ICH Analytical Method Validation
How do I validate an analytical method under ICH guidelines?
You validate an analytical method under ICH by showing it is fit for its intended purpose, following ICH Q2(R2) (Validation of Analytical Procedures, adopted 1 November 2023) together with its companion ICH Q14 (Analytical Procedure Development). In practice that means: define the purpose and performance criteria, write a protocol, test the performance characteristics that apply to your type of test, and report the results against the criteria in the CTD [1][2].
Approach: I read the full ICH texts of Q2(R2) and Q14, and checked how EMA and the FDA have implemented Q2(R2) on their official pages.
Scope
Q2(R2) covers procedures used for release and stability testing of commercial drug substances and products. It can also be applied, on a risk basis, to other procedures in the control strategy, and in a phase-appropriate way during clinical development. It covers assay, potency, purity, impurity (quantitative or limit test), identity and other measurements [1].
Step 1: Define the purpose and criteria (Q14)
- Minimal (traditional) approach: at a minimum you identify the attributes to test, select the technology, study specificity, accuracy, precision over the reportable range, and robustness, and document the procedure and its control strategy [2].
- Enhanced approach: you record the expected performance criteria in an Analytical Target Profile (ATP), and add risk assessment, univariate or multivariate experiments (DoE), and set-points, PARs or MODRs. Formal submission of an ATP is optional [2].
- Robustness: this is normally evaluated during development (deliberate parameter variations, plus stability of sample preparations and reagents). If it was done during development, it does not need to be repeated in validation [2].
- Control strategy: define it, including the system suitability test (SST), before validation, and confirm it after validation [2].
Step 2: Write the validation protocol
Before the study, a protocol "should be generated." It sets out the intended purpose, the performance characteristics to validate and their acceptance criteria. If you rely on prior knowledge, you need to justify it. The experimental design should use the same number of replicates as routine analysis uses to produce a reportable result. Results go into a validation report. You can use development data as validation data, and you can abbreviate testing for an established platform procedure if you can justify it scientifically [1].
Step 3: Choose the tests for your type of procedure (Q2(R2) Table 1)
| Characteristic | Identity | Impurity, quantitative | Impurity, limit test | Assay (content/potency) |
|---|---|---|---|---|
| Specificity | + | + | + | + |
| Response (calibration model) | - | + | - | + |
| Lower range limit | - | QL (DL in some complex cases) | DL | - |
| Accuracy | - | + | - | + |
| Repeatability / intermediate precision | - | + | - | + |
Key: "+" means the test is normally conducted; "-" means it is not normally conducted. You can use a combined accuracy/precision approach instead. If reproducibility data exist, you can derive intermediate precision from them [1].
Step 4: Run the studies (Q2(R2) Section 3)
-
Specificity/selectivity: show there is no interference, or compare with an orthogonal procedure, or justify by the technology itself (e.g. NMR chemical shifts). For chromatography, you can show it by the resolution of the two closest-eluting components. A stability-indicating test needs samples containing relevant degradants, such as spiked or stressed samples [1].
-
Range and response: at least five concentrations are recommended for linearity. Report the plot, correlation coefficient (or R²), y-intercept, slope and a residuals assessment. Non-linear responses (e.g. 4PL/5PL for bioassays) use a suitable model [1].
-
Reportable ranges in Table 2:
- Assay: 80-120% of declared content or of the specification limits.
- Content uniformity: 70-130%.
- Impurities: reporting threshold to 120% of the specification criterion.
- Potency: specification limits ±20%.
Other ranges are acceptable if justified [1].
-
DL/QL: four ways to set them:
- visual evaluation;
- signal-to-noise (3:1 for DL, at least 10:1 for QL);
- DL = 3.3σ/S and QL = 10σ/S;
- direct accuracy and precision at the QL.
For impurity tests, the QL should be at or below the reporting threshold [1].
-
Accuracy: test across the reportable range (e.g. 3 concentrations × 3 replicates) by comparison with a reference material, a spiking study or an orthogonal procedure. Report mean % recovery or bias with a 100(1-α)% confidence interval [1].
-
Precision:
- Repeatability: at least 9 determinations across the range, or at least 6 at 100% of the test concentration.
- Intermediate precision: vary days, analysts, equipment and environment as relevant, preferably by DoE.
- Reproducibility: inter-laboratory, usually not required for submission.
- Report SD, RSD and a confidence interval [1].
-
Multivariate methods: validate on an independent set using RMSEP compared with RMSEC [1].
-
Technique examples (Annex 2): for HPLC, GC and CE, if a relative response factor falls outside 0.8-1.2, "a correction factor should be applied" [1].
Step 5: Manage the method over its lifecycle
- Changes: a change may need partial or full revalidation, scaled to the performance characteristics it affects. A transfer to another site needs revalidation and/or comparative testing, or a justification for skipping it. Co-validation across sites can also satisfy transfer [1].
- Bridging (Q14 Table 2): if you change the measurement principle, you need full validation of the new procedure plus comparative analysis.
- Approved ranges: moving within an approved PAR or MODR "does not require regulatory notification." Changes to parameters that are not established conditions do not require regulatory communication [2].
Regional status
- EU: EMA lists Q2(R2) (EMA/CHMP/ICH/82072/2006) as adopted, effective 14 June 2024 (EMA).
- US: the FDA issued Q2(R2) as final guidance in March 2024 (docket FDA-2022-D-1503), alongside Q14 (FDA Q2(R2)).
Sources:
How does analytical procedure development fit in?
Development comes before validation and sets it up. Under ICH Q14 (adopted 1 November 2023), development turns your measurement needs into a procedure and control strategy that you can then validate under Q2(R2). The knowledge you gain also decides how much flexibility you get for post-approval changes. Q14 calls itself complementary to Q2: Q14 covers development and change management, and Q2 covers validation. It also notes that the order and extent of each lifecycle step can vary, and steps can overlap [2].
Approach: I read the full ICH Q14 guideline text, including the 4 April 2025 error corrections that align its glossary with Q2(R2).
How development feeds into validation
| Development output (Q14) | How it is used in validation (Q2(R2)) |
|---|---|
| Analytical Target Profile (ATP): the intended purpose, the attributes to measure, and the performance characteristics with their criteria | Once you have picked a technology, the ATP "serves as a foundation" for the performance criteria in validation. Submitting it is optional. |
| Robustness study: deliberate changes to parameters, plus how long sample preparations and reagents stay stable | If you did it during development, "it does not need to be repeated during validation". |
| Development data (for example, DoE robustness data) | Can count as validation data for the related performance characteristics. |
| PARs / MODRs | Validation is "required only for those performance characteristics not covered by data from analytical procedure development". The part you will run routinely (typically the set point) "must be covered by validation data". |
| Control strategy, including the system suitability test (SST) | "Should be defined before validation" and "confirmed after validation has been finalised". |
| Platform procedures | Development can be shortened, and some validation tests left out, if you justify it on science and risk grounds. |
Minimal vs enhanced approach
-
Minimal (traditional). Q14 says development "should include" at least these steps: identify the attributes to test, select the technology and instruments, study specificity, accuracy, precision over the reportable range and robustness, and document the procedure and its control strategy. This remains an acceptable approach.
-
Enhanced. You document the anticipated performance criteria in an ATP. You can also add:
- a risk assessment and a review of prior knowledge to find the parameters that matter (for example, with an Ishikawa diagram or the ICH Q9 tools);
- univariate or multivariate experiments;
- set points, PARs and/or MODRs.
In return you can get wider operating ranges and a better-targeted set of established conditions (ECs) with matching reporting categories.
Effect on lifecycle and post-approval changes
- Established conditions. If you use the minimal approach, you may end up with many ECs, all fixed parameters and set points. With the enhanced approach, ECs can be cut down and focused on performance (for example, ranges and performance criteria), as long as your understanding of the procedure and your risk management justify it.
- Moving within ranges. Set points, PARs and MODRs are subject to regulatory approval. Once approved, moving within them "does not require regulatory notification" and is handled in your pharmaceutical quality system. Changes outside the approved ranges, or widening them, require regulatory communication.
- Changes to parameters that are not ECs. These "do not require regulatory communication".
- Changes between technologies. The ATP can form the basis of a post-approval change management protocol (PACMP). That lets you report a change (for example, a switch between technologies) at a lower reporting category, provided the requirements you set in advance are met.
- Bridging studies. How much evidence a change needs depends on how much you know and how risky the change is (Table 1). For example:
- a change of measurement principle needs full validation of the new procedure, plus comparative analysis and/or a demonstration that the procedure still separates acceptable from non-acceptable results equally well (Table 2);
- a site transfer follows a similar bridging strategy.
- Ongoing monitoring. Monitoring selected outputs of the procedure is recommended so you can spot trends.
Where it goes in the CTD (Q14 Section 10)
- Procedure description: 3.2.S.4.2 for drug substance and 3.2.P.5.2 for drug product. Under the enhanced approach, this is also where the ATP criteria, the PARs or MODRs, and any proposed ECs go. Using the enhanced approach "should not lead to providing a less detailed description" of the procedure.
- Validation data and supporting information: 3.2.S.4.3 and 3.2.P.5.3. These sections also hold the summaries of your risk assessment and development studies that support your lifecycle strategy, and the justification for which parameters are and are not ECs.
- Development data as justification: may be appropriate to submit in some cases, depending on the intended purpose (for example, dissolution testing) or the technique.
- Multivariate models: if a model is used in manufacturing development or for in-process controls, its development information goes in the process development section (for example, 3.2.S.2.6 or 3.2.P.2).