Bioanalytical Method Transfer: Partial Validation for a New Matrix
A validated LC-MS/MS method for quantifying drug X in human plasma (K2EDTA) is being transferred to a new laboratory for a clinical trial that will also collect urine and dried blood spot (DBS) samples. The original validation included calibration curves (1–1000 ng/mL), QC samples at 3, 30, and 800 ng/mL, and stability in plasma. The receiving lab performs a partial validation: calibration curves in urine and DBS, precision and accuracy at 3, 30, and 800 ng/mL in each matrix, and stability in urine and DBS. Results show: urine calibration curves linear (r=0.998), QC accuracy within ±12%, precision ≤10%; DBS calibration curves linear (r=0.995), but QC at 3 ng/mL shows accuracy −18% and precision 15% (n=6). Stability in DBS at room temperature for 24 h shows −20% change at 3 ng/mL. The lab wants to proceed with all matrices using the same calibration range and QC acceptance criteria. The clinical team needs to report low-concentration DBS samples (expected 5–50 ng/mL).
1. Evaluate whether the partial validation results support using the method for DBS samples in the range 5–50 ng/mL. What are the key deficiencies and what actions would you recommend?
Read the worked answer
The DBS partial validation fails at the low QC (3 ng/mL): accuracy −18% (outside typical ±15% acceptance) and precision 15% (exceeds 15% CV). Stability at 3 ng/mL in DBS shows −20% change after 24 h at room temperature, indicating instability or extraction issues at low concentrations. Since the clinical range of interest is 5–50 ng/mL, the low-end performance is critical. The method cannot be considered validated for DBS at 5 ng/mL because the lowest QC (3 ng/mL) fails and stability is unacceptable. Recommended actions: (1) Investigate the cause of low-concentration failure-e.g., nonspecific binding, hematocrit effects, extraction recovery, or matrix interference. (2) Optimize the DBS extraction procedure (e.g., use of internal standard, longer extraction, different solvent). (3) Re-validate DBS with QCs at 5, 15, and 40 ng/mL (within the clinical range) and demonstrate accuracy and precision within ±15% (±20% at LLOQ) and stability at relevant temperatures and times. (4) If low-end sensitivity cannot be improved, consider a separate low-range calibration curve for DBS or alternative sampling. (5) Do not report DBS data until validation is complete. Urine validation appears acceptable, but ensure stability in urine is also within criteria. The same calibration range may not be suitable for DBS; a matrix-specific range may be needed.
2. The lab proposes to use the original plasma calibration curve for DBS by applying a conversion factor (DBS concentration = plasma concentration × 1.2). Discuss the scientific validity of this approach and what additional experiments would be required.
Read the worked answer
Using a fixed conversion factor from plasma to DBS assumes a constant relationship across concentrations, hematocrit values, and individuals. This is generally invalid because DBS concentrations depend on hematocrit, spot volume, and distribution into red blood cells. A conversion factor derived from a few samples may not hold across the clinical range (5–50 ng/mL) or patient population. To validate such an approach, one would need to: (1) Demonstrate a consistent correlation between DBS and plasma concentrations across the expected range and hematocrit range (e.g., 30–55%). (2) Perform a bridging study with paired plasma and DBS samples from the same patients (n≥20) and show that the conversion factor is reproducible and that the 95% confidence interval of the ratio is within acceptance criteria (e.g., ±20%). (3) Evaluate the effect of hematocrit on spot volume and recovery. (4) Validate DBS-specific stability and extraction. Without these, the conversion factor approach is not acceptable for regulated bioanalysis. The partial validation failures at low concentrations further undermine this approach. It is more scientifically sound to develop and validate a DBS-specific method with its own calibration curve and QCs.