Leveraging Digitalization: Insights from Biopharma CDMOs
Contract development and manufacturing organizations (CDMOs) that embrace data analytics are at a competitive advantage when it comes to fulfilling the growing demands for biologics and the push toward digitalization in the pharmaceutical industry.
This article is posted on our Science Snippets Blog.
The concepts of digital transformation and Pharma 4.0 have finally made their way into the budgets of pharmaceutical manufacturers, with ABI Research forecasting that spending on data analytics will grow by 27% annually and reach $4.5 billion by 2030. At the same time, the popularity of contract manufacturing continues to grow – providing an economical and efficient way for pharmaceutical companies to meet the increasing global demand for biotherapeutics.
When we talk about Biopharma 4.0, we’re including a number of ideas that foster the move toward digitalization, including:
- Big Data: The efficient generation, extraction, management and storage of data
- AI (Artificial Intelligence): Computer-based decision making including deep learning and machine learning
- Advanced Analytics: The dynamic visualization and modeling of data, including continuous process verification, process simulation, in-silico process development and digital twins
- IoT (internet of things): Being able to deliver information to people and systems through better connectivity of all things
- Cloud: Connecting people and devices to scale platforms and support SaaS services
- Process Analytical Technologies (PAT): Combining sensors and analytics and offering closed-loop control for processes
Pharma Companies Seeking Process Knowledge
The growing global demand for biologics and biosimilars is one of the key reasons that pharma companies continue to outsource manufacturing to CDMOs. When pharmaceutical manufacturers turn to CDMOs, they are increasingly looking for specialized process knowledge in biologics production. This presents an opportunity for CDMOs to differentiate themselves by developing process analytical technologies supported by advanced data analytics.
It is worth mentioning that in the case of cell-based therapies, the product is the process. Your ability to prepare and deliver characterization assays is critical. Product characterization helps in predicting the appropriate scale-up approaches and plays a key role in release criteria definition. The availability of product characterization data facilitates also quick technology transfer.
“Creating a strong process that is platform agnostic but is built around a data package creates value for us. This will shorten the development process because we don’t need a large number of runs to establish a baseline process – since we can use a data model built on multivariate analysis instead.”
– Kai Touw, Technical Lead at Batavia Biosciences
Panelists from BioPharma Industry Share Insights
Sartorius recently took a closer look at Biopharma 4.0 and held a panel discussion with representatives from across the biopharma industry to share and discuss the opportunities and barriers of digitalization and data analytics at CDMOs.
The panelists included:
- Graham McCreath, Senior Director Data Science at FUJIFILM Diosynth Biotechnologies
- Paul Lemoi, MSAT Data Science Manager at Lonza
- Kai Touw, Technical Lead at Batavia Biosciences
- Mark Demesmaeker, Head of Sartorius Data Analytics
Along with moderators:
- Tiffany McLeod, Life Science Market Manager, Sartorius Data Analytics, and
- Lennart Eriksson, Senior Lecturer and Principal Data Scientist at Sartorius Data Analytics
Highlights from the Panel Discussion
Our industry experts answered a number of questions about the value of data analytics and digital transformation for CDMOs.
Q: What challenges do CDMOs face in regard to digitalization projects?
One consideration, of course, is that CDMOs may have a lot of variation in the types of projects they handle and need to maintain flexibility. What are some others?
Kai: For us at Batavia, one of the challenges is that the prediction platforms for viral vectors and other new modalities are much less established when compared to conventional MAP processes. The wide variety of modalities, cell lines and viral vectors that we work with means that our prediction platforms and all the tools that come with it need to be flexible in order to work across projects. Our customer base mainly consists of small biotech, which has tremendous pressure from the market forces and their investors – they want to develop a high-quality process as quickly as possible. This is also where I see a key benefit of digitalization and data analytics application: in early-stage process development.
Mark: One challenge for CDMOs is that the depth of tech transfer can be limited. It really depends on whether the originator provides the depth of information necessary for scaling and tech transfer because this is a very data-driven process. And success here really depends on the level of information you get.
Graham: It’s not unusual for us to have a process transferred that has only been run once, versus one that has been run 25 times. There is a huge variation in the data we see. Data transfer can be an issue. We often get data transfer from clients via SharePoint or Excel spreadsheets, or even paper, which can be challenging for digitalization.
Tiffany: There is definitely an opportunity there to improve that process.
Q: What do you see as the differential needs of CDMOs vs. the broader biopharma market?
Mark: Predominantly there is a need for greater agility and speed in commercial process manufacturing and execution. Because we typically observe very tight timelines for CDMOs in order to get production up to process and to get the PD and commercial manufacturing executed. We also see an even tougher crunch on the cost of goods sold.
Having documented product robustness and critical quality attributes can be a competitive differentiator when bidding for projects. This is important for a CDMO in their value proposition to potential customers. For those CDMOs with a long track record of successful processes, having a lot of historical process development and process data from previous projects on similar molecules or similar platforms helps develop a robust QBD approach, and with winning projects and successfully executing them on time.
And last but not least, CDMOs should consider new business models. So essentially create a process as a service to their customers. We have seen this in the clinical CRO space already some 5-10 years ago where large CROs have started to share analytics with their customers so they could get a live view on clinical trials. In a very similar manner, we see evolving business models for CDMOs to share process data LIVE with their customers.
Kai: I agree with that point. I see huge potential to use data as a service in the different projects we run. We work a lot with fixed bioreactor systems, which are hard to sample, so if we can create good growth curve models from one cell line to the other, and also really leverage that from one project to another, then we can sell that to our customers as a data package. And we can shorten the development time because we don’t need an extraordinary number of runs to establish a baseline process for them. This type of data analytics modeling is creating value for us.
Q: How do you get your team onboard with digitalization? Who needs to be involved? Are there dedicated transition teams?
Kai: For us at Batavia, there is no dedicated team, but that’s also due to the size of the company. We are a small-to-mid-sized company, so we have two sides working closely together. For us, digitalization is run more as part of small-sized projects rather than a big effort toward digital transformation. All these small incremental steps have taken us a long way already.
Graham: At FUJIFILM, we have our enterprise global project management office get involved in any digitalization implementation projects. So when we first introduced our EOA into our PD labs in four locations, it was our global project management team that looked after it. That was a very large deployment. And other digitization efforts to be rolled out on a global basis would also be looked after by that team.
Mark: From a vendor perspective, we see the full spectrum: we see CDMOs that are digitally enabled in the sense that they have teams ready to take on the challenge of data-driven process management and manufacturing, and we also see those that are almost at point zero in digital transformation.
Also, as we know, staff, like data scientists and simulation and modeling experts, are quite scarce in the market and there is a huge competition for this talent. That is something that is also one of the biggest bottlenecks that we observe to the digitalization of bioprocess at the moment.
Having said that, it’s always great to see some lighthouses, for example at Lonza, Paul who is driving the digitalization in Portsmouth, and that is typically how it starts off. You need to find a team within your CDMO who lives digital transformation and understands what the value is in order to make it happen across the company.
Paul: I think Lonza has changed a lot in the last few years with regards to coming on board as a global company. One of the difficulties can be getting people on board with learning with these new toolsets. A lot of other toolsets have been in place for a long time that may be kind of clunky, but teams have grown used to them, so getting them on board with new technologies can be challenging. We have dedicated teams now to get new technologies incorporated into our portfolio. It is a process that is being developed and rolled out globally.
Q: What role does data analytics play within the digital transformation story at your respective companies?
Paul: I think the more efficient we can become at things like automated process monitoring, predictive analytics, and fault detection, the more faith the customer will have in our ability to manage processes, avoid deviations, troubleshoot, and prevent atypical problems from happening with the process before they do. Customers need to know we are looking out for their interests and will provide the highest quality product possible. Transparency is a big deal, and data analytics helps us provide that.
Kai: Data analytics also helps improve the quality of what we can deliver as a CDMO. Incorporating data analytics into the early phases of the development cycle helps us increase process understanding. And when transferring it to our customer’s late-stage manufacturing sites, there is already a thorough understanding of the control strategy needed, which will benefit the customer.
Mark: Coming back to the point of transparency, there is also a unique opportunity for CDMOs to prove their level of partnership, and even, as I mentioned before, to develop new business models around sharing process data that gives full transparency into ongoing processes. That is also a good asset for CDMOs to work on.
Q: Do you think increasing data transparency will help you secure higher value contracts with end partners?
Graham: At FUJIFILM in Copenhagen, they have been doing that for some time now. They are sharing online production data with clients using SIMCA®-online where they can show the customer what’s happing using multivariate models. The feedback they have on that has been very good. We are getting more and more requests from customers to see this.
Paul: Lonza have historically already provided access to data. That is part of our contract: to have a GMP-validated interface into a qualified, easy-to-use, self-service data analytics tool. For the future, we are making improvements to that system, and we see customers wanting a live transfer of data. Or a near real-time transfer of data. One concern that has been raised, is what if the customer spots a problem before we do?
Graham: You usually find it first. So far, our clients have good feedback. It’s more used as a tool for reporting out, so clients can see that we did follow the process. If it deviated, here is how we handled it.
Paul: Customers are very focused on efficient data management processes. They are more and more interested in the co-development of solutions, and the interest seems to be in a near-real-time transfer of information from our electronic systems to their electronic systems. They are also interested in seeing other types of data, not just process data. Information from tracking wise, documentation and management systems, SAP, so looking at how to provide all of that information together to provide value. The raw data transfer seems to be the most talked-about approach currently.
Customers from remote locations often need to have a person on the plant, so there is a need to offer plant and data systems that could potentially allow them to feel more in touch with the product without having to have someone here. That was a big issue during COVID where they couldn’t send someone there.
Q: Do you see adopting data analytics as a way to differentiate yourself from other CDMOs?
Kai: I think so. The bigger CDMOs are perhaps further down the line with this, but there is still a moment to differentiate yourself from competition. In the near future, it will be like the QBD paradigm: that it will be expected by partners that we apply some sort of data analytics in our processes.
Paul: Definitely. Customers are developing these digital and data analytics skillsets alongside us. They are becoming very knowledgeable, and we are trying to do the same.
Graham: The most important thing for the clients is that their drug substance is delivered at the right time and at the right cost. But if you can give them more data and develop a more efficient and reliable process that they are confident in, you will retain them as customers for the long term.
Mark: Whether you are an established CDMO or start-up, you are selling a service, and in this digital age, you need to wrap data analytics around your base service (which is essentially delivering a certain amount of a drug). These digital companion services are going to be very important moving forward. We see this in other industries: retail, financial, oil and gas, semiconductors – all of these have adopted digital wraparound offerings many years ago, sometimes decades ago. And if you look at what clinical CROs are doing, they started nearly 10 years ago to create digital wrap-around offerings going so far as mobile apps for patients involved in clinical trials.
CDMOs need to start thinking a little bit out of the box and to develop these new digital business models that are still quite disruptive.
Leverage Data Analytics Tools as a Differentiator
The CDMOs that will succeed in the age of digital transformation are those that do the best job of proving they are trustworthy long-term partners and provide transparency in their processes. That means adopting next-generation methods of quality management and easing partners’ concerns about loss of control and process visibility. Incorporating advanced data analytics tools can reduce these concerns and help CDMOs create processes that are more transparent, controlled and cost-effective.
Read more: How CDMOs Can Use Data Analytics as a New Source of Revenue
Listen to the full panel discussion now by watching the recording