Effective HPLC method development(Part III)

Effective HPLC method development(Part III)

7. Standardization

7.1 Area % method

If the response of the active pharmaceutical ingredient is linear from LOQ to the nominal sample concentration, use the % area approach where the related substance is reported as % area.  This is the most straightforward approach, and doesn’t require the preparation of standard solutions.  It also has the highest precision since preparation to preparation variation will not affect the results. However, in order to ensure the concentration is linear within this range, the sample concentration is usually limited and this will reduce the method sensitivity (i.e., increase LOQ).

In general, use this approach as long as the desired LOQ can be achieved.

7.2 External Standard method

Use the external standard method if the response of the active pharmaceutical ingredient is not linear throughout the whole range, or the desired LOQ can not be achieved by the area % method.  The concentration of standard solution should be high enough to ensure the standard solution can be prepared accurately and precisely on a routine basis, it should be low enough to approximate the concentration of related substance in the sample solution.  In general, the standard concentration should correspond to about 5 % of related substances.

7.3 Wavelength Selection and Relative Response Factor

Generate the linearity plot of API and related substances at different wavelengths.  At this point, Photodiode Array Detector can be used to investigate the linearity of the active pharmaceutical ingredient and related substances in the proposed concentration range.  By comparing the linearity slopes of the active pharmaceutical ingredient and the related substances, one can estimate the relative response factors of the related substances at different wavelengths. Disregard of whether Area % or External Standard approach is used, if the relative response factors of some significant related substances are far from unity, a response factor correction must be applied.

The optimum wavelength of detection is the wavelength that gives the highest sensitivity (lmax) for the significant related substances and minimizes the difference in response factors between those of the active pharmaceutical ingredient and the related substances. After the optimum wavelength is determined, use a highly stressed sample (e.g., 5% degradation) to verify that the selected wavelength will give the highest % related substance results.

7.4 Overall accuracy

A final check of the method performance is to determine the overall accuracy of the method. Unlike the accuracy from sample preparation (section 6.1.1), which simply compares the response of the analyte with and without spiking with matrix, the overall accuracy compares the % related substances calculated from an accuracy solution with that of the theoretical value. The accuracy solutions are the solutions spiked with known concentrations of related substances and matrix.  Since the extraction efficiency, choice of wavelength and the bias in standardization influence the calculated related substance result, this is the best way to investigate the accuracy of the method.  Overall accuracy reflects the true accuracy of the method.

8. Method Optimization/ Robustness

After the individual components of the method are optimized, perform the final optimization of the method to improve the accuracy, precision and LOQ.  Use an experimental design approach to determine the experimental factors that have significant impact on the method.  This is very important in determining what factors need to be investigated in the robustness testing during the method validation (see section 9).  To streamline the method optimization process, use Plackett Burmann Design (or similar approach) to simultaneously determine the main effects of many experimental factors.

Some of the typical experimental factors that need to be investigated are:

HPLC conditions: % organic, pH, flow rate, temperature, wavelength, column age. Sample preparation:  % organic, pH, shaking/sonication, sample size, sample age. Calculation/standardization: integration, wavelength, standard concentration, response

factor correction.

Typical responses that need to be investigated are:

Results: precision (%RSD), % related substance of significant related substances, total related substances.

Chromatography: resolution, tailing factor, separation of all related substances (section 3.1.1 and 3.1.2).

9. Method validation

9.1 Robustness

Method validation should be treated as a “final verification” of the method performance and should not be used as part of the method development.  Some of the typical method validation parameters should be studied thoroughly in the previous steps.  In some cases, robustness can be completed in the final method optimization before method validation.  At this point, the robustness experiments should be limited only to the most significant factors (usually less than 4 factors).  In addition, unlike the final method optimization (see section 8), the experimental factors should be varied within a narrow range to reflect normal day to day variation.  During the method validation, the purpose is to demonstrate that the method performance will not be significantly impacted by slight variations of the method conditions.

9.2 Linearity, Accuracy, Response Factor

Linearity, accuracy and response factors should be established for the significant related substances (section 3.1.1) during the method validation. In order to limit the workload of method development, usually less than 3 significant related substances should be selected in a method.

Therefore, the other related substances (section 3.1.2) should not be included in these

experiments.

9.3 System suitability criteria

It is advisable to run system suitability tests in these robustness experiments.  During the robustness testing of the method validation, critical method parameters such as mobile phase composition and column temperature are varied to mimic the day-to-day variability. Therefore, the system suitability results from these robustness experiments should reflect the expected range.  Consequently, the limits for system suitability tests can be estimated from these experiments.