Current Thoughts on Critical Process Parameters and API Synthesis

A task force formed under the auspices of the Pharmaceutical Research and Manufacturers of America (PhRMA) Active Pharmaceutical Ingredient (API) Technical Group recently published an article about four aspects of the US Food and Drug Administration's February 1987 Guideline for Submitting Supporting Documentation in Drug Applications for the Manufacture of Drug Substances (1, 2). One of these aspects was the "control of critical steps and intermediates," with particular reference to Section 3.2.S.2.4 of the Common Technical Document (CTD) (3). Although this article was published before FDA issued its 2003–2004 draft guidances for drug products and drug substances (4), the authors recognized the potential value of greater clarity and commonality of understanding among PhRMA industry members and health authorities on the subject of critical process parameters (CPPs).

Definitions

This article reviews critical quality attributes (CQAs), CPPs, and the associated concepts of critical in-process controls (CIPCs) and critical steps. The intent of this article is twofold:

• to provide common language at the technical and practical level among industry and health authorities about how CQA and CPP concepts relate to current industry practice in API and drug substance manufacturing;

• to illustrate how applying these concepts in a focused way supports drug quality good manufacturing practice (GMP) systems and regulatory dossiers and aligns with on-going science- and risk-based regulatory initiatives.

A common understanding of these concepts—from product development and registration through life cycle management and subsequent quality evaluations of process changes—demonstrates that the industry has extensive technical and practical knowledge about the API process. Thus, presentation of only key portions of this technical knowledge in Section 3.2.S.2.4 of the CTD or in support of internal quality evaluations could logically lead to more streamlined and science-based regulatory dossiers. Consequently, this approach will help increase patient safety and drug efficacy without creating an undue burden of added submission or review activities as a result of excessive or unfocused dossier information requirements.

For the purposes of this article, the terms drug substance and API are used interchangeably, because both are used in various regulatory guidance documents. The terms guidance and guideline also are used interchangeably for similar reasons.

Objectives

The concept of critical is generally understood to be a feature or aspect of a process that requires careful control or execution to ensure that the API is of acceptable quality, with the desired end result being a safe and efficacious drug product. Despite this general understanding, the task of assigning some features of a process as critical is inherently subjective, and there are considerable differences of interpretation in the industry and health authorities.

Therefore, objectives of this article are the following:

• Provide a common language and understanding to facilitate future discussions of the topics noted above.

• Define critical quality attributes, critical process parameters, and critical in-process controls and show their interdependency in the context of chemical and physical characteristics of the drug substances in a particular drug product.

• Establish the principle that specifications may contain parameters or attributes other than CQAs and that in-process tests often exist for reasons that are not critical to quality.

• Provide a methodology for defining CQAs, CPPs, and CIPC, in a given synthetic process.

• Discuss the relationship between process performance and CQAs, CPPs, and CIPCs.

• Clarify the identification of CQAs and CPPs in regulatory submissions.

• Provide linkage to on-going industry initiatives in quality risk assessment and the integration of technical knowledge with quality systems.

• Identify the regulatory opportunity of reducing supplemental filings for postapproval changes that do not significantly affect API quality.

PhRMA has published its opinion that the focus of process understanding and control must center around CQAs and CPPs. Nevertheless, the term critical step appears in various regulatory guidances. For example, the International Conference on Harmonization (ICH) Q7A Good Manufacturing Practices for Active Pharmaceutical Ingredients states that industry must validate critical process steps and the quality unit must review and approve batch records for critical process steps (5). Therefore, this article also provides a practical discussion of critical steps.

With respect to the remaining concepts of CQAs, CPPs, CIPCs, and the importance of characterizing the significant aspects of a particular process, this knowledge provides additional business benefits. Some examples are:

• Clarifying what constitutes a robust process during process development and the typical industry approaches to generating data that support a robust process (thereby illustrating that current industry practice is consistent with FDA's "quality by design" initiative).

• Providing a reference point to methods used by many companies in managing risk, thus laying the foundation for future science- and risk-based decisions. This knowledge also is a key element in the principles outlined in the draft "ICH Q9 Quality Risk Management" document.

• Providing a framework for deciding which technical information to include in CTD Section 3.2.S.2.4, "Control of Critical Steps and Intermediates" and which more appropriately remain part of the supportive technical body of knowledge found in a process's history.

• Clarifying the relationship between CTD Section 3.2.S.2.4, "Control of Critical Steps and Intermediates" and the selected validation acceptance criteria, and, in particular, identifying why there may be a difference between the two.

A brief history of applicable regulations

The 1987 FDA guidance refers to critical steps and establishes the connections among the critical steps, the testing regimens for intermediates, and the ultimate validation of the process (2). Section 2.E, "Intermediates and In-Process Controls," states:

"The regulations require that controls ... be employed ... to assure that ... procedures are operating properly and that the intermediate tested is suitable for subsequent processing. The choice [of intermediates, steps, and testing] is the responsibility of the applicant .... As experience is gained with the synthesis, the critical reaction steps and intermediates to be monitored are selected .... The whole operation is part of the process validation of the synthesis."

Since 1987, the concept of process validation has evolved substantially. It is now standard practice to predefine acceptance criteria for the steps being validated. These validation acceptance criteria typically arise from the need to demonstrate the appropriate control of critical operations within the process. Acceptance criteria typically consist of a requirement for selected critical unit operations to remain within predefined acceptable operating limits during processing and/or in-process or intermediate tests and specification limits that ensure acceptable API quality will be achieved.

In January 2004, FDA published a draft guidance for industry for drug substances (4) that defined critical as "a process step or process control (e.g., process condition, test requirement, or other relevant parameter or item) that must be controlled within predetermined criteria to ensure that the drug substance meets its specification" (lines 2122–2124) (4).

This definition is consistent with the ICH Q7A definition of critical as "a process step, process condition, test requirement, or other relevant parameter or item that must be controlled within predetermined criteria to ensure that the API meets its specification" (5).

The ICH Common Technical Document M4Q Quality, Section 3.2.S.2.4, "Controls of Critical Steps and Intermediates," defines critical steps as "tests and acceptance criteria (with justification including experimental data) performed at critical steps identified in 3.2.S.2.2 of the manufacturing process to ensure that the process is controlled should be provided" (3).

A common theme among these regulatory documents is that the process leading to the drug substance itself should be controlled, not just the resulting final drug substance or API. In other words, there should be an adequate evaluation of quality during the process to ensure the acceptable quality of the final drug substance. Another common theme is one of a critical step—the idea that there is a subset of steps in a chemical process that must be carefully controlled or executed to ensure the API produced is of acceptable quality.

Introducing a science-and risk-based approach

The foundation of a science-based and risk-based approach can be taken from many recent draft guidance documents. Common to these documents is the use of a science- and risk-based approach for defining and controlling products and processes. The three main components of such an approach are:

• a well-defined product and a well-characterized process;

• the use of sound risk-based decision principles;

• the integration with well-defined quality systems.

This is consistent with FDA expectations articulated in the 2002 initiative Pharmaceutical CGMPs for the 21st Century: A Risk-Based Approach and in the process analytical technology (PAT) final guidance PAT—A Framework for Innovative Pharmaceutical Development, Manufacturing and Quality Assurance (6, 7). These documents state the goals are intended to ensure that:

• "The most up-to-date concepts of risk management and quality systems approaches are incorporated into the manufacture of pharmaceuticals while maintaining product quality."

• "Manufacturers are encouraged to use the latest scientific advances in pharmaceutical manufacturing and technology."

FDA's current thinking on well-characterized is summarized in the PAT guidance (7):

"A process is generally considered well understood when (1) all critical sources of variability are identified and explained; (2) variability is managed by the process; and, (3) product quality can be accurately predicted over the design space established for materials used, process parameters, manufacturing, environmental, and other conditions."

In the guidance, FDA further expands on the concept of mitigating and reducing product risk through proper understanding of the process and placement of controls appropriate to the risk. The use of sound risk management principles coupled with scientific understanding can identify and control sources of risk to the product. As the PAT guidance states,

"The approach is based on science and engineering principles for assessing and mitigating risks related to poor product and process quality. In this regard the desired state of future pharmaceutical manufacturing and regulation may be characterized as follows:

• Product quality and performance are ensured through the design of effective and efficient manufacturing processes

• Product and process specifications are based on a mechanistic understanding of how formulation and process factors affect product performance ....

• Risk-based regulatory approaches recognize ... the capability of process control strategies to prevent or mitigate the risk of producing poor quality product."

ICH also has recognized the importance of process knowledge and identifying and controlling risks associated with the product by forming Expert Working Groups on Quality Risk Management (Q9) and Pharmaceutical Development (Q8).

Irrespective of whether one views the API regulatory landscape from the initial development and registration perspective or from the viewpoint of subsequent quality management and process improvement (life cycle management to more easily implement process improvements is a logical extension of using process knowledge), the science- and risk-based approach can be an appropriate model. To identify the well-defined product and well-characterized processes that this approach requires, one can first focus on the CQAs associated with the product.

Critical quality attributes as the core concept

The position presented in this article is that the CQA is the core concept linking any proposed science- and risk-based regulatory approach to a basic technical understanding of an API process and related terms such as critical steps, CPPs, and CIPCs.

A typical API process. A typical API process consists of several stages or steps, either chemical transformations or purifications that may involve isolated or nonisolated intermediates. Each stage or step comprises numerous unit operations, most of which are typically operated within certain defined ranges according to a batch sheet or a manufacturing order. In-process testing at various points within a stage may be conducted to confirm that a specific point or condition has been achieved before progressing to the next step of the process. In addition, isolated intermediates can be analyzed, and their quality can be determined against specifications. All of the individual steps, their respective unit operations and associated ranges, in-process controls, and specifications make up the API process.

Purposes of process control. Operating ranges for the unit operations in a process, in-process testing regimens, and specifications for reaction intermediates can be defined for several reasons, including:

• quality: A given range is necessary to ensure that the product of a stage is of appropriate quality to produce acceptable API.

• yield: A given range may maximize the yield of a required product.

• efficiency: A given range may ensure the process is completed in a short period of time.

• safety: a given range may be necessary to minimize the hazards associated with a process.

• consistency: A range may be defined for no other reason than to ensure consistency of operation from one batch to the next.

• environmental: A range may be defined to ensure compliance with state, local, or federal environmental standards and permits.

Rationale to focus on CQAs. Of these many valid reasons, only those which ultimately affect patient safety or product efficacy through API quality are relevant to critical steps, CQAs, and CPPs. By defining and focusing on criticality according to quality only, it is possible to closely follow FDA's science-based and risk-based regulatory approach, which also is consistent with the ICH Q7A Section 12.1. This approach also makes it possible to focus the registration (NDA/CTD filing) of CQAs, specifications, in-process controls, and CPPs on those values related to the ultimate quality of the API while understanding that a larger body of other information may have been developed during product and process characterization.

"The critical quality attributes process-assessment approach

Fundamental in the determination of CQAs and any resultant critical steps and CPPs is relating process variables to their potential effects on the end product, the API. As noted in ICH Q7A Section 12.1 (5),

The critical parameters/attributes should normally be identified during the development stage or from historical data, and the ranges necessary for the reproducible operation should be defined. This should include:

• defining the API in terms of its critical quality attributes;

• identifying process parameters that could affect the critical quality attributes of the API;

• determining the range for each critical process parameter expected to be used during routine manufacturing and process control."

This approach is summarized in Figure 1. First, the drug product is defined. Different drug products have different API quality needs. For example, particle-size homogeneity may be important for tablet dosage forms but not of importance to the quality of solutions. Similarly, microbiological attributes are especially important for quality sterile products. Dosage forms that have complex formulation technologies may suggest the need for APIs possessing certain physical characteristics. It is the drug product—the end use—that determines the CQAs of the API. This focus on the CQAs of the API provides the basis for science-based decisions during product development and life cycle management.

Figure 1: The critical quality attributes (CQAs) process-assessment approach.

Conducting this exercise yields a list of CQAs, including critical impurities that have been observed at some point in the final API. These should be the observed synthesis-related impurities and not the result of speculation. Degradation impurities that are not synthesis-related but storage-related are not included for further CQA assessment. Although synthesis steps or process parameters may be important and remain part of the final API specification, if they cannot lead to the formation or removal of these degradation impurities, then they should not be considered CQAs for the purpose of process risk assessment.

Once CQAs are identified, each API synthesis step is judged to assess its potential effect on a given CQA because each synthesis step does not equally influence individual CQAs. Then, potential CPPs are identified within each chemical synthesis step. Several iterations may be needed to assess whether potentially critical steps or parameters can be eliminated by means of risk mitigation measures or increased process understanding. Synthesis steps not affecting the CQAs by definition do not contain CPPs and do not require further risk assessment.

This approach moves the discussion of CPPs into the scientific realm of mechanisms, chemical properties, and physical properties. Product and process evaluation becomes focused around how, when, and where to control the synthesis to achieve the desired CQAs in the final API. Mitigation of risk is linked to technical understanding of the process and product. Further, it may be possible through robust process design, control, or other risk-mitigation measures to eliminate the CPPs at a given synthesis step so that, ultimately, the step is no longer critical and no control is needed.

Each step in this process-assessment approach is elaborated in the following sections.

Defining CQAs and specifications. CQAs are API characteristics that are desired or needed to ensure patient safety and benefit, with an unsafe API defined as one of unacceptable quality risk. Quality attributes are not the same as specifications, however. A quality attribute is an intrinsic quality characteristic of the API. A specification is "a list of tests, references to analytical procedures, and appropriate acceptance criteria, which are numerical limits, ranges, or other criteria for the tests described" that serve as a measurement of that attribute (8). First, one must identify and list all of the quality attributes of an API and then identify those considered CQAs based on the API's use in the drug product. ICH Q6A can serve as a basis for this determination.

According to Q6A, some attributes should be considered critical, regardless of the drug product end use. Additional attributes should be considered critical, depending on the final drug product. Table I lists the attributes found in Q6A along with an outline for determining the CQAs.

Table I: Quality attributes.

Identification, physicochemical properties, appearance, assay, and purity are applicable to all drug products. Particle size, microbial purity, and polymorphism depend on the drug product. Q6A decision trees can be used to determine the criticality of these quality attributes for solids, solutions, or sterile products. Only those specifications controlling CQAs should be listed in the CTD.

If one chooses to evaluate legacy products, it is important to first confirm the CQAs of the API and not from any existing specifications. It is not uncommon to define legacy products with more than one specification measuring the same attribute (e.g., base assay and HPLC assay) overlaid with specifications linked to historical practices. Specifications for the API should be linked to a specific quality attribute being measured and controlled to develop a set of current, science-based specifications.

CQAs of APIs or intermediates. CQAs of an API sometimes may be controlled by controlling a specific intermediate, either isolated or nonisolated. This control can be achieved by using an in-process test and control limit or by means of an in-process operating range (i.e., parameter), which will be elaborated later in this article. Most important, the control must ultimately be linked to the API's acceptable quality—as the intermediate's CQA must, by definition, affect the API. In this context, we introduce one additional element into CQA discussion as a major subset of possible CQAs—the critical impurity. This concept can be useful to justify appropriate control at an intermediate process point instead of at the final API stage.

Critical impurities: typical CQA for intermediates. The concept of a critical impurity, defined as "an impurity which affects the impurity profile of the API or leads to an impurity that affects the impurity profile of the API," allows further focusing on potential CQAs through initial examination of the API process. Two examples follow.

An example of the fate of impurities for a noncritical impurity is shown in Figure 2. In this case, impurity A in the first synthesis step leads to impurity B in the second step and so on. The impurity eventually is removed in the fourth step by means of selective crystallization. Because none of these impurities affects the impurity profile of the API, they are noncritical and do not lead to a CQA for the API or an earlier intermediate.

Figure 2: Example of the progression of a noncritical impurity.

An example of the origin and fate of impurities for a critical impurity is shown in Figure 3. In this case, impurity A in the first synthesis step is not removed by later process steps and gives rise to impurity D in the final API. In this case, not only are the impurities in the API critical, but each of the impurities B, C, and D in the chain is potentially critical as well.

Critical impurities can be identified through knowledge of the chemistry and process capability, batch data reviews, spiking experiments carried out in development, or other appropriate means. In Figure 3, critical impurity A is formed at the first step. This step should be examined to determine the parameters that lead to the formation of impurity A. These parameters would be potential CPPs. The other steps in the synthesis should be examined to determine which have the greatest effect on removing impurities B, C, or D in the chain to ensure that the API meets its specification for organic purity. Any subsequent process step that could potentially remove the critical impurity may also lead to potential CPPs. Thus, it is apparent that CPPs can address either the formation or the removal of critical impurities or their precursors.

Figure 3: Examples of the progression of a critical impurity.

CQA process-impact assessment: process control versus quality system control. Not all CQAs are affected by the chemical synthesis process. The process affects some CQAs, but others are quality system- (i.e., GMP-) controlled. For example, the synthesis process can affect organic purity because temperature or mole ratio of a certain intermediate step can lead to the formation of a critical impurity. Because the process can affect the organic purity, each of the synthesis and purification steps should be evaluated for its effect on organic purity.

The attribute "identification," however, is not influenced by the chemical synthesis itself, but by the GMP control system that ensures that appropriate batch records, raw materials, and labeling are used. Process parameters such as temperature, mole ratio, or time do not affect the identification attribute.

Only CQAs that can be influenced by the synthesis process must be considered further in the CQA process assessment for defining CPPs (critical process parameters) and CIPCs (critical in-process controls). The next step in determining criticality is to identify potentially critical parameters in the chemical synthesis steps of a given process. Once identified, these parameters are evaluated for the probability of their effect on API quality. From this reduction process, the CPPs and associated critical synthesis steps and controls or risk-mitigation measures are identified.

Critical process parameters in a risk-based approach

Many steps are associated with a quality risk-management process. This article focuses on the following two recognized key components of risk assessment:

• identification of the possible events or failure modes;

• estimation of the probability of the event occurring.

In the case of CPPs, the first phase is to identify those potential CPPs that may affect the CQAs of the API (i.e., the severity). In this phase, process capability or robustness is not taken into account. The second phase of the assessment is to estimate the probability of a parameter affecting the API (i.e., the risk), taking into account process robustness and equipment capability. The parameters with a high probability of influencing the API with enough severity to produce unacceptable API are critical parameters.

The general trend is that potential risk to the final API increases as the synthesis moves closer to the final API. It is often the case to find more potential CPPs closer to the end of the synthesis. This trend is in agreement with the accepted Q7A philosophy that "the level of GMP should increase as the API synthesis progresses."

Identifying potential CPPs. The first phase is to identify potential CPPs—that is, the specific process parameters that may affect particular CQAs. The evaluation takes into account experimental knowledge as well as practical experience. This exercise is rooted in an understanding of the synthetic process and the process equipment used. If impurity A is critical and is formed in a given step, then specific questions are asked for the purpose of identifying what influences the formation or removal of impurity A. The answers are derived from process knowledge and understanding.

There is an important distinction between an operation that is difficult to perform and one that contains potential critical parameters that may affect API quality. Process parameters (i.e., operating ranges) are set for several reasons (e.g., process safety), as noted in the previous discussion of a typical API process. A process operation may be difficult or an operating range may be narrow to optimize a variable that is not associated with API quality. Only those parameters with potential CQA or API influence are taken into account for further analysis as CPPs.

In addition, certain ground rules can help focus the evaluation of potential CPPs. It is safe to assume that all parameters, if taken to improbable extremes, will affect quality. For example, consider a reaction that may be routinely operated successfully between 30 and 50 °C and can tolerate temperatures well in excess of 50 °C without affecting quality. If heated to 150 °C, however, materials will decompose, the reaction will turn black, and quality will be affected. Because temperature excursions of this magnitude are not likely, there are no circumstances in which this process would be described as "sensitive" to temperature. Chasing such hypothetical discussions is fruitless. A reasonable approach to identifying potential CPPs would be to:

• focus on the CQAs being influenced by the synthesis step;

• avoid absurd and extreme guessing and what-if scenarios;

• assume each process step is there for a reason and will be performed as written;

• avoid consideration of the influence of items that are controlled by the GMP quality system. (It is part of the initial assumptions that the GMP system will ensure calibrated and clean equipment, identification and labeling of bulk raw materials, and so forth.)

These steps compose a practical and logical approach to selecting the potential CPPs that become the focus of the next stage of the evaluation.

CPPs risk assessment through operating ranges. Once potential CPPs are identified in the first part of the risk-assessment process (severity being linked to acceptable API quality), they are evaluated for their probability of influencing API quality (i.e., the likelihood they might occur). To do this, the routine process-operating range and process control capability for the parameter is assessed against the proven range.

The proven range for a process parameter is the range in that parameter shown to yield acceptable product. The proven range represents the area in which one has process knowledge, with or without knowing the edge of process failure. Indeed, the proven range may not be close to the edge of failure. A process may be sufficiently robust that further experimentation beyond a given proven range, to find the edge of failure, is not value added.

The critical nature of a process parameter is not defined by the wideness or narrowness of the operating range. If the defined operating range were wide and close to the proven range for a potential CPP, then—according to the definition—it would be assigned as a CPP. This is logical because the process still must be carefully controlled within the defined, albeit wide, range. Similarly, it is not appropriate to set an unrealistically narrow range (i.e., a selected operating range that is not readily and reproducibly achievable by the equipment) for an operating parameter to claim that the process is not close to the proven range for a potential CPP and therefore not critical. In this case, the likelihood of repeated process deviations increases. This approach does not represent proper process control and is a questionable GMP practice.

Establishing a proven range does not automatically imply that excursion beyond the proven range leads to a failing API, because the proven range need not be the edge of failure. Process development should ideally provide an adequate proven range to minimize the likelihood of operating outside the proven range. This approach greatly reduces the chance of adversely affecting the quality of the API and, consequently, reduces the number of critical parameters. Thus, it is conceivable, and certainly desirable, that a robust process could have very few CPPs relevant to the CQAs of the API.

It also is possible that only one end of an operating range is a potential CPP. For example, a critical impurity may be formed as a result of temperature being too high during the reaction. In this case, the upper limit is a potential CPP. Lower temperatures may only lead to longer reaction times and thus decrease productivity. The lower limit would not be considered a potential CPP. If the upper operating limit is close to the proven limit, then the upper limit is critical. If the process has a high process capability (i.e., the operating limit is not close to the proven limit) and equipment capability is sufficient, then the process parameter is not critical.

These concepts can be illustrated in two scenarios in which the upper temperature is a potential CPP because it contributes to the formation of a critical impurity (i.e., a CQA). In the first scenario, a process temperature range of 30–40 °C is specified, and the proven range for the parameter has an upper limit of 43 °C. Operating the process in the range immediately beyond the operating range of 40 °C will result in failure (e.g., exceeding the proven limit of 43 °C). The 40 °C upper end of the operating range is close to the proven limit for the process, and it is important that the temperature be held below 40 °C. Therefore, based on the defined operating range for the temperature for this reaction, temperature is a critical parameter.

In a second scenario, the temperature range specified for this process is 30–40 °C, and the proven range for the parameter has an upper limit of 60 °C. In this case, the temperature is not a critical parameter because there is no need to constrain the temperature between 30–40 °C to ensure API quality. The temperature could rise as high as 60 °C before adversely affecting API quality. In other words, 40 °C is not close to the proven limit or edge of failure. Note that the identification of a critical parameter is determined by a combination of the relationship between the defined operating range and the proven range along with its effect on the defined CQA.

Completing the risk assessment. The previous discussion leads to a gradient of risk associated with the effect on the CQAs of the API. Figure 4 provides a crude overview of the risk assessment. For this figure, the following are defined:

• trigger event: Is it a potential CPP because of the effect on the API?

• severity: For "potential" CPP, severity is HIGH; for a "not potential" CPP, severity is LOW.

• probability: If the operating range is far from the proven range, then the probability is LOW. If the process is operating close to the proven range probability, then the probability is MEDIUM. If the process is operating close to the proven range and at the edge of failure, then the probability is HIGH.

• overall risk: the combination of severity and probability.

• risk assessment: the impact of the process on the API.

Figure 4: Risk assessment of process parameters.

It is important that the quality system appropriately manage the risk and designation of "critical" during the life cycle of the product. Later data may show the proven range is wider or narrower than first expected. As such, the designation of critical may change for potential CPPs. Future process optimizations or changes to equipment may bring the operating range closer to or farther away from the proven range, thus changing the designation of critical. The operating range may be moved as a corrective action to repeated process deviations. Again, this may affect the designation of critical. New potential CPPs can be identified as a result of changes in the impurity profile, thus increasing the severity of once noncritical process parameters.

Appropriate regulatory and quality decisions can be made through an agreement on the definition of the terms and with a fundamental understanding that the CQAs of an API drive the designation of CPPs. By justifying action or inaction in terms of the likely affect on API quality, as indicated by the CQAs, a science- and risk-based approach is maintained.

The final selection of the operating range depends on many factors in addition to API quality. The proven range represents the area of known acceptable API quality. The final operating range may be anywhere within the proven range to optimize the process, ensure operator and equipment safety, reduce environmental impact, improve yield, or reduce cycle time. The operating range must take into account the type of equipment and control system used.

Although operating parameters are given in ranges, the parameters themselves may be controlled to a set point within the range. Thus, the target temperature in the batch record might be set at 35 °C, and during the process the temperature may fluctuate between 32 and 38 °C. The purpose of the specified operating range is not to provide a range of target set points but to ensure that excursions around the set point are identified, evaluated, and controlled. The selection of or any adjustment to a particular set point within the range must take into account normal fluctuations in the operation because of equipment capability or other factors so that proper control is maintained within the operating range. An appropriately defined operating range will encompass the expected level of normal variation.

Linking critical process parameters with critical synthesis steps.

The previous sections have described how an evaluation of the synthesis steps leads initially to a pool of potential CPPs and then to specific CPPs associated with the CQAs of the final API. Consistent with this analysis, a critical synthesis step is defined as a synthesis step that contains a CPP.

The role of impurity formation and removal in CPPs and critical synthesis steps. Experience has shown that most CQAs are influenced significantly by the processing that occurs after the last true solution step of the synthesis (i.e., the final crystallization and API handling). Examples include particle size, polymorph, microbial purity, and residual solvent. Late-stage intermediates and the crude API, if isolated, are primarily responsible for the potential impurity profile of the final API, in the form of organic impurities, inorganic impurities, or optical impurities. Nevertheless, because the formation and removal of critical impurities are so important to defining CPPs throughout the synthesis, a brief discussion about them in impurity formation and removal is provided.

In some cases, impurity formation is totally independent of the process. For example, related compounds may enter with the raw materials and react to form impurities in spite of varying process conditions. In these circumstances, control of the raw-material quality may be critical, but no potential CPPs will be identified in the synthesis step unless one of these steps fortuitously reduces the impurity level. The rationale for these decisions should be documented. If the process does not affect the ultimate content of a particular impurity in the final API, however, further evaluation of the process for CPPs affecting this impurity is not needed.

In most cases, at least one process parameter (e.g., temperature, time, or mole ratio) will affect the formation or removal of impurities. As a simple example, it may be known that the amount of a dibrominated impurity is a function of stirring rate during the bromine addition as well as the rate of bromine addition. Dibromination may be favored by "locally concentrated" bromine. The temperature of the reaction as well as the final stirring time may make no contribution to the impurity formation. In this case, stirring rate and bromine addition rate are potential CPPs. The rate of stirring is in question, not whether the reaction is unstirred. One of the basic assumptions noted earlier in this article is that all operations will be performed as written. Therefore, considering that the reaction will be stirred, does it make a difference whether the stirring rate is slow, medium, or fast? In this example, stirring rate does have an influence and is a potential CPP.

Table II shows the potential influence of each chemical synthesis step of a hypothetical process on the process-dependent CQAs of an API. In this example, the final crystallization between crude and final API has the greatest influence on API quality. This step affects all of the CQAs and is the sole influence on five of the CQAs. For this example synthesis, two of the CQAs—organic purity and impurity profile—are potentially influenced by steps in the synthesis before the final intermediate step. In addition, two CQAs—residual solvent and particle size—are influenced by operations (e.g., milling) subsequent to the final API isolation step. The analysis was conducted using origin and fate of impurity information developed as discussed.

Table II: Does the synthesis step influence the critical quality attributes (CQAs) of an active pharmaceutical ingredient (API)?

To illustrate the overall selection process for CPPs described earlier, a more detailed evaluation of the hypothetical synthesis is provided. Suppose intermediate step D1–D2 shows two chemical conversions in the same synthesis step. D1 is not isolated but converted in situ into D2. During this process, a critical impurity is formed. Parameters that influence the formation of this critical impurity are potential CPPs. The chemical transformation in step E does not produce a critical impurity. Furthermore, intermediate E is isolated and used without purification. Therefore, there is no influence on either the formation or the removal of critical impurities. As such, step E has no influence on the CQAs of the API and therefore contains no CPPs and is a noncritical synthesis step. The conversion to the final intermediate involves a chemical transformation and optical resolution. Critical organic impurities are reduced during highly selective crystallization. This crystallization also is enantioselective. Accordingly, this is a critical synthesis step with potential CPPs that drive the efficiency of the recrystallization and removal of impurities. Conversion to the crude API involves a catalytic hydrogenation in which another critical impurity is formed. A heavy metal catalyst is used for the catalytic hydrogenation. The potential CPPs at this step are those that affect the formation and removal of the critical impurity and the proper removal of the catalyst.

The outcome of this analysis is the identification of the critical process steps with the greatest potential to affect the API. Moreover, the potential risk is linked to specific CQAs. The link to the final API and the scientific rationale for establishing this link form the foundation of future risk-based decisions. Table II makes it easier to understand the decision-making process and to adjust the evaluation if changes are made. In the same way, the relative risks of using various potential starting material sources or different synthesis routes can be assessed and compared.

Additional risk mitigation: specifications and process controls.

At this point, we have identified the critical process parameters within the process and which critical impurities and critical quality attributes in the API these parameters affect. Thus, a process operated within the defined operating range (and proven acceptable range) should provide an acceptable API. Nonetheless, for several reasons, it may be prudent to establish control points in the synthesis where confirmation of successful processing can be carried out. From a quality perspective, these control points would be strategically located to verify that CQAs are in control throughout the synthesis process.

These control measures can be placed to mitigate and reduce product risk. This phase of the risk-management process may combine:

• formal specifications and testing for starting materials, intermediates, and final product;

• in-process testing;

• processing limits (operating ranges);

• process monitoring and controls to include process analytical technologies where appropriate;

• quality system controls such as approved vendors, auditing, deviation handling, process validation, and change control.

The level of control should be commensurate with the level of risk to the defined CQAs of the API and may include combinations of controls as appropriate.

Specifications. Formal specifications and testing are implemented in key places in the synthesis to control risk associated with the CQAs of the API. As with process parameters, these controls should be linked to a particular CQA.

The starting material and the final API can act as bookends to the quality of the synthesis. The final API may be considered a critical control point because it is the last place where fitness for use can be demonstrated before release for secondary manufacture. All of the CQAs of the API should be confirmed on the final API. Likewise, the quality of the starting material may be considered critical because it confirms "fit for use" for the beginning of the synthesis. Risks to the process that are present before the starting material may be controlled at the starting material step. Controls on the starting material quality must address any relevant CQAs required to ensure API quality.

In-process testing and process controls. The following are some considerations for instituting control points for intermediates or in-process controls:

• highly selective purification steps, those essential to reduce the level of impurities;

• steps that introduce an impurity that is difficult to remove;

• steps that are the sole contributors toward ensuring the conformance of CQAs;

• in-process testing that provides critical process control (e.g., reaction end-point or drying).

An example of a CIPC (critical in-process control) follows. Suppose a reaction must proceed until a starting material is consumed to a specified low level to ensure that the residual starting material does not exceed a certain level in the intermediate, and ultimately in the API. In this case, the level of the starting material in the intermediate would be a critical impurity and affect organic purity (a CQA). This means the CIPC test confirms that the required low level of starting material has been reached. Typically, the reaction conditions (e.g., temperature and time) would be set such that this condition is routinely reached at the time of in-process testing. The in-process control confirms that the condition has been met. Should the reaction not have proceeded sufficiently, then the reaction typically would be continued until more of the starting material has been consumed. In this case, the CQA for the intermediate is linked to the CIPC as proof of the completion of the process step.

Regulatory uses of critical process data

Registration. As mentioned previously, the CTD (common technical document) format provides a placeholder for information to be included in Section 3.2.S.2.4, "Control of Critical Steps and Intermediates." Included in this information could be those CIPCs or process parameters defined relative to the CQAs for the API process. Thus, some of the CIPCs or CPPs developed during process characterization could be included in the original registration dossier to form the basis for product approval and regulatory change management postapproval. The remaining body of technical knowledge (i.e., the bulk of the technical data and process history and experience) would serve as background information available from the sponsor should the need arise during regulatory inspections or for any GMP-quality activities encompassed by sponsor quality systems (outlined briefly later in this article).

Validation. According to ICH Q7A, critical steps (i.e., synthesis steps with potential influence on the CQAs of the API) should be validated. For intermediates, the steps can be those in which critical impurities are formed or removed. Synthesis steps with no influence on the API do not require validation, although validation can still be performed. In the example shown in Table II, five of the nine steps would be validated. Other situations could lead to other validation decisions. In any case, meeting the CPPs should be part of the validation exercise.

Revalidation of established synthetic processes is event-driven and may be needed in the following circumstances:

• the addition of a new critical synthesis step;

• the addition of a new potential CPP or CPP;

• changes to the operating range of a potential CPP;

• the addition of or changes to a specification on a CQA;

• the introduction of a different starting material source.

It is important to stress that a definitive list of CPPs is needed before validation. This list is based on a full review of all "prevalidation" process knowledge and experience. Because validation of any or all steps may not be completed at the time that a registration dossier is being prepared, the CTD dossier may not necessarily contain all the CPPs that may ultimately be defined during validation. The registration dossier could include a discussion of intermediate CQAs and their link to API quality and the identification of the CPPs known at the time. The optimal time for a sponsor to finalize a selection of CPPs and the specific ranges that will serve as acceptance criteria is at the time of validation.

Should the CPPs listed in the CTD dossier require modification—either as a result of the validation exercise or through knowledge gained during product life cycle management—it is expected that appropriate updates to the regulatory dossier would be made. The routes for such updates need confirmation (discussed later in this article).

Integration with quality systems. The last part of the CQA core strategy is integrating science- and risk-based decision-making into the firm's quality systems to ensure overall compliance and fitness for use. This process knowledge is the central factor in taking science- and risk-based strategies and applying them to the life cycle of the product. These quality systems provide the daily checks and balances to ensure product quality on a lot-by-lot basis and systems to ensure proper change control and management review. These are not new topics, thus they will not be elaborated on in this article. Figure 5 shows the strategic link of the CQA risk-assessment process with the overall quality system. The two-phase approach of identifying potential CQA and CPPs provides the framework for maintaining the synthesis in an on-going state of compliance. This approach integrates well into the system of validation and assessment for the need of revalidation, change control, deviation handling, and annual product review.

Figure 5: Risk assessment linkage to quality systems.

The evolving regulatory landscape. Numerous FDA and ICH guidances or draft guidance initiatives were discussed previously in this article, providing evidence of the current environment of change surrounding regulatory expectations. Both industry and health authorities benefit from having a consistent set of expectations surrounding the definition and reporting of CQAs, CPPs, and the associated concepts in registration dossiers (i.e., NDA, CTD). In the context of the science- and risk-based approach currently under development, it makes great sense to focus on those parameters having the greatest potential influence on patient safety and efficacy and to provide summaries and conclusions rather than extensive data tabulations. This approach is consistent with concerns expressed by industry during a review of draft FDA guidances on chemistry, manufacturing, and controls information for drug substances and drug products, in which questions were raised about the value of proposed increased levels of detail for typical small-molecule APIs and drug products (e.g., equipment listings, environmental controls, and detailed synthesis conditions) (9).

After items have been listed in Section 3.2.S.2.4, "Controls of Critical Steps and Intermediates" of an original CTD, the regulatory paths to be used postapproval to implement changes to these parameters must be clarified. As noted, FDA has provided some guidance in the current Bulk Active Postapproval Changes (BACPAC I) guidance for postapproval changes for intermediates (10) and in Guidance Changes to an Approved NDA or ANDA, Revision 1, of June 2004 (11). In the United States, sponsors can make changes by means of annual report or changes-being-effected supplements in which the change is expected to have only moderate influence on the safety and efficacy of the drug product.

These documents, however, do not specifically address changes to the new information requested by the new draft drug substance and drug product FDA guidances. It is in the interest of both industry and health authorities that the regulatory paths be simple, allow for continuous improvement without unnecessary delay, and be consistent with existing improvements such as BACPAC I as well as the new risk-based approaches being initiated. One might briefly state the expectation that a greater body of knowledge about the process and its CPPs could justify a simpler regulatory route for changes to the process.

In addition to explaining a method of developing sound understanding of the synthesis process, the CQA–CPP approach outlined in this article also discusses several benefits in the context of postapproval changes. First, the focus on critical aspects of the process that control quality helps ensure that the appropriate amount of information is provided in the CTD and that excess information not relevant to API quality is not included. This ensures appropriate notification to regulators of postapproval changes to items that affect API quality. Second, the overall analysis is entirely consistent with existing guidance such as BACPAC I, which calls for comparisons at the final intermediate or earlier in the synthesis. Because we are interested in ensuring consistent API quality, the CQAs of these intermediates should be compared because these are the attributes that govern API quality. As discussed previously, for intermediates, this activity is likely to revolve around a comparison of critical impurity levels. Furthermore, this approach also is consistent with the BACPAC II proposals that were recently submitted to FDA by the Product Quality Research Institute and PhRMA's API Technical Committee, allowing for comparison of API quality (i.e., critical impurity levels) before the "last true solution."

Summary

This article discusses a science- and risk-based approach to determining which process parameters and controls are critical to quality. The primary points are:

• CQAs of APIs must be defined by the requirements of the specific drug product (e.g., dosage form, strength, formulation constraints).

• Some CQAs are affected by the API synthetic process, and others are controlled by overall quality systems.

• The steps before the last true solution typically affect the impurity profile of an API rather than the physical properties. Therefore, the process parameters for these steps are generally chosen to mitigate the effect of critical impurities on the final API.

• CPPs must meet two criteria in a risk-assessment evaluation: an established link between the process parameter and its effect on the CQAs of an API and a risk-assessment evaluation conclusion that the operating limit of this process parameter being close to the proven limit increases the probability of going outside the proven range for obtaining an acceptable API.

• Strategic controls at key points in the synthesis confirm the quality of CQAs. These form the basis of filed controls for the intermediates.

• Controls exist for reasons in addition to quality. Controls unrelated to the CQAs do not need to be in the regulatory submission.

• The analysis and application of CPPs presented in this article are entirely consistent with BACPAC I, the PhRMA and PQRI proposals for BACPAC II, and FDA's Guidance Changes to an Approved NDA or ANDA.

• Focus on the CQAs will permit registration files to address the quality-critical parameters, steps, and controls consistent with ongoing risk- and science-based initiatives, thereby providing an opportunity to reduce future FDA submissions by clarifying the value of parameters initially registered and avoiding those for which future regulatory consequences are vague.

CPPs typically are uncovered during process development. Where possible, the process is optimized to eliminate these critical parameters. This is a key goal during process development and an example of quality by design for drug-substance processes. Thus, it is not unusual to see relatively few critical parameters in a well-designed process. This does not imply that unit operations in the process are not important; all steps in a process are important and must be conducted, but not all are critical in the context of CQAs.

This article has clarified several of the concepts and interdependencies related to CQAs, CPPs, and CIPCs. This discussion should provide a basis for a common understanding in the industry and among health authorities. The authors hope this common understanding can help improve product safety and efficacy while supporting technically sound and historically valid regulatory practices. The authors also hope that by including additional discussion that demonstrates the sponsor's understanding of the CPPs that affect the CQAs of the API in the original CTD submission, there will be some opportunity for streamlining and simplifying the regulatory requirements for postapproval changes to processes.

William P. Ganzer* is the divisional vice-president, Global Pharmaceutical Operations, Quality Assurance Services, Abbott Laboratories, 200 Abbott Park Road, D-0485, Bldg. J-23, Abbott Park, IL 60064, tel. 847.937.5716, fax 847.938.7453, bill.ganzer@ abbott.comJoan A. Materna is a senior associate director, Global Regulatory CMC, Novartis Pharmaceutical Corp. (East Hanover, NJ). Michael B. Mitchell is the vice-president, Pharmaceutical Development, Schering-Plough Research Institute (Union, NJ). L. Kevin Wall is director, Strategic Improvement, Quality Janssen Pharmaceutica (Belgium).

*To whom all correspondence should be addressed.

References

1. T. Cupps et al., "PhRMA Perspectives on Drug Substance Regulatory Filing Issues: Starting Material, Reprocessing, Retesting, and Critical Controls," Pharm. Technol. 27 (2), 27-34 (2003).

2. FDA, Guideline for Submitting Supporting Documentation in Drug Applications for the Manufacture of Drug Substances (FDA, Rockville, MD, Feb. 1987), www.fda.gov/cder/guidance/drugsub.pdf.

3. International Conference on Harmonization, Guideline Quality M4, The Common Technical Document for Registration of Pharmaceuticals for Human Use (Sep. 2002) and Guideline M4Q, The CTD — Quality (ICH, Geneva, Switzerland, Aug. 2001), www.ich.org

4. FDA, Draft Guidance for Industry Drug Substance; Chemistry, Manufacturing and Controls Information (Jan. 2004), www.fda.gov/cber/gdlns/drugsubcmc.pdf and Draft Guidance on Drug Product CMC Information (FDA, Rockville, MD, Jan. 2003), www.fda.gov/cber/gdlns/drugcmc.pdf.

5. International Conference on Harmonization, Guideline Q7A: Good Manufacturing Practices for Active Pharmaceutical Ingredients (ICH, Geneva, Switzerland, Nov. 2000), www.ich.org.

6. FDA, Pharmaceutical CGMPs for the 21st Century: A Risk-Based Approach (FDA, Rockville, MD, Aug. 2002), www.fda.gov/oc/guidance/gmp.html.

7. FDA, Guidance for Industry: PAT—A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance (FDA, Rockville, MD, Sept. 2004), p. 6, www.fda.gov/cder/guidance/6419fnl.pdf.

8. International Conference on Harmonization, Guideline Q6A: Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances (ICH, Geneva, Switzerland, Oct. 1999), www.ich.org.

9. PhRMA comments to FR Docket 2003D-0571, FDA Draft Guidance for Industry on Drug Substance; Chemistry, Manufacturing and Controls Information, Jan. 2004, letter dated July 8, 2004, www.fda.gov/ohrms/dockets/dailys/04/july04/071204/03d-0571-c00025-vol2.pdf.

10. FDA, Guidance for Industry: BACPAC I—Intermediates in Drug Substance Synthesis; Bulk Actives Post approval Changes: Chemistry, Manufacturing, and Controls Documentation, Feb. 2001, www.fda.gov/cder/guidance/3629fnl.htm.

11. FDA, Guidance Changes to Approved NDA or ANDA, June 2004 (Revision 1), www.fda.gov/cder/guidance/3516fnl.pdf.