Key points
- Calibration standards: within ±15% of nominal (±20% at the LLOQ); at least 75% of the standards and at least six non-zero levels must pass.
- QCs: at least three levels (low, medium, high) in duplicate; at least two-thirds of all QCs and at least 50% at each level within ±15%.
- The number of QCs must be at least 5% of the study samples in the run, or six QCs - whichever is higher.
Why run acceptance matters
Every analytical run in a regulated bioanalytical study produces two kinds of results: the concentrations of the study samples, and the evidence that the run itself performed as validated. ICH M10 (Bioanalytical Method Validation and Study Sample Analysis, adopted 2022) sets the acceptance criteria for that evidence. If a run fails, none of its study sample results can be reported.
In most laboratories the check is still done by hand: the run is exported from the chromatography data system, pasted into a spreadsheet, and each calibration standard and QC is compared with its nominal value. It is repetitive, it is checked a second time by QC, and it is exactly where transcription errors creep in.
Calibration standards
For chromatographic methods, the back-calculated concentration of each calibration standard should be within ±15% of the nominal concentration, except at the LLOQ, where ±20% is accepted. At least 75% of the calibration standards, with a minimum of six calibration standard levels, must meet these criteria.
- The blank and the zero sample (blank with internal standard) are analysed but are not part of the regression.
- Standards that fail are excluded and the curve is re-calculated without them - the rejection is documented, not hidden.
- The regression model and weighting (for example linear, 1/x²) are those established during method validation.
Quality control samples
QCs are prepared independently of the calibration standards and show whether the run quantifies unknown samples correctly.
| Rule | ICH M10 (chromatography) |
|---|---|
| QC levels | At least three: low (≤3× LLOQ), medium (around 30-50% of the range) and high (at least 75% of the ULOQ) |
| Replicates | At least duplicate at each level |
| Number of QCs | At least 5% of the study samples in the run, or six QCs - whichever is higher |
| Accuracy | Within ±15% of nominal |
| Overall | At least two-thirds (67%) of all QCs within ±15% |
| Per level | At least 50% of the QCs at each concentration level within ±15% |
When study sample concentrations cluster in a narrow part of the range, ICH M10 expects the QC levels to be adjusted so they reflect the concentrations actually measured.
Worked example
A run with eight non-zero calibration standards (LLOQ 1 ng/mL to ULOQ 1000 ng/mL) and six QCs:
| Sample | Nominal (ng/mL) | Found (ng/mL) | Bias | Limit | Result |
|---|---|---|---|---|---|
| CS1 (LLOQ) | 1.00 | 0.83 | -17.0% | ±20% | Pass |
| CS2 | 2.00 | 2.21 | +10.5% | ±15% | Pass |
| CS3 | 10.0 | 11.8 | +18.0% | ±15% | Fail |
| CS4-CS8 | 50-1000 | - | within ±9% | ±15% | Pass |
| LQC 1 / 2 | 3.00 | 2.81 / 3.12 | -6.3% / +4.0% | ±15% | Pass / Pass |
| MQC 1 / 2 | 400 | 421 / 389 | +5.3% / -2.8% | ±15% | Pass / Pass |
| HQC 1 / 2 | 800 | 938 / 772 | +17.3% / -3.5% | ±15% | Fail / Pass |
- Calibration: 7 of 8 standards pass (87.5% ≥ 75%) and seven levels remain (≥ 6) - accepted. CS3 is excluded and the curve re-fitted.
- QCs overall: 5 of 6 pass (83% ≥ 67%) - accepted.
- QCs per level: HQC has 1 of 2 within limits (50%) - accepted.
The run is accepted. Had both HQCs failed, the per-level rule (0% at HQC) would reject the run, even though four of six QCs (67%) passed overall - this is the check most often missed in manual review.
When the LLOQ or ULOQ standard fails
If the calibration standard at the LLOQ or the ULOQ is rejected, the calibration range of that run is narrowed: the next accepted standard becomes the lowest or highest limit of the run. Study samples outside the narrowed range cannot be quantified from this run - samples above it are re-analysed with dilution, samples below it are handled as defined in your SOP and the study plan. The QCs must still cover the narrowed range as required.
Common mistakes
- Forgetting the per-level QC rule - checking only the overall two-thirds.
- Counting levels, not standards (or the reverse) - both the 75% and the six-level conditions must hold.
- Too few QCs for large runs - 5% of 200 study samples is 10 QCs, not six.
- Silent exclusion - a rejected standard must be documented with the reason; an audit trail should show it.
- Copying values by hand from the CDS into a spreadsheet - the most common source of errors QA finds.
Educational summary for chromatographic (LC-MS/MS) bioanalysis. Always apply the current ICH M10 text, the expectations of the regulator concerned and your laboratory's SOPs. Ligand-binding assays have wider limits in ICH M10.
ARISENTO Insights is published by Autofai Technologies Private Limited, the makers of ARISENTO bioanalytical software.