Advances in Analytics for Bioprocessing

BILLION PHOTOS/SHUTTERSTOCK.COM

When FDA’s process analytical technology (PAT) initiative was first unveiled, the aim was to have the industry move away from a mindset of “testing quality” at the end of production to instead develop methods that monitor, assess, predict, and control quality from the start of the manufacturing process. PAT tools have facilitated ongoing monitoring and adjustment of bioprocesses to ensure that biologic drug products are produced with high quality, meeting the required specifications all the time. BioPharm International spoke to Jim Mills, senior vice-president of Technical Operations at Abzena; Karl Rix, vice-president, Business Unit Bioprocess, Eppendorf; Tommy Smith, lead scientist, Cell Culture, GE Healthcare Life Sciences; Richard Moseley, chief technologist, Microsaic Systems; and Christian Grimm, R&D director, Process Analytical Technologies, Sartorius Stedim Biotech, about the advances in PAT tools for bioprocessing. 

Advances in PAT tools

BioPharm: In your opinion, what are the three most significant advances in analytical tools for monitoring processes in biopharmaceutical manufacturing? 

Mills (Abzena): The three significant advances are on-line biomass measurement, at-line substrate and catabolite measurement, and the recent promise of at-line product measurement.

Rix (Eppendorf): In my opinion, the important advances are in peripherals, such as sensors, pumps, valves, and motors, being designed with additional feedback and intelligence, allowing more precise monitoring of the entire process. This goes hand in hand with advances in bioprocess software that allow customized use of the gathered data, for example, in tailored feedback control loops. 

Significant advances are also caused by the application of multivariate data analysis tools to interpret bioprocess data, develop a thorough process understanding, and use this knowledge for process design and ideally in-process corrective actions during manufacturing. 

Finally, I consider scale-down models as important analytical tools; although they are not directly used during manufacturing, they are still of high relevance for it. I think efforts in the creation of scalable bioprocess systems have advanced and will further advance the possibilities for bioprocess analysis, both in development and for troubleshooting.  

Smith (GE Healthcare): The three most significant advances in analytical tools for monitoring upstream processes in biopharmaceutical manufacturing are near-infrared (NIR) spectroscopy, Raman, and radio-frequency (RF) impedance. In upstream processing, NIR and Raman are useful for real-time monitoring of cell culture parameters including pH, dissolved oxygen, and various metabolites. After calibration for a given process and media, these two tools provide constant information about the culture without the need for manual sampling. The same is true of RF impedance in relation to cell population density within a culture. 

These tools are important for keeping the entire cell culture process under close surveillance. An additional advantage to using these three in-process analytical instruments is a reduced risk for contamination due to manual intervention and off-line analyses.

Moseley (Microsaic Systems): The analytical tools commonly used to monitor upstream and downstream bioprocessing have been traditional, simple instruments. For upstream process monitoring, these include pH monitors, glucose measurements, and gas pressure gauges. For downstream purification, the tools are slightly more sophisticated, such as enzyme-linked immunosorbent assay (ELISA) and high-performance liquid chromatography with ultraviolet detection (HPLC–UV). However, these tools are not suitable for the current challenges faced by the biopharmaceutical industry, particularly those driven by the stricter regulations surrounding biologics, or biopharmaceuticals. Efficient quality-by-design (QbD) strategies are now needed, giving bioprocessing operators and managers access to more detailed product and process information at the point-of-need, in other words, where the reaction and process is taking place. 

The first significant advancement in analytical tools is the development of better multivariate data analysis (MVA). The biopharmaceutical industry, previously familiar in making simpler small-molecule drugs, has found that MVA is suitable for the challenge of safely controlling a bioprocess. MVA is used to link the critical quality attributes (CQAs) of the target product (as an example, glycosolyation is a CQA for monoclonal antibody safety and potency) to a set of critical process parameters (CPPs). CPPs can include temperature, pressures, and flow rates. 

Adopted industry-wide as the preferred choice of analysis, mass spectrometry (MS) can provide the information needed to monitor and control the processing of biologics using MVA, as it measures CQAs and CCPs directly. Until recently, MS instruments were large, power hungry, and capital-intensive equipment to purchase and maintain. These instruments still form much of the centralized-laboratory infrastructure, where samples from the bioprocessing workflow are analyzed. However, turning a sample analysis around can often take days or even weeks. A significant advancement is the emergence of compact, miniaturized MS instruments containing powerful analytical software tools, more suited to point-of-need applications. Such compact instruments may be deployed at almost any step in the bioprocess (at the ‘point-of-need’) and can be used to analyze the contents of the cell media, or buffer solution. As a result, analysis times become instantaneous, speeding up time for manufacture and also the time to market for new biologic drug candidates.

Grimm (Sartorius): From my perspective, the most significant developments of analytical tools for use in biopharmaceutical processes are the full integration of ready-to-use single-use inline sensors such as in-line capacitance for biomass monitoring in bioreactor bags, the use of optical spectroscopic equipment and methods for inline integration into bioprocess solutions (e.g., NIR, Raman, UV–VIS or fluorescence), and the use of data analytics such as multivariate statistical and other chemometric methods to transfer data into knowledge.

PAT in biopharmaceutical manufacturing

BioPharm: How widespread, would you say, is the use of PAT tools in biopharmaceutical manufacturing? 

Grimm (Sartorius): Obviously, the implementation of PAT tools is not as widespread in biopharmaceutical production as in chemical or pharmaceutical manufacturing. In biopharmaceutical processes, we have to distinguish between the status within upstream and downstream processes. Interestingly, the implementation of relatively new tools such as spectroscopy is more mature in upstream than in downstream, even if the matrix under test is far more complex in a cell culture for example than in a later polishing step. Compared to classical pharmaceutical processes, however, we are far behind in the implementation of PAT tools. This may change significantly in the coming period especially with the increased use of continuous and intensified manufacturing approaches within the industry.

Moseley (Microsaic Systems): Simple, low cost, and robust sensors have been used for PAT in the chemical industry for many years, for example, Fourier-transform infrared (FTIR) spectroscopy is used to monitor and control small-molecule API synthesis. However, the data produced from these types of simple sensors has very limited value for MVA in a bioprocessing context. Instruments suited to analyzing complex biologics such as MS have been proposed and explored in the past by biopharmaceutical companies, but the lack of deployability of the large MS instruments, and the limited software analysis that were available at the time did not lend themselves well to being PAT tools. The advent of miniaturized mass spectrometers has presented an important step forward for the adoption of point-of-need MS by biopharmaceutical manufacturers. 

Rix (Eppendorf): I think the majority of the biopharmaceutical companies are interested in implementing PAT tools in manufacturing even if they are not using them now. I expect it will take a couple of years more until PAT is widespread in bioprocessing. Furthermore, PAT tools will get increasingly important at earlier development stages, I believe, because gaining sufficient process and product knowledge during the research and development phase helps implementing PAT at later stages. Regulatory agencies encourage PAT implementation, and we might see even more incentives from their side in the future.  

Mills (Abzena): I agree that the concept of PAT is not new in pharmaceutical manufacturing. However, the ability to consistently and robustly measure, monitor, and control such complex processes as biologics manufacturing are still largely outside of the state-of-the art. It has much to do with the difficulty in directly measuring macromolecular characteristics in-process. Many developmental PAT tools have been proposed using indirect measurement methods relying on complex statistical correlations between traditional physical on-line measurements and the parameter being inferred. Because of the difficulty in direct measurement, trust of such inferred measurements has been hard to gain. In a highly regulated industry such as the biopharmaceutical sector, which also has high costs and long cycle times for product development, these challenges have acted as a significant barrier to PAT adoption for most companies, especially small and medium enterprises (SMEs) and contract manufacturing organizations (CMOs).

I think PAT tools are beginning to gain a foot-hold in bioprocess development to save time and cost during the early transition from research into development. One aspect that has helped with this is more vendors, such as Sartorius, providing already integrated solution for PAT in their product offerings (e.g., single-use biomass measurement), which makes adoption simpler for companies. However, even though PAT is starting to be used during process development, generally for GMP manufacturing, traditional quality control testing is still relied upon.

BioPharm: What positive changes have PAT tools contributed to in biopharmaceutical manufacturing over the past 10 years?

Smith (GE Healthcare): In upstream processes, periodic sampling (daily or twice daily) is the traditional norm. However, with the advent of PAT, constant pH, dissolved oxygen, and metabolite monitoring are now possible. Such monitoring allows for proactive management of the system instead of a reactive response. Now the operator can have an insight into what the status of the system is at any time as well as an opportunity to add feeds, bases, and acids when needed instead of at or after the traditional sampling time during the day. The gain in this instance is a more consistent quality of product due to the ability to keep parameters that affect quality attributes steady or within desired ranges over the course of a production run.

The increased use of PAT has led to a reduction of contamination incidences in manufacturing due to the ability to monitor processes with minimal manual intervention or sampling. In upstream work, this includes non-invasive pH and dissolved oxygen monitoring through optical devices that do not need to be inserted into the bioreactor post steam-in-place or irradiation. Another example is the monitoring of cell density via an RF impedance probe that is inserted into the system before steam, autoclave, or irradiation. Overall, this monitoring inevitably results in a cost savings to the company during the life of the production cycle. Tighter controls over the process are also a result of PAT tools. 

Grimm (Sartorius): Working with complex biological systems producing the target molecule or drug substance makes the implementation of PAT tools in biopharmaceutical production processes far more difficult than in chemical synthesis. However, the recent implementation of several PAT tools in terms of design of experiment (DoE), inline sensors, spectroscopy probes, and the use of multivariate statistical methods have increased fundamental process-knowledge and given insight into the mechanisms of CPPs on the CQAs. This enabled more reliable process automation and was a major step forward in the direction of QbD implementation in bioprocesses.

Rix (Eppendorf): PAT is a system for designing, analyzing, and controlling manufacturing through timely measurements of critical quality and performance attributes, with the goal of ensuring final product quality. I think this exactly describes the most important positive changes PAT tools contribute to. They allow achieving better process understanding and therefore predictability. Associated with process control, they can help to minimize process variations. As a result, a more consistent product quality can be achieved. I assume PAT tools also help to save costs, as they can help to reduce the number of failed batches.

Moseley (Microsaic Systems): There is broad recognition that PAT tools are essential for helping to simplify manufacturing, and this work has already led to the adoption of relatively simple instruments being integrated into process steps where PAT is relatively straightforward. However, the complexity of bioprocessing means important decision points in the process are not all routinely covered by PAT. Important improvements to MS hardware and software means the remaining difficult bioprocessing steps can be monitored and controlled using miniaturized MS-based PAT systems. 

Challenges of implementing PAT

BioPharm: What challenges still exist in PAT implementation for bioprocessing?

Rix (Eppendorf): One important challenge I see is transferability. PAT already plays a key role during research and development, where it helps in increasing process knowledge and developing a process control strategy, but it usually happens at small scales. To be applied in manufacturing, PAT tools and control strategies need to be validated and transferred across scales and facilities. 

Secondly, PAT implementation requires specialized expertise of how to integrate sensors, how to validate models, and how to program feedback control loops. It will often require the combined efforts of vendors and users to tackle this challenge.

Moseley (Microsaic Systems): The understanding of PAT and its relevance to complex bioprocessing has improved enormously. There are, however, many challenges that need to be tackled before there can be full implementation of PAT tools into the whole of bioprocessing. For the implementation of a miniaturized MS PAT tool, the key challenges are: 

• To integrate the mass spectrometer in environments normally hostile to the instrument without damaging the instrument or compromising the safety of the product 

• Acquire reliable data over an acceptable time 

• Provide accurate and useful information to the end-user. 

Mills (Abzena): I think the main challenge is still the implementation of robust, reliable direct measurement tools that can be validated and applied in the GMP manufacturing environment.

Grimm (Sartorius): The remaining hurdles for PAT implementation for bioprocessing could be split into three categories. There are, of course, residual technical and application challenges due to the fact that we have to monitor and control complex biological systems, whose responses to changes of macroscopic environmental factors cannot be predicted completely. In addition, the number and complexity of important parameters are quite high, hence, requiring analytical methods that are selective and sensitive enough to measure these parameters in an online way. Furthermore, the current trend in the industry to use single-use equipment-not only in process development but going further to manufacturing scale-requires an implementation of PAT tools in single-use systems as well, which is a technical challenge per se. 

Secondly, the implementation of PAT in bioprocess requires mindset changes of the acting people. There is still, in some cases, the perception of the operators, that they are faced with a black box batch process, which they better not touch if they were able to run it in a semi-optimal way. If our industry is moving forward to continuous or intensified processes, there is a clear demand for implementing PAT tools so that the processes can be run in a stable and robust regime. 

The third important aspect for the hesitation of PAT implementation in bioprocessing is the lack of a strong economical driving force. We still have legacy production processes running in the industry that are far away from optimal in terms of efficiency and quality. The recent increase of more biosimilars coming to market, higher consolidation in the industry, and changes in local markets will give the demand of more cost competitive manufacturing. These drivers will lead to a higher degree of implementation of PAT in the near future.

Smith (GE Healthcare): The vast number of potential analytical instruments that could be integrated into a biopharmaceutical process presents a logistical challenge. A single unit that implements NIR or Raman for metabolite monitoring is typically not an issue. However, when instrumentation such as HPLC/MS, metabolite/cell density analyzers, and sample purification modules are all individually attached to a sterile sampling mechanism, the footprint of the equipment and distance of travel of the sample becomes a burden.

Future outlook

BioPharm: What advances in analytical tools for biopharmaceutical manufacturing would you like to see within the next five years?

Moseley (Microsaic Systems): We believe that any advances should be aimed at bringing cheaper and safer biologics to all the people who require them. For analytics, this means the tools must: 

• Add important value to bioprocessing by maximizing product yields, ensure product safety, and minimize process downtime

• Be reliable and robust to suit the point-of-need applications needed for PAT. 

In the next five years, we would like to see miniaturized mass spectrometers routinely integrated into upstream and downstream bioprocessing environments, where they can both monitor and control the production of biologics, through their unique ability to detect and analyse proteins, and small molecules.

Grimm (Sartorius): Next PAT advances should be on further implementation of inline sensors for monitoring and controlling the CPPs, including more multi attribute methods. For example, HPLC or MS could be used to get online access to CQAs and mechanistic or hybrid process models could improve advanced process control. In addition, the industry has concentrated mainly on implementation of PAT tools in the upstream area, but we would likely see more development in the downstream part to set the basis for the vision of real-time release.

Smith (GE Healthcare): It would be optimal to see more all-in-one modular offerings allowing the targeted industry to implement a single instrument for all, or most, desired analysis parameters.

Rix (Eppendorf): In my opinion, biopharmaceutical manufacturing would benefit from additional sensors for online monitoring of metabolites and CQAs of the product. These parameters of course can be analyzed offline and online-integration of some sensor types is feasible; however, I hope to see, for example, metabolite sensors in the future, which can be as easily used as today’s pH and dissolved oxygen sensors. Often single-use technology is applied in manufacturing; it would be also desirable for new sensor technology to simplify handling and avoid sterility issues.

Mills (Abzena): I’d like to see actual real-time measurement of product concentration and product quality. 

Article Details

BioPharm International
Vol. 31, Number 7
July 2018
Pages: 39–43

Citation

When referring to this article, please cite as A. Siew, “Advances in Analytics for Bioprocessing,” BioPharm International 31 (7) 2018.