Failed Analytical Method Transfer for a Fixed-Dose Combination Tablet
A validated reversed-phase HPLC assay for a fixed-dose tablet containing metformin 500 mg and sitagliptin 50 mg is being transferred from a development laboratory to a contract QC laboratory. The method specifies a C18 column (150 × 4.6 mm, 5 µm), phosphate buffer at pH 3.0, and 20 minutes of shaking for sample extraction. Development laboratory results are sitagliptin 99.4% and 99.8% of label claim, metformin 100.2%. The receiving laboratory, using its own qualified column and a buffer whose measured pH is 3.18, obtains sitagliptin 93.6% and 94.1%, while metformin is 99.9%. System suitability in the receiving laboratory passes (sitagliptin tailing 1.3, resolution from metformin 2.4, %RSD of five injections 0.6%), and standard responses differ by less than 2% between the two laboratories. No degradation peak above 0.05% is seen. The receiving laboratory proposes to report 94% and to apply a correction factor derived from the standard.
1. Explain why acceptable system suitability and matched standard response constrain the root cause, and set out the investigation sequence. Address explicitly whether the buffer pH difference explains the result.
Read the worked answer
What passing system suitability rules out: tailing 1.3, resolution 2.4 and %RSD 0.6% exclude gross chromatographic failure - bad column, wrong mobile phase composition, injection problems.
What the matched standard rules out: if the same standard gives equivalent peak response (within 2%) in both laboratories, then standard concentration, dilution, detector sensitivity and mobile phase strength are all behaving the same. The analyte reaches the detector correctly; the loss must occur before injection, i.e. in sample preparation.
The pH question: sitagliptin is a weak base with a pKa near 7.7. The fraction ionised at pH 3.00 versus 3.18 differs by well under 0.1%, so the buffer difference cannot account for a 5–6% loss. It should be documented as not causative rather than 'fixed'.
Sequence: Phase I - verify pH meter calibration, review the analyst's preparation record, confirm column brand and lot, re-inject the same vial. If no assignable cause, Phase II - repeat using the transferring laboratory's column and buffer on the receiving instrument, then analyse a common sample prepared independently in each laboratory.
2. Design an experiment to test the extraction hypothesis, with acceptance criteria, and state what to do if extraction proves complete.
Read the worked answer
Design: prepare one homogeneous sample, split into aliquots, and vary only the extraction step: 20 min shaking, 30 min shaking, 20 min sonication, and a second extraction of the filtered solids. Assay all against the same standard and express each as a percentage of the 30-min sonication value.
Criteria: if the second extraction of the solids recovers more than 2% of label claim, extraction is incomplete. Run a parallel filter-spiking test - pass a standard solution through three filter types and compare with unfiltered solution; more than about 2% adsorption means the filter must be changed or the first 2 mL discarded. Repeat in both laboratories so the comparison is like-for-like.
If extraction is complete in both: the remaining difference points to column chemistry (end-capping and residual silanols altering sitagliptin peak shape and integration) or to integration parameters. In that case the method should specify the column by brand and chemistry, and the transfer should be co-validated with that column. Metformin passing at 99.9% throughout is a useful internal control - it shows the diluent and volumetric steps are sound.
3. Evaluate the proposal to report 94% and apply a correction factor. What is the correct regulatory path?
Read the worked answer
The correction factor is not acceptable. Because the standard already gave equivalent response in both laboratories, a factor of about 1.07 would simply multiply every sample result and mask a real sample-preparation loss. It has no mechanistic basis, would require revalidation, and would convert a legitimate investigation into a data-integrity problem - the same class of failure as discarding an inconvenient result.
Correct path:
- Retain the out-of-trend result and complete the laboratory investigation with documented findings.
- Identify the assignable cause (extraction, filtration, column chemistry) and correct it.
- Perform a formal co-validation per USP <1224> with predefined acceptance criteria, typically a difference between laboratories of no more than 2.0% absolute for each analyte, across at least three lots spanning the specification range.
- Report the results obtained under the co-validation and only then qualify the receiving laboratory.
Until the co-validation passes, the contract laboratory is not qualified for this assay and its results cannot support batch release.