Research Project

Combating Tumor Heterogeneity


Program Title

Combating Tumor Heterogeneity


Drs. Andrea Bild, Fred Adler, and Adam Cohen

Lab Websites

Project Description

Our research program focuses on finding effective treatments for resistant tumors by studying the evolution of phenotypes emergent in late-stage breast and ovarian cancer. Our team develops systems-based strategies to understand how genomic diversity, clonal evolution, and phenotypic change interact in the progression toward chemoresistant cancer. We will carry out clinical trials aimed at blocking transition of tumors to a resistant state by targeting critical resistant phenotypes. We will test our dynamic models in these clinical trials to determine if we can predict the “resistant state” of a tumor and a patient’s response to drug. We hypothesize that acquired resistance emerges from selection acting on phenotypes during tumor evolution, and that simultaneously measuring and modeling subclone genotypes and phenotypes will identify new, and testable, therapeutic targets.

The summer research project will provide the opportunity to work closely in a multi-disciplinary team of a systems cancer biologist (Bild), mathematician (Adler), and medical oncologist clinician (Cohen) to carry out translational research. We will together execute a project that focuses on understanding how single cells with a patient’s tumor change during therapy and gain resistance phenotypes to chemotherapy. We will then block those phenotypes using targeted cancer drugs to reverse the resistant state and re-instate a tumor’s response to chemotherapy. As we know very little about the cancer that actually kills the patient, we expect our studies will illuminate the landscape of end-stage ovarian and breast cancer. The outcome from this research will be new approaches to block or reverse the transition to a resistant state for advanced stage breast and ovarian cancer patients

Here are a few examples of our research focus areas that students could participate in:

1) The study of tumor heterogeneity evolution

As tumor cells continue to evolve in the metastatic setting, the study of longitudinal cohorts of metastatic tumor cells over time provides a more accurate characterization of how metastatic disease either innately has pic1resistance or acquires resistance and is ultimately terminal; the use of a single snapshot of metastatic cancer would not provide this information. For each patient, we determine subclonal evolution through bulk and/or single-cell DNA sequencing at various points in the patient’s treatment history shown to the left. As indicated, each unique subclone, as identified by the different large cell “circles” (various smaller colored circles within cells represent different mutations), evolved over time and treatment, leading to a dominant subclone at time of the patient’s death. These data enable us to identify the mechanism by which tumors evolve to evade response to drug, and to ask if there are specific genetic (or phenotypic) events that confer a survival advantage to progressive tumor cells.

2) Development of cancer integral projection models

Progression of a cancer involves the dynamic interaction among many cell genotypes and phenotypes with each other and with their environment. In this proposal, we integrate these intracellular and extracellular effects using patient samples and experimental modeling to create a holistic understanding of chemo-resistant state. To delay, prevent, or reverse the emergence of resistance cell states, we must identify the key factors controlling those dynamics. Our modeling approach builds on the emergent method of Integral Projection Models (IPM) which have yet, to our knowledge, to be applied in cancer biology. pic2These models use statistical analyses to quantify the factors that control dynamics and couple them directly to mathematical models to project the future course of populations and their environment. We will extend this method, developed initially for plant populations, to model cancer by linking genetic and phenotypic information in a population of cells that cannot be followed individually over time, as most plants can be. We will develop extensions that can combine data across patients, much as ecologists combine data across different growth sites for plant populations. The strengths of this approach are its use of empirical data to evaluate the importance of cell-cell and TME interactions in the changing state of the tumor, and to then predict the effects of different processes or interventions. In particular, we can incorporate genetic and phenotypic changes in cell populations, how cell populations change in response to cooperation or competition among subclones and phenotypes, and interactions with the tumor environment that the cells themselves partially shape. In addition to testing the effects of interventions, these mechanistic models identify points of vulnerability to target with new interventions. This flexible approach can unravel the complex web of interactions and enable efficient in silico experiments to evaluate therapies.

3) Single-cell RNA sequencing phenotyping of cancer and normal cells

To identify phenotypes that may promote acquired drug resistance, we performed single-cell RNA sequencing (scRNA-Seq) on pre- and post-treatment pleural effusions from advanced stage breast cancer patients. Since bulk RNA-Seq data may misreport phenotypes due to the frequent presence of normal cell contaminants (10-80% of the cell population) or specific cancer sub-populations, we considered scRNA-Seq as a potentially superior approach to address this question. We measured the presence of 3,331 signatures in the Molecular Signatures Database (C2 and Cancer Hallmarks signatures) by applying ssGSEA to each single cell’s expression data, and compared the resulting enrichment scores between pre- and post-treatment single cells using Student’s t-test. The top 25 most differentially regulated signatures between each patient’s pre- and post-treatment samples were highly enriched in EMT-associated signatures (red arrows) and immune-associated signatures (blue arrows), such as TNF and antigen presentation. Indeed, EMT was significantly increased in 3 out of 4 patients while a TNF signature was significantly decreased in 3 out of 4 patients. These phenotypes could potentially provide a selective advantage and promote drug resistance, as EMT has been linked to an aggressive, stem-like state while loss of TNF signaling could prevent T cell activation after chemotherapy-induced cell death. Thus EMT and immune evasion may be key phenotypes promoting acquired drug resistance in breast cancer, as evidenced by our single cell RNA sequencing data analyses.


Dr. Andrea Bild

Dr. Andrea Bild trained as a pharmacologist with a specialization in genomics and cancer cell biology. Her research program, currently part of the Integrative Cancer Biology Program through a U01 grant, has established the development and clinical translation of a systems biology framework for personalized medicine genomics. Specifically, her research has enabled 1) investigations of signaling pathways and networks in a physiologically relevant setting: patient tumors; 2) algorithms to personalize matching of effective drugs to patients; and 3) systems-guided clinical trials with novel therapeutic strategies (see biosketch). Dr. Bild has also developed and founded the Genome Science Program at the University of Utah in order to train students and scientists in genomics and systems biology as well as create a rich collaborative structure for faculty across campus with expertise in this field.

Dr. Fred Adler

Dr. Fred Adler trained as an applied mathematician with specialization in dynamical systems and mathematical ecology, and is Director of the Center for Quantitative Biology. His research program, through his joint appointment in the Departments of Mathematics and Biology, ranges from evolutionary ecology to mathematical oncology. These areas are unified through a commitment to finding key mechanisms in complex systems through mathematical modeling and close interaction with experimentalists and data. He has trained 17 graduate students across this broad range of topics, and his most influential research includes modeling the dynamics of cystic fibrosis to optimize the timing of lung transplantation, the dynamics of populations in spatially subdivided landscapes, and modeling of biodiversity in interacting communities. His expertise in collaboration across disciplinary boundaries, linking dynamical systems with data, and mentoring of trainees will effectively integrate the modeling research and computational core into all aspects of this project.

Dr. Adam Cohen

Dr. Adam Cohen is a board certified medical oncologist with a master’s degree in mathematics. His research specializes in clinical trials, genomics, and biomarkers. He has developed and been the PI for three investigator-initiated clinical trials and has been the local PI for many multi-site trials. Dr. Cohen has successfully applied genomic biomarkers in his completed clinical trials, and is the PI for a tissue acquisition protocol that has been used in our successful collection of pleural effusion samples.