1. Purpose
The purpose of this international guideline is to outline minimum mandatory requirements as well as recommendations for the identification and reaction to trends found in stability data.
2. Scope and Applicability
This guideline is applicable to all Commercial Stability Sites and contractors performing stability studies on commercial drug substances and drug products.
3. Definitions
3.1 Out of Specification (OOS)
Result A laboratory test result that is outside its regulatory or compendial limits.
3.2 Trend
A pattern of data that indicates change over time. This data may demonstrate either an increasing or decreasing trend (change of mean) for the stability indicating parameter over time or the data may indicate no discernible change at all. The change may be linear or non-linear.
3.3 Trending in Stability Studies
The evaluation of stability data (not necessarily statistical) in order to identify trends and their impact on the stability of a product. 3.3 Significant Trend An average typical trend for a parameter that, in relation to release result variability and specification limits, may lead to an OOS result before or at end of shelf life for any batch released. The definition of the criterion for significant trend might differ between products, but could for example be defined as a certain proportion of the specification interval.
3.4 Out of Trend (OOT)
A single result or a number of results that do not follow the expected trend, for a particular batch or series of batches, either in comparison with other stability studies or with respect to previous results collected during a stability study. There are three types of OOT situations identified in this guideline: Atypical result, Atypical trend and Adverse trend.
3.5 Atypical Result
A single result that does not follow the expected trend for a stability indicating parameter compared to previous results from the same study.
3.6 Atypical Trend
A set of data points for a stability indicating parameter, from one stability study or several studies, showing an atypical pattern. The detection of an atypical trend requires knowledge of the expected trend, i.e. the typical trend of stability data from other studies on the product.
3.7 Adverse Trend
A series of results showing a decrease/increase that indicates a probability of an OOS result before the registered retest period or shelf life of the batch is reached.
3.8 Release Alert Limit
An in-house limit set for the purpose of alerting a formulation site to a batch, whose release data is near enough to the release specification limit, that there is an increased probability that the batch will not meet one or more of it’s registered lifetime specifications before the retest/expiry date of the batch is reached.
3.9 Stability Study Alert Limit
A predefined limit for a parameter in a stability study that is designed to identify an atypical result, an atypical trend or an adverse trend.
4. Responsibilities
4.1 Commercial Stability Site
It is the responsibility of the Commercial Stability Site to: Perform trending on commercial stability studies at least once a year. Identify and justify stability indicating parameters that are to be trended, Inform QP/senior QA Management at the site manufacturing the product, if a significant trend has been observed Report adverse trends to appropriate management Recommend the establishment of release alert limits, if necessary review and evaluate release alert limits periodically and recommend revision of these limits where justified by available data.
4.2 Contractor
It is the responsibility of the contractor (if specified in the Quality Assurance Agreement) to: Perform trending on commercial stability studies at least once a year Inform QP/senior QA Management at the site managing the contractor, if a significant trend has been observed Report adverse trends according to QA Agreement.
4.3 Lead Site
It is the responsibility of the Lead Team/Site (acting as the Commercial Stability Site) assigned to a contractor to ensure that the contractor has procedures in place to comply with this guideline. This responsibility shall be clearly documented in the relevant Quality Assurance Agreement.
4.4 QP/Senior QA Management
It is the responsibility of QP/senior QA Management at the formulation site, or site managing the contractor to: Review the outcome of the trending reports issued by the Commercial Stability Site or contractor Ensure that the product release specification is constructed in a way that provides a high probability that any batch released will remain within the registered specification throughout the retest period/shelf-life of the drug substance/drug product when stored according to the labeled storage conditions React to reported adverse trends shown by a single batch or series of batches according to procedure.
5. Guideline
5.1 Introduction
The registered retest period/shelf life of a drug substance/product will have been set taking into account the specification to be registered and the trends seen in stability studies completed or ongoing at the time of new product and/or new primary pack registration.
However, these studies will have been conducted on relatively few batches, some made only at pilot scale. Therefore, the Commercial Stability Site must use trending of stability data to support the retest period/shelf-life of a drug substance/product as well as to indicate when a change to retest period/shelf-life and/or cautionary labeling statement is required. Trending must also be used as a tool to identify significant trends and recommend release alert limits when necessary.
The formulation site must take appropriate action to ensure that drug substance/ product release procedures are updated to take into account the latest stability trend analysis reports issued by the Commercial Stability Site. This includes the establishment or revision of release alert limits if recommended by the Commercial Stability Site. See section 5.5 for further guidance. During ongoing studies Commercial Stability Sites may identify OOT results.
In that case, the formulation site(s) must be involved to investigate and whenever possible reverse the cause of the deterioration in stability characteristics of the product. If a trend is identified as adverse, the Commercial Stability Site must report this.
5.2 Trending of stability data
Historical data for stability indicating parameters must be evaluated in order to assess the typical trend over time. This can include characterization of the typical trend, such as decreasing/increasing, linear/non-linear behavior, and calculation of the average change of the parameter characteristics from time of release to end of shelf life. The typical trend must be evaluated using statistical methods and be reviewed against existing specification limits.
If a significant trend is identified, the probability of an OOS must be assessed and if necessary, release alert limits for the product be recommended (see section 5.5). The alert limits should be evaluated periodically. Trending of stability data must be performed at least annually, however, it is recommended that stability data be reviewed as soon as possible after analysis in order to identify atypical results.
5.3 Type of stability study
Understanding the nature of the stability studies is required in order to appropriately assess the stability trend. There are two types of studies: typical stability studies and special stability studies. Typical stability studies are those that have been initiated on product that has been manufactured and packaged according to established processes. These studies encompass set-downs of e.g. annual maintenance batches. Data generated for these studies can be compared to the reference data where there have been no changes to processes or procedures of manufacture and packaging.
Special stability studies are divided in two categories: set-downs due to process deviations and as required by a change control. Studies initiated for batches where there have been process deviations are required to assess the impact of the process deviation on stability indicating parameters. The data generated for batches set down due to deviations should be evaluated against the reference data for the product. Data from studies on process deviations must not be included in the periodic review of stability data from typical production.
Trends on batches required by a change control must be compared with the reference data. Changes in stability profile must be reported to the formulation site(s) involved for further investigation as to the possible cause(s).If a change is being made permanent and the stability data after the change is different from reference data, the post change stability data must, when sufficient data is available, be considered to replace the reference data as the basis for setting or revising registered release limits or release alert limits.
5.4 Identification of OOTs
When new results from a stability study are available, the data must be evaluated as typical or atypical against the reference data. A visual inspection of the data, for example by using a scatter plot, may be sufficient to identify typical or atypical trends and to predict adverse trends.
If a visual inspection is not sufficient, further statistical evaluation should be initiated. It is important to take into consideration the amount of data that is being evaluated, as it may be too early in the study to have sufficient data to provide confidence that OOT predictions are accurate.
5.4.1 Atypical result
The review of individual results should be performed as soon as possible after the results have been obtained and ideally by the responsible laboratory. By doing so it is possible to either confirm an atypical result or identify the probable cause of the atypical result.
An atypical result is observed when a single result, while within specification limits, is aberrant, i.e. outside normal analytical and sampling variation as well as exhibiting a difference in the typical change over time. (See figure 1-2, appendix 1). When an analytical result appears to be atypical during evaluation of stability data, the actions must follow the steps outlined in appendix 2.
As the first step, the analytical result must be reviewed to confirm the result as atypical or due to an analytical error. If an atypical result is confirmed, the probability of an OOS must be evaluated. The minimum action required when an atypical result is identified, is to monitor the next time point. The need for adding an extra time point ahead of the next scheduled time point must also be considered.
5.4.2 Atypical trend and/or adverse trend
Examples of atypical trend situations are when one or more of the following events occur:
– A series of results within a study show a trend that is different in comparison with the trend of other studies
– At least two consecutive results within a study are confirmed as atypical
– An atypical result occurs in at least two studies
Examples of atypical and adverse trends are illustrated in simple scatter plots in appendix 3. When an atypical trend is identified, the actions must follow the steps outlined in appendix 4.
According to the flow diagram in appendix 4, an assessment must be made in order to confirm whether an observed trend is adverse or not, i.e. whether there is a probability for an OOS to occur. Studies where adverse trends have been identified must be monitored closely. Additional time points may be required before the next scheduled time point to further confirm the trend.
When a trend is regarded as adverse it is important to identify the reason for the study set down (typical batch, set down due to deviation or change control). It is also important to determine the scope of the adverse result or trend, that is to determine whether the adverse trend is isolated to one batch or whether the atypical trend is affecting many batches. In either case, an investigation must be initiated.
A detailed investigation should document the review and assessment of the data, the statistical models chosen and follow-up actions that may be required. For example, the registered release limits or release alert limits may require review to ensure that the established limits are set at an appropriate level or if changes are required as a result of the investigation.
Actions may also include, but are not limited to: requesting an investigation to determine process, formulation or testing changes, new analyst, equipment or instrument changes, or deviations associated with the batch. When an atypical trend that is not regarded as adverse is identified, the probable cause should be investigated. An example of such a case is shown in appendix 3, figure 8. When a trend is considered as not atypical, there can still be a possible OOS, which must be assessed according to the flow diagram in appendix 4.
5.4.3 Compilation of reference data
Data used for OOT evaluation must be gathered and compiled electronically with adequate precision. An appropriate number of studies must be selected in order to characterize the stability profile of a parameter. As the analytical variability increases, more studies will be required to serve as reference. The reference data should be reviewed periodically and the set of data revised if required.
When results from the last time point of a study are available, an evaluation must be done regarding if to include the data from this study in the reference data. The data can be incorporated in the reference data for that product, if it is considered typical. If the data is atypical, the reason for the study set down must be identified and if necessary the cause of the atypical results be clarified. Reference data provides a good background for precision that includes product, method and laboratory variability in order to accurately assess the trending for a particular batch. It may be necessary to include more decimals than are reported according to the specification, especially for parameters like degradation products and impurities.
5.4.4 Statistical evaluation
The data should be presented initially as a scatter plot, allowing simple visual inspection of the characteristics of the parameter over time. Make sure that the time axis is correct.
This method can be used to identify both atypical results and atypical trends by comparison with normal/typical stability results both within and between studies. If appropriate, the individual results or batches that are identified as atypical during visual inspection may be more closely evaluated using more sophisticated statistical methods.
Comparisons with other data at the corresponding time point or comparisons of slopes or other parameters that quantify the change rate are some examples. The model used depends on the data that is being evaluated.
5.4.5 Stability Study Alert Limits
Alert limits for stability data could be either non-statistical (absolute settings) or statistical (based on reference data).
Alert limits, based on reference data, can be used as an aid in identifying atypical results or atypical trends. When setting alert limits it is recommended that a statistical approach is used. There are several different methods, although none of them can be generally Applied.
The following items should however generally be considered:
Appropriate reference data (representative for the normal stability profile of the product’s stability indicating parameter). Availability and amount of data. Can data from several product/package combinations be used as reference? Type of data (single, multiple, near detection limits). Appropriate model fitted to the normal degradation profile (e.g. linear or non-linear).
Time points. When can action be taken? Choosing a suitable level of risk (avoiding false alarms). Logistics of implementation (different market specifications, pack sizes, strengths, time points etc.). Adequate precision, i.e. enough decimal places to enable evaluation. Once stability study alert limits have been implemented, they must be reviewed on a regular basis and revised if needed.
5.5 Release Alert Limits
Qualified Persons are responsible for releasing drug products with a high level of certainty and thereby assure that the quality of any batch released will remain within the registered life-time specification for the market that it is intended, throughout the registered shelf life when stored within the labeled storage requirements.
It may be necessary to apply tighter or additional controls at the time of release in cases where products show a significant trend in one or more test parameters studied during long term stability studies on the packaged product. This should be achieved by establishing ‘Alert Limits’ for release that take into account the following: The product and formulation site.
The level of humidity protection provided by the primary pack types in which the product may be packed. The climatic zones in which the product may be marketed Market specific registered release and/or end of shelf-life specifications, shelf lives and cautionary labeling statements. These variables may give rise to a series of alert limits for a test parameter on the same product with batch release being restricted where necessary to more protective packs (where applicable), markets with wider end of shelf life specifications and/or markets in less severe climatic zones.
In cases where the formulation site is uncertain of the downstream packing/marketing options to which a particular batch may be allocated (typical for many sold oral dosage forms) the alert limits should be set on a worst case scenario.