Data, machine learning and the road to continuous bioprocessing

Martin Smith, chief technology officer of Pall Corporation, explores how data is finally unlocking the door to continuous bioprocessing in medicine manufacture.

The journey from tried and tested batch manufacture to continuous bioprocessing has been a long and gradual process. It began around two decades ago when manufacturers started to shift from stainless steel systems to single-use components, eventually leading to the introduction of modular systems – but the pace of change has been slow.

Pharmaceutical manufacturers have historically been wary of departing from batch processing – largely because of perceived risk of contamination and control issues, large investments of capital already made in stainless steel plants, and lingering concerns around regulatory barriers to process change.

Now, finally, things are changing. Our burgeoning ability to harness the power of data is giving us means by which to demonstrate the remarkable cost-efficiency, consistency and scalability of continuous bioprocessing, while eradicating the perceived disadvantages of regulatory risk and compliance challenges. 

With increased cost pressures, greater competition and public criticism for medicine shortages, manufacturers have had to stop and consider the huge potential benefits that continuous processes can bring. Investing in continuous manufacturing methods is now critical to longevity, profitability and survival, as well as the key to finding a more efficient way to ensure the right patient receives the right medicine, at the right time.

Playing catch up

A wide range of industries have experienced the benefits of continuous for many years, including increased efficacy and reliability, reduced capital costs and failure rates, and improved process control.

Data and technology have expedited the pace of change within industries such as food and automotive. While automotive manufacturers have been embracing Industry 4.0 for many years now, including utilising AI, robotics and data analytics technologies as a matter of course, these transformative technologies are still nascent in biopharma. The pharmaceutical laboratory has remained largely unchanged for decades; only now is the industry genuinely talking about the ‘Lab of the Future’ with real conviction.

This has hindered the sector’s move to continuous processes. As was the case with single-use components, pharmaceutical manufacturers want the reassurance of hard data and evidence of scalability before they make a change. But since adoption of cutting-edge technology has been slow, data on continuous exists on a very small scale, making it a challenge to demonstrate production-scale viability. This cycle has been hard to break.

Continuous chromatography

Nothing demonstrates value of the continuous approach more than the possible improvements it can bring to chromatography. Chromatography systems have been critical in the production of new medicines for many years. The technology is well established and many manufacturers have been reluctant to go against the status quo, but the potential for huge cost savings are driving a major shift to continuous chromatology.

Chromatography is usually the most cost-intensive aspect of downstream bioprocessing. Protein A sorbents used in the primary capture step, for example, can cost over $10,000 per litre. However, by optimising the number of columns needed to operate this process, manufacturers can enable reductions in buffer and decrease the volume of sorbent used by up to 90%. This vastly improves the efficiency of consumable use, while also reducing the need for large tanks, buffer-storage bags and other equipment – all of which takes considerable cost out of the manufacturing process.

Of course, this is often easier said than done. Column numbers and configurations need to be optimised on a case-by-case basis, and every bioprocess is different. Many different factors including binding capacity (which permits reduction of buffer consumption), productivity (which enables minimisation of sorbent use) and workflow (which minimises time) impact the decision and need careful consideration to achieve the best results.

To determine column numbers and configurations effectively, you need data. At Pall, we have developed a number of tools to calculate the optimal column numbers from any given parameters, enabling us to predict the best configuration for a particular situation.

Unlocking the future

Data analytics and monitoring, as well the application of cutting-edge automation have a key role to play when it comes to regulatory issues too.

There are several perceived risks that have hindered uptake in the past. In continuous processing, for example, unit operations mostly run for longer periods than in batch systems, which raises questions about bioburden management and process consistency. There are also concerns that the use of more complex instrumentation in continuous bioprocessing could generate additional risk of equipment failure.

These fears, while understandable, do not bear scrutiny. From a regulatory perspective, continuous processing is inherently no riskier than batch processing – indeed, the current regulations make no distinction between batch and continuous processing. And on the subject of complexity, the sophisticated nature of continuous systems reduces risk, rather than creating it: the quality of data that manufacturers can generate around process parameters using these advanced instruments makes it infinitely easier to show consistent operation within acceptable ranges.

These robust datasets can speed up the process for clinical approvals and have the potential to form the foundation of a new approach to regulation. Instead of the classic model of regulators approving each stage of batch processing, we could see the emergence of a more seamless, integrated drug approval process, improving speed to market.

Most importantly, it is patients that will ultimately feel the true benefit of an industry-wide shift to continuous bioprocessing. Where traditional batch processing could take up to seven weeks, continuous bioprocessing could deliver the same product in just one week. This vast improvement in the speed at which biopharmaceutical companies can evaluate the viability of a new product, produce clinical trial material and reach potential failure points will help tackle some of the world’s most pressing medicine shortages. This, coupled with the remarkable financial benefits manufacturers can realise, makes going continuous a question of when not if, and something the industry can no longer afford to ignore.