Harshbir Singh Sandhu1, Amy Kent1, Tom Snowden1, Viji Chelliah1, Suruchi Bakshi1,2, Piet H. van der Graaf1,2,3
1Certara QSP, 2Systems Pharmacology and Pharmacy, LACDR, Leiden University , 3Cincinnati Children’s Hospital Medical Center
Introduction: The increase in recent regulatory approvals demonstrate the growing contribution of gene therapies. Arguably, the role of model-informed drug discovery and development (MID3) in this area is lagging behind. In most cases, “gene therapy” implies that a therapeutic gene is packaged into a virus vector, typically adeno-associated virus (AAV) vector [1]. This is a complex modality with several steps between dosing and effect including pharmacokinetic (PK)/biodistribution of the virus vector, viral transduction (i.e. infection of host cells with virus leading to consequent intracellular processes from virus uncoating to stable DNA transfection), transgene expression, followed by secretion and biodistribution of the gene product. Given the complexity of this modality as well as the stable effect post a single dose requires modelling approaches different to the traditional PK/pharmacodynamics (PD). Furthermore, it was shown that typical allometric interspecies scaling approaches do not translate well for this modality [2-6]. Therefore, it has been argued that mechanistic modelling approaches are required to support translation of AAV gene therapy dose-response [3,5,7,8]. Objectives: We developed a generic framework for mechanistic modelling of AAV gene therapeutics in order to capture their non-trivial PK/PD behaviour, and to support translation of dose-response across species. Methods: Our framework consists of three distinct modules 1) AAV PK/biodistribution, 2) AAV viral transduction model, and lastly, 3) transgene product biodistribution model. A fourth module linking expressed protein to disease pathway can be added where necessary. The model is based on ordinary differential equations. The first module describes PK-like distribution and clearance of the administered viral particles. The AAV viral transduction model involves virus internalization and downstream intracellular processes, which have been described previously [9-14]. The stable transduction is linked to transcription and protein synthesis from the transgene. Secretion of the synthesized protein and its downstream biodistribution is modelled in module 3. The model is largely parameterized using physiological parameters and mechanistic knowledge around viral transduction pathway. Vector and transgene-specific model parameters may be tuned using preclinical data, where available. Literature data on AAV PK and protein therapeutics and their biodistribution can be leveraged for parameterizing respective modules. Interspecies scaling approaches are applied to each module separately to arrive at the final scaled model. We demonstrated the application of our framework using a case study based on data pooled from various AAV gene therapeutic examples. Results: The modular framework allows us to tailor the generic model to each bespoke AAV gene therapy program in an efficient manner. Relevant tissues from standpoint of viral transduction, transgene product biodistribution and data sampling can be added as necessary. The individual modules are calibrate using various sources of preclinical data, including in vivo preclinical studies measuring AAV biodistribution/transduction and protein expression. Interspecies scaling approaches are applied to individual modules reducing uncertainty in scaling of the overall dose-response. Literature data is leveraged to support scaling of individual modules. We applied our framework to an anonymized case study based on data from multiple species including mouse, rat, non-human primates and human. In this example, AAV PK is described using compartmental modelling and is scaled using standard allometric approaches and the predictions are validated using PK data from multiple species. Protein biodistribution (of transgene product) in preclinical species is parameterized using exposure data in plasma and various tissues after administration of recombinant protein therapeutic to mice. This module is scaled using standard mechanistic/physiology-based approaches and the predictions validated using available clinical data using the recombinant protein therapeutic. Based on multispecies preclinical data we found that the mechanistic links between the viral transduction and transgene protein expression do not scale adequately using standard allometric approaches and a new power-law-based approach with an exponent of -2 was necessary to translate this module. The combined scaled model including all the modules was used to predict human dose-response relationship from preclinical data. Using this modular approach, we were able to isolate the processes where traditional scaling approaches do not apply in AAV gene therapy, and we used this insight to scale individual modules appropriately. This approach allowed us to reduce the parameter uncertainty involved in scaling the dose-response for this modality and helped identify key experimental results useful for translation. Lastly, the modular framework is agnostic to the type of AAV vector, the transgene, tissues of relevance and therapeutic indication, and can be tailored to any preclinical species being used in the drug development. Conclusions: We have developed a modular framework to support development of AAV gene therapies. This framework allows for describing the complex PK/PD relationships and translation of dose-response for this modality. Further, it can be tailored to each AAV gene therapy program and allows us to draw on literature data/knowledge to reduce uncertainty in scaling of dose-response for better prediction of efficacious human doses. Due to the transgene/therapeutic area agnostic nature of this framework, this can be used as a platform to support development of a diverse portfolio of AAV-based gene therapies. Our model framework has already successfully supported FIH starting dose justifications for IND submissions.
[1] Wang, J., Gessler, D., Zhan, W. et al., (2024) Adeno-associated virus as a delivery vector for gene therapy of human diseases. Signal Transduction and Targeted Therapy, 9(78). https://doi.org/10.1038/s41392-024-01780-w [2] Tang, F., Wong, H., & Ng, C. M. (2021). Rational clinical dose selection of adeno-associated virus-mediated gene therapy based on allometric principles. Clinical Pharmacology & Therapeutics, 110(4), 803–807. https://doi.org/10.1002/cpt.2269 [3] Zou, P. (2023). Interspecies normalization of dose-response relationship for adeno-associated virus-mediated haemophilia gene therapy—Application to human dose prediction. British Journal of Clinical Pharmacology, 89(4), 1393–1401. https://doi.org/10.1111/bcp.15597 [4] Burr, A., Erickson, P., Bento, R., Shama, K., Roth, C., & Parekkadan, B. (2022). Allometric-like scaling of AAV gene therapy for systemic protein delivery. Molecular Therapy Methods & Clinical Development, 27, 368–379. https://doi.org/10.1016/j.omtm.2022.10.011 [5] Aksenov, S., Roberts, J. C., Mugundu, G., Mueller, K. T., Bhattacharya, I., & Tortorici, M. A. (2021). Current and next steps toward prediction of human dose for gene therapy using translational dose-response studies. Clinical Pharmacology & Therapeutics, 110(5), 1176–1179. https://doi.org/10.1002/cpt.2374 [6] Zhang, T., & Zou, P. (2023). Interspecies scaling of transgene products for viral vector gene therapies: Method assessment using data from eleven viral vectors. AAPS Journal, 25(101). https://doi.org/10.1208/s12248-023-00867-6 [7] Mitra, A., Ahmed, M.A., Krishna, R., Sun, K., Gibbons, F.D., Campagne, O., Rayad, N., Roman, Y.M., Albusaysi, S., Burian, M. and Younis, I.R. (2023), Model-Informed Approaches and Innovative Clinical Trial Design for Adeno-Associated Viral Vector-Based Gene Therapy Product Development: A White Paper. Clin Pharmacol Ther, 114: 515-529. https://doi.org/10.1002/cpt.2972 [8] Rao, S., Narula, J., Ko, G., Luo, H., Zhang, Z., Musante, C., & Tania, N. (2023). Quantitative systems pharmacology modeling of adeno-associated virus gene therapies: Mechanistic identification and species translation using preclinical and clinical data. In Quantitative Systems Pharmacology (pp. 61–86). Taylor & Francis. https://doi.org/10.1201/9781003230342-4 [9] https://2010.igem.org/Team:Freiburg_Bioware/Modeling/Virus_Infection [10] Kamiya, H., Akita, H., & Harashima, H. (2003). Pharmacokinetic and pharmacodynamic considerations in gene therapy. Drug Discovery Today, 8(21), 990–996. https://doi.org/10.1016/s1359-6446(03)02889-7 [11] Lagache, T., Danos, O., & Holcman, D. (2012). Modeling the step of endosomal escape during cell infection by a nonenveloped virus. Biophysical Journal, 102(5), 980–989. https://doi.org/10.1016/j.bpj.2011.12.037 [12] Varga, C. M., Hong, K., & Lauffenburger, D. A. (2001). Quantitative analysis of synthetic gene delivery vector design properties. Molecular Therapy, 4(5), 438–446. https://doi.org/10.1006/mthe.2001.0475 [13] Varga, C. M., Tedford, N. C., Thomas, M., Klibanov, A. M., Griffith, L. G., & Lauffenburger, D. A. (2005). Quantitative comparison of polyethylenimine formulations and adenoviral vectors in terms of intracellular gene delivery processes. Gene Therapy, 12(13), 1023–1032. https://doi.org/10.1038/sj.gt.3302495 [14] Choules, M. P., Bonate, P. L., Heo, N., & Weddell, J. (2024). Prospective approaches to gene therapy computational modeling—Spotlight on viral gene therapy. Journal of Pharmacokinetics and Pharmacodynamics, 51(5), 399–416. https://doi.org/10.1007/s10928-023-09889-1
Reference: PAGE 33 (2025) Abstr 11662 [www.page-meeting.org/?abstract=11662]
Poster: Drug/Disease Modelling - Other Topics