Publications
Published work.
67 peer-reviewed articles, reviews and book chapters, newest first, each with a short summary. Each title links to the published paper. Where we posted a preprint, it is linked alongside, and so is any blog post about the work. Names in bold are members of the group, past and present (see Team); tags link to the matching theme on the Research page.
2026
-
Multifactorial sheltering in peristromal niches shapes in vivo responses of lung cancers to targeted therapies
Nature Communications 17(1):7003
In models of ALK-positive lung cancer, tumor cells next to stroma survive targeted therapy through many juxtacrine and paracrine signals acting together, an ecological rescue that sustains residual disease. Rather than blocking those signals one at a time, the residual cells' adaptive HER2 upregulation can be exploited with the antibody-drug conjugate T-DXd, which strongly improved responses and suppressed relapse.
-
Overcoming vascular niche–mediated TKI resistance in acute myeloid leukemia through miR-126 inhibition
npj Systems Biology and Applications 12(1):38
An agent-based model of the leukemic bone marrow, parameterized with in vitro and in vivo data, captures how endothelial miR-126 lets leukemic stem cells return to quiescence and escape tyrosine kinase inhibitors. Simulations show that adding the miR-126 inhibitor miRisten on a defined schedule dismantles this protective vascular niche.
-
The future of mathematical oncology in the age of AI
npj Systems Biology and Applications 12(1):22
A perspective on how data-driven AI and mechanistic modeling can work together: the opportunities, the limitations of current AI approaches, and a mathematical oncology in which AI augments mechanistic rigor and clinical relevance.
2025
-
The adaptive state determines the impact of mutations on evolving populations
Proceedings of the National Academy of Sciences 122(26):e2427070122
In a spatial agent-based model of a tumor population evolving resistance to targeted therapy, favorable mutations were over-represented at the extinction bottleneck but depleted at the adaptive peak. Experiments confirmed the prediction: raising the mutation rate helped cancer cells under therapy but was moderately harmful at baseline, suggesting that restricting mutational variability could be therapeutic.
2024
-
Dissecting the Spatially Restricted Effects of Microenvironment-Mediated Resistance on Targeted Therapy Responses
Cancers 16(13):2405
Combines spatial measurements of proliferation and death in an animal model of targeted therapy with spatial mathematical models to dissect how the amount, reach and layout of stroma affect tumor cells' ability to escape elimination. The spatial distribution of stroma within the tumor had a particularly strong effect.
-
A seven‐step guide to spatial, agent‐based modelling of tumour evolution
Evolutionary Applications 17(5):e13687
A non-technical, step-by-step guide to building spatial agent-based models of tumor evolution, from the Eden growth model to off-lattice simulations with diffusible factors, covering implementation, parameterization, visualization and reproducibility, and arguing for models that match the complexity of the question rather than of the whole system.
-
The bone ecosystem facilitates multiple myeloma relapse and the evolution of heterogeneous drug resistant disease
Nature Communications 15(1):2458
A hybrid agent-based model of multiple myeloma in bone shows that resistance arises from myeloma-intrinsic mechanisms, but that environment-mediated drug resistance in the bone makes resistant clones more likely and more heterogeneous. The model predicts, and in vivo studies support, that targeting this protection deepens responses by eliminating sensitive cells sheltered near stroma and bone.
2023
-
Stroma-Mediated Breast Cancer Cell Proliferation Indirectly Drives Chemoresistance by Accelerating Tumor Recovery between Chemotherapy Cycles
Cancer Research 83(22):3681–3692
In triple-negative breast cancer, stroma did not directly reduce chemosensitivity in vitro, but stromal factors increased tumor-cell proliferation, and proximity to stroma was associated with faster proliferation in vivo. A spatial agent-based model parameterized from histology showed that this lets residual tumors recover between chemotherapy cycles: an indirect form of stroma-mediated chemoresistance.
-
A survey of open questions in adaptive therapy: Bridging mathematics and clinical translation
eLife 12:e84263
Eleven open questions in the mathematical modeling of adaptive therapy, which keeps tumor burden high so that treatment-sensitive cells suppress resistant ones. The questions cover what models should include, how dosing protocols are designed and validated, and the challenges of clinical translation.
2022
-
Integrated computational and in vivo models reveal Key Insights into macrophage behavior during bone healing
PLOS Computational Biology 18(5):e1009839
Coupled ODE models fitted to in vivo bone-injury data test hypotheses about monocyte and macrophage dynamics during bone healing. They indicate that anti-inflammatory macrophages are key to early inhibition of osteoclasts and that pro-inflammatory cells rise in two waves separated by an inhibitory phase; the model was checked against an independent dataset.
-
Roadmap on plasticity and epigenetics in cancer
Physical Biology 19(3):031501
A multi-author roadmap of mathematical and experimental approaches to non-genetic plasticity in cancer: how it relates to mutation-driven evolution, how the microenvironment controls it, how to measure and model it, and what it means for the clinic.
2021
-
Coexistence of “Cream Skimmer” and “Crumb Picker” Phenotypes in Nature and in Cancer
Frontiers in Ecology and Evolution 9:697618
Fluctuating resources can let two phenotypes coexist on one resource. Models of 'cream skimmers', which do better in rich times, and 'crumb pickers', which do better in poor times, together with the argument that nutrients, oxygen and pH fluctuate in tumors, suggest this coexistence mechanism is likely in cancer; candidate biomarkers for the two phenotypes are proposed and explored.
-
The Evolutionary Ecology of Dormancy in Nature and in Cancer
Frontiers in Ecology and Evolution 9:676802
Compares dormancy in natural ecosystems and in cancer, where the term is used inconsistently, and argues that clearer shared terminology would let ecologists and cancer biologists learn from each other.
-
Cancer immunoediting: A game theoretical approach
In Silico Biology 14(1-2):1–12
An evolutionary game of proliferating, quiescent and immune cells that captures the elimination, equilibrium and escape phases of cancer immunoediting, with an analysis of the stability of its equilibria.
-
Computational modeling reveals a key role for polarized myeloid cells in controlling osteoclast activity during bone injury repair
Scientific Reports 11(1):6055
An ODE model fitted to data from injured mouse tibiae shows that bone cannot return to homeostasis after injury unless pro- and anti-inflammatory myeloid cells are included, and that osteoclast activity must vary more than 17-fold, so osteoblast and osteoclast numbers alone do not predict how bone repairs.
-
Quantification and Optimization of Standard-of-Care Therapy to Delay the Emergence of Resistant Bone Metastatic Prostate Cancer
Cancers 13(4):677
A model of bone-metastatic prostate cancer built from patient biopsies taken before and after abiraterone tracks resistant phenotypes from first principles. Linked to PSA, it reproduced individual responses in an independent cohort of 24 patients and, combined with a decision algorithm, proposes patient-specific treatment sequences to delay resistance.
-
Spontaneous cell fusions as a mechanism of parasexual recombination in tumour cell populations
Nature Ecology & Evolution 5(3):379–391
Breast cancer cells spontaneously fuse at low frequency in vitro and in vivo, and many hybrids partially reduce their ploidy, producing progeny that inherit mixtures of parental alleles. Modeling shows that these fusion rates can amplify clonal heterogeneity and speed adaptation to new selective pressures.
-
Selection-driven tumor evolution with public goods leads to patterns of clonal expansion consistent with neutral growth
iScience 24(1):101901
In an agent-based model, cells that degrade extracellular matrix open up new space, a public good that also benefits non-producers, while other cells exploit it by proliferating. Although both strategies were under strong selection, the resulting clonal architecture looked consistent with neutral growth, a caution for inferring neutral evolution from tumor sequencing.
2020
-
Evidence for hypoxia increasing the tempo of evolution in glioblastoma
British Journal of Cancer 123(10):1562–1569
An agent-based model of a proliferative hierarchy under heterogeneous oxygen, compared with glioblastoma histology, indicates that cell division is effectively upregulated in hypoxic niches, which give tumor cells avenues to spread. Hypoxia may therefore raise the tempo of evolution and make key mutations more likely.
-
Modeling Osteocyte Network Formation: Healthy and Cancerous Environments
Frontiers in Bioengineering and Biotechnology 8:757
A stochastic agent-based model of bone formation by osteoblasts and osteocytes, with matching mean-field equations, shows how bone secretion, osteoblast differentiation and cancer change the density and topology of the osteocyte network, and explores how bisphosphonates such as zoledronic acid affect it.
-
High School Internship Program in Integrated Mathematical Oncology (HIP IMO): Five-Year Experience at Moffitt Cancer Center
Bulletin of Mathematical Biology 82(7):91
The structure, curriculum and outcomes of Moffitt's High school Internship Program in Integrated Mathematical Oncology, which trained 59 students between 2015 and 2019.
-
What Is the Storage Effect, Why Should It Occur in Cancers, and How Can It Inform Cancer Therapy?
Cancer Control 27(3):1073274820941968
The storage effect, an ecological mechanism of coexistence, needs a temporally variable environment, buffered population growth and species-specific responses to the environment. Tumors appear to meet all three conditions, so the storage effect probably contributes to intratumor diversity, with consequences for treatment.
-
Stromal reactivity differentially drives tumour cell evolution and prostate cancer progression
Nature Ecology & Evolution 4(6):870–884
A multiscale model integrating biology and pathology shows that stromal ecology, whether inhibitory, highly reactive or non-reactive, correlates directly with prostate tumor growth but inversely with tumor evolution. Aggressive, environment-independent prostate cancer may thus result from poor stromal ecology, and stromal markers could improve prognosis.
-
Hybrid Automata Library: A flexible platform for hybrid modeling with real-time visualization
PLOS Computational Biology 16(3):e1007635
HAL is an open-source Java library for building spatial hybrid models, with on- and off-lattice agents, finite-difference diffusion fields and real-time visualization, available on GitHub under the MIT license.
2019
-
Fibroblasts and alectinib switch the evolutionary games played by non-small cell lung cancer
Nature Ecology & Evolution 3(3):450–456
A game assay measures the evolutionary games played in co-cultures of alectinib-sensitive and -resistant lung cancer cells. Both the drug and cancer-associated fibroblasts switch the type of game, from Leader to Deadlock: empirical support for the idea that therapy can change not only the players but the game.
-
Simulated ablation for detection of cells impacting paracrine signalling in histology analysis
Mathematical Medicine and Biology: A Journal of the IMA 36(1):93–112
A PDE model that simulates removing each cell from a histology image estimates whether it absorbs or secretes diffusible signals. Applied to a breast cancer specimen stained for HER2, it helps separate responses to paracrine gradients from contact-mediated and (epi)genetic sources of heterogeneity.
2018
-
Bifurcation Mechanism Design—From Optimal Flat Taxes to Better Cancer Treatments
Games 9(2):21
Bifurcations, abrupt changes in behavior after small parameter changes, can be used by design: transient parameter changes can lock a system into a better equilibrium. The idea is applied to flat taxation and then to a game of tumor metabolism to derive candidate treatment strategies.
-
Fractionated Follow-Up Chemotherapy Delays the Onset of Resistance in Bone Metastatic Prostate Cancer
Games 9(2):19
An evolutionary game of bone remodeling and its co-option by prostate cancer, including the physical bone environment, shows resistant cells taking over under conventional chemotherapy. 'Fractionated follow-up' (one block of chemotherapy followed by alternating smaller doses and holidays) controlled the tumor and pathological bone better than continuous or periodic schedules.
-
Size Matters: Metastatic Cluster Size and Stromal Recruitment in the Establishment of Successful Prostate Cancer to Bone Metastases
Bulletin of Mathematical Biology 80(5):1046–1058
In a hybrid cellular automaton of bone, prostate cancer clusters seeding the bone must fall within a size range to establish a metastasis: large enough to succeed, but not so large that cancer and stromal cells compete for scarce resources. Recruiting mesenchymal stem cells can compensate for small seeding clusters.
-
Testing Three Hypotheses of the Contribution of Geometry and Migration Dynamics to Intestine Crypt Evolution
The 2018 Conference on Artificial Life pp. 420–427Conference paper
An agent-based model of intestinal crypts tests three hypotheses for what governs how fast a crypt becomes monoclonal, a key metric for the establishment of cancer-initiating mutations. The number of basal stem cells had a direct influence, with a lesser role for extracellular influences.
2017
-
The Influence of Cellular Characteristics on the Evolution of Shape Homeostasis
Artificial Life 23(3):424–448
In an individual-based model in which an evolving intracellular network controls each cell, 96 combinations of cell capabilities were compared for their ability to evolve robust, homeostatic structures. Long-range growth factors helped most, and solutions evolved in unstable environments were more robust and adaptive.
-
Mean-field approach to evolving spatial networks, with an application to osteocyte network formation
Physical Review E 96(1):012301
Derives a mean-field integro-differential equation for evolving networks whose nodes carry a state such as position, validated against stochastic simulations, and applies it to osteocyte networks in bone: faster differentiation gives denser osteocytes with fewer dendrites.
-
Evolutionary emergence of angiogenesis in avascular tumors using a spatial public goods game
PLOS ONE 12(4):e0175063
Treats secretion of pro-angiogenic factors as a public good in a spatial evolutionary game on graphs. Population structure shapes whether pro-angiogenic clones emerge; they form clusters running from the surrounding vasculature toward the tumor center, which thin out after anti-angiogenic therapy.
-
Homeostasis Back and Forth: An Ecoevolutionary Perspective of Cancer
Cold Spring Harbor Perspectives in Medicine 7(9):a028332
Argues that an evolutionary view of cancer must be complemented by an ecological one that accounts for the tissue homeostasis tumors disrupt and co-opt, that therapies aiming to restore homeostasis deserve clinical attention, and that the mechanistic models needed must embrace more complexity than modelers are often comfortable with.
-
Cancer treatment scheduling and dynamic heterogeneity in social dilemmas of tumour acidity and vasculature
British Journal of Cancer 116(6):785–792
The 'double goods game' treats glycolytic acid as a public good and oxygen from VEGF-driven vascularization as a club good, giving three phenotypes. Dynamics fall into fully glycolytic, fully angiogenic or cyclic polyclonal regimes, so the timing and order of treatments such as lonidamine or vascular normalization matter.
2016
-
Predictive computational modeling to define effective treatment strategies for bone metastatic prostate cancer
Scientific Reports 6(1):29384
A hybrid cellular automaton of bone-metastatic prostate cancer predicted that TGF-β inhibition works best before metastases are established, a prediction validated in two in vivo models. Seeded with patient data, the model predicts how TGF-β inhibitors would affect disease evolution.
2015
-
Edge effects in game-theoretic dynamics of spatially structured tumours
Journal of The Royal Society Interface 12(108):20150154
Adds spatial structure to the go-versus-grow game with the Ohtsuki–Nowak transform. Space can favor the invasive strategy, and at static boundaries such as vessels or basement membranes an edge effect allows invasive cells at the tumor edge even when the bulk has none, so pathology that ignores position can undercount invasive cells.
-
Current and Emerging Therapies for Bone Metastatic Castration-Resistant Prostate Cancer
Cancer Control 22(1):109–120
A review of approved and emerging therapies for metastatic castration-resistant prostate cancer, including drugs targeting androgen signaling and the bone microenvironment, and of how computational biology could help with tumor heterogeneity and patient selection.
2014
-
Improving treatment strategies for patients with metastatic castrate resistant prostate cancer through personalized computational modeling
Clinical & Experimental Metastasis 31(8):991–999
A perspective, written by clinicians, researchers and mathematicians, arguing that personalized computational models and optimization algorithms could help oncologists sequence therapies for metastatic castration-resistant prostate cancer in bone, with a worked example using genetic algorithms.
-
Evolutionary game theory: molecules as players
Molecular BioSystems 10(12):3066–3074
The second of two reviews of evolutionary game theory below the level of organisms, with viruses, genes, catalytic RNA, transposable elements and proteins as the players.
-
Evolutionary game theory: cells as players
Molecular BioSystems 10(12):3044–3065
The first of two reviews of evolutionary game theory below the level of organisms, covering cells as players: metabolic strategies, public goods and cross-feeding, cyclic competition, signaling, pathogens and the immune system, and cancer.
-
Invasion and proliferation kinetics in enhancing gliomas predict IDH1 mutation status
Neuro-Oncology 16(6):779–786
In 172 patients with contrast-enhancing glioma, a patient-specific growth model based on MRI found IDH1-mutant tumors relatively more invasive, mainly because of lower net proliferation. The model's invasion profile predicted IDH1 status (area under the curve 0.91), suggesting a non-invasive test before surgery.
-
An Integrated Computational Model of the Bone Microenvironment in Bone-Metastatic Prostate Cancer
Cancer Research 74(9):2391–2401
A hybrid cellular automaton of bone remodeling reproduces the coupling of bone resorption and formation and, seeded with a single prostate cancer cell, the vicious cycle of bone metastasis. It reveals distinct osteolytic and osteogenic phases and a key role for mesenchymal stromal cells; simulated anti-RANKL therapy suggests better targeting or dosing could improve outcomes.
-
Microenvironmental Variables Must Influence Intrinsic Phenotypic Parameters of Cancer Stem Cells to Affect Tumourigenicity
PLoS Computational Biology 10(1):e1003433
A hybrid cellular automaton of a generic stem-cell-driven tissue maps which traits of tumor-initiating cells disrupt homeostasis and cause overgrowth. The simple microenvironmental variable had little effect, suggesting that models need direct feedback from the microenvironment onto stem-cell behavior.
2013
-
Senescent Fibroblasts in Melanoma Initiation and Progression: An Integrated Theoretical, Experimental, and Clinical Approach
Cancer Research 73(23):6874–6885
A hybrid cellular automaton of skin, combined with experiments and human histology, shows that senescent fibroblasts create a pro-tumorigenic microenvironment, with proteases as a key factor, that cooperates with mutant melanocytes to drive melanoma initiation and progression. Simulated stroma-targeting therapy delayed both.
-
Exploiting ecological principles to better understand cancer progression and treatment
Interface Focus 3(4):20130020
Argues that cancer is better seen as a disruption of the balance among interacting cells and microenvironment in an organ, an ecosystem in homeostasis, than as a group of mutated cells, and uses several mathematical and computational models to show how ecological principles can inform progression and treatment.
-
A mathematical model of tumour self-seeding reveals secondary metastatic deposits as drivers of primary tumour growth
Journal of The Royal Society Interface 10(82):20130011
A model of circulating tumor cells that first meet the lung capillary bed shows that 'secondary seeding', in which metastatic deposits shed cells that return to the primary tumor, is several orders of magnitude more likely than direct self-seeding by the primary.
-
Mathematical Modeling of the Metastatic Process
Experimental Metastasis: Modeling and Analysis pp. 189–208Book chapter
A short review of mathematical and computational models of metastasis: the emergence of the metastatic phenotype, the timing and size distribution of metastases, the dormancy of micrometastases and patterns of spread.
2012
-
Exploiting Evolution To Treat Drug Resistance: Combination Therapy and the Double Bind
Molecular Pharmaceutics 9(4):914–921
An evolutionary game framework for the evolutionary double bind, in which resistance to one therapy increases sensitivity to another, reproduces experimental data on the synergy between a p53 cancer vaccine and chemotherapy and explains it through a commensal relationship between tumor phenotypes.
-
Investigating prostate cancer tumour–stroma interactions: clinical and biological insights from an evolutionary game
British Journal of Cancer 106(1):174–181
An evolutionary game of stromal cells, self-reliant tumor cells and stroma-dependent tumor cells explains three observed courses of prostate cancer (indolent, aggressive and stroma-independent, or stroma-co-opting) and how treatments such as hormone therapy change which phenotypes are selected.
-
'Go or Grow': the key to the emergence of invasion in tumour progression?
Mathematical Medicine and Biology 29(1):49–65
Argues that the rapid recurrence of glioblastoma after resection cannot be explained by mutations alone, and uses a lattice-gas cellular automaton to show that the go-or-grow dichotomy combined with hypoxia can explain the transition to invasive phenotypes.
2011
-
Evolving homeostatic tissue using genetic algorithms
Progress in Biophysics and Molecular Biology 106(2):414–425
An evolutionary algorithm evolves cell-signaling networks in a 3D hybrid cellular automaton to build and maintain a monolayer tissue. Two contrasting homeostatic strategies emerged, high turnover versus dividing only when needed, with different robustness to wounding and mutation.
-
Production of 2-hydroxyglutarate by isocitrate dehydrogenase 1-mutated gliomas: an evolutionary alternative to the Warburg shift?
Neuro-Oncology 13(12):1262–1264
A letter asking whether 2-hydroxyglutarate, produced in large amounts by IDH1-mutant gliomas and actively exported from cells, could act as an evolutionary alternative to the Warburg shift to aerobic glycolysis.
-
The role of IDH1 mutated tumour cells in secondary glioblastomas: an evolutionary game theoretical view
Physical Biology 8(1):015016
An evolutionary game of four phenotypes (autonomous growth, invasive, glycolytic and invasive-glycolytic) in IDH1-mutant secondary glioblastoma reproduces two experimental findings and predicts how invasive growth depends on vascularity.
-
Computational analysis of the influence of the microenvironment on carcinogenesis
Mathematical Biosciences 229(1):22–29
A two-dimensional cellular automaton of a cancer colony growing in vitro shows how oxygen, nutrients and crowding shape the order in which cells acquire cancer traits, with environmental stress determining which phenotypes dominate.
2010
-
Evolution, Regulation and Disruption of Homeostasis and Its Role in Carcinogenesis
Multiscale Cancer Modeling (Chapman & Hall/CRC Mathematical & Computational Biology) pp. 1–30Book chapter
-
Perspectives on Tissue Interactions in Development and Disease
Current Molecular Medicine 10(1):95–112
A review of the biological models used to study stromal–epithelial interactions, and of attempts to describe these interactions as ecological relationships and model them mathematically.
2009
-
Microenvironmental Independence Associated with Tumor Progression
Cancer Research 69(22):8797–8806
A hybrid discrete-continuum model parameterized with traits of related normal, transformed and tumorigenic mammary cell lines shows transformed cells winning in resource-rich environments and tumorigenic phenotypes winning when space or growth factors are limited. A game-theory model supports microenvironmental independence as an advantage under scarcity.
-
The Role of Transforming Growth Factor-β–Mediated Tumor-Stroma Interactions in Prostate Cancer Progression: An Integrative Approach
Cancer Research 69(17):7111–7120
A hybrid cellular automaton based on the structure of the human prostate shows how stochastic tumor–stroma interactions involving TGF-β, matrix-degrading enzymes and stromal activation can lead to suppression, modest proliferation or unopposed growth. A prediction about epithelial TGF-β and glandular stability was confirmed in vivo.
2008
-
Evolutionary game theory elucidates the role of glycolysis in glioma progression and invasion
Cell Proliferation 41(6):980–987
An evolutionary game of autonomous-growth, glycolytic and invasive phenotypes in glioma shows that invasion is more likely to evolve after glycolysis appears, suggesting that therapies making the switch to glycolysis more costly could reduce the emergence of invasive cells.
-
Studying the emergence of invasiveness in tumours using game theory
The European Physical Journal B 63(3):393–397
A game-theoretical model, with cellular automaton simulations, of when mutations that increase cell motility can spread through a tumor of rapidly proliferating cells, and the therapies this suggests for preventing progression to invasion.
-
The Evolution of Robust Development and Homeostasis in Artificial Organisms
PLoS Computational Biology 4(3):e1000030
Artificially evolved cellular automaton organisms with distinct embryonic and homeostatic phases all recovered quickly from wounds, most robustly with a stratified, tissue-like architecture, suggesting that such robustness may be a by-product of the evolution of development rather than directly selected.
-
A Game Theoretical Perspective on the Somatic Evolution of cancer
Selected Topics in Cancer Modeling (Modeling and Simulation in Science, Engineering and Technology) pp. 1–16Book chapter
Introduces game theory as a tool for studying how cell phenotypes emerge in a tumor and critically reviews its applications in cancer research.
2005
-
Using genetic algorithms to evolve three-dimensional microstructures from two-dimensional micrographs
Metallurgical and Materials Transactions A 36(7):1643–1652
2004
-
Evolving 3D Microstructures Using a Genetic Algorithm
Materials Science Forum 467-470:1019–1024
Introduces MicroConstructor, a genetic algorithm that generates 3D material microstructures whose stereological parameters match those measured in 2D micrographs, avoiding costly 3D imaging.
-
Evolving and growing microstructures of materials using biologically inspired CA
Proceedings. 2004 NASA/DoD Conference on Evolvable Hardware pp. 275–282Conference paper
MicroConstructor uses a genetic algorithm to evolve cellular automata, inspired by developmental biology, that self-organize into 3D microstructures for materials analysis.
2003
-
Evolving Cellular Automata to Grow Microstructures
Genetic Programming (Lecture Notes in Computer Science) pp. 1–10Book chapter