September 15, 2016 (Vol. 36, No. 16)

K. John John Morrow Jr. Ph.D. President Newport Biotech

Beware of Obstructions in the Upstream Processing of Biologics

A surging confluence of upstream processing technologies is sweeping biological drug candidates toward the clinic. In recent years, cell lines, media, and disposable components have undergone marked improvement, in parallel with hardware design, improving consistency and robustness.

Advances in automated bioreactor design now provide tools for a multivariant approach with the same design as previous larger scale protocols, quickly and with substantial savings in resources.  With fewer failures and better tools for monitoring culture parameters, companies are able to eliminate unsatisfactory candidates early on, and move their best molecules forward rapidly.

These bioprocessing innovations were assessed at the recent European Upstream and Downstream Technology Forum. The event was held at Sartorius College in Göttingen, Germany.

“The knowledge base that we are developing is focused on bringing together the different technologies including molecular and cellular biology in the upstream space,” said Christel Fenge, Ph.D., vice president of marketing for fermentation technologies at Sartorius Stedim Biotech (SSB). “We are integrating the Quality by Design concept into the software solutions, using multivariate statistical tools and automated approaches to deliver the strongest improvements possible. In this fashion, the results of many studies are combined to develop a product that responds to customers’ needs.”

Data Analysis

Doug Marsh, a senior scientist for process research at GSK Pharma working on cell-line development, presented data comparing large- and small-scale culture performance for a number of cell lines using the ambr® 250 system. GSK has expanded their bioreactor capability, and the company now has 72 microscale bioreactors that can be operated by just two people.

“Using several dozen bioreactors, we can generate a huge array of data quickly from the various cell lines under investigation,” Marsh said. “The design space investigations allow us to assess the robustness and developmental potential of a given cell line.” He added that an informatics approach is used to enhance cell lines.

To handle the deluge of data generated through ambr, Marsh and his colleagues use JMP (“jump”), a computer program for statistical analysis. JMP is employed in a variety of applications including Six Sigma, quality control, experimental design, and numerous research protocols. JMP incorporates a custom scripting user interface that allows the program to accept data flows directly from the ambr software.

Marsh indicated that once he and colleagues have the data organized and in a platform in which it can be interpreted, they have the opportunity to advance and apply some of the advanced techniques of data analysis, such as a partial least squares regression.

 “We can now get insight into our data a few hours after the experiment is concluded,” asserted Marsh. “This kind of analytical capability enables us to predict several product quality outcomes prior to the cell culture being harvested.”

Media Optimization

Michael Gillmeister, senior scientist at Lonza Biologics, discussed independent case studies of a new media optimization approach that his company has developed together with Sartorius Stedim Biotech. This approach uses the ambr 15 microbioreactor system to evaluate chemically defined and custom media of nonanimal origin. According to Gillmeister, the system comes into play when commercially available media fall short of expectations and optimized custom media must be developed as rapidly as possible for high-producing cell lines.

This new line of attack provides a media-optimization solution for clients, taking into account different genetic constructs, clone-to-clone variation, host-to-host variation, or even molecule-to-molecule variation among candidate cell lines. The approach recognizes that many different techniques of platform manipulation exist, from minor correction of off-the-shelf media to their full-blown redesign. While having the in-house knowledge of their constituency is essential, this requires the time and resources to adjust and fine-tune their components.

“We have taken the media-optimization information and put it together in a single integrative process running in an ambr 15, resulting in the formulation optimization strategy,” Gillmeister explained. “This protocol partners the client together with Lonza and Sartorius.

“Furthermore, we have put together the alternatives for completing the whole media-optimization process. For organizations that do not have the instrumentation, we supply a ‘traveling ambr 15’ unit. The application specialists can go to customer sites, making it unnecessary to bring client cell lines to their laboratory for redesign.”


Scientists at Lonza Biologics worked with Sartorius Stedim Biotech on using the ambr15 microbioreactor system to evaluate chemically defined and custom media of nonanimal origin. This approach to media optimization can take into account different genetic constructs, clone-to-clone variation, or even molecule-to-molecule variation among candidate cell lines.

Matching Hardware and Software

At the upstream end of bioprocessing, monoclonal antibodies present special challenges. Monoclonal antibodies are large, complex molecules, and they show substantial glycosylation. Accordingly, they can be hard to produce.

The production of monoclonal antibodies was addressed by Roel de Waard, a process engineer at Synthon. He described Synthon’s efforts to generate monoclonal antibodies with a native conformation, free from aggregation and other defects that would compromise the performance of their products. Currently, the company has a focus on anticancer agents and therapies for autoimmune diseases.

Working with a 50 L bioreactor, de Waard and his team adapted technical solutions for inadequate oxygen control, a recurrent problem in regulating cell-line growth and macromolecular synthesis in culture. For the best possible growth, it is critical to maintain an optimal concentration of dissolved oxygen (DO) by sparging the bioreactor with air or pure oxygen.

“Through our investigations we determined that one mass flow controller is not sufficient to achieve accurate DO regulation,” de Waard explained. “Furthermore, we find that an optical DO probe is a feasible alternative to the traditional DO probe.”

Single-use cell culture technology involves disposable bags with capacities up to 2,000 L, and the fit between the bag and the stainless-steel holding apparatus must be perfect. Since it is always a struggle to deflate the bag following a run, it is essential that the match be as seamless as possible. “The take-home message,” De Waard advised, “is that we must match the hardware to the bag design and the bag design to the hardware.”

Pushing Cell-Density Limits

Jurjen de Jong, a process scientist at Patheon Biologics, presented his analysis of the company’s software package, the Multi Fermenter Control Software (MFCS/win). “Using this system, the scientist can send a pH set point (or profile) and monitor the values,” de Jong stated. “Afterward, the controlled value can be plotted or exported for review or approval.”

With the MFCS unit switched on, data is collected continuously, allowing monitoring of the system’s behavior. However, variables can be changed if circumstances warrant. For instance, if the oxygen level falls below a certain optimized level or the pH needs adjustment, the investigator can introduce appropriate modifications, and such modifications can be made on an ongoing basis.

As the variables can be adjusted to different setpoints conforming to optimized levels, they constitute a specific protocol, referred to as a recipe. In fact, the programmer can indicate how individual variables should be adjusted to ensure a superior response on a repeatable basis. The MFCS software is configured to process data as it is collected (and the data, as noted above, is collected continuously), and the functions run as long as they are started—until the recipe alters or terminates their function.

Perhaps the most significant feature of the program is its ability to respond to changes in cell density and adjust conditions so as to optimize the yield. After inoculation of the culture, the user can set pH alarms and initiate gravimetrical flow controllers. This phase allows control during the exponential growth phase of the culture.

“Overall, the main benefit of the recipe control is its standardization of the process,” de Jong concluded. “On top of that, the software has the ability to perform calculations that are not part of the control unit itself.”

Geometric Dissimilarities

The widespread application of single-use bioreactors relies on off-the-shelf designs, a requirement that limits the ability of the engineer to match these devices to the geometry of his or her existing stirred tank reactor. Even small differences in geometry can have profound effects on culture performance, and these geometric dissimilarities can override physicochemical tactics designed to equalize the performances.

As discussed by Colin Jaques, senior principal scientist at Lonza Biologics, impeller placement, sparger hole placement, and the lower volumes of single-use bioreactors necessitate the use of multivariate data analysis to compare different geometries and different fill volumes.

An important question is whether the employment of complex analytical programs may generate insights that lead to real performance improvements. This was indeed found to be the case. The application of multivariate data analysis revealed substantial differences in cell culture performance between the different stirred-tank reactors.

The basis of the analysis is the Principle Components Analysis (PCA) through the use of mass-transfer studies. PCA is an unbiased data analysis method,” said Jaques. “It tells what is there, allowing a search for natural grouping in the data.”

Jaques’ team compared a single-use bioreactor with a stirred-tank reactor, and established that these reactors had a high degree of geometric similarity at full volume, but not at half volume.

“With single-use bioreactors, you don’t get to design the vessel,” noted Jaques, “but you do need to control the design if you want to follow the principles of quality by design.”

As a group, all the speakers emphasized that the most critical challenge for biologic drug manufacturers is moving molecules rapidly to the clinic. This requires systems that can generate large quantities of high-value homogeneous material consistently, reliably, and at manageable cost.

But in the race to compress timelines, it is of the utmost importance that process development and optimization be accomplished as expeditiously as possible. Doing so helps drive product quality and makes scaleup as straightforward as possible. All biologic drug manufacturers, but especially those with many development programs, need to reach the read out in patients fast, with a reliable process and decent titers to minimize later rework if the product shows effect in patients.

Previous articleDiscovery of Rapidly Adapting Antibody Opens Door for Universal Flu Vaccines
Next articleGenomic Signatures Clearly Dictate Schizophrenic Drug Responses