A Modeling-Based Personalized Screening Strategy Combining Circulating Biomarker and Imaging Data for Breast Cancer Early Detection
Abstract
For a breast cancer screening method to be most effective, it should not only detect the presence of a lesion at the earliest stage before it becomes lethal, but it should also predict whether it will ever become invasive. Mammography is the current standard of care, but mammograms often detect lesions that turn out to be benign (false positives) or miss lesions that are harmful (false negatives). It is also not known how frequently mammography should be performed so that the benefits of screening outweigh the harms (e.g., radiation exposure, unnecessary invasive procedures, and anxiety) for each individual woman. This proposal addresses the Breast Cancer Research Program’s Overarching Challenges of conquering overdiagnosis and overtreatment and distinguishing deadly from benign breast cancers. We focus on the use of circulating biomarkers to report on the presence and aggressive nature of a tumor. Circulating biomarkers are proteins or other molecules (such as DNA) that are secreted or released from tumor cells into blood, urine, or other body fluids. In this project, we propose to design breast cancer screening schedules that are personalized for each woman, using her individual blood and urine biomarker measurements to determine her baseline (normal) biomarker levels and to establish how frequently she should be monitored to find early lesions that would threaten her health. To do this, we use a mathematical model – a set of equations that describes the size and growth rate of a potential lesion, based on the amount of biomarker measured – to predict not only the presence of a cancer, but also its potential to grow quickly and become lethal. The model can be used to better understand each patient’s individual biology and, unlike screening mammography, is free of any harm. This approach is also applicable to all patients, regardless of age or breast density. To show that our model-based personalized cancer screening strategy is effective, we will study a population of mice that naturally develop mammary cancer. These mice are genetically engineered so that all cancer cells that arise are programmed to release tumor-specific biomarkers that are otherwise not found in blood or urine; this enables us to directly monitor the development and growth of a tumor from the biomarker measurements alone. We will first use the mathematical model to study aggressive and benign mammary tumor growth rates in a group of mice, based on regular blood and urine biomarker sampling, as well as imaging of the potential lesions. We will then use these studied tumor growth rates to design personalized screening schedules for a separate group of mice and to evaluate the model’s ability to accurately predict the emergence of cancer and, more importantly, distinguish the presence of aggressive vs. benign disease in these mice. To truly minimize overdiagnosis and overtreatment while improving accuracy of the screening results, the model can be used to determine when, and how frequently, additional biomarker or imaging measurements should be taken in a specific individual to improve the accuracy of a breast cancer diagnosis. The accuracy of the model predictions can therefore be honed to an acceptable “threshold” of certainty, which in the clinical setting, can be decided jointly by the clinician and the patient. Novel biomarkers and diagnostic technologies are currently being developed to screen for early-stage breast cancers in the blood, urine, and body fluids. Once approved for clinical use, we will need to know how to best utilize these tools. This mathematical approach to personalizing breast cancer screening will help remove the guesswork as to whether a patient’s biomarker measurements are abnormal. The mathematical model will enable us to truly leverage a patient’s personal measurements to more precisely detect when a potential lesion may be present and how quickly it may grow. Importantly, the model can be applied to v
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 29, 2018
- Source ID
- W81XWH1810342
Entities
People
- Sharon Hori
Organizations
- Stanford University
- United States Army