Developing Digital Pathology Biomarkers for Breast Cancer Recurrence Score
Abstract
Study Rationale: Distinguishing lethal from non-lethal breast cancers is of utmost importance. Some patients are at a low risk of recurrence or death after surgery, whereas others are at a high risk of experiencing morbidity and mortality even with aggressive chemotherapy and hormonal therapy. OncotypeDx (ODX) is an increasingly utilized test that measures patterns of how genes are expressed in cancer cells to better determine the risk of a patient’s disease. ODX is clinically indicated for the majority of patients with early-stage hormone receptor positive breast cancer, the most common subtype of breast cancer, and it is used to aid in the appropriate use of chemotherapy in such patients. However, ODX testing has a number of limitations. ODX costs approximately $3,000 per assay, resulting in an estimated $231 million in testing costs on a yearly basis. Breast cancer tissue must be sent to a central laboratory in California for ODX testing, and results can take several weeks to obtain, resulting in unnecessary treatment delays. Yet, the most pressing challenge associated with the use of the ODX assay is its contribution to healthcare disparities. Breast cancer mortality is approximately 40% higher in African American women than white women, and a breast cancer case is approximately three times as likely to be lethal when diagnosed in Africa rather than North America. ODX testing is less likely to be performed in African American and Hispanic patients in the United States, and ODX testing is not available in Africa and other low resource settings, where the greatest benefit from this technology could be realized. In addition, the ODX assay has lower predictive accuracy in African Americans than in white patients. While not all breast cancer patients receive ODX testing, all cancer patients undergo a biopsy with routine examination by pathologists. Pathologists are able to describe the specific type of breast cancer and certain features of disease aggressiveness by microscopic examination of breast cancer tissue. Recently, our group and others have demonstrated that by using the latest artificial intelligence technology, subtle features present in microscopic images of cancer can be used to identify patterns of gene expression, which is used in ODX assay. We have previously used microscopic images of cancer to develop highly accurate clinical grade predictors of tests similar to ODX in gastric and intestinal cancers. Our preliminary data suggest that digital images could similarly be used to predict ODX. However, no repository currently exists with the necessary number of patients with matched digital histology slides and ODX scores to develop such a tool. Additionally, publicly available cancer repositories are notably lacking in racial and ethnic diversity, which is essential to the development of an equitable artificial intelligence tool. Objectives and Aims: To address these critical gaps in clinical research, we plan to use data from patients enrolled in the well-annotated and diverse Chicago Multi-ethnic Cohort of Breast Cancer (ChiMEC), to develop a clinical grade predictor of ODX (Aim 1). We will test this predictor in patients from the Northshore University Health System and Ingalls Memorial Hospital, which serve two unique and diverse patient populations within Chicago, to ensure that our model can be equitably applied to patients of all ancestries (Aim 2). Finally, since the ultimate goal of our proposal is to develop an accurate predictor of recurrence and chemotherapy benefit, we will compare our artificial intelligence model to actual ODX scores to determine if we can predict long term outcomes more accurately than ODX (Aim 3). Applicability of the Research: This study addresses two overarching challenges: Conquer the problems of overdiagnosis and overtreatment and Distinguish deadly from non-deadly breast cancers. Accurate determinants of breast cancer recurrence are needed to help health care
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 28, 2022
- Source ID
- W81XWH2210792
Entities
People
- Alexander T Pearson
Organizations
- United States Army
- University of Chicago