Our current approach to cancer treatment has been largely driven by finding molecular targets, those patients fortunate enough to have a targetable mutation will receive a fixed treatment schedule designed to deliver the maximum tolerated dose (MTD). Cancers are complex evolving systems that adapt to therapeutic intervention through a suite of resistance mechanisms, therefore whilst MTD therapies generally achieve impressive short-term responses, they unfortunately give way to treatment resistance and tumor relapse. The importance of evolution during both tumor progression, metastasis and treatment response is becoming more widely accepted. However, MTD treatment strategies continue to dominate the precision oncology landscape. Evolutionary therapy is a new evolution inspired treatment paradigm that seeks to exploit how a cancer evolves under treatment through smart drug dosing and sequencing often informed by mathematical modelling. Adaptive therapy is an evolutionary therapy that aims to slow down the emergence of drug resistance by controlling tumor burden through competition between drug sensitive and resistant cell populations. Adaptive therapy specifically alters the treatment schedule (timing and dose) in response to a patient’s disease dynamics, often stopping therapy or deescalating dose when burden is low and starting therapy or increasing dose when burden is high. This approach was inspired by pest management and developed through mathematical model driven insights and has been shown to work in preclinical animal models (prostate, ovarian, melanoma, breast) and in pilot clinical trials (NCT02415621; NCT05189457; NCT03543969). Recently, phase 2 adaptive therapy trials in prostate (NCT05393791) and ovarian cancer (NCT05080556) are testing the treatment break and treatment deescalation approaches respectively. In this talk we will utilize differential equation and cellular automata models as well deep reinforcement learning (DRL) approaches to tackle different aspects of (i) Adaptive therapy including: How best to optimize the treatment switch threshold; The importance of appointment frequency, and (ii) DRL integration with mathematical models including: How best to train DRL models with limited clinical data; How to exploit transfer learning as models increase in complexity; The importance of memory in DRL models.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Successful treatment both of microbial infections and of cancer is often hindered by the evolution of resistance. A prominent approach to limit the risk of resistance is the administration of multiple drugs during treatment. In the absence of cross resistance, such multi-drug treatments increase the genetic barrier to resistance, which is a priori expected to prevent or at least delay the evolution of resistance. In classical combination therapy, two or more drugs are administered together, whereas in sequential therapy, drugs are cycled throughout treatment. While the idea of multi-drug therapy is straightforward, it is nevertheless not clear whether it is always superior to mono-therapy nor what an optimal design looks like, given the myriad of options for how to combine different drugs. In this talk, I will present results from two mathematical models for multi-drug treatment strategies of bacterial infections. Complementing each other, the models are analysed to explore the effect of a variety of factors on combination and on sequential therapy. I will end the talk by a discussion of conceptual challenges in modeling the joint effects of multiple drugs. While the talk focuses on treatment of bacterial infections, similar concepts are relevant for the evolution of resistance in cancer and many of the insights likely extend to cancer chemotherapy.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

development by quantitatively linking drug exposure to tumor response. Tumor growth inhibition (TGI) models are widely used to describe and predict tumor dynamics under treatment, enabling the evaluation of drug efficacy and optimization of dosing regimens. In this talk, we will review several commonly used TGI models, as well as develop new ones, and demonstrate their application to preclinical data.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

In this overview talk I want to present three current collaborative research topics in medicine where mathematical modelling might be useful: 1. Understanding the formation & dynamics of in-vitro heart-like organoids (Mendjan group, IMBA), 2. Predicting the supply & demand dynamics of lactating mothers (Willow breast pumps company), 3. Investigating the effect of heterogeneity in cancer cells (Winkler group, MedUni). All three topics are early stage collaborations with experimentalists and are grounded in experimental data. The suggested mathematical methods range from vertex models, nonlinear differential equations, data analysis to optimisation. I will discuss the challenges and potential insights to be gained from using mathematical modelling.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Glioblastoma (GBM) is a highly aggressive brain tumour marked by cellular heterogeneity, therapy resistance, and rapid progression. Molecular profiling has revealed that a subpopulation of GBM cells with stem cell-like properties, including self-renewal, multipotency, and resilience to treatment, likely drives tumour growth and recurrence.While many mathematical models describe spatial aspects of tumour growth, they often neglect underlying cellular hierarchies. In this talk, I present a new model of glioblastoma that extends our previous work on neural stem cell dynamics, incorporating cellular states identified in patient-derived single-cell transcriptomic data. Our integrated approach reveals that GBM cells recapitulate an activation state architecture (ASA) similar to that of healthy NSCs, but fail to regulate transitions between cell states, maintaining aberrant signalling activity (e.g. Wnt) throughout. The model identifies the stem cell activation rate as a key parameter linked to tumour aggressiveness. Interestingly, some tumours appear to compensate for reduced activation via enhanced self-renewal, reflecting behaviours also observed in aging neurogenesis under some perturbations. We conclude by outlining open modelling challenges, in particular, how to link cellular architecture to the spatial structure of growing tumours in a mathematically coherent way.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Despite the revolutionary impact of immune checkpoint inhibition (ICI) on cancer therapy, for most indications the majority of patients do not sustain a durable clinical benefit. In this work, we explore the theoretical consequences of the existence of multiple states of immune cell exhaustion on response to ICI therapy. In particular, we consider the emerging understanding that T cells can exist in various states: fully functioning cytotoxic cells, reversibly exhausted cells that are minimally cytotoxic but targetable by ICIs, and terminally exhausted cells that are cytotoxic yet not targetable by ICIs. Under the assumption that tumor-induced inflammation triggers the transition between these T cell phenotypes, we developed a conceptual mathematical model of tumor progression subject to treatment with an ICI that accounts for multi-stage immune cell exhaustion. Simulations reveal that treatment response sensitively depends on both the dose and frequency of drug administration, with standard high-dose, low-frequency protocol failing a significant fraction of a heterogeneous population. Conversely, a metronomic-like strategy that distributes a fixed amount of drug over many doses given close together is predicted to be effective across the largest proportion of a heterogeneous population. A sensitivity analysis identifies that anti-inflammatory drugs make the ideal combination partner to enhance the efficacy of ICIs across protocol space. We conclude we recent supporting clinical evidence, and plans to iterate on the model using this emerging clinical data.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

T lymphocytes are crucial cells of the immune system that recognize infected cells or tumor cells with high specificity through their T-cell receptor. Among T lymphocytes, cytotoxic T lymphocytes are endowed with the potential to kill infected cells or tumor cells via the release of lytic granules containing cytotoxic molecules. The inability to control a viral infection or the growth of a tumor is often related to suboptimal responses of cytotoxic T lymphocytes. Understanding the factors that restrict cytotoxic T lymphocyte function is key to understand disease progression and to design T lymphocyte-centered therapies such as CAR-T cell immunotherapy. Our laboratory studies the tuning of T lymphocyte activation and function via quantitative cell imaging approaches. Our aim is to capture the spatial organization of key molecular events such as activation of adhesion receptors and organization of the actin cytoskeleton at the immunological synapse that forms between T lymphocytes and target cells. In this seminar, we will present our recent findings pertaining to human T lymphocyte biology. We will exemplify the importance of partnerships with computational scientists. Such partnerships are key to design advanced analytical solutions for automated lymphocyte imaging. They are key to understand the topography and calibration of T lymphocyte adhesion. We are also investing in developing ML models to predict efficacy of immunotherapy in individual patients.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

The Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs) are a group of slowly progressing hematological malignancies primarily characterised by an overproduction of myeloid blood cells. Without treatment, they result in severe complications such as thrombosis, bleeding, infections, bone marrow failure, and progression to acute myelogenous leukemia. The hematopoietic system is responsible for the formation of blood cells. It consists of cells of different maturity levels, starting with the least mature hematopoietic stem cells (HSCs) in the bone marrow, continuing with the more mature so-called progenitor and precursor cells, and ending with the fully mature cells in the peripheral blood. All hematopoietic cells are derived from HSCs. HSC proliferation needs to fulfil two roles: maintaining the HSC pool and producing more mature committed cells that will eventually become fully mature. The hematopoietic system is subjected to a complex regulatory network which adapts the production of mature cells to the current state of the organism. It is believed that MPNs develop from a single mutated stem cell that proliferates and slowly produces both mutated stem cells, mutated progenitors, and consequently also mutated mature cells. We construct a mathematical model relying on key mechanisms of the hematopoietic system and include mechnisms of several drugs to model response to treatment. A major focus of the talk is the open problem of using mathematical modeling to design a ’STOP interferon’-trial where well responding patients will be exposed to treatment holidays. The work is based on close collaborations with clinicians at Department of Hematology, Zealand University Hospital, Denmark.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Neuroblastoma is one of the deadliest pediatric malignancies. High-risk patients have a 50% risk of relapse and refractory disease to multimodal therapy. One way to understand the potential reason for high-risk neuroblastoma severity is to search for its roots. Neuroblastoma most often occurs in the adrenal gland and might be associated with errors in the development of this organ. Studying the healthy development of the adrenal gland with the power of single-cell omics provided a reference for the lineage progression and the developmental delay in neuroblastomas. These large datasets provide insights into tumor heterogeneity, the potential cell of origin of neuroblastoma, and allow us to hypothesize about the mechanism of tumor initiation and healthy cells’ plasticity. Currently in Kameneva Lab at St.Anna CCRI, we harness this data to build models of tumor initiation using pluripotent stem cells, with the idea to unpick cell-type-specific vulnerabilities for tumor initiation. In my talk, I will highlight the recent progress in this field and the challenges that need to be met for the faithful modeling of tumor initiation. 11

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Hematopoietic stem cell transplantation (HSCT) is the only curative option for a wide range of hematologic malignancies and inherited disorders of the blood-forming system. However, its clinical benefits are often limited by severe short- and long-term complications. The risk and nature of these complications vary significantly depending on the patient’s age and overall physical condition. Elderly or frail patients are often considered ineligible for HSCT due to the risk of transient immune and coagulation suppression in the early post-transplantation phase. In younger individuals, longterm complications are of greater concern. These include the expansion of donor- or host-derived pre-malignant clones and chronic graft-versus-host disease (cGVHD). The latter arises following allogeneic transplantation when donor-derived immune cells recognize host tissues as foreign. This immune mismatch triggers chronic inflammation and tissue fibrosis, potentially resulting in life-long morbidity. Predicting the occurrence and clinical trajectory of cGVHD remains a major challenge, as robust biomarkers are currently lacking. Another rare but life-threatening complication is graft failure, characterized by the cessation of blood cell production. The first part of the presentation focuses on the development and calibration of mechanistic mathematical models to describe the expansion dynamics of transplanted stem and progenitor cells within the recipient’s bone marrow. These models incorporate multiple immature blood cell types and their regulation through feedback mechanisms. The aim is to use these models to explore how complications may be mitigated – either by increasing the number of transplanted cells or by enriching grafts with specific non-stem cell populations. The second part is devoted to the clinical course of cGVHD. A basic immune-metabolism model is integrated with novel clinical trial data to identify potential biomarkers for the onset and progression of cGVHD. The goal is to better understand the underlying mechanisms driving the high inter-individual variability in cGVHD symptoms and to support more personalized approaches to post-transplant care

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Cancer cachexia is the loss of muscle and adipose tissues with advanced stages of cancer, that causes fatigue, weakness, pain, poor quality of life, and poor treatment options and survival outcomes. In this talk I will present work on modelling and analysis of cancer cachexia. We have developed two frameworks to describe the response of muscle tissue to cancer. First, we model the role of stem cells in tissue maintenance and use the model to examine mechanisms of cancer induced muscle loss, including disruption of the differentiation pathway. Next, we model the interactions of the immune response with cancer and muscle mass to explore how the cancer-dysregulated immune reaction affects healthy muscle mass. Finally, I will conclude with a discussion of model-informed potential treatment targets that may help preserve lean mass during cancer treatment.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Immune cell infiltration into solid tumours was initially considered a positive sign of the host attempting to eliminate malignant threats. However, it is now recognized that tumours can subvert immune cells, reprogramming them to support rather than suppress tumour growth. Immunotherapy aims to reverse this process by enhancing or restoring effective immune responses. Despite notable successes, the complex and dynamic nature of tumour-immune interactions poses significant challenges, particularly in understanding why some patients respond to immunotherapy while others do not. In this talk, we will show how a range of mathematical tools – including mathematical modelling and topological data analysis – can shed light on these complexities. We will share recent findings that highlight the complementary strengths of these approaches in advancing our understanding of immunotherapy and improving patient outcomes.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

The pregnane X receptor (PXR) is a liver-enriched, ligand-activated transcription factor that is considered a crucial sensor in the xenobiotic response. PXR controls the transcription of drughandling genes, including those from the cytochrome P450 family, such as, CYP3A4, CYP2C9, and CYP2B6. CYP3A4, for example, is the most important enzyme that is involved in the metabolism of approximately 50% of all clinically used drugs. Several mathematical models exist that describe the effect of ligands on PXR-activated CYP3A4 gene expression in primary human hepatocytes (PHHs). However, these studies have only focused on CYP3A isomorphs and have employed data evaluated from PHH monolayer cultures, which degenerate shortly after exposure to ligands. We propose a mathematical model that describes PXR activation and the expression kinetics of select target genes that are upregulated. We calibrate this model using long-term target gene messenger RNA (mRNA) expression data from a state-of-the-art in vitro system of 3D PHH spheroids. We discuss the effects of two PXR activators - a model PXR ligand, rifampicin, and its primary metabolite, 25-desacetyl rifampicin - on CYP mRNA expression. Finally, we discuss the experimental validation of the theoretical results.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Bispecific T cell engagers (TCEs) can exhibit bell-shaped efficacy curves, where increasing the dose beyond a certain point leads to reduced, not improved, efficacy. This counterintuitive behavior arises when efficacy depends on forming a trimeric complex between drug, tumor target, and T cell receptor, as is the case with teclistamab, a bispecific targeting BCMA and CD3 in multiple myeloma. Using a semi-mechanistic PK/PD model and a virtual patient population, we demonstrate that apparent loss of response may reflect overdosing rather than true resistance. We explore how measurable pre-treatment biomarkers, such as soluble and membrane-bound BCMA, can guide dose optimization and patient stratification. The model supports a shift toward semi-personalized dosing strategies and highlights that lowering the dose may, in some patients, restore efficacy.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Cancer is an evolving system, controlled by the forces of mutation, selection and genetic drift. Darwinian evolution underpins the life history of all cancers from their malignant transformation, to metastatic colonisation and therapy resistance. Understanding these dynamics is therefore crucial to tackling the disease. Chromosomal instability (CIN), where chromosomal structure is frequently altered during cell division, is a cancer hallmark and linked to aggressive disease. CIN helps fuel cancer evolution by acting as a mutational process that can cause dramatic changes to the genomes of cells, but is poorly understood in terms of its evolutionary dynamics. In my talk I will introduce work in various adult and paediatric cancers that demonstrate the insights that understanding CIN and cancer evolution can give us. I will propose how new technologies and modelling approaches can help us quantify this important evolutionary fuel in cancer.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

Radiotherapy is the most applied cancer treatment with over 60% of cancer patients receiving radiation at some point in their clinical care. Yet, we have no biomarker to understand and predict why patients with comparable clinical presentations may have different radiation responses and outcomes. Mathematical and computational models, trained on patient-specific clinical data, can help understand and predict interpatient heterogeneity and personalize radiotherapy protocols. We deploy different mathematical (ordinary and partial differential equations), spatial statistical (2- point correlation function, power spectral density), computational (agent-based models), machine learning (neural networks, generative adversarial networks), and data-driven approaches to analyze pre-treatment tissue biopsies and radiology images to characterize the spatial architecture of the tumor, immune infiltration, immune interplay, and tumor volume dynamics of different cancers. From these, we can create digital twins and simulate in silico clinical trials to predict response and outcome to different radiation protocols and derive optimal radiation regimens for each patient. Routinely collected tissue biopsies can predict pan-cancer radiotherapy response and outcome with high accuracy. Pre-treatment radiology can predict the optimal radiation fractionation protocol, and on-treatment response dynamics can predict the optimal total dose to maximize tumor control and minimize radiation-associated comorbidities. Integrated scientific modeling is well-positioned to guide clinical decision-making for individual cancer patients. Prospective clinical trials will be needed to validate the prediction accuracy of the presented methodologies.

• Thematic program: Quantitative Methods in Biology and Medicine (2025/2026)
• Event: 8th Workshop "Mathematics for Medicine (against Cancer)" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

Cancer invasion and metastasis are inherently multiscale processes, shaped by complex interactions between cancer cells and the tumour microenvironment. A central mechanism driving cancer heterogeneity is the Epithelial to Mesenchymal Transition (EMT), which enables cancer cells to switch between phenotypic states with distinct migratory and invasive capabilities. I will first describe a hybrid multiscale model that couples individual-based representations of migrating cancer cells with continuum descriptions of the evolving tumour mass, illustrating how local EMT-driven phenotype changes influence macroscopic invasion patterns. I will then discuss a phenotype-structured population model that incorporates continuous transitions along the epithelial–mesenchymal spectrum, providing a tractable means to study the emergence and maintenance of phenotypic diversity within the tumour.

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)

• Event: Workshop "Mathematical modeling in biology and medicine" (2025)