Grant: Project Grant

Magnetic resonance imaging (MRI) of the breast can detect breast cancer earlier than mammography or ultrasound. However, because MRI is very expensive and more likely to mistake benign breast structures for cancer, MRI is usually only offered to women with the highest risk of developing breast cancer based on genetic testing, family history, or previous radiation to the chest. The High-Risk Ontario Breast Screening Program (OBSP) is a unique screening program which provides annual screening with MRI to high-risk women. There is still a lot of variation in this high-risk group. At one extreme are young women with an inherited BRCA1 mutation who tend to develop aggressive, rapidly growing tumours. These women may benefit from more frequent screening or prophylactic mastectomy. At the other extreme are women with a strong family history associated with a currently unknown genetic abnormality who could actually be at average risk and who may not need annual screening. There is, therefore, a need for improved risk prediction methods. We propose to develop new artificial intelligence methods to extract information about breast tissue volume and texture from MRI. With MRI we can also measure how much contrast agent is taken up in normal breast tissue; this is known as background parenchymal enhancement, BPE, and it has been shown that there is a correlation between BPE and risk. We will assess whether a combination of these MRI properties can be used to predict cancer risk. We will also determine whether they are more strongly correlated with breast cancer risk than mammographic density which is already known to be useful in identifying patients at intermediate risk. Our methods will be tested on data from the OBSP which screens thousands of high-risk women each year at 30 different centres in Ontario. A more accurate assessment of risk could allow screening programs to be tailored to the needs of individual women and would also make better use of limited resources.