Keys to successful process development transformation: Proven strategies from industry leaders

It’s a brave new world for process development teams in the biotech industry. R&D pipelines once dominated by small molecules and monoclonal antibodies now include a diverse portfolio of next-generation biologics. Accelerated regulatory pathways are cutting the time that advanced medicinal products are spending in development. A flood of data science innovations such as generative AI, large language models, and digital twins have created opportunities to increase efficiency and deliver better therapeutics for patients. Process development teams need to evolve their approaches to stay at the cutting edge: everything from the technology they use, the data architecture they build around, and the way they empower scientists to do their best work. 

We had the opportunity to sit down with three industry leaders at Benchtalk 2024 on a panel entitled “Driving effective transformation across process development” to learn more about what’s motivating these changes and what has been effective so far. Here are the top takeaways. 

Our speakers included (from left to right): Manasi Nawathe (Development Associate, MacroGenics), Joel Nichols (Senior Vice President, Digital Systems, Tessera Therapeutics), Gang Xue, Ph.D. (Senior Director, Data Integration and Predictive Technologies, Johnson & Johnson), and moderated by Michael Schwartz (Benchling). 

Process development gets a makeover (and a wake-up call)

Process development teams carry the role of turning innovative discoveries into products that can be reliably produced to meet performance, safety, and techno-economic requirements. While these goals remain the same, the environment they happen in has changed considerably in the past five years. The types of therapeutic products that process teams work on has been steadily shifting towards biologics, hitting a peak of 61% of all novel drug launches in 2022.¹

Having these complex therapeutics in your process development portfolio can “make things very interesting,” according to Joel Nichols, whose company Tessera Therapeutics is at the forefront of gene editing. “Some things still feel the same as when I started out developing small molecules. But so much has changed. The tailwinds and the headwinds of the science itself create a very dynamic environment. It’s exciting and challenging at the same time.”

The enormous potential for these advanced therapeutics has been accompanied by strong demand to move these products into the clinic faster. “I think COVID was a wake-up call to the entire industry,” remarked Gang Xue, Ph.D. from J&J. “We cannot just tread through the old way of doing things, with years to deliver one molecule. Process development and CMC needs to provide clinical supply to trials much faster now.” This also means being able to work side by side with regulators, demonstrating understanding of how these novel processes perform. According to one study, 70% of all biologics are now using at least one of the FDA’s expedited development and review programs.²

This means that process development and CMC teams need to have well-documented, well-understood processes that they can readily communicate internally and externally, and earlier in the overall lifecycle than in prior generations of therapeutic development. Accomplishing this requires the ability to capture structured data and metadata throughout process development, leading Xue to contemplate “How can we leverage the exciting developments in the technology space to drive a change in our digital capabilities, so we can deliver more medicines to patients in shorter time frames?”   

Let’s focus on the end goal: contextualized data to help teams reach milestones faster

“Data is the focus of everything we do within digital,” commented Nichols, who leads the digital systems function at Tessera, “and that’s what excites me about process development.” The intense focus on data is understandable. Process teams want to learn from their past experiments, avoid unnecessary work, and generate predictive insights, all towards the goal of reaching milestones more efficiently. 

Process data is complex, with a high volume of parameters, test conditions, and metadata required to contextualize the captured results. For example, “stability studies, especially ones that are going to be expiry setting for clinical supply, could run for up to 48 months,” noted Manasi Nawathe from MacroGenics, a company focused on innovative monoclonal antibody-based therapeutics for the treatment of cancer, “and there are likely multiple individuals that will oversee the study over that duration. 

There has to be consistent documentation at each time point so you can consistently repeat that at subsequent time points. At the end, you have to be able to collate all of that data across test conditions in a consistent manner to get a true sense of the molecule’s stability.”  

Xue (who is also a cofounder of the Allotrope Foundation) advised to initially think less about the systems used to capture data, and more about the right way to unify data across an organization. “We have so many siloed systems in our enterprises,” he noted, “but the system really just becomes a means to capture the knowledge. We are moving in the direction of what we call ‘model-based architecture’. We want all teams to follow the same data model as they progress through their iterative learning.” 

One advantage of a harmonized data model is that it helps different process development stakeholders collaborate around their data. This goes hand-in-hand with having broad access to the data, another historical challenge within process development. “We're realizing that if our goal is to accelerate IND timelines, we really need to have consistent, clean data and communication across different teams,” Nawathe remarked. “There may be a lot of communication within individual teams, but it becomes really hard to query data from what other groups are doing if you don’t have direct access to it. Requesting a colleague’s lab notebook that is upstairs in the quality archive is not something that’s going to be useful to me.”

With data structure, context, and accessibility in mind, you can then proceed to thinking about digital initiatives that help deliver on these aims. For companies early in their digital journey, this might entail retiring paper notebooks in favor of a structured digital notebook and registration system. Other companies might think more about consolidating redundant systems and building capabilities to support structured data capture across core process development functions. 

“One of the things we prioritized was having an experiment registration system that used a consistent ontology,” Nichols shared. “My organization is also in charge of automation, so we’re always looking for ways to scale our highest throughput operations such as amplicon sequencing, and even less exciting but still important things like label printing. All of this helps us drive effective data management.”

AI is going to help us with all of this, right?

The rapid surge of generative AI is poised to make an impact in biopharmaceutical R&D. A recent McKinsey report predicts that generative AI will produce $28-53B in economic impact across the R&D phase of the pharma value chain.³ Xue noted, “We think of our innovation wheel as the ability to design, execute, and analyze. Our digital transformation strategy is focused on how we can reduce the initial design time or iterate faster through fewer cycles of this wheel before we can ultimately deliver clinical supply into patients. The more realistic expectation of what Gen AI and ML can do for us right now is not to replace these steps, but to make us more efficient and faster.”

The same McKinsey report also cautioned that Gen AI cannot deliver results unless a proper data architecture is in place. “This is one of the key challenges we have tried to address in the past few years,” Xue shared, “The first step is how we democratize the data, making it FAIR, and ensuring contextualization. This is necessary before we can feed our data into machine learning and other AI tools.” Nichols agreed that the value of this kind of transformation, inclusive of AI, has enormous potential. “There is potential impact to the scientific aspect of our work, such as process and parameter optimization. And there is a productivity aspect, helping our teams complete their work more efficiently. There’s clearly a link there, where each can catalyze change for the other.”

When it comes to change management, the carrots might work better than the sticks

All of these transformation initiatives, from democratizing data access to adopting new Gen AI capabilities, often necessitate some measure of change management. While there are no ‘one size fits all’ models, our panel shared their reflections on how to bring teams forward together on new platforms. A great place to start is simply listening and understanding how teams currently operate. “I think the most important thing I’ve learned is to talk to everyone,” Nawathe noted. “Especially end-users who might not be comfortable using new technology or software applications. It really helps to understand what their pain points are and what issues they are running into, and explain things in ways they are already familiar with.” 

Knowing that there will be a range of expectations and emotions across teams, it helps to develop advocates for any new initiative. Having people well-versed in the technology is helpful, aka ‘super-users’, but even better is having people that can truly evangelize for the outcomes the business is looking for. “We’ve started a program we’re calling ‘digital champions’,” Nichols shared. “These are people who are natural leaders and promoters at all levels of the organization. They can articulate the initiatives that are coming and show others how things are easier when you do it this way.”

This of course requires that these initiatives actually deliver end-value for users. Sharing these ‘carrots’ is key to driving change and adoption. Xue offered an example, “Perhaps you can show a scientist that they can reduce the amount of manual entry needed into their notebook. They can just scan a sample ID and all the information flows in automatically. These are the types of carrots that help the end-users get on board with the digitization journey.”

Xue clarified that digitizing processes is one thing, but full digital transformation is a broader set of changes designed to drive true evolution of the business. “This really cannot happen without executive sponsorship.” So what helps drive this kind of alignment at the top? Xue suggested starting with education and awareness. He offered reading suggestions to help executives come up to speed quickly on new areas such as AI. A strong business case is essential to rationalizing these broader initiatives. Nichols encouraged his executive peers to understand the true value of the data that accompanies new transformations. 

“When data is generated, there is a current value and a potential value. For example, the measurement of an analyte in a bioprocess condition may have a current value that is limited to that experiment. But in aggregate, this data has enormous potential value as it contributes to the fidelity of future experiments and as the product and process moves towards IND.”

Top takeaways

To recap our top learnings from these process development leaders…

1. The expectations on process development teams is constantly changing as the science and industry evolves. Teams should recognize when their current technology needs to evolve as well, and prioritize systems that offer adaptability and extensibility 

2. Model-based architecture is a useful way of thinking about how to standardize data across organizations as a first principle, with systems for data capture feeding into that

3. AI is poised to accelerate process development cycles, but require common data architecture and accessibility to supply model development and utilization

4. Digital initiatives require extensive understanding of end-user goals and pain points, and consider use of ‘digital champions’ to help evangelize the business impact of new approaches.

References

1. IQVIA Global Trends in R&D 2024: Activity, productivity, and enablers 

2. JAMA Netw Open. 2022 Nov; 5(11): e2239336.

3. McKinsey & Company: Generative AI in the pharmaceutical industry: Moving from hype to reality