Workshop 4: Tumor Heterogeneity and the Microenvironment

(February 2,2015 - February 6,2015 )

Organizers


Alexander Anderson
Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center & Research Institute
Trevor Graham
Tumour Biology, Barts Cancer Institute, QMUL
Michael Ostrowski
Ohio State Comprehensive Cancer Center, Ohio State University
Charlie Swanton
London Research Institute

Heterogeneity in cancer is an observed fact, both genetically and phenotypically. Cell-cell variation is seen in almost all aspects of cancer from early development through to invasion and subsequent metastasis. Our current understanding of this heterogeneity has mainly focused at the genetic scale with little information on how this variation translates to actual changes in cell phenotypic behavior. Given that many genotypes can lead to the same cellular phenotype, we must also understand the range and scope of this heterogeneity at the phenotypic scale as ultimately this variability will dictate the aggressiveness of the tumor and its treatability. Central to our understanding of this heterogeneity is how the tumor cells interact with each other and with their microenvironment.

The tumor microenvironment is not simply the extra cellular matrix, but a complex milieu consisting of growth promoting and inhibiting factors, nutrients (including oxygen and glucose), chemokines, and importantly other cell types including (but not limited to) fibroblasts, immune cells, endothelial cells and normal epithelial cells. These microenvironmental factors and different cell types interact with one another and the tumor as it grows. The role of endothelial cells and the immune system in cancer development are fairly well established, but less is known about the function of host fibroblasts in this process. Most solid tumors present as dense fibrotic masses, which suggests that fibroblasts contribute to tumor growth by infiltrating and depositing extracellular matrix proteins. In addition, the phenotype of fibroblasts found within and around tumors (activated fibroblasts or cancer associated fibroblasts: CAFs) is different to normal fibroblasts, and closely resembles myofibroblasts. Fibroblasts act in wound healing, angiogenesis and tissue remodeling by releasing growth factors and proteases such as matrix metalloproteinases. They also deposit matrix proteins such as laminin, tenascin and fibronectin. Therefore, if the growing tumor can co-opt such fibroblasts it has an unlimited source of many of the fundamental elements required for growth and invasion.

The two central themes of this workshop are:

  • Heterogeneity (be it phenotypic, signaling or genotypic), and
  • Microenvironment (ECM, nutrients, fibroblasts and immune cells).

Since a highly heterogeneous tumor has the potential to adapt to any microenvironment, understanding how interactions between the growing tumor and its microenvironment modulate tumor heterogeneity is critical to unraveling the mechanisms of cancer initiation.

Accepted Speakers

Alexander Anderson
Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center & Research Institute
Fran Balkwill
Centre for Cancer and Inflammation, Queen Mary University of London
Mary Helen Barcellos-Hoff
Radiation Oncology, New York University School of Medicine
David Basanta
Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center
Robert Clarke
Office of the Dean for Research, Georgetown University Medical Center
Joseph Costello
Department of Neurological Surgery, University of California, San Francisco
Christina Curtis
Medicine and Genetics, Stanford University
Elza De Bruin
Research Dep of Oncology, University College London Cancer Institute
Andrew Ewald
Cell Biology, Johns Hopkins University
Jude Fitzgibbon
Center for Haemato-Oncology, Queen Mary, University of London
Robert Gatenby
H. Lee Moffitt Cancer Center & Research Institute
Philip Gerlee
Mathematical Sciences, Chalmers University of Technology
Trevor Graham
Tumour Biology, Barts Cancer Institute, QMUL
Simon Hayward
Surgery, NorthShore University HealthSystem
Sui Huang
., Institute for Systems Biology
Shelley Hwang
Surgery, Duke University
Simon Leedham
Wellcome Trust Centre for Human Genetics, University of Oxford
Gustavo Leone
Molecular Virology, Immunology, and Medical Genetics, The Ohio State University
Morag Park
Biochemistry and Oncology, McGill University
Vito Quaranta
Vanderbilt Ingram Cancer Biology Center, Vanderbilt University
Sergio Quezada
Research department of haematology, University College London Cancer Institute
Erik Sahai
Tumor Cell Biology Lab, London Research Institute
Owen Sansom
Colorectal Cancer and Wnt Signalling, Cancer Research UK Beatson Insititue
Andrea Sottoriva
Centre for Evolution and Cancer, The Institute of Cancer Research
Thea Tlsty
Department of Pathology, University of California, San Francisco
Richard White
Gastrointestinal Oncology Service, Memorial Sloan-Kettering Cancer Center
Daniel Worthley
Medicine, University of Adelaide
Yinyin Yuan
Centre for Evolution and Cancer, The Institute of Cancer Research
Monday, February 2, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:30 AM

Greetings and info from MBI - Marty Golubitsky

09:30 AM
10:15 AM
Mary Helen Barcellos-Hoff - Carcinogenesis in Context: Tumor Heterogeneity as a Function of Host Biology

Both clinical and experimental data show that the stroma is highly involved in early malignancy, supporting the idea of reciprocal evolution of the malignant cell and the tumor microenvironment. Although it is clear that stroma composition and signaling is altered in cancer, less is known about how and when stroma contributes to carcinogenesis and how carcinogens, like radiation, might alter these processes. To investigate how tumor diversity evolves in the context of different physiological states (e.g. as a function of age and ionizing radiation exposure) we developed a mammary chimera model in which histologically and genomically diverse cancer originate from a p53 null epithelium orthotopically transplanted to a syngeneic wildtype host. We use expression profiling of tissue and tumors to identify the biological ‘bookends’, those processes initiated in normal tissue that appear to contribute to the development and are recapitulated in particular types of tumors. These studies show that the spectrum of mammary tumor types is strongly influenced by the host biology, opening new perspective on the drivers of aggressive cancer. The mounting evidence from these and other studies that cancer results from a systemic failure in which cells other than those with oncogenic alterations determine the frequency and type of clinical cancer is changing the cancer paradigm. To account for this, we propose that the tumor microenvironment is built through rate-limiting steps during multi-stage carcinogenesis [Barcellos-Hoff, 2013 #18417]. In this model, construction of a ‘pre-cancer niche’ is a necessary early step required for initiated cells to survive and evolve; subsequent niche expansion and maturation accompany promotion and progression respectively. The model postulates that cancer cell survival and proliferation is as much a function of the successful niche construction as it is of the natural selection for specific cancer cell mutations. Consequently, cancer represents an emergent property that requires a comprehensive analysis of the cell-cell interactions during the course of carcinogenesis. Moreover, in contrast to initiation, which is stochastic by nature, niche construction represents a robust target for native immunosuppression and a potent target for cancer prevention.

10:15 AM
11:00 AM
Christina Curtis - A Big Bang model of human colorectal tumor growth

What happens in the early and still undetectable human malignancy is unknown because direct observations are impractical. Here I will describe a novel “Big Bang” model, whereby a tumor grows predominantly as a single expansion producing numerous intermixed sub-clones, which are not subject to stringent clonal selection. In this model, both public and most detectable private mutations arise during the earliest phase of tumor growth. Multi-scale genomic profiling of 349 individual glands sampled from 15 colorectal tumors revealed the absence of selective sweeps, uniformly high intra-tumor heterogeneity, and sub-clone mixing in distant tumor regions, as postulated by the Big Bang. By integrating the data in a spatial model of tumor growth and statistical inference framework we also verified the most striking prediction of our model, namely that most detectable intra-tumor heterogeneity originates from private alterations acquired early during growth, and not from the later expansion of selected sub-clones. Hence, early sub-clones define the genomic profile of colorectal carcinomas and advanced adenomas, whereas potentially dangerous late-arising sub-clones will go undetected. Moreover, our results suggest that sub-clone mixing may be a biomarker of malignant potential. This new model provides a quantitative framework that explains the origins of intra-tumor heterogeneity and tumor growth dynamics with significant clinical implications for treatment resistance and metastatic progression, as I will discuss.

11:00 AM
11:30 AM

Break

11:30 AM
12:15 PM
Thea Tlsty - Environmental Control of Cellular States Associated with Plasticity and Cell Fate Decisions

The product of the p16INK4a/CDKN2A locus encodes a cyclin-dependent kinase (CDK) inhibitor that functions as a negative regulator of cyclin/CDK complexes. In addition to governing regulation through the cell cycle, this protein also regulates the production of chromatin remodeling proteins downstream of E2F. We identified cell surface markers associated with repression of p16INK4a/CDKN2A and found that they allowed direct isolation of rare cells from healthy human breast tissue that exhibit extensive lineage plasticity. This subpopulation of cells has the ability to enter a state that can transcribe pluripotency markers, Oct3/4, Sox2 and Nanog at levels similar to those measured in human embryonic stem cells and to acquire an epigenetically plastic state sensitive to environmental programming. In vitro, in vivo and teratoma assays demonstrated that either a directly-sorted (uncultured) or a single cell-derived (clonogenic) cell population from primary human tissue can differentiate into functional derivatives of each germ layer, ectodermal, endodermal and mesodermal. In contrast to other cells that express Oct3/4, Sox2 and Nanog, these human endogenous Plastic Somatic cells (ePS cells) are mortal, express low telomerase activity, expand for an extensive but finite number of population doublings, and maintain a diploid karyotype before arresting in G1. These types of cellular states, exhibiting elevated plasticity, can be directed towards multiple cell fates dictated by the microenvironment.

12:15 PM
01:00 PM
Sui Huang - Tumor Progression: Non-Genetic Cancer Cell Population Heterogeneity and Plasticity Drive Non-Darwinian Somatic Evolution

Tumor progression and generation of therapy resistant cells following therapy are not simply a manifestation of Darwin’s “Survival Of The Fittest” but rather, of Nietzsche’s “What Does Not Kill Me Makes Me Stronger”. The robust arrow of progression towards increasingly sophisticated aggressiveness of cancer cells during the course of treatment, including therapy resistance, is typically explained by a somatic Darwinian evolution: selection of cancer cell clones that carry random genomic mutations that (by chance) confer the new more malignant phenotype. However, the immense non-genetic heterogeneity and the plasticity of cell phenotype offer a new level for evolutionary mechanism that transcends (but does not exclude) the Neo-Darwinian scheme. By relaxing the rigid genotype-phenotype relationship that is tacitly assumed to explain the Darwinian expansion of the fittest mutant clone, cell phenotype plasticity accelerates evolution. The plasticity of cell phenotype is not random: it is constrained and channeled by the principles of stochastic, nonlinear dynamics through which the gene regulatory network (GRN) coordinates gene expression to produce complex, coherent phenotypic states of the cell. In this framework we postulate that cancer cells are “stuck” in “cancer attractors”: latent, normally unused stable states in the vast theoretical space of all possible gene expression patterns of the genome. Cancer attractors are the inevitableby-products of the trial/error scheme of rewiring the GRN during evolution. They can be imagined as the normally unoccupied side-valleys in Waddington’s epigenetic landscape that must be avoided during the developmental descent of immature (stem) cells to the mature states of differentiation, the lowest point in the main valleys. Near-lethal therapeutic intervention on cells stuck in such abnormal attractor states opens –via symmetry-breaking bifurcations that epitomize the desired destabilization of the cancerous cell states– access to yet more aberrant attractors in the unused regions of the landscape further away from the normal developmental trajectories. They encode even less mature, “degenerate” states, and following such bifurcations are inevitably occupied by some cancerous cells for simple entropy reasons.

In this talk I will explain the basic principles of non-Darwinian dynamics on the quasi-potential (=epigenetic) landscape of the GRN and why therapy, or any attempt to destabilize the cancerous state, will, in addition to intended apoptosis also cause “rebellious cells” to spill over into attractors of increased stemness and hence, promote malignancy and produce therapy resistance. Experimental data that document this phenomenon will be presented as well as results of first attempts to curb such evasive behavior.

01:00 PM
02:15 PM

Lunch Break

02:15 PM
03:00 PM
Joseph Costello - Under Pressure: Tumor evolution after therapy

Tumor recurrence is a leading cause of cancer mortality. Therapies for recurrent disease may fail, at least in part, because the genomic alterations driving the growth of recurrences are distinct from those in the initial tumor. At diagnosis, low grade brain tumors have similar histology and driver mutations, but following surgical resection their clinical behavior is highly variable. Infiltrating tumor cells comprising the residual disease may remain indolent for more than a decade after the surgical resection, or may rapidly transform into an aggressive rapidly growing malignant tumor. Adjuvant chemotherapies such as temozolomide (TMZ) are frequently used, especially in cases with subtotal surgical resections or when deferring the use of radiation therapy is preferable. We used genome and epigenome sequencing technologies to infer patterns of intratumoral heterogeneity and tumor evolution over time. In this presentation I will discuss new insight into diverging genetic pathways arising from initially similar tumors, and the profound influence of chemotherapy in shaping tumor aggressiveness.

03:00 PM
03:45 PM
Elza De Bruin - Genetic intra-tumour heterogeneity sheds light on lung cancer evolution

Lung cancer is the most common cancer worldwide, and the 5-years overall survival remains poor. Understanding the genetic evolution of the most common type of lung cancer, non-small lung cancer (NSCLC), and resolving which mutations occur early in a tumour’s evolution (clonal) and which mutations are involved at later stages (subclonal) may improve therapeutic interventions. We performed multi-region exome and/or genome sequencing at a mean depthof >200x to analyse the intra-tumour heterogeneity in a total of 44 tumour regions from fourteen primary NSCLC samples. Non-silent mutations were validated using ultra-seek amplicon sequencing. All samples revealed potential driver mutations occurring on the trunk (clonal mutations) and on the branches of the phylogenetic trees (subclonal mutations), although the majority of driver mutations and large-scale genomic events such as genome doubling occurred predominantly early in the life history of these tumours. On-going chromosomal instability enhanced the intra-tumour heterogeneity at later stages of tumour development, and also potential driver mutations could be identified. Detailed analyses of the mutational spectra identified several mutational processes involved in NSCLC tumour evolution, with some processes predominantly involved in early mutations, and others driving later mutations. The relative contribution of these later processes show regional variation, thereby contributing to the intra-tumour heterogeneity in NSCLC.

03:45 PM
04:00 PM

Break

04:00 PM
04:45 PM
Robert Gatenby - Evolutionary triage governs fitness in driver and passenger mutations and suggests targeting never mutations

Genetic and epigenetic changes in cancer cells are typically divided into "drivers" and "passengers" Drug development strategies target driver mutations, but inter- and intra-tumoral heterogeneity usually results in emergence of resistance. Here we model intratumoral evolution in the context of a fecundity/survivorship trade-off. Simulations demonstrate the fitness value, of any genetic change is not fixed but dependent on evolutionary triage governed by initial cell properties, current selection forces, and prior genotypic/phenotypic trajectories. We demonstrate spatial variations in molecular properties of tumor cells are the result of changes in environmental selection forces such as blood flow. Simulated therapies targeting fitness-increasing (driver) mutations usually decrease the tumor burden but almost inevitably fail due to population heterogeneity. An alternative strategy targets gene mutations that are never observed. Because up or down regulation of these genes unconditionally reduces cellular fitness, they are eliminated by evolutionary triage but can be exploited for targeted therapy.

04:45 PM
05:00 PM

Discussion

05:00 PM
07:00 PM

Reception and Poster Session

07:00 PM

Shuttle pick-up from MBI

Tuesday, February 3, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Richard White - Assessing normal and cancer somatic variation with RAD-seq

Both normal and cancerous tissues are able to acquire genetic variation over time. The spectrum of cancer mutations has been well characterized through projects such as the TCGA, yet the degree of normal somatic variation within an individual, called somatic mosaicism, has been poorly characterized. To understand cancer phylogenies will require a deeper appreciation and understanding of normal somatic variation, but techniques such as whole-genome and exome sequencing are not efficient for the construction of such trees. To overcome this, we are using a reduced representation sequencing technique called RAD-seq to assess genetic variation across normal and cancer tissues. In this approach, DNA samples are digested with a restriction enzyme to give a desired amount of genome coverage, and then only those portions of the genome containing those cut sites undergo Illumina sequencing. We have applied the RAD-seq pipeline to a series of normal and cancer tissues from both humans (pancreatic cancer patients) as well as zebrafish (melanoma bearing animals), sequencing both the tumor itself, individual metastases, and several normal tissues such as lung and brain. We are using this data to understand the patterns of SNVs and copy number changes in each of these tissues, and across species. The application of RAD-seq will allow for high-throughput, deep coverage of genomic variation at a fraction of the cost of traditional sequencing approaches, and will be broadly applicable to experiments focusing on cancer evolution.

09:45 AM
10:30 AM
Alexander Anderson - Predicting Cancer Evolution: Heterogeneity and the Genotype-Phenotype Mapping

Heterogeneity in cancer is an observed fact, both genetically and phenotypically. Intercellular variation is seen at all scales and stages of development, and has significant implications for prognosis. At present, our understanding of this heterogeneity is mainly restricted to the genetic scale with little information regarding the relationship between genetic and phenotypic heterogeneity. Further, little is known about how cells alter their microenvironment or how these changes drive selection and feedback to further drive cancer evolution.

Strong selective pressures imposed by a milieu of microenvironmental factors combined with high profileration rates and high mutation rates inevitably lead to the rapid emergence of resistance to therapy. Hence, the failure of cancer therapies is often attributed to Darwinian evolution. To understand and predict cancer evolution we must understand not only the mutations which drive evolution but also the mechanisms through which these mutations manifest themselves in phenotypic change. Thus, our success in predicting cancer progression and designing effective therapy is contingent on understanding the junction at which genes and environment meet to produce phenotypes, the genotype-phenotype (GP) map.

Experimental studies have revealed the complexity inherent within the GP-map which is responsible for the difficulty in predicting evolution; many genotypes produce identical phenotypes and further many phenotypes can emerge from a single phenotype. Indeed, recent experimental evidence shows that this mapping produces phenotypic heterogeneity through a variety of genetic and non--genetic mechanisms. Heterogeneity can be driven through phenotypic plasticity, the phenomenon whereby isogenic cells in different environments display different phenotypes. Further, isogenic cells in identical environments can display phenotypic heterogeneity which is the manifestation of intra--cellular noise amplified through the complex machinery of the cell signalling pathways. In this talk I will present a collection of related mathematical models which explore genetic, environmental, phenotypic and morphological heterogeneity through the unifying lens of the GP-map, outline the key mechanisms which could be responsible for generating adaptive phenotypes -- mutation, plasticity and stochasticity -- and explore the implications of these mechanisms for developing novel and effective cancer therapies.

10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Simon Leedham - Morphogenic and microenvironmental control of intestinal stem cell fate

The rapid turnover of the intestinal epithelium is supported by tissue specific stem cells thought to be located at the base of the intestinal crypt - the basic functional unit of the gut. Stem cell progeny progressively proliferate and differentiate as they move out of the crypt base stem cell niche and along the crypt-villus axis of the intestine. This system is carefully orchestrated by interacting endogenous epithelial and paracrine secreted morphogen gradients, with cell fate coupled to, and determined by, a cells position within these gradients. It has been recently suggested that stem cell fate is similarly influenced by microenvironmental factors, and that stem cell function is not cell autonomous. Recent work has shown that cells outside of the stem cell niche can reacquire stem cell properties to assist in crypt regeneration and that disruption of mucosal morphogenic gradients can cause tumour initiation from cells that have exited the crypt base stem cell niche.

This talk will explore the exciting developments in the homeostatic and pathological microenvironmental control of intestinal stem cells and discuss the implications of this for colorectal cancer heterogeneity and chemotherapeutic strategy.

11:45 AM
12:30 PM
Daniel Worthley - Connective tissue stem cells in the tumor microenvironment

The tumor microenvironment presents an exciting opportunity for innovative therapeutic approaches to cancer. The origin and exact biological contribution of the peritumoral connective tissue at the primary and metastatic sites, however, is still uncertain. Just as one would not assess the average of “hematopoietic” contribution to the tumor microenvironment without considering distinct cell types, so too lumping together all “peritumoral mesenchyme” or “cancer-associated fibroblasts” on the basis of broad markers is likely to oversimplify a diverse connective-tissue contribution to cancer. In this presentation I will discuss the types of connective tissue cells in cancer. I will outline recent studies addressing the biological function of these cells in cancer and discuss our own research on the heterogeneity of connective tissue stem cells in the intestine, the bone and in several cancer models. Understanding the biological heterogeneity of mesenchymal cells in cancer will provide new opportunities for targeted cancer prevention and therapy.

12:30 PM
01:45 PM

Lunch Break

01:45 PM
02:30 PM
Owen Sansom - The influence on tumour driver mutations on the microenvironment

It is has become clear that the microenvironment plays a very important role in tumour progression with recent studies showing both tumour suppressive and promoting roles. Our previous work has identified the importance of neutrophils in the both inflammation and spontaneous models of carcinogenesis.

I will discuss our latest findings examining how changes in the microenvironment affect epithelial cancer development. I will show latest data on the interaction between epithelial cancer cells and surrounding stromal cells and how manipulation of key signalling pathways can either accelerate or suppress tumor development.

02:30 PM
04:15 PM

Breakout Session

04:15 PM
05:00 PM
Philip Gerlee - The evolution of carrying capacity in constrained and expanding tumour cell populations

Cancer cells are known to modify their micro-environment such that it can sustain a larger population, or, in ecological terms, they construct a niche which increases the carrying capacity of the population. It hashowever been argued that niche construction, which benefits all cells in the tumour, would be selected against since cheaters could reap the benefits without paying the cost. We have investigated the impact of niche specificity on tumour evolution using an individual based model of breast tumour growth, in which the carrying capacity of each cell consists of two components: an intrinsic, subclone-specific part and a contribution from all neighbouring cells. Analysis of the model shows that the ability of a mutant to invade a resident population depends strongly on the specificity. When specificity is low selection is mostly on growth rate, while high specificity shifts selection towards increased carrying capacity. Further, we show that the long-term evolution of the system can be predicted using adaptive dynamics. By comparing the results from a spatially structured vs. well-mixed population we show that spatial structure restores selection for carrying capacity even at zero specificity, which poses a possible solution to the niche construction dilemma. Lastly, we show that an expanding population exhibits spatially variable selection pressure, where cells at the leading edge exhibit higher growth rate and lower carrying capacity than those at the centre of the tumour.

05:00 PM

Shuttle pick-up from MBI

Wednesday, February 4, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Vito Quaranta - Data-driven modeling of the heterogeneity of targeted therapy response in oncogene addicted cancers links single-cell level to cell-population outcomes

With a high-throughput colony Fractional Proliferation (cFP) assay, we simultaneously track in real-time the proliferation dynamics of hundreds to thousands of single-cell derived clones in a cell population exposed to perturbations (Frick et al, 2015, DOI: 10.1002/jcp.24888). In the mutant EGFR-addicted PC9 lung cancer cell line treated with erlotinib, cell fates (death, quiescence, continued proliferation) within each clone vary from cell-to-cell, even between siblings. This widespread heterogeneity of drug response is captured by a new metric, the drug-induced proliferation (DIP) rate, which encapsulates single-cell variation into a dynamic measure of drug response outcomes.

DIP rates variation from colony to colony in PC9 is approximately normally distributed, a strong indication it arises from stochastic sources. Measurement error or mixed colony ancestry could not account for this variation, since DIP rates of PC9 sublines isolated from single cells and propagated in long-term culture (PC9-DS1/95) exhibited the same normal distribution and maintained it for over 25 generations. Similar distributions were obtained from many additional oncogene-addicted cell lines, rigorously re-derived from single cells. Thus, a mutated driver oncogene does not ensure cell-to-cell homogeneity of response, even when genetic background diversity is minimized.

To explore whether these distributions are of consequence to treatment, we constructed a Polyclonal Growth (PG) mathematical model able to incorporate theoretical or experimental DIP rates as parameters. Since DIP rate distributions are normal, they are entirely defined by two parameters, mean and variance. Inputting the average DIP rate of parental PC9 predicts that the cell line as a whole will completely succumb to treatment. In contrast, with the DIP rate distribution parameters as input, a completely different result was obtained: the size of the erlotinib treated population rebounded to initial values after ~11 days, after an initial drop to half the value at 5 days. The PG model predicted similar dynamics of erlotinib response for several mutant EGFR-addicted cell lines: in every cell line tested, rebound occurred within days to weeks, after initial drops to varying depths. Time to rebound is affected primarily by the extent to which the right tail of the DIP rate distribution extends into positive territory. Using stochastic simulations of the PG model, we are able to differentiate the effects of clonal heterogeneity from those of stochastic cell fate decisions (intrinsic noise) that cause significant variability in the response trajectories, including response depth and duration. Predictions were validated experimentally in PC9. It is unlikely that conventional acquired resistance was responsible for the rebound, since SNaPshot multigene assays were negative and response dynamics were inconsistent with a model of rare drug resistant clones. These findings suggest that, even in the absence of acquired genetic resistance, heterogeneity of drug response promotes rebound of the treated population. We propose that these experimental and modeling tools (cFP assay and PG model) enable realistic evaluation of depth and duration of response to targeted drug treatment. Expected and unexpected PG model predictions and suggested avenues for treatment, especially drug combinations, will be discussed.

09:45 AM
10:30 AM
Andrea Sottoriva - Neutral evolution and star-like phylogenies in next-generation sequencing data

Despite extraordinary efforts to profile cancer genomes on a large scale, interpreting the vast amount of genomic data in light of cancer evolution and in a clinically relevant manner remains challenging. This is complicated by interpatient variation and extensive intra-tumor heterogeneity. In particular, the relationship between the cancer genotype and phenotype remains largely unknown as phenotypical characteristics are hard to measure. A critical issue is the lack of a theoretical framework of reference able to make predictions on existing data. Using a multi-sampling strategy we have recently shown how colorectal cancers grow as a single “Big Bang” expansion populated by many intermixed sub-clones that are not subject to stringent selection. Here we demonstrate that this signature of effectively-neutral evolution can be detected in next-generation sequencing data from bulk samples as well, as a result of the star-like tumor phylogenies predicted by the Big Bang growth. In particular, we present a simple mathematical model of cancer expansion based on neutral evolutionary dynamics that predicts the spectrum of alterations reported by nextgeneration sequencing in a large proportion of cases. This hidden feature of cancer evolution is common to multiple tumor types and can be detected in different independent cohorts. Importantly, this allows the direct measurement in each individual patient of the in vivo mutation rate per division and the timing of mutations using currently available sequencing data. This result provides a new way to interpret the wealth of cancer genomic data available to date, shedding new light on the relationship between the cancer genotype and phenotype.

10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Andrew Ewald
11:45 AM
12:30 PM
David Basanta
12:30 PM
01:45 PM

Lunch Break

01:45 PM
02:30 PM
Erik Sahai - Intravital imaging reveals how BRaf inhibition generates drug tolerant microenvironments

Many tumours show an initial response to targeted therapies before genetic resistance emerges, however little is known about how tumour cells tolerate therapy before genetic resistance dominates. In this study, we have used both intravial ratiometic FRET and FLIM of an ERK/MAP kinase biosensor to investigate heterogeneity in signalling in melanoma models. Moreover, we have longitudinally monitored responses to targeted therapy and identified areas that become refractory to drug action. BRaf mutant melanoma cells can rapidly become tolerant to PLX4720 in areas of high stroma. The rapid kinetics of this process indicate that it is not caused by genetic events. We demonstrate that PLX4720 has an unexpected effect on the tumour stroma leading to enhanced matrix remodelling. The remodelled matrix then provides signals that enable melanoma cells to tolerate PLX4720. We propose that this safe haven enhances the population of cancer cells from which genetically resistance emerges. This work highlights the utility of intravital imaging in understanding the reason for therapy failure.

02:30 PM
04:15 PM

Breakout Discussion

04:15 PM
05:00 PM
Robert Clarke
05:00 PM

Shuttle pick-up from MBI

Thursday, February 5, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Sergio Quezada - Targeting immune-checkpoints in cancer: New mechanistic insights

The continual interplay between the immune system and cancerous cells is thought to result in the establishment of a dynamic state of equilibrium. This equilibrium depends on the balance between subsets of effector and regulatory lymphocytes. Whereas the overall mechanisms underpinning the establishment and maintenance of the intra-tumour balance between Teff and Treg cells remain unknown, in many solid cancers it is characterized by the dominant infiltration of regulatory T cells over effector T cells resulting in a low Teff/Treg ratio. Furthermore, different subtypes of regulatory cells and inhibitory molecules such as CTLA-4 tightly control the few effector T lymphocytes that manage to infiltrate the tumour. The outcome of this balance is critical to survival, and while in a few cases the equilibrium resolves in the elimination of the tumour by the immune system, in many other cases the tumour manages to escape immune control.

Remarkably, antibodies against CTLA-4, a key immune modulatory receptor expressed on T cells, efficiently modify this balance, driving effector T cell expansion and increasing the ratio of Teff/Treg within the tumour. Whilst the high Teff/Treg ratio driven by anti-CTLA-4 directly correlates with tumour destruction in mice and humans, the mechanisms underpinning this phenomenon remain unknown.

By focusing in the study of effector and regulatory tumour-reactive CD4+ T cells my group is interested in the mechanism underpinning the activity of different immune-modulatory antibodies within the tumour microenvironment, and the potential positive and negative impact that the tumour microenvironment may have in the recruitment, survival and function of different T cell subsets. In this context and using a murine model of melanoma we have recently demonstrated that both, the change in the Teff/Treg balance as well as tumour rejection, depend on the selective depletion of tumour-infiltrating Treg cells expressing high levels of surface CTLA-4. Regulatory T cell depletion is mediated by ADCC and completely depends on the expression of FcRIV on tumour infiltrating CD11b+ myeloid cells. These results reveal novel and unexpected mechanistic insight into the activity of anti-CTLA-4-based cancer immunotherapy, and illustrate the importance of specific features of the tumour microenvironment on the final outcome of antibody-based immune-modulatory therapies.

09:45 AM
10:30 AM
Fran Balkwill - The evolving tumour microenvironment of high grade serous ovarian cancer, HGSC

I will be describing our studies on the tumour microenvironment of high grade serous ovarian cancer and inflammatory cytokine targets in this cancer.

10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Yinyin Yuan - The Ecosystem Diversity Landscape of Breast Cancer

It is increasingly recognised that the tumor microenvironment is an important determinant in cancer progression and evolution, where tumors act as complex ecosytems involving interactions between cancer cells, stromal cells and their physical environment. To study this we developed a statistical model to systematically quantify the spatial heterogeneity of the tumor ecosystem based on automated image analysis of 1,026 Hematoxylin & Eosin (H&E) stained primary breast tumors. I will discuss the clinical implication of heterogeneous tumour ecosystem, how ecosystem diversity was capable of predicting prognosis independent to known clinical and cancer heterogeneity parameters, before moving on to the bioinformatics integration of this measurement with whole-genome genomic profiling data for all of these tumours. This helped us reveal enrichment of specific copy number alterations for certain genes in this subtype, for which an RNAi screen showed an overall pattern of increased cell invasion, suggesting that tumour ecosystem heterogeneity may aid cancer progression through its interplay with specific genomic alterations. Taken together, these results support the use of statistical modelling of spatial pathological data for a quantitative understanding of ecosystem heterogeneity and provide initial evidences of the clinical implication of tumour ecosystem diversity.

11:45 AM
12:30 PM
Jude Fitzgibbon - Sequential monitoring of Follicular lymphoma uncovers Rich and Sparse patterns of evolution

Follicular lymphoma is an incurable malignancy, with transformation to an aggressive subtype representing a critical event during disease progression. We performed whole-genome or whole-exome sequencing on 10 follicular lymphoma-transformed follicular lymphoma pairs followed by deep sequencing of 28 genes in an extension cohort, and we report the key events and evolutionary processes governing tumor initiation and transformation. Tumor evolution occurred through either a 'rich' or 'sparse' ancestral common progenitor clone (CPC). We identified recurrent mutations in linker histone, JAK-STAT signaling, NF-κB signaling and B cell developmental genes. Longitudinal analyses identified early driver mutations in chromatin regulator genes (CREBBP, EZH2 and KMT2D (MLL2)), whereas mutations in EBF1 and regulators of NF-κB signaling (MYD88 and TNFAIP3) were gained at transformation. Collectively, this study provides new insights into the genetic basis of follicular lymphoma and the clonal dynamics of transformation and suggests that personalizing therapies to target key genetic alterations in the CPC represents an attractive therapeutic strategy.

12:30 PM
01:45 PM

Lunch Break

01:45 PM
02:30 PM
Trevor Graham - Quantifying intra-tumour heterogeneity as a prognostic marker

Carcinogenesis is an evolutionary process; establishing the prognosis for a cancer therefore requires predicting the future course of cancer evolution. The same is true in pre-cancerous conditions: the risk of developing cancer is determined by how the pre-cancerous lesion is evolving.

The level of heterogeneity within a population measures the evolvability of the population: if there is no diversity natural selection cannot operate, whereas diverse populations are likely to contain well-adapted individuals that can prosper in changing environments. Consequently, quantification of within-tumour heterogeneity is likely to be a proxy-measure of the rate of the underlying evolutionary process that drives carcinogenesis, and so be an effective prognostic marker. In this talk, I will describe how we have measured within-tumour diversity, both genetically and phenotypically, to successfully determine prognosis in both established cancers and in premalignant lesions.

In addition, I will describe how we begun to search for the most prognostic measures of intra-tumour heterogeneity by constructing simple computational models of cancer development, and using the models to perform an exhaustive search of possible heterogeneity measures.

02:30 PM
04:15 PM

Breakout Discussion

04:15 PM
05:00 PM
Simon Hayward - In vivo and in vitro modeling of epithelial and stromal heterogeneity in tumors

Both stromal and epithelial cells in tumors exhibit marked heterogeneity that can affect communication within and between tissue layers resulting in alterations in cell proliferation, survival and death. Many of these interactions are cooperative rather than competitive. The various signaling axes represent a complex network of altered ligand/receptor availability that can be modified to alter growth and progression. A more complete understanding of cellular complexity in both stromal and epithelial tissues should allow us to model these intercellular communications with a view to identifying key nodes that can be coordinately modulated to restrict tumor growth.

05:00 PM

Shuttle pick-up from MBI

06:30 PM
07:00 PM

Cash Bar at Crowne Plaza

07:00 PM
08:30 PM

Banquet at Crowne Plaza

Friday, February 6, 2015
Time Session
08:00 AM

Shuttle to MBI

08:15 AM
09:00 AM

Breakfast

09:00 AM
09:45 AM
Gustavo Leone
09:45 AM
10:30 AM
Morag Park
10:30 AM
11:00 AM

Break

11:00 AM
11:45 AM
Shelley Hwang - Less is More: Building a cancer heterogeneity model to guide breast cancer treatment

Genetic diversity in a population is the fuel for natural selection and is a key determinant of the rate of evolution. The more genetic and microenvironmental diversity, the more opportunities for selection to drive clonal expansions and for the neoplasm to adapt to new selective pressures, including interventions. Measures of the somatic evolutionary process itself represent new forms of biomarkers. Because all neoplasms progress through a process of somatic evolution, measures of somatic evolution have the potential to predict progression across many cancers, not just breast cancer. In essence, they may be universal biomarkers. We are applying this approach to study the earliest breast cancers, ductal carcinoma in situ and propose applying this concept to more precisely guide the management of early stage breast cancer. The heuristic is based upon application of metapopulation and dispersal theories from ecology to predict cancer progression based on their effects on natural selection. This framework represents an innovative approach that is a significant departure from traditional cancer biology, and could yield a universally applicable construct for understanding interactions between tumors and their environments.

11:45 AM

Shuttle pick-up from MBI (One back to hotel and one to airport)

Name Email Affiliation
Abdel-rahman, Mohamed mohamed.abdel-rahman@osumc.edu Ophthalmology / Division of Human Genetics, The Ohio State University
Altrock, Philipp paltrock@jimmy.harvard.edu Program for Evolutionary Dynamics, Harvard University
Anderson, Alexander alexander.Anderson@moffitt.org Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center & Research Institute
Avendano, Alex avendano.8@osu.edu Mechanical and Aerospace Engineering, The Ohio State University
Baker, Ann-Marie a.m.c.baker@qmul.ac.uk Tumour Biology, Barts Cancer Institute, Queen Mary University of London
Balakrishnan, Subhasree subhasree.balakrishnan@osumc.edu Molecular Cellular and Developmental Biology, The Ohio State University
Balkwill, Fran f.balkwill@qmul.ac.uk Centre for Cancer and Inflammation, Queen Mary University of London
Barcellos-Hoff, Mary Helen MHBarcellos-Hoff@nyumc.org Radiation Oncology, New York University School of Medicine
Basanta, David david@cancerevo.org Integrated Mathematical Oncology, H. Lee Moffitt Cancer Center
Botesteanu, Dana dboteste@math.umd.edu Applied Mathematics & Scientific Computing, University of Maryland
Cebulla, Colleen colleen.cebulla@osumc.edu Ophthalmology and Visual Science, The Ohio State University
Clarke, Robert clarker@georgetown.edu Office of the Dean for Research, Georgetown University Medical Center
Costello, Joseph jcostello@cc.ucsf.edu Department of Neurological Surgery, University of California, San Francisco
Curtis, Christina ccurtis2@stanford.edu Medicine and Genetics, Stanford University
de Bruin, Elza e.debruin@ucl.ac.uk Research Dep of Oncology, University College London Cancer Institute
Dong, Shuai dong.209@osu.edu Pharmacy, The Ohio State University
Durrett, Rick rtd@math.duke.edu Department of Mathematics, Duke University
Ewald, Andrew andrew.ewald@jhmi.edu Cell Biology, Johns Hopkins University
Fitzgibbon, Jude j.fitzgibbon@qmul.ac.uk Center for Haemato-Oncology, Queen Mary, University of London
Gallaher, Jill jill.gallaher@moffitt.org Integrated Mathematical Oncology, Moffitt Cancer Center
Gatenby, Robert robert.gatenby@moffitt.org H. Lee Moffitt Cancer Center & Research Institute
Gerlee, Philip philipgerlee@gmail.com Mathematical Sciences, Chalmers University of Technology
Govinder, Kesh govinder@ukzn.ac.za Mathematics, Statistics and Computer Science, University of KwaZulu-Natal
Graham, Trevor t.graham@qmul.ac.uk Tumour Biology, Barts Cancer Institute, QMUL
Hayward, Simon simon.hayward@vanderbilt.edu Surgery, NorthShore University HealthSystem
Huang, Sui sui.huang@systemsbiology.org ., Institute for Systems Biology
Hwang, Shelley shelley.hwang@duke.edu Surgery, Duke University
Irshad, Shazia shazia.irshad@well.ox.ac.uk
Jacobsen, Karly jacobsen.50@mbi.osu.edu Mathematical Biosciences Institute, The Ohio State University
Jenkins, Robert robert.jenkins@cancer.org.uk Tumour Cell Biology Laboratory, Cancer Research UK London Research Institute
Kan, Rebecca Pei Qi kanrebecca@gmail.com Clinical Oncology, University of Hong Kong
Kapoor, Aastha aastha_phd@iitb.ac.in Biosciences and bioengineering, Indian Institute of Technology, Bombay
Kaveh, Kamran kkavehma@gmail.com School of Mathematics, University of Minnesota
Kianercy, Ardeshir akianer1@jhmi.edu Urology, Johns Hopkins Hospital
Klimov, Sergey sklimov3@gmail.com Math Bio, Georgia State University
Knippler, Christina knippler.1@osu.edu Comprehensive Cancer Center, The Ohio State University
Konstorum, Anna akonstor@uci.edu Mathematics, University of California, Irvine
Le, Jenny le.252@osu.edu Department of Mechanical and Aerospace Engineering, The Ohio State University
Leedham, Simon simon.leedham@cancer.org.uk Wellcome Trust Centre for Human Genetics, University of Oxford
Leone, Gustavo gustavo.leone@osumc.edu Molecular Virology, Immunology, and Medical Genetics, The Ohio State University
Mahdipour Shirayeh, Ali ali.mahdipour@gmail.com Applied Mathematics, University of Waterloo, University of Waterloo
Marras, Alexander marras.3@osu.edu Mechanical and Aerospace Engineering, The Ohio State University
Martinez, Pierre p.martinez@qmul.ac.uk Tumour Biology, Evolution and Cancer laboratory, Queen Mary University of London
Mollica, Molly mollica.15@osu.edu Mechanical and Aerospace Engineering, The Ohio State University
Nichol, Dan daniel.nichol@univ.ox.ac.uk Department of Computer Science, University of Oxford
Oberai, Assad oberaa@rpi.edu Scientific Computation Research Center, Rensselaer Polytechnic Institute
Ostrowski, Michael michael.ostrowski@osumc.edu Ohio State Comprehensive Cancer Center, Ohio State University
Owusu, Benjamin bnyowusu@uab.edu Biochemisty & Cancer Biology, University of Alabama at Birmingham & Southern Research Institute
Park, Morag morag.park@mcgill.ca Biochemistry and Oncology, McGill University
Perez-Castro, Antonio antonio.perez-castro@osumc.edu molecular virology, immunology and genetics, OSU
Pitarresi, Jason jasonpitarresi@gmail.com Molecular Virology, Immunology, and Molecular Genetics, Ohio State University Medical Center
Quaranta, Vito vito.quaranta@vanderbilt.edu Vanderbilt Ingram Cancer Biology Center, Vanderbilt University
Quezada, Sergio s.quezada@ucl.ac.uk Research department of haematology, University College London Cancer Institute
Rezaei Yousefi, Mohammadmahdi rezaeiyousefi.1@osu.edu Electrical and Computer Engineering, The Ohio State University
Robertson Tessi, Mark Mark.Robertsontessi@moffitt.org Integrated Mathematical Oncology, Moffitt Cancer Center
Sahai, Erik erik.sahai@cancer.org.uk Tumor Cell Biology Lab, London Research Institute
Sansom, Owen o.sansom@beatson.gla.ac.uk Colorectal Cancer and Wnt Signalling, Cancer Research UK Beatson Insititue
Shahriyari, Leili shahriyari.1@osu.edu Mathematical Biosciences Institute, Ohio State University
Shibata, Darryl dshibata@usc.edu Department of Pathology, University of Southern California
Sizemore, Gina Gina.Sizemore@osumc.edu OSU-Comprehensive Cancer Center, Ohio State University
Song, Jonathan song.1069@osu.edu Mechanical And Aerospace Engineering, The Ohio State University
Sottoriva, Andrea andrea.sottoriva@icr.ac.uk Centre for Evolution and Cancer, The Institute of Cancer Research
Subramaniam, Vish subramaniam.1@osu.edu Mechanical & Aerospace Engineering, The Ohio State University
Taslim, Cenny taslim.2@osu.edu Statistics, The Ohio State University
Thies, Katie katie.thies@osumc.edu MVIMG, The Ohio State University
Tlsty, Thea thea.tlsty@ucsf.edu Department of Pathology, University of California, San Francisco
Valenciaga, Anisley anisley.valenciaga@osumc.edu CCC, The Ohio State University
White, Richard whiter@mskcc.org Gastrointestinal Oncology Service, Memorial Sloan-Kettering Cancer Center
Worthley, Daniel daniel.worthley@adelaide.edu.au Medicine, University of Adelaide
Wu, Jinghai jinghai.wu@osumc.edu The Comprehensive Cancer Center, The Ohio State University
Yuan, Yinyin yinyin.yuan@icr.ac.uk Centre for Evolution and Cancer, The Institute of Cancer Research
Predicting Cancer Evolution: Heterogeneity and the Genotype-Phenotype Mapping

Heterogeneity in cancer is an observed fact, both genetically and phenotypically. Intercellular variation is seen at all scales and stages of development, and has significant implications for prognosis. At present, our understanding of this heterogeneity is mainly restricted to the genetic scale with little information regarding the relationship between genetic and phenotypic heterogeneity. Further, little is known about how cells alter their microenvironment or how these changes drive selection and feedback to further drive cancer evolution.

Strong selective pressures imposed by a milieu of microenvironmental factors combined with high profileration rates and high mutation rates inevitably lead to the rapid emergence of resistance to therapy. Hence, the failure of cancer therapies is often attributed to Darwinian evolution. To understand and predict cancer evolution we must understand not only the mutations which drive evolution but also the mechanisms through which these mutations manifest themselves in phenotypic change. Thus, our success in predicting cancer progression and designing effective therapy is contingent on understanding the junction at which genes and environment meet to produce phenotypes, the genotype-phenotype (GP) map.

Experimental studies have revealed the complexity inherent within the GP-map which is responsible for the difficulty in predicting evolution; many genotypes produce identical phenotypes and further many phenotypes can emerge from a single phenotype. Indeed, recent experimental evidence shows that this mapping produces phenotypic heterogeneity through a variety of genetic and non--genetic mechanisms. Heterogeneity can be driven through phenotypic plasticity, the phenomenon whereby isogenic cells in different environments display different phenotypes. Further, isogenic cells in identical environments can display phenotypic heterogeneity which is the manifestation of intra--cellular noise amplified through the complex machinery of the cell signalling pathways. In this talk I will present a collection of related mathematical models which explore genetic, environmental, phenotypic and morphological heterogeneity through the unifying lens of the GP-map, outline the key mechanisms which could be responsible for generating adaptive phenotypes -- mutation, plasticity and stochasticity -- and explore the implications of these mechanisms for developing novel and effective cancer therapies.

The evolving tumour microenvironment of high grade serous ovarian cancer, HGSC

I will be describing our studies on the tumour microenvironment of high grade serous ovarian cancer and inflammatory cytokine targets in this cancer.

Carcinogenesis in Context: Tumor Heterogeneity as a Function of Host Biology

Both clinical and experimental data show that the stroma is highly involved in early malignancy, supporting the idea of reciprocal evolution of the malignant cell and the tumor microenvironment. Although it is clear that stroma composition and signaling is altered in cancer, less is known about how and when stroma contributes to carcinogenesis and how carcinogens, like radiation, might alter these processes. To investigate how tumor diversity evolves in the context of different physiological states (e.g. as a function of age and ionizing radiation exposure) we developed a mammary chimera model in which histologically and genomically diverse cancer originate from a p53 null epithelium orthotopically transplanted to a syngeneic wildtype host. We use expression profiling of tissue and tumors to identify the biological ‘bookends’, those processes initiated in normal tissue that appear to contribute to the development and are recapitulated in particular types of tumors. These studies show that the spectrum of mammary tumor types is strongly influenced by the host biology, opening new perspective on the drivers of aggressive cancer. The mounting evidence from these and other studies that cancer results from a systemic failure in which cells other than those with oncogenic alterations determine the frequency and type of clinical cancer is changing the cancer paradigm. To account for this, we propose that the tumor microenvironment is built through rate-limiting steps during multi-stage carcinogenesis [Barcellos-Hoff, 2013 #18417]. In this model, construction of a ‘pre-cancer niche’ is a necessary early step required for initiated cells to survive and evolve; subsequent niche expansion and maturation accompany promotion and progression respectively. The model postulates that cancer cell survival and proliferation is as much a function of the successful niche construction as it is of the natural selection for specific cancer cell mutations. Consequently, cancer represents an emergent property that requires a comprehensive analysis of the cell-cell interactions during the course of carcinogenesis. Moreover, in contrast to initiation, which is stochastic by nature, niche construction represents a robust target for native immunosuppression and a potent target for cancer prevention.

Under Pressure: Tumor evolution after therapy

Tumor recurrence is a leading cause of cancer mortality. Therapies for recurrent disease may fail, at least in part, because the genomic alterations driving the growth of recurrences are distinct from those in the initial tumor. At diagnosis, low grade brain tumors have similar histology and driver mutations, but following surgical resection their clinical behavior is highly variable. Infiltrating tumor cells comprising the residual disease may remain indolent for more than a decade after the surgical resection, or may rapidly transform into an aggressive rapidly growing malignant tumor. Adjuvant chemotherapies such as temozolomide (TMZ) are frequently used, especially in cases with subtotal surgical resections or when deferring the use of radiation therapy is preferable. We used genome and epigenome sequencing technologies to infer patterns of intratumoral heterogeneity and tumor evolution over time. In this presentation I will discuss new insight into diverging genetic pathways arising from initially similar tumors, and the profound influence of chemotherapy in shaping tumor aggressiveness.

A Big Bang model of human colorectal tumor growth

What happens in the early and still undetectable human malignancy is unknown because direct observations are impractical. Here I will describe a novel “Big Bang” model, whereby a tumor grows predominantly as a single expansion producing numerous intermixed sub-clones, which are not subject to stringent clonal selection. In this model, both public and most detectable private mutations arise during the earliest phase of tumor growth. Multi-scale genomic profiling of 349 individual glands sampled from 15 colorectal tumors revealed the absence of selective sweeps, uniformly high intra-tumor heterogeneity, and sub-clone mixing in distant tumor regions, as postulated by the Big Bang. By integrating the data in a spatial model of tumor growth and statistical inference framework we also verified the most striking prediction of our model, namely that most detectable intra-tumor heterogeneity originates from private alterations acquired early during growth, and not from the later expansion of selected sub-clones. Hence, early sub-clones define the genomic profile of colorectal carcinomas and advanced adenomas, whereas potentially dangerous late-arising sub-clones will go undetected. Moreover, our results suggest that sub-clone mixing may be a biomarker of malignant potential. This new model provides a quantitative framework that explains the origins of intra-tumor heterogeneity and tumor growth dynamics with significant clinical implications for treatment resistance and metastatic progression, as I will discuss.

Genetic intra-tumour heterogeneity sheds light on lung cancer evolution

Lung cancer is the most common cancer worldwide, and the 5-years overall survival remains poor. Understanding the genetic evolution of the most common type of lung cancer, non-small lung cancer (NSCLC), and resolving which mutations occur early in a tumour’s evolution (clonal) and which mutations are involved at later stages (subclonal) may improve therapeutic interventions. We performed multi-region exome and/or genome sequencing at a mean depthof >200x to analyse the intra-tumour heterogeneity in a total of 44 tumour regions from fourteen primary NSCLC samples. Non-silent mutations were validated using ultra-seek amplicon sequencing. All samples revealed potential driver mutations occurring on the trunk (clonal mutations) and on the branches of the phylogenetic trees (subclonal mutations), although the majority of driver mutations and large-scale genomic events such as genome doubling occurred predominantly early in the life history of these tumours. On-going chromosomal instability enhanced the intra-tumour heterogeneity at later stages of tumour development, and also potential driver mutations could be identified. Detailed analyses of the mutational spectra identified several mutational processes involved in NSCLC tumour evolution, with some processes predominantly involved in early mutations, and others driving later mutations. The relative contribution of these later processes show regional variation, thereby contributing to the intra-tumour heterogeneity in NSCLC.

Sequential monitoring of Follicular lymphoma uncovers Rich and Sparse patterns of evolution

Follicular lymphoma is an incurable malignancy, with transformation to an aggressive subtype representing a critical event during disease progression. We performed whole-genome or whole-exome sequencing on 10 follicular lymphoma-transformed follicular lymphoma pairs followed by deep sequencing of 28 genes in an extension cohort, and we report the key events and evolutionary processes governing tumor initiation and transformation. Tumor evolution occurred through either a 'rich' or 'sparse' ancestral common progenitor clone (CPC). We identified recurrent mutations in linker histone, JAK-STAT signaling, NF-κB signaling and B cell developmental genes. Longitudinal analyses identified early driver mutations in chromatin regulator genes (CREBBP, EZH2 and KMT2D (MLL2)), whereas mutations in EBF1 and regulators of NF-κB signaling (MYD88 and TNFAIP3) were gained at transformation. Collectively, this study provides new insights into the genetic basis of follicular lymphoma and the clonal dynamics of transformation and suggests that personalizing therapies to target key genetic alterations in the CPC represents an attractive therapeutic strategy.

Evolutionary triage governs fitness in driver and passenger mutations and suggests targeting never mutations

Genetic and epigenetic changes in cancer cells are typically divided into "drivers" and "passengers" Drug development strategies target driver mutations, but inter- and intra-tumoral heterogeneity usually results in emergence of resistance. Here we model intratumoral evolution in the context of a fecundity/survivorship trade-off. Simulations demonstrate the fitness value, of any genetic change is not fixed but dependent on evolutionary triage governed by initial cell properties, current selection forces, and prior genotypic/phenotypic trajectories. We demonstrate spatial variations in molecular properties of tumor cells are the result of changes in environmental selection forces such as blood flow. Simulated therapies targeting fitness-increasing (driver) mutations usually decrease the tumor burden but almost inevitably fail due to population heterogeneity. An alternative strategy targets gene mutations that are never observed. Because up or down regulation of these genes unconditionally reduces cellular fitness, they are eliminated by evolutionary triage but can be exploited for targeted therapy.

The evolution of carrying capacity in constrained and expanding tumour cell populations

Cancer cells are known to modify their micro-environment such that it can sustain a larger population, or, in ecological terms, they construct a niche which increases the carrying capacity of the population. It hashowever been argued that niche construction, which benefits all cells in the tumour, would be selected against since cheaters could reap the benefits without paying the cost. We have investigated the impact of niche specificity on tumour evolution using an individual based model of breast tumour growth, in which the carrying capacity of each cell consists of two components: an intrinsic, subclone-specific part and a contribution from all neighbouring cells. Analysis of the model shows that the ability of a mutant to invade a resident population depends strongly on the specificity. When specificity is low selection is mostly on growth rate, while high specificity shifts selection towards increased carrying capacity. Further, we show that the long-term evolution of the system can be predicted using adaptive dynamics. By comparing the results from a spatially structured vs. well-mixed population we show that spatial structure restores selection for carrying capacity even at zero specificity, which poses a possible solution to the niche construction dilemma. Lastly, we show that an expanding population exhibits spatially variable selection pressure, where cells at the leading edge exhibit higher growth rate and lower carrying capacity than those at the centre of the tumour.

Quantifying intra-tumour heterogeneity as a prognostic marker

Carcinogenesis is an evolutionary process; establishing the prognosis for a cancer therefore requires predicting the future course of cancer evolution. The same is true in pre-cancerous conditions: the risk of developing cancer is determined by how the pre-cancerous lesion is evolving.

The level of heterogeneity within a population measures the evolvability of the population: if there is no diversity natural selection cannot operate, whereas diverse populations are likely to contain well-adapted individuals that can prosper in changing environments. Consequently, quantification of within-tumour heterogeneity is likely to be a proxy-measure of the rate of the underlying evolutionary process that drives carcinogenesis, and so be an effective prognostic marker. In this talk, I will describe how we have measured within-tumour diversity, both genetically and phenotypically, to successfully determine prognosis in both established cancers and in premalignant lesions.

In addition, I will describe how we begun to search for the most prognostic measures of intra-tumour heterogeneity by constructing simple computational models of cancer development, and using the models to perform an exhaustive search of possible heterogeneity measures.

In vivo and in vitro modeling of epithelial and stromal heterogeneity in tumors

Both stromal and epithelial cells in tumors exhibit marked heterogeneity that can affect communication within and between tissue layers resulting in alterations in cell proliferation, survival and death. Many of these interactions are cooperative rather than competitive. The various signaling axes represent a complex network of altered ligand/receptor availability that can be modified to alter growth and progression. A more complete understanding of cellular complexity in both stromal and epithelial tissues should allow us to model these intercellular communications with a view to identifying key nodes that can be coordinately modulated to restrict tumor growth.

Tumor Progression: Non-Genetic Cancer Cell Population Heterogeneity and Plasticity Drive Non-Darwinian Somatic Evolution

Tumor progression and generation of therapy resistant cells following therapy are not simply a manifestation of Darwin’s “Survival Of The Fittest” but rather, of Nietzsche’s “What Does Not Kill Me Makes Me Stronger”. The robust arrow of progression towards increasingly sophisticated aggressiveness of cancer cells during the course of treatment, including therapy resistance, is typically explained by a somatic Darwinian evolution: selection of cancer cell clones that carry random genomic mutations that (by chance) confer the new more malignant phenotype. However, the immense non-genetic heterogeneity and the plasticity of cell phenotype offer a new level for evolutionary mechanism that transcends (but does not exclude) the Neo-Darwinian scheme. By relaxing the rigid genotype-phenotype relationship that is tacitly assumed to explain the Darwinian expansion of the fittest mutant clone, cell phenotype plasticity accelerates evolution. The plasticity of cell phenotype is not random: it is constrained and channeled by the principles of stochastic, nonlinear dynamics through which the gene regulatory network (GRN) coordinates gene expression to produce complex, coherent phenotypic states of the cell. In this framework we postulate that cancer cells are “stuck” in “cancer attractors”: latent, normally unused stable states in the vast theoretical space of all possible gene expression patterns of the genome. Cancer attractors are the inevitableby-products of the trial/error scheme of rewiring the GRN during evolution. They can be imagined as the normally unoccupied side-valleys in Waddington’s epigenetic landscape that must be avoided during the developmental descent of immature (stem) cells to the mature states of differentiation, the lowest point in the main valleys. Near-lethal therapeutic intervention on cells stuck in such abnormal attractor states opens –via symmetry-breaking bifurcations that epitomize the desired destabilization of the cancerous cell states– access to yet more aberrant attractors in the unused regions of the landscape further away from the normal developmental trajectories. They encode even less mature, “degenerate” states, and following such bifurcations are inevitably occupied by some cancerous cells for simple entropy reasons.

In this talk I will explain the basic principles of non-Darwinian dynamics on the quasi-potential (=epigenetic) landscape of the GRN and why therapy, or any attempt to destabilize the cancerous state, will, in addition to intended apoptosis also cause “rebellious cells” to spill over into attractors of increased stemness and hence, promote malignancy and produce therapy resistance. Experimental data that document this phenomenon will be presented as well as results of first attempts to curb such evasive behavior.

Less is More: Building a cancer heterogeneity model to guide breast cancer treatment

Genetic diversity in a population is the fuel for natural selection and is a key determinant of the rate of evolution. The more genetic and microenvironmental diversity, the more opportunities for selection to drive clonal expansions and for the neoplasm to adapt to new selective pressures, including interventions. Measures of the somatic evolutionary process itself represent new forms of biomarkers. Because all neoplasms progress through a process of somatic evolution, measures of somatic evolution have the potential to predict progression across many cancers, not just breast cancer. In essence, they may be universal biomarkers. We are applying this approach to study the earliest breast cancers, ductal carcinoma in situ and propose applying this concept to more precisely guide the management of early stage breast cancer. The heuristic is based upon application of metapopulation and dispersal theories from ecology to predict cancer progression based on their effects on natural selection. This framework represents an innovative approach that is a significant departure from traditional cancer biology, and could yield a universally applicable construct for understanding interactions between tumors and their environments.

Morphogenic and microenvironmental control of intestinal stem cell fate

The rapid turnover of the intestinal epithelium is supported by tissue specific stem cells thought to be located at the base of the intestinal crypt - the basic functional unit of the gut. Stem cell progeny progressively proliferate and differentiate as they move out of the crypt base stem cell niche and along the crypt-villus axis of the intestine. This system is carefully orchestrated by interacting endogenous epithelial and paracrine secreted morphogen gradients, with cell fate coupled to, and determined by, a cells position within these gradients. It has been recently suggested that stem cell fate is similarly influenced by microenvironmental factors, and that stem cell function is not cell autonomous. Recent work has shown that cells outside of the stem cell niche can reacquire stem cell properties to assist in crypt regeneration and that disruption of mucosal morphogenic gradients can cause tumour initiation from cells that have exited the crypt base stem cell niche.

This talk will explore the exciting developments in the homeostatic and pathological microenvironmental control of intestinal stem cells and discuss the implications of this for colorectal cancer heterogeneity and chemotherapeutic strategy.

Mechanical heterogeneity of invasive breast carcinomas

Heterogeneity is a hallmark of cancer whether one considers the genotype of cancerous cells, the composition of their microenvironment, the distribution of blood and lymphatic microvasculature, or the spatial distribution of the desmoplastic reaction. It is logical to expect that this heterogeneity in tumor microenvironment will lead to spatial heterogeneity in its mechanical properties. In this study we seek to quantify the mechanical heterogeneity within malignant and benign tumors using ultrasound based elasticity imaging. By creating in-vivo elastic modulus images for ten human subjects with breast tumors, we show that Young’s modulus distribution in cancerous breast tumors is more heterogeneous when compared with tumors that are not malignant, and that this signature may be used to distinguish malignant breast tumors. Our results complement the view of cancer as a heterogeneous disease on multiple length scales by demonstrating that mechanical properties within cancerous tumors are also spatially heterogeneous.

Anticancer Activity of SRI31215, a Novel Inhibitor of Hepatocyte Growth Factor (HGF) Activation

Aberrant Hepatocyte growth factor (HGF)/MET signaling has been implicated in the development and progression of several types of human cancers and in the development of resistance to therapy. Both HGF and its receptor tyrosine kinase, MET, have therefore emerged as targets for cancer therapy. Several MET kinase inhibitors are available, but their clinical efficacy is hindered by emergence of acquired resistance. Thus, there is the need to adopt new therapeutic strategies to interfere with oncogenic HGF/MET signaling.

The rate-limiting step that regulates the biological activity of HGF is the processing of pro-HGF to the mature/active form by one or more of the serine proteases: hepsin, matriptase and hepatocyte growth factor activator (HGFA). These enzymes are tightly regulated by the endogenous inhibitor, HGF activator inhibitor 1 (HAI-1), a protein with the tumor suppressor activity.

We developed a novel small molecule, SRI31215, which inhibits the activity of matriptase, hepsin and HGFA and thereby mimics the activity of HAI1. We demonstrated that SRI31215 inhibits HGF activation and its biological activity in vitro. SRI31215 prevents HGF-mediated scattering and migration of cancer cells and inhibits epithelial-mesenchymal transition (EMT), a crucial step in the metastatic spread of cancer cells.

HGF/MET signaling drives primary and acquired resistance to anticancer kinase inhibitors. We have shown that SRI31215 inhibits autocrine HGF/MET signaling and sensitizes colon cancer cells to the epidermal growth factor inhibitor, Cetuximab (CET). SRI31215 also prevents fibroblast-mediated resistance to CET by inhibiting HGF-dependent crosstalk between tumor cells and tumor associated fibroblasts.

Taken together, our data demonstrate that inhibitors of HGF activation such as SRI31215 offer a novel approach to interfere with oncogenic HGF/MET signaling and HGF-dependent tumor progression and to prevent primary and acquired resistance of colon cancer cells to targeted therapy.

Data-driven modeling of the heterogeneity of targeted therapy response in oncogene addicted cancers links single-cell level to cell-population outcomes

With a high-throughput colony Fractional Proliferation (cFP) assay, we simultaneously track in real-time the proliferation dynamics of hundreds to thousands of single-cell derived clones in a cell population exposed to perturbations (Frick et al, 2015, DOI: 10.1002/jcp.24888). In the mutant EGFR-addicted PC9 lung cancer cell line treated with erlotinib, cell fates (death, quiescence, continued proliferation) within each clone vary from cell-to-cell, even between siblings. This widespread heterogeneity of drug response is captured by a new metric, the drug-induced proliferation (DIP) rate, which encapsulates single-cell variation into a dynamic measure of drug response outcomes.

DIP rates variation from colony to colony in PC9 is approximately normally distributed, a strong indication it arises from stochastic sources. Measurement error or mixed colony ancestry could not account for this variation, since DIP rates of PC9 sublines isolated from single cells and propagated in long-term culture (PC9-DS1/95) exhibited the same normal distribution and maintained it for over 25 generations. Similar distributions were obtained from many additional oncogene-addicted cell lines, rigorously re-derived from single cells. Thus, a mutated driver oncogene does not ensure cell-to-cell homogeneity of response, even when genetic background diversity is minimized.

To explore whether these distributions are of consequence to treatment, we constructed a Polyclonal Growth (PG) mathematical model able to incorporate theoretical or experimental DIP rates as parameters. Since DIP rate distributions are normal, they are entirely defined by two parameters, mean and variance. Inputting the average DIP rate of parental PC9 predicts that the cell line as a whole will completely succumb to treatment. In contrast, with the DIP rate distribution parameters as input, a completely different result was obtained: the size of the erlotinib treated population rebounded to initial values after ~11 days, after an initial drop to half the value at 5 days. The PG model predicted similar dynamics of erlotinib response for several mutant EGFR-addicted cell lines: in every cell line tested, rebound occurred within days to weeks, after initial drops to varying depths. Time to rebound is affected primarily by the extent to which the right tail of the DIP rate distribution extends into positive territory. Using stochastic simulations of the PG model, we are able to differentiate the effects of clonal heterogeneity from those of stochastic cell fate decisions (intrinsic noise) that cause significant variability in the response trajectories, including response depth and duration. Predictions were validated experimentally in PC9. It is unlikely that conventional acquired resistance was responsible for the rebound, since SNaPshot multigene assays were negative and response dynamics were inconsistent with a model of rare drug resistant clones. These findings suggest that, even in the absence of acquired genetic resistance, heterogeneity of drug response promotes rebound of the treated population. We propose that these experimental and modeling tools (cFP assay and PG model) enable realistic evaluation of depth and duration of response to targeted drug treatment. Expected and unexpected PG model predictions and suggested avenues for treatment, especially drug combinations, will be discussed.

Targeting immune-checkpoints in cancer: New mechanistic insights

The continual interplay between the immune system and cancerous cells is thought to result in the establishment of a dynamic state of equilibrium. This equilibrium depends on the balance between subsets of effector and regulatory lymphocytes. Whereas the overall mechanisms underpinning the establishment and maintenance of the intra-tumour balance between Teff and Treg cells remain unknown, in many solid cancers it is characterized by the dominant infiltration of regulatory T cells over effector T cells resulting in a low Teff/Treg ratio. Furthermore, different subtypes of regulatory cells and inhibitory molecules such as CTLA-4 tightly control the few effector T lymphocytes that manage to infiltrate the tumour. The outcome of this balance is critical to survival, and while in a few cases the equilibrium resolves in the elimination of the tumour by the immune system, in many other cases the tumour manages to escape immune control.

Remarkably, antibodies against CTLA-4, a key immune modulatory receptor expressed on T cells, efficiently modify this balance, driving effector T cell expansion and increasing the ratio of Teff/Treg within the tumour. Whilst the high Teff/Treg ratio driven by anti-CTLA-4 directly correlates with tumour destruction in mice and humans, the mechanisms underpinning this phenomenon remain unknown.

By focusing in the study of effector and regulatory tumour-reactive CD4+ T cells my group is interested in the mechanism underpinning the activity of different immune-modulatory antibodies within the tumour microenvironment, and the potential positive and negative impact that the tumour microenvironment may have in the recruitment, survival and function of different T cell subsets. In this context and using a murine model of melanoma we have recently demonstrated that both, the change in the Teff/Treg balance as well as tumour rejection, depend on the selective depletion of tumour-infiltrating Treg cells expressing high levels of surface CTLA-4. Regulatory T cell depletion is mediated by ADCC and completely depends on the expression of FcRIV on tumour infiltrating CD11b+ myeloid cells. These results reveal novel and unexpected mechanistic insight into the activity of anti-CTLA-4-based cancer immunotherapy, and illustrate the importance of specific features of the tumour microenvironment on the final outcome of antibody-based immune-modulatory therapies.

Intravital imaging reveals how BRaf inhibition generates drug tolerant microenvironments

Many tumours show an initial response to targeted therapies before genetic resistance emerges, however little is known about how tumour cells tolerate therapy before genetic resistance dominates. In this study, we have used both intravial ratiometic FRET and FLIM of an ERK/MAP kinase biosensor to investigate heterogeneity in signalling in melanoma models. Moreover, we have longitudinally monitored responses to targeted therapy and identified areas that become refractory to drug action. BRaf mutant melanoma cells can rapidly become tolerant to PLX4720 in areas of high stroma. The rapid kinetics of this process indicate that it is not caused by genetic events. We demonstrate that PLX4720 has an unexpected effect on the tumour stroma leading to enhanced matrix remodelling. The remodelled matrix then provides signals that enable melanoma cells to tolerate PLX4720. We propose that this safe haven enhances the population of cancer cells from which genetically resistance emerges. This work highlights the utility of intravital imaging in understanding the reason for therapy failure.

The influence on tumour driver mutations on the microenvironment

It is has become clear that the microenvironment plays a very important role in tumour progression with recent studies showing both tumour suppressive and promoting roles. Our previous work has identified the importance of neutrophils in the both inflammation and spontaneous models of carcinogenesis.

I will discuss our latest findings examining how changes in the microenvironment affect epithelial cancer development. I will show latest data on the interaction between epithelial cancer cells and surrounding stromal cells and how manipulation of key signalling pathways can either accelerate or suppress tumor development.

Neutral evolution and star-like phylogenies in next-generation sequencing data

Despite extraordinary efforts to profile cancer genomes on a large scale, interpreting the vast amount of genomic data in light of cancer evolution and in a clinically relevant manner remains challenging. This is complicated by interpatient variation and extensive intra-tumor heterogeneity. In particular, the relationship between the cancer genotype and phenotype remains largely unknown as phenotypical characteristics are hard to measure. A critical issue is the lack of a theoretical framework of reference able to make predictions on existing data. Using a multi-sampling strategy we have recently shown how colorectal cancers grow as a single “Big Bang” expansion populated by many intermixed sub-clones that are not subject to stringent selection. Here we demonstrate that this signature of effectively-neutral evolution can be detected in next-generation sequencing data from bulk samples as well, as a result of the star-like tumor phylogenies predicted by the Big Bang growth. In particular, we present a simple mathematical model of cancer expansion based on neutral evolutionary dynamics that predicts the spectrum of alterations reported by nextgeneration sequencing in a large proportion of cases. This hidden feature of cancer evolution is common to multiple tumor types and can be detected in different independent cohorts. Importantly, this allows the direct measurement in each individual patient of the in vivo mutation rate per division and the timing of mutations using currently available sequencing data. This result provides a new way to interpret the wealth of cancer genomic data available to date, shedding new light on the relationship between the cancer genotype and phenotype.

Environmental Control of Cellular States Associated with Plasticity and Cell Fate Decisions

The product of the p16INK4a/CDKN2A locus encodes a cyclin-dependent kinase (CDK) inhibitor that functions as a negative regulator of cyclin/CDK complexes. In addition to governing regulation through the cell cycle, this protein also regulates the production of chromatin remodeling proteins downstream of E2F. We identified cell surface markers associated with repression of p16INK4a/CDKN2A and found that they allowed direct isolation of rare cells from healthy human breast tissue that exhibit extensive lineage plasticity. This subpopulation of cells has the ability to enter a state that can transcribe pluripotency markers, Oct3/4, Sox2 and Nanog at levels similar to those measured in human embryonic stem cells and to acquire an epigenetically plastic state sensitive to environmental programming. In vitro, in vivo and teratoma assays demonstrated that either a directly-sorted (uncultured) or a single cell-derived (clonogenic) cell population from primary human tissue can differentiate into functional derivatives of each germ layer, ectodermal, endodermal and mesodermal. In contrast to other cells that express Oct3/4, Sox2 and Nanog, these human endogenous Plastic Somatic cells (ePS cells) are mortal, express low telomerase activity, expand for an extensive but finite number of population doublings, and maintain a diploid karyotype before arresting in G1. These types of cellular states, exhibiting elevated plasticity, can be directed towards multiple cell fates dictated by the microenvironment.

Assessing normal and cancer somatic variation with RAD-seq

Both normal and cancerous tissues are able to acquire genetic variation over time. The spectrum of cancer mutations has been well characterized through projects such as the TCGA, yet the degree of normal somatic variation within an individual, called somatic mosaicism, has been poorly characterized. To understand cancer phylogenies will require a deeper appreciation and understanding of normal somatic variation, but techniques such as whole-genome and exome sequencing are not efficient for the construction of such trees. To overcome this, we are using a reduced representation sequencing technique called RAD-seq to assess genetic variation across normal and cancer tissues. In this approach, DNA samples are digested with a restriction enzyme to give a desired amount of genome coverage, and then only those portions of the genome containing those cut sites undergo Illumina sequencing. We have applied the RAD-seq pipeline to a series of normal and cancer tissues from both humans (pancreatic cancer patients) as well as zebrafish (melanoma bearing animals), sequencing both the tumor itself, individual metastases, and several normal tissues such as lung and brain. We are using this data to understand the patterns of SNVs and copy number changes in each of these tissues, and across species. The application of RAD-seq will allow for high-throughput, deep coverage of genomic variation at a fraction of the cost of traditional sequencing approaches, and will be broadly applicable to experiments focusing on cancer evolution.

Connective tissue stem cells in the tumor microenvironment

The tumor microenvironment presents an exciting opportunity for innovative therapeutic approaches to cancer. The origin and exact biological contribution of the peritumoral connective tissue at the primary and metastatic sites, however, is still uncertain. Just as one would not assess the average of “hematopoietic” contribution to the tumor microenvironment without considering distinct cell types, so too lumping together all “peritumoral mesenchyme” or “cancer-associated fibroblasts” on the basis of broad markers is likely to oversimplify a diverse connective-tissue contribution to cancer. In this presentation I will discuss the types of connective tissue cells in cancer. I will outline recent studies addressing the biological function of these cells in cancer and discuss our own research on the heterogeneity of connective tissue stem cells in the intestine, the bone and in several cancer models. Understanding the biological heterogeneity of mesenchymal cells in cancer will provide new opportunities for targeted cancer prevention and therapy.

The Ecosystem Diversity Landscape of Breast Cancer

It is increasingly recognised that the tumor microenvironment is an important determinant in cancer progression and evolution, where tumors act as complex ecosytems involving interactions between cancer cells, stromal cells and their physical environment. To study this we developed a statistical model to systematically quantify the spatial heterogeneity of the tumor ecosystem based on automated image analysis of 1,026 Hematoxylin & Eosin (H&E) stained primary breast tumors. I will discuss the clinical implication of heterogeneous tumour ecosystem, how ecosystem diversity was capable of predicting prognosis independent to known clinical and cancer heterogeneity parameters, before moving on to the bioinformatics integration of this measurement with whole-genome genomic profiling data for all of these tumours. This helped us reveal enrichment of specific copy number alterations for certain genes in this subtype, for which an RNAi screen showed an overall pattern of increased cell invasion, suggesting that tumour ecosystem heterogeneity may aid cancer progression through its interplay with specific genomic alterations. Taken together, these results support the use of statistical modelling of spatial pathological data for a quantitative understanding of ecosystem heterogeneity and provide initial evidences of the clinical implication of tumour ecosystem diversity.