Enrico Mastrostefano(1), Alessandro Ravoni(1), Elia Onofri(1), Paolo Tieri(1) and Filippo Castiglione(2)
(1) Istituto per le Applicazioni del Calcolo
Introductions: The immune system’s significance in Tuberculosis (TB) drug development is often underestimated due to experimental complexities. In this study, we use an in silico approach to understand how the immune system influences therapy effectiveness in managing the infection and reducing relapse risk.
Objectives:
- Incorporate pharmacokinetic-pharmacodynamic (PKPD) modeling into an Agent-Based Immune System Simulator.
- Simulate Mycobacterium Tuberculosis (Mtb) infection and therapy on a large virtual patient population to represent real TB outcomes.
- Qualitatively and quantitatively estimate the contribution of the immune system to the clearance and/or resurgence of the disease in the presence of therapy.
Methods: We employed a validated agent-based Immune System (IS) simulator, C-IMMSIM [1], to simulate long-term outcomes of TB over 50 years for a virtual cohort infected with Mtb. C-IMMSIM utilizes a 3D stochastic cellular automaton to model interactions between lymphoid and myeloid lineage cells during the IS’s response to the pathogen. The in silico experiments involve four steps: generating pathological and immunological trajectories, calibrating the simulator to replicate a realistic epidemiological distribution, selecting virtual patients with active disease, and administering standard anti-TB therapy over 50 years. The therapy includes rifampin (RIF), isoniazid (INH), pyrazinamide (PZA), and ethambutol (EMB) in specified plasma concentrations to mimic real treatment doses [2].
Results: Our simulations show that following drug therapy, patients successfully cleared tuberculosis bacteria, except for dormant ones, aligning with real-world trials and previous studies. While non-granuloma bacteria were eliminated within 60 to 120 days, dormant bacteria could sporadically reactivate, causing new infections. Post-therapy outcomes depend on the patient’s immune response, with some experiencing ongoing tuberculosis despite completing treatment. However, short-term relapses, affecting around 4% of patients [3], were not observed in our simulations, posing a challenge in predicting and understanding their occurrence. Weaknesses in the IS are suggested to play a role in these relapses [4]. To delve deeper, we simulated compromised IS in the final months of therapy. Around 20% of patients experienced disease reactivation, leading to death without further treatment. We found two main scenarios: one where standard therapy failed to clear tuberculosis bacteria, causing a rapid increase after treatment, and another where dormant bacteria remained, potentially leading to fatal infections. These findings emphasize the crucial role of the IS post-treatment. A strong IS prevents relapse, but deficiencies increase the risk of relapse and death. Ongoing research aims to understand this relationship, focusing on predictors like cytokine concentrations.
Conclusions: We presented initial findings from our study using a computational approach which integrates an Agent-Based Simulator and PKPD modeling to understand how the IS influences tuberculosis treatment outcomes. Our model reproduced realistic disease outcomes and highlighted that IS deficiencies lead to short-term relapse and death. Our results suggest that computational methods can advance tuberculosis understanding and inform drug development and treatment strategies.
References: [1] Rapin N et al. Computational Immunology Meets Bioinformatics: The Use of Prediction Tools for Molecular Binding in the Simulation of the Immune System. PLOS ONE. 2010 Apr 28;5(4):e9862. [2] Mitchison DA et al. Assessment of the efficacy of new anti-tuberculosis drugs. Open Infect Dis J. 2008;2:59. [3] Menzies D, et al. Effect of duration and intermittency of rifampin on tuberculosis treatment outcomes: a systematic review and meta-analysis. PLoS Med. 2009;6(9):e1000146. [4] Cadena AM et al. Heterogeneity in tuberculosis. Nat Rev Immunol. 2017;17(11):691-702.
Reference: PAGE 32 (2024) Abstr 11269 [www.page-meeting.org/?abstract=11269]
Poster: Methodology - New Modelling Approaches