Speed or quality? When it comes to two-step chromatography purification, biopharmaceutical manufacturers want both, despite knowing, realistically, that each choice involves tradeoffs.
With purity, stability, toxicity, processing times, and costs hanging in the balance, the key optimization questions, therefore, are which purification efforts deliver the greatest return and how they can be combined to achieve the ultimate, optimal balance.
A small multinational team of researchers is among the first to address that question with an analytical model “to jointly manage speed-quality tradeoffs and stage-specific lead-time constraints in purification operations,” Yasemin Limon, PhD, assistant professor, Bilkent University, tells GEN. This method guides optimization decisions, helping biomanufacturers decide how aggressively to intervene at each purification step of a serial, two-step chromatographic purification process based upon the costs of the intervention and the time constraints of the purification steps.
The model, based on queueing network theory, captures what the authors call “practically relevant” tradeoffs, correlating intervention efforts, their effects on stability timeframes, and the probability of quality enhancement. It was developed by Limon and colleagues, Tugce Martagan, PhD, associate professor, Northeastern University, and Ananth Krishnamurthy, PhD, professor, Indian Institute of Management Bangalore.
“Understanding how much and at which stations interventions should be applied allows biomanufacturers to optimize system performance without compromising on manufacturing lead times,” the team reports. Thus, the risk of long wait times between steps that may cause product deterioration is reduced.
They divided purification optimization steps into two categories: Type I—those that improve batch quality without increasing purification processing time (such as selecting better resins or reagents)—and Type II—those that increase both batch purity and purification processing times (such as reducing flow rates).
For each category, they evaluated how each optimization affected stage-specific lead-time constraints and how those constraints varied between the two categories of interventions.
Choices are interrelated
“Optimal intervention efforts change with costs,” they acknowledge. Here are the key takeaways:
- Under-investing in upstream purification pushes purification downstream, where increasing the polishing time may risk product stability
- For Type I interventions, put maximum effort into the least expensive options until product stability becomes a constraint
- For Type II interventions, each decision affects both quality and processing times. Characterize process times at each chromatography step and document stability-based time windows to create a reference chart that can be used repeatedly
- Shortening the stability window for step two necessitates more aggressive purification at step one. Fresh time constraints—related to new molecular stability data, for example—should not be evaluated in isolation
- Create a reference map for the range of operating conditions typically encountered in your facilities, along with possible interventions, their costs, and stability-based time effects. Use this as a real-time reference on the manufacturing floor
“The optimal policy depends on costs, processing times, and lead-time constraints,” Limon says. “Decisions at the first and second chromatography steps are interdependent.” Map those effects early to guide decisions in real time.
She recommends turning the model into a decision map. “A manufacturer can estimate its own process parameters (batch arrival rates, processing times at each purification step, stability-based time limits, intervention costs, and the effect of each intervention on quality and processing time) and use the model to identify which intervention policy is optimal under those conditions.
“Distinguish carefully between interventions that improve quality without increasing processing time and interventions that improve quality but slow the process,” Limon continues. “The first type affects lead time mainly through congestion at the downstream step, while the second type directly affects processing time and can make stage-specific lead-time constraints restrictive. Therefore, firms should quantify how interventions change processing time, congestion, and feasibility with respect to stability-based time windows.”
