Elio Campanile1,2, Elisa Pettinà 2,3, Lorena Leonardelli2, Stefano Giampiccolo2, Luca Marchetti2,3
1Department of Mathematics, University of Trento , 2Fondazione the Microsoft Research – University of Trento Centre for Computational and Systems Biology, 3Department of Cellular, Computational and Integrative Biology (CIBIO), University of Trento
Introduction/Objectives: The interest in mRNA-encoded therapeutics has surged in recent years, particularly in the context of tumor therapies. Despite promising experimental results, the understanding of the biological processes occurring from the injection of mRNA encapsulated in lipid nanoparticles (LNPs) to the production of the translated proteins remains incomplete. To address this gap, we developed a phenomenological model to simulate LNP trafficking and key mRNA reactions. By integrating it with an existing Physiologically Based Pharmacokinetic (PBPK) model for antibody transport [1], we developed a multiscale PBPK model capable of describing the biodistribution of mRNA-encoded therapeutics. This model allows us to: •Better understand the correlations among different mRNA-LNP formulations by mapping their physiochemical properties to relevant model parameters. •Determine optimal dose scheduling for both mRNA-encoded and recombinant proteins through predictive concentration profiles over time. Methods: We further developed a multiscale PBPK model previously implemented by the group, which extends the work of Sepp et al. 2019 [1], describing antibody trafficking with an additional layer to account for LNP trafficking and mRNA uptake, translation and degradation. A transit compartment model was employed to phenomenologically mimic LNP trafficking in the organism while preserving a parsimonious approach to estimate model parameters. The mRNA translation was modeled using three differential equations that describe the appearance and degradation of mRNA in hepatocytes and the production of the desired antibody in the liver, governed by an efficacy parameter. All the investigated mRNA-encoded therapeutics target tumors and undergo translation in the liver, but they vary significantly in size (55-150 kDa), presence of Fc region and interstitial space distribution. By incorporating the two-pore hypothesis and FcRn binding kinetics, our platform adapts to these protein-specific characteristics. The current version of the model was implemented in MATLAB2024b Simbiology and supports a wide range of mRNA therapeutics: B7H3xCD3 BiTE [2], RiboMab02.1 [3], and XA-1 [4] without Fc binding; Pembrolizumab [5], TrastuzumAb [6], and RiboMab01 [7] with Fc binding. Each paper provided a time series for both the recombinant and mRNA-encoded antibodies, which were used to fit the model, along with several time points from different dosages and time series from multi-dose regimens, which were used for validation. All mRNA-related parameters were also subjected to an identifiability analysis using GENSSI [8], confirming global structural identifiability. Results: Utilizing datasets from the literature, we successfully calibrated and validated our model across six different therapeutics and corresponding mRNA-LNP formulations. Further simplifications of the mRNA layer were tested, but they resulted in a loss of predictive accuracy, failing to sufficiently capture the observed data. In contrast, our model accurately captured pharmacokinetic time-series data for each drug analyzed and successfully reproduced a series of validation scenarios that were excluded from the fitting process. This finding was consistent across a range of dosages and treatment protocols, including both single- and multi-dose regimens. Across the datasets, the parameters were fitted weighted on corresponding standard deviations, resulting in favorable AIC values and simulations that reflect the variability observed in the experimental data. Our model demonstrated strong predictive performance for mRNA-encoded antibodies under various dosing schedules, proving to be an effective tool for this purpose. Moreover, the few calibrated parameters provided important insights, i.e., estimates of LNP trafficking time within the organism before reaching the liver and recurrent performance patterns among similar LNP formulations. Conclusions: We have developed a general-purpose, flexible, multiscale PBPK platform that accurately predicts the trafficking of mRNA-encoded therapeutics in mice following intravenous injection. The model accommodates different LNP and mRNA formulations and accurately predicts concentration-time profiles for various cancer therapeutics. Our platform enables quantitative predictions of mRNA-encoded therapeutic distribution across organs, particularly at the site of action, facilitating optimized dose scheduling for various formulations. Furthermore, the PBPK model presented here sets the basis for model extensions in other species. Future work will extend model predictions to animals of higher complexity (NHPs) and humans. Moreover, we plan to include different administration routes and different tropisms.
[1] Sepp A, Meno-Tetang G, Weber A, Sanderson A, Schon O, Berges A. Computer-assembled cross-species/cross-modalities two-pore physiologically based pharmacokinetic model for biologics in mice and rats. J Pharmacokinet Pharmacodyn. 2019 Aug;46(4):339-359. DOI: 10.1007/s10928-019-09640-9 [2] Huang C, Duan X, Wang J, Tian Q, Ren Y, Chen K, Zhang Z, Li Y, Feng Y, Zhong K, Wang Y, Zhou L, Guo G, Song X, Tong A. Lipid Nanoparticle Delivery System for mRNA Encoding B7H3-redirected Bispecific Antibody Displays Potent Antitumor Effects on Malignant Tumors. Adv Sci (Weinh). 2023 Jan;10(3):e2205532. DOI: 10.1002/advs.202205532 [3] Stadler CR, Ellinghaus U, Fischer L, Bähr-Mahmud H, Rao M, Lindemann C, Chaturvedi A, Scharf C, Biermann I, Hebich B, Malz A, Beresin G, Falck G, Häcker A, Houben A, Erdeljan M, Wolf K, Kullmann M, Chang P, Türeci Ö, Sahin U. Preclinical efficacy and pharmacokinetics of an RNA-encoded T cell-engaging bispecific antibody targeting human claudin 6. Sci Transl Med. 2024 May 22;16(748):eadl2720. doi: 10.1126/scitranslmed.adl2720. Epub 2024 May 22. PMID: 38776391. [4] Wu L, Wang W, Tian J, Qi C, Cai Z, Yan W, Xuan S, Shang A. Engineered mRNA-expressed bispecific antibody prevent intestinal cancer via lipid nanoparticle delivery. Bioengineered. 2021 Dec;12(2):12383-12393. doi: 10.1080/21655979.2021.2003666. PMID: 34895063; PMCID: PMC8810065. [5] Wu L, Wang W, Tian J, Qi C, Cai Z, Yan W, Xuan S, Shang A. Intravenous Delivery of RNA Encoding Anti-PD-1 Human Monoclonal Antibody for Treating Intestinal Cancer. J Cancer. 2022 Jan 1;13(2):579-588. DOI: 10.7150/jca.63991 [6] Rybakova Y, Kowalski PS, Huang Y, Gonzalez JT, Heartlein MW, DeRosa F, Delcassian D, Anderson DG. mRNA Delivery for Therapeutic Anti-HER2 Antibody Expression In Vivo. Mol Ther. 2019 Aug 7;27(8):1415-1423. doi: 10.1016/j.ymthe.2019.05.012. Epub 2019 May 18. PMID: 31160223; PMCID: PMC6698250. [7] Bähr-Mahmud H, Ellinghaus U, Stadler CR, Fischer L, Lindemann C, Chaturvedi A, Diekmann J, Wöll S, Biermann I, Hebich B, Scharf C, Siefke M, Roth AS, Rao M, Brettschneider K, Ewen EM, Sahin U, Türeci Ö. Preclinical characterization of an mRNA-encoded anti-Claudin 18.2 antibody. Oncoimmunology. 2023 Oct 16;12(1):2255041. doi: 10.1080/2162402X.2023.2255041. PMID: 37860278; PMCID: PMC10583639. [8] Chis O, Banga JR, Balsa-Canto E. GenSSI: a software toolbox for structural identifiability analysis of biological models. Bioinformatics. 2011 Sep 15;27(18):2610-1. doi: 10.1093/bioinformatics/btr431. Epub 2011 Jul 22. PMID: 21784792; PMCID: PMC3167050.
Reference: PAGE 33 (2025) Abstr 11360 [www.page-meeting.org/?abstract=11360]
Poster: Drug/Disease Modelling - Oncology