Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

Review

Ofcial USP Reference Standards: Metrology concepts, overview, and scientic issues and opportunities
Roger L. Williams ,1 Project Team 42 the 20002005 Reference Standards Committee of the USP Council of Experts and Its Advisory Panel3 USP Staff and Consultant4
United States Pharmacopeia, 12601 Twinbrook Parkway, Rockville, MD 20852-1790, USA Received 11 May 2005; received in revised form 26 June 2005; accepted 6 July 2005 Available online 21 November 2005

Abstract The United States Pharmacopeia (USP) is a private standards-setting body created in 1820 by practitioners who wished to promote the quality of therapeutic products in commerce. The principal product of USP, then and now, is the United States Pharmacopeia (USP), to which was added the National Formulary (NF) in 1975. The two compendia are published as a combined text annually (USPNF). Originally a book of process standards, USPNF evolved over time into compendia containing primarily product standards that are expressed in monographs for therapeutic ingredients, products, and excipients. As a public health service, USP supplies ofcial USP Reference Standards to manufacturers and others who wish to test an article according to selected procedures of a monograph or General Chapter. During the past decade, understanding of USP monographs and ofcial USP Reference Standards as a means of controlling the quality of a therapeutic article has evolved, based on advances in metrology, on activities in the International Conference on Harmonization of Technical Requirements for the Registration of Pharmaceuticals for Human Use (ICH), and on considerations by the USP Council of Experts and its Expert Committees and USP staff. This article discusses the evolution of this understanding, focusing on drug substances and excipients for well-characterized small molecules and their corresponding dosage forms. 2005 Elsevier B.V. All rights reserved.
Keywords: United States Pharmacopeia; Reference Standards; Specications; USPNF; Metrology; Monographs

Contents
1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nomenclature and legal implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Legal implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 4 4 5 5

3.

Tel.: +1 301 816 8300; fax: +1 301 816 8373. E-mail address: [email protected]. 1 Council of Experts Chair., 2 Lise Alaimo, Mark Bailey, Matthew Borer, Raymond Cox, David Fay, Gregory Kaster (Chair), Reenie Parris, and Jeffrey Singer. 3 Reference Standard Committee: Judy P. Boehlert, Raymond A. Cox, Thomas S. Foster, Samir A. Hanna, Krishan Kumar, Pauline M. Lacroix, Guirag K. Poochikian, Susan J. Schniepp, Eric Sheinin (Chair), Manfred E. Wolff, and Wesley E. Workman; Advisory Panel: David A. Fay, Ralph Gomez, Vivian A. Gray, Ruth Homan, Maria Ines, Gregory T. Kaster, Judy A. Lee, Michael Ribick, and R.M. Santoro. 4 Todd Cecil, Shawn Dressman, John Esker, L.Valentin Feyns , Walter Hauck (Consultant), Ronald Manning, Stefan Schuber, and Eric Sheinin. 0731-7085/$ see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jpba.2005.07.017

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

4.

Collection overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Sources of candidate material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Evaluation of candidate materials: ingredients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Evaluation of candidate materials: impurities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Assignment of content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Content assignment: ingredients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Content assignment: impurities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Value reassignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. Statistical topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.2. Theoretical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.3. Pilot study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A.4. Analysis of 12 representative data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 6 6 7 8 9 9 9 9 11 11 11 12 14 14

1. Introduction The United States Pharmacopeia (USP) is a private standards-setting body created in 1820 by practitioners who wished to promote the quality of therapeutic products in commerce [1]. The principal product of USP, then and now, is the United States Pharmacopeia, to which was added the National Formulary (NF) in 1975. Together the two compendia are published as a combined text annually (USPNF) with two Supplements. Originally a book of process standards (recipes for preparations), USPNF evolved over time into compendia containing primarily product standards [2]. These standards are expressed in monographs for drug substances, excipients, dosage forms and other articles, and in General Chapters, which are dedicated to procedures, general information, and general requirements widely used throughout the compendium. A monograph contains introductory statements, packaging, storage, and other labeling statements, and the articles public specication, which consists of tests, procedures, and acceptance criteria. As a public health service, USP supplies ofcial USP Reference Standards to manufacturers, regulatory agencies, and other interested parties who wish to test an article according to selected procedures of a monograph or of a General Chapter. During approximately the past 10 years, advances in metrology [3], activities in International Conference on Harmonization of Technical Requirements for the Registration of Pharmaceuticals for Human Use (ICH), and considerations by the USP Council of Experts and its Expert Committees with USP staff have advanced understanding of the role of USP monographs and ofcial USP Reference Standards as a means of controlling the quality of a therapeutic article [4]. This article discusses the evolution of this understanding, focusing on drug substances and excipients for well-characterized small molecules and their corresponding dosage forms. An Appendix A considers special statistical issues with regard to collaborative testing.

2. Nomenclature and legal implications 2.1. Nomenclature Modern metrology concepts have developed during the past decade and longer through research and standardssetting activities in government, academia, industry, the pharmacopeias, and elsewhere. Harmonizing nomenclature for these concepts has been presented in publications of International Standards Organization (ISO) guides and also in publications of the National Institute of Standards and Technology (NIST), the International Union of Pure and Applied Chemistry (IUPAC), EURACHEM, Co-Operation on International Traceability in Analytical Chemistry (CITAC), and International Laboratory Accreditation Cooperation (ILAC). Successful application of modern metrologic approaches, when used in ratio-method measurements in the pharmacopeia, are associated with a reference material (RM), dened as a material or substance one or more of whose property values are sufciently homogeneous and well established to be used for the calibration of an apparatus, the assessment of a measurement method, or for assigning values to materials [5]. A certied reference material (CRM) is dened as a reference material, accompanied by a certicate, one or more of whose property values are certied by a procedure which establishes its traceability to an accurate realization of the unit in which the property values are expressed, and for which each certied value is accompanied by an uncertainty at a stated level of condence [6]. A general objective of modern metrology allows the content (purity) of an RM to be expressed in the Syst` me International dUnit s (SI) e e units of mass (e.g., kilogram) and/or amount (e.g., mole) [7]. When used appropriately, RMs and CRMs allow value assignment for a measurand in SI units. Although USP offers RMs (ofcial USP Reference Standards) for calibration and assessment of a measurement method (procedural standard), the bulk of the USP collection is composed of neat materials used in quality control and similar laboratories to assign a valueexpressed in terms of massto measurands. These

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

are pharmaceutical ingredients (drug substances, excipients, and other ingredients, e.g., antimicrobial agents, preservatives, and anti-oxidants) as well as dosage forms. The use of ofcial USP reference standards has risen over the past several decades as a result of the increasing use of primary ratio methods in instrumental techniques. 2.2. Legal implications A complex relationship exists between USPs ofcial publications, USPNF, and the Federal Food, Drug, and Cosmetic Act (FDCA) [8]. For enforcement purposes, the FDCA states that a drug shall be deemed to be adulterated if it purports to be or is represented as a drug, the name of which is recognized in an ofcial compendium and its strength differs from, or its quality or purity falls below, the standards set forth in such compendium [9]. Such determination regarding strength, quality, or purity shall be made in accordance with the tests or methods of assay set forth in such compendium. USP is not itself a regulatory body, but it provides Reference Standards that support the regulatory framework for therapeutic agents. Reference Standards are chemicalsnot drugs or devices. They are not intended for use in the diagnosis, cure, mitigation, or treatment of diseases but are offered for use as comparison standards in monograph procedures. The use of a molecular formula for the active ingredient named in dening the required strength is intended to designate the chemical or chemicals having absolute (100%) potency, e.g., cortisone acetate tablets contain 90110% of the labeled amount of cortisone acetate, not 90110% of cortisone acetate (C23 H30 O6 ) itself, which, as the drug substance, must separately meet an acceptance criterion of 97102% of C23 H30 O6 . The ofcial USP Reference Standard is C23 H30 O6, labeled to indicate the estimated content that is in fact C23 H30 O6 . Procedures in new USP or NF monographs requiring the use of Reference Standards are not in effect until the Reference Standard is available [10]. USPs ofcial lot system presupposes all users are testing to procedures traceable to the ofcial USP Reference Standard. Using validated analytical procedures, users should obtain the same result in testing in accordance with sound metrologic principles. This centrally controlled system with a single lot in commerce is dissimilar to that for a therapeutic article, for which many different lots may be in commerce at a specied point in time. 3. History The USP Reference Standard collection began with an announcement of the availability of standards in USP X (1926). Early Reference Standards frequently were complex materials for biological assays, moving over time with the rise in modern pharmaceutical chemistry and manufacturing, toward well-characterized articles, and returning more recently to less-well-characterized articles (botanical dietary supplements and natural-source or recombinant DNA bioFig. 1. Growth of USP Reference Standards Collection (number of Reference Standards per year).

logicals). This evolution is evident through the years: pepsin (1936); cod liver oil (1942); sulfanilamide, insulin, posterior pituitary (1942); melting point standards (1946); penicillin G sodium, heparin sodium (1950); negative control plastic (1966); dissolution calibrators (1978); endotoxin (1981); insulin human (1985); particle count set (1990); oxytocin (1996); and powdered ginger, Alliin (1999). Advances in analytical procedures during the past several decades have moved from those relying on direct measurement to those that rely on instrumental techniques (e.g., spectroscopic or chromatographic procedures) that frequently rely on physical standards for comparisons. With this transition, the USP catalog of Reference Standards has grown substantially (Fig. 1). FDA generally has welcomed USPs efforts in making Reference Standards available to facilitate analytical testing, with transfer of selected biologic standards in the 1930s and, more recently, the shift of antibiotic Reference Standards from FDA to USP in the 1970s [11].

4. Collection overview The collection of ofcial USP Reference Standards usually consists of highly characterized specimens of articles (drug substances, excipients, and impurities, including degradation products), and procedural standards, e.g., melting point, dissolution, particle size, and other calibrators. They range widely in appearance (crystalline or amorphous powders, volatile or viscous liquids, solutions or suspensions, gels or pastes, plastic sheets, and photomicrographs (in development)), structure (from simple inorganic salts to proteins produced by recombinant technology and cells (in development)), and composition (from single components to complex mixtures drawn from plant or animal sources). Their qualitative and quantitative uses also range widely according to the tests and procedures of the USPNF monograph, including: Identication (qualitative); Assay (quantitative); Impurities (quantitative as feasible); System Suitability tests; and blanks and controls. Despite their varying applications, the primary uses of ofcial USP Reference Standards are for spectroscopic and chromatographic procedures. They are used less often in other situations, including microbial assays (antibiotics), enzymatic reactions, animal tests, in vitro biochemical tests, titrations, and thermal analysis. Compendial uses of ofcial USP Reference standards and detailed information about their storage are provided in USPNF General Chapter 11 USP Reference Standards.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

4.1. Sources of candidate material Until the early 1980s, USP acquired most of its candidate Reference Standard materials from innovator companies that manufactured both active ingredients and dosage forms. As pharmaceutical manufacturing has moved offshore, so has USPs Reference Standards candidate acquisition. For the ve-year period between 1999 and 2004, one-third of the USP Reference Standards candidate materials were obtained from non-US companies that had a manufacturing site within the US, and another third came from US companies that have manufacturing plants overseas as well as in the US. The remaining third comes from international companies without manufacturing sites in the US. As long as a Reference Standard candidate represents an article that is legally approved for marketing in the US, USPs Reference Standards Laboratory is able to seek it in both domestic and global commerce [4]. At times, USP will resort to custom synthesis of a reference standard. The custom-synthesis laboratories are qualied under USPs vendor qualication program (which is governed by SOPs in accordance with USPs certication to ISO 9001 and 17025 standards), and the material is tested according to the same laboratory process as are candidate materials obtained from innovator and generic manufacturers. As part of its ISO 9001compliant quality management program, USP has implemented a supplier evaluation program. This program ensures that USP selects suppliers based on their ability to provide products or services that are in accordance with predened internal requirements. A specic audit checklist, developed to ensure the quality of bulks provided by companies other than the innovator and primary generic manufacturer, is completed prior to the approval of any of these bulks for use as Reference Standards. Completed checklists are audited and reviewed by USP Quality Assurance staff to ensure that the appropriate quality systems are in place at these facilities. In addition to this checklist, an on-site verication of quality systems may be conducted, as needed. In addition, the selection process includes a review of all FDA and ISO compliance audits that have occurred at the manufacturing site. Review of these audits adds to USPs condence in the suppliers ability to deliver a quality product. 4.2. Evaluation of candidate materials: ingredients All candidate ingredient materials received for consideration as an ofcial USP Reference Standard are evaluated by a minimum of three laboratories. These include USPs Reference Standard Laboratory, an FDA laboratory, and an independent, third-party laboratory that may belong to the primary manufacturer. Since September 2004 USP has been involved in a collaborative research agreement with Health Canadas Health Products and Food Branch Inspectorate. This agreement formalizes long-standing cooperation between the two nations and species that Health Canada laboratories will assist in the testing and characterization of

candidate Reference Standards materials. USP, FDA, Health Canada, and any commissioned contract laboratories tend to follow an identical protocol, but prior testing conducted in a primary manufacturers laboratory may have followed a different protocol. For a drug substance, the evaluation consists of at least one laboratorys performing a full set of tests according to the monograph specication (and additional tests as required), and all laboratories carry out additional studies to estimate content. USP relies on comprehensive testing and careful review of generated data by USP staff and by the USP Reference Standard Committee to ensure the quality of an ofcial USP Reference Standard. Additional continued suitability for use testing is done to ensure that this quality is maintained. The processes by which a candidate ingredient material becomes an ofcial USP Reference Standard is at times complex and necessarily varies according to the article, its compendial uses, the analytical procedures utilized, whether or not the article has undergone a regulatory approval, and other factors. A general path for an article that has undergone a regulatory approval is shown in Fig. 2. In this most common approach, candidate material accompanies a manufacturers Request for Revision [12], using information reective of the regulatory ling and approval (e.g., characterization studies, methods validations, and the private specication concluded between the sponsor and the regulatory body). In many instances, the USPNF monograph procedures are standardized approaches that for convenience and

Fig. 2. Process by which a candidate Reference Standard Material (RSM) becomes an ofcial USP Reference Standard (PF, Pharmacopeial Forum; RSL, Reference Standards Laboratory; RDL, Research and Development Laboratory; RSC, Reference Standards Committee; CoE, Council of Experts; RSO, Reference Standards Operations; CDER, Center for Biologics Evaluation and Research; CDER, Center for Drug Evaluation and Research; CVM, Center for Veterinary Medicine; EPA, Environmental Protection Agency; CVB, Center for Veterinary Biologics; CDRH, Center for Devices and Radiological Health). Note that more than one FDA laboratory may participate in methods validation, and Reference Standards Validation Sponsors may include other organizations. For example, USP has a collaborative research agreement with the Health Products and Food Branch Inspectorate of Canada to evaluate candidate Reference Standards.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315 Table 1 Four general approaches to assess the purity of a candidate reference material Approach Functional group analysis (e.g., titration) Assay against another standard Pros Accurate and precise; Absolute determination; does not require a standard The standard is evaluated in the conditions in which it will be used; Selective; Ensures continuity between different lots Potentially low error Cons

Mass balance (subtracting from 100 the sum of volatiles, residue on ignition, and chromatographic impurities)

All (or some) of the above

Flexibility in making the decision

Nonselective Requires standardized titrant, calibrated volumetric glassware Not very precise; requires large number of experiments; Assumes that existing standard has not changed and that original purity determination was valid In many cases the chromatographic impurities are unknown, or standards for them are not available. The impurities are then determined by the very risky assumption that the impurities have the same response factor as the main component. Flexibility in making the decision

general applicability are expressed in General Chapters (e.g., 281 Residue on Ignition or 921 Water Determination) that can be adapted and veried as suitable for a particular article. In other instances, a specic compendial test must be developed both for the private and public specication, particularly for a monographs Identication, Impurities, and Assay testsand other article-specic tests as appropriate. This generally occurs after an FDA review and acceptance of the private specication, coupled with methods validation testing in FDA laboratories [13]. USP uses at least four general approaches to assess the purity of a candidate RM (Table 1). The most commonly employed is mass balance, which is based on measurement of impurities, including water, and subtracting from 100. For noncomplex active drug substances and excipients, usually results from three separate laboratories are combined to yield an estimate of purity (which may be expressed as a calculation value, for example as mg/mg). USP General Chapter 11 USP Reference Standards notes that ofcial USP Reference Standards are established and released under the authority of the Board of Trustees upon recommendation of the USP Reference Standards Committee (RSC),5 which evaluates the selection and suitability of each lot [4]. Based on review of data from characterization and collaborative testing studies, RSC balloting occurs and must be unanimous for a positive decision.6 With a positive decision, the material is subdivided and labeled, quality control checks are performed, and the material, now an ofcial USP Reference Standard, is listed in the USP catalog and becomes available for distribution. Statements provided include safety warnings, required information for controlled substances,
5 In the 20002005 cycle and in prior cycles, the RSC at USP traditionally was an appointed rather than elected committee of the Council of Experts, chaired by staff. In the 20052010 cycle, it is an elected Expert Committee. 6 From 1999 to 2004, 13 donated candidate materials were rejected out of 1399 received. Twelve items did not successfully complete the collaborative testing process and were not presented to the RSC for review. The RSC rejected one item. The success rate for custom synthesis is similarly high. Of the 10 items synthesized, only one was synthesized improperly (Succinylmonocholine Chloride was synthesized with the wrong salt. During this period, 37 items from India, China, and Japan completed the full evaluation process, and none was rejected by either Standards Operations at USP or by the Reference Standards Committee.

content or potency for quantitative-use standards, acceptance ranges as needed, and other useful information (Fig. 3 shows a proposed Reference Standard certicate). USP does not provide information from characterization studies and collaborative testing, which might be included for a certied reference material, because all the information that the user needs for the ofcial applications of the standard is provided in the label text and, as necessary, in the additional documentation provided. 4.3. Evaluation of candidate materials: impurities In the mid-1980s, based on various advances USP reconsidered its approaches to the measurement and control of impurities in drug substances and other articles. In 1986 USP held an open conference (its rst) on impurities. Most participants supported a need for improved impurities approaches. The 1986 open conference set the stage for USPNFs General Chapters 461 Ordinary Impurities and 1086 Impurities in Ofcial Articles. General Chapter 466 was created to provide detailed instructions for carrying out thin-layer chromatography (TLC) to measure impurities. General Chapter 1086 dened various impurities (ordinary, toxic, signal, concomitant, related substances, foreign substances, and process contaminants). Via a Pharmacopeial Forum (PF) Headquarters Column [14], USP invited manufacturers to submit TLC procedures for drug substance impurities. In the absence of a strong response, USP laboratories tested candidate Reference Standard material for impurities using TLC. Based on data from these analyses, the Subcommittee on Impurities proposed impurity tests for several drug substance monographs via PF. As many as 4050 monographs in a single year were revised to include TLC testing for ordinary impurities using this approach, with approximately 200 concluded during several years. These early efforts form the basis for an approach, present in many USP monographs, to control impurities via the Related Substances and Chromatographic Purity tests. These approaches did not always control specic impurities and generally stated that all impurities taken together should not exceed 2% unless otherwise indicated in the monograph. This approach was less than optimal, not only because of

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

Fig. 3. Example of a proposed Reference Standard certicate.

its de minimus character but also because it was based on a one-size-ts-all approach without regard for different impurities arising from different routes of synthesis. The approach has generally been superceded by approaches in the Quality documents of ICH. For noncomplex drug substances and dosage forms, these documents focus on stability testing (Q1 documents), analytical validation (Q2 documents), and impurities (Q3 documents). The ICH Q6A document denes characterization studies that lead to private specications concluded between an applicant and a regulatory agency [15]. The private specicationand subsequently the public oneconsists of universal tests (Description, Identication, Assay, and Impurities) and specic tests that can vary depending on the intended use of the ingredient and the type of dosage form. ICH Q3A(R), Q3B(R), and Q3C provide a comprehensive set of approaches for controlling impurities in ingredients and dosage forms [1618]. Although impurities may be controlled by good system suitability testing and by response factors, as often is the case in the USP Chromatographic Purity test, general agreement exists that optimal control for specied impurities is an Impurity test procedure that adequately identies impurities. This is best accomplished when the widest range of impurity Reference Standards is available [19]. USP, as part of its mission to advance public health, is exploring options to obtain candidate impurity Reference Standards materials

and to advance them to ofcial status, recognizing that the increasing globalization of the pharmaceutical industry offers both opportunities for the rewards of free markets and the dangers of counterfeiting.

5. Assignment of content A key conclusion of the RSC, based on collaborative testing, is the value assignment for content of an ofcial USP Reference Standard. In the past, USP usually applied a value of 100% using either two or three signicant gures.7 Over time, this practice came to be challenged for several scientic and pragmatic reasons. To address the issue, USP formed Project Team 4 (Reference Standards), which worked in association with the Prescription/Nonprescription Stakeholder Forum.8 In addition, USP conducted some statistical
7 The number of signicant gures in the labeled calculation value is a function of the use of the standard and the number of signicant gures in the acceptance range or limit. Generally, Reference Standards used in assays are labeled with three signicant gures, and standards used in limit tests have two signicant gures. Reference Standards that have multiple applications in different methodologies may require separate assay-specic assignments. For calibration standards, the labeled value is determined by a statistical analysis. 8 For members of the Project Team, see author list.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

studies on this topic; the more detailed work is presented in the Appendix A. 5.1. Content assignment: ingredients The Project Team met on four occasions (October 2002, January 2003, May 2003, and October 2003), at times concurrently with the Reference Standards Committee. Their deliberations and recommendations provide further insights into the technical details of characterizing, maintaining, and validating Reference Standards. One of the Project Teams earliest objectives was to develop a characterization protocol to outline the type of collaborative testing required for each category of Reference Standard and to dene how Reference Standard assignments should be made. Characterization protocols are based on the three ways that Reference Standards are used: (1) quantitatively; (2) nonquantitatively; and (3) special applications. The protocol tests include: identication, impurities, ROI, water, testing against Reference Standards, testing against previous lot(s) of USP Reference Standards, and monograph tests. The Project Team recommended that the assignment strategy for USP Reference Standards should be based on mass balance when possible. Comparisons to previous lot(s) of USP Reference Standards generally should not be used for assignment purposes but should be completed to conrm an assignment and to determine if Reference Standards users observe a shift in results. When mass balance is not practical and an international Reference Standard is not available, direct methods can be used for assignment as well as for assay-specic assignments. Finally, the Project Team agreed that there should be no fewer than two collaborators and that replicates should be based on Reference Standard type; further, if the variability of the two collaborators work was greater than the relative standard deviation (R.S.D.) of the method as submitted by the monograph sponsor, further studies should be conducted. For a quantitative Reference Standard value, the Project Team recommended using a determined (actual) value for the labeled calculation value, not a threshold above which 100% is assigned. This approach has several positive aspects, including: (1) it needs no justication (unlike the case when one is assigning thresholds) because one is assigning determined (i.e., actual) values; (2) there are no rounding issues with the use of a determined (actual) value (e.g., 99.45 100%); (3) for assays of active pharmaceutical ingredients (APIs), potential problems arise when one assigns threshold values in the context of tight API limits; (4) materials with obvious impurities are assigned a value <100%; (5) a slight purity change does not result in a larger assignment change (e.g., if 99.5 100% and the USP Reference Standard changes to 99.4%, then the labeled value changes 100 99.4%). 5.2. Content assignment: impurities The Project Team evaluated impurity Reference Standard values and recommended that USP should use a threshold

above which 100% is assigned (e.g., if 98% 100%). Positive implications of this approach include: (1) error caused by rounding is insignicant compared to error associated with the method for impurities at low levels; (2) error in rounding an impurity Reference Standard results in a more conservative estimate of impurities in drug substances and drug products; and (3) accurate quantication of an impurity is not as critical as it is for the ingredient. The recommendation is qualied in that: (1) using thresholds is not a true scientic approach; (2) if rounding occurs near a threshold, the difference could be detectable with the method; (3) if an error in rounding an impurity Reference Standard results in a more conservative estimate of impurities in drug substances and products, one faces the potential of rejecting material that actually meets specications. 5.3. Value reassignment Project Team 4 considered the reassignment of labeled Reference Standard values; i.e., what is an appropriate threshold difference beyond which relabeling is required? The team concluded that no xed threshold exists for all productsdecisions must be made on a case-by-case basis. However, for a typical well-characterized small molecule, reassignment might be necessary if the change is >0.5% for a material of purity >98% or if the change is >1% for a material of purity <98%. If assignment values are three signicant gures and retesting (at the USP laboratory) differs by more than 0.5%, then further data should be collected, possibly by a collaborating laboratory. How these data will be collected, assessed, and compared raises signicant questions involving statistical analysis and collaborative testing. Re-evaluation typically takes place at a single USP laboratory.

6. Discussion In offering a USPNF monograph and ofcial USP Reference Standard, USP wishes to base its activities with the latest principles of sound metrologic science. These principles speak to tness for purpose, as follows: The tness for purpose of chemical measurements is formally dened as the degree to which data produced by a measurement process enable a user to make technically and administratively correct decisions for a stated purpose [20]. A key element in the concept is for the interested parties to dene in advance the acceptable degree of measurement uncertainty and desired degree of identication condence [21]. In addition to being the criterion for assessing when any aspect of the measurement effort is adequately complete, t-for-purpose considerations are central to the prospective design of a measurement study [22,23]. The better dened the purpose, the more realistic the forecast of analytical effort required to achieve tness. An unrealistic, unclear, or overly broad purpose may result in unnecessary costs, delay, or failure of measurement study.

10

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

Fig. 4. Chart demonstrating some linkages between USP monographs for non-complex actives with ICH Quality approaches.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

11

Compendial approaches have a strong role in assuring practitioners and patients and the public at large that a therapeutic article is t for purpose, i.e., is safe and/or effective in the maintenance of health and treatment of disease [13]. In this context, modern metrologic science supports USPs long-standing objective of ensuring the identity of an article via the test procedures and other standards of the monograph, regardless of who is manufacturing the article, who is testing it, and when or where it is tested. As USP works with rst-, second-, and third-party quality control laboratories, it is important to understand roles and responsibilities. USPs monographs and ofcial USP Reference Standards are most commonly used in hypothesis testing studies by quality control laboratories. USP does not engage in hypothesis testing itself but rather provides the measurement study (monograph) and ofcial USP Reference Standard to facilitate testing. The hypothesis of a quality control laboratory is that the article, when tested, yields a result that either does or does not fall within the acceptance criteria. If results fall within the acceptance criteria, the article is deemed acceptable. If not, the result may be deemed out of specication [24]. The article tested is an ingredient or a dosage form. The measurand is the active pharmaceutical ingredient and/or its impurity(ies) in the drug substance or dosage form. As a rule, the SI units for small molecules are almost invariably expressed in terms of mass (kilogram units). The general issue of either accepting or rejecting the testing hypothesis paradigm leads to issues of consumer and producer risk. In the past USP has not always carefully considered these two aspects of hypothesis testing. Recent work [25] has suggested this is a fruitful area of study for both manufacturers and compounding professionals, as well as practitioners and patients. In the coming cycle, USP wishes to explore the value of estimating the uncertainty in the estimation of content in selected ofcial USP Reference Standards. This uncertainly could be added to the overall uncertainty in hypothesis testing of, for example, measurand mass. USP and other pharmacopeias have not done this in the past on the assumption that uncertainly for a pharmacopeial reference material is negligible in relation to the dened limits of the method-specic assays . . . [26]. In fact, based on preliminary studies, USP believes that this may not always be the case and, further, that the denition of negligible might at times require careful consideration. USP believes that a better understanding of an uncertainty statement for an ofcial USP Reference Standard, coupled with an uncertainty estimate of tests applied to the measurand, may help quality control laboratories reduce the likelihood of failing to meet acceptance criteria. Uncertainty in measurement should also be considered in setting acceptance criteria [3]. During the 20002005 cycle, USP created a Guideline for Submitting Requests for Revision to USPNF [10]. The Guideline provides instructions to Sponsors intending to submit Requests for Revision and harmonizes many elements of the USP monograph with the ICH Quality approaches

(Fig. 4). USP expects to revise this document continuously. For selected candidate reference materials, USP will add a section that provides a protocol with study design and analysis approaches. This protocol will focus initially on smallmolecule ingredient and impurity candidate materials and can be expanded subsequently for other candidate materials as needed. The integrity of the USP monograph and allied ofcial USP Reference Standards is of paramount importance to USP. The authority to create and maintain both the documentary standards in USPNF and the physical reference materials in the USP collection is a privilege in the United States. In other countries, the authority is reserved to the government. USP takes this privilege seriously and continuously works to merit it through close attention both to its processes, e.g., those of the RSC, and to the laboratory activities that support the availability of an ofcial USP Reference Standard. In this regard, USP is especially proud of its attainment of ISO 9001 and 17025 certication during the 20002005 cycle. As this article demonstrates, USP also wishes to work closely with rst, second, and third parties who rely on USPs ofcial Reference Standards as a means of assuring the public that the articles they place in commerce meet optimal quality standards. Appendix A. Statistical topics A.1. Introduction During the 20002005 cycle, USP engaged in a statistical analysis of various issues associated with collaborative testing of candidate reference material for value assignment of content for a small molecule ingredient. This effort involved theoretical considerations, a pilot study, and evaluation of data from current collaborative testing. A.2. Theoretical considerations The current USP collaborative study design for smallmolecule Reference Standards typically uses three laboratories (L), with one experiment (E) per laboratory and 16 determinations (D) per laboratory. The precision of the estimated content depends on three sources of variability: 1) interlaboratory variability; 2) inter-experiment variability within laboratory; and 3) intra-experiment variability (these can be based, e.g., on determinations, including inter-injection and inter-preparation for HPLC). If L, E, and D are all >1, all three variances can be estimated. Otherwise, only combinations can be estimated, which may be acceptable depending on the purpose of the study. For example, in the USP design for non complex ingredient Reference Standards, only one experiment per laboratory is usually performed, which precludes determination of a separate variance for experiment; any inter-experiment variability cannot be separated from inter-laboratory variability. As a means of simplifying presentation of formulas, a design is called balanced if there

12

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

are equal numbers of determinations per experiment and an equal number of experiments per laboratory. For each component to be measured in a mass-balance determination, the precision of that measurement is needed. Let S2 denote estimated variances, with subscripts L, E, and D for laboratory, experiment, and determination, respectively. The standard error (SE) of the estimated measurement for a single component of a mass-balance determination from a balanced design is then: SE =
2 2 2 SL SE SD + + L LE LED

(1)

For condence intervals, the degrees of freedom of the tstatistic is L-1, regardless of the magnitude of E and D. If the design is not balanced, then there is no simple formula for SE, and statistical software must be used. To determine the precision of estimated content determined by mass balance, the standard errors from the various components need to be combined. Suppose there are C components to be assayed. For each component, this results in a standard error. The standard error for the content is then found by adding on the variance scale: SEP = SE2 + + SE2 . C 1 (2)

increasing sample size; i.e., to increase precision, the sample size is increased with attention to each source of variability. Increasing the number of determinations, for example, does nothing to reduce the contribution from inter-laboratory or inter-experiment variability (see formula (1)). The only way to reduce the contribution from inter-laboratory variability is to increase the number of laboratories collaborating. If the goal of the study is to have a precise estimate of the content, then part of the design of the collaborative study should be specication of the desired level of precision. This could be stated in terms of a desired standard error or of a desired condence interval width. The condence interval is preferable because it incorporates information about how well the variances are estimated. There usually is no unique approach for choosing L, E, and D; choices typically are determined by cost, logistics, and other factors. The following two sections summarize work conducted by USP to assess whether the current design was adequate for the purposes of assigning a content to USP Reference Standards. A.3. Pilot study A small pilot study of three candidate Reference Standard materials was conducted with the objective of determining the effect of increasing the number of collaborating laboratories on a) the nal assignment by mass balance calculation of the content of candidate Reference Standard materials and b) the precision of the assignment. The selection criteria for three candidate Reference Standard materials were: 1) the standards had to have an ofcial quantitative application in an assay (thus requiring an assignment with three signicant gures; 2) the standards must not yet have been evaluated in any laboratory (to avoid any potential bias); 3) the standards had to be on the USP list of priority candidates (so that the data could be submitted in a timely manner to RSC for approval); 4) the projected content was to be assigned by mass balance using results from at least two tests (practically, that meant that the standards had to be used as-is with water content being one of the elements used in the mass balance calculation; and 5) the candidates must not have been rst-time standards. Study protocols included number of preparations, number of replicates, and sample size. Participating laboratories were USPs Reference Standards Laboratory and FDA laboratories; RDL and accredited contract laboratories also participated. Candidates selected were bendroumethiazide, sotalol hydrochloride, and valproic acid. All three materials presented special analytical challenges: (1) bendroumethiazide testing requires two separate HPLC chromatographic purity tests and decomposes so rapidly in one of them that only singlet injections can be made from each preparation; (2) sotalol has three identied related compounds that are determined by external standard testing using the respective USP Reference Standards; and (3) valproic acid is hygroscopic, and the water content determination requires special precautions.

Here the subscript P is for the nal content (purity) estimate and the other C SEs are for the C components. SEp is the combined standard uncertainty of the estimated content.9 Determination of a condence interval is more difcult. A conservative approach is to proceed as above using the t distribution with L-1 degrees of freedom. An alternative is to use Satterthwaites approximation, as suggested by the National Institute of Standards and Technology (NIST) [3]. The condence interval is of the form C kSEp , where C is the estimated content and k is a coverage factor that depends on the degrees of freedom. (USP does not use the conventional choice of k = 2, due to the small number of degrees of freedom). The term kSEp is the expanded uncertainty of the estimated content. One implication of these considerations is that we can identify situations in which USP is unable to determine the precision of an estimated content. For example, if any non negligible component of a mass balance determination is measured by only one laboratory, it is not possible to estimate inter-laboratory variability, and hence the precision cannot be determined. Also, because in most studies the design is not balanced due to unequal numbers of determinations at the laboratories, USP needs information about the individual replicate values from all laboratories. A second implication is an understanding of how to consider the design of these studies. Variability is addressed by
9 In ISO and related documents, the combined standard uncertainty is described as a standard deviation. The standard deviation of the estimated content is more commonly called its standard error, the term the authors keep to.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315 Table A1 Results for bendroumethiazide Number of laboratories 5 4 3 2 Average impurities D (%) 0.05 0.06 0.05 0.04 Average other impurities (%) 0.26 0.26 0.26 0.26 Average water (%) Estimated Standard error 0.21 0.21 0.22 0.28 99.47 99.47 99.47 99.43 Average (%) 0.04 0.05 0.06 0.03 99.37, 99.58 99.32, 99.62 99.20, 99.73 99.06, 99.80

13

Content 95% Conf. Intl.

Table A2 Results for sotalol hydrochloride Number of laboratories Average impurities (%) Average water (%) Estimated Average (%) 5 4 3 2 Table A3 Results for valproic acid Number of laboratories Average impurities (%) Average water (%) Estimated Average (%) 5 4 3 2 0.03 0.04 0.04 0.06 0.20 0.25 0.14 0.06 99.77 99.71 99.82 99.87 Standard error 0.11 0.12 0.08 0.03 99.47, 100.06 99.32, 100.10 99.46, 100.17 99.46, 100.28 Content 95% Conf. Conf. Intl. 0.17 0.17 0.17 0.16 0.08 0.09 0.08 0.10 99.76 99.74 99.74 99.74 Standard error 0.02 0.02 0.02 0.01 99.70, 99.81 99.69, 99.80 99.64, 99.84 99.62, 99.85 Content 95% Conf. Intl.

For each compound, the results were determined with ve sets of laboratories. Results shown for two laboratories used laboratories labeled #1 and #2. Those for three laboratories used results for laboratories labeled #13, and similarly for four and ve laboratories. Condence intervals were determined using a t-distribution with degrees of freedom equal to the number of laboratories minus one. Tables 13 show the results for the three compounds. Standard errors for the three compounds range from 0.02 to 0.12. All are precisely estimated. The condence intervals widths reect the magnitude of the standard error and the number of laboratories (an indicator of how well the standard error is determined). Normal expectation would be that the condence intervals become narrower as the number of laboratories increases, and the tdistribution multiplier drops from 12.71 to 2.78 as the number of degrees of freedom increases from 1 (two laboratories) to 4 (ve laboratories). That does not always happen here to the extent expected Laboratories 1 and 2 are more consistent with each other than are the other laboratories; notice that the smallest standard error in each table is with two laboratories. With three important caveats, the results of this pilot study do not support routinely increasing the number of laboratories from the current three for the type of candidate materials selected for study. The caveats are: the pilot study results apply only to compounds whose content is determined by mass balance,

the pilot results refer only to three compounds that may not be representative of all compounds that may be measured by mass balance, and the results apply only to data collected as described herenamely with a standardized protocol and data collection form. One can speculate that the standardized protocol and data collection may be contributing to the good precision seen here. Results from other compounds following current standard practice may be informative about this point. In any event, this pilot study brings us to a consideration of important ways in which modern metrology informs and improves

Table A4 Reviewed evaluation packages RSCEP 719 808v01 850 856 862 878 1112 1127 1130 1143 1144 1164 Compound Cefpodoxome Proxetil Naratriptan HCl Clonidine Related Compound B Metoprolol Succinate 2E,4E-Hexadienoic Acid Isobutylamide Leuprolide Aetate Amiodarone Hydrochloride Fexofenadine Hydrochloride Mangafodipir Trisodium Losartan Potassium Tolcapone Morantel Tartrate

14

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315

Table A5 Condence interval results for seven RSCEPs by mass balance determination RSC Water (%) Average 878 (Values in water columns are for acetic acid) 1112 1127 1130 1143 1144 1164 7.690 0.094 0.220 0.193 0.045 0.205 SE 0.053 0.027 0.044 0.011 0.006 0.038 Impurities (%) Average 1.492 0.253 0.078 0.402 0.046 0.006 0.098 SE 0.151 0.003 0.014 0.080 0.033 0.023 Negligibles (%) Average 0.1 0.1 0.11 Estimated (%) Average 90.7 99.6 99.6 99.6 99.8 99.9 99.7 SE 0.16 0.03 0.05 0.08 0.03 0.01 0.04 Approx. 95% Conf. Intl. (90.0, 91.4) (99.4, 99.7) (99.4, 99.8) (99.3, 99.9) (99.6, 99.9) (99.9, 100.0) (99.5, 99.9) Monograph Accept. Range 3%

3% 1.5%, +1.0% 1.5%

0.025

Negligibles are residue on ignition (878, 1112, and 1164), and sulfated ash and solvents (1127).

the processes involved in identifying, preparing, and working with Reference Materials.
[4]

A.4. Analysis of 12 representative data sets

To assess current practice, USP conducted a statistical analysis to obtain the condence intervals for the estimated contents for 22 rst-time Reference Standard candidate evaluation packages (RSCEPs) from two periods, December 2003 to January 2004 and August to October 2004. Here we consider only 12 assay standards with content determined by mass balance (Table 4). Other Reference Standards were for impurities or did not use mass balance. Uncertainty (standard error and condence interval) could be determined for 7 of the 12 candidates. Common problems are the lack of detail from Collaborator C, absence of replicate information about water determination from Laboratory A, and determination of non-negligible components by only a single laboratoryether in one case and water in another. For the AugustOctober set, USP staff obtained replicate information from Collaborators A and C. For the rst set, no contact was made. Table 5 shows condence-interval results for the seven candidates for which condence intervals could be determined. For three others from the rst set of candidates, it seems likely that replicate information could have been obtained if sought, so there are only 2 of the 12 candidates for which uncertainty could not be determined from the studies as currently conducted.

[6]

[7]

[8] [9] [10] [11] [12]

[13]

[14] [15]

References
[1] L. Anderson, G.J. Higby, The Spirit of VoluntarismA Legacy of Commitment and Contribution: The United States Pharmacopeia 18201995, The United States Pharmacopeial Convention, Inc., Rockville, MD, 1995. [2] H.G. Wolfe, Charles Rice (18411901), an immigrant in pharmacy, Paper presented at Joint meeting of American Pharmaceutical Association and American Institute of the History of Pharmacy, April 1949. Jacksonville, FL. [3] B.N. Taylor, C.E. Kuyatt, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, National [16] [17] [18] [19] [20]

Institute of Standards and Technology, Gaithersburg, MD, 1994. General Chapter 11 USP Reference Standards. USP 28NF 23. The United States Pharmacopeial Convention, Inc., Rockville, MD, 2005. p. 2204. W. May, R. Parris, C. Beck, J. Fassett, R. Greenberg, F. Guenther, G. Kramer, S. Wise, T. Gills, J. Colbert, R. Gettings, B. MacDonald, Denitions of Terms and Modes Used at NIST for Value-Assignment of Reference Materials for Chemical Measurement, National Institute of Standards and Technology Special Publication 260-136, Gaithersburg, MD, 2000, p. 10. S.L.R. Ellison, M. Rosslein, A. Williams (Eds.), EURACHEM/ CITAC Guide: Quantifying Uncertainty in Analytical Measurement, second ed., National Institute of Standards and Technology, Gaithersburg, MD, 2000, p. 9. D.L. Deuwer, R.M. Parris, E.W. White, V.W.W. May, H. Elbaum, An Approach to the Metrologically Sound Traceable Assessment of the Chemical Purity of Organic Referenxce Materials, NIST Special Publication 1012, National Institute of Standards and Technology, Gaithersburg, MD, 2004, pp. 3. Federal Food, Drug, and Cosmetic Act (Public Law No. 75-717 52 Stat. 1040), 1938. 21 U.S.C. 321(g). General Notices, USP 28NF 23, United States Pharmacopeia, Rockville, MD, 2005. p. 5. Valentino J.G., Pharm. Technol. 3 (9) (1979) 6164. Department of Standards Development. Guideline for Submitting Requests for Revision to USPNF. United States Pharmacopeia, Rockville, MD, 2004. Also available at http://www.usp.org/ standards/revisionguideline/index.html (accessed 05/09/05). T. Layloff, M. Nasr, R. Baldwin, M. Caphart, H. Drew, J. Hanig, C. Hoiberg, S. Koepke, G. Lunn, J.T. MacGregor, Y. Mille, E. Murphy, L. Ng, R. Rajagopalan, E. Sheinin, M. Smela, M. Welschenbach, H. Winkle, R. Williams, Pharm. Technol. 24 (1) (2000) 3042. L.T. Grady, W.M. Heller, Headquarters Column. Pharm. Forum. 11 (5) (1985) 620621. International Conference on Harmonization. Q6A: Specications: Test Procedures and Acceptance Criteria for New Drug Substance and New Drug Products: Chemical Substances, Geneva, Switzerland, 2000. International Conference on Harmonization. Q3A (R): Impurities in New Drug Substances, Geneva, Switzerland, 2000. International Conference on Harmonization. Q3B (R): Impurities in New Drug Products, Geneva, Switzerland, 2000. International Conference on Harmonization. Q3C: Impurities: Residual Solvents, Geneva, Switzerland, 2000. L. Bhattacharyya, H. Pappa, K.A. Russo, E. Sheinin, R.L. Williams, Pharm. Forum. 31 (3) (2005) 960966. IUPAC. Compendium of Analytical Nomenclature, Denitive Rules 1997. (Orange Book), third ed. Blackwell, Oxford, 1998.

R.L. Williams / Journal of Pharmaceutical and Biomedical Analysis 40 (2006) 315 [21] R. Bethem, J. Boison, J. Gale, D. Heller, S. Lehotay, J. Loo, S. Musser, P. Price, J. Stein. J. Am. Soc. Mass Spectrometry 14 (2003) 528541. [22] T. Fearn, S.A. Fisher, M. Thompson, S.L.R. Ellison, Analyst 127 (2002) 818824. [23] L. Bhattacharyya, T. Cecil, R. Dabbah, D. Roll, S. Schuber, E.B. Sheinin, R.L. Williams, USP Council of Experts Expert Committee, Pharm. Res. 21 (10) (2004) 17251731.

15

[24] Food and Drug Administration, Center for Drug Evaluation and Research. Guidance for Industry (Draft Guidance), ANDAs: Impurities in Drug Substances, Rockville, MD, 2004. [25] W.W. Hauck, T. Foster, E. Sheinin, T. Cecil, W. Brown, M. Marques, R.L. Williams, Pharm. Res. 22 (2) (2005) 182187. [26] ISO. Guide 34 General Requirements for the Competence of Reference Material Producers. second ed. Geneva, Switzerland: ISO Press; 2000:v.