Upstream advances and evolving antibody formats are reshaping bioprocessing at a fundamental level. As these changes move downstream, they are placing new and often competing demands on purification strategies—particularly in chromatography polishing.
Antibody diversification and process intensification are reshaping downstream processing
As upstream productivity continues to increase and antibody formats become more diverse, chromatography polishing is facing a new set of demands. What worked well for platform monoclonal antibody (mAb) processes in the past is not always enough for today’s molecules, where higher titers, greater structural complexity, and more challenging impurities are changing the downstream landscape.
In a recent Cytiva panel discussion, I was joined by Eva Heldin (Principal Scientist), Niklas Jungnelius (Director of Process Modeling), and Jakob Liderfelt (Global Program Leader) to explore how these shifts are affecting downstream processing. Their discussion highlighted a clear theme: polishing is becoming a more critical point of differentiation in process design, not only for impurity clearance, but also for maintaining productivity and process economy going forward.
When intensification moves the bottleneck downstream
One of the panel’s starting points was the impact of process intensification. Over the past 10 to 15 yr, the industry has seen steadily increasing titers upstream, along with a broader range of molecules entering development. While these changes are positive in many respects, they also place greater pressure on downstream purification from both a process intensification and selectivity perspective.
As Niklas Jungnelius explained, the effect on polishing depends on the type of intensified process being used. In fully connected continuous processing, where the output from the bioreactor sets the pace, polishing may not always be a productivity limiting step. But in more conventional intensified processes with higher titers and larger mass loads coming downstream, polishing can quickly become a bottleneck. More product mass, coupled with higher concentrations after capture, increases the burden on subsequent steps. Simply scaling up the column or adding cycles is not always practical, particularly when process time is constrained by upstream cadence.
The panel made an important distinction here: the strategies that have supported performance gains in the capture steps do not necessarily translate directly to the polishing steps. Running at higher dynamic binding capacities or pushing for faster operation may be effective in capture, whereas in polishing the same approaches can compromise resolution and lead to losses in both purity and yield. In other words, optimizing one part of the process may simply shift the bottleneck elsewhere if the entire process is not considered more holistically.
Selectivity remains central
If productivity is one pressure point, selectivity is the other. Eva Heldin noted that polishing has always been different from capture because the challenges are not only to bind product efficiently, but to discriminate between the product and the impurities that remain after capture. That requires careful process design, including decisions around step order, loading strategy, and acceptable trade-offs between yield and purity.
This is one reason why polishing cannot be reduced to a simple “capacity” question. While higher-capacity resins can help, the full capacity is not always usable when separation performance must be maintained. A process developer needs to determine where the optimum lies and how much can be loaded before performance begins to decline. The right answer will depend on the molecule, the impurity profile, and the economics of the overall process.
The panel also discussed the flow-through polishing mode, which continues to attract strong interest thanks to its potential to simplify operations, reduce column size, and improve processing speed. Eva Heldin described flow-through as a promising polishing approach, but also emphasized that its success depends entirely on selectivity. If the target molecule can pass through while impurities bind, the step can be highly efficient. However, that is not always achievable, particularly for more complex molecules where impurities closely resemble the product. For that reason, bind-elute polishing is unlikely to disappear. In many cases, combining flow-through and bind-elute steps may be the most effective way to address specific purification challenges.
New molecule formats are raising the bar
The increasing diversity of antibody-based therapeutics was another major theme of the discussion. While standard mAbs still represent an important part of the market, the growth of bispecific antibodies and other antibody variants are introducing new purification challenges. According to Jakob Liderfelt, these more complex molecules often generate product-related impurities such as unwanted homodimers or heterodimers that have properties very similar to those of the target molecule. That makes separation significantly more difficult.
At the same time, high-expression systems can also increase levels of free light chain or other species that are difficult to remove during polishing. Host cell proteins remain an additional concern, particularly those that can affect final product stability or shelf life. Some of these impurities are also time-consuming to analyze, which adds another layer of complexity to development and process verification.
The panel’s view was that future solutions will likely involve both upstream and downstream innovation. Improved cell lines and expression systems may help reduce some impurity issues at source, but the complexity of emerging antibody formats means that robust polishing will remain essential. For many molecules, polishing will still be the step that determines whether the required purity can be achieved efficiently.
Moving beyond the platform mindset
A particularly relevant point for process developers was the changing role of platform processes. For many conventional mAbs, platform purification strategies have enabled rapid progression with only limited adjustments. But as antibody properties become more engineered and less “standard”, that model is becoming harder to apply. Jakob Liderfelt pointed out that today’s molecules are more likely to differ in characteristics such as hydrophobicity, which can force teams to move away from familiar platform conditions and carry out more extensive resin and condition screening to find the optimal process conditions.
This has clear implications for process development timelines. The panel noted that some developers are already screening very large numbers of molecules each week to identify suitable purification conditions early enough to support clinical progress. In that context, process development itself must become more efficient. Eva Heldin highlighted the value of miniaturized formats, such as 96-well plates and small prepacked columns, as well as design of experiments and mechanistic modeling to support higher-throughput development. At the same time, she noted that analytics remain a limiting factor: running many experiments is only useful if the resulting fractions can be analyzed quickly enough to guide decisions.
A broader toolbox—and the need to use it well
The panel returned several times to the idea of a broader chromatography toolbox. Where standard ion exchange approaches are not selective enough, multimodal chromatography can offer additional interaction modes that improve the separation. The discussion specifically referenced Capto™ adhere and Capto MMC ImpRes resins as examples of resins designed to provide different selectivity options.
What matters, however, is not only having access to more tools, but also understanding how to apply them. Buffer composition, pH, conductivity, and loading conditions all influence performance, and finding the right operating window often depends on smart screening rather than overly complex experimentation. The panel’s perspective was that the challenge is increasingly to identify the right tool for the specific molecule and impurity problem at hand.
Looking ahead, the group also touched on the possible role of machine learning and artificial intelligence (AI) in helping process developers navigate this complexity. The expectation was cautiously positive: these technologies are likely to become useful in guiding resin and condition selection, but only as the underlying data foundation matures. Until then, developers will continue to rely on a combination of process understanding, targeted experimentation, and practical experience.
Polishing will remain a strategic step
If there was one overarching conclusion from the panel, it was that polishing is no longer a routine downstream step that simply follows a familiar template. It is becoming a strategic part of process development, especially as molecule diversity increases and productivity targets continue to rise.
Maintaining impurity clearance, yield, throughput, and cost-effectiveness under those conditions requires a more selective, more flexible, and more data-driven approach. In many cases, it is within the polishing steps that the balance between purity and productivity must be achieved.
For process developers, that means looking beyond standard workflows and focusing on how polishing can be designed to address the specific challenges presented by each molecule. As a result, polishing is increasingly recognized as a process step where critical process trade-offs have to be resolved and where new opportunities for improvement can be realized.
In a fast-changing therapeutic landscape, the ability to rethink polishing strategies may be what determines the success of a process in terms of performance and economic goals—highlighting the need for continued innovation in how polishing is approached.
Additional resources
CY60597