GMPs for Method Validation in Early Development: An Industry Perspective (Part II)

The authors, part of the International Consortium on Innovation and Quality in Pharmaceutical Development (IQ Consortium), explore and define common industry approaches and practices when applying GMPs in early development. A working group of the consortium aims to develop a set of recommendations that can help the industry identify opportunities to improve lead time to first-in-human studies and reduce development costs while maintaining required quality standards and ensuring patient safety. This article is the second in the paper series and focuses on method validation in early-stage development.

The International Consortium on Innovation and Quality in Pharmaceutical Development (IQ) was formed in 2010 as an association of over 25 pharmaceutical and biotechnology companies with a mission to advance science-based and scientifically-driven standards and regulations for medicinal products worldwide. In the June 2012 issue of Pharmaceutical Technology, a paper was presented which described an overview of IQs consolidated recommendations from the Good Manufacturing Practices (GMPs) in Early Development working group (WG) (1). The focus of this IQ WG has been to develop recommended approaches on how to apply GMPs in early phase CMC development activities covering Phase I through Phase IIa. A key premise of the GMPs in Early Development WG is that existing GMP guidances for early development are vague and that improved clarity in the definition of GMP expectations would advance innovation in small-molecule pharmaceutical development by improving cycle times and reducing costs, while maintaining appropriate product quality and ensuring patient safety.

A consequence of the absence of clarity surrounding early phase GMP expectations has been varied in interpretation and application of existing GMP guidances across the industry depending on an individual company's own culture and risk tolerance. Internal debates within a company have frequently resulted in inappropriate application of conservative "one-size-fits-all" interpretations that rely on guidelines from the International Conference on Harmonization (ICH) that are more appropriate for pharmaceutical products approaching the point of marketing authorization application. In many cases, erroneous application of these commercial ICH GMP expectations during early clinical development does not distinguish the distinct differences in requirements between early development and late-stage development (Phase IIb and beyond). A key objective of this IQ WG, therefore, has been to collectively define in early development-within acceptable industry practices-some GMP expectations that allow for appropriate flexibility and that are consistent with existing regulatory guidances and statutes (2).

As outlined in the previous introductory paper, the efforts of the GMPs in Early Development WG have focused on the following four areas of CMC activities: analytical method validation, specifications, drug-product manufacturing, and stability. The initial scope of these efforts has been limited to small-molecule drug development which supports First in Human (FIH) through Phase IIa (Proof-of-Concept) clinical studies. A series of papers describing a recommended approach to applying GMPs in each of these areas is being published within this journal in the coming months. In this month's edition, the authors advocate for a life-cycle approach to method validation, which is iterative in nature in order to align with the evolution of the manufacturing process and expanding product knowledge space.

A pharmaceutical industry collective perspective on analytical method validation with regard to phase of development has not been published since 2004 (3). Genesis of the 2004 paper occurred during a set of workshops sponsored by the Analytical Technical Group of the Pharmaceutical Research and Manufacturers of America (PhRMA) in September 2003. The referenced paper summarized recommendations for a phased approach to method validation for small-molecule drug substance and drug products in early clinical development. Although a few other reviews on method validation practices have been published (4), this paper provides a current, broad-based industry perspective on appropriate method validation approaches during the early phases of drug-product development.

This broad industry assessment of method validation also uncovered the need to clearly differentiate the context of the terms of "validation" and "qualification." Method qualification is based on the type, intended purpose, and scientific understanding of the type of method in use during the early development experience. Although not used for GMP release of clinical materials, qualified methods are reliable experimental methods that may be used for characterization work, such as reference standards and the scientific prediction of shelf-life.

A perspective on some recent analytical method challenges and strategies, such as genotoxic impurity methods, use of generic methods, and methods used for testing toxicology materials or stability samples to determine labeled storage conditions, retest periods and shelf life of APIs and drug products are also presented. The approach to method validation described herein is based on what were considered current best practices used by development organizations participating in the IQ consortium. In addition, this approach contains some aspects which represent new scientifically sound and appropriate approaches that could enable development scientists to be more efficient without compromising product quality or patient safety. These science-driven acceptable best practices are presented to provide guidance and a benchmark for collaborative teams of analytical scientists, regulatory colleagues, and compliance experts who are developing standards of practice to be used during early phases of pharmaceutical development. The views expressed in this article are based on the cumulative industry experience of the members of the IQ working group and do not reflect the official policy of their respective companies.

Early-phase method parameters requiring validation

CLICK TABLE TO ENLARGE Table I: Summary of proposed approach to method validation for early- and late-stage development.

In early development, one of the major purposes of analytical methods is to determine the potency of APIs and drug products to ensure that the correct dose is delivered in the clinic. Methods should also be stability indicating, able to identify impurities and degradants, and allow characterization of key attributes, such as drug release, content uniformity, and form-related properties. These methods are needed to ensure that batches have a consistent safety profile and to build knowledge of key process parameters in order to control and ensure consistent manufacturing and bioavailability in the clinic. In the later stages of drug development when processes are locked and need to be transferred to worldwide manufacturing facilities, methods need to be cost-effective, operationally viable, and suitably robust such that the methods will perform consistently irrespective of where they are executed. In considering the purpose of methods in early versus late development, the authors advocate that the same amount of rigorous and extensive method-validation experiments, as described in ICH Q2 Analytical Validation is not needed for methods used to support early-stage drug development (5). This approach is consistent with ICH Q7 Good Manufacturing Practice, which advocates the use of scientifically sound (rather than validated) laboratory controls for API in clinical trials (6). Additionally, an FDA draft guidance on analytical procedures and method validation advocates that the amount of information on analytical procedures and methods validation necessary will vary with the phase of the investigation (7).

IQ's perspective regarding which method parameters should be validated for both early- and late-stage methods is summarized in Table I. In this table, identification methods are considered to be those that discriminate the analyte of interest from compounds with similar (or dissimilar) structures or from a mixture of other compounds to assure identity. This category includes, but is not limited to identification methods using high-performance liquid chromatography (HPLC), Fourier transform infrared spectroscopy (FTIR), and Raman Spectroscopy. Assay methods are used to quantitate the major component of interest. This category includes, but is not limited to drug assay, content uniformity, counter-ion assay, preservative's assay, and dissolution measurements. Impurity methods are used for the determination of impurities and degradants and include methods for organic impurities, inorganic impurities, degradation products, and total volatiles. To further differentiate this category of methods, separate recommendations are provided for quantitative and limit test methods, which measure impurities. The category of "physical tests" in Table I can include particle size, droplet distribution, spray pattern, optical rotation, and methodologies, such as X-Ray Diffraction and Raman Spectroscopy. Although representative recommendations of potential parameters to consider for validation are provided for these physical tests, the specific parameters to be evaluated are likely to differ for each test type.

When comparing the method-validation approach outlined for early development versus the method-validation studies conducted to support NDA filings and control of commercial products, parameters involving inter-laboratory studies (i.e., intermediate precision, reproducibility, and robustness) are not typically performed during early-phase development. Inter-laboratory studies can be replaced by appropriate method-transfer assessments and verified by system suitability requirements that ensure that the method performs as intended across laboratories. Because of changes in synthetic routes and formulations, the impurities and degradation products formed may change during development. Accordingly, related substances are often determined using area percentage by assuming that the relative response factors are similar to that of the API. If the same assumption is used to conduct the analyses and in toxicological impurity evaluation and qualification, any subsequent impurity level corrections using relative response factors are self-corrective and hence mitigate the risk that subjects would be exposed to unqualified impurities. As a result, extensive studies to demonstrate mass balance are typically not conducted during early development.

In addition to a smaller number of parameters being evaluated in preclinical and early development, it is also typical to reduce the extent of evaluation of each parameter and to use broader acceptance criteria to demonstrate the suitability of a method. Within early development, the approach to validation or qualification also differs by what is being tested, with more stringent expectations for methods supporting release and clinical stability specifications, than for methods aimed at gaining knowledge of processes (i.e., in-process testing, and so forth). An assessment of the requirements for release- and clinical-stability methods follows. Definitions of each parameter are provided in the ICH guidelines and will not be repeated herein (5). The assessment advocated allows for an appropriate reduced testing regimen. Although IQ advocates for conducting validation of release and stability methods as presented herein, the details are presented as a general approach, with the understanding that the number of replicates and acceptance criteria may differ on a case-by-case basis. As such, the following approach is not intended to offer complete guidance.

Specificity. Specificity typically provides the largest challenge in early-phase methods because each component to be measured must be measured as a single chemical entity. This challenge is also true for later methods, but is amplified during early-phase methods for assay and impurities in that:

• The chemical knowledge regarding related substances is limited.

• There are frequently a greater number of related substances than in commercial synthetic routes.

• The related substances that need to be quantified may differ significantly from lot-to-lot as syntheses change and new formulations are introduced.

A common approach to demonstrating specificity for assay and impurity analysis is based on performing forced decomposition and excipient compatibility experiments to generate potential degradation products, and to develop a method that separates the potential degradation products, process impurities , drug product excipients (where applicable), and the API. Notably, requirements are less stringent for methods where impurities are not quantified such as assay or dissolution methods. In these cases, specificity is required only for the API.

Accuracy. For methods used in early development, accuracy is usually assessed but typically with fewer replicates than would be conducted for a method intended to support late-stage clinical studies. To determine the API in drug product, placebo-spiking experiments can be performed in triplicate at 100% of the nominal concentration and the recoveries determined. Average recoveries of 95–105% are acceptable for drug product methods (with 90–110% label claim specifications). Tighter validation acceptance criteria are required for drug products with tighter specifications. For impurities, accuracy can be assessed using the API as a surrogate, assuming that the surrogate is indicative of the behavior of all impurities, including the same response factor. Accuracy can be performed at the specification limit (or reporting threshold) by spiking in triplicate. Recoveries of 80-120% are generally considered acceptable, but will depend on the concentration level of the impurity. For tests where the measurements are made at different concentrations (versus at a nominal concentration), such as dissolution testing, it may be necessary to evaluate accuracy at more than one level.

Precision. For early-phase methods, only injection and analysis repeatability is examined. Area % relative standard deviation (RSD) is typically determined from 5 replicates. Repeatability is determined at 100% of nominal concentration for the API with impurities being evaluated at the reporting threshold using the API as a surrogate. Acceptance criteria of 1% RSD (injection repeatability) or 2% RSD (analysis repeatability) for API are frequently targeted. For impurities, higher precision limits (e.g., 10–20%) are acceptable and should consider the level of the impurity being measured (injection and analysis repeatability). For tests where the measurements are made at different concentrations (versus at a nominal concentration), such as dissolution testing, it may be necessary to evaluate repeatability at more than one level.

Limit of detection and limit of quantitation. A sensitivity assessment is necessary to determine the level at which impurities can be observed. Using the API as a surrogate, a "practical" assessment can be made by demonstrating that the signal of a sample prepared at the reporting threshold produces a signal-to-noise ratio of greater than 10. A limit of quantitation can be determined from this assessment by calculating the concentration that would be required to produce a signal to noise ratio of 10:1. Similarly, a limit of detection can be calculated as the concentration that would produce a signal-to-noise ratio of 3:1. However, it is emphasized that the "practical limit of quantitation" at which it is verified that the lowest level of interest (reporting threshold) provides a signal at least 10 times noise and thus can be quantitated, is of paramount importance.

Linearity. Linearity can be determined from 3-point calibration curves at test concentrations of 70, 100, and 130% of nominal (API) for assay or from 3 points ranging from the reporting threshold to 130% of the specification limit for impurities. API is used as the surrogate analyte for impurities. For both analyses, a validation criterion can be established as R 2 > 0.995. For tests where the measurements are needed over broader concentration ranges, such as dissolution testing, a broader linear range may be examined using a 3-point calibration.

Range. As for late-phase methods, the range is inferred from the accuracy, precision, and linearity studies.

Robustness. Full robustness testing is not conducted during early development. However, an assessment of solution stability should be conducted to demonstrate the viable lifetime of standards and samples. Specifically, solutions should be considered stable when the following conditions are met:

• The API assay changes by not more than 2%

• No new impurities greater than the reporting threshold are observed

• Impurities at the reporting threshold change by not more than 30%; impurities at levels between the reporting threshold and the specification limit change by not more than 20%; and impurities at or above the specification limit change by not more than 15%.

Notably, if validation is performed concurrently with sample analysis as an extended system suitability, solution stability must be assessed separately. This assessment is typically conducted as part of method development.

Early-phase methods requiring validation

During discussions held to develop this approach to early-phase method validation, it was evident that the context of the terms "validation" and "qualification" was not universally used within all the IQ member companies. To facilitate a common understanding of this approach, the authors will therefore refer to "validated methods" as those methods which perform as expected when subjected to the series of analytical tests described in this approach. "Qualified methods" are considered to be analytical methods which are subjected to less stringent testing to demonstrate that they are scientifically sound for their intended use. In the following sections, the authors recommend which types of methods typically employed in early development require either validation or qualification.

Methods for release testing and to support GMP manufacturing. In early development, specifications are used to control the quality of APIs and drug products. Consideration of specifications places great emphasis on patient safety since knowledge of the API or drug product process is limited due to the low number of batches produced at this stage of development. Specifications typically contain a number of different analytical tests that must be performed to ensure the quality of the API or drug product. Typical material attributes, such as appearance, potency, purity, identity, uniformity, residual solvents, water content, and organic/inorganic impurities, are tested against established acceptance criteria. The API and drug-product specific methods for potency, impurity, uniformity, and others should be validated as described above and demonstrated to be suitable for their intended use in early phase development prior to release. If compendial methods are used to test against a specification (e.g., FTIR for identification and Karl Fischer titration [KF] for water content), they should be evaluated and/or qualified to be suitable for testing the API or drug product prior to use without validation. Materials used in the manufacture of GMP drug substance and drug product used for early-phase clinical studies for which specifications are not outlined in a regulatory filing (e.g., penultimates, starting materials, isolated intermediates, reagents, and excipients) need only to be qualified for their intended use. Method transfer is less rigorous at this early stage of development and may be accomplished using covalidation experiments or simplified assessments.

As mentioned, method qualification is often differentiated from method validation. The experiments to demonstrate method qualification are based on intended purpose of the method, scientific understanding of the method gained during method development and method type. It is an important step in ensuring that reliable data can be generated reproducibly for investigational new drugs in early development stages. The qualified methods should not be used for API or drug product release against specifications and concurrent stability studies. However, reference material characterization may be done with qualified methods.

Generation of process knowledge in early development is rapidly evolving. Numerous samples are tested during early development to acquire knowledge of the product at various stages of the process. The results from these samples are for information only (FIO) and methods used for this type of testing are not required to be validated or qualified. However, to ensure the accuracy of the knowledge being generated, sound scientific judgment should be used to ensure the appropriateness of any analytical method used for FIO purposes.

"Generic" or "general" methods. A common analytical strategy often employed in early development is the use of fit-for-purpose generic or general methods for a specific test across multiple products (e.g., gas chromatography for residual solvents). These methods should be validated if they are used to test against an established specification. The suggested approach to validating these methods in early development is typically performed in two stages. Stage 1 involves validating the parameters that are common for every product with which the method can be used. Linearity of standard solutions and injection repeatability belong to this stage. Stage 2 of the validation involves identifying the parameters that are specific to individual product, such as accuracy. Specificity may be demonstrated at Stage 1 for nonproduct related attributes and at Stage 2 for product related attributes. Stage 1 validation occurs prior to GMP testing. Stage 2 validation can happen prior to or concurrent with GMP testing. This approach to validation of fit-for-purpose methods can provide efficiency for drug development by conserving resources in the early phases of development and can ensure reliability of the method's intended application.

Methods for GTI and tox batch qualification. The need for analytical methods to demonstrate the control of genotoxic impurities (GTI) has developed recently because of expectations and guidances provided by regulatory authorities (8, 9). Often, these methods require high sensitivity with limits of quantitation in the parts-per-million (ppm) range. Although the control levels for GTIs (referred to as the threshold of toxicological concern) is less stringent for early clinical studies (e.g., patient intake < 50 ug/day for clinical studies < 30 days vs 1.5 ug/day for longer clinical studies), regulatory authorities expect that GTI control is demonstrated during early development. Depending on when a GTI is potentially generated during an API synthesis, GTIs may be listed in specifications. Validation of these methods is again dependent upon the intended use of the method. Methods used for assessment may be qualified unless they are used to test against a specification as part of clinical release. Method qualification is also considered appropriate if the method is intended for characterization or release of test articles for a toxicology study.

Methods for stability of APIs and drug products. Batches of API and drug product are typically exposed to accelerated stress conditions and tested at timed intervals to assess whether any degradation has occurred. The shelf-life of the API or drug product-that is, the time period of storage at a specified condition within which the drug substance and drug product still meets its established specifications, is based on analytical data generated from these studies. For this application, analytical methods need to be stability-indicating (e.g., capable of detection and quantitation of the degradants) to ensure quality, safety, and efficacy of a drug substance and drug product. Often, the analytical methods used to perform stability tests are the same methods used to test against a specification for release testing; these methods should be validated. However, if additional tests are performed which are not included in the established specification, they may be qualified for their intended use, rather than validated.

In-process testing methods. In-process testing (IPT) during manufacturing of drug substance and drug product can be done on-line, in-line, or off-line. The results generated from IPT are used to monitor processes involving reaction completion, removal of solvents, removal of impurities, and blend content uniformity. Manufacturing parameters may be adjusted based on IPT results. IPT methods are often very limited in scope. In early development, the primary benefit of performing IPTs is the generation of process knowledge, and not as a control or specification. As a result, even though IPT is essential for manufacture of drug substance and drug product, method qualification for an IPT method is appropriate in early-phase development.

Documentation and other requirements. The extent of documentation and associated practices in early development should be aligned with the appropriate level of method validation as discussed above. In this paper, the authors provide a perspective on the appropriate level of documentation, protocol and acceptance-criteria generation, instrument qualification, and oversight of the quality assurance unit for early-phase method validation and qualification. This approach provides development scientists with flexibility to efficiently adapt to the dynamic environment typical within early phase pharmaceutical development, while ensuring patient safety and the scientific integrity of the validation process.

With respect to documentation, it the IQ perspective that the raw data which is generated during early phase method validation should be generated and maintained in a compliant data storage format. The integrity of raw data should be controlled such that it can be retrieved to address future technical and compliance-related questions. Proper documentation of data and validation experiments should also be considered an important aspect of early phase validation. The availability of electronic notebook (ELN) systems has provided a viable, more efficient alternative to the use of traditional bound-paper notebooks. In developing policies to implement ELNs, the goal should not be that all documentation practices used with paper notebooks are replicated. Rather, the ELN should possess sufficient controls for the intended use of the data. In many cases, electronic systems such as ELNs will transform the work process, and the controls it provides will be achieved in a completely novel manner compared to the outdated system being replaced.

Although data needs to be documented as described above, it is the authors' position that formal, detailed method and validation reports are not required to ensure compliance in early development. Adequate controls need to be in place to ensure method parameters used to execute validated methods are equivalent to parameters used during validation. Generation of brief method and validation summary reports are required only when needed to fulfill regulatory filing requirements or to address requests or questions from health authorities. Validation summaries are not required to present all of the validation data, but rather a summary of the pertinent studies sufficient to demonstrate that the method is validated to meet the requirements of its intended use. Once reports are generated and approved internally, approved change control procedures should be available and followed to maintain an appropriate state of control over method execution and report availability.

Although the authors' perspective is that a validation plan needs to exist for early phase method validation, analytical organizations could consider different mechanisms to fulfill this need. For example, internal guidelines or best practice documents may sufficiently outline validation requirements such that a separate validation plan need not be generated for each method. In the absence of such a guideline or procedure, a validation plan could be documented in a laboratory notebook or ELN which includes a brief description of validation elements and procedures to be evaluated. Validation plans should ensure that the method will be appropriate for its intended use. The use of strict validation criteria within the validation plan should be limited at these early stages of development. Validation studies for early development methods may be performed on fit-for-purpose instruments which are calibrated and maintained, but not necessarily qualified or under strict change-control standards.

The role of the pharmaceutical quality system and the oversight over early phase method validation practices and documentation is another area for consideration. In the pharmaceutical industry, quality management is overseen by a "Quality Unit" that qualifies and oversees activities in the areas of GMP materials such as laboratory controls. In practice, the size and complexity of the Quality Unit overseeing GMP manufacturing varies based on a manufacturer's size and stage of drug development. Regardless, the basic aspects of a quality system must be in place. In early development, IQ's position is that, because API and drug-product manufacturing processes are evolving, the analytical methods do not yet require full validation as prescribed in ICH Q2. Correspondingly, the quality system implemented during early phases could consider that evolving analytical methods are intrinsic to the work being performed to develop the final API and drug product processes and could allow flexibility to readily implement method changes during early development. For example the Quality Unit should delegate oversight for validation plan approval, change control, approval of deviations and reports to the analytical departments prior to finalization and performing full ICH Q2 validation of the analytical methods. This approach would be consistent with Chapter 19 of ICH Q7A. However, analytical departments must ensure that early phase validation studies are conducted by qualified personnel with supervisory oversight who follow approved departmental procedures. Clearly, agreements between Quality Units and analytical departments to implement an appropriate strategic, phase-based quality oversight system would provide many benefits within the industry.

Conclusions

Within this paper, IQ representatives have presented an industry perspective on appropriate requirements and considerations for early phase analytical method validation. A suggested outline of acceptable experiments that ensure analytical procedures developed to support API and drug product production of early phase clinical materials are suitable for their intended use has been presented. Additionally, the authors have provided a position on phased approaches to other aspects of method validation such as documentation requirements, generation of method validation plans, validation criteria, and the strategic involvement of quality unit oversight. When applied appropriately, this approach can help to ensure pharmaceutical development organizations provide appropriate analytical controls for API and drug product processes which will serve the ultimate goal of ensuring patient safety. Although the extent of early-phase method validation experiments is appropriately less than employed in the later stages of development, we view that any risks related to this approach will not be realized, especially when considering the overall quality and safety approach used by pharmaceutical companies for early phase clinical studies.

It is the authors' hope that providing such an approach to early-phase method validation, along with the approaches outlined in this series of early-phase GMP papers, will serve as a springboard to stimulate discussions on these approaches within the industry and with worldwide health authorities. To encourage further dialogue, this IQ working group is planning on conducting a workshop in the near future to promote robust debate and discussion on these recommended approaches to GMPs in early development. These discussions will ideally enable improved alignment between R&D development, Quality, and CMC regulatory organizations across the pharmaceutical industry, and most importantly with worldwide regulatory authorities. Agreement between industry and health authorities regarding acceptable practices to applying GMPs in the early phases of drug development would clearly be beneficial to CMC pharmaceutical development scientists and allow for a more nimble and flexible approach to better address the dynamic environment typical of the early phases of clinical development, while still guaranteeing appropriate controls to ensure patient safety during early development.

About the Authors

Donald Chambers is in analytical sciences at Merck Research Laboratories, Gary Guo is in analytical R&D at Amgen, Brent Kleintop* is in analytical and bioanalytical development at Bristol-Myers Squibb Co., Henrik Rasmussen is in analytical development at Vertex Pharmaceuticals, Steve Deegan is in GMP Quality Assurance Operations at Abbott, Steven Nowak is in NCE Analytical R&D, Global Pharmaceutical R&D, at Abbott, Kristin Patterson is in emerging markets R&D at GlaxoSmithKline, John Spicuzza is in analytical development at Baxter, Michael Szulc is in analytical development at Bioden Idec, Karla Tombaugh is in R&D/Commercialization Quality, Merck Manufacturing Division, at Merck & Co., Mark D. Trone is in analytical development, small molecule, at Millennium Pharmaceuticals, and Zhanna Yuabova is in analytical development, US, at Boehringer Ingelheim Pharmaceuticals.

*To whom all correspondence should be addressed.

References

1. A. Eylath at al., Pharm. Technol. 36 (5) 54–58 (2012).
2. FDA, Guidance for Industry: cGMP for Phase 1 Investigational Drugs (July 2008 FDA).
3. S.P. Boudreau et al., Pharm. Technol. 28 (11) 54–66 (2004).
4. M. Bloch, "Validation During Drug Product Development – Considerations as a Function of the Stage of Drug Development," Method Validation in Pharmaceutical Analysis, a Guide to Best Practice, Eds. J. Ermer, J.H. Miller (Wiley, 2005), pp. 243–264.
5. ICH, Q2 (R1) Validation of Analytical Procedures: Text and Methodology (Nov. 2005).
6. ICH, Q7A Good Manufacturing Practice Guidelines for Active Pharmaceutical Ingredients (Aug. 2001).
7. FDA, Draft Guidance for Industry: Analytical Procedures and Method Validation, Chemistry, Manufacturing and Controls Documentation (Aug. 2000).
8. EMA, Guideline on the Limits of Genotoxic Impurities (June 2006).
9. FDA, Draft Guidance for Industry: Genotoxic and Carcinogenic Impurities in Drug Substances and Products: Recommended Approaches (2008).