Epigenetic state inheritance drivers drug-tolerant persister-induced resistance in solid tumors: A stochastic agent-based model

The efficacy of anti-cancer therapies is severely limited by the emergence of drug resistance. While genetic drivers are well-characterized, growing evidence suggests that non-genetic mechanisms, particularly those involving drug-tolerant persisters (DTPs), play a pivotal role in solid tumor relapse. To elucidate the evolutionary dynamics of DTP-induced resistance, we develop a stochastic agent-based model (ABM) of solid tumor evolution that couples macroscopic population dynamics with microscopic epigenetic state inheritance during the cell cycle. Our simulations accurately reproduce the temporal progression of relapse observed in experimental studies, capturing the dynamic transition from sensitive cells to DTPs, and ultimately to stable resistant phenotypes under prolonged therapy. By explicitly modeling the epigenetic plasticity of individual cells, our model bridges the gap between cellular heterogeneity and population-level tumor evolution. Furthermore, we performed \textit{in silico} clinical trials using virtual patient cohorts to evaluate therapeutic outcomes, demonstrating that optimized adaptive treatment strategies can significantly delay tumor relapse compared to standard dosing. This study provides a quantitative framework for dissecting DTP-driven resistance mechanisms and designing more effective, biologically informed therapeutic strategies.

💡 Research Summary

The paper addresses a critical gap in cancer resistance modeling by focusing on non‑genetic mechanisms, specifically drug‑tolerant persisters (DTPs), which are known to drive relapse in solid tumors such as non‑small cell lung cancer (NSCLC). Traditional models based on ordinary or partial differential equations treat cell populations as a few discrete compartments (sensitive, tolerant, resistant) and therefore cannot capture the continuous phenotypic drift observed in single‑cell transcriptomic data. To overcome this limitation, the authors develop a stochastic agent‑based model (ABM) that explicitly incorporates epigenetic state inheritance during cell division.

Each simulated cell carries a three‑dimensional continuous epigenetic vector x = (x₁, x₂, x₃):

• x₁ – activity of the drug‑targeted pathway (e.g., EGFR), inversely related to drug sensitivity;
• x₂ – stemness/plasticity index, high values correspond to a slow‑cycling, dormant DTP phenotype;
• x₃ – activity of alternative bypass pathways (e.g., IGF‑1R, acetylcholine signaling) that enable proliferation despite drug pressure.

The cell cycle is split into a resting G0 phase and a fixed‑duration proliferative phase (τ). Transition rates—re‑entry into the cycle (β), apoptosis during proliferation (μ), and removal/differentiation (κ)—are functions of the epigenetic state. Notably, the intrinsic proliferation rate β₀(x) is a unimodal function of x₂ (maximal at intermediate stemness), while the differentiation rate κ(x) declines monotonically with x₂, reflecting biological observations that quiescent stem‑like cells divide slowly but are less likely to differentiate.

A key innovation is the inheritance probability kernel p(x, y), a multivariate normal density that describes how a mother cell with state y produces a daughter with state x. The kernel’s variance controls the “memory loss” of epigenetic information during mitosis, allowing stochastic drift toward DTP or resistant states. This mechanism reproduces the experimentally observed sequence: (i) early drug exposure suppresses x₁, selecting for cells with high x₂ (DTPs); (ii) prolonged exposure permits stochastic increases in x₃, leading to the emergence of drug‑resistant cells (DRCs).

Parameter values are calibrated using published NSCLC PC9‑erlotinib data (initial DTP fraction, timing of relapse, marker expression). Simulations accurately recapitulate the temporal dynamics: a small DTP subpopulation appears within 1–2 weeks of treatment, remains dormant, and after 4–6 weeks the proportion of DRCs rises sharply as x₃ crosses a threshold. The model also reproduces the overall tumor burden curve observed in vitro.

To demonstrate translational relevance, the authors conduct in silico clinical trials with 200 virtual patients, comparing standard fixed‑dose EGFR‑TKI therapy to an adaptive regimen that modulates dose based on tumor size and estimated DTP fraction. The adaptive strategy—lowering dose when tumor burden is low but increasing it when DTPs exceed a preset level—extends median progression‑free survival by roughly 45 % relative to the standard arm. This result underscores the therapeutic advantage of accounting for epigenetic plasticity and persister dynamics.

The discussion acknowledges limitations: the model does not explicitly include stromal components (CAF, TAM) or spatial diffusion, and the inheritance kernel is phenomenological rather than derived from direct epigenetic measurements. Future work is suggested to integrate single‑cell omics for data‑driven kernel estimation and to develop hybrid PDE‑ABM frameworks that capture spatial heterogeneity.

In conclusion, the study provides a rigorous multiscale quantitative framework that links epigenetic state inheritance to DTP‑mediated drug resistance, validates the model against experimental relapse kinetics, and demonstrates that adaptive treatment schedules informed by persister dynamics can substantially delay tumor relapse. This work advances our mechanistic understanding of non‑genetic resistance and offers a practical computational tool for designing next‑generation cancer therapies.