Messenger RNA (mRNA) has changed the world over the past few years with COVID-19 vaccines and products like Givlaari and Oxlumo being the obvious examples. To maintain this momentum the sector must develop more scalable production methods.
So says Zoltán Kis, PhD, a senior lecturer at the department of chemical and biological engineering at the University of Sheffield in the U.K., who told GEN there is significant scope for the development of mRNA platforms.
“Currently, mRNA vaccines and therapeutics are made in batch processes using equipment largely adopted from protein biopharmaceutical manufacturing. Some companies use bespoke production processes, while others develop platform processes, although the definition of a platform varies widely across the industry,” explained Kis.
“There are certainly opportunities to develop more broadly applicable platform technologies, for example, by using the Quality by Design (QbD) framework and by establishing a multi-product QbD design space.”
Kis and colleagues wrote about one such approach in a paper in May pointing out that, in combination with computational models, QbD can better identify and quantify the relationship between critical quality attributes (CQAs) and process parameters (CPPs).
And this is important because quality control is a major production bottleneck for the mRNA industry Kis says.
Rate limiting steps
“The rate-limiting steps are the production of template DNA and the quality control of the drug substance and the drug product,” according to Kis. “There are also challenges for establishing a multi-product QbD design space, for example, defining what is part of the platform, i.e., sequence agnostic vs what is product- or sequence-specific.”
And there are other areas of mRNA production that would benefit from platform technologies that incorporate QbD, notes Kis, who cites the use of analytical methods and the removal of product-related impurities as examples.
“PAT that enables continuous manufacturing and soft sensors, digital twins, and software solutions that allow real-time monitoring and advanced control of the product CQAs and manufacturing KPIs would be of benefit to the mRNA developers,” continued Kis.
He talked about a £2 million project his team is working on with Amodo Design–focused on the development of a fully-automated compact RNA manufacturing process–as an example of the type of innovation needed.
AI for better RNA?
Artificial intelligence also has a potential role to play in helping the RNA vaccine and therapeutics sector accelerate and streamline manufacturing. The key, Kis says, is amassing enough high-quality data to train the software systems.
“AI can play a role when sufficient data is available that is suitable for AI applications. As this is still a relatively new field, we are predominantly using models that capture the mechanisms of the underlying processes,” he explained.
“These models can guide process development by being able to simulate a wide range of scenarios-experiments, substantially saving experimental costs and time. Based on these simulations, only a reduced number of experiments need to be performed to confirm optimal conditions and reduce uncertainties. The obtained experimental data is fed back to the model to improve model predictions.”
Once available, AI models can be used to simulate a new round of scenarios, and the iterative modelling and the experimental process development cycle can continue. And this iterative approach has implications for process control.
“In the end, a high-performing process and models are obtained. The obtained models can be adapted for monitoring and automation. For example, we can use the models with real-time PAT data to predict product CQAs in the near future,” says Kis. “If the model predicts CQAs going out of spec 5–10 minutes from now, we can take corrective measures at the present time when CQAs are still in range to prevent CQAs from going out of spec.”
