Out-of-Specification Titer and Purity After Sample Stagnation
At day 12 of a 2,000 L fed-batch CHO culture, the in-process titer by Protein A HPLC is 2.1 g/L against a historical range of 3.0–3.5 g/L. The sample was drawn from a sample valve after 45 minutes without flushing and held 24 hours at 2–8°C before analysis. The same sample shows Protein A purity of 91.2% (in-process expectation ≥97.0%) and SEC aggregates of 1.8% (specification ≤2.0%). The reference standard run in the same sequence is within 2% of its assigned value. A repeat sample drawn after a 5-minute flush and analyzed immediately gives 2.9 g/L and 96.8% purity. The bioreactor has 36 hours left in the run. The harvest pool would be 1,800 L at 2.9 g/L, which meets the minimum plant yield of 4.5 kg, but the release specification for purity after purification is ≥97.0%. The deviation system requires a root cause before batch disposition.
1. How should the team interpret the discrepancy between the first and second samples? List at least three plausible root causes and the data that would distinguish them.
Read the worked answer
The first result should be treated as a suspect out-of-specification result, not as proof that the bioreactor failed. The second sample is more credible because it was drawn after a proper flush and analyzed immediately, but the difference must still be explained.
Plausible root causes and discriminating data:
- Stagnant sample valve or dead-leg. The first draw may have contained fluid that sat in the valve and adsorbed protein or allowed aggregation. Data: compare titer and purity from a 5-, 10-, and 20-second flush series, and sample from a second valve or a sterile sampling device.
- Hold-time degradation or adsorption during 24-hour storage. Protein may have adsorbed to the container or aggregated at 2–8°C. Data: hold a fresh sample for 0, 4, 8, and 24 hours and measure titer, purity, and SEC in the same sequence.
- Analytical carryover or column fouling in the first run. The reference standard passed, but that does not rule out sample-specific carryover or injection order effects. Data: re-inject the original sample in a new sequence with a blank and a duplicate, and inspect chromatography for carryover peaks.
- True bioreactor heterogeneity or a real decline. Data: test a third independent sample from a different location and compare viable cell density, viability, and metabolite trends over the last 72 hours.
The investigation should not stop at the retest. A root cause is required before disposition.
2. The repeat sample meets titer and nearly meets purity. Can the batch be released on the repeat result? Explain.
Read the worked answer
No. A repeat result does not erase an out-of-specification result. The original result must be investigated and invalidated with a documented root cause before the repeat can be used for disposition. If the root cause is a stagnant sample valve or a hold-time artifact, the original result can be classified as a sampling or analytical failure, and the repeat result can support the in-process decision. If the root cause is a real decline in culture performance, the repeat result is simply a better estimate of a marginal batch.
The purity result is also not a release result. The 96.8% Protein A purity is an in-process value; the release specification of ≥97.0% applies after purification. A slightly low in-process purity may still be acceptable if downstream clearance is validated and the aggregate level is controlled. However, the team should not release the batch on a single repeat. It should require a third independent sample, a hold-time study, and a review of the purification step's clearance data. If the investigation cannot establish a root cause, the conservative disposition is to hold the harvest pool pending further testing rather than release it.
3. What harvest decision do you make at 36 hours, and what data would change it?
Read the worked answer
I would proceed to harvest at 36 hours only if a third independent sample confirms titer near 2.9 g/L and in-process purity at or above about 96%. The yield calculation matters: 1,800 L at 2.9 g/L is about 5.2 kg, which exceeds the 4.5 kg minimum, so the batch is not automatically a yield failure. The harvest should proceed with the pool held for disposition, not released.
Data that would change the decision:
- A third sample with titer below 2.5 g/L or purity below 95% would suggest a real decline, and I would consider extending the culture only if viability and metabolites support it.
- A confirmed hold-time or valve artifact would support harvest and release of the in-process result after deviation closure.
- A rising aggregate trend or a purification clearance failure would push toward discarding the batch rather than harvesting a marginal pool.
- Contamination or filter integrity failure would stop harvest immediately.
The decision should be documented as a risk-based disposition, not as a simple pass/fail on the retest.