The hidden challenges of cell culture process development

If your shake-flask-to-bioreactor transfer failed the first time, you are not alone. Even when the biology behaves as expected, experienced cell culture labs encounter the same recurring challenges: unexplained drops in viability, inconsistent pH control, CO₂ accumulation that quietly erodes product quality, and process data that does not reproduce at the next vessel or the next site. 

The cells may behave as expected, but what you learn in a shake flask does not necessarily predict how those same cells will behave in a stirred bioreactor. Likewise, what works at bench scale does not always transfer cleanly to pilot scale. Each transition introduces a new set of variables. This guide explains why that happens and what changes when the bioreactor, control software, and process record are treated as a single system.

The shake flask: capable, but not the full picture

Shake flasks remain the right starting point for cell line evaluation, clone selection, and initial media optimization. At this stage, the ability to run many parallel conditions and sample them manually often delivers the best balance between throughput and cost. The Multitron incubator shaker is designed to make this stage as controlled and reproducible as possible, providing precise temperature, CO₂, and humidity management across the full stack, from screening flasks through larger working volumes. Minitron extends this capability to smaller setups for entry-level labs, while Celltron, INFORS HT's shaker system for CO₂ incubators, provides the same controlled environment within a laboratory's existing incubator infrastructure.

Many labs complete substantial portions of their workflow in shake flasks, including early process screening and production activities where parallelism and cost outweigh the limitations of offline sampling. For deeper process development, including feeding strategy development, parameter optimization, and cascade tuning, the bioreactor provides a much more detailed understanding of process behavior. Cell density can be measured quickly with a cell analyzer, but pH, dissolved oxygen, CO₂ accumulation, and metabolite trajectories are not continuously visible in a shake flask. They are sampled periodically, and what happens between measurements remains unknown.

Once a process moves beyond initial screening and into feeding strategy development, parameter optimization, or transfer activities, the lack of continuous process data quickly becomes a bottleneck. The transition to a bioreactor is not simply a scale-up step. It represents a move from limited process visibility to a fully instrumented environment, significantly increasing the amount of information gained from every run.

Three parameters that get worse before they get better

When mammalian cells (CHO, HEK, Vero, and others) enter a stirred bioreactor for the first time, three parameters tend to behave unexpectedly.

Batch, fed-batch, or perfusion: choosing the right process mode

Once the physical environment is under control, the question shifts to feeding strategy. This is where the bioreactor earns its place in a cell culture process development workflow.

Fed-batch remains the dominant cultivation strategy in mammalian cell culture and is used for most therapeutic protein production. As a result, it accounts for much of today's bench-top process development activity. Perfusion has become an established cultivation strategy and continues to gain momentum. It is widely used for high-cell-density processes, reducing manufacturing footprints, and supporting continuous biomanufacturing approaches.

At bench scale, the practical challenges of perfusion often involve selecting the right cell retention method, configuring pumps, maintaining aseptic operation during extended runs, and integrating continuous feeding strategies. Across all three modes, the same principle applies: process development becomes more valuable when more conditions can be tested in parallel, with more instrumentation and more complete data capture.

Reproducibility is a system property, not a protocol property

This is where many cell culture process development conversations become difficult. Labs document protocols. They define setpoints. They train operators. Yet variability still appears:

• Between operators
• Between runs
• Between laboratories
• Between sites

In most cases, the root cause is not the setpoint itself. It is everything surrounding the setpoint, including cascade configuration, PID tuning, sensor calibration, and hardware assignment. Two operators can run the same protocol on the same bioreactor and achieve different outcomes if their controller configurations differ, their sensor calibrations are inconsistent, or their gas-control strategies are configured differently. None of those details are fully captured in a paper protocol. All of them affect the data. What needs to be managed, versioned, audited, and transferred is the complete controller configuration together with the process record. That is what makes a process development result reproducible across operators, runs, and sites.

Beyond data capture, the platform helps reduce operator variability. When controller configurations are versioned and applied automatically, one of the most common sources of run-to-run variability, human error during cascade or sensor setup, is significantly reduced. Increased automation reduces the number of manual interventions required, which in turn decreases opportunities for operators to unintentionally deviate from the same protocol.

Reproducibility is not achieved through a single feature. It is the result of cultivation hardware, control logic, and data systems working together as an integrated platform. In cell culture process development, where small configuration differences can have a significant impact on outcomes, that level of integration often determines whether a process scales successfully or requires rework at each transition.

What the full development path looks like in practice

A typical cell culture process development pathway in a biopharma, CDMO, or advanced therapy lab follows three key stages. Each stage builds on the data, control strategy, and process understanding developed in the previous one.

Where to go next

If you are working through a specific stage of the pathway above, these resources provide a useful next step.