Speeding up the move to PAT

Published on: 

Pharmaceutical Technology Europe

Pharmaceutical Technology Europe, Pharmaceutical Technology Europe-07-01-2006, Volume 18, Issue 7

The big challenge for pharmaceutical companies is to mathematically model their production processes, combining both data from the laboratory and the production process.

The era of process analytical technology (PAT) in pharmaceutical manufacturing is upon us.1 Many companies are high up on the overall PAT learning curve, but all are grappling with the more critical challenge of determining the best way to introduce PAT and to select the right tools and infrastructure.

Companies need to address wider business issues to determine the answers to some of the key questions that arise when introducing PAT: which facilities to prioritize? How best to phase investment? How to achieve the best product quality and mitigate risks? How to migrate the control infrastructure? How best to scale-up? Determining the answers to these questions is vital for companies as they face the challenge of implementing PAT.

The impetus for PAT stems primarily from FDA guidance issued in its final form in September 2004. In many respects, the pharmaceutical industry is way behind other industries in its adoption of PAT. This is understandable in a sector where much is conditioned by the regulatory environment. The proactive stance of FDA is only recent and the regulatory momentum in other territories is less driven. While European regulation already contains the concept of PAT and the EU has established a PAT team, the EU is expecting the industry to lead rather than taking the more pro-active stance of FDA.

Key benefits of PAT

However, companies that delay PAT implementation risk being left behind on the business as well as the regulatory front. While regulation may be the catalyst it is competitive advantage that is the compelling prize. Early adopters can secure cost and competitive advantage gains. For an existing production process, PAT adoption and implementation tends to be seen in terms of reduced cost, lower inventory levels and a move towards 'just in time' production and supply.

For new processes the benefit of PAT is the ability to quickly develop the manufacturing process, upscale to a robust process and perform validation more easily. These are tremendously important, but the full competitive advantage potential also needs to be seen in the context of the end-consumer or patient.

The 'process understanding' that lies at the heart of PAT means that manufacturers can gain improved knowledge and production capability to create products to match the quality and therapeutic expectations of patients. Bioavailability is a critical patient variable but it is not a current production variable. PAT enables manufacturers to gain improved control of their product quality but it also enhances a company's ability to better match different patient needs. This is significant because bioavailability differences are, of course, important for drugs with a relatively narrow therapeutic index (TI) and also because of the potential of PAT to deliver competitive advantage by influencing new product development.

Decision-making in PAT

However, the prize of longer term gain comes at significant short-term hazard and cost. PAT investment comes at a time when the pharmaceutical sector in general is facing significant pressure to reduce costs to maintain profitability levels. Blockbusters are rarer and the industry faces the twin challenges of price pressure from consumers and health buyers, and competitive pressure from generics. Introducing PAT investment and change in a well judged and timely manner is crucial. How can companies feel more confident and certain about their PAT journey?

A clear understanding of the business goals of PAT implementation is a vital starting point. There are many points to consider and their exact fit with a particular company will be dependent on the company's current manufacturing base and its future strategy.

PAT can fix or improve existing processes, speed up new product development, reduce site-to-site transfer risk and times, reduce validation costs and, through quality reliability, enhance a company's image. At the end of a drug's life cycle, PAT can help prolong patent life through the development of new formulations.

Most companies will want to realize a blend of these business benefits but, at the same time, they will also want to take a view on which one is their most important priority. This will be determined by the current state of play of its manufacturing and wider market activities. In turn, a company's decision on which business benefit to prioritize will determine the focus and pace of PAT investment.

A company's view on the business benefits it wants to prioritize will be the lens through which it evaluates its return on investment (ROI) in PAT. For example, a company may want to focus on improvements in existing processes. Some companies choose an easy and stable production process for PAT introduction.

However, this may not be the choice that will yield the quickest benefits. An alternative focus might be on a troublesome process where the gains of better process understanding and control will be far higher.


Other companies may be at a stage in their product portfolio where they may want to make new drug development the focus of their PAT investment. For these companies, the potential gains in revenue of using PAT to speed the time to market of drugs in the development pipeline may be greater than prioritizing the application of PAT to the manufacture of existing drugs.

Each company's situation will be different and judgements on the focus and pace of PAT implementation will vary according to the ROI analysis of the different options open to them. Companies will need to carry out ROI analysis of all options to validate their prior assumptions about the lead business benefits.

In some cases, they may decide that these prior assumptions need to be adjusted with some options, on examination, emerging as better contenders than previously expected. Such an examination of priorities is crucial since one of the key issues for companies is where and how best to start implementing PAT.

Start small and scale-up later

Some companies are so overwhelmed by the regulatory guidance and the nature of the change that they don't always remember that you can easily start small and gradually evolve towards a comprehensive PAT infrastructure.

Identifying the optimal starting location and purpose is part of a prioritization exercise that not only maximizes ROI but makes the introduction of PAT manageable. The importance of starting small and then scaling up enables companies to avoid very high investments early on that may prove not to have the best fit with their overall plan.

The 'starting small and scaling up approach' needs to be guided both by the ROI analysis discussed above and a clear step by step programme to build PAT capability. In the rest of this article we examine eight key implementation steps.

Step 1: Data mining and investigation

Data is the bedrock of PAT. Typically, companies will already possess a wealth of data. Throughout the manufacturing organization, important data is captured by various tools, from routine logbooks and spreadsheet files, more sophisticated data managers to laboratory information management systems (LIMS) and process historian databases. With data of previous trials or clinical/manufacturing batches available, multivariate data analysis (MVDA) tools can be applied to discover relationships and interdependencies. This information, in turn, is used to design the PAT parameters. PAT implementation costs may be reduced if companies examine the wealth of data they already have, reducing the need for data collection.

The big challenge for pharmaceutical companies is to mathematically model their production processes, combining both data from the laboratory and the production process. Part of the challenge is to bring the right skills and experience together, including knowledge of production processes, chemometrics, software, laboratory chemists and, of course, the hardware connections with process equipment and spectroscopic equipment.

Step 2: Design of experiment

The insight and understanding from step 1 allows companies to establish a dedicated design of experiment (DoE). By varying multiple parameters at the same time, the 'design space' for a particular process can be explored and identified, leading to a better understanding of process behaviour and its impact on product quality and performance.

This delivers significant regulatory advantage for companies. Once the design space is established and agreed with the regulator, companies have greater flexibility. Because the design space has already been thoroughly explored and tested, they do not have to get further approvals if they subsequently want to make changes to the control space as long any moves they make are within the design space (Figure 1).

Figure 1 The process understanding context.

Step 3: Additional monitoring

The next step is to check whether new sensors need to be added to the process to get additional information in order to have a real-time picture of the process. Process analysers can be positioned as antenna reflecting the actual status of the process. Possibly more than one type of analyser could be installed to allow for a full picture of the process and increase process knowledge.

Hence it is crucial to select the optimal one or select the appropriate combination. Companies need to be careful. We have, for example, seen simple problems arise such as investment in the wrong type of probes. One way to avoid mistakes is to simply talk to more people, including companies in other industries such as chemicals and even food and beverages who have a great deal of experience in installation of PAT systems.

At regular intervals the process analyser is delivering data that has to be stored, managed and processed (through MVDA) to extract the meaningful information. A PAT data management system will organize this data traffic and serve as interface to the analyser and the MVDA tool on the one side and the process control system on the other side.

Another task is to monitor the analyser performance to avoid making decisions on results generated by an erratic analyser. The data management system will also deal with multiple data flows coming from various analysers and sensors and even different unit operations.

Step 4: Integration of the PAT solution into the control landscape

Most pharmaceutical companies in Europe have installed analysers and are trying to build mathematical models but these are not yet linked to control. Their next step is to interface the PAT solution with the process control and automation infrastructure.

When the design space and more detailed control space is determined, the optimal process envelope can also be determined and critical process parameters defined. A fully automated control strategy can then be established that excludes human intervention and potential sources for errors.

Step 5: Batch-to-batch comparison

Because the batch profile has been identified and characterized, each batch can be monitored on-line and compared with previous batches. This newly generated information allows acceptable process variance to be defined, making the control space more robust and further developing the boundaries of the production envelope.

The sensors and the use of MVDA in process control allow diagnosis of process operating deviations from the optimal derived trajectory. This automated supervision enables the manufacturing process to be kept within the defined boundaries of the trajectories so producing a consistent product quality and in-built process quality control. More intelligence on quality decision-making is built into the control system.

Step 6: Real-time decision and release

This is the stage at which companies can move to immediate release of the batch to market straight from production. The need for laboratory tests is removed since, by reaching this stage, companies are able to have complete control over the production process.

The in-line sensors, data analysis and control infrastructure, in effect, mean that the quality checking task, previously performed by the laboratory, is built into the production control system. Not only does this speed manufacturing throughput, but the continuous, real-time control allows much greater production flexibility and the use of continuous manufacturing. It also allows companies to introduce innovative manufacturing strategies.

Step 7: Integrated quality system and transparent manufacturing

Once process understanding and associated real-time release is accomplished and automated feedback and feed forward controls are in place, one can adjust the other aspects of supply chain, like just-in-time raw material purchasing, as well as final product order processing.

A company will be able to reduce inventories and adjust its planning quicker and hence be more agile in reacting to a changing market. The whole pharmaceutical business model and related quality system can now be optimized and the company can manage its working capital better. Quality is no longer tested after the event but is part of the process of production.

Step 8: Continuous optimization and improvement

The final stage is an ongoing one with a company using the increased process knowledge and understanding, resulting in optimized processes in terms of throughput, yield, asset utilization and utility consumption. Continuous collection of data and integration of the data into further optimization process models will allow more accurate and consistent quality of the final product.


Pharmaceutical managers, particularly those running manufacturing operations, face some tough decisions and dilemmas. We have looked at how they might meet the technological challenge of integrating PAT with existing infrastructure. A focus on the lead business benefits is key together with the realization that you can start small and scale-up later.

Allied to this, one of the biggest challenges facing manufacturing decision-makers is how to convince top management. Again, this is where an unrelenting focus on the business case is vital. This allows greater certainty around which facilities and processes to prioritize for PAT and a clearer roadmap for ROI measurement.

Manufacturing managers need also to be mindful of the opportunities for flexibility that an ultimate move to PAT and continuous production can deliver. The investment capital needed to put another smaller continuous manufacturing line in place, for example, is small compared to struggling to get the same output from an oversized discontinuous batch production facility as many companies do currently.

Finally, pharmaceutical companies should not be shy of learning lessons and talking directly to companies in other industries such as food or petrochemicals, who have learned how to implement PAT-like principles in the past. Such dialogue can be helpful in avoiding mistakes. These industries have long been transformed by PAT in a way that pharmaceutical manufacturing is set to be over the next five years.

Ingrid Maes is senior consultant in the Pharma Competence Center of Siemens A&D, Belgium.

Beatrijs Van Liedekerke is associate director at PricewaterhouseCoopers, China.


1. US Food and Drug Administration (5600 Fishers Lane, Rockville, MA 20857, USA) A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance, September 2004.