This is a ready-to-use validation protocol for a ligand-binding ELISA. It handles the parts specific to ELISAs that a general potency-bioassay protocol does not: choosing the correct validation framework, the hook effect, minimum required dilution, matrix and selectivity, and dilutional linearity. Replace every <<FILL: ...>> placeholder and route it through document control. A filled specimen follows. This is general guidance to adapt and verify, not legal or regulatory advice. For the relative-potency curve and parallelism logic shared with cell-based assays, see also Protocol: Relative Potency Bioassay Validation.
Approval page
| Role | Name | Signature | Date |
|---|---|---|---|
| Author (assay SME) | <<FILL>> | ||
| Biostatistician | <<FILL>> | ||
| QC management | <<FILL>> | ||
| Quality Assurance | <<FILL>> |
1. Objective
Validate the <<FILL: ELISA name>> for its intended use, <<FILL: intended use>>, against the analytical target profile in section 3, demonstrating it is fit for that purpose.
2. Choose the framework first (this drives everything)
The acceptance framework depends on what the ELISA reports. Select one and do not mix them.
| ELISA type | Reportable value | Governing framework | Key acceptance basis |
|---|---|---|---|
| Relative-potency release ELISA | Relative potency vs reference | USP <1032>-<1034>, ICH Q2(R2), Q6B | Relative accuracy, precision, parallelism by equivalence |
| PK / concentration ELISA | Absolute concentration in matrix | FDA Bioanalytical Method Validation (2018) | Accuracy/precision limits, selectivity, matrix effect |
| Immunogenicity (anti-drug antibody) | Titer / cut point | FDA immunogenicity guidance; USP <1106>/<1106.1> | Screening/confirmatory cut points, sensitivity, drug tolerance |
This protocol is written for a <<FILL: relative-potency release / PK / immunogenicity>> ELISA. Applying PK accuracy/precision limits to a relative-potency release assay (or vice versa) is a conceptual error that surfaces in audits.
3. Analytical target profile
| Attribute | Target |
|---|---|
| Intended use | <<FILL>> |
| Reportable value format | <<FILL: e.g. relative potency, geometric mean of N assays / concentration in ng/mL>> |
| Required range | <<FILL>> |
| Required accuracy | <<FILL>> |
| Required precision | <<FILL>> |
| Matrix | <<FILL: buffer / serum / plasma / cell-culture>> |
4. System description and prerequisites
- Plate format, reader, and software:
<<FILL>>. - Reference standard and critical reagents (capture antibody, detection conjugate, coating reagent), with lot and assigned value:
<<FILL>>. - Locked curve model and weighting:
<<FILL: 4PL or 5PL; 1/Y^2 or other>>. - Prerequisites: qualified instruments, trained analysts, approved method draft, reference standard qualified.
<<FILL>>.
5. Roles
| Role | Responsibility |
|---|---|
| Assay SME | Executes development-to-validation, owns model and reagent strategy |
| Biostatistician | Sets equivalence bounds (relative-potency) or cut points (immunogenicity), variance/weighting model, analyzes precision/accuracy |
| Analysts | Execute the validation runs per the draft method |
| QA | Reviews and approves protocol and report, confirms no undocumented data exclusion |
6. Validation characteristics and test cases
Fill actual and pass/fail during execution.
| Test case ID | Characteristic | Design | Acceptance criterion | Actual | Pass/Fail | Tester/date |
|---|---|---|---|---|---|---|
<<FILL: TC-01>> | Curve model fit | Fit reference curves; residual analysis; confirm asymptotes defined | <<FILL: random residuals; fit diagnostics within limits>> | |||
<<FILL: TC-02>> | Hook effect (prozone) boundary | Run a high-concentration series to find where signal paradoxically drops | <<FILL: hook boundary characterized; assay range below it>> | |||
<<FILL: TC-03>> | Minimum required dilution (MRD) | Test dilutions to find where matrix interference is acceptable | <<FILL: MRD established and justified>> | |||
<<FILL: TC-04>> | Matrix effect / selectivity | Compare response in matrix vs buffer; test blank matrix lots | <<FILL: matrix effect controlled at MRD; blank produces no measurable result>> | |||
<<FILL: TC-05>> | Relative accuracy (trueness) | Prepare samples at known relative potencies (50/70/100/140/200%) | <<FILL: recovered value within X% of nominal across range>> | |||
<<FILL: TC-06>> | Dilutional linearity | Serially dilute high samples; confirm recovered values track | <<FILL: recovery within X% across dilutions>> | |||
<<FILL: TC-07>> | Spike recovery | Spike known amounts into matrix; recover | <<FILL: recovery within X%>> | |||
<<FILL: TC-08>> | Parallelism (relative-potency ELISA) | Equivalence test of sample vs reference curve shape | <<FILL: CI within registered equivalence bounds>> | |||
<<FILL: TC-09>> | Intermediate precision | Vary analyst, day, plate reader, reagent lot | <<FILL: GCV of reportable value NMT X%>> | |||
<<FILL: TC-10>> | Specificity / stability-indicating | Test stressed/degraded material and interferents | <<FILL: stressed reads reduced; interferents do not bias>> | |||
<<FILL: TC-11>> | Range | Interval where accuracy and precision both hold | <<FILL: e.g. 50-200% RP>> | |||
<<FILL: TC-12>> | Plate/edge effect control | Residual-by-position analysis; edge strategy | <<FILL: no significant positional bias; edge strategy documented>> |
7. The reportable value drives the criteria
Decide the reportable-value format before setting acceptance criteria, because precision improves with the square root of N independent assays. State N and lock it: <<FILL: e.g. geometric mean of 3 independent assays>>. Dropping to fewer assays in routine use requires a change control and a precision re-justification.
8. Deviation handling
Any deviation during execution is recorded, assessed for impact on the validation conclusion, and dispositioned before the report is finalized. <<FILL: reference deviation SOP>>.
9. Acceptance criteria (summary)
The method is validated when every test case in section 6 passes at the reportable-value format in section 7, the framework in section 2 is applied consistently, and no undocumented data exclusion occurred. Specific numeric criteria are derived from the method’s own development data and the ATP, not copied from a template.
10. Summary and conclusion
<<FILL: state whether the method met the ATP and is released for use, with the validated range, reportable-value format, and any limitations>>.
11. References
21 CFR 610.10, 600.3(s) (potency), where the ELISA is a potency method. USP <1032>-<1034>, <111> (bioassay design/validation/analysis) for a relative-potency ELISA. ICH Q2(R2), Q14, Q6B. FDA guidance, Bioanalytical Method Validation (2018) for a PK/concentration ELISA. FDA immunogenicity guidance and USP <1106>/<1106.1> for an anti-drug-antibody ELISA.
Describe compendial chapters in your own words; do not paste their text. Confirm the current version of every reference before use.
12. Attachments
- Analytical target profile.
- Locked method draft.
- Statistical analysis plan (equivalence bounds or cut points, variance model).
- Raw data and fitted-parameter records.
Filled specimen (excerpt)
The following shows two test cases completed for an example relative-potency release ELISA in buffer. Illustrative only.
| Test case | Characteristic | Acceptance criterion | Actual | Pass/Fail |
|---|---|---|---|---|
| TC-02 | Hook effect boundary | Assay range must sit below the hook; hook characterized | Signal plateaued then dropped above 500 ng/mL; working range capped at 200 ng/mL, well below hook | Pass |
| TC-05 | Relative accuracy | Recovered RP within 90-110% of nominal at 50/70/100/140/200% | Recovered 96, 103, 99, 101, 104%; all within 90-110% | Pass |
Note on TC-02: the hook (prozone) matters because a very high sample would otherwise read falsely low, a high result masquerading as a low one. Characterizing the hook boundary and capping the working range below it is what makes a high sample get flagged for dilution rather than mis-reported. This is an ELISA-specific failure mode a cell-based potency protocol would not address.
Common inspection findings this protocol prevents
- Wrong framework: PK accuracy/precision limits applied to a relative-potency release ELISA, or vice versa.
- Hook effect never characterized, so a high sample reads falsely low.
- Minimum required dilution and matrix effect not established for a matrix-based assay.
- Equivalence bounds or cut points set to software defaults with no data basis.
- Reportable-value format validated at N assays, then release on fewer without change control.
- Post-hoc well exclusion or model re-selection per run.
How to adapt this protocol
- Select and state the single framework in section 2; delete the rows that do not apply.
- Fill the ATP and the numeric acceptance criteria from your development data.
- For a buffer-based release ELISA, the matrix/MRD test cases simplify; for a serum PK ELISA, they are central.
- Get the biostatistician to own equivalence bounds (relative-potency) or cut points (immunogenicity) before execution.
- Confirm the current version of every cited reference before use.