The development of new anti-cancer therapies is time-intensive and costly, and traditionally relies on extensive laboratory and clinical experimentation. To complement these efforts, mathematical and computational approaches are increasingly used to address key research questions, reduce experimental burden, and guide decision-making. In oncology, a range of emerging treatment modalities-including immunotherapy, gene therapy, targeted therapies, nanomedicine, and oncolytic virotherapy-represent a significant shift towards more precise and personalised cancer care. These approaches are designed to preferentially target cancer cells while limiting damage to healthy tissue, often by engaging or enhancing the body's own immune response. Mathematical modeling offers a unifying framework for studying the complex, multi-scale interactions between tumours, therapies, and the immune system. By integrating experimental and clinical data, such models enable the simulation, analysis, and optimisation of treatment strategies, while also helping to identify biological constraints and practical challenges related to therapy delivery. As such, modeling serves as a critical bridge between experimental research and clinical application, supporting collaboration across disciplines 1.Figure 1 provides a visual roadmap for the argument developed in this editorial. We propose that the four articles in this collection are not isolated contributions but exemplars of a broader, unified pipeline. This pipeline begins with the recognition that current clinical practice faces structural limitations -costly trials, one-size-fits-all dosing, and static protocols, among others -which create demand for quantitative approaches. The central hub, mathematical modeling, supplies a diverse toolkit from which five therapeutic This Research Topic aims to increase understanding of how mathematical modeling can serve as a pivotal tool in the evolution of cancer treatment. Specific goals include optimizing existing treatment strategies, closely examining the dynamics of emerging therapies, elucidating their impact on tumor growth, and predicting responses to different treatment modalities. Through various modeling techniques such as ordinary differential equations and agent-based modeling, this topic seeks to illuminate the intricate mechanisms of action of these therapies and facilitate the development of improved treatment protocols.The collection includes contributions that develop and simulate mathematical models to predict treatment efficacy, explore inter-treatment interactions within a mathematical framework, analyze the influence of the tumor microenvironment on treatment outcomes, optimize treatment schedules and dosage plans, and construct and validate models that predict outcomes of combination therapies. Importantly, the articles focus on robust model validation, precise parameter estimation, the utilization of models in clinical or virtual trials, and strategies for integrating multiple treatment modalities.Understanding the complexity of cancer treatment necessitates multiscale approaches that can capture both stochastic cellular dynamics and deterministic population-level behaviors. The four articles in this collection illustrate the breadth of mathematical modeling approaches that can be used to address these challenges, each focusing on different aspects of cancer therapy: from viral dynamics and spatiotemporal pattern formation to surrogate model construction, evolutionary principles, and causal statistical inference. Table 1 provides a structured summary of each contribution and an editorial commentary on its methodological significance and clinical relevance.Bansod and Hillen analyse the spatiotemporal dynamics of oncolytic virotherapy using a reactiondiffusion model that couples uninfected tumour cells, infected tumour cells, and free virus particles.By applying a centre-manifold reduction to a classical cell-infection-virus (C-I-V) model, they derive a complex Ginzburg-Landau equation whose parameters depend on the underlying biological kinetics.Their analysis shows that spatial patterns previously observed in such models (e.g., rings and fragmented structures) evolve over longer timescales into spiral waves and turbulent dynamics, originating from a Hopf bifurcation in the reaction kinetics. The authors identify the asynchrony parameter as the key determinant of pattern stability and demonstrate that biologically relevant parameter values place the system within the turbulent regime. These findings establish spatiotemporal complexity as an intrinsic feature of oncolytic virotherapy and suggest fundamental limits on tumour eradication by viral monotherapy alone. The authors show that the equilibrium cancer cell density decreases with increasing initial immune cell fraction and with increasing immune cell competitive strength, demonstrating that combining a larger initial immune cell dosage with greater competitive ability yields compounded therapeutic benefit. Kim presents a hybrid modeling framework that combines stochastic trait evolution with birth-death population dynamics to study lineage diversification under stabilising selection and random drift. The model is calibrated using long-term evolution data from three Escherichia coli lineages (wild-type, priA, and recG), allowing estimation of key parameters governing typical trait values, variability, and the strength of stabilising forces. This approach distinguishes three evolutionary regimes: stable behaviour in the wild-type, highly variable and plastic dynamics in priA, and instability with frequent collapse in recG.The priA lineage, characterised by high variability and weak stabilisation, is proposed as an analogue for hypermutable tumour subclones that can drive paediatric cancer relapse despite low mutation burdens, while recG resembles fragile lineages under replication stress. Illustrative treatment simulations based on the priA regime show how cyclic therapy can produce oscillatory traits, suppression-rebound dynamics, and clonal loss. These simulations are presented as hypothesis-generating, with the framework designed for future calibration to longitudinal clinical data to enable evolution-aware, patient-specific modeling.Liao et al. address causal inference challenges in longitudinal studies where treatments are assigned sequentially and time-varying covariates are influenced by earlier treatments. This feedback may confound the choice of subsequent treatments, posing significant difficulties for standard methods. The authors propose an approach based on using standardized point effects to estimate blip effects, defined as the net causal effect of a treatment on the outcome when all subsequent treatments are set to control. Their key innovation is to group point-treatment effects into a small number of scientifically motivated strata, enabling efficient regression estimation of structural nested mean model (SNMM) parameters without requiring restrictive treatment assignment assumptions. The method is illustrated using a medical study of 1,067 stomach cancer patients, examining how diagnosis and treatment in large versus small hospitals affect one-year survival and how these effects vary with age. A simulation study, using treatment sequences of length T = 3 with normal, dichotomous, and Poisson outcomes, confirms unbiased estimates, nominal coverage probability, and high statistical power. Indeed, their approach matches the performance of methods that require treatment assignment conditions while offering greater flexibility for targeted analysis and cross-time SNMM constraints. Collectively, these four articles showcase the ability of mathematical modeling to increase understanding of complex cancer dynamics and its potential to identify optimal therapeutic strategies. From pattern formation in virotherapy and surrogate modeling of immune interactions to evolutionary principles in pediatric tumors and statistical methods for sequential treatments, this collection advances our quantitative understanding of cancer therapy. The methodological diversity, encompassing bifurcation analysis and amplitude equation reduction, equation learning, hybrid stochastic branching processes, and causal inference for longitudinal data, reflects the multifaceted nature of cancer research and the necessity of interdisciplinary approaches.Each contribution identifies key parameters governing treatment outcomes: the asynchrony parameter that controls viral spread patterns, the competition value and initial immune cell dosage required to reduce cancer to manageable levels, the stabilizing strength and diffusion scale governing tumor evolutionary plasticity, and the blip effects of sequential treatment decisions modifiable by patient age. As these
Malinzi et al. (Thu,) studied this question.