Avoiding pitfalls in scaling up biopharmaceutical production

All biotechnologists appreciate the complexity of developing production processes for recombinant proteins and the intimate link between the process and the product. However, the development of the protein and its laboratory-scale production is just the beginning for a biotechnology company with ambition beyond discovery. Taking a biomanufacturing process from laboratory to clinical production can be a precarious journey, littered with a multitude of possible pitfalls, some of which could prove fatal.

Companies seeking to scale-up their biopharmaceutical production need to be aware of the potential pitfalls so as to avoid them — not only to provide a successful scale-up, but also to ensure that failure does not lead to significant financial loss. Successful scale-up requires careful planning to anticipate all possible issues, from process design and associated costs, to raw materials and regulatory requirements. This is particularly true for biopharmaceutical companies looking to move their pipeline of monoclonal antibodies (mAbs) beyond discovery to therapeutic applications.

Benefits and requirements

There are several reasons why a company would seek to scale-up its production process. Most obviously, there is a requirement for larger quantities of the protein; for example, to perform appropriate preclinical toxicology tests and advance to clinical development. Later in development, as a company looks to commercial supply, it will start to think of how to scale-up its production process specifically to meet the quantities predicted by market demand. At the same time, there are cost of goods benefits associated with large-scale production.

If an existing process operates at low product productivity then a company will also often be making process improvements as part of its ongoing process development strategy. The large-scale purification of antibodies as drug products is a highly complex process that often involves numerous unit operations, which must be sequenced and integrated using a rational approach that maintains the requirements of purity, yield and throughput. This process development is an important part of the never-ending search for better results, better productivity and reduced cost of goods.

However, it must be remembered that a good bioprocess development scientist's motto will often be "how good is good enough?" and care must be taken to ensure that this 'never-ending search' itself does not hinder manufacturing and scaling-up processes. It is particularly valuable to bear this in mind at the planning stage when development targets are being set.

Three key principles

We generally follow three key principles when designing large-scale processes for novel mAbs:

• Design for scalability at the beginning of the development process.

• Tailor a generic purification scheme.

• Keep it simple.

Designing for scalability. It is possible to choose from a range of different strategies for upstream and downstream processing; thus, if the ultimate goal is commercial supply then it is valuable and important to start with this large-scale process objective in mind. It is also important to remember that the design of a large-scale manufacturing process for a biopharmaceutical is different to the creation of processes in the laboratory in early stage R&D. Although this important concept may seem obvious, it is often neglected when taking a process from the bench to clinical production and beyond. Aspects that are clearly different include the quantities required (scale), regulatory requirements, raw materials, equipment, process costs (economics) and formulation, and stability of the final product.

In the scale-up of a manufacturing process, you first need to ask: "How much product is needed?" This clearly depends upon the potency of the antibody, the proposed clinical indication (patient population), the dose and the route of administration. Often, this information is not available when entering a scale-up programme for the manufacturing process. However, it is important to understand the scale required when creating a strategy. For example, gel filtration chromatography (GFC) is limited compared with other modes of chromatography in its throughput and capacity. Thus, if the process uses this step, alternatives may have to be found before production at very large scale can be justified economically.

The equipment and methods used at scale clearly have to be very different from those used within a laboratory environment and need to be considered at the outset. An obvious example is the method of clarification of fermenter harvests. Filtration is frequently used in laboratories and is scalable to a degree, but commercialized biotechnology processes create new challenges in filtration technology. Centrifugation may produce efficient economies of scale and may be the unit operation of choice for large-scale processes. It is important, therefore, to understand the scale-up and validation issues associated with centrifuges early in the development of a manufacturing process.

Another substantial difference between processes used in therapeutic production and those in research is the requirement of the manufacturer to follow a framework of regulatory requirements for the production and quality control of the biologic being produced. These legal requirements define the steps in the development process that take a protein therapeutic into large-scale production. Consequently, process design must refer to these requirements continually during the development phase so as to ensure that the process being created is capable of meeting them. This is often neglected in the research environment and is a clear distinction between laboratory and manufacturing.

Customizing generic purification schemes. The similarities between mAbs make the use of a generic purification process the obvious choice when starting development and scale-up of a process for a novel antibody. Typical large-scale purification processes are built around the use of immobilized Protein A as the primary capture and purification step combined with ion exchange chromatography (IEC) for polishing and ultrafiltration for the formulation stages.

However, in scale-up, it must be remembered that the generic process is just a guideline and cannot always be followed explicitly. For example, generic elution conditions from Protein A, followed by low pH viral inactivation and subsequent neutralization often involve significant volume increases to manage the required changes in pH. On scale-up this can result in volumes that are difficult to handle or even in extreme cases are too large for existing vessels. Thus, 'fit (of process) to plant' is an important concept to bear in mind, even when working with generic processes. That is, small changes can have large consequences on scale-up and the majority of problems are simply caused by scale-up of volumes and the implications that this has for liquid handling at scale.

Keep it simple. Economic pressures on processes to increase speed, efficiency and yield have, unsurprisingly, had a big influence on process scale-up during recent years. Clearly, there are advantages in using fewer steps and in increasing the benefit of each of these steps. This reduction in complexity generates improvements to yields, increases process robustness (as there are fewer things to go wrong in manufacturing!) and reduces the overall processing time.

Process intensification (PI) can be achieved with rationalization and reduction of unnecessary steps such as ultrafiltration for conditioning of chromatography loads or unnecessary concentration steps. The re-ordering of process steps can often result in feed streams that are suitable for the next chromatography step without further modification.

By redesigning existing generic strategies, we can:

• Remove unnecessary concentration and diafiltration steps.

• Reduce buffer volumes.

• Increase column capacities and flow rates.

• Utilize a common buffer system for all steps.

• Replace flow-through polishing IEC with charged membrane filtration.

These relatively simple changes can have a dramatic effect on costs and recovery, but more importantly for scale-up, they add simplicity and process robustness.

Managing risk and change

Scaling up, in itself, is a process change. As manufacturing processes for protein therapeutics are both complex and critical in defining the product (as the process is the product) any process change can affect the product, including scale-up. Thus, scale-up can affect the product. This is something that must not be underestimated — many biotechnologists will be familiar with examples of how changes in process-scale alone have caused unexpected effects to the product. This unpredictability is a challenge to manage as great effort is generally taken in scale-up not to introduce process conditions or parameters that will impact the product and thus, most of the differences encountered are unexpected.

The unpredictable nature of scale-up can be reduced by spending time understanding the process, which leads to a better understanding of the product. This is facilitated through the use of generic processes as discussed previously — familiar processes with familiar outcomes gives the production scientist a sense of confidence when running large-scale production for the first time. Additionally, building a body of experience with the process also helps when things go wrong because these are often situations that have been encountered before and have common solutions.

The desire to build confidence in processes as they are scaled up has also led to the use of computer-based process models as an important tool for the downstream process developer. We use a detailed computer model that encompasses process parameters, equipment, facility and materials details and flows, to build up a manufacturing model that simulates real-time process operation. This allows examination of changes in activity levels or scale in a facility to observe where constraints may appear. These may be related to equipment, people, space, utilities or other resources. As these activities do not necessarily take place in a smooth continuous manner, identifying the impact of intermittent high-level activities using this type of process simulation can indicate peaks in resource and constraints of the facility (and their impact). This type of modelling is highly valuable in process scale-up as it predicts current plant capacity, identifies any facility constraints, helps with identification of cost effective solutions and identifies when changes have to be implemented to accommodate the scaled process.

Although change is inevitable with scale-up, changing a process too frequently should be avoided. Although this seems obvious at first glance, it leads to an unexpected conclusion, namely, save up the changes and make all (well-considered) changes in one go. The benefit of this approach is the considerable saving of effort associated with the quality and regulatory aspects of the process. For example, process change or scale-up might require a repeat of a process viral clearance study or of a product stability study, which can be lengthy and costly to perform. However, some caution is required and we suggest that in combining process change the synergies of all changes in combination should be well considered and tested at pilot scale.

Failure in the scale-up of a process during product development is often not an option for a company. Failed batches, with their cost and time implications, can be crippling and may delay clinical development. Hence the importance of concentrating on the basics when scaling up to ensure that the process is well-thought-out, robust and reproducible. Thus, good risk management and change control procedures are important elements of scale-up. Process engineers must evaluate all the risks in change-in-process and change-in-scale.

Surprise

A company with a well-characterized and understood generic process, combined with a suitable cell line, might hope to have no surprises during scale-up. However, the reality is that manufacturing biopharmaceuticals is complicated, and it is 'nature' that does all the hard work. The biotechnologist simply exploits cells to do what they do naturally — synthesize proteins. The task of the biotechnologist is to control the cellular growth and biochemistry to produce the desired protein, to recover that protein as a highly purified entity and to formulate it as a stable medicine in a way that can be safely delivered to the patient. All of which can lead to unexpected events and surprises. A thorough understanding of your process and your product, including any critical process parameters, can help to limit the amount of damage caused by these surprises.

Conclusions

Biomanufacturing is a complex process and scale-up of such processes is not always straightforward and without incident. Minimizing risks of process failure on scale-up is supported by taking a holistic approach to the whole process scale-up through attention to designing in scalability and fit to plant from day one, keeping processes and scale-up simple, and a thorough understanding of your process and the potential impact of any change on the product scale.

Key Points

• Biomanufacturing is a complex process and scale-up of such processes is not always straightforward and without incident.

• Understanding your process and the key process parameters will help manage uncertainty on scale-up.

• Thinking about scalability early in development, tailoring a generic process and keeping things simple are key principles that support rapid and low-risk process scale-up.

Dr Brendan Fish is director of bioprocess sciences, biopharmaceutical development at Cambridge Antibody Technology (UK). He has spent 15 years working in the bioprocess industry on purification development, formulation and analytical characterisation for a variety of biologics entering commercial pharmaceutical manufacturing. Before joining CAT, Brendan held positions at Delta Biotechnology Ltd and the University of Toronto (Canada).

Dr Richard Williams is manufacturing manager at Cambridge Antibody Technology (UK). He has spent 8 years working in the biotechnology industry on CMC project management and manufacturing, scale-up/technology transfer, and upstream/downstream process development. Prior to joining CAT, Richard held positions at Schering Plough Animal Health and Acambis Ltd working on a variety of programmes ranging from live vaccines to monoclonal antibodies.