article

On the edge of glory: how close are we to fully automated continuous biopharma production?

Posted: 3 September 2019 | | No comments yet

We’re getting closer to something special with regard to how we manufacture biopharmaceuticals. Loe Cameron explores where the continuous bioprocessing journey could take us and what it might mean for medicine manufacturers worldwide.

continuous automated bioprocessing image

FEW WOULD SAY the journey to continuous bioprocessing has been easy. Lagging behind other manufacturing sectors including automotive, aerospace and food and beverage, it has taken decades for pharma, and now biopharma, to shift from the status quo of batch processing to a continuous approach for biotech medicine manufacture.

Now that manufacturers are successfully navigating the initial roadblocks (either real or imagined), they are beginning to experience the benefits that continuous bioprocessing brings. Even partial adoption of continuous approaches can improve cost efficiency, reduce CapEx requirements and increase output using a smaller footprint. These improvements in productivity benefit both manufacturers and the patient and could help tackle some of the world’s most pressing medicine shortages.

However, we are nowhere near the end of our continuous bioprocessing journey and the continuous evolution requires ongoing and careful consideration of many questions. What is the next step in our journey? How long will it take to get there? What obstacles stand in our way and what will it take to overcome them?

The sky is the limit for continuous bioprocessing

The logical endpoint for continuous bioprocessing is real-time release – where a product emerges from a fully automated manufacturing line already sampled, tested and ready to be distributed. When we reach the endpoint, the benefits will be beyond measure by current standards. We will be able to carry out qualification assessments mid-manufacturing run, enabling a product to go into the next unit operation without any intermediate holding period. Every hour spent holding a product represents lost revenue, decreased productivity and added risk, so the positive impact of removing these delays will be considerable. Perhaps more importantly, we will rapidly increase the speed to market, without any negative impact on quality or safety.

Every hour spent holding a product represents lost revenue, decreased productivity and added risk

Regulation challenges will be the first thing on everyone’s mind when they consider the implications of real-time release of continuously manufactured products, notwithstanding the fact that regulators are highly enthusiastic about the prospect. In dealings with members of the US Food and Drug Administration (FDA), I have found them to be extremely encouraging. The loud and clear message is that while the onus is on the industry to come up with workable solutions and devise the technological advances to facilitate this, they will take a collaborative approach and work closely to get it right.

Back down to earth

We expect the continuous road we are taking to lead somewhere special, but there are still major obstacles ahead. To overcome them will require investment, innovation and shared commitment from the industry.

The biggest challenge relates to quality testing. The current approach of pulling samples and sending them to an analytical laboratory is simply not quick enough for true real-time release. Even when the analytical lab is on site, it takes at least a couple of hours (sometimes longer) to test a sample and deliver results.

To resolve this, the obvious first step is to embed analytical equipment into the manufacturing suite. But here again, there are technical challenges. Most analytical equipment is highly sensitive and unsuitable for most environments outside of the laboratory. Manufacturing facilities require full washdown capabilities with disinfectant, water, hydrogen peroxide, etc – processes that are far too harsh for a sophisticated piece of analytical equipment 

The current approach of pulling samples and sending them to an analytical laboratory is simply not quick enough for true real-time release.

It is not just a matter of equipment challenges. Some tests, such as bioburden, can take days to complete using current methods, which is not fast enough for a process that is moving continuously through all the unit operations. This is especially critical because the risk of bioburden has the potential to be higher in continuous bioprocessing due to unit operations running for days and weeks, rather than the shorter campaigns seen in batch processing. There are other obstacles too, relating to data and how well we can manage and use information. In real-time release the quantity of data would be vast: the chromatography stage, for example, might have 100 data pools to analyse, compared to the one we have today. It is not viable for humans to process this amount of data – it would require machine-learning algorithms capable of sufficiently advanced multivariate data analysis (MVDA) and modelling, as well as standard approaches to validating these models. Only when the technologies meet these requirements will there be complete trust in the product at the end of the line.

How do we get there?

We need to go back to basics on some of the technologies we are using, both on the testing side using process analytical technologies (PAT) and our machine-learning capabilities.

In testing, it would mean a different approach to equipment design and deployment; for example, creating simpler technologies dedicated to a single type of analysis that could be carried out extremely quickly and without requiring autosampling, so it can be run in-line. On the data side, specialist, bespoke software would be needed that builds on the significant advances seen in other industries, but which is tailored to the specific regulatory and performance requirements of the biopharmaceutical manufacturing sector.

This is the direction in which PAT is headed and where I believe the future of biopharma manufacturing lies. Many in the sector do not believe we’ll see anything close to real-time release in the next five years, but I am not so sure. The analytical technologies and machinelearning capabilities do not currently exist, but the pace of technological advancement is always accelerating and there is huge investment coming in these areas. Indeed, this could be the most significant revolution in medicine manufacture for decades – the possibilities are simply too great for us to ignore.

About the author

Loe Cameron is Senior Director of Analytics & Controls at Pall Corporation, leading development and strategy for the backbone of Industry 4.0 including process analytical technologies (PAT), instrumentation and automation. Her multidisciplinary approach to industry challenges is shaped by a Bioengineering degree and 17 years of biotechnology experience in both technical and business roles.

Send this to a friend