Cell-free tumor DNA in CSF decodes breast cancer brain mets

Institution: Stanford University
Investigator(s): Melanie Hayden Gephart, MD -
Award Cycle: 2015 (Cycle 21) Grant #: 21IB-0105 Award: $236,541
Award Type: IDEA
Research Priorities
Biology of the Breast Cell>Pathogenesis: understanding the disease

Initial Award Abstract (2015)

Development of a less invasive, genetic brain tumor marker will allow treating physicians to track how the genetic spectrum of breast cancer tumors in the central nervous system (CNS; i.e. brain and spinal cord) change over time and with therapy. Tumors originating from breast cancer have a marked propensity to seed the brain, eventually present in 30% of breast cancer patients with stage IV disease. While treatments have improved for primary tumors, the development of CNS metastases severely limits life expectancy and few therapeutic options exist.

Tumors carry unique mutations in their DNA that enable them to grow and invade inappropriately, with disastrous consequences for the patient. Usually only the cancer harbors these mutations, so the mutations can be used to uniquely identify the tumors, like a fingerprint. This tumor fingerprint can suggest which therapeutic drugs would best stop the tumor, potentially personalizing the treatment plan for a particular patient. Currently, determining how the mutation fingerprint may have changed as the tumor grows and its DNA mutates would require an invasive opening of the skull or spine to remove tumor tissue. The high risk and cost of such a procedure, let alone repetitive ones, is prohibitive. In addition, the underlying mechanism by which breast cancer metastasizes to the CNS, engrafts and flourishes in this unique niche is still poorly understood. While we have made gains in characterizing primary breast cancers based on gene expression (e.g. ER, PR, HER2, etc.), we still do not understand how mutations within the metastatic brain tumor may be different from the primary tumor. Translating this genetic insight to applying currently available medications not known to be relevant in breast cancer, or developing new treatment for breast cancer patients requires more specific laboratory models for human breast cancer which has metastasized to the brain.

The question(s) or central hypotheses of the research: We have developed a less invasive method to obtain a tumorís DNA from blood and fluid samples, which we hypothesize has the potential to guide clinical decision making regarding the most appropriate medications to treat breast cancer which has metastasized to the CNS. As tumor cells die, they shed their DNA, which can be detected in the patientís blood (cell-free DNA; cfDNA). Unfortunately, the plasma cfDNA may not represent the tumor mutation profile within the protected environment of the CNS. Cutting-edge DNA sequencing techniques allow us to identify and study tumor cfDNA at very low levels, such as those present in cerebral spinal fluid (CSF). As we begin to study breast cancer cells within the CNS, we will need to test our hypotheses in cell lines reflective of the unique brain metastasis environment. We believe our new cell lines, derived directly from breast cancer cells which have been growing within the brain, will allow us to identify novel treatments.

The general methodology: Our preliminary data show we can identify brain metastasis tumor cfDNA in CSF, and use it as an important diagnostic and scientific tool. The breast cancer brain tumor cfDNA fingerprint will be followed prospectively to elucidate novel mutations during disease progression and response to treatment. The changed tumor cells may be susceptible to treatments that would not have affected the original tumor, so our method allows matching tumors with specific chemotherapeutics. Patients who undergo surgery to have a breast cancer removed from the brain will have those same brain metastases cells grown in the laboratory, allowing us to test theories of tumor biology and potential therapeutics.

Innovative elements of the project: We propose the first application of cfDNA sequencing to better characterize breast cancer tumors metastatic to the CNS. Circulating cfDNA in CSF could provide a powerful, specific, real-time surrogate for cytology or tissue-based biomarkers, with the added benefit of a holistic sampling of a tumorís mutation profile. Identifying and following the tumor cfDNA fingerprint in patient CSF and plasma ultimately enables more sophisticated, personalized, and targeted therapies. This insight will provide new information regarding tumor biology and how some tumors escape treatment. Our creation of breast cancer brain metastases-specific cell culture tools will allow us to test hypotheses derived from the cfDNA sequencing data, as well as determine the mechanism and therapeutic potential of genes previously implicated in breast cancer metastases and brain tumors. These tangible, translational tools have immediate potential applications for remodeling the approach to diagnostic intervention, scientific discovery, and drug targeting of breast cancer metastases to the brain.

Final Report (2017)

Overview: We are developing an innovative, minimally invasive method to study the genetic spectrum of breast cancer brain metastases though the sampling of cerebral spinal fluid (CSF). Cell free DNA (cfDNA) acts as a tumor fingerprint, allowing detection of genetic alterations that mark the variation and progression of disease. As a clinical tool, it may supplant the currently available and relatively insensitive methods of brain tumor diagnosis. As a scientific tool, novel tumor cfDNA mutations identified in CSF will allow for great insight into tumor biology. To study a particularly devastating type of breast cancer brain metastasis, leptomeningeal disease (LMD), we have developed a new animal model using a human breast cancer cell line. We continue to develop techniques to study genes important in breast cancer brain metastases using primary human cell cultures.

Completion of Project Aims: Aim 1. Sequence tumor cfDNA in the CSF of patients with solid and LMD breast cancer brain metastases Ė we have compared the cfDNA of breast cancer CSF to primary breast cancer tumors and other metastatic brain tumors. We have found cfDNA can track the response to therapy of a patient with LMD, and identify mutations unique to the brain tumor. Aim 2. We have developed a novel mouse model of LMD, which we will use in future projects to test potential therapeutics for breast cancer brain metastases. We continue to develop primary human cell cultures for breast cancer brain metastases.

Barriers and Solutions: The level of cfDNA in CSF is very low. We are continuing to optimize our techniques to extract the brain tumor cfDNA and analyze it. As blood contamination could lead to false positive CSF samples, we have also tested plasma for breast cancer cfDNA. If the patient has controlled systemic disease, the breast cancer brain tumor cfDNA will only be found in the CSF. To ensure our cell cultures maintain the characteristics of the tumors from which they were derived, we freeze samples at various stages and test them for gene expression levels.

Accomplishments: We have been able to collect and test or sequence nearly a dozen matched brain tumor-plasma or brain tumor-DNA specimens. We are completing exome sequencing of metastatic and primary brain tumors to then ensure similar mutations can be detected in the matched patient CSF. Importantly, we have been able to collect several patients with LMD at multiple time points over the course of treatment, allowing us to track how a tumor mutation profile changes over time and with response to therapy or relapse. We continue to collect these samples whenever available. Yingmei Li, PhD is a talented postdoctoral fellow with a PhD in chemistry, who is an expert in DNA and has been leading our cfDNA project. We have also recruited Sophia Chernikova, PhD, an expert in breast cancer brain tumor metastases, who has since developed the mouse model of LMD using human breast cancer metastases-derived cells. We have determined the sensitivity of our technique, and the variability of cfDNA between samples. Our data so far suggest a correlation between tumor volume/burden and genomic copies/mL in CSF. So far, in patients with tumor cells in CSF, we are able to detect tumor specific mutations from cfDNA in CSF. We have not experienced a false-negative result, but we have small numbers for this award.

Plans for Continuation: The techniques accelerated here may revolutionize the recognition and study of new therapeutics for LMD. We hypothesize breast cancer brain metastases would be more reliably defined and targeted by their expression pattern (e.g. ER/PR/HER2 in breast cancer) rather then by a specific mutation profile (e.g. EGFR mutations in lung cancer, BRAF mutations in melanoma), that would be more dependent upon the detection of cell free RNA. This has not yet been detected and validated in CSF, but this is something we are actively pursuing. Ideally, through a CBCRP Translational Research Award, we will be able to test our breast cancer brain metastasis specific gene targets in our cell culture and mouse models.