PIs: Eric Holland, MD, PhD, Franziska Michor, PhD

Key personnel: Eric Holland, MD, PhD, Franziska Michor, PhD, Kornelia Polyak, MD, PhD, David Scadden, MD

The principal mission of the Dana-Farber Cancer Institute Physical Sciences - Oncology Center (DFCI PS-OC) is to promote the understanding of cancer evolution and treatment responses utilizing approaches from the physical sciences intertwined with cancer biology and oncology. This center will investigate the treatment response of leukemic, brain and breast cancer cells and their microenvironment to classical and novel treatment approaches. To this end, the DFCI PS-OC will develop novel mathematical frameworks describing the evolutionary dynamics of tumor progression and treatment response that are parameterized using the experimental systems, and predict and validate optimum intervention strategies that will ultimately be implemented as prospective clinical trials. This approach will incorporate the spatio-temporal organization of evolving, heterogeneous tumor cell populations within their microenvironment. Collaborations between physical and experimental scientists in the PS-OC will bridge the chasm between the physical sciences and oncology with the ultimate goal of uncovering novel biology and enhancing patient care.


The goal of the PS-OC is to advance our understanding of the physical principles that govern cancer initiation, progression, response to treatment, and the emergence of resistance. The PS-OC brings together an expert team from the fields of mathematics, computational biology, physics, and engineering as well as cancer biology, oncology and surgery, to develop the infrastructure, research programs, and the network required to enable team research converging the physical sciences with cancer biology. Our PS-OC fosters a new trans-disciplinary environment and research that: (1) originates and tests novel, non-traditional physical science-based approaches to understanding and controlling cancer; (2) generates physical measurements and integrates them with existing knowledge of cancer; and (3) develops and evaluates theoretical evolutionary approaches to provide a comprehensive and dynamic picture of cancer. Ultimately, through iterative trans-network development and testing of innovative approaches to cancer processes, our PS-OC is expected to generate new bodies of knowledge and fields of cancer study that define the critical aspects of the physical sciences that operate at all levels in cancer processes.

The overarching framework of our PS-OC is Exploring and Understanding Evolution and Evolutionary Theory in Cancer from a Physics Perspective through the incorporation of theories of Darwinian and somatic evolution with experimental approaches from the physical sciences to better define, understand and control cancer at all levels. In terms of the physical sciences, cancer should be considered as a complex adaptive system that is most appropriately studied in the context of evolution and evolutionary theory. Tumors can be viewed from an evolutionary standpoint as collections of cells that accumulate genetic and epigenetic changes, which are then subjected to the selection pressures within a tissue. These normally heritable variations can lead to adaptations of the cells such as induction of angiogenesis, evolution of resistance or evasion of the immune system. Beneficial heritable changes can cause rapid expansion of the mutant clone since they enable their carriers to outcompete cells that have not accumulated similar improvements. Mutations advantageous to the cancer cell are normally detrimental to the organism, ultimately causing death of both the patient and the tumor. Therefore, neoplastic processes serve as an example for selection acting on different hierarchical levels: clonal evolution generally selects for increased proliferation, survival, and evolvability on the cellular level and leads to progression, invasion, and resistance; the latter effects are selected against on the level of multicellular organisms. 

The investigation of cancer evolution requires mechanistic, quantitative models that incorporate realistic properties of biological systems such as stochasticity and nonlinearity. As the outcomes of such interactions cannot be determined by purely phenomenological reasoning alone, they must be computed from general integrative models of carcinogenesis. Theoretical approaches to tumorigenesis can lead to considerable insights into the natural history of the disease and will enhance the rational and predictive aspects of cancer research.

As a PS-OC, we are developing experimental and theoretical models that can support the advancement of an evolutionary construct for understanding, predicting, and controlling cancer. This construct takes advantage of genomics data, in vitro and in vivo modeling along with appropriate physical measurements for evaluating and testing our theoretical models. The PS-OC adheres to the following general approach (Figure 1): tumors in their temporal, spatial and compositional heterogeneity are being decomposed to various aspects such as time (when do events happen?), space (where do events happen?), and components (what is the distribution of events across all components of the system?).


(Figure 1)

(Figure 2)

This reductionist step allows for the identification of different dimensions of a tumor that are considered as individual parts of the overarching theoretical frameworks. We then build a mathematical representation of the system that explicitly considers the dimensions outlined above. This mathematical framework leads to predictions which will be tested in vitro and in vivo utilizing novel physical measurements such as the distribution of traits in a population of cancer cells, migration of cells along gradients, a temporal and spatial resolution of events, and others. These physical measurements are aimed at representing the in vivo tumor and serve as validation of the mathematical framework. There is be a continuous iterative process between physical measurements and the mathematical framework that allows us to update the theoretical representation of the system to be as accurate as possible (Figure 2).

We first build a basic mathematical framework that allows us to suggest an initial set of experiments to obtain physical measurements of the system, through investigation of both in vitro and in vivo models and with particular emphasis on single-cell measurements to adequately capture the diversity and heterogeneity of the system. These measurements are then used to update the mathematical framework and to decide between mutually exclusive assumptions of the models. As a third step, the updated mathematical framework is used to generate suggestions for the next set of experiments. These rationally planned experiments in turn inform about the validity of the mathematical framework, suggest improvements and modifications and provide more quantitative estimates of parameters that are used for modifying the next version of the mathematical framework, which is in turn utilized to propose novel experiments.

The work of the PS-OC is performed by a key team of investigators with highly diverse backgrounds.


Our PS-OC consists of an interdisciplinary team composed of mathematical and computational biologists, physicists, cancer biologists and oncologists as well as cancer surgeons (Figure 3). Collaboration between investigators with such diverse backgrounds will bridge the divide between the physical sciences and oncology and lead to new evolutionary theories of cancer initiation, progression, response to therapy, and the emergence of resistance.

(Figure 3)


There exists a major economic and humanitarian opportunity to shift interventions in cancer from late stages of refractory disease to earlier stages (potentially pre-surgery). Such early interventions might render surgical treatments unnecessary and would reduce cancer-associated morbidity and mortality by a large amount. To drive the development of earlier cancer interventions and therapies, three major changes in scientific approach, politics, and current ways of implementation are necessary: (i) Endeavors to identify possibilities for early interventions critically depend on an understanding of the temporal sequence in which genetic and epigenetic alterations arise during tumorigenesis. This knowledge can be gained by direct sequencing of cancer cells at different stages of their transformation to malignant invasive tumors in cancer types for which clinicopathologically defined samples are available. However, in tumor types for which such stages cannot be obtained, the reconstruction of the order of events arising during tumorigenesis has not been possible. (ii) Furthermore, additional screening and detection methods must be developed such that tumors can be diagnosed at earlier stages of their progression. (iii) Lastly, another important step of transformation of cancer therapy towards earlier intervention in the tumorigenic process is given by changes in the culture and politics of cancer therapy. The rules of the FDA concerning approval of anti-cancer therapies need to be revised such that more clinical trials are permitted that would test agents in earlier stages of cancer.

The last item discussed above cannot be addressed within the scope of this PS-OC. However, the team of the Center is contributing to the first two items by employing our joint expertise in the physical sciences and oncology to design, test and validate an evolutionary mathematical model to identify the temporal sequence in which pathogenic mutations arise during tumorigenesis. The identification of early mutations in cancer progression will then enable the development of tumor markers, screening and detection methods for early interventions. As outlined below and in greater detail in Project 1, the combined expertise of the PS-OC investigators is utilized to develop and, iteratively with physical, single-cell based measurements in in vivo models, extend and revise a theoretical framework to retrace the order of tumorigenic events. This framework includes considerations like tumor heterogeneity, integration of different data types such as sequence information, copy number alterations, and others to obtain an integrated view of cancer as an evolutionary process. The evolutionary model describes the shift in cell populations that occurs during tumorigenesis, and the genetic and epigenetic alterations that occur in each subpopulation at different times and that lead to different selective effects within the population.


Project Leader: David Scadden, MD (Massachusetts General Hospital)

Acute myeloid leukemia (AML) is a rapidly fatal disease that can quickly be brought into complete remission, but with high relapse rates. The genetic evolution of the disease has been defined, but the basis for resistance to therapy and effective strategies to overcome it are lacking. This project seeks to take advantage of well-defined murine models where an analogous form of highly lethal, human AML can be temporarily brought into remission by traditional chemotherapy agents. Combining these basic features with novel strategies for quantitatively assessing oncological behavior and cell growth patterns, this project will provide multidimensional datasets for mathematical modeling of disease progression and susceptibility to therapeutic approaches. Ultimately, these models will be used to predict and test unique vulnerabilities of the disease that can be exploited therapeutically to reduce AML relapse.

The focus of interventions on earlier stages of cancer development requires more sensitive and quantitative measurements down to the level of single cells. To obtain such non-traditional physical measurements of cells, single cell profiling will be performed. There will be two areas of focus: 1) development of quantitative assays for tumor growth and death rates for input to as well as validation and refinement of the mathematical modeling effort; and 2) development of new single-cell measurements to assess the heterogeneity of cell signaling, cell cycle regulation, and apoptosis in primary tumors. An appreciation of phenotypic diversity in a clonal population of cells that drive population dynamics of tumors is essential for obtaining an integrated view of cancer initiation and progression and will be addressed by the investigators of this PS-OC. Our approach is designed to identify vulnerabilities of cancer cells that arise at different stages of disease; these vulnerabilities can be broken down with respect to different subpopulations of cells and different phenotypic effects that these changes exert on cells.

Apart from identifying the temporal sequence of events arising during tumorigenesis, it is also important to identify the spatial location and type of cells initiating and driving cancer progression. The question about the cell of origin of human cancers is essential in that this answer influences treatment choices and increases the understanding of the natural history of tumors and their evolution. In Project 2, we design a mathematical framework of the evolutionary dynamics of cell populations that can be used to identify the cell of origin of cancers. This framework is then iteratively updated and revised in conjunction with in vivo experiments in brain cancer and leukemia models. The framework is being extended to include stochastic three-dimensional computer simulations, investigations of heterogeneity due to differentiation, evolution, and spatial location, a study of motility and gradients of factors within the tissue, and single cell measurements performed.



Project Leader: Eric Holland, MD, PhD (Fred Hutchinson Cancer Research Center)

Glioblastomas (GBMs) are the most common and most malignant primary brain tumors and can be divided into several molecular subgroups. All subtypes demonstrate high insensitivity to standard therapeutic approaches and, unfortunately, recurrence inevitably occurs. This project aims to better understand particular subtypes of GBMs and optimize intervention strategies. Particularly, the investigators hope to elucidate the combined effects of chemotherapy and radiation treatment on proneural-GBMs, to maximize radiation therapy responses in mesenchymal-GBMs and improve overall GBM treatment strategies independent of tumor subtype. Utilizing a multidisciplinary approach of basic biological and novel mathematical models, the investigators hope to further the understanding of therapeutic dynamics of treatment response to ultimately provide optimized and novel intervention strategies for GBM and improve the standard of care for patients.

Again, single cell measurements are essential for identifying the cell of origin of human cancers and the parameters that influence the likelihood that cells of a particular differentiation stage accumulate the mutations leading to cancer. Single cell measurements are also used to approach these questions and provides the investigators of this project with measurements and parameter values to refine the mathematical model and validate its predictions.

Connected to the sequence of events arising during tumorigenesis is the point at which resistance to conventional or targeted therapies arises. The nucleation of resistance might arise much earlier than the physiological manifestation of such resistance; therefore the progression to therapy-refractory disease must be built into the population dynamics models which can then be used to predict and eventually prevent the emergence of acquired resistance. This question is at the heart of Project 3 of our PS-OC, in which we design and analyze a stochastic mathematical model of the evolution of resistance in heterogeneous populations of cells. This mathematical model, together with parameter values gleaned from disease-specific cell line and mouse models as well as pharmacokinetic data from human patients, is used to identify the optimum treatment administration strategy to maximally delay the emergence of resistance against targeted drugs and radiation therapy. The predictions of the mathematical framework are then tested in disease-specific in vitro and in vivo models.



Project Leader: Kornelia Polyak, MD, PhD (Dana-Farber Cancer Institute)

Breast cancer is the most commonly diagnosed cancer and is the main cause of cancer-related mortality in women worldwide. Achieving a meaningful global impact on breast cancer-associated morbidity and mortality requires a better understanding of drivers of metastatic progression and therapeutic resistance at the cellular and molecular level, a way to recognize these features in diagnostic patient samples, and early intervention and treatment strategies that prevent the emergence of metastatic and treatment-resistant tumors. The goals of this project are to tackle these key clinical problems in breast cancer through a combined mathematical modeling and experimental approach utilizing single cell analyses and xenograft models to identify optimal therapeutic strategies to ultimately prevent metastatic outgrowth and treatment resistance. Ultimately, these models will be validated in clinical and xenograft samples and they will be utilized to combat primary tumor and metastases formation.

An integrated approach of physical sciences and oncology is essential for driving the understanding of cancer biology since it allows for the development of overarching theories that are aimed at predicting and preventing cancer initiation and progression as well as the emergence of resistance. Cancer research today is limited by two principal difficulties: (i) measurements are limited to investigations of the bulk of tumor cells rather than stratified cell populations, separated by differentiation status, number of pathogenic mutations, spatial location, response to therapy etc. It has been proved difficult, both experimentally and theoretically, to get to the fine details of the heterogeneity of human tumors, and the extent to which a small number of progenitor cells drives tumor formation. Similarly, surgical practice has been focused on the tumor as a whole. Pathology is traditionally tissue-oriented and is only slowly beginning to investigate tumors at a deep molecular resolution. This discipline is still lagging behind in the general application and research of immunohistochemical markers, sequencing techniques and modern genomics approaches to cancer. (ii) Early detection has been a major limiting factor. It is a reasonable hypothesis to think that the early events in tumorigenesis will inform about the key cellular factors in tumor development.

We believe that there are two ways of approaching these limitations. First, the focus of investigations can be predominantly experimental, through physical measurement of tumor heterogeneity, spatial location of cells, and the temporal sequence of events that lead to a mature tumor. Therefore, a part of our approach is to perform such measurements to elucidate the cellular and molecular details of cancer. While those techniques are being developed, another approach is to develop a conceptual mathematical framework of oncogenesis. This approach is time tested and generally applied in the physical and natural sciences, and no major area in physics lacks a general mathematical framework used to predict the properties of systems. Without such general mathematical frameworks, investigations remain phenomenological and rational, and rational predictions cannot be made. We are therefore developing a theoretical framework, cast in mathematical form, that is specifically tested and validated with rationally designed experimental approaches. These experiments are chosen to be the investigations most critical to unraveling parameters that are central to the investigations, and that are able to discriminate between alternative mechanistic models. With these two integrated approaches, we intend to unravel the key molecular events that contribute to tumor progression.

The intended impact of our PS-OC is that its operations – the investigation of mutations arising during tumorigenesis, their cellular origin and strategies to prevent the emergence of resistance – are carried out in close collaboration with clinical investigators and will eventually lead to theoretical investigations in conjunction with clinical trials. This paradigm shift from segregated research environments to integrated analyses of physicists, engineers, mathematicians and clinical oncologists, cancer biologists and surgeons is the ultimate goal of the PS-OC. So far such interdisciplinary approaches have been implemented anecdotally but have not been carried out systematically. We envision that our detailed theoretical investigations of the evolutionary processes leading to cancer, in close collaboration with cancer biologists, oncologists, and surgeons, will eventually lead to rationally designed molecular profiling and interventions in clinical settings.


S5 Box

Login Form