Choosing the polishing step for a mAb platform that misses aggregate and HCP specifications
An IgG1 monoclonal antibody is produced in CHO cells at 4.0 g/L in a 2,000 L bioreactor. The current platform is: Protein A affinity capture (yield 92%; eluate 12 g/L IgG, 5,000 ppm HCP, 8% aggregate by SEC, 800 ppb DNA), a low-pH hold at pH 3.6 for 60 min for retrovirus inactivation (no yield loss, but aggregate rises from 8% to 11%), flow-through anion exchange (yield 90%; removes 1.0 log HCP and 2 log DNA; no aggregate removal), and a final UF/DF (yield 95%). Release specifications are aggregate ≤ 2.0%, HCP ≤ 100 ppm, DNA ≤ 10 ppb, and overall yield ≥ 60%. The team can afford one additional chromatography step and is choosing between: cation exchange bind-elute (yield 85%; removes 1.5 log HCP; aggregate 11% to 1.8%) and a second flow-through anion exchange (yield 95%; removes 1.0 log HCP; no aggregate removal).
1. Calculate the overall yield and the final HCP, aggregate and DNA levels for the current process and for each option. State which specifications pass or fail.
Read the worked answer
Current process. Yield = 0.92 × 0.90 × 0.95 = 0.787, i.e. 78.7%, which passes the ≥60% criterion. HCP = 5,000 / 10 (1.0 log) = 500 ppm, which fails the 100 ppm limit. Aggregate = 11%, which fails the 2.0% limit. DNA = 800 ppb / 100 (2 log) = 8 ppb, which passes.
Option A, cation exchange added before UF/DF. Yield = 0.92 × 0.90 × 0.85 × 0.95 = 0.669, i.e. 66.9%, passes. HCP = 500 / 31.6 = 16 ppm, passes. Aggregate = 1.8%, passes. DNA remains ≤ 10 ppb.
Option B, second flow-through AEX. Yield = 0.92 × 0.90 × 0.95 × 0.95 = 0.747, i.e. 74.7%, passes. HCP = 500 / 10 = 50 ppm, passes. Aggregate stays at 11%, fails. DNA unaffected.
Conclusion. Both options clear HCP; only cation exchange clears the aggregate specification. Option B is cheaper in yield but does not deliver a releasable product, so its higher yield is irrelevant. Option A is the only one that satisfies all four criteria simultaneously.
2. Justify the choice on process grounds beyond the arithmetic, including what the choice implies about where the real problem lies and what operational risks the chosen step introduces.
Read the worked answer
Why cation exchange wins. Aggregate clearance is the binding constraint, and flow-through AEX is structurally incapable of removing an IgG aggregate because both species flow through. Only a binding step that discriminates on surface hydrophobicity or charge can separate aggregate from monomer, so the decision is driven by the specification, not by yield.
Where the problem really lies. The upstream pool already enters Protein A at 8% aggregate, and the low-pH hold adds three more points. A polishing step is treating a symptom. The 11% aggregate load also erodes the CEX yield, because aggregate and monomer compete for binding capacity and aggregate can foul the resin.
Risks introduced by CEX. Bind-elute requires the AEX flow-through pool to be diluted or buffer-exchanged to reach a loading conductivity the resin tolerates, adding a dilution, tank volume and possibly a UF step; the elution pool needs salt adjustment before UF/DF. Resin cleaning and lifetime for a bind-elute step are more demanding than for flow-through. Yield of 85% on a step that loads 12 g/L IgG means a large column and significant buffer volumes, so cost of goods rises.
Recommendation. Adopt cation exchange, but run a parallel experiment on low-pH hold time and stabilisers (for example arginine or sucrose) to attack the 11% at source; if that reduces aggregate to under 5%, a cheaper flow-through option may later become viable.
3. If the low-pH hold is confirmed as the main source of aggregation, propose an alternative viral-safety strategy and state the data you would need before switching.
Read the worked answer
Options. (1) Shorten the hold: determine the minimum time at pH 3.6 that still gives ≥ 1.5 log retrovirus inactivation, rather than fixing 60 min. (2) Add a stabiliser such as arginine, glycine or sucrose during the hold to suppress acid-induced unfolding, then remove it at UF/DF. (3) Replace the low-pH step with a solvent/detergent inactivation (for example 0.5–1% Triton X-100 or polysorbate 80) which works on enveloped viruses and is essentially non-denaturing for IgG, provided the detergent is cleared to acceptable levels downstream. (4) Move the low-pH hold to a point where the protein is more dilute, since aggregation is concentration-dependent.
Data needed before switching. Virus inactivation kinetics must be re-established with a model retrovirus (e.g. MuLV) at the new conditions, with at least two orthogonal logs, because regulatory expectations for orthogonal viral clearance are strict. Aggregate must be measured as a function of hold time and pH to show a real reduction, not just a shift. Product quality attributes (charge variants, glycan profile, potency) must be unchanged. For the detergent route, you need a validated clearance study showing the detergent is below its permitted residual limit, plus confirmation that the detergent does not interfere with the subsequent chromatography steps.
Decision rule. Switch only if viral clearance remains ≥ the current validated log reduction value while aggregate at the Protein A pool falls enough to make the cheaper flow-through polishing option viable.