Introduction
This guidance provides information on demonstrating batch homogeneity of final APIs (small and large molecules) and critical intermediates.
This procedure provides guidance for performing a homogeneity evaluation in support of API process validation. The following components of the evaluation are described:
– Materials to be tested
– Selection of test methods for examining homogeneity
– Sampling plan – when to collect samples, from what locations, and the number of samples
– Selecting acceptance criteria for evaluating homogeneity test results.
Recommendations & Rationale for Recommendations
General Comments
Homogeneity is the acceptable distribution of chemical and physical properties within a batch, based on predefined criteria. The intent of examining homogeneity during the validation is to demonstrate that the quality of a sample collected from any location within a batch is representative of the quality of the entire batch.
For large molecules the evaluation of homogeneity must consider the consistency of the profile of heterogeneity of product-related molecular variants. This profile should be consistent throughout a batch and similar between batches.
Unless previously performed in another study, examination of API homogeneity must be performed during process validation. If homogeneity was shown in a previous study, the following should be considered to determine if this study is still applicable:
– Was the API prepared by the same process?
– What processing changes have been made, and what potential impact (if any) do these changes have on API homogeneity?
– Was the API prepared in the same (or equivalent) equipment?
– Was the API prepared at approximately the same batch size (e.g. within +25% linear scale of the validated batch size)?
Requirements, acceptance criteria, and conclusions for the homogeneity study may be included in the process validation documents, or may be presented in separate documentation that is referenced in the validation documents.
Materials to be tested
– Homogeneity shall be demonstrated for finished APIs unless otherwise justified and documented.
– The need to show homogeneity of isolated critical intermediates should be considered on a case-by-case basis depending on how the intermediate is used in subsequent processing. In general, studying the homogeneity of an intermediate is of less importance than that of a final API, especially if the intermediate will be dissolved in the next step of the processing. If homogeneity of the intermediate is critical to the quality of the final API prepared from it, demonstration of intermediate homogeneity should be considered.
– Homogeneity testing is typically not needed when the API is a liquid, because of the inherent homogeneous nature of such materials.
Selection of Test Methods for Examining Homogeneity
Three measurements are typically considered for a given homogeneity study: one to demonstrate chemical homogeneity, one to demonstrate physical homogeneity (if appropriate), and one to demonstrate the effectiveness of the drying process (if appropriate). Appropriately chosen analytical tests in these categories usually eliminate the need to perform other analytical tests to show homogeneity.
Chemical Homogeneity: Process impurity testing is generally the preferred analytical methodology for examining chemical homogeneity of small molecules, but other analytical techniques may be used. For large molecules, the consistency of the profile of heterogeneity of product-related molecular variants is demonstrated by appropriate techniques.
Physical Homogeneity: Evaluation for physical homogeneity may include tests such as those for particle size distribution, crystal form uniformity, and/or bulk volume. Where physical quality attributes have not been established for the API, physical homogeneity need not be demonstrated. If this is the case, it should be explained in the validation protocol.
Effectiveness of Drying: This may be important because residual solvents (including water) are considered process impurities. This assessment is especially important for higher risk cases such as vacuum tray driers used for static drying operations. Attention should be focused on either the last solvent used in the process, and/or the solvent that is most difficult to remove from the API. Examination of drying effectiveness is usually performed at the stage where all processing operations that influence dryness of the API have been completed.
Development of the Homogeneity Sampling Plan General
The following questions should be considered in preparing a homogeneity sampling plan:
When during the processing should samples be collected?
From what locations should samples be collected? The sample location should take into account such factors as the batch size, the type of processing equipment involved, and how the API is packaged.
How many samples are needed?
When during Processing Should Homogeneity Samples be Collected?
It is important to collect the homogeneity samples at the appropriate point during processing.
Sampling should target potential in homogeneity. An API process generally prepares a material that is a single substance rather than a mixture of materials (such as that found in a drug product), and a well-mixed API batch is typically obtained near the end of processing. Operations that potentially introduce inhomogeneity in APIs are readily identified (examples include collection of caked material from a crystallization tank, non-agitated washing of filter cakes, and non-agitated drying of filter cakes), so sampling and analytical testing may be targeted at investigating suspected potential for inhomogeneity.
Example 1: An API being examined for homogeneity is dried using stationary (non-agitated) conditions prior to milling. Because selective sampling from the drying equipment is possible, examination of effectiveness of drying on the unmilled dried material is preferred. Chemical homogeneity may be assessed using the same samples. Examination of physical homogeneity must be deferred until after the subsequent milling operation is completed.
Consider the effect of physical manipulation of the API. For example, samples to show chemical homogeneity of the API should be collected after size reduction of the material, if degradation of the API is known to be possible because of size reduction, deagglomeration, or material handling conditions. If such degradation has not been observed in the past and is not suspected, sampling for chemical homogeneity testing may be performed before physical manipulation of the API, and it is unnecessary to repeat this study after the physical manipulations.
Sample for effectiveness of drying upon completing the process step where the final drying of the API occurs.
Example 2: An API is wet-milled, but other factors make it desirable to examine the chemical homogeneity of partially dried material before milling. Examination of the effectiveness of final drying (and of physical homogeneity) should be deferred until all operations impacting these characteristics are completed.
Evidence of physical inhomogeneity after movement or shipment of packaged material does not necessarily implicate the ability of the process to provide homogeneous product.
Physical segregation is possible due to settling of fine particulates. It is preferable to collect samples for testing of physical homogeneity as soon as possible after packaging to minimize the incidence of settling that may occur after processing of the material has been completed.
Number of Samples
A minimum of six samples is recommended for a homogeneity study (3). Graphed simulations of the distribution of F ratios for various sample sizes and method variances indicate that results for at least six API samples are sufficient for statistical confidence that variance of the sample set is not significantly different than six replicate determinations of the test method.
When packaging of the API batch provides a large number (much greater than ten) of containers, it is not necessary to sample and test every container.
When one, or only a few, packaged API containers are produced, multiple samples may be collected from each container for homogeneity testing.
Sample Location
When determining sample location, consider the manufacturing equipment used for processing the material.
For equipment (such as agitated filter/dryers, mills, and blenders) that provide some mixing of material (whether intentional or not), sampling during discharge of material from equipment or during packaging of material is acceptable. In these cases, collect samples from the first and last containers and from locations dispersed throughout the batch.
For equipment that processes material in a static (non-agitated) manner, there is a potential of introducing non-uniform quality characteristics. In these cases, collect samples from locations that are suspected to be both easy-to-dry and difficult-to-dry within the batch. It is recommended to also sample from areas both deep within the mass of material and near its edges, where washing and/or drying operations may have different effectiveness. To aid in identifying easy-to-dry and difficult-to-dry locations in static processing equipment, it may be helpful to complete mapping studies (of temperature, pressure, or air flow) on freeze driers, shelf driers, and ovens.
Acceptance Criteria for Evaluating Homogeneity Test Results
Chemical homogeneity results are often evaluated using a statistical criterion. An F-Test is appropriate for this evaluation and is used to compare the variability of the results of homogeneity samples from the batch to the variability of the test method (4). The purpose of the evaluation is to see if the variability of results seen with the samples from the batch is significantly greater than the variability that arises because of the test method. An assessment of method precision that examines the full analytical system is not needed for this F-test comparison, because the sample set from a batch is typically run within a short period of time, and often by just one analyst. A set of six replicate determinations (including separate sample preparations) or recent system suitability results are generally regarded as a sufficient assessment of the method variability for this purpose.
Some considerations for use of the F-test:
When performing the F-Test, a generally accepted criterion is based on 95% confidence that the variability between results for the set of homogeneity samples is not different than the variability exhibited by the test method. Critical F values for a one-tailed test are used since one is only testing the data to determine if the variability of samples results is greater than method variability.
When the magnitude of the numbers being examined approaches that of the uncertainty of the measurement, the F-test is less reliable and may not be appropriate. For instance, it is recommended that HPLC impurity amounts below about 0.2% not be evaluated using the F-test. In these cases, rather than requiring a statistical analysis on a different measure (such as API purity), examine the range of impurity results for the individual samples from a given batch. When all results for the batch are qualitatively similar, the results support homogeneity of the material.
There is no indication of inhomogeneity in the case of method variability being greater than variability of sample results, thus obviating the need for a two-tailed test.
Appendix I provides an example of the application of the F-test.
Effectiveness of drying: A statistical criterion such as the F-test recommended for evaluating chemical homogeneity may be used for evaluating effectiveness of drying (especially for residual solvents), such as when experience with the process indicates consistent control of drying. Some processes, most notably those involving hydrates or solvates, may exhibit more variability in moisture levels, and thus may not be not well suited for statistical analysis of drying. In any event, results should also be reviewed for practical significance.
For an API quality attribute that is considered acceptable when it conforms to a limit or range (as is true for many physical properties), homogeneity may be established by demonstration that each sample of the API lot conforms to the limit(s) for that property. Analytical test results that are normally reported as “Meets Test” or “Acceptable” for chemical properties are not suitable for evaluation by statistical methods.
Where used, a statistical evaluation should be regarded as the first level of examination. It is possible that the statistical criterion may not be met, for reasons that do not mean that the material is inhomogeneous, such as when test method variability is greater than the variability of the sample results. In this event, evaluating the practical significance can determine if the sample results support the assumption that the batch is homogeneous, despite the failure of the statistical test to confirm this conclusion. Some considerations of practical significance include:
– The release decision is the same across the range of values observed;
– Homogeneity results from within a given API batch should be examined for possible trends during preparation of the batch. For instance, during a milling operation, the particle size distribution may have changed during the time period over which the milling was performed.
In the event that the F-test criterion is not met, an alternative statistical test should be considered for evaluating the homogeneity results. Often not meeting the F-test criterion will be due to the presence of a potential outlier result that might appear to represent inhomogeneity. An alternate test that may be used in this instance is to determine the mean and standard deviation for the sample set with this potential outlier value excluded from the set, and then see if the potential outlier value is within an acceptable confidence interval (such as +3standarddeviations) of the mean for the other values. A finding that all values for the batch (including the potential outlier) fall within the confidence interval supports the conclusion of homogeneity, but having the potential outlier value fall beyond this confidence interval may make the conclusion of homogeneity questionable without an adequate reason to explain the unusual result.