1 Department of Cellular, Computational and Integrative Biology (CIBIO), University of Trento, Trento, Italy
2 Fondazione the Microsoft Research—University of Trento Centre for Computational and Systems Biology (COSBI), Rovereto, Italy
3 Department of Mathematics, University of Trento, Trento, Italy
4 Department of Information Engineering and Computer Science, University of Trento, Trento, Italy
Background: Messenger RNA (mRNA)-based therapeutics offer scalable, cost-effective platforms for delivering vaccines and protein therapeutics. However, unlike conventional drugs, mRNA therapeutics require cellular uptake and translation before exerting their effects. This multi-step cascade benefits from QSP or, in general, mechanistic descriptions capturing delivery vehicle dynamics, intracellular trafficking, and translation kinetics[1]. We developed a platform of two complementary models supporting mRNA-based vaccines and monoclonal antibodies (mAbs), providing a unified modelling solution for respectively passive and active mRNA immunotherapy.
Methods: We present a QSP model describing the events following the administration of mRNA vaccines, from product injection to the appearance of antibodies in the blood. It consists of a molecular layer describing LNP uptake, mRNA translation, and antigen presentation, and a tissue layer encompassing the injection site, draining lymph node, and blood. This multiscale framework mechanistically links vaccine-specific parameters with system-level immune dynamics, including dendritic cell maturation, T cell activation, and B cell affinity maturation. The model was calibrated and validated using BNT162b2[2-4] and mRNA-1273[4] clinical trial data. Completing the platform, we developed a multiscale physiologically based pharmacokinetic (PBPK) model for mRNA-encoded mAbs. It extends an established framework for the trafficking of recombinant mAbs, while also incorporating mechanistic liver equations for LNP-mRNA metabolism. It supports three preclinical products (RiboMab02.1[10], Pembrolizumab[11], B7H3×CD3[12]) spanning 55-140kDa and varying FcRn affinity.
Results: The vaccine model predicted antibody dynamics across 10-100μg doses and multiple schedules[2,5-7]. Model simulations identified optimal second-dose timing ensuring continuous protection. Age-stratified analysis quantified slower B cell maturation in elderly populations, while cross-platform validation with Moderna mRNA-1273[8-9] demonstrated adaptability. The PBPK model accurately reproduced plasma concentration-time profiles for both mRNA-encoded and recombinant formats across the three calibrated products. Multi-dose validation in mice confirmed predictive accuracy across dosing schedules. Successful validation in NHP for RiboMab02.1[10] demonstrated cross-species translatability. The mechanistic liver layer enabled quantitative tracking of LNP-mRNA metabolism.
Conclusions: This platform establishes frameworks for rational design and optimization of mRNA therapeutics. Modularity supports cross-species scaling and product-specific features. The vaccine model's ability to identify age-related immune deficiencies and optimal dosing windows demonstrates utility for stratified population approaches, while the PBPK platform positions as a versatile tool for accelerating mRNA therapeutic development, thanks to its adaptability to different therapeutics.