1. Home
2. Precision Medicine
3. Projects
4. Integrated Machine-Learning Platform to Inform Precision Therapy in Breast Cancer Patients

Integrated Machine-Learning Platform to Inform Precision Therapy in Breast Cancer Patients

Principal Investigators

Pablo Tamayo, PhD
Professor in Residence, Medicine, UC San Diego

William Kim, PhD
Assistant Professor, Medicine, UC San Diego

Background

Most breast cancers are associated with one of three proteins acting abnormally. These proteins are estrogen receptor (ER), progesterone receptor (PR), or a protein called HER2. Treatments have been developed that block the action of these proteins, which stops tumors cells from continually multiplying. However, a triple-negative breast cancer (TNBC) tumor produces little to no ER, PR, or HER2, so the three most-likely candidates that would cause a tumor are not responsible. The causes of TNBC are more elusive, and therefore harder to treat. Consequently, TNBC has a much higher mortality rate than other types of breast cancer.

Outcomes for Hispanic women with TNBC are worse than for non-Hispanic Caucasian women, for a number of reasons. Hispanic women are 30% more likely to be diagnosed with TNBC than non-Hispanic Caucasian women. The poor outcomes may also be due to discrepancies in social determinants of health such as income, education, and diet.

Project CELSUS

Project CELSUS is named after Aulus Cornelius Celsus, an early pioneer of evidenced-based medicine in the 1st century A.D. The project brings together a community of researchers, clinicians, patients, and community advocates who will join forces and apply DNA and RNA profiling, along with sophisticated analytic approaches, to identify new treatments, and implement a more precise system to match TNBC tumors and cancer therapies.

Project CELSUS is an opportunity for breast oncology practices to provide underserved patients with advanced precision medicine capabilities, while at the same time fostering relationships and dialogue between UCSD, community cancer physicians, community hospitals, patients, and other stakeholders to ensure equitable participation and benefit.

Using machine learning and publicly available databases, like the Cancer Cell Line Encyclopedia (CCLE) and The Cancer Genome Atlas (TCGA), as well as new samples from Hispanic patients, the researchers will define new multi-dimensional subtypes, or "cellular states" of TNBC. The team will then use computational models to predict the effectiveness of different treatments, alone and in combination, on cells in the discrete cellular states. The predictions will then be tested using several different models—from established cell lines grown in a dish, to mice with engineered genetic backgrounds, and patient-derived samples.

The research team will genetically sequence TNBC tumor cells from both Hispanic and non-Hispanic patients, and use the information from the computational, cellular, and mouse-based models to generate a Patient Analytic Profile, which will be used to assess prognosis and inform the best possible course of treatment.

The Moores Cancer Center Community Outreach and Engagement Core has built strong relationships with the San Diego community, and will be a central part of the Project CELSUS implementation, interpretation, and dissemination plan. Along with collaborators at San Diego State University, the project team will conduct interviews and focus groups throughout the project period and address issues related to trust, patient consent, data sharing, privacy, and security. The Community Outreach and Engagement Core will also be utilized to disseminate the study processes and results through in-person meetings, social media, and through the CancerDAT platform, which provides neighborhood-specific data about the health and well-being of residents as well as information about evidence-based practices and local academic and community resources available to address health issues.

Research Team and Collaborators