(436h) In Vitro Transcription Platform Process Modelling to Ensure mRNA Vaccine Quality

mRNA is emerging as one of the most promising technologies in the medical field. Beyond COVID-19 vaccines, these new products are raising high hopes for new clinical application, including many unmet medical needs. Over a hundred new RNA-based vaccines have now entered the clinical pipeline. Strikingly, this technology offers the potential for modular, disease-agnostic manufacturing platforms. However, current production processes still present a high level of impurities and by-products in comparison to conventional vaccine technologies. Double-stranded RNA (dsRNA) and small RNA are considered particularly critical, as they are highly immunogenic. Disease-specific requirements, such as higher dose regimen or lower immunogenicity, could potentially prevent clinical translations. These products are also heterogenous in size and formation mechanisms, posing additional analytical challenges and downstream burdens. Furthermore, when switching from a sequence to another, significant variations in the RNA active substance quality profile have been reported. A deper process understanding appears thus essential to further optimize In Vitro Transcription (IvT) and ensure consistent product quality.

This study shows how process modelling tools can be used to enhance this knowledge and unlock RNA platform potential.

In line with Quality by Design (QbD) principles, a holistic modelling approach was implemented for multi-product IvT. Over 400 reactions and 30 RNA constructs were used to calibrate this platform process model. A system of differential algebraic equations (DAEs) was implemented and solved in Pyomo 5.7 (Python). The dynamics of more than 30 chemical species in solutions are thus captured under partial-equilibrium assumptions. This modelling framework represents the mechanistic backbone for the prediction of additional critical quality attributes (CQAs), including the formation of dsRNA and small RNAs. Then, an analytical workflow based on RNA capillary gel electrophoresis, ELISA, RP-IP-HPLC and spectrum analysis was developed to assess product quality profile at different sampling times. In addition, a screening design was implemented to identify the non-linear effects of plasmid DNA, T7 RNA polymerase (T7RNAP), magnesium, nucleotides and buffer concentrations for 12 novel RNA products. Purified mRNA molecules at different concentrations were also incubated without plasmid DNA and under different conditions to isolate templated from non-templated RNA degradation and by-products formation.

This work provides new insights into mRNA quality and a more complete view of the multiple species that make up the final IvT product. Model predictions were validated through cross validation algorithms. Such a mechanistic approach guarantees extrapolation capability with future RNA vaccine candidates and process innovations. Additionally, a multi-objective design space optimization workflow is presented. Operating regions meeting quality requirements are showcased and highlight the central role of T7RNAP activity. Eventually, a model-based strategy is introduced to effectively explore the IvT design space for regulatory purposes. The integration of a maximal covering design can provide a process validation tool, ensuring greater process flexibility and confidence within the normal operating regions.

Overall, this study provides a new model-based approach for IvT process optimization, capturing the kinetics of the formation of the most critical RNA quality attributes. It is also a step towards the deployment of versatile RNA production platforms. In the near future, the integration of sequence effects into this modelling framework could bridge the gap between process development and sequence design. It can also lay the foundations for new feeding strategies to ensure better control over process conditions, further guaranteeing the safety and efficacy of any new mRNA candidates.