How Biogen Integrated Process Analytics Technology With End-To-End Supply Chain Transparency
By Dave Kolwyck, Biogen
The inherent unpredictability of biologic drug development continues to present the industry with new challenges when it comes to ensuring consistent process performance. As next-generation manufacturing platforms become implemented in biomanufacturing networks, real-time inline monitoring will be a critical element of maintaining control in bioprocessing. When we understand more about material attributes, we can then use that information to drive appropriate and dependable critical process control requirements.
It is for these reasons that Biogen’s raw material characterization team decided to explore several novel strategies that could potentially increase their process and raw material control and optimize the communication of data throughout the supply chain. As a result, they were able to better understand sources of variation in their raw materials, thus enhancing and streamlining Biogen’s manufacturing process control strategy.
Media Solution Preparation
While Biogen buys most of its raw materials as solids, they must be converted into liquid in order to grow cells for protein therapeutic production. Often, there can be sources of variability during the dissolution process, depending on how easily those chemicals dissolve when converted into a liquid state. This process is traditionally monitored using an endpoint system of either pH, conductivity, or osmolarity as a measure of dissolution. However, that is not ideal. Many components in media solution preparation are pH neutral or are not strongly dissociative or ionic and, therefore, have a weak conductivity or osmolarity signal. Because of this, variance in those raw materials cannot be accurately measured by these metrics.
For Biogen to understand the impact of raw material variability on dissolution, detect process variance, and help ensure media suitability and consistency, a cost-effective, real-time measurement method was needed. The application goals included:
- a qualitative method to improve process understanding and control while confirming media suitability and consistency
- operation with a simple sensor that is fully automated, rugged, and capable of direct, offline contingency
- low upfront cost and LTCO
- repeatable, reproducible detection of major solid components and potential abnormalities that are specific to each media
While several technologies were considered, such as a more sensitive pH reader, refractometers, and turbidity, Biogen eventually decided upon a refractive index (RI). RI measures how light propagates through a specified medium and is used in the food and chemical industries to measure the consistency of formulations. It is designed for tank-mounted and stream applications using both flush and immersion sensors with a wide range of probe diameters. RI offers a high degree of sensitivity with the ability to detect the refractive index shift from individual component additions like salts, amino acids, and base media as each one is added, allowing the trending of complex hydrations over multiple batches. Figure 1 shows examples of target media that exhibited repeatable and reproducible profiles using inline RI.
Figure 2 shows a comparison between an inline RI profile and an inline conductivity profile after 15 different component additions, showing that RI is more sensitive for detecting individual components.
The driver for this change is not just gaining a better image of what dissolution looks like. With titers increasing, cells are now in high-density environments and need more nutrients, so feeds have become much more concentrated, creating potential issues with what can and cannot pass through sterile filters during hydration. With a more detailed profile, Biogen could identify potential issues before a batch is added to the production process when more time and money have been invested.
Comparison Between Supplier Materials
The team also used RI to compare raw materials between different suppliers. While all of the suppliers were using pin mill technology to make their materials, they are produced at different sites using different settings, different flow rates, etc., and particle size can have an impact on dissolution rate. If the particles are too small, the media can take longer to go into the solution. If they are too big, the volume ratio is different, again affecting the length of time it takes for dissolution. In its analysis of multiple suppliers of the same material, Biogen found that the particle size distribution of the material from Supplier 1 was different than that of Suppliers 2 and 3. (Supplier 1 material appeared to have a larger particle size.) Figure 3 shows this comparison.
When they noticed this difference, they were then able to link that to the attributes of those raw materials and determine that Supplier 1’s material RI value stabilized approximately 1 to 2 minutes earlier than the other supplier materials. In the end, though, the final RI value was comparable (see Figure 4), indicating that media samples reached a similar final dissolution state.
As information such as this is collected from suppliers, it creates an opportunity to pull together raw material data from various sources in one organization that would otherwise operate in siloes. It can then be put into a platform that can be accessed across a network, offering a flexible, visual representation of genealogy on demand from any starting point.
Identifying The Genealogy Of Raw Materials
Genealogy and processing data from a supplier can be used to understand the differences in raw materials and link the inputs with that raw material data to process outputs. The challenge, though, is developing a common syntax across all systems. For example, if one database labels a raw material “CDM-6” but in another database it is labeled Media-6, a database integrator will not be able to recognize it is the same material. To address this, Biogen has begun to take steps to create an integrated system where that information can be collected and controlled more effectively with standardized labeling syntax.
First, Biogen has started to collect information from their suppliers to link to their local database and then use logic in data analytics software to identify supplier lots and even sub-supplier lots. This allows them to tie information together to trace raw material performance and use the platform and analytics to identify any trends. So far, the team has built a library of about 1,400 raw material samples and their attributes in order to understand what the normal range of variations is for a given raw material. When a process is being developed, they can then use samples that represent the high and low ends of that range for a variety of attributes, run them in a process as it is being developed, and determine if the process can handle that range of variation. If not, either the process needs to be adjusted to accommodate the normal range of raw material variation or, if this is not possible, a custom specification range for that critical raw material attribute needs to be agreed to with the supplier(s).
This technology has already been leveraged at Biogen’s site in Research Triangle Park in an investigation related to product quality. The team suspected the problem was a potential raw material issue, but after reviewing the raw material lot trend data and investigating further, they were able to determine an operational deviation resulted in extra material being added to the batch. Without a proactive and integrated raw material characterization system, the approach to a resolution would have been to send raw material lots out for analysis, likely taking weeks to complete instead of days. The goal for this system is to get to a continuous state where Biogen can predict the process performance of any given raw material ahead of its use in development.
In the end, manufacturing consistency is a function of both process and raw material variation control and transparency of data from supply chain partners. Not only do Biogen’s efforts to understand and control sources of variation from materials mitigate adverse effects on its manufacturing process, but they also become a valuable tool in ensuring a consistent supply of safe and effective drugs to their patients.