Worst-case scenario simulations ensure manufacturing is prepared for all contingencies, but over-sizing or under-sizing may ensue. This results in larger than necessary filters and columns that may limit the ability to load minimum amounts of protein for some operations, impair filtration, reduce yields, and increase the cost of resins, filters, and other consumables. Conversely, under-sizing may cause excessive numbers of chromatography cycles, which increases the time needed to complete a batch.
By using Monte Carlo simulation to model a monoclonal antibody downstream manufacturing process to improve facility fit, Bristol Myers Squibb (BMS) avoided those risks while improving consumable usage and lowering manufacturing costs.
The study, reported in Biotechnology Progress, compared Monte Carlo simulation to a worst-case scenario approach in right-sizing facilities for downstream manufacturing and tech transfer.
“The Monte Carlo method computes outcomes based on random sampling of defined probability distributions for process inputs…and was used to provide a more realistic assessment of expected mass and volume at each step of the process,” first author Christopher Furcht, PhD, associate director of downstream process development, BMS, says. “It reduces the propensity to oversize downstream unit operations when accounting for the highest possible protein mass that could be generated by the bioreactor, which is generally unlikely to occur.”
Reduced consumable costs and buffer utilization
By right-sizing the chromatography and filtration steps, BMS reduced consumable costs and buffer utilization. For perspective, when applied to protein A column sizing, the Monte Carlo approach showed that right-sizing the column reduced the quantity of required resin by 19 L. With resin costing more than $10,000 per liter, the savings are substantial.
To apply this approach to manufacturing scaleup, it is important that the process development and manufacturing teams agree on the ranges to simulate for each process parameter as well as the fraction of scenarios to exclude for scale-up planning. “For example,” Furcht says, “while excluding the upper 5% of mass or volume projections may reduce the size of chromatography columns or filters required, there should be agreement on whether this 5% cutoff introduces excess risk of under-sizing the downstream process.
“The main caveat to choosing Monte Carlo simulation for facility fit is that there is an increased probability (relative to the worst-case approach) that a specific unit operation may be over- or under-sized,” Furcht says. That’s because the Monte Carlo simulation approach may accommodate about 90% of the projected mass or volume scenarios, while the worst-case approach accounts for 100%. “However, the percentage of scenarios the Monte Carlo approach should account for can be increased or decreased depending on the risk tolerance for the specific process to be scaled,” he adds.