Acceptable Analytical Practices for Justification of Specifications

Published on: 
Pharmaceutical Technology, Pharmaceutical Technology-04-02-2007, Volume 31, Issue 4

The concept of Acceptable Analytical Practices (AAPs) was developed by the Analytical Technical Group of the Pharmaceutical Research and Manufacturers of America to share information about how the pharmaceutical industry has implemented chemistry, manufacturing, and controls and quality guidances of the International Conference on Harmonization and worldwide regulatory authorities. The AAP process identifies and addresses critical issues in which guidance is lacking, ambiguous, or contradictory. AAPs were designed to provide a forum where one could learn from the experience of experts in pharmaceutical analysis and enhance the understanding of analytical practices that reflect good science and sound regulatory compliance. This article summarizes the discussion points from a meeting regarding the Justification of Specifications topic.

Many factors are typically considered in setting quality specifications for pharmaceutical drug substances and drug products. The extent of development and commercial-scale manufacturing experience plays a key role in decisions about how acceptance limits are established. As part of the regulatory dossier, it is important that the sponsor provide key information that justifies the tests, analytical methods used, and acceptance criteria applied for quality assessment of a given drug substance or drug product. Appropriate justification data and scientific background can be critical to successful regulatory approval of the proposed specifications. Although prescriptive guidance on how to set specifications is not universally possible nor practical, greater clarity is desirable regarding what body of data and scientific information provides acceptable justification for a given specification and what general approaches are suitable for evaluating these data.

The contents of such a justification may encompass relevant development data, pharmacopeial standards, International Conference on Harmonization (ICH) requirements, batch data for materials used in toxicology and clinical studies, stability information, and data from validation and commercial batches. In some cases (e.g., content uniformity) there are accepted standard criteria articulated in either pharmacopeias or regulatory guidances. In other cases, such as impurities, there is an expectation that the limits are derived in some manner from the development experience.

In considering what development batch data to include, the first clinical batch and beyond are the critical ones, with greater emphasis placed on data for drug-substance batches made by the same synthetic route as that registered in the application. In the case of a drug product, the development batches that represent the same formulation and composition as that being registered are emphasized. Process experience at various scales of operation and the depth of process understanding also should be clearly communicated in the regulatory filing. In deciding what attributes require a specification, it can be just as important to justify which attributes do not require regulatory specifications as it is for those that do and which are deemed important in controlling quality, safety, and efficacy of pharmaceutical products.

The intent of specifications and relationship to "fitness for use"

One of the most difficult challenges in establishing and subsequently justifying specifications is achieving the appropriate balance among all factors, including patient safety and efficacy, scientific data, analytical variability, process knowledge and capability, regulatory requirements, and business issues. There is clear agreement that the safety and efficacy of a drug to the patient are the primary considerations, but beyond these absolute requirements there is less guidance on how to weigh the other aspects in establishing that a drug is fit for use. The greater the extent of the scientific data and understanding available for a quality attribute, the more prominent scientific considerations become in justifying the specification for that attribute.

When only limited manufacturing data are available at the time of filing, a key question is how to position the acceptance criterion for the attribute under consideration with respect to safety results (including acceptable exposure limits), efficacy, and the demonstrated process capability. It is very common that the limit established solely on safety considerations would be considered too loose by regulators while that based on a regulator's perception of process capability would be considered overly restrictive by a sponsor. A regulator's assessment of process capability is based on limited information available at the time of filing and very likely does not reflect the actual process capability. (Throughout the remainder of this article process capability will be used to represent the zone described by both the variability and the position of the mean of the process, both at development or commercial scale. There is no generally accepted definition for this term in the context of pharmaceutical specifications, leading to widely varying interpretations of this concept, particularly between industry and regulators.)

Tighter specifications result in limited ability of the sponsor to extend the shelf-life when additional long-term data become available and an increased risk of drug shortages caused by the rejection of safe and effective batches. The scenario of having limited data at the time of registration is a frequent situation for new products and lends itself to the consideration of alternative specification modes such as adopting an interim specification that is again reviewed when additional data are available. On the other hand, some tests that are not driven by process capability, including assay, content uniformity, and pharmacopeial tests, would not require this strategy and could be unambiguously set at the time of filing. In those cases, the development task is to determine whether the product can meet these standard specification requirements.

By contrast, the conventional approach used to establish acceptance criteria that are not clearly dictated by pharmacopeias or guidance is as follows:

  • Establish a safety threshold or set of product performance-based requirements that provide outer boundaries for possible acceptance limits.

  • Gather development information concerning the robustness of the proposed process to understand the stability behavior of the drug substance and product.

  • Based on the above information and the manufacture of a relatively small number of development batches, propose a limit that reflects the process capability of the proposed process.

Several current initiatives are prompting considerable discussion on other approaches to setting specifications. However, because the conventional approach is often followed and was discussed extensively at the September 2003 AAP workshop, this paradigm is assumed for much of this article.

Approaches to data treatment and process capability

If acceptance criteria are to be set on the basis of process capability, then the data used must be representative of the proposed commercial process (active pharmaceutical ingredient [API] or drug product). As stated previously, this requirement limits the scope of batches for consideration to API batches prepared by the final process and drug-product batches of the intended composition prepared by the formulation process to be used for commercial production. Consideration should be given to the scale of the relevant lots and the ruggedness (insensitivity to changes) of the process in question. The net result is that at the time of filing a new drug application a relatively small number of relevant batches are typically available for assessing the process capability of either an API or drug product process.

Ruggedness of acceptance criteria

In general terms, establishing a limit based on process capability starts with estimating both the mean and the variance (or some measure of variability) of the data for a particular quality attribute. In setting the limit, these estimates are then used to establish a criterion that provides both an acceptable operational window and addresses all safety and efficacy considerations. Several approaches for calculating a limit have been proposed. The ICH Q6A and Q6B guidances (1, 2) suggest using the mean plus three standard deviations for impurities or a related variant for degradation products. A tolerance limit or prediction-interval approach, which considers individual values rather than focusing on the mean, is preferred by industry representatives. (Tolerance limits or prediction intervals are concerned with inference about single observations. The concept is to establish limits on expected individual values based on knowledge of the population parameters. These statistics estimate a range in which future individual results are expected to fall within a specified confidence.) The most conservative approach is to set the limit not to exceed the values associated with lots in the development program. Note that the situation is more complex for specifications where some change over time is expected and must be accommodated, such as for degradation products.


The industry advocates that statistical evaluation is of value in establishing acceptance criteria for impurities, including degradation products and residual solvents. Although the mean can be reasonably estimated from relatively small datasets, estimating variability inherently requires more data. To illustrate and expand on this point, a hypothetical case (see sidebar "Ruggedness of acceptance criteria") was used to consider the ruggedness of acceptance criteria derived from three general approaches: mean ± 3 standard deviations, tolerance limits, and maximum value. The general industry consensus is that on the order of 15–30 relevant batches are required to estimate statistically meaningful and robust acceptance criteria for a given test. Concerning the analysis of data to estimate appropriate acceptance criteria, the industry preference was to use a tolerance limit approach, which is considered an appropriate statistical approach to making inference about individual values and is therefore relevant for establishing statistically justified limits based on process capability.

Shelf-life considerations

As noted previously, the situation is more complex for specifications in which a parameter is expected to change over time. The amount of change with time must be estimated, and specifications will need to accommodate this change. This is often the case for degradation products, for active ingredient assays or potency estimations, and for other important quality parameters. It then becomes important to distinguish between those specifications that apply throughout the expiry period and those that should apply only at the time of batch release. Where use-related limits are specified in advance (i.e., there is already a range linked to safety and efficacy concerns), appropriate allowances must be calculated to take the change with time into consideration. Requirements may need to be more stringent at the time of release, and manufacturing capability to this requirement would need to be scrutinized. Where limits cannot be specified in advance, and the process capability paradigm described above is used to set release specifications, the effect of the change with time needs to be accommodated in developing the appropriate shelf life specifications (3).

Tests to consider for potential skip testing*

During stability studies, many batches are tested on many occasions. Even without any change with time, this additional testing may carry with it additional risk through the simple compounding of the probability of an out-of-specification (OOS) result. The design of the shelf life specification must appropriately accommodate such risks to avoid unexpected and unintended consequences.

Specifying impurities

Justification of the specifications established for impurities and degradation products is an area that has received considerable attention. ICH guidances Q3A(R) and Q3B(R) address the regulatory filing requirements for the reporting, identification, and qualification of impurities and degradants, together with a general outline of the impurity specifications (4, 5). The guidelines further indicate that the rationale for specifying an impurity (or not) should be explained on the basis of process understanding, data from development batches, stability studies, and batches made by the commercial process.

During development, it is common for impurities to be reported and tracked as processes are modified, optimized, and then scaled up. Impurities initially observed in development batches may be minimized or no longer be present in material prepared by the final commercial process. The industry agrees that potential impurities not observed in the commercial process should generally not be specified. Nonetheless, when there is limited manufacturing experience or outsourced materials or intermediates are used with limited experience, it may be appropriate to include an impurity specification for a potential impurity, possibly as an interim or sunset specification, until additional data are available.

Interim specifications

An interim acceptance criteria (specification) for a specific test generally involves setting a provisional limit on a quality characteristic of the drug substance or drug product at the time of approval. This provisional limit usually involves a postapproval commitment to review the limit as more data become available and is made by prior agreement with FDA. In general, there is strong industry support for this concept of interim specifications, but some reservations exist about how this would be specifically implemented. There is a need for additional clarity from the regulators regarding implementation of this concept.

As discussed previously, limited batch data do not allow a robust estimate of process capability. If criteria are to be derived from process capability, then this lends itself to an interim approach that allows industry to gain more experience with the commercial process and better estimate the process capability. Note that limited data can both under-estimate or over-estimate the appropriate limits. The need for interim specifications should be justified based on safety, process capability, and statistical considerations. Industry would prefer a wider limit until more data are available. For example, limited scale experience is cause for manufacturers to be concerned about committing to an official limit for a specific impurity, especially when safety and efficacy are not an issue. Generous interim limits are a desirable approach to the manufacturer that does not want to encounter needless rejection of a batch where there is no risk to the patient. Setting specifications very close to the manufacturing capability also may impact the ability of the manufacturer to improve other aspects of the process (e.g., environmental impact, cost reduction, stability of supply chain). Specific parameters where an interim specification may be appropriate include impurities, degradation products, dissolution (e.g., especially for highly water-soluble drugs or revision or replacement of a test for an existing product), residual solvents, and particle size.

Tests to consider for potential sunset testing

The industry supports the use of interim specifications, where appropriate and when safety is not an issue, at the time of filing. This could be justified at the time of regulatory filing with an interim limit up to the safety threshold. In turn, FDA expects a specific Phase 4 commitment from the new-drug-application holder that defines the number of batches or the time frame over which further experience will be obtained, together with the data-analysis plan to facilitate a timely justification of the final specifications.

Postapproval acceptance criteria changes

The industry considers multiple options viable for specification changes postapproval, including follow-up on interim specifications. One option may be to make a commitment to re-evaluate the specification, but leave the specific timing for this re-evaluation open until data and statistical justification are available. However, providing a protocol at the time of filing with plans for the number of batches, statistical treatment of the data, and so forth is a proposed approach. For example, revision of dissolution acceptance criteria in response to improvements of the dissolution test method for an existing product might best be handled with a comparability protocol covering the expected number of batches to be evaluated before finalization of the dissolution specification. Similarly, for a new impurities method in an older product, a commitment may be needed to define the number of lots needed.

Periodic or skip testing

Periodic or skip testing involves performing specific tests at predetermined intervals, rather than on a batch-to-batch basis, for release of the drug substance or drug product. Those batches not being tested still must meet all acceptance criteria established for the drug substance or drug product (1). The obvious benefit is significant reduction in the cost of testing and release cycle time. On the other hand, the sponsor assumes the risk that if a product fails to meet specifications at a given test point, products released to the market since the last testing point may potentially be considered out-of-specification. Depending on the outcome of an OOS investigation into root causes and an evaluation of product impact, there is a potential for product recall from the market. Thus, skip testing can be used only when extensive experience shows that there is low risk of failure and that a potential failure would result in low risk to the patient. The use of process analytical technologies (PAT) for in-process monitoring may be an acceptable justification for consideration of skip testing in that particular case.

Although the concept is recognized in the ICH guideline, skip testing of finished drug products is not in widespread use because of a perceived regulatory risk or reluctance of regulatory agencies to approve skip-testing release protocols (1). Most firms, however, do use skip testing in conjunction with supplier certification programs for the release of excipients. Risk is limited because the qualification or certification process usually requires the supplier to test these materials on a batch-by-batch basis, and each batch has been tested at least once. Unlike raw materials, finished products are at greater risk because actual batch analysis data are not available on skipped lots. This risk may be limited by the availability of data from related tests or by in-process testing or by data from process analytical monitoring.

One example of a case in which skip testing is routinely used is for the monitoring of microbial attributes for certain low-risk solid oral-dosage forms. Generally, this practice is widely accepted in many European Union countries. Other examples of tests that the industry considers candidates for skip testing are provided in the sidebar "Tests to consider for potential skip testing."

The industry strongly supports the application of skip testing when appropriate. Because only limited data are usually available at the time of filing, skip testing typically would be implemented as a postapproval change, requiring justification and preapproval by a regulatory agency before use. Nonetheless, conditions and criteria for the introduction of skip testing can be specified in the original registration applications.

Sunset specifications

The concept of sunset testing involves establishing a predetermined time interval or number of batches as a postmarketing commitment after which testing of certain attributes would be discontinued if specific criteria were met. It differs from skip testing insofar as after the commitment is met, the test would no longer be part of the product specifications. This strategy is particularly useful when only limited data are available at the time of submission and there is low risk to the patient. In some cases, a sunset-testing approach could be used in conjunction with skip testing (i.e., skip-testing intervals are gradually increased until finally the test is retired from routine testing). Many of the same considerations discussed for skip testing apply to sunset testing. Examples of tests the industry considers candidates for sunset testing are listed in the sidebar "Tests to consider for potential sunset testing."

Parametric release

The concept of parametric release involves making release decisions by measuring critical operational parameters for a given process in lieu of direct measurement of the quality attribute itself (1, 6). A typical example of a parametric release is sterility testing for terminally sterilized products. In this case, the release of each batch is based on satisfactory results from monitoring specific parameters such as temperature, pressure, and time during the sterilization process instead of conducting actual sterility testing. Parametric release is also commonly used to release in-process materials. For example, solids mixed in a V-blender are essentially released for further processing by the measurement of operational parameters, including the mixing time and the rotation speed of the blender, provided that the parametric endpoints have been validated to ensure adequate mixing. In some cases, blend uniformity is verified by off-line chemical testing of the powder blend or stratified testing of the finished product (7). In principal, the concept can be extended to other types of testing.

Building on parametric release, the FDA has proposed a concept of real-time release (8) in which quality attributes are measured either directly or indirectly in real time during processing using techniques of PAT. Thus, in the blending example given above, parametric confirmation of the blending operation would be replaced with data from real-time sensors that would establish when satisfactory blending had been achieved.

Through the use of in-process testing and process analytical technology (PAT) controls, there is a potential to substantially decrease or eliminate end-product testing while improving compliance and reducing rework or rejection. The implementation of PAT would either eliminate the need for conducting finished-product testing of particular attributes or would support skip-testing or sunset-testing protocols by greatly reducing the risk of failure and risk to the patient. Using a "science and risk-based approach to good manufacturing practices" in conjunction with strategies to reduce testing costs will provide benefits to both the industry and the patient (9).


Fundamental questions about setting specifications for pharmaceutical products are currently receiving much attention for both traditional therapeutic agents and biopharmaceuticals. Through its initiatives on PAT and CGMPs for the 21st Century, the US Food and Drug Administration has provided the pharmaceutical industry with a clear mandate to take a new look at the way product quality is defined and controlled (8, 9). These initiatives are stimulating considerable discussion with regard to better ways to set standards that ensure the safety and efficacy of pharmaceutical products. The process of establishing and maintaining meaningful product specifications throughout the product life cycle is an important part of that process. Moreover, traditional concepts of demonstrating that processes are robust and operated in a state of control using process validation and end-product testing are being challenged by the concept of real-time process monitoring, which will potentially lead to increased process robustness, enhanced in-process controls, and presumably less of a need for verification through end-product testing.

In response to these evolving concepts of quality management, other approaches and tools to establish specifications have been proposed and currently are under consideration. Some of these concepts such as periodic or skip-lot testing, parametric release (i.e., real-time release), and the use of interim specifications to address inherent problems in deriving specifications from limited data sets have been acknowledged in principle in the ICH Q6A and Q6B guidance on setting specifications. On the basis of the results of the AAP Workshop on the Justification of Specifications, there is great interest and support for these initiatives on the part of industry. Other initiatives, including sunset testing, also were strongly supported, particularly for products in which a significant body of data from postmarketing experience indicated that the tests were no longer value-added controls or were adequately controlled by another test (i.e., either a finished product or in-process control test). Finally, the use of appropriate statistical approaches to determining acceptance criteria to set robust specifications deserves further emphasis. Such approaches must balance the evaluation of process capability, shelf-life considerations, and the ultimate safety and efficacy requirements in the face of limited-size datasets.

Further dialog between industry and regulatory agencies regarding these and other initiatives is needed and will lead to new approaches to establishing performance-based specifications and controls that hopefully will shorten development time to market for new products and ultimately reduce the overall cost of development and manufacturing.

Paul Kurtulik, PhD, is executive director, analytical R&D and quality operations, at Celgene Corporation. Terry Tougas works at Boehringer Ingelheim Pharma's CMC development office. Paul K.S. Tsang, PhD, is director and quality site head at Amgen. Edward Warner is director of global quality statistical services, Schering Plough Corporation. Jean M. Wyvratt, PhD,* is vice-president, analytical science and quality testing, global pharmaceutical commercialization at Merck & Co., tel. 732.594.7174,

*To whom all correspondence should be addressed.

Submitted: Oct. 18, 2006. Accepted: Dec. 8, 2006.Keywords: analysis, compliance, guidance, PhRMA, specifications.


1. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, ICH Harmonized Tripartite Guideline Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances Q6A (ICH, Geneva, Switzerland), Oct. 1999.

2. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, ICH Harmonized Tripartite Guideline Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products Q6B, Mar. 1999.

3. G.C. Davis et al., "Rational Approaches to Specification Setting: An Opportunity for Harmonization," Pharm. Technol .15 (9) (1991).

4. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, ICH Harmonized Tripartite Guideline: Impurities in New Drug Substances Q3A(R) (ICH, Geneva, Switzerland), Feb. 2003.

5. International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, ICH Harmonized Tripartite Guideline Impurities in New Drug Products Q3B(R) (ICH, Geneva, Switzerland), Nov. 2003.

6. EMEA Committee for Proprietary Medicinal Products, "Note for Guidance on Parametric Release," CPMP/QWP/ 3015/99, Mar. 2004.

7. FDA, Guidance for Industry: Powder Blending and Finished Dosage units - Stratified In-Process Dosage Unit Sampling and Assessment, Draft Guidance Oct. 2003.

8. FDA, Guidance for Industry: PAT—A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance, Draft Guidance, p. 17 (Aug. 2003).

9. PQRI/FDA Report, presented at "A Drug Quality System for the 21st Century," Apr. 22–24, 2003, Washington DC (report prepared on June 16, 2003).