Multiscale modelling of drug–host–pathogen interaction: quantifying drug and immune contributions to treatment response
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background and Objective
Predicting treatment outcomes in infectious diseases requires accounting for the interplay between drug effects, pathogen dynamics, and host immunity. Integrating pharmacological and immunological approaches into a single simulation environment remains a fundamental challenge in both theory and practice. We aimed to develop and validate a multiscale in silico framework coupling these processes, and to quantify their respective contributions to bacterial clearance.
Methods
We present the Drug–Host–Pathogen Interaction (DHPI) frame-work, combining three independent mechanistic components: a physiologically based pharmacokinetic model of drug disposition, a pharmacokinetic– pharmacodynamic model of drug-induced bacterial killing, and a stochastic agent-based model of the immune response. Continuous concentration profiles are time-averaged onto the agent-based time grid, assigned to bacterial phenotypic states, and converted into per-agent killing probabilities, so that drug-mediated and immune-mediated death events are recorded separately at each step. The framework was applied to simulate symptomatic pulmonary tuberculosis. Phenotype-specific drug-efficacy parameters were inferred using Approximate Bayesian Computation from historical clinical data on eight weeks of 600 mg rifampicin monotherapy, and validated against independent early bactericidal activity data over a disjoint time window.
Results
The calibrated framework reproduced the observed decline in bacterial load, and matched reported early bactericidal activity over the first week. In a virtual cohort of symptomatic patients, drug-mediated killing accounted for 81–88% and immune-mediated killing for 12–19% of total bacterial elimination over the 60-day treatment course, while the dormant, granuloma-contained fraction rose from 0.20–0.29 in the first week to 0.85– 0.89 at treatment completion. Over a follow-up of up to 50 years, patients reaching clinical cure had accumulated more memory lymphocytes during treatment than those progressing to clinical failure or death; moreover, the final outcome depended on the immune changes occurring during therapy rather than on the initial disease stage.
Conclusions
The results show that the DHPI framework can reproduce treatment dynamics observed in patients and enable the analysis of how therapy reshapes host immune responses and subsequent disease trajectories. By explicitly representing drug–host–pathogen interactions, it provides a mechanistic basis for in silico treatment simulations and for the study of long-term immune consequences of antimicrobial therapy.