Brian Davies

Firms using biotechnology to manufacture medicines are embracing process analytical technology (PAT) at an ever-quickening pace as part of their quality by design (QBD) initiatives. 

They are often faced with complex and fragile processes lasting many hours or days, and are looking towards PAT as a means of not only understanding the fundamentals of their processes, but also to offer more incisive process monitoring technologies, control and optimization strategies.  

Commercial payback comes from both R&D — from being able to get more robust processes to market quicker — and from manufacturing — by being able to mitigate process risk, batch-on-batch, as well as through process optimization, which improves subprocess and final product yields. 

This article will describe how aspects of PAT are applied to a fundamental process stream in biotechnology manufacturing — fermentation. 

Biological processes often have high variation levels, making the development of a robust and reliable process inherently difficult. Couple this with the high intrinsic value of the end product, and a biotech firm has to bear a degree of risk during its manufacturing process that is unseen in other parts of pharmaceutical manufacturing. Mitigating this manufacturing risk has been a biotech goal for many years, and current approaches correlate with the principles QBD and PAT: 					

To identify critical quality attributes (CQAs) of the product and their acceptability limits.

To identify the sources that lead to the variability in the CQAs.

To identify the factors within the process that can be adjusted to control the variation in the CQAs.

To determine a control strategy to ensure that the CQAs are held within predetermined limits during the manufacture and lifetime of the product.

A typical fermentation-based biotech process flow is shown in Figure 1. From an economic and quality risk perspective, the critical parts of this process are the production of the active agent through the fermentation step, the recovery of the active agent in the purification step, and the stabilizing of the finished product prior to shipment during the freeze drying/lyophilization step.  

Figure 1 A typical fermentation-based biotech process flow.			

QBD and PAT efforts are being put into this process, not only to aid understanding, but to control the critical steps and go on to optimize the subprocess yields from the fermentation, purification and lyophilization steps. 

Understanding and monitoring what is happening during the fermentation process is vital to ensure the healthy progression of the culture, as its health determines the conversion rate of nutrients to product, the formation of unwanted by-products, and in extreme cases, the onset of poisoning. Growth kinetics and substrate consumption are key indicators of the progress. 

Process mass spectrometers (MS) and process FT-NIR spectrometers are widely used as the measurement technology components in these monitoring, control and optimization systems.  

Figure 2 illustrates a biotech fermentor with a process MS and process FT-NIR attached. Each technique monitors different aspects of the process as it progresses.  

Figure 2 A biotech fermentor with a process MS and process FT-NIR attached. Each technique can be used to monitor different aspects of the process as it progresses.			

The MS allows for highly sensitive and precise measurements of the components of the gas streams into and out of the fermentor. Analysis of the respiratory gases being fed to and removed from the fermentor is an ideal means of characterizing the fermentation state. A key indicator of the fermentation progression is the respiratory quotient (RQ). Respiration is the process by which an organism oxidizes food to produce energy. The RQ is the ratio of the carbon dioxide evolution rate (CER) to the oxygen uptake rate (OUR).  

Many fermentations are characterized by small changes in oxygen and carbon dioxide concentrations at critical phases of the fermentation. Therefore, the process MS (with ability to simultaneously measure multiple gas components, and use other components such as nitrogen and argon in the gas stream as reference gases) can very rapidly calculate this rate and feed it into a process control strategy to intervene, keeping the fermentation on track.  

When equipped with a sample stream multiplexer, a single process MS can be attached to multiple fermentation vessels, offering a high degree of operating and economic efficiency. 

Figure 3 shows the process MS generated trace of RQ versus fermentation time. Here, the RQ value falling below a predetermined level is used to trigger the addition of nutrients into the fermentor to keep the fermentation at an optimum state.  

Figure 3 Fermentation control: nutrient addition triggered from process MS data.			

Making direct measurements of fermentation components within the fermentor itself allows additional information to be obtained on the progress and state of the fermentation.  

Here, FT-NIR using a fibre optic probe immersed in the bulk of the fermentor liquor is used. The components measured by the FT-NIR vary by fermentation type, but can include: 					

Making measurements directly in a fermentation liquor can be inherently difficult because of the changing chemical and physical properties of the medium during the process. For example, as biomass is produced during a process, the liquor may go from being a relatively clear liquid to a turbid liquid with a high degree of suspended solids; this can confound the measurements of the critical components.  

Where transmission spectroscopy would work well at the beginning of this process, it is likely that reflectance spectroscopy would be the best technique to use as the medium turbidity increases.  

The latest generation of multipurpose fibre optic probes with both transmission and reflectance optics can be linked to a spectrometer with multiplexed measurement channels. This allows the system to be controlled through adaptive software work flows and can easily track these physical changes automatically. This enables the system to work at the optimum sampling regime for the most accurate and precise component concentration; ideal for automated process measurement and linking to a process control system. 

Figure 4 shows the data from using an FT-NIR to determine the concentration of a product during the fermentation process cycle.  

Figure 4 Monitoring fermentation product concentration using a process FT-NIR spectrometer.			

A fermentation stage with process MS and FT-NIR capturing large quantities of chemical component data coupled with classical process parameters such as temperature, pH, stir speed and gas rates being logged elsewhere, is ideal for investigating with multivariate data analysis tools. Currently, work is being conducted on this rich data pool and process fingerprints, multiplicative statistical process control, 'biological time' and model predictive control approaches are developing for use in control and optimization strategies.  

Although the processes involved in biotechnology manufacturing differ significantly from those of small molecule medicines, the underlying principles of QBD and PAT are being applied enthusiastically.  

The nature of biotechnology subprocesses allows control strategies to be extended into optimization strategies, and create sound economic value for firms.  

We can expect to see more of this pragmatic application of PAT in this part of the pharma industry, as measurement and control technologies develop, and are applied to other sections of the process, especially those downstream of the fermentation step that carry a high degree of residual risk.  

is director, PAT at Thermo Electron Corporation, Madison (WI, USA).			

Articles

Quality by design and PAT approaches are increasingly being used for the biotech manufacturing of medicines. Complex manufacturing processes can not only be controlled using PAT principles, but optimized with respect to both product quality and economic value. This column describes how the fermentation process is often the first to benefit from this type of implementation.

The development of PAT in biotech manufacturing

This editorial is about the process analytical technology (PAT) market — which doesn't exist! 

Markets, in general, exist for products and/or services, and can be described in terms of the concepts of market evolution and development. However, PAT is not a product or service. It is a concept, a working principle or a framework for operating, depending on your point of view and how you implement it. 

It is not an analytical system, which is only a small part of PAT. Nor is it something you can go out and buy. You can purchase the tools and technologies that make up the framework, you can buy the services and capabilities that you need to implement and it is these that have a market that exists and is growing. But you can't go out and buy 'PAT' even though some people believe they have! 

Saying this, the following editorial will discuss PAT in marketing terms as though it is a product and has a product marketing life cycle. I will attempt to describe how PAT is evolving in the pharmaceutical industry from the perspective of a product and service provider. 

In marketing terms the most appropriate way to look at how PAT is evolving is through the analogy of a product life cycle adoption curve. Here, a product or market is described in terms of uptake phases through its life from introduction, through maturity and into its decline. Figure 1 illustrates the typical four phases of a market evolution. 

Innovators have existed for many decades, well before the inception of PAT. In many areas and companies within the pharmaceutical industry, groups of innovative scientists and engineers struggled to apply measurement and even control techniques to pharmaceutical processes. 

 Figure 1 The PAT ÃÂ´marketÃÂ´ adoption curve.			

Often these innovations were stopped or stifled by the then current regulatory regime, but once the PAT initiative kicked in many of these groups blossomed into well funded and internally supported PAT groups. Although now encouraged by the regulators to be innovative they found that the necessary measurement and control tools, technologies or methodologies didn't exist and set about developing them themselves. From necessity, these groups often had to set out and develop instrument and control systems before they could apply them to PAT applications, taking nascent measurement technologies and creating bespoke systems. This was an expensive and time consuming exercise for groups whose core competency was pharmaceutical product development, engineering or analytics. 

This period of PAT development is almost over, vendors awakened to the fact that a gap in the supply market existed and a raft of technologies were developed that are now available for purchase. Innovators are realizing their core competencies should be applied to their products and processes rather than emulating vendors. 

As time moves on, each generation of vendor-supplied technology needs less and less customization to fit the PAT needs. Innovators are now moving into exploiting available technologies and moving their innovation into application, not inventing measurement systems. 

The more progressive innovators have realized that deep, mutually beneficial partnerships with their vendors offer them a fast track to getting the tools they need. 

As the early innovators move application of PAT tools into the mainstream they are realizing that tackling the standardization of analytical, engineering and data infrastructures is the next hurdle to embedding PAT into their organizations. 

Unfortunately, from a marketing point of view, the innovators, although powerful and necessary advocates of PAT, don't actually represent a large percentage of the pharma industry but they are the leaders, they set the trends. 

Early adoptors follow the innovators once the product market begins to develop and stabilize; PAT uptake is not different. Seeing the support from the regulators and the evidence from the innovators that PAT can work and bring more consistent product quality and business benefits, these people are now beginning to implement PAT. The realization that a well thought-through PAT strategy can be implemented without the establishment of large and expensive internal groups and that commercial measurement and control systems are available off-the-shelf from vendors, with full after-sale services if needed has given these early adoptors the confidence to engage in PAT. 

It is vital that these implementations succeed because the majority of the pharma industry, yet to engage with PAT, will be watching to see how these more practical and pragmatic projects fare with respect to regulatory scrutiny and the creation of real and tangible business benefits for the pharma company. The market is currently at the beginning this phase. 

The majority of pharma companies are sitting on the fence, deciding if or when to jump into PAT. Not necessarily the industry leaders or the largest of companies, this majority group contains the greatest number of individual pharma companies i.e., the middle- to small-size research-based companies, generic manufacturers and contract manufacturing companies. It is likely that the engagement of this community will bring the explosion in the use of PAT that will be necessary for its success. 

Typically this market group is highly pragmatic, very risk averse and very financially aware of return on investment. Therefore, they are likely to wait to see the results of the early adoptors and how easy it becomes to purchase the necessary technologies and services to implement a successful PAT project. To this end, it is important that vendors invest in creating goods and services that provide a high chance of success, good return on investment and low-risk implementation. 

In marketing terms the 'laggards' are those that, for whatever reason, follow late in the adoption life cycle. Typically these are highly conservative, risk-averse organizations that need to see rock-solid reasons to change and engage. In pharma company terms it is unclear who these will be. 

Perhaps companies that cannot create internal cultures that respond to commercial and regulatory pressures will lie in this group or those with products and processes so well understood that PAT will bring little or no benefit? 

However, it's likely there will be laggards. The regulators have made it clear where regulatory scrutiny will be focused; at companies that chose  not to understand the scientific and engineering  aspects of their products and processes and cling to  the 'old' three batch validation paradigm. 

It is clear that PAT is evolving and maturing and that the pace is quickening. It is driven by the twin forces of regulatory encouragement and a pharma industry that has an increasingly more profound understanding of the potential business benefits of PAT. It is also being fuelled by the vendor community through an ever increasing number of off-the-shelf PAT technologies, tools and services. 

However, looked at across the whole pharma industry, uptake of PAT remains patchy and hesitant in 2006. Typically it is some combination of fear, uncertainty and doubt that stops people from engaging in change, even one such as PAT with its clear benefits to the industry and its patients. Understanding and removing these barriers to PAT uptake will require the close-working partnership of the regulators, the pharma industry and the vendors. For all of these parties the development of a large and strong PAT 'market' is a common cause.  

In marketing terms, the Innovators will show the way to the Early Adaptors who in turn will prove to the Majority that investing in PAT can be done with a high degree of success and low risk. 

The Majority must be engaged if PAT is to become a way of life in the pharma industry and a success, the earlier and faster the better. 

Of course, if the market adoption analogy is applied strictly to PAT, it like all 'products' or 'markets' has a finite lifetime before it goes into decline and is replaced by the 'next generation'. What this next generation of PAT will be and look like will depend on the success of this first generation. No doubt the next generation of regulatory, industry, and vendor Innovators are thinking about this already.  

is director of PAT at Thermo Electron Corporation, USA. 

The PAT market is developing and evolving rapidly as pharmaceutical firms strive to implement the framework set in place by the regulators. As with all new markets, people will engage at different times, at differing rates, and for different reasons. This editorial seeks to treat PAT in classical marketing terms, describing four phases of the PAT market evolution and discussing how the PAT market will likely mature.

The process analytical technology market

Today's pharmaceutical company is driven by the need to create and execute effective and efficient processes if its business model of exploiting the revenue generation of the new product pipeline is to be realized. The challenges within drug discovery are to discover new entities, screen them for safety and efficacy, and then pass them swiftly into the product development process.  

In product development the challenge is to handle the product pipeline effectively. Early culling of entity and product failures is important if the focus is to be on recognizing new products with high success potential and driving them through the product pipeline, ultimately leading to routine manufacture and sale. 

Technological and scientific innovation has been very high in the discovery phase of the process, but, until recently, not followed through to the same extent into the later stages of product development and the routine manufacturing parts of the pharmaceutical product life cycle.  

The pharmaceutical industry has been performing process analytics in one form or another for well over 20 years, but had stated that innovation is not encouraged, enabled or supported by the traditional good manufacturing practice (GMP) used in pharmaceutical manufacturing. Likewise, the industry regulators (e.g., FDA) had been similarly frustrated by a lack of scientific basis for product development, the utilization of innovative manufacturing practises and the reliance of the industry on traditional quality assurance/quality control (QA/QC) strategies. 

Things have changed. Coming out of this dual frustration is the process analytical technology (PAT) initiative, which is driven by both the regulators and the industry. PAT is a framework for innovative pharmaceutical development, manufacturing and QA, and promises to revolutionize the way the industry operates its product development, manufacturing and quality systems. 

Ultimately, PAT is about ensuring consistent product quality, from which springs manufacturing business benefits. Creating consistent product quality means designing in quality at the earliest part of the product and process development phases of a new product. Design for manufacturing, a well-established concept in most manufacturing industries, is now brought into the pharmaceutical manufacturing context via PAT. The key steps in this process are summarized and illustrated in Figure 1.

 Figure 1: The PAT process for creating consistent product quality.						

 Gaining a deep and fundamental understanding of the manufacturing process is at the heart of PAT. This is about truly understanding how processes create products from input materials and what influences the unit operations have over critical product quality attributes. Measuring and understanding sources of variability through the process are often the first steps in this. 

Getting to this understanding is facilitated by the use of PAT principles, tools and techniques, and usually starts with looking at the raw materials used in the formulation and then working through each unit operation before finishing with the final product.  

Once the process is understood from a scientific and engineering context, then the risks to product quality can be identified, quantified and rated. Links between product quality and process quality are established, and the iterative process of steadily gaining more and more knowledge as the new product formulation and manufacturing process are developed is started. Process understanding allows the high-risk process steps to be identified, leading to the identification of critical control points.  

A simple starting point in gaining process understanding is to start at the beginning of any process — knowing and understanding the quality impact of the sources of variation in the raw materials. 			

Traditional approaches to measuring the quality of raw materials are based around identity, chemical purity and pharmacopoeial methodologies. Clearly, using the correct material and it being of a suitable chemical quality is fundamental to creating a superior product, but in the tablet manufacturing process steps (which are essentially based around the handling, mixing, modifying and compression of powdered materials) it will be the physical attributes of raw materials that have substantial process impact.  

It is not unknown for the cause of a failed tablet batch to be traced back to a raw material ingredient physical property issue; the raw material having passed all the current traditional incoming materials inspection process! 

A basic tool that can be used to help process understanding within PAT is FT-near infrared (FT-NIR) spectroscopy.  

Well established as an analytical tool in pharmaceutical testing, FT-NIR has the ability to provide information on both chemical quality attributes and physical properties quickly, easily and without destroying the critical information. Its speed of deployment and operation makes it a fast way of creating a database of raw material information that contains valuable chemical, physical and batch-to-batch variance information, and one that can give rapid paybacks to an organization. 

Figure 2 shows the FT-NIR spectra for a series of lactose raw materials of different grades. Both chemical information, from the absorption peak positions and shapes, and basic particle size information, from the broad baseline and relative offsets of the spectra, are obtained from a single, simple measurement. 

Figure 2:  Process understanding - raw material chemical and physical information from FT-NIR.			

Once a raw material database is established, then links between quality performance later in the process and ingredient FT-NIR spectra can often be established. From this, causal relationships can be explored and investigated, or the FT-NIR system can be deployed with the ability to 'screen' incoming raw materials for suitability in the process. Raw material qualification by FT-NIR is often the first process critical control point. 

From process understanding comes the ability to identify the critical control points (CCPs) — the risk to product quality at those points and the appropriate limits that the process must be within at these points. CCPs in a process are where product quality attributes are created and product quality may be at its highest risk.  			

The points offer a position in the process where appropriate monitoring and active feedback or feed forward control interventions can be deployed to ensure consistent quality. 

Strategic placement of measurement and control systems.

 Once CCPs are identified, then appropriate monitoring and control strategies and techniques can be deployed. The placement of these depends on the type of product being manufactured, the degree of process knowledge in place at that time and the quality risk profile of the process. 

Measurement and control systems can range from simple monitoring of raw material process feedstock (the example used in this article) through to integrated monitoring and active feed forward or feedback process control loops implemented within the manufacturing stages. 

Monitoring systems can be in-line, on-line or at-line depending on the required speed of decision making necessary to intervene effectively in the process. The process knowledge and risk assessments performed earlier in the deployment process will guide decision making here.  

The scientific, technical and business benefits of PAT are manifold, and can be realized throughout the product development and manufacturing value chains. Tablet manufacturing is a challenging process with process steps that have high product risk associated with them (used here as an example process). 

Figure 3: Example of NIR applications in the tablet manufacturing process.						

This example is based around the use of FT-NIR as the principal monitoring and control tool for a tabletting process (Figure 3). The business benefits of this example are summarized in Figure 4.

Figure 4: The benefits of PAT in the product development and manufacturing of tablets.			

A robust manufacturing process that leads to a consistently high quality product is a major business goal for pharmaceutical R&D. Another major objective is meeting the time-to-market challenges of ensuring that the product licence is created and approved as fast as possible, to exploit the revenue generation period given by the patent process. 			

PAT, used correctly, can not only lead to a better manufacturing process, but also speed up the R&D process itself, especially formulation development, scale up, clinical trial manufacture and the technology transfer into the manufacturing part of the business. 

As PAT tools are used in R&D to gain process and product knowledge, the same tools can speed up the development process. Using the FT-NIR deployment example, raw material knowledge can be created rapidly, blenders can be monitored and tablets analysed swiftly — allowing formulation development experiments to be performed faster compared with using traditional laboratory support techniques to analyse blends or finished tablets. Imaging techniques can be used on powder blends or finished tablets to identify causes of failure during formulation development and speed up remediation or optimization. 

PAT also allows R&D teams to gain sufficient knowledge to challenge the current approaches to specification setting. This could have a significant positive impact in defining raw material specifications in terms that represent not only chemical quality, but also attributes that guarantee successful processability and in-process specifications that ensure onward process success.  

As product and process knowledge is gained, and as PAT tools are introduced into product licence applications and rolled out into manufacturing technology, transfer times and costs should reduce. The process is likely to be better understood and more robust than previous generations — and offer less regulatory risk. 

Once implemented in manufacturing facilities, PAT-enabled processes should produce significant business benefits, typical of those seen in other industries adopting widespread monitoring and control, and leveraging it to adopt lean manufacturing approaches. 			

Looking at the tablet manufacturing process example again these would include: 

 Faster testing of incoming raw materials, judged against both material identity and processability criteria. This leads to a faster disposition of the material and release for manufacturing. Exploited effectively this can be part of a just-in-time manufacturing regime and lead to: 					

Lower inventory levels — quarantine stock and manufacturing available stock.

 Higher velocity of raw materials into the process. 

 Lower poor quality and QA testing costs. 

 Smaller storage buildings needed and/or reduced expansion of current facilities. 

 Increased consistency in finished product quality. 

Dryer and blender monitoring and control.

 Monitoring the blender and dyer allows the process steps to be terminated at the optimum product dryness or blend homogeneity for further processing. These are both high-risk processes and process failure here often results in quality failures in the finished product. As well as leading to a more consistent product quality, the batch is released faster and passed on to the next process step quicker. Other benefits can be: 						

 Higher process step velocity/faster stock turns. 

 At-line or in-line tablet analysis to ensure compliance with product conformity specifications offers a near real-time confirmation of a successful blend compression and early warning of problems.  

Used in conjunction with other at-line or in-line physical measurement techniques, most of the critical quality attributes of the finished tablet can be measured in the manufacturing area. Batches can be forwarded to the packing stage faster with higher confidence that any subsequent QA release testing will be within specification, allowing the product to be released to the market. 

Measuring, understanding and ultimately controlling manufacturing processes offers pharmaceutical companies a route to greatly enhance the business effectiveness of both their product development process and facilities. PAT creates technological, business process and regulatory frameworks to enable this. This paper looks at a PAT overview within the context of developing and manufacturing a tablet product, highlighting the potential of FT-NIR spectroscopy.

The Route to Process Understanding and Control

A look at how process analytical technology promises to be the most radical change in pharmaceutical manufacturing in 30 years...