Jeanette Doerr (JD): In the last two years, we saw increased demand globally for therapeutics, and the timelines for their development and commercialization were extremely accelerated. That is going to be the norm now that the industry’s ability to meet those short timeframes has been established.
Haiou Yang (HY): Regulatory agencies want to see a more in-depth understanding of the process. Rather than simple linear evaluations, we want to know about complete operational zones, and, rather than individual parameters alone, we want an understanding of how those parameters affect one another. To get all that information requires more experiments using design of experiment (DoE) and quality-by-design (QbD) approaches. High-throughput solutions are absolutely essential to increase the number of experiments that can be completed in shorter and shorter periods of time.
HY: Typically, processes have been developed at Avid Bioservices in approximately six months, much shorter if a platform method can be applied. That has now shrunk to three months. The key is to cut the process development timeline in half without negatively impacting quality. That requires a faster turnaround, which can be addressed with high-throughput (HTP) solutions, because they allow the evaluation of many ideas much more quickly. An HTP approach is also beneficial when there is limited starting material to work with, which is typically the case for biologics.
JD: Upstream process development is actually really labor-intensive and time-consuming. At Avid, we have 30 benchtop bioreactors, so we can do large sets of experiments, but it is very resource-intensive. Leveraging automation, such as the Ambr® 250 system from Sartorius, we can run 24 bioreactors with a maximum working volume of 250 mL simultaneously without intense technician involvement and with reduced risk of human error. That helps us achieve our goal of generating more data with less resources in less time, leading to faster process development.
JD: It really depends on where we’re starting. If we’re starting pre-IND, then we would normally do several rounds of 10–12 bioreactors in a DOE format to really look at the design space of the process and evaluate and optimize parameter settings.
For clients wanting to enter the clinic quicker, a risk-based approach might be used in which Avid performs one set of runs and then performs a verification run and a pilot run, followed by tech transfer to manufacturing. In this case, a process that is phase appropriate is used to get into the clinic, and then we circle back to do more optimization during phase I trials.
For phase III projects, the work often involves scale-down model qualification and process characterization. That is also very intensive and involves running numerous bioreactors simultaneously to obtain the necessary depth of process knowledge.
HY: There are several ways. One is to first understand the molecules involved in the process using commercially available modeling software. For instance, before developing a chromatography process, modeling can be used to learn about the physiological and biochemical properties, such as the hydrophobicity/hydrophilicity, of the molecules involved in the process and thus figure out how they will interact with different types of resins. This understanding is generally lacking right now at the process development stage and could really help accelerate column selection.
As another example, the current approach to process development involves a significant amount of trial and error with regard to determining the optimum process equipment, materials, and conditions. When a good result is obtained, we move forward without conducting any further tests. Because there isn’t time to explore other materials and conditions, we don’t fully understand the variability of the process. A high-throughput approach gives you the ability to fully explore the design space of processes.
There is also software that can help accelerate process development by evaluating how processes will perform in existing larger-scale commercial equipment using process-development data. Potential issues with performance can be identified earlier and guide the development of processes optimized for that equipment.
High-throughput methods can also be used to accelerate process scale-up. Using chromatography as an example again, it’s easier to scale up using an HTP method evaluating different column heights and diameters to determine the optimal plate numbers for a given purification.
JD: We currently use JMP® data analysis software from the SAS Institute to analyze our DOE data and are evaluating software from another company that would provide additional capabilities and work with both the Sartorius and JMP solutions.
HY: We are definitely expecting that data analysis will become a real bottleneck that must be addressed. The goal with the evaluation of this additional software is to identify a solution that will avoid that bottleneck — ideally within the coming year.
JD: Yes. Along with the upstream and downstream investment at Avid in high-throughput technology, we’ve also invested in the analytical development department with the purchase of a Sartorius Octet HTX, some Waters UPLCs, a Maurice Bi-Protein Simple, and a Hamilton Liquid Handler. We have had to increase throughput in the analytics group in order to avoid bottlenecks there. Luckily, we have been able to invest in upstream and downstream process development and analytical development hand-in-hand.
JD: Upstream, we are also adopting the Sartorius Univessel® SU 2L single-use bench-scale bioreactor for transferring from the Ambr®vessels to bench scale for verification.
HY: Downstream, we’re thinking about implementing a new high-capacity membrane chromatography technology from Cytiva with the potential to speed up this important unit operation — not just for development, but in manufacturing as well. We are also considering the implementation of more inline solutions in the process development lab.
In general, we are always looking for innovative technologies to assess in process development, for use in the lab and/or manufacturing. Avid has been very open-minded in terms of leveraging changes and advances in technology, with the company leadership generally excited for us to pursue new opportunities and transfer them to the plant where appropriate.
JD: Avid is expanding its clinical and commercial manufacturing capabilities. All clients enter Avid through process development before tech transfer to ensure their processes fit our facility and any gaps are addressed. Therefore, to expand commercial manufacturing, we had to invest in process development to ensure that we can keep pace.
JD: Two big things that come to my mind. We are very creative and we always make things work. No matter what the gap is or what the issue may be, we always seem to find a solution that works. I’ve been at Avid for 17 years, and that creative culture has been one of the key reasons for me staying at the company. At the end of the day, we get the process and the manufacturing right, no matter what it takes.
HY: I totally agree that creativity is our key asset, and we’re also problem solvers. Whenever we have any problem, we feel like we have a way to get there. In addition, we’re very transparent with the client. We go through every step that we have done and the thought process, how we see the problem, and what we expect the results to be. As a result, clients are informed all the time about the developments, which makes them more comfortable. I think this communication is very important.
JD: Avid has made a sizeable investment into improving our process and analytical development capabilities so that we can be efficient and deliver robust, reproducible processes in a short amount of time and really be competitive with other CDMOs. We have not only added new equipment and instruments but reconstructed the labs to accommodate them and expanded our headcount. In fact, we still have some openings to fill.
JD: For upstream, that would be modeling solutions for scale-up. Going from bench scale in the process development lab to manufacturing scale without having any issues or impacts on product quality is rare. That is the biggest gap, but fortunately it is being filled in by companies such as Sartorius and Thermo Fisher Scientific. Running bioreactors in both the process development lab and manufacturing helps enable modeling to predict performance at larger scale. Avid is working to gather data at the small and large scales to develop and optimize our own models.
HY: I would add to that the concept of continuous processing, which, although it requires a huge investment, represents the future of bioprocessing. The idea is for material to flow from the bioreactor through the different downstream purification steps on a continual basis over an extended period of time.
JD: For upstream process development, we are still constrained by how fast cells grow — generally a 24-hour doubling time. There is only so much of the timeline that can be accelerated beyond that point. If there is a way to get the data faster, that would be great.
HY: I always feel like there’s room to improve. Process development work today is mostly done in batch mode. We analyze offline samples in order to determine what to do next. That issue is being addressed with inline process analytical technology to enable real-time monitoring in manufacturing. For process development, however, we don’t have those inline analytical capabilities. If such small inline analytical methods were available, we could potentially integrate process development and analytical and thus speed things up even further.
Originally published on PharmasAlmanac.com on November 9, 2021.