September 30, 2022 PAO-09-022-CL-08
Process analytical technologies (PAT) comprise a range of analytical tools used to facilitate the development of dynamic manufacturing processes that can account for variability in raw materials and equipment used to produce drug substances and drug products. The use of PAT is guided by a regulatory framework established by the U.S. Food and Drug Administration (FDA) to encourage innovation in pharmaceutical development, manufacturing, and quality assurance.
The goal of PAT is to build quality into biopharmaceutical manufacturing processes by monitoring and controlling the process inline and in real time. Once the critical process parameters (CPPs) that impact the critical quality attributes (CQAs) of the drug substance or drug product have been identified and fully characterized, appropriate analytical methods are developed and corresponding technologies are used to monitor those CPPs so that they can be controlled and thus both the CPPs and CQAs be maintained within a specified design space. In this manner, quality is built into the process from the start, supporting the principle of quality by design (QbD), rather than assessing the quality of products after they have already been produced.
Most current techniques and technologies employed as PATs in biopharmaceutical manufacturing focus on monitoring CPPs in upstream cell culture or microbial fermentation processes. They include sensors for pH, dissolved oxygen (DO), carbon dioxide, temperature, pressure, and capacitance measurements throughout these upstream processes. There are also many technologies that can go beyond these more traditional sensor measurements, such as various types of spectroscopies, spectrometry, and chromatography. Advances in some of these more complex technologies are enabling their incorporation inline and atline for use as PAT tools. These advances are major steps toward unlocking the potential of the Pharma 4.0 and Biopharma 4.0 paradigms and being able to implement self-monitoring and regulating autonomous processes.
The use of PAT varies significantly across the biopharma industry, with different companies, facilities, and operations reaching different stages of maturity. Similarly, analytical technologies themselves are evolving as new solutions are introduced that enable online and atline sampling (if needed) and analysis. PAT is used not only for the monitoring of batch processes but also semi-continuous and continuous (perfusion) upstream processes as a means for intensifying processing.
In the future, the real-time data collected upstream will be channeled downstream to realize true dynamic manufacturing processes with appropriate adjustments made across the entire process. For instance, detection of increased aggregation in the bioreactor could lead to modification of the chromatography step to ensure effective and efficient elimination of aggregates from the process. Currently, PAT is predominantly used for monitoring, with a few companies starting to implement control aspects for upstream processes. Such activities are encouraged by regulatory authorities, and this encouragement builds on the growing recognition of the value that PAT brings.
While PAT implementation is currently more widespread in small molecule manufacturing, there has been a rapid uptick in implementation for upstream processes within the biopharma sector. The ability to take measurements at many more time points and monitor changes within the bioreactor allows implementation of control solutions much more quickly, which adds real value.
The main challenge going forward is to move existing analytical technologies to inline or online operation. Fit-for-purpose versions of existing techniques must be developed specifically for use as PAT tools. When it comes to real-time product release, the biggest hindrance at present is in the area of biosafety analysis — bioburden, mycoplasma, and endotoxin testing. Rapid tests are needed so that these analyses can be implemented inline or even online.
The greatest advances have been made with optical spectroscopic methods including infrared, ultraviolet, and Raman spectroscopy. Recent improvements in signal processing analysis and computation have moved those technologies forward and enabled inline applications.
Raman is an optical spectroscopy technique that essentially provides a molecular fingerprint of a sample. As is the case with human fingerprints serving as unique and consistent identifiers of individual people, individual molecules have unique and reproducible molecular Raman spectra. With Raman spectroscopy, therefore, it is possible to identify which molecules are present in a sample at a very high resolution. When Raman spectroscopy is used as a PAT tool, it allows for monitoring of the molecular composition of a sample over time. As an optical method, this technology is able to determine chemical composition and molecular structure information in a nondestructive and reproducible manner, avoiding the loss of chemical information due to degradation, instability, or sample preparation. In addition, there is no need to collect a sample and send it to the QC lab for analysis. It only requires the insertion of a probe into the process along with the other sensors that are already widely used.
For bioprocessing in particular, Raman spectroscopy also benefits from the weakness of the bands produced by water. The water peaks do not interfere with the peaks associated with the analytes of interest, which allows for high-quality analysis of aqueous bioprocess solutions. This feature — combined with the molecular specificity and nondestructive nature of Raman spectroscopy — has made it the primary spectroscopic technique employed to date as a PAT tool in the biopharmaceutical industry.
Currently, the most significant use is in process development, with Raman less widely deployed in the GMP space. But that is changing, and there is definitely movement toward wider adoption in GMP manufacturing.
The key limiting factor of Raman spectroscopy is that it lacks the sensitivity to analyze molecules that are only present in trace amounts. A sufficient concentration and variation of concentration is needed so that multivariate modeling can be performed. In addition, an offline reference measurement must be available for which the Raman spectrum has previously been obtained with which to match the fingerprint of a given analyte. As long as those two criteria are satisfied, Raman can be applied for the analysis of many types of molecules.
Raman spectroscopy is predominantly applied as a PAT tool for the monitoring of upstream cell culture processes performed in bioreactors. There remains significant potential to employ Raman in the downstream space as well, an area of ongoing exploration within the biopharma industry given the robustness of the technology. Upstream applications have included all aspects of biopharma drug development, from single-cell research to process control in GMP manufacturing environments. In fact, Raman is a technology that is applicable to the entire workflow, provided that the key criteria are met.
The ability to monitor and control bioprocesses in real time has tremendous advantages for manufacturers. Rather than waiting until the end of a run to perform quality control testing, outliers can be identified quickly. Process conditions can then be modified to address the issue, or, in the worst-case scenario, the batch can be scrapped before too much time and resources have been invested in it. Process challenges can be better managed, and operators can have greater confidence that processes are proceeding as expected.
There are situations where Raman spectroscopy might not be the best PAT tool or when one or more other techniques must also be employed. As mentioned above, detection of components in a bioprocess solution that are present in trace amounts is an obvious example. When the time for analysis is highly limited, Raman may also not be optimal, because the acquisition time is longer than for some other methods. In these cases, infrared or fluorescence spectroscopy may be a better option. For most applications, however, at least in the upstream space, Raman has proved to be a robust and well-employed technology, even for fairly challenging tasks, such as identifying the different glycoforms generated in a process.
Successful use of Raman spectroscopy as a PAT tool in biopharma processing is dependent on successful multivariate modeling. The model must be built before Raman can be deployed as a PAT tool, and each model is specific to the molecule of interest and the process. Fortunately, software advances have been achieved that greatly streamline the steps involved and make this modeling activity intuitive and nearly fool-proof.
With the Raman technology developed by MilliporeSigma, the Raman probe in the bioreactor records signals. At set time intervals, offline samples are collected and sent to the QC lab for analysis, and the time of sample collection is input into the software. At the conclusion of the run, the information from the offline samples is entered in the software to account for variations attributable to biological differences that take place in a batch. This data is preprocessed and analyzed to make sure that there are no apparent outliers. The process is repeated for at least three runs, and then the model is developed and validated. After that, it can be deployed.
The quality and accuracy of Raman models can be improved using data from additional samples taken from subsequent runs. Such ongoing calibration is not necessary in most instances, but it may be appropriate if a process is being moved from a smaller vessel to a larger one. The need for additional data can be determined by performing a calibration run at larger scale using the model developed for the small-scale process.
In addition to a comprehensive software solution from spectral acquisition through chemometric modeling, MilliporeSigma provides extensive support and training in model development. For companies that don’t want to build their own models, MilliporeSigma also offers modeling as a service.
As the biopharma industry moves toward Biopharma 4.0, different analytical tools will need to be integrated into bioprocesses to enable real-time monitoring and control. Integrating PAT solutions is a clear necessity in the path forward to realizing the anticipated autonomous Facility of the Future.
Because Raman spectroscopy is such a strong technology in bioprocessing owing to its capability to monitor multiple parameters simultaneously, and because the presence of water in aqueous solutions does not interfere with analysis, this technology was determined to be important for meeting internal MilliporeSigma and external customer needs for monitoring processes in real time.
The acquisition of RESOLUTION Spectra Systems in mid-2020 provided access to GMP-ready Raman instrumentation and software. It also supported MilliporeSigma’s Bio4C® Software Suite, a first-of-its-kind ecosystem that combines process control, analytics, and plant-level automation that was launched shortly beforehand. The data provided by the Raman probes is collected by the Bio4C® PAT Raman Software and combined with other real-time PAT across multiple CPPs to support integrated bioprocessing.
The Raman PAT tool offered by MilliporeSigma, the ProCellics™ Raman Analyzer, was designed specifically for the bioprocessing industry. It is a compact instrument with a very small footprint suitable for use on the manufacturing floor, unlike most other available technologies. The user interface is very intuitive and easy to use, so there are no requirements for data science expertise, and the technology can be accessible anywhere, from process development to the GMP space. To facilitate seamless transition from R&D to manufacturing in a GMP environment, the Raman PAT Platform is equipped with access control, a workflow control layer, an audit trail, and record management capabilities which comply with GAMP 5 and GMP requirements. The user experience team within MilliporeSigma is dedicated to continuously improving the customer user experience with existing technology as well as technology in development.
The Raman software is specific to the ProCellics™ Raman Analyzer and complements the Bio4C® suite of products, including Orchestrator and ProcessPad. For instance, time series data generated by the ProCellics™ Raman analyzer can be imported into the ProcessPad software, enabling operators to gain a better understanding of the process.
There are two analyzer options: single and multichannel. The systems are the same size, but with the latter it is possible to run up to four probes simultaneously, which allows monitoring of multiple reactors at the same time. With this system, it is possible to build the multivariate model more quickly since the data from three runs can be collected at the same time. It is also possible to monitor different processes in different reactors.
Most of the applications evaluated to date for the ProCellics™ Raman Analyzer have been focused on the bioreactor and the upstream space. However, it is not limited to just upstream uses, and there are potential opportunities for deployment of the technology in the downstream space. There may need to be slight modifications to the instrument to accommodate process differences, but the fundamental Raman technology itself is not limited.
Feedback regarding the overall customer experience with the ProCellics™ Raman Analyzer and its associated software have been very positive. The platform has proven to be easy to use, quite approachable, and very intuitive. The technology is also backed by a team of chemometric experts available to help customers implement Raman technology as a PAT tool for use in process development and manufacturing. The team helps ensure that implementation is as seamless as possible by guiding customers through the model-building steps and streamlining the entire process.
The potential applications for Raman technology as a PAT tool are fairly broad, although other techniques may still be more suitable in certain instances. MilliporeSigma is therefore actively pursuing other methods that can be developed as atline or inline solutions to fill any gaps. The goal is to ensure that customers have the PAT tools they need to monitor all of the possible CPPs and CQAs associated with their bioprocesses, whether those tools involve Raman spectroscopy or another technology. By providing these solutions, we are ultimately providing customers with the ability to transition seamlessly to Biopharma 4.0 and to operate facilities of the future.
Stacy Shollenberger holds an MS degree in biochemistry from the University of Virginia and an MBA from Penn State University. She has over 12 years of experience in analytical and bioanalytical product and method development. Currently, Stacy is a Senior Manager of Process Analytical Technologies (PAT) at MilliporeSigma and is responsible for defining the PAT strategy to move testing from traditional off-line quality control (QC) to QC on the bioprocessing floor.