Out-of-Specification Assay with a Passing Retest at Release
Batch MR-2291 of a modified-release tablet (480,000 tablets) is at release testing. The assay specification is 95.0-105.0% of label claim. The first preparation gave 92.4% (duplicate injections 92.1% and 92.7%). The analyst re-injected the same vial and obtained 99.6%; a freshly prepared sample then gave 100.2%. The laboratory supervisor proposes to report 100.2% as the batch result, noting that the first dilution 'looked wrong' and that the second preparation was made by a more experienced analyst. No formal Phase I investigation has been opened. System suitability passed on all injections, and the standard and sample chromatograms show no integration anomalies. The batch is one of three made from the same API lot; the other two passed. A distributor is waiting, the product is on allocation, and the QC manager asks you, as QP, to approve release so the batch can ship tomorrow.
1. Is it acceptable to release MR-2291 on the basis of the 100.2% result? Justify your answer.
Read the worked answer
No. The 100.2% figure is a retest obtained after an initial out-of-specification result, and reporting it as the batch result without a documented, assignable cause is testing into compliance.
Several features make the passing result especially hard to defend. The duplicate injections at 92.1% and 92.7% are precise and mutually consistent, which argues against an injection or integration error. The chromatograms show no anomalies, and system suitability passed, so the usual laboratory excuses are not supported by the raw data. The claim that the first dilution 'looked wrong' is an observation made after the result was known, and the supervisor's argument that the second analyst was more experienced is not evidence.
Under OOS principles the initial result stands as the result of record until it is invalidated by an investigation with a demonstrated cause. Averaging, selecting the highest result, or re-testing until a pass is obtained is prohibited. The correct action is to open an investigation immediately, keep the batch on hold, and tell the distributor that the ship date cannot be confirmed. Commercial allocation pressure is irrelevant to the scientific and legal question.
2. What must a Phase I laboratory investigation establish before any retest could be considered valid, and what would you do if it identifies a plausible dilution error?
Read the worked answer
The Phase I investigation must test the laboratory hypothesis against evidence, not against the desired outcome. It should cover:
- The analyst's worksheet, calculations, diluent volumes, glassware calibration and pipette checks.
- Standard preparation and potency, balance calibration, and the sample's storage and homogeneity.
- The full chromatographic raw data and audit trail, including any reintegration or sequence reprocessing.
- An interview with the analyst and a documented reconstruction of the preparation.
A retest can only be accepted if an assignable cause is demonstrated, for example a documented 10 mL pipette used where the method specifies 5 mL, with the error reproduced or otherwise proven. Even then, the retest must follow a predefined, approved plan: a specified number of independent preparations, performed by a qualified analyst, with the original data retained and the invalidated result clearly explained.
If the cause is only 'likely' or cannot be demonstrated, the OOS is confirmed. In that case the batch is rejected, and the investigation moves to Phase II manufacturing review: blend uniformity, compression, API lot, and whether the other two batches are genuinely unaffected or were simply tested less critically.
3. What are the implications for the other two batches from the same API lot, and how would you handle the distributor's allocation?
Read the worked answer
The other two batches cannot be assumed safe merely because they passed. If the confirmed OOS has a manufacturing cause, the same API lot, blend or compression campaign is implicated, and the passing results may reflect sampling rather than a genuine difference. I would review their batch records for blend uniformity, content uniformity, compression parameters and any deviations, and consider targeted testing of retained samples or additional units.
If the cause is laboratory-only and the API and manufacturing process are exonerated, the other two batches may remain releasable, but the reasoning must be documented.
For the distributor, allocation pressure is not a release criterion. I would not release MR-2291 to fill allocation. I would give the distributor a clear, non-speculative status update, prioritise release of the unaffected batches if their assessment supports it, and escalate to the supply team so that a shortage is managed transparently. If the investigation ultimately shows a manufacturing cause affecting multiple batches, regulatory notification and a field action may be required, and the QP should raise that possibility early rather than after stock has shipped.