Validation of Analytical Procedures

3.1.1.1 Absence of Interference

Specificity/selectivity can be shown by demonstrating that the identification and/or quantitation of an analyte is not impacted by the presence of other substances (e.g., impurities, degradation products, related substances, matrices, or other components likely to be present).

3.1.1.2 Orthogonal Procedure Comparison

Specificity/selectivity can be verified by demonstrating that the measured result of an analyte is comparable to the measured result of a second, well characterised analytical procedure that
ideally applies a different measurement principle.

3.1.1.3 Technology Inherent Justification

In some cases where the specificity of the analytical technology can be ensured and predicted by technical parameters (e.g., resolution of isotopes in mass spectrometry, chemical shifts in NMR spectroscopy), additional experimental studies may not be required, if justified.

3.1.2.1 Identification

For identification tests, a critical aspect is to demonstrate the capability to identify the analyte of interest based on unique aspects of its molecular structure and/or other specific properties.
The capability of an analytical procedure to identify an analyte can be confirmed by obtaining positive results comparable to a reference material using samples containing the analyte, along with negative results from samples which do not contain the analyte. In addition, the identification test should be applied to materials structurally similar to or closely related to the analyte to confirm that a positive result is not obtained. The choice of such potentially interfering materials should be based on scientific judgement with a consideration of interferences that could occur.

3.1.2.2 Assay, Purity and Impurity Test(s)

The specificity/selectivity of an analytical procedure should be demonstrated to fulfil the accuracy requirements for the content or potency of an analyte in the sample.
‍
Representative data (e.g., chromatograms, electropherograms, spectra, biological response) should be used to demonstrate specificity and relevant components should be labelled, if appropriate.
‍
For separation techniques, suitable discrimination should be investigated at an appropriate level (e.g., for critical separations in chromatography, specificity can be demonstrated by the resolution of the two components which elute closest to each other). Alternatively, spectra of different components could be compared to assess the possibility of interference.
‍
For non-separation techniques (e.g., bioassay, ELISA, qPCR), specificity can be demonstrated through the use of reference materials or other suitably characterised materials to confirm the absence of interference in relation to the analyte. In cases where the analyte is a process related impurity, specificity (non-interference) must also be confirmed against the product.
‍
In case a single procedure is not considered specific or sufficiently selective, an additional procedure should be used to ensure adequate discrimination. For example, where a titration is used to assay a drug substance for release, the combination of the assay and a suitable test for impurities may be used.
‍
Impurities or related substances are available or can be intentionally created:
‍
For assay or potency, discrimination of the analyte in the presence of impurities and/or excipients should be demonstrated. Practically, this can be performed by spiking product with appropriate amounts of impurities and consequently demonstrating that the assay result is unaffected by the presence of these materials (e.g., by comparison with the assay result obtained on unmanipulated samples). Alternatively, samples containing appropriate amounts of impurities could be generated through deliberate stressing of product materials.
‍
For a purity or impurity test, discrimination can be established by stressing or spiking product to achieve appropriate levels of impurities or related substances and demonstrating the absence of interference.
‍
Impurities or related substances are not available:
‍
If impurities, related substances or degradation products cannot be prepared or isolated, specificity can be demonstrated by comparing the test results of samples containing typical impurities, related substances or degradation products with an orthogonal procedure. The approach taken should be justified.

3.2.2.1 Linear Response

A linear relationship between analyte concentration and response should be evaluated across the range of the analytical procedure to confirm the suitability of the procedure for the intended purpose. The response can be demonstrated directly on the product or suitable reference materials, separate weighings of analyte, or predefined mixtures of the components (e.g., by dilution of a solution of known content), using the proposed procedure.
‍
Linearity can be evaluated with a plot of signals as a function of analyte concentration or content, and should demonstrate the analytical procedure capability across a given range to obtain values that are proportional to the true (known or theoretical) sample values. Test results should be evaluated by an appropriate statistical method (e.g., by calculation of a regression line by the method of least squares).
‍
Data derived from the regression line may help to provide mathematical estimates of the linearity. A plot of the data, the correlation coefficient or coefficient of determination, y-intercept and slope of the regression line should be provided. An analysis of the deviation of the actual data points from the regression line is helpful for evaluating linearity (e.g., for a linear response, the impact of any non-random pattern in the residuals plot from the regression analysis should be assessed).
‍
To assess linearity during validation, a minimum of five concentrations appropriately distributed across the range is recommended.
‍
The measured data can be mathematically transformed if necessary (e.g., through the use of a log function).
‍
Other approaches to the assessment of linearity should be justified.

3.2.2.2 Non-linear Response

Some analytical procedures may show non-linear responses. In these cases, a model or function which can describe the relationship between the activity/concentration present and the response of the analytical procedure is necessary. The suitability of the model should be assessed by means of non-linear regression analysis (e.g., coefficient of determination).
‍
For example, immunoassays or cell-based assays may show an S-shaped response. S-shaped test curves occur when the range of concentrations is wide enough that responses are constrained by upper and lower asymptotes. Common models used in this case are four- or five-parameter logistic functions, though other acceptable models exist.

For these analytical procedures, the evaluation of linearity is separate from consideration of the shape of the concentration-response curve. Thus, linearity of the concentration-response relationship is not required. Instead, analytical procedure performance should be evaluated across a given range to obtain values that are proportional to the true (known or theoretical) sample values.

3.2.2.3 Multivariate Calibration

Algorithms used for construction of multivariate calibration models can be linear or non-linear, as long as the model is appropriate for establishing the relationship between the signal and the quality attribute of interest. The accuracy of a multivariate procedure is dependent on multiple factors, such as the distribution of calibration samples across the calibration range and the reference analytical procedure error.
‍
In multivariate analysis, the measured data are commonly pre-treated through derivatives or normalisation.
‍
Linearity assessment, apart from comparison of reference and predicted results, should include information on how the analytical procedure error (residuals) changes across the calibration range. Graphical plots can be used to assess the residuals of the model prediction across the working range.

3.2.3.1 Based on Visual Evaluation

Visual evaluation can be used for both non-instrumental and instrumental procedures. The limit is determined by the analysis of samples with known concentrations and by establishing the minimum level at which the analyte can be reliably resolved and detected or quantitated.

3.2.3.2 Based on Signal-to-Noise

This approach is relevant for analytical procedures which exhibit baseline noise. Determination of the signal-to-noise ratio is performed by comparing measured signals from samples with known low concentrations of analyte with those of blank samples. Alternatively, signals in an appropriate baseline region can be used instead of blank samples. The DL or QL are the minimum concentrations at which the analyte can be reliably detected or quantitated, respectively. A signal-to-noise ratio of 3:1 is generally considered acceptable for estimating the DL. For QL, a ratio of at least 10:1 is considered acceptable.
‍
The signal-to-noise ratio should be determined within a defined region and, if possible, situated equally around the place where the peak of interest would be found.

3.2.3.3 Based on the Standard Deviation of a Linear Response and a Slope

The detection limit (DL) can be expressed as:

while the quantitation limit (QL) can be expressed as:

The slope S can be estimated from the regression line of the analyte. The estimate of σ can be carried out in a variety of ways, for example:

Based on the Standard Deviation of the Blank
‍
Measurement of the magnitude of background response is performed by analysing an appropriate number of blank samples and calculating the standard deviation of the responses.
‍
Based on the Calibration Curve
‍
A specific calibration curve should be evaluated using samples containing an analyte in the range of the DL and QL. The residual standard deviation of a regression line (i.e., root mean square error/deviation) or the standard deviation of y-intercepts of the regression lines can be used as the standard deviation.

3.2.3.4 Based on Accuracy and Precision at Lower Range Limits

Instead of using estimated values as described in the previous approaches, the QL can be directly validated by accuracy and precision measurements.

3.2.3.5 Recommended Data

The DL and the approach used for its determination should be presented. If the DL is determined based on visual evaluation or based on signal-to-noise ratio, the presentation of the relevant data is considered an acceptable justification.
‍
In cases where an estimated value for the DL is obtained by calculation or extrapolation, this estimate can subsequently be validated by the independent analysis of a suitable number of samples known to be near or prepared at the DL.
‍
The QL and the approach used for its determination should also be presented. If the QL was estimated, the limit should be subsequently validated by the analysis of a suitable number of samples known to be near or at the QL. In cases where the QL is well below (e.g., approximately 10 times lower than) the reporting limit, this confirmatory validation can be omitted with justification.
‍
For impurity tests, the QL for the analytical procedure should be equal to or below the reporting threshold.

3.3.1.1 Reference Material Comparison

The analytical procedure is applied to an analyte of known purity (e.g., a reference material, a well characterised impurity or a related substance) and the measured versus theoretically expected results are evaluated.

3.3.1.2 Spiking Study

The analytical procedure is applied to a matrix of all components except the analyte where a known amount of the analyte of interest has been added. In cases where all the expected components are impossible to reproduce, the analyte can be added to or enriched in the test sample. The results from measurements on unspiked and spiked/enriched samples are evaluated.

3.3.1.3 Orthogonal Procedure Comparison

The results of the proposed analytical procedure are compared with those of an orthogonal procedure. The accuracy of the orthogonal procedure should be reported. Orthogonal procedures can be used with quantitative impurity measurements to verify primary measurement values in cases where obtaining samples of all relevant components needed to mimic the matrix for spiking studies is not possible.

3.3.1.4 Recommended Data

Accuracy should be assessed using an appropriate number of determinations and concentration levels covering the reportable range (e.g., 3 concentrations/3 replicates each of the full analytical procedure).
‍
Accuracy should be reported as the mean percent recovery of a known added amount of analyte in the sample or as the difference between the mean and the accepted true value, together with an appropriate 100(1-α) % confidence interval (or justified alternative statistical interval). The observed interval should be compatible with the corresponding accuracy acceptance criteria, unless otherwise justified.
‍
For impurity tests, the approach for the determination of individual or total impurities should be described (e.g., weight/weight or area percent with respect to the major analyte).

For quantitative applications of multivariate analytical procedures, appropriate metrics, e.g., root mean-squared error of prediction (RMSEP), should be used. If RMSEP is found to be comparable to acceptable root mean-squared error of calibration (RMSEC) then this indicates that the model is sufficiently accurate when tested with an independent test set. Qualitative applications such as classification, misclassification rate or positive prediction rate can be used to characterise accuracy.

3.3.2.1 Repeatability

Repeatability should be assessed using:
‍
a) a minimum of 9 determinations covering the reportable range for the procedure (e.g., 3 concentrations/3 replicates each)
‍
or
‍
b) a minimum of 6 determinations at 100% of the test concentration.

3.3.2.2 Intermediate Precision

The extent to which intermediate precision should be established depends on the circumstances under which the procedure is intended to be used. The applicant should establish the effects of random events on the precision of the analytical procedure. Typical variations to be studied include different days, environmental conditions, analysts and equipment, as relevant. Ideally, the variations tested should be based on and justified by using analytical procedure understanding from development and risk assessment (ICH Q14). Studying these effects individually is not necessary. The use of design of experiments studies is encouraged.

3.3.2.3 Reproducibility

Reproducibility is assessed by means of an inter-laboratory trial. Investigation of reproducibility is usually not required for regulatory submission but should be considered in cases of standardisation of an analytical procedure, for instance, for inclusion of procedures in pharmacopoeias and in cases where analytical procedures are conducted at multiple sites.

3.3.2.4 Recommended Data

The standard deviation, relative standard deviation (coefficient of variation), and an appropriate 100(1-α) % confidence interval (or justified alternative statistical interval) should be reported. The observed interval should be compatible with the corresponding precision acceptance criteria, unless otherwise justified.

Additionally, for multivariate analytical procedures, the routine metrics of RMSEP encompass accuracy and precision.

3.3.3.1 Recommended Data

If a combined performance criterion is chosen, results should be reported as a combined value to provide appropriate overall knowledge of the suitability of the analytical procedure. If relevant to justify the suitability of the analytical procedure, the individual results for accuracy and precision should be provided as supplemental information. The approach used should be described.