Using Multi-Modal Data Integration Across Breast Cancer Molecular Subtypes to Develop Predictors of Treatment Response, Early Relapse, and Survival and to Explore Early Signals of Later Metastatic Rec
Abstract
Some of the critical questions that women and men with breast cancer ask during their treatment journey are: How will I know if I am responding to treatment; When will I know if I am responding to treatment; and How will I know if my disease will come back. It is our opinion that, in order to answer these questions, we need to use a more cohesive approach by bringing together all the information and data generated about each individual patient. Specific information from each type of investigation, such as genetic tests, mammograms, or pathology specimens, can give us clues as to why some are likely to respond to treatment or progress to metastatic disease. However, currently we look at all this information in a disconnected and siloed manner. Thanks to the generosity of >6500 patients, who have provided samples and data through participating in clinical trials, we are proposing to build a unique data resource that links up the information from multiple types of investigations in each individual patient. By doing this with thousands of patient data and samples, we will create a platform with sufficient power to develop specific tools that help predict whether someone might respond to treatment or be at risk of their disease spreading. This study will address three of the overarching challenges that we face: (1) revolutionize treatment regimens by replacing them with ones that are more effective and less toxic; (2) distinguish deadly from non-deadly breast cancers; and (3) identify why some breast cancers become metastatic while others do not. In addition to tests done in the clinic, such as mammograms, etc., we will also begin to look in fine detail, down at the level of single cells, to find out if specific factors or characteristics can help with predicting the response to treatment or metastatic spread. To do this, we will use cutting edge technologies that allow us to look at the breast cancer cells in minute detail and assess their relationships with cells around them and within the environment in which the tumor developed. The team is composed of scientists with world-class expertise in this area and clinicians with over 20 years of clinical trials experience. They have a range of experts with knowledge of all the different data types to be explored. The team is led by a clinician advised by two consumer patient advocates who will ensure that all prediction tools are developed with the patient and clinical implementation in mind. This study aims to: (1) identify factors that predict for treatment response and early metastatic spread; (2) develop predictive tests using and combining these factors; and (3) create a unique breast cancer research resource that is comprehensive, that links together data from different investigations, and that can used by the breast cancer community to ask and answer many of the other challenges in breast cancer. This type of resource will undoubtedly speed up finding a path to ending breast cancer as a lethal disease. Our team is split into working groups who focus on specific types of data, e.g., data from mammograms or other radiological images, which are called radiomics. Each working group will select key factors they believe influence prediction of response to treatment or risk of metastatic spread. The selected factors from each working group are then carefully combined using special mathematical models and tests, called machine learning and artificial intelligence. We will then understand what factors are truly important for these predictions. These assumptions will then be re-tested in an independent set of data. We intend to complete this process in several different types of breast cancers starting with triple-negative breast cancer (TNBC). Here, some of the fine detail work at the cellular level has already been established through allied projects. This allows us to use TNBC as our prototype model within which to develop our predictors. We wou
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2024
- Source ID
- HT94252310767
Entities
People
- Jean Abraham
Organizations
- United States Army
- University of Cambridge