyze approximately 36,000 screening tomosynthesis examinations from women ages 40 to 74 across four large cohorts. Researchers will extract more than 2,000 imaging features to identify and validate radiomic phenotypes in a racially and ethnically diverse population.
Investigators will then examine how these imaging phenotypes relate to the future development of invasive and advanced breast cancer in a nested case-control study involving approximately 8,500 women diagnosed with invasive breast cancer and 17,000 matched controls. The study will also evaluate how these imaging biomarkers interact with breast density, body mass index, tumor characteristics, and genetic risk factors.
Toward More Personalized Screening
A key goal of the project is to determine whether radiomic phenotypes can enhance existing clinical breast cancer risk models and FDA-approved artificial intelligence algorithms used for risk assessment. By improving the ability to identify women at elevated risk for invasive and advanced disease, researchers hope to support more personalized approaches to breast cancer screening and prevention.
“Elucidating and characterizing novel radiomic phenotypes from screening digital breast tomosynthesis has the potential to improve our ability to identify women at differential breast cancer risk and inform more personalized screening and prevention strategies,” said Kontos.