Drug Information Journal, Vol. 34, pp. 919–959, 2000 0092-8615/2000Printed in the USA. All rights reserved. Copyright 2000 Drug Information Association Inc.

REGULATORY AND DEVELOPMENTISSUES IN THE DEMONSTRATION OFTHERAPEUTIC EQUIVALENCE FORMULTISOURCE BIOTECH-DERIVED

PHARMACEUTICALS

ROBERT L. ZEID

Principal Consultant, TLI Development, Oak Island, North Carolina

The exclusion of biological products (approved under the Public Health Service Act)from Title I of the Hatch-Waxman Amendments is being revisited today in the light ofphilosophical shifts in regulatory review, cost containment, and scientific consistency.The definition of drug versus biologic—be it scientific, legal, or administrative—is evolv-ing and these overlapping working definitions further blur the regulatory distinctionbetween drugs versus biologics. This is most clearly seen with the emergence of specifiedbiotechnology products and the precedent of Center for Drug Evaluation and Research(CDER)-approved recombinant deoxyribonucleic acid (rDNA)-derived products underSection 505 of the Food Drug & Cosmetic (FD&C) Act. Improvements in productionmethods, process controls, and analytical test methods have given the Food and DrugAdministration (FDA) the impetus for greater harmonization in regulatory review such asthe Biologics License Application (BLA), which replaced the Product License Application(PLA) and Establishment License Application (ELA). Industry and regulators collabo-rated to create fast-track approvals and the ‘specified biotech’ paradigm to shed thelong-standing “product = process” dogma. In doing so, entire classes of biotech-derivedproducts are being brought to market faster, and significant manufacturing changes arebeing supported without repeating extensive Phase III clinical studies, and as somepostulate—closer to the realm of multisource (or interchangeable products) competitionby noninnovator companies.

Though specified biotech and comparability protocols do not equal multisource bio-tech, there are many common conceptual elements, parallel tracks, and intersections thatsupport cost savings and faster route to market for noninnovator manufacturers, especiallywith regards to demonstrating therapeutic equivalence. Though each situation is case-dependent, what are some common features that could allow the same manufacturingchanges under specified biotech as those supported for noninnovator multisource biotechpharmaceuticals? Given the range of complexity of moieties under the biotech umbrella,no single approach of demonstrating therapeutic equivalence would suffice for all. Itshould be evaluated first on a product class basis (eg, monoclonal antibodies, interferon,fibrinolytics, etc.) and then on a case-by-case basis using a tiered approach of combinedanalytical characterization + pharmacokinetic/pharmacodynamic (PK/PD) assessments+ surrogate endpoint equivalence, if required. Assessment of immunogenicity, development

Reprint address: Robert L. Zeid, TLI Development, 113 SE 5th St., Oak Island, NC 28465.

919

920 Robert L. Zeid

of neutralizing antibodies, or other hypersensitivity reactions may require additionalclinical studies for both innovator and noninnovators alike.

To accomplish this goal, a sponsor may develop extensive structure-activity relation-ship (SAR) databases for major classes of compounds (eg, somatotropins, fibrinolytics,monoclonal antibodies, etc.) expanding upon principles outlined in current FDA andInternational Conference on Harmonization (ICH) chemistry, manufacturing, and controls(CMC) guidance documents. Throughout development a manufacturer (innovator or multi-source firm) may be able to support those physico-chemical differences that do not impactclinical safety or therapeutic equivalence. The expansion of these comparability protocolsinto SAR databases may eventually allow for class-specific guidance documents or mono-graphs and facilitate a unified regulatory review model for drugs and certain specifiedbiologics.

Currently, there are at least four potential regulatory filing strategies for multisourcebiotech pharmaceuticals: a 505(b)(2), a BLA, a Canadian New Drug Submission (NDS),or a European Marketing Authorization Application (MAA). There are other internationalfilings available (eg, Japan, Latin America, etc.) as well. The 505(b)(2) allows a multi-source biotech approval route—with an AB rating—when compared against innovatorproducts previously approved under Section 505 of the FD&C Act. Albeit the 505(b)(2)may be the most clearly defined route, it has limited utility in that the number of Section505-approved biotech products available to multisource competition is small when com-pared to those specified biotech products approved under the PHS Act. The BLA is alsoa new application that will allow multisource biotech products to compare therapeuticequivalence to innovator products approved under the PHS Act. Demonstrating therapeu-tic equivalence to an innovator compound, however, does not result in an AB rating sincethere is no mechanism under the Hatch-Waxman Amendments or current FDA policythat would allow assignment. Hospital formularies, however, can support interchangeabil-ity with greater latitude based on efficacy data and physician input. The MAA must befiled via the Centralized Procedure but it is not clear that there would be consensusamong the Member States as to an adequate level of testing. Thus, a scientificallysound and consistent approach is lacking due to variations in legal, administrative, andregulatory definitions. Despite these limitations, could a single standard be developedand applied to such a multitude of biologic product scenarios as is currently done forAbbreviated New Drug Applications (ANDAs)?

While the multisource biotech dossiers could not follow the exact same approachcurrently used for generic drug applications, there are many common elements. Thephysico-chemical characterization and PK/PD comparisons would be to a listed productor external reference standard. Where changes were observed, those aspects would behighly integrated into an existing SAR development database for their impact on safetyand efficacy. Where those data did not exist, the applicant would be compelled to provideadditional data/studies. Thus, a tiered approach of demonstrating therapeutic equivalencewill be applied and become increasingly harder to show based on: 1. Product complexityand characterization issues, 2. Clinical indications, and 3. Demonstration of additionalsafety for observed changes. This process allows for the greatest linkages to establishedsafety and efficacy data by first demonstrating physico-chemical comparability for bothinnovator and multisource competition alike. Where differences exist, those changes mustbe evaluated in a scientifically sound and consistent manner.

This article reviews technical considerations for analytical, pharmacokinetic, clinical,and regulatory aspects in demonstrating therapeutic equivalence (of various classes ofbiologics) in the light of current FDA policy and law. Patent law and manufacturing/process development issues are not covered in this article but are detailed in severalreferences.

Key Words: Generic biotech; Multisource biotech; Therapeutic equivalence; Structure-activity-relationship; Surrogate endpoints

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 921

INTRODUCTION provals of both biologics and drugs. The tidehas turned for greater consistency in their regu-lation (eg, Section 123(f) of FDAMA) (4).HISTORICALLY, BIOLOGICS HAVE been

regulated a world apart from classically syn- With all of these changes—plus the prece-dent of rDNA-derived products approved un-thesized drugs. The distinctions—evident in

public health law (1–4), FDA policy (5–8), der Section 505 of the FD&C Act—the clas-sic distinction of drug versus biologic isand industry practice—reflect a history of

substances that were complex and difficult increasingly blurred. The review and ap-proval of biotech-derived products in theto characterize (9,10) and/or used potentially

contaminated source materials (11,12,13). classic ‘drug’ domain challenges one to de-fine a true biologic—and in doing so, hasLimited analytical tools and product-specific

processes galvanized the long-standing ap- brought entire classes of products closer tothe realm of multisource (or interchangeable)proach that biologics were defined by the

manufacturing process. Thus, the “product = biotech from noninnovator companies (32–35). In the quest to chart a course for anprocess” dogma permeated throughout bio-

logics development, leaving it with some expedited review process for specified bio-tech products, the intrepid have churned upunique milestones like an ELA. Process

changes often required clinical ‘bridging’ calm channels that have long separated bio-logics from drugs. With others now follow-studies to demonstrate equivalency of pilot

scale material used in pivotal efficacy trials ing in their wake, some recognize—perhapstoo late—the chance to return to familiarversus the commercial scale material for

market. It was this process-specific distinc- shores is gone.tion more than anything else that excludedbiologics from Title I but not Title II of the

REGULATORY FILING STRATEGIESformal ANDA process (3). This logic, how-

FOR MULTISOURCE BIOTECHever, is being challenged with the emergenceof the ‘specified (or well-characterized) bio- There are currently two regulatory filing

strategies that could lead to approval of alogic,’ comparability protocols, and the con-solidation of separate establishment licenses therapeutically equivalent biotech product in

the United States: a 505(b)(2) or a BLA. Aand product license approvals into a singleBLA (9). third filing strategy is a Marketing Authori-

zation Application (MAA) via a CentralizedIn the last 15 years, the international regu-latory standards for biologics (14–18) have Procedure that would be used for a European

dossier. The fourth filing strategy is the Ca-become considerably more sophisticated andcircumspect—a reflection of improved pro- nadian NDS. Other international filing strate-

gies for Japan, Latin America, the Middleduction methods, and process controls, ana-lytical methods, as well as the regulatory re- East, and so forth are available but are not

discussed in this article. The developmentviewers’ increased familiarity with rDNAproduct safety, and current Good Manufac- path for the United States filings is depicted

in a schematic flowchart (see Figure 1) thatturing Practice (cGMP) issues (19–30). Thislearning curve—combined with the FDA captures some of the major considerations

and milestones. Note that some terminologyModernization Act of 1997 (FDAMA) andthe impetus for streamlined government— will differ for the MAA or NDS but the con-

cepts are highly conserved across all submis-has forged greater harmonized review be-tween CDER and the Center for Biologics sion types.Evaluation & Research (CBER) for specifiedbiologic drugs. When combined with man-

The 505(b)(2)aged health care, increased patient access tonew therapies (31), and saving money, these The abbreviated New Drug Application, or

505(j), is a well-established mechanism forevents have created an impetus for faster ap-

922 Robert L. Zeid

FIGURE 1. Multisource biotech development flowchart.

classically synthesized drugs (37), but it is that derived by conventional technology. Thepremise of this policy is that biotechnologynot possible to use it for rDNA-derived prod-

ucts. An enduring FDA policy (5, 36) has has the potential to induce new structuralfeatures in the product, lead to microhetero-required either a BLA or NDA for multi-

source rDNA-derived products, even if the geneity, or introduce new contaminants—each of which could impact the safety, effi-active ingredient/drug product is identical to

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 923

FIGURE 1. Continued.

924 Robert L. Zeid

cacy, or stability of the product. While Title the mainstream 505(j) approach, but thenhow does the regulation allow ‘essentiallyI (of the Hatch–Waxman Amendment) ex-

cludes biologics approved under the PHS Act similar’ products to be approved under theHatch-Waxman Amendments? Enter thefrom a 505(j) process, the 505(b)(2) process

is a reflection of 21 CFR 314.54 that articu- 505(b)(2).Section 505(b)(2) permits approval forlates this regulatory requirement for a prod-

uct different from a listed product. The FDA products other than a duplicate product (eg,not analytically equivalent) and “permits re-policy has consistently interpreted biotech-

nology-derived products as sufficiently dif- liance for such approvals on literature or onan agency finding of safety and/or effective-ferent from a conventional technology-de-

rived product (approved under Section 505 ness for an approved drug product.” A505(b)(2) application is characterized by oneof the FD&C Act) so that a 505(b)(2) process

is mandated. or more investigations “relied upon by theapplicant for approval that were not con-Although there is scientific merit in the

policy, its application is constantly evolving ducted by or for the applicant and for whichthe applicant has not obtained a right of refer-and thus, unpredictable in interpretation—

more a reflection of current technology sup- ence or use from the person by or for whomthe investigations were conducted (21 U.S.C.porting product and process changes. Thus,

some innovator companies might be allowed 355(b)(2).” The October 1999 draft guidance(38) clarified the types of applications cov-significant manufacturing changes under

Prior Approval Supplements (PAS) while ered by Section 505(b)(2) of the FD&C Act,a provision added by the Hatch-Waxmanother firms would need to file an entire NDA

or BLA to demonstrate the safety and effi- Amendments (3). For the first time, the FDAhas clearly articulated a regulatory route forcacy of the same proposed changes. What

product/process features compel a PAS ver- filing multisource biotech products seekingto replace drug products made from conven-sus a NDA or BLA is the $64,000 question.

Clearly, FDA considers the manufacturing tional technology or other biotech-derivedpharmaceuticals—a route that will derive anhistory (eg, process history, specifications,

impurities, etc.) of the innovator company in AB rating for adequate comparisons againstan innovator standard.these case-by-case evaluations, but if the new

process is a radical departure from prior ex- In conjunction with the 505(b)(2) draftguidance, FDA is also working on issuingperience, then what scientific evaluations are

being consistently applied across manufac- PK/PD guidances for insulin and somato-tropin—the two most prominent candidatesturing scenarios? What prior manufacturing

experience could be extrapolated to spare the of the Section 505 arena that would be ripefor multisource competition. These PK/PDinnovator clinical bridging studies? Purely

analytical equivalence? Additional PK/PD guidance documents—which may be issued inthe year 2000—most likely reflect the currentstudies? What enduring features might be ex-

trapolated from the innovator proposal and FDA philosophy for demonstrating therapeuticequivalence as well as considerations forapplied to a comparator firm wanting to do

the same? What differences could allow an safety evaluation of product variation.As encouraging as all of this might seem,acceptable comparability protocol for one

firm but halt a multisource biotech route for there are a number of serious constraints andfurther complications. First and foremost,another? These are vexing questions that rely

on case-by-case analysis, but also beg for the number of multisource biotech productsthat would be available via a 505(b)(2) mech-consistent scientific evaluation for variations

in manufacturing scenarios across suppliers. anism is limited to a small number of biotechproducts currently approved under SectionIt is not so much about what the differences

are but more about what the differences 505, paltry by comparison to the volume ofspecified biotech products approved undermean. These considerations are well outside

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 925

the PHS Act. Given that the regulatory and physicians among the various growth hor-mones?scientific distinctions of drug (FD&C) versus

biologic (PHS) are becoming increasingly What is apparent is that an AB rating isa critical component for some applicationsblurred and that a prime FDAMA directive

is to minimize the differences in evaluation to facilitate easier substitution, but it is clearthat not all factions of the health care industryand approval of drugs versus biologics, the

505(b)(2) application is notably restrictive adhere to those criteria when making a deter-mination of therapeutic equivalence. Withoutsince it only considers a narrow legal defini-

tion of biotech-derived drug product from a clear therapeutic standard designated fromthe FDA, comparator companies may selectHatch-Waxman—a law that is almost 20

years old. There are no obvious comparable a reference standard by its portion of marketshare, not highly conserved physico-chemi-avenues of showing therapeutic equivalence

under a BLA mechanism that would allow cal features that best suit widespread substi-tution. The result may be a conglomerationroutine formulary substitution (eg, AB rat-

ing). Current public health law only makes of various multisource products that are onlyinterchangeable via a complicated algorithm.that inequity all the more apparent. There

will most likely need to be a Congressional This is certainly not consistent with the intentof the Hatch-Waxman Amendments or partremedy to expand the 505(b)(2) concept to

include specified biotech products filed as of the quid pro quo that established the505(b)(2) mechanism for nonduplicate prod-BLAs and approved under the PHS Act.

Second, the AB rating would only be valid ucts in exchange for exclusivity of thoseproducts. Please be aware that a number offor those innovator products that were includ-

ed in a multisource comparator’s 505(b)(2) patent exclusivity issues associated with the505(b)(2) mechanism that may preclude ap-application. If there were already seven ap-

proved innovator sources of somatotropin, proval of the application itself or subsequentapplications also use a 505(b)(2) application.would a multisource firm need to include all

seven products in the PK/PD studies in order Last, the draft guidance on 505(b)(2) ap-plications may need to be expanded to clarifyto achieve the AB rating, even if the seven

approved products are therapeutically equiv- what types of studies may be exempt fromthe IND requirements per Title 21 of thealent already? If yes, few firms would want

to tackle such a massive complicated study, Code of Federal Regulations, Parts 312.2(c)and subsequent references to 21 CFR 320.31,so they would pick some top contenders. This

a la carte approach can complicate routine 314.108(a), 320.38(b), and 320.63. Is a bio-tech-derived product similar enough to aformulary substitution for the following sce-

nario. The innovator product (from Company listed product to enjoy exemptions from INDrequirements and be tested via expanded PK/A) is picked by a multisource firm (Company

B) as the innovator standard. Later in time, PD studies, but yet distinct enough to requirean NDA? If these PK/PD studies include clin-another multisource (Company C) selects the

Company B product as the standard. Thus, ical surrogate markers, is that interpreted asefficacy analysis and thus compels the firmif A = B and B = C, then can A = C under

the AB rating system? Can Company C prod- to perform these studies under an IND? This‘neither fish nor fowl’ standard creates a lotuct be AB-rated to the Company A product

when it never included the Company A stan- of confusion among the various companiesseeking to market their own versions of pre-dard in its application? If so, how can this

be established uniformly across the varying viously approved Section 505 drug products.The sum of these issues and questionsstate formulary criteria when the 505(b)(2)

guidance does not suggest any broader suggest that the FDA may need to start plan-ning what pieces of the Hatch-Waxman legis-consideration? How substantial can this

formulary impediment be when there is al- lation can be incorporated for a multisourcebiotech model that may encompass BLAs—ready widespread therapeutic substitution by

926 Robert L. Zeid

in anticipation of the torrent of company sce- to repeat ‘proof-of-principle’ studies. Whenphysico-chemical, SAR, and characterizationnarios that are likely to unfold. These issues

should include: data show few differences (innovator tomultisource product), there may be savingsin time and money by not repeating extensive1. Selection criteria for an innovator standard

for AB ratings, preclinical studies. There may also be re-duced numbers of patients required for vari-2. Key SAR data that must be met in a

505(b)(2) in order to stay out of IND ous efficacy trials. As time goes on, theremay be greater savings from a number ofstudies,

3. Labeling issues regarding what innovator companies demonstrating this approach. Anumber of the same concepts from thedata a multisource firm can incorporate

into its prescribing information, and 505(b)(2) filing apply here as well; they useextensive physico-chemical characterization4. Types of process changes that require an

NDA or BLA versus a PAS. data in conjunction with literature. The ex-tent of these savings would be a reflection ofmany variables, such as product complexity,Obviously, FDA will take a more holistic

approach rather than simply the comparator’s clinical indication, PK/PD relationship, anddemonstration of safety and efficacy for ob-finished product against an innovator stan-

dard. This need was probably the impetus served differences from innovator standard.Note that any clinical studies performed forbehind creating the FDA’s Complex Drug

Substance Coordinating Committee, which a BLA—unlike an ANDA or 505(b)(2)—would be outside of the exemptions specifiedevaluates issues surrounding complex mole-

cules, particularly the definition of ‘same- in 21 CFR 312.2(b) and would need to beconducted under an Investigational Newness.’ Perhaps this body of regulatory scien-

tists can clarify those differences among the Drug (IND).There have been anecdotal remarks of se-various products that do not impact substitu-

tion or therapeutic equivalence. nior FDA officials alluding to the possibilityof a ‘paper BLA’ (39), but it is not clear ifthis meant:

Biologics License Application

The BLA is more than just a successor to 1. A multisource firm using a BLA to com-pare against an innovator drug approvedthe PLA and ELA—it is a revolutionary step

with its roots in the evolution of the specified under Section 505,2. A parallel mechanism like the 505(b)(2)biotech paradigm (9). Clearly, there are areas

of specified biotech, comparability proto- but used for products approved under thePHS Act, orcols, and multisource biotech that converge

and diverge. It is these areas that will be 3. Expanding the 505(b)(2) process to in-clude both drug and biologic products—evaluated on a case-by-case basis that varies

with product complexity, PK/PD relation- regardless of whether the innovator com-pound was approved under the FD&C Actships, clinical indications, and adequate char-

acterization of differences from innovator or PHS Act.standards.

While there is no such animal as an ‘ab- The last scenario would most likely requireCongressional remedy before it could takebreviated BLA,’ there are tangible compo-

nents of an approved product profile that effect. All of these ‘paper BLA’ scenariosare extremely speculative—certainly notmultisource biotech companies can incorpo-

rate to save time and money in develop- enough to act upon without some concur-rence or clarification from FDA. Note thatment—similar to what is done by an innova-

tor firm for comparability protocols. For while biologics (approved under the PHSAct) are excluded from Title I of the Hatch-example, there may be savings in not having

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 927

Waxman Amendment (3), there is no provi- The regulation that established the CP ap-plication process also established the Euro-sion in the PHS Act that specifically excludes

a multisource biologic mechanism (1). Thus, pean Agency for the Evaluation of MedicinalProducts (EMEA). Within the EMEA, thethere is ongoing legal analysis of what is

supported by current law under either the Committee for Proprietary Medicinal Prod-ucts (CPMP) operates a system of partialPHS or FD&C Acts.

Although the multisource BLA filing will mutual recognition of national authoriza-tions—the multistate process that was super-be a full application, it should incorporate a

body of data that compares its product perfor- seded by the concertation procedure that isnow superseded by the CP. The CPMP hasmance against an external reference standard,

an innovator product, or compendial stan- published a number of guidance documentsregarding biotechnology and assessment ofdard. While the multisource BLA does not

rely on innovator data for approval per se, the comparability by product class or manufac-turing stages (see Table 1).comparisons to innovator physico-chemical

data, bioassay performance, PK/PD study When it comes to preparing the Summaryof Product Characteristics (SPC), there maydata, and surrogate endpoint analysis all lend

a critical reference point to how the compara- be more harmonization in the data dossiersthan is currently seen with drugs. This specu-tor product stacks up when handled and

tested in the comparator’s hands. This type lation is based on the premise that European,Canadian, and Japanese regulatory reviewersof evaluation helps the regulatory reviewers

considerably since they can assess: will probably want to see the same level ofdetailed information submitted to the FDA.While the outcome of the decisions may vary1. A comparator company’s testing of inno-

vator product against the original innova- a lot, the actual multisource dossier might beleading the way toward a harmonized techni-tor application test data (provided they

have access to those innovator application cal document. Simply put, because it is socomplex, everybody will want to see every-data and they are applicable/comparable

methods), and thing.A key consideration in the evolving Euro-2. The comparator performance against an

innovator standard when tested with the pean model is that individual Member Statesmay require varying degrees of demonstrat-comparator methods.ing therapeutic equivalence. Thus, a PK/PDstudy for somatotropin that is acceptable for

Marketing Authorization Applicationsthe Medicines Control Agency (MCA) re-viewers may be unacceptable for the French,The MAA for biotech-derived pharmaceuti-

cals is automatically via the Centralized Pro- Belgium, or German regulatory reviewers.This is to be expected. What needs to becedure (CP). The CP route is also applicable

(upon sponsor request) for new drug sub- considered is that one member’s requirementfor a very high standard of demonstratedstances, new delivery systems, new indi-

cations, biotechnology innovations that are therapeutic equivalence (eg, pivotal efficacystudies) may become the rate-limiting stepsignificantly new, or blood/human plasma

products. As with the United States regula- in approval via the CP route. If the companyacquiesces and performs the trials, it maytory landscape, there are a number of legal,

scientific, and political factors that are at have the precedent of becoming the defaultminimum standard for all products in thatodds with a harmonized European approval

process. It is a balance of individual national class. Without a concerted effort to harmo-nize the comparability protocol standards bydifferences in a review process versus a har-

monized European standard that does not product class, the prevarication of case-by-case could negatively impact the barrier toseek to subjugate approval on the individual

member states. entry. Although some of these differences

928 Robert L. Zeid

TABLE 1CPMP Guidelines on Biotech-derived Pharmaceuticals

CPMP Guideline Description

Production and Quality This guideline outlines the requirements for murine, human,Control of Monoclonal and engineered monoclonal antibodies for therapeuticAntibodies and in vivo diagnostic use. It was originally published as

two separate guidelines: Quality Control of HumanMonoclonal Antibodies and Production and QualityControl of Monoclonal Antibodies of Murine Origin. Notethat Part 15 of the current guideline reviews issuessurrounding product equivalence

Production and Quality This guidance is to facilitate collection and submission ofControl of Medicinal data for MAA within the EEC for polypeptide-basedProducts Derived by products derived by rDNA technology. It concerns theRecombinant DNA application of Part 2, Sections A-E of the Annex toTechnology Directive 75/318/EEC

Production and Quality This guidance is to facilitate collection and submission ofControl of Cytokine data for MAA within the EEC for cytokine productsProducts Derived by derived from biotechnological processes. It concerns theBiotechnological Processes application of Part 2, Sections A-E of the Annex to

Directive 75/318/EECAnalysis of the Expression This guidance is from the ICH Q5 B: Genetic Stability

Construct in Cells Used for guidance that reviews the type of characterization dataProduction of rDNA- for expression constructs to produce rDNA proteins inDerived Protein Products eukaryotic or prokaryotic cell lines

Concept Paper on the This concept paper (published June 1998) notes the needDevelopment of a CPMP for a similar regulatory review mechanism for MAA as isGuideline on Comparability currently considered for BLAs under the comparabilityof Biotechnology-derived protocol. The guideline will focus on demonstratingProducts comparability via SAR, assessing impact by

manufacturing process changes, assessing the nature ofthe changes for possible impact on safety and efficacy,and where possible to establish ‘decision trees’ forassisting in applying a uniform standard for evaluation ofthe changes

may be settled in arbitration, some firms may appeal with supporting data. By anticipatingthese issues and scenarios and workingchoose to avoid this altogether. The net result

is that a de facto “process = product” stan- within the confines of class or product-spe-cific SAR databases, regulators may achievedard would prevail and stifle further innova-

tion as well as cost-competitive features from some profoundly useful tools rather quickly.As industry data builds, these decision treesmultisource biotech firms.

As with the FDA, it may behoove the Eu- can be modified and the resulting compendialmonographs updated too.ropean, Canadian, and Japanese regulatory

communities to anticipate the types of murky While all of this is extremely speculative,it is important to note that European andquestions arising from multisource applica-

tions and try to work with international com- Canadian regulatory reviewers, while not en-cumbered by the semantics of Hatch-Wax-pendial authorities to arrive at critical bodies

of SAR data that would allow decision trees man or United States law, are still fairly con-servative and may actually require a higherto be established. If the multisource firm dis-

agreed with the regulators’ decision, it could standard of demonstrated clinical safety and

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 929

efficacy. Here again, it is also a reflection of doing clinical efficacy studies, which are as-sumed to be more discriminating than analyt-product class complexity and acceptance of

SAR relating to safety and efficacy. ical or pharmacokinetic methods. Point #1automatically invalidates any supportivecomparability data by stipulating a state-of-

DEMONSTRATING THERAPEUTICthe-art standard that is yet to be established.

EQUIVALENCE: OVERVIEW OFOn the basis of this assumption, it makes

POINT AND COUNTERPOINTmoot any data supporting comparabilitysince it questions if the data are accurate orMost would agree that demonstrating thera-

peutic equivalence for biologics is techni- truly reflective of safety and efficacy param-eters. Under this premise, any data thatcally feasible; it is the amount of data that

are required that is subject to a great deal showed ‘no difference’ between innovatorand comparator product may be misleading.of interpretation, speculation, and conten-

tion. Some feel that analytical equivalence The requirement of a clinical trial appears tobe an extension of an established scientific/and bioequivalence should be sufficient—as

is currently done for an ANDA. Others con- regulatory premise that an observation of noeffect is not the same as “proof-of-no-effect.”tend clinical efficacy studies should always

be required, either to assess any undetected Since the current state-of-the-art methodswill always be questionable in giving accu-differences from the analytical and pharma-

cokinetics studies or assess any observed rate data, one will never be able to achievethe holy grail of ‘proof of no clinically signif-differences from the analytical or pharmaco-

kinetics studies—a veritable modern-day icant effect on safety or efficacy’ on the basisof analytical or pharmacokinetic data alone.Hobson’s choice. Obviously, powerful eco-

nomic and political factors are influencing Thus, clinical studies are the only provenmeans of demonstrating comparability. Therethe scientific merits of each side. Perhaps

they are both right—and both wrong. Maybe is a certain circular logic or chasing your tailbuilt into this premise since the state-of-the-we are trying to fit a square peg into a round

hole. Perhaps a deeper analysis of some ma- art standards are constantly evolving and onewill never arrive at a point where they are final.jor points of contention will reveal a sign, a

new direction. The dissonance of this logic is most evi-dent in the fact that current analytical andpharmacokinetic methods—whatever limita-

Point #1: Some Contend that Clinicaltions they may have—are widely used today

Trials Should Always be Required forby innovator firms for comparability testing

Multisource Biotech Because Currentprotocols in lieu of clinical bridging studies.

Analytical Methods andThen how can the same analytical or pharma-

Pharmacokinetic Studies are not Elegantcokinetic tools be valid for the innovator

Enough to Detect all the Possible Subtlecomparability protocols but inherently un-

Differences that Might Impact Safetysuitable as applied to multisource biotech?

or EfficacySome might argue that this is due to distinc-tions between the innovator and comparatorThis argument makes a number of assump-

tions that are either not universally true or databases—the innovator with extensivelinks to clinical safety and efficacy that mightinconsistently applied. The first assumption

is that current methods showing comparabil- allow such an exception—but that still doesnot address the issue of data validity fority are inherently suspect or flawed since

they do not possess the combined prowess either purpose. If current methods yield in-complete data when used by a comparatorto truly show comparability of safety and

efficacy. Something is missing but we just for multisource biotech, would they also beinsufficient for innovator use in comparabil-do not have the tools to see it yet without

930 Robert L. Zeid

ity protocols? If so, why has this issue never may occur in a coincidental fashion to theobserved measurement such as levels thatbeen raised for comparability protocols?

Maybe because it is not an issue, but more occur below the LOD or proteins beyondthe elution profile for the method. Methodof a red herring.

The argument in Point #1 muddies the development is applying the state of the artto achieve a validation of accurate measure-waters with a question of analytical validity

for a multisource biotech purpose but does ments against the unobserved or spuriousfeatures that cannot be consistently measurednot do so for comparability protocols. It is a

question of scientific validity based solely or characterized. It is a balancing act thatdemands constant attention and indeed, moston the intended purpose of the end result. The

resulting confusion stems from a selective methods are dynamic—constantly evolving.Despite these limitations, it is the ob-interpretation or application of the strengths

and weaknesses from both sides of the ar- served measurements that are trended andcorrelated with other features in the develop-gument—a slight-of-hand. Analytical meth-

ods and pharmacokinetic studies have both ment program such as purification, preclini-cal safety, clinical safety, or efficacy. It isstrengths and limitations. The strengths may

be touted to support the comparability proto- the observed changes such as shifts in themicroheterogeneity profile, impurities pro-cols by the innovator industry—validated

links to extensive safety and efficacy data- file, reduced potency, or altered stability pro-files that need to be assessed for impact onbases (which the comparator does not have).

On the other hand, the methods’ limitations safety and efficacy. FDA and other regulatoryreviewers apply their learning curve andand the unknown factors outside those limits

are used to question the validity of the results evaluate the comparator’s methods to ensurethey capture crucial information that givesor safety and efficacy conclusions as it ap-

plies to multisource biotech despite the fact validity to the observed measurements—andthey will do so with innovator method capa-that the data may be used to support identical

process changes (by either the innovator or bility in mind. Thus, features captured in aninnovator method may come to bear in acomparator). It is an interesting mix of le-

gitimate scientific issues and psychologi- comparator program so that over time, fewerand fewer parameters (with impact on safetycal interplay that leads one to doubt one

group’s assertions while believing another— and efficacy) remain unobserved. There mayalso be situations where comparator methodsall based on the same set of analytical/phar-

macokinetic data but cast in a very different are considerably improved over the innova-tor; the innovator may be compelled to de-light.

Analytical methods will always have lim- velop new technology/methods to monitorpreviously unobserved parameters or includeits of utility by virtue of sensitivity, accuracy,

precision, limits of detection (LOD), limits them in comparability protocols.Product development is a learning pro-of quantitation (LOQ), and ruggedness; that

is why validation is such a key requirement cess—one where lots of variables are fleshedout for their impact on yield, purity, stability,in demonstrating these limits for each prod-

uct or application. It is also why ‘state-of- and so forth. Unobserved parameters maynot be realized until there is a loss of correla-the-art’ is a constantly evolving standard. So,

there will always be something that a method tion (eg, process optimization or purity) withobserved measurements. Normally, thesecannot prove today, but perhaps tomorrow.

Drawing attention to that aspect does not ne- variables are identified early in process de-velopment but there are some late bloomersgate its utility or diminish its validity for

observed measurements. On both practical that occur in scale-up to commercial lot sizes.Even in the event of sweeping changes inand theoretical levels it is well accepted that

the analytical method may not measure a cer- unobserved parameters, the combined testingprofile should be sufficiently sensitive andtain protein or that an unobserved feature

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 931

reproducible to detect any impact to observed current analytical and pharmacokinetic meth-ods. Innovator firms have successfully usedparameters and allow a meaningful assess-

ment. Otherwise, the validity of the charac- comparability protocols to support signifi-cant manufacturing changes without repeat-terization testing, release testing, stability,

and comparability protocol parameters come ing extensive clinical trials—and the trendis growing. This indicates an evolving regu-into question. The focus should be on assess-

ments of the impact from both observed and latory mindset that when CMC areas (knownto impact SAR to clinical safety or efficacy)unobserved changes with the current meth-

ods in hand—not a capricious application of are assessed throughout development, SARdatabases can be established to support com-a theoretical analytical or pharmacokinetic

standard. If the nonclinical assessments are parability studies without extensive repeat ofclinical studies. Similar constructs have beeninconclusive, then clinical bridging studies

come into play. applied to classically synthesized drugs, suchas Scale-Up and Post-Approval ChangesThough ‘proof of no effect’ is a valid point

and a laudable goal, it is not mutually exclu- (SUPAC) for solid oral dosage forms, semi-solids, and modified release forms, or Bulksive of the philosophy that FDA and industry

have used to support comparability protocols Actives Post-Approval Changes (BACPAC)for bulk active pharmaceutical ingredients(10)—that critical parameters impacting pu-

rity, potency, safety, and efficacy can be iden- (API).Thus, if an innovator firm is currently al-tified and assessed throughout development

using the analytical tools of today. The focus lowed to support significant postapprovalmanufacturing process changes on the basisfor comparability testing protocols is to as-

sess the impact of the combined observed of controlling an established, albeit limitedbut demonstrated, set of criteria known toand unobserved product changes on an estab-

lished battery of tests—which have already impact product potency, purity, or safety,then how can those same analytical/pharma-been linked to safety and efficacy data pools.

Though every comparability testing protocol cokinetic methods be inherently unsuitablefor a competitor making the same product?is unique to the product/process and evalu-

ated on a case-by-case basis, they all strive Some might say the difference lies in the factthat the innovator has demonstrated thoseto assess whether the modified product char-

acteristics, however subtle or significant they manufacturing changes via extensive links tosafety and efficacy databases; that it is notmay be, will impact clinical safety or effi-

cacy. The innovator must provide sufficient entirely the methods, but the supporting doc-umentation as well. Thus, could Point #1 pre-data to show that its case is an exception to

Point #1—otherwise it may be compelled clude multisource biotech but still allowcomparability protocols for an innovatorto perform clinical bridging studies. Life-

threatening indications most likely demand based upon the links to safety and efficacydatabases? No, because the central issue inmore stringent data requirements. If Point #1

were uniformly true, then the invalidity of Point #1 is whether the quality of the datacan be trusted at all—regardless of its appli-the data would require the innovator firm to

reproduce clinical efficacy studies in support cation. If it cannot, then all the extensivelinks to safety and efficacy do not add to itsof virtually every manufacturing change be-

cause no analytical method/pharmacokinetic validity; it is just a bigger house of cards.There is no argument that the innovator maystudy would be sufficient to reveal a differ-

ence and any observed differences should be have a number of advantages from its data-base, which could support an expedited com-evaluated clinically.

Thus, the scientific premise for Point #1 parability testing study, versus a compara-tor’s comparability study (eg, impuritiesis uniformly modified/negated by a rather

large exception—the reality of FDA/industry qualification). This advantage, however, doesnot negate the utility of the comparator’s dataprecedent with comparability protocols using

932 Robert L. Zeid

when comparability is demonstrated. Again, above, there are many possible scenarioswhere an innovator comparability protocola selective interpretation of Point #1 is based

on an attempt to muddy the waters of data may mirror the same types of manufacturingor process changes proposed by a comparatorquality by its intended use. Regardless of

whether one accepts Point #1 as true or not, firm. In these situations, the assessment ofsupporting data should be done in a consis-it needs to be consistently applied. Using

a “process = product” regulatory review for tent manner—not in an arbitrary fashion tocontrol competition.multisource biotech applications of the same

innovator products that are currently allowedsimilar changes under comparability proto-cols would be inconsistent and scientifically Point #2: Today’s Analytical Methods

are Insufficient to Support Multisourceimpossible to defend.A second assumption of Point #1 is that Biotech on the Basis of Analytical

Equivalence Alone; Therefore,a clinical study is always a more discriminat-ing or definitive measure of safety and effi- a Pharmacokinetic Study is

Always Neededcacy; a clinical study will reveal what com-bined analytical and pharmacokinetic studies

This appears to be a less strident version ofdid not. While clinical studies offer certainthe previous point: that it is mandatory toadvantages over nonclinical programs, theyhave a pharmacokinetic assessment for a bio-also have a number of limitations. Dependinglogic. Here again, this point seeks to applyupon the design and size of the study, it maya higher or different standard for the pur-be difficult to draw any definite conclusionsposes of multisource biotech than may beabout comparability of safety and efficacycurrently applied to a wide variety of otherother than broad, sweeping observations. Forproducts of the same class. For instance, itinstance, it is well accepted that even largeis technically and scientifically possible toclinical programs may be too small or selec-have a complex generic drug approved ontive to identify every possible serious, un-the basis of analytical data alone (eg, paren-identified adverse event that could occur interals). It is also possible to have significanta general population. So, if even the largestmanufacturing changes for specified biotechclinical studies fall short in design or numberproducts approved on the basis of analyticalto allow identification of key safety and effi-data alone, as is already done under compara-cacy features, how can a clinical bridgingbility protocols (10). It really depends on thestudy fill this gap? It cannot and that is whyclass of compounds, the quality of CMC/postmarketing surveillance programs are sodevelopment data, the nature of any differ-critical. Furthermore, a clinical study is onlyences to innovator/reference standard, andas good as the tools used to measure it. If aclinical applications.pharmacokinetic study is flawed in its appli-

Certainly not all biologics could be sub-cation for multisource biotech, then how canject to generic in-roads by analytical datathe same method be reliable when applied inalone, but neither is that the case for a greata clinical study evaluating PK data relatingnumber of classically synthesized drugs.to efficacy?Within the context of the current ANDA drugThe third assumption of Point #1 is thatmodel, an “A” rating (eg, AA, AB, AN, AO,clinical studies should be required for anyAP, AT, etc.) can be assigned:multisource biotech initiative, regardless of

product class or therapeutic indication. Thisgeneralization suggests a standard based 1. Solely on the basis of analytical equiva-

lence (eg, parenterals, ophthalmics, etc.),more on the intended use of the data ratherthan a circumspect consideration of product 2. Analytical equivalence + bioequivalence

(eg, solid oral dosage forms), orclass complexity, clinical indications, manu-facturing issues, and so forth. As discussed 3. Analytical equivalence + clinical efficacy

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 933

studies against an innovator standard (eg, Since this is still an evolving area, one needsto evaluate:anti-fungal cream) (37).

In the classic drug realm, there is prece- 1. The necessity of doing a PK/PD study,dent and scientific interpretation for at least 2. The value of the data from a current drugthree different levels of establishing thera- study design versus a biotherapeutic proto-peutic equivalence, so it would seem unlikely col, anda single standard could be applied to biolog- 3. The inherent flaws in either approach thatics considering the breadth of complexity. might mitigate additional clinical studiesJust because this analytical equivalence stan- if those data were inconclusive.dard may not be applicable for all biologicsdoes not mean that it is impossible to use A hasty approach might leave a developmentfor any biotech product. Indeed, as analytical team with an invalid study design and wastedmethods improve, there may well be large data (along with squandered time andnumbers of multisource biologic classes on money).the basis of analytical equivalence alone—ashas already been done with antibiotics. Al-beit, antibiotics were reviewed and approved Point #3: Differences in the Purity andunder Section 507 of the FD&C Act (Certifi- Impurity Profiles Between Innovatorcation of Antibiotics) until FDAMA, but the and Comparator Products Preventcharacterization and lot release criteria bear Multisource Biotech Since Thesemany parallels with biologics. (Differences) Could be a Safety Risk if

Another issue that comes up is the relative Not Properly Assessed via Preclinicalvalue of assessing biologics using a drug Studies and/or Clinical Safety Studiespharmacokinetic model. Currently, the com-

Point #3 asserts correctly that purity and im-parability guidance notes that PK/PD studiespurity profiles must be evaluated for impactshould be performed as a complement to theon safety or efficacy, but this should occuranalytical characterization data, but there areregardless of whether it is innovator or com-limitations to the utility of the drug modelparator product. This point, however, is link-for some biologics. This is particularly theing a rigorous testing standard to a premisecase for some biologics where pharmacody-that multisource biotech is only possible bynamic effects (eg, immunomodulation) aremeeting an analytical equivalence standardlong lasting and durable long after the phar-of the innovator versus comparator prod-macokinetics phase of dosing. So it is notuct—a definition of ‘sameness’ that is onlyclear that performing or meeting the require-one of many working models used by regula-ments for a standard PK/PD study (random-tory reviewers and industry. Although thisized, two-way crossover) is a definitiveargument raises legitimate scientific and safetymeasure, especially when there may be im-issues about those differences, it makes an in-munogenicity or persistent pharmacologicaltractable conclusion that extensive testing isactivity that are not assessed by current pro-required for multisource biotech when theretocol designs. Depending upon the productis ample precedent otherwise, particularlyclass and known interactions, toxicities, andwhen tiered testing approaches are widelyspecial considerations a PK/PD study for aused by the innovator industry. This criticalbiotherapeutic may be a very different designarea needs to be fleshed out in terms of:than those seen for classic drug comparisons.

This has been the focus of recent industry/FDA meetings (40,41), to identify issues sur- 1. Definitions of ‘sameness’ (eg, analytical

equivalence versus therapeutic equiva-rounding PK/PD studies for biotherapeuticsand alternative designs for assessing site-spe- lence),

2. Considerations for each level of tiered test-cific activity via in vitro or ex vivo methods.

934 Robert L. Zeid

ing to evaluate purity and impurity pro- trials, regardless of the fact that they are notanalytical equivalents. Some ANDAs forfiles, and

3. Considerations for the differences in data- complex substances have been shown to betherapeutically equivalent despite physico-bases available to innovator or comparator

when evaluating risks that impact on chemical differences from the innovator ref-erence standard (eg, menotropins), expand-points #1 and #2.ing FD&C case law on FDA’s interpretationsof sameness. Based on this precedent, itWith regards to using only an analytical

equivalence definition for sameness, one ob- seems the working definition of ‘sameness’has been amply broadened to include bothvious problem might be establishing a uni-

form application of ‘different.’ This could analytical differences as well as therapeuticequivalence. The degree to which a sponsorquickly become problematic since it would

most likely vary according to reviewer, prod- must show this—the amount of data re-quired—is a regulatory reviewer judgmentuct class, and perhaps FDA division. This

very issue has plagued many ANDAs for call that will consider the sponsor’s existingCMC data for SAR, sensitivity of testing cri-classically synthesized drugs (eg, qualifying

a second supplier for marketed product) so teria, clinical indication, duration of dosing,demonstrated safety/efficacy in animal mod-one can only imagine the complexity sur-

rounding its application to biotech. For in- els or clinical studies, and so forth.With regards to the second point, tieredstance, is a two-fold increase of a previously

observed impurity—but still well below testing based on existing databases, this isan area where the innovator has a clear ad-0.5%—different or the same? What about a

10% increase in aggregated dimer compared vantage over a comparator. The innovator ismore likely to have a substantial databaseto innovator on stability? What about an in-

crease in electrophoretic (eg, SDS PAGE) linking observed impurity and purity profilesto preclinical and clinical studies, thus hav-densitometry reading of an impurity from

0.1% to 0.5%? What about an 18% increase ing qualified their safety and potency. De-pending upon how many different lots werein a variant form of peptide that does not

impact the critical SAR for potency? One can used in clinical trials, preclinical studies, andmarketed product, the clinical support forquickly digress from a clinically significant

change to artifactual differences, assay vari- duration of impurity profile exposure/qualifi-cation could be enormous (eg, products usedability, or even manufacturing variability.

In actuality, the criteria for ‘sameness’ in long-term chronic disorders such as formultiple sclerosis, cancer, or AIDS) or rela-does not lie solely in analytical equivalence

but in demonstrating therapeutic equivalence tively abbreviated (eg, products for emer-gency use/acute trauma such as for acuteto a standard battery of tests that have been

shown to be sensitive enough to detect subtle myocardial infarction [AMI], stroke, or sep-sis). Long-term clinical qualification of adifferences in SAR and have some value in

predicting impact on safety and efficacy. For wide spectrum of purity and impurity profileswill go a long way toward supporting compa-example, it is accepted that some product

microheterogeneity differences may have lit- rability protocol testing without extensivepreclinical or clinical testing.tle or no safety/potency impact; this has been

shown without extensive preclinical or clini- If a comparator’s product has a purity andimpurity profile similar to the innovator, thecal testing. Comparability protocols have al-

ready forged a broader, more circumspect regulatory bodies may not compel extensivetesting since they already know these impuri-definition of therapeutic equivalence even in

the light of significant analytical differences. ties have been clinically qualified (eg, com-parison to a compendial standard). That too,Many marketed versions of biotech products

have been demonstrated to be therapeutically however, is conditional on factors such ascomparability of profiles and quantities ofequivalent to the material used in clinical

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 935

differences, immunogenicity of the innova- sume a shift in the heterogeneity profile orimpurity profile automatically means a nega-tor’s product, toxicity associated with certain

impurities, and so forth. If so, regulatory au- tive impact on safety or efficacy. There aremany examples of product changes with sig-thorities may require a comparator to do ex-

tended safety testing, which could be in vitro, nificant shifts without any impact on potencyor safety as amply shown via comparabilityex vivo, or preclinical studies, in addition to

any human studies. If a comparator’s product protocols. The manufacturer needs to assessthe heterogeneity profile throughout devel-had unique impurities that had never been

seen before in the innovator profile, then reg- opment to assure that it is consistent withthe product used in preclinical and clinicalulators might require a tiered testing plan

of preclinical and/or clinical. The extent of studies (eg, PK/PD or surrogate endpoint ef-ficacy studies). When heterogeneity profilestesting would depend on the class of biologic,

the type of impurity and quantity, the spon- change from material used in preclinical orclinical studies, the significance of thesor’s SAR database, and other product-spe-

cific features (eg, dosing). change should be evaluated, which may beaccomplished via a tiered approach of phys-Regulatory guidances note that the extent

of purification of rDNA-derived products ico-chemical analyses, bioassays, and/or pre-clinical studies (18). If that is insufficient toshould be consistent with the intended use

of the product. Drugs and biologics, which make a determination, then clinical safetystudies may be required.may be administered repeatedly or at high

concentrations, should be adequately pure to For a comparator product that did not pos-sess the extensive database of an innovator,prevent development of immune or toxic re-

actions (18). If the sponsor has sufficient the relative comparison to the innovator prod-uct would be the starting point. When thosephysico-chemical characterization data, it

may be able to avoid a preclinical (toxicity) product profiles differ, what supporting in-formation does the comparator have to assessstudy. It will always be evaluated on a case-

by-case basis but hopefully as product class the safety and efficacy impact of differentanalytical and pharmacokinetic profiles? Isand SAR data grow, there will be greater

consistency for all biotech-derived products the elegance and sensitivity of the methodssufficient to make a valid comparison? If not,in that class.

With regards to the final point—differ- then a clinical evaluation of the differenceswould be warranted but would it have to beences in risk assessment based on the SAR

databases available to the innovator or com- a classically defined pivotal efficacy study?Perhaps it would not if validated surrogateparator—this is case-specific and reflects a

number of features already discussed. Ana- endpoints can be used.It is obvious that these issues are germanelytical differences will be heavily scruti-

nized. There has always been considerable to both multisource biotech and innovator tosupport manufacturing changes/product pro-focus on differences in innovator and compa-

rator impurity profiles for drugs, so there is file differences in an international market.Following postapproval changes, a likelyno reason to assume it would be fundamen-

tally different for biotech-derived products. scenario could be that the tiered testing thatsufficed for one country’s regulatory authori-It is also recognized that biologics are inher-

ently complex and there is often tremendous ties is inadequate for another. Pointing to anexisting clinical database may not be suffi-heterogeneity (eg, isoforms, glycoforms) that

cannot be controlled. The analytical differ- cient.What if all of the nonclinical testing crite-ences need support from existing proprietary

or public SAR databases that they are clini- ria were not sensitive enough to truly predictthe consequences? Would significant healthcally qualified or pose minimal/no risk.

What if those SAR data are not available? risks be missed if the varying impurity pro-files were not clinically qualified first beforeIn terms of calculating risk, one cannot as-

936 Robert L. Zeid

going to market? Obviously yes, but in one In general, toxicological potential couldbe seen as a compound that:instance—antibiotics—the answer appears

no. And the supporting data are about 50years in the making. 1. Antagonizes the desired therapeutic bene-

fit directly or indirectly,The development and marketing of antibi-otics after World War II, under Section 507 2. Prolongs a toxic metabolite from being

excreted,of the FD&C Act, was considerably lax com-pared to current standards. Antibiotics were 3. Potentiates the desired pharmacological

effect into a toxic effect, orassessed primarily by potency standards. Itis not clear from the historical record of lot- 4. Has some unrelated toxic effect through

synergism with another body/enzymeto-lot certification requirements for antibiot-ics if impurities were part of the evaluation, system.monitored, or even controlled. Even if theywere, what is the probability that each mar- Chances are if the impurity does not possess

biological activity in the sponsor’s screeningketed version of penicillin, streptomycin,chlortetracycline, chloramphenicol, bacitra- assays or characterization/bioassay test-

ing—or general biochemical toxicity assayscin, cefazolin, and so forth for the past 30years have had identical or even similar im- (eg, tissue culture)—then its likelihood of

systemic toxicity becomes even more re-purity profiles compared to the competitorproducts in the same compendial class, espe- mote. It is not out of the realm, but the risk is

considerably reduced. Note that while manycially, given the scale-up and process changesthat must have occurred before SUPAC was proteins are antigenic and elicit an immune

response, the clinical significance of theseeven a glimmer in the FDA’s eye? The oddsof this occurring—this level of serendip- observations is still under investigation. Im-

munogenicity is not inherently toxic but thereity—would be rather extraordinary, to saythe least. is the concern of long-term sensitization to

all products within a certain class or the de-Assuming there was considerable ‘hetero-geneity’ in the purification schemes/impurity velopment of neutralizing antibodies that

could nullify or attenuate the clinical benefitprofiles, what health risks have emerged aftergenerations of patients taking antibiotics— of the drug product (40,41).

One cannot throw caution to the windvia oral, IV, IM, IP, rectal and vaginalsuppository, otic, and ophthalmic prepara- about qualification of new or unusual impu-

rity profiles, but the litmus test for determin-tions—some quantities in excess of 1 gram/day—from dozens of different manufactur- ing additional preclinical studies must be

evaluated in a circumspect light of the ob-ers—all with little or no characterization ofthe impurity profiles in these rather complex served quantities (of impurity), pharmaco-

logical profile in screening or bioassays, ana-structures? Is there some correlative evalua-tion to show toxicity of the varying impuri- lytical sensitivity, and proposed clinical

indication. Otherwise, one might be alarmistties? Are there any data, anecdotal or other-wise, that one can draw upon to help identify in challenging minute differences in impurity

profiles without supporting data to indicateor target possible clinical safety issues fornew and emerging biologics? If not, then there is a risk. For instance, how many drug

manufacturers would recall marketed prod-what does that suggest about the impact ofimpurities on clinical safety and efficacy for uct based on a new impurity appearing, while

on stability, but well below the regulatoryantibiotics? The meta-message from the anti-biotic saga may be that patients are probably specification for ‘unknown’ impurities? This

fine balance of scientifically-supported bene-much more resilient in dealing with impuri-ties than we give them credit for even patients fits versus calculated risk is a laudable goal.

What remains to be seen is if it can be appliedwho are given large quantities of antibioticsin serious or critically ill conditions. in a uniform fashion across the breadth of

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 937

biotech products. Perhaps the only way that changes. It just depends. The greater the dif-ferences, the greater the effort to show com-will come about is via the development of

SAR profiles—and impurity profiles—for parability.While there may be some advantages toclasses of biotech products.

An epilogue to the antibiotic saga is that producing multisource biotech by the sameprocess, it is by no means the only way tomost current compendial monographs (for

antibiotics) are notably missing impurity pro- technically derive a therapeutically equiva-lent product. Considering the explosion offile specifications. The reason is not clear

but appears to be a combination of historical biotechnology and genetic engineering, someapproved product processes and methodsprecedent and clinical experience. One in-

dustry colleague noted the FDA’s approach may not be the safest, most cost-effective, ortechnically efficient means of resulting in theon this has been primarily historical; antibiot-

ics have always been poorly characterized. same product. The same is true for analyticalmethods and, of course, the resulting dataThe winds of change, however, are stirring.

Some proposed changes to antibiotic mono- and specifications. Different manufacturingapproaches with somatotropins and othergraphs (eg, Pharmacopeal Forum) include

impurity assessments for both quantity and specified biotech products, as well as com-plex drug substances, have demonstrated thatprofile. In recent years, FDA has required

manufacturers of antibiotic drug substances a number of different manufacturing scenar-ios can derive therapeutically equivalentand drug products to perform impurity analy-

ses and start trending them in stability stud- products. Indeed, this was one of the compel-ling arguments presented to FDA that gaveies. These initiatives, however, are slow to

come about. It would appear that the issue impetus to the notion of a well-characterizedbiologic. When there is a compendial stan-of patient safety from unqualified impurities

in antibiotics seems far less compelling to dard available (eg, human insulin), this argu-ment becomes even less scientifically defen-some parts of the industry than those of

multisource biotech. sible. Although the FDA precedent withcomparability protocols and innovator manu-facturing changes already negates Point #4,

Point #4: Multisource Biotech isit bears closer examination.

Virtually Impossible Without theThe logic in Point #4 hopes to mandate a

Innovator and Comparator Usinglink between a multisource biotech process

Identical Processes, Analytical Methods,and an innovator’s (current proprietary) pro-

and Finished Product Releasecess by challenging the comparability of a

Specifications. The Resulting Productcompetitor’s product that was derived from

from Such Varied Processes Could Nota different route and thus, compel the multi-

be Accurately Comparedsource biotech to use the same methods andlimit its access to market. It is an interestingThe point that any change in process, meth-

ods, or specifications would preclude multi- premise that links legitimate scientific ques-tions about any product differences with thesource biotech seems especially ironic since

innovator manufacturers have successfully assumption that a competitor’s product ismanufactured in such an inherently differentsupported an opposite rationale for their con-

tinual process improvements since 1982. In- manner that they cannot be adequately com-pared as therapeutically equivalent. Candeed, allowing process and product change

without repeating clinical studies was the im- there be product differences arising from thetwo processes that cannot be detected by cur-petus driving the specified biotech paradigm.

As with so many case-by-case scenarios, not rent analytical methods or PK evaluationsbut still have an impact on safety and effi-all changes are equal. A switch in master cell

bank (MCB) lines is much more sweeping cacy? At the outset of the premise, it makesmoot the reliability of any data that wouldthan some downstream purification process

938 Robert L. Zeid

be favorable to a competitor’s product—and complete characterization testing at everystage and repeat clinical efficacy studies forit does so on the basis of questioning the

validity of current scientific criteria by in- even modest changes. Even then, there wouldnot be any confidence that the data actuallyvoking a new unknown standard beyond to-

day’s state-of-the-art technology. supported the safety and efficacy observa-tions, since those evaluation standards wouldPoint #4 uses a slight variation of the

themes from Points #1 and #2. The notion is be insufficient too. Obviously, this is an un-tenable premise for any drug development.that an innovator’s product safety and effi-

cacy stem from its SAR development data- Second, absolute physico-chemical com-parability is not a necessity for therapeuticbase, which is a reflection of the process and

controls. This much is true. An innovator equivalence, regardless of whether materialwas made from the same versus differentmay claim that though a competitor’s product

may be the same end-result, as shown by processes. The concept of microheteroge-neity in biologics is well established andphysico-chemical testing, PK, surrogate end-

points, and so forth, it could lack therapeutic thus, several forms of the same enzyme (eg,isoforms, glycoforms, etc.) may be intention-equivalence to the innovator since it is not

performed by the same process, analytical ally equal in activity. Some have postulatedthat heterogeneity in biological systems ismethods, and so forth. The comparison argu-

ment suggests there may be undetected dif- a reflection of the evolutionary process—apositive selection factor—that favored or-ferences in impurity profiles, heterogeneity,

or bioassay comparisons that cannot be ob- ganisms with redundancies in enzymatic sys-tems. Thus, genetic defects or mutations thatserved yet because the analytical tools are

insufficient to discern these subtle differ- resulted in the loss of an isoform would notbe lethal since there were other isoforms thatences. Although the products are assumed

to be therapeutically equivalent, it is only were interchangeable. Organisms that devel-oped more redundancies in systems survivedbecause we do not have the tools yet to show

the differences are there—differences that the selection pressures over those that hadsingle or limited pathways.could impact safety and efficacy. It is another

level of a Hobson’s choice, but is not a viable Finally, if a competitor cannot justify theuse of a different process from an innovatorpremise for several reasons already covered

under the Point #2 discussion. on the basis of extensive characterizationdata, then how is it scientifically justifiedThe conjecture forming the premise of

Point #4, that certain product differences ex- for an innovator to use significantly differentprocesses for clinical trial material versusist which currently cannot be measured (to

prove or disprove the question of compara- commercial scale on the basis of limited SARcomparability protocol data? If product com-bility) could impact safety and efficacy in

ways that also cannot be distinguished by parability cannot be demonstrated satisfacto-rily by today’s analytical, pharmacokinetic,current methods—cannot be proven either

way. It offers no measurable means of solv- and surrogate endpoint equivalence for acompetitor, then how could the innovator ac-ing the issue one way or the other, since it

automatically questions any data demonstrat- complish this for its comparability protocols?The conclusion is an obvious double stan-ing therapeutic comparability at the outset.

First and foremost, if Point #4 were valid, dard. It suggests that the “product = process”dogma is appropriate when applied to a com-then comparability protocols would be mutu-

ally exclusive and neither the innovator nor petitor’s processes/methods wishing to de-velop a multisource version, but still allowscompetitor would have analytical tools or

surrogate endpoints that could meet the stan- an innovator use of comparability protocolsto achieve the same effect in-house (eg, pro-dard, which has yet to be defined, that could

support any development changes. On this cess improvements, cost-effective scale-upmeasures, etc.).basis, an innovator would have to perform

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 939

The intent of Point #4 is obvious. The physico-chemical differences from previousiterations of preclinical and clinical materialmore a multisource biotech process is linked

to a proprietary process, the harder it is to there are, the more extensive repeats of keystudies will be required.manufacture any multisource biotech product

since any product outside of the “equivalent Third, biotech encompasses a wide vari-ety of compounds ranging from simple pep-process” would not be approvable. In any

multisource scenario, however, a competi- tides to complex proteins requiring posttrans-lational modifications to conjugated vaccinestor’s product differences will be evaluated

against the innovator/reference standard. It and so forth. It would be impossible to gener-alize a product development strategy thatis up to the sponsor to demonstrate to regula-

tory authorities that those differences do not could apply to all, as it would be for diversityseen with classically synthesized drugs. Theimpact safety or efficacy. Each product’s dif-

ferences are decided on a case-by-case basis. realm of multisource biotech is so extraordi-narily complex that possibly the only way toSome products are therapeutically equivalent

even in the light of major differences, let approach it is via one product class at a time.Key issues for somatotropins may not applyalone major process differences.to tissue plasminogen activator, hepatitis vac-cine, or insulin. By restricting focus to spe-

A Pause at the Crossroadscific product categories one can scrutinizecritical CMC issues that impact sensitiveThough there are at least a dozen more dis-

cussion points, these four issues have fleshed SAR for clinical safety and efficacy. Bybuilding on the foundation of the existingout some rather intriguing concepts. First and

foremost, it is apparent that demonstrating knowledge from the innovator database andgeneral publications, regulatory officials willtherapeutic equivalence of biotech-derived

products will entail an exhaustive multi-dis- have a better standard to apply toward anyinnovator comparability protocols—and multi-ciplined approach to supporting SAR links

with preclinical safety, clinical safety and ef- source biotech.Finally, this is an emerging area for prod-ficacy, and CMC changes throughout the de-

velopment program. Analytical methods will uct development and there are likely to bemore failures than successes. It would bebe key windows to the process that will vali-

date these SAR links. prudent to reserve judgment of the feasibilityof some aspects until there are enough dataSecond, there is much to be done to estab-

lish decision trees that will allow a more to reach a reasonable conclusion.The following sections detail technical as-apparent and consistent review process for

regulatory reviewers. These decision trees pects of analytical equivalence, structure ac-tivity relationships, pharmacokinetics, andwill be a useful tool to help distinguish those

scientific nuances that may give innovators surrogate endpoints.latitude in their comparability protocol test-ing versus the kind of supporting data re-quired from a multisource biotech company ANALYTICAL EQUIVALENCEtrying to do the same task. While the innova- OVERVIEWtor has considerably more latitude for someof these manufacturing changes—based on Bioanalytical characterization testing is com-

plex, fraught with dead ends, and furtherextensive SAR databases and impurity quali-fication in preclinical/clinical studies—the compounded by unrealistic deadlines of fast-

moving projects. First, there are the technicalcomparator is not far behind. The closer thecomparator product is to the innovator’s ref- difficulties of method development and vali-

dation followed by data interpretation oference standard, the more the comparator cansupport its program without extensive repeat anomalous findings or spurious results. Sec-

ond, there are many levels of analytical sup-of those same studies. Obviously, the more

940 Robert L. Zeid

port needed for the various programs—all Master Cell Bank (MCB) andManufacturers’ Working Celloperating at different speeds and with few

people keeping tabs on all of it until the bitter Bank (MWCB)end. Finally, the momentum of a fast-movingproject may leave precious little time to dwell The creation and characterization of a MCB

and a MWCB is one of the most fundamentalon critical areas that seriously impact CMC,preclinical, or clinical areas. The result can distinctions between drugs and biologics

manufacturing. Although there are many as-be chaotic and the final days of preparing fora submission may resemble a Godzilla movie pects that classic drug development people

are not familiar with, there are a great manywhere the scientists are racing against timefor an answer. In the case of drug develop- parallels to the classic modern-day API drug

manufacturer. Instead of a 100000 squarement, Godzilla may often come disguised assenior management. Nonetheless, the analyt- foot facility, the biotech plant is more on the

order of 100–300 microns wide. The type ofical program can be subdivided into majorareas (see Table 2). separation of activity called for in cGMPs is

TABLE 2Analytical Characterization Programs

Master Cell Bank (MCB) PlasmidManufacturer’s Working Cell Bank (MWCB) Host Cell System

VectorExpression SystemEnd-of-Production Cells (EPC)Late-Extended Cells (LEC)

Process Development: In-Process Control Lab ScaleTesting Pilot Scale

Commercial ScaleComparability of Process Changes

Bulk API Characterization TestingRelease TestingComparison to Innovator Material

Finished Product Characterization TestingRelease TestingComparison to Innovator Material

Stability Program Characterization TestingComparison to Innovator Material

Bioassay Method Development & Validation SAR Links to Bioassay ActivityComparison to Innovator Material

Analytical Methods Development & Purity AssaysValidation Potency Assays

Impurity AssaysStability-indicating AssaysSafety Assays

Structure-Activity Relationship (SAR) Links to Manufacturing Changes & ProcessDatabase and ControlsQuantitative SAR Links to Bulk API and Finished Product

Links to Stability ProgramLinks to Bioassay TestingLinks to PK/PD DataLinks to Surrogate Endpoints

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 941

also observed in the biotech plant, albeit on early on in a host system adapted for suspen-sion (eg, adsorbed microbeads, etc.) rathera much smaller scale. The documentation as-

sociated with drug substance raw materials than roller bottles and then change duringscale-up. It should be a holistic approach.qualification and use in production is similar

to biotech, although it tends to be focused Another critical aspect to note is that dif-ferences in multisource biotech MCB or cellon plasmid, vector, and host cell systems as

they are combined to create a single expres- lines (versus the innovator MCB) will oftenhave profound impact on product character-sion system. As with any production process,

there are limits to duration, speed, and other istics—more so than a lot of differences indownstream purification procedures. Therefactors which impact on the resulting product

quality—aspects fleshed out in validation may, however, be compelling reasons to havethese differences such as easier purification,studies. Similarly for biotech, the end-of-pro-

duction cell (EPC) and late-extended cell greater production in a certain cell line,avoidance of patent infringement, and so(LEC) characterization profile lend credence

to a validated number of generations that will forth. Whatever the differences are, thesewill be scrutinized by regulatory reviewersreproducibly create the protein/peptide of

choice within certain parameters. Superim- for their potential impact on product compa-rability.posed on these microscopic production is-

sues are many parallels of routine cGMPfacility operation and qualification of equip-

Process Development: In-Processment, personnel, and processes.

Control, Bulk API, and FinishedA number of FDA and ICH guidance doc-

Product Testinguments (7, 14, 16, 20, 22, 28, 29) outline cellline characterization issues and are briefly Analytical support for lab scale, pilot scale,

and commercial scale process developmentsummarized here (see Table 3). Key charac-terization areas include genetic (plasmid) sta- will be extensive since there will be a number

of comparability studies for various changes.bility in EPC and LEC stages as well as theresulting protein from those extended growth The time for evaluating these differences will

reflect the complexity of the product charac-phases. The expression system insertion andcopy number need to be assessed (44). Multi- teristics, differences from innovator product,

and SAR links supporting the significanceple integration events scattered about the hostgenome can result in the tandem insertion of of the changes.

Since demonstrating physico-chemicalrepeated copies at a single locus. There maybe plasmid loss or rearrangement in high equivalence across process scale will be done

on a product class-specific basis (eg, vac-yielding strains. The combined data will con-firm that a commercial process that typically cines, monoclonal antibodies, etc.), a good

starting point is to review the existing FDAruns for 30 to 40 generations (EPC) will pro-duce protein per critical parameters. The and ICH guidance documents on CMC issues

for the particular product class (see Table 4).LEC stage confirms that the process is notoperating near an edge of failure, which A number of these CMC guidance docu-

ments (20–29) allude to structure-activity ormight be observed with mutations or rever-sion at 50 generations. If this is the case, potency/purity issues surrounding certain

CMC processes. Pay particular attention toit will be heavily scrutinized and may needadditional testing or possibly regenerating of these areas. A second trove of analytical test-

ing can be found in the public archives onthe Master Cell Bank. The best approach isto work closely with an experienced team in related topics, which will include Freedom

of Information (FOI), European Public As-developing and characterizing cell lines withthe proposed commercial scale of manufac- sessment Report (EPAR) summaries of prod-

uct approvals in both the United States andture in mind. For instance, if using a tissueculture cell line, develop the MCB cell lines Europe, and general scientific articles. Last

942 Robert L. Zeid

TABLE 3Characterization Testing of the Host Cell System, Plasmid, Vector,

and Expression System

CMC Information for Host CellsA Therapeutic A description of the source, relevant phenotype, and genotyperDNA-Derived should be provided for the host cell used to construct theProduct or a biological production system. The results of the characterizationMonoclonal of the host cell for phenotypic and genotypic markers, includingAntibody Product those that will be monitored for cell stability, purity, and selectionFor In Vivo Use should be included(August 1996) Gene Construct (Plasmid)

A detailed description of the gene that was introduced into thehost cells, including both the cell type and origin of the sourcematerial, should be provided. A description of the method(s)used to prepare the gene construct and a restriction enzymedigestion map of the construct should be included. The completenucleotide sequence of the coding region and regulatoryelements of the expression construct, with translated amino acidsequence, should be provided, including annotation designatingall important sequence features

VectorDetailed information regarding the vector and genetic elements

should be provided, including a description of the source andfunction of the component parts of the vector, eg, origins ofreplication, antibiotic resistance genes, promoters, enhancers. Arestriction enzyme digestion map indicating at least those sitesused in construction of the vector should be provided. Thegenetic markers critical for the characterization of the productioncells should be indicated

Final Gene Construct (Expression System)A detailed description should be provided of the cloning process

that resulted in the final recombinant gene construct. Theinformation should include a step-by-step description of theassembly of the gene fragments and vector or other geneticelements to form the final gene construct. A restriction enzymedigestion map indicating at least those sites used in constructionof the final product construct should be provided

Cloning and Establishment of the Recombinant Cell LinesDepending on the methods to be utilized to transfer a final gene

construct or isolated gene fragments into its host, themechanism of transfer, copy number, and the physical state ofthe final construct inside the host cell (ie, integrated orextrachromosomal), should be provided. In addition, theamplification of the gene construct, if applicable, selection of therecombinant cell clone, and establishment of the seed should becompletely described

Monoclonal AntibodiesA detailed description of the development of the monoclonal

antibody should be provided including characterization of theparent cells, donor history for human cells, immunogen,immortalization procedures, screening, and cell cloningprocedures

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 943

TABLE 3Continued

Master Cell Bank (MCB)A detailed description of the preparation and testing of the MCB,

as outlined below and in the ICH guideline “Analysis of theExpression Construct in Cells used for Production of R-DNADerived Protein Products,” should be submitted. The MCBshould be described in detail, including methods, reagents andmedia used, date of creation, quantity of the cell bank, in-process controls, and storage conditions. The results of thecharacterization of the MCB for identity and purity usingappropriate phenotypic markers such as morphology,auxotrophy, isoenzyme, etc. should be included. Restrictionenzyme analysis and DNA sequencing data supporting theintegrity of the introduced genetic sequence and data supportingthe stability of both the host cell and final gene construct duringstorage should also be submitted. For bacterial cells, the resultsof tests for contamination with both lytic and lysogenicbacteriophages and non-host microorganism(s) should beincluded

Working Cell BankA detailed description of the preparation and testing of the WCB

such as those outlined in the applicable guidance documentsshould be submitted. The production of the Working Cell Bankshould be described in detail, including methods, reagents andmedia used, date of creation, quantity of the cell bank, numberof cell doublings from the MCB and storage conditions. If thereis no MCB, the results of the characterization of the WCB shouldbe provided in the format detailed for the MCB

End of Production Cells (EPC)A detailed description of the characterization of the EPC that

demonstrates that the biological production system is consistentduring growth should be provided. The results of the analysis ofthe EPC for phenotypic or genotypic markers to confirm identityand purity should be included. This section should also containthe results of testing supporting the freedom of the EPC fromcontamination by adventitious agents. The results of restrictionenzyme analysis of the gene constructs in the EPC should besubmitted. Further guidance can be obtained from the ICHdocument, on “Analysis of the Expression Construct in CellsUsed for Production of R-DNA Derived Protein Products” and“Points to Consider in the Manufacture and Testing ofMonoclonal Antibody Products for Human Use”

but not least, contact the regulatory reviewers as in-process control (IPC) stages. The ana-lytical links provide windows to the entirefor early input. If there are colleagues famil-

iar with these product lines, contact them as process so that when any process changesoccur, one can examine the immediate stagewell. Search out people or facilities familiar

with testing these products. Weave that input and downstream stages with a minimum ofdisruption to finished product testing. Thisearly on into the program.

Analytical testing may start with the fin- line of logic is consistent with other SUPACand comparability guidance for assessingished product but it should not end there.

There needs to be additional testing/methods manufacturing changes (10, 18, 19). It is alsocritical to be able to identify/isolate impuri-qualification for bulk drug substance as well

944 Robert L. Zeid

TABLE 4Overview of CMC Guidance for Biotech-derived Products: Characterization Testing

CMC Format & Characterization/Proof of StructureContent for The synthetic peptide should be characterized by, at a minimum,Synthetic Peptide amino acid analysis, MS, and peptide sequencing. Amino acidSubstances analysis of the purified peptide provides information on whether

the peptide has the correct composition. Mass spectroscopybased on a number of techniques, such as fast atombombardment, electrospray, plasma desorption, or laserdesorption, is frequently used to provide the molecular weight orsequence information.

Edman degradation can be used for the determination of thecorrect sequence of the synthetic peptide. If the N-terminus ofthe synthetic peptide is blocked, the N-terminal moiety should beremoved by an appropriate method prior to Edman degradation.If a C-terminal modification, eg, amidation, arises from amodification of the synthetic resin (see section III.A.2.b), thisneed not be characterized in the final peptide.

If the peptide contains more than 20 amino acid residues, apeptide map should be provided. The fragments should beisolated and characterized by either MS or peptide sequencing.When unnatural amino acids (ie, besides the generally accepted20 “naturally” occurring amino acids) are used in synthesis, theirchromatographic behavior (for amino acid analysis) and that oftheir derivatives (for peptide sequencing) should be establishedto verify the presence of these amino acids in the syntheticpeptide.

Disulfide bond(s), if present in a peptide, should be determined byreaction with a suitable reducing agent, such as dithiothreitol.The free-SH groups can then be detected by a suitable reagent,such as Ellman’s reagent. When a peptide contains two or moredisulfide bonds, the correct disulfide linkages should beestablished.

The chemical structure can be further substantiated by 1H-NMRand 13C-NMR spectroscopy. Other test methods used tocharacterize the drug substance should be fully described (eg,circular dichroism and fluorescence spectroscopy).

For applications to CBER for in vitro diagnostic (IVD) test kits,peptides should be shown to be devoid of blocking groups onthe side-chains, or else should be highly characterized regardingthe presence of such side-chain blocking groups. Also, it isimportant to demonstrate the presence of multiple epitopes.

CMC and Characterization/Proof of StructureEstablishment Potency assays, chromatographic assays, electrophoresis (eg,Description SDS-PAGE), immunoblot analysis, Fluorescence Activated CellInformation for Sorter (FACS) analysis, Enzyme-Linked Immunosorbent AssayHuman Plasma- (ELISA), Biological Activity Potency: Biological testing performedDerived Biological on the manufacturer’s reference (standard lot or test lot);Products or description of the potency assay should include the methodsAnimal Plasma or and standards used and the variability and acceptable limits ofSerum-derived the assay.Products

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 945

TABLE 4Continued

CMC Information Characterization/Proof of Structureand Establishment UV/visible or mass spectrometry; amino acid analysis; amino acidDescription or nucleic acid sequencing; carbohydrate analysis and, ifInformation for a appropriate, sequencing; peptide mapping; determination ofVaccine or disulfide linkage; SDS-PAGE (reduced and non-reduced);Related Product isoelectric focusing (1D or 2D); various chromatographic

methods such as HPLC, GC, LC, or TLC; NMR spectroscopy;and/or assays to detect related proteins including deamidated,oxidized, processed, and aggregated forms and other variants,such as amino acid substitutions and adducts/derivatives, andother process contaminants such as sulfhydryl reagents, urea,residual host proteins, residual DNA, and endotoxin.

Additional physicochemical characterization may be required formodified drug substances such as conjugates, multiple antigenpeptides (MAP), or those undergoing further chemical orenzymatic modifications. The information provided shouldinclude the degree of derivatization or conjugation, the amountof unmodified substance, removal of free materials (eg, toxins,linkers, etc.), and the stability of the modified substance.

Biological potency/ bioassay should include specific identity testingsuch as Western blot analysis or ELISA; cytometric analysis;neurovirulence testing, if appropriate; serotyping; electrophoretictyping; inactivation studies; neutralization assays; and titrations.A description and results of all relevant in vivo and in vitrobiological testing (bioassays) performed on the manufacturer’sreference standard lot or other relevant lots to demonstrate thepotency and activity(ies) of the drug substance should beprovided. This section should include a complete description ofthe protocol used for each bioassay, the control standards used,the validation of the inherent variability of the test, and theestablished acceptance limits for each assay. Thecharacteristics of specific antibodies used in theimmunochemical or serological assays should also be included.

CMC Information for Physicochemical Characterization of Reference Standard andA Therapeutic Qualifying LotsrDNA-Derived A description and the results of all the analytical testing performedProduct or a on the manufacturer’s reference standard lot and qualifying lotsMonoclonal to characterize the drug substance should be included.Antibody Product Information from specific tests regarding identity, purity, stabilityFor In Vivo Use and consistency of manufacture of the drug substance should

be provided. Examples of analyses for which information may besubmitted include, but are not necessarily limited to thefollowing: amino acid analysis, amino acid sequencing, entiresequence or amino- and carboxy-terminal sequences, peptidemapping, determination of disulfide linkage, SDS-PAGE(reduced and nonreduced), isoelectric focusing, Conventionaland HPLC (eg, reverse-phase, size exclusion, ion-exchange,etc.), mass spectroscopy, assays to detect product-relatedproteins including deamidated, oxidized, cleaved, andaggregated forms and other variants (eg, amino acidsubstitutions, adducts/derivatives), assays to detect residual hostproteins, DNA, reagents, immunochemical analyses, and assaysto quantitate bioburden, endotoxin.

(continued)

946 Robert L. Zeid

TABLE 4Continued

Additional physicochemical characterization may be required forproducts undergoing posttranslational modifications, forexample, glycosylation, sulfation, phosphorylation, orformylation. Additional physicochemical characterization mayalso be required for products derivatized with other agents,including other proteins, toxins, drugs, radionuclides, orchemicals. The information submitted should include the degreeof derivatization or conjugation, the amount of unmodifiedproduct, removal of free materials (eg, toxins, radionuclides,linkers, etc.), and the stability of the modified product. All testmethods should be fully described and the results provided. Theapplication should also include the actual data such as legiblecopies of chromatograms, photographs of SDS-PAGE oragarose gel, spectra, etc.

Biological ActivityA description and results of all relevant in vivo and in vitro

biological testing performed on the manufacturer’s referencestandard lot to show the potency and activity (ies) of the drugsubstance should be provided. Results of relevant testingperformed on lots other than the reference standard lot thatmight have been used in establishing the biological activity ofthe product should also be included. The description andvalidation of the bioassays should include the methods andstandards used, the inter- and intra-assay variability, and theacceptable limits of the assay.

ties or variants at the earliest stage in manu- a combined battery of physico-chemicalidentity tests. Potency is assessed via bioas-facturing so that one can modify or develop

the process more selectively. Thus, analytical say(s), which may be compendial or proprie-tary. Safety is usually assessed via compen-testing that is only geared to finished product

will not serve the long-term needs for devel- dial test methods such as sterility, bioburden,endotoxin, residual host cell protein, host cellopment and postapproval changes. Just as

importantly, the analytical data need to be DNA, impurities, and so forth.A number of guidance documents existlinked to other aspects of the product pa-

rameters such as bioassay, preclinical, and for various product classes that clarify thetype/extent of testing for regulatory dossiers.surrogate endpoints. Using a tiered testing

approach for demonstrating therapeutic equiv- Note that these guidance documents shouldbe used as a starting point, not a check list.alence, the program should assess physico-

chemical differences for any impact on safety Some guidance documents are considerablydated, so some proposed test methods mayand efficacy measures. If the development

program is designed correctly, it will explore not be state-of-the-art. Look at the most re-cent methods for characterization profileareas in anticipation of the regulatory re-

viewer questions and issues—weaving testing. Find testing guidance in FOI docu-ments such as summary basis of approvalsthe answers in as the program unfolds (see

Table 5). (SBA) for older innovator products or morerecent products in EPAR summaries, whichWhen performing analytical testing, the

holy trinity is purity, potency, and safety are posted on the Internet. In the case of somecompendial test methods, considerable effort(see Table 6). Purity is assessed in the same

fashion as currently conducted for ‘proof of may be spent doing crossover studies show-ing comparability of data sets. Finally, it maystructure’ characterization testing in drugs—

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 947

TABLE 5Tiered Approach of Testing for Demonstrating Therapeutic Equivalence

Analytical Equivalence • Physico-chemical comparisons• Confirmation of primary, secondary, and tertiary structures• Analysis of differences: Purity vs. impurities• Links with manufacturing and stability• Links to bioassay equivalence and surrogate endpoints• Compendial reference standards: Cross-over studies

Bioequivalence • Bioassay equivalence• Bioequivalence

Surrogate Endpoint • Use of validated surrogate endpointsEquivalence • Establishing new surrogate endpoints

be useful to comprise an exhaustive list of Bioassay Method Developmentand Validationproposed testing for any meetings with regu-

latory reviewers and get their input on cur-rent standards. A brief overview of CMC The regulatory requirement for determining

potency—21 CFR 610.10—is unique to bio-guidance documents and analytical testing isprovided (see Table 4). logics and when it comes to development

TABLE 6Analytical Equivalence Test Methods

Purity • Amino acid analysis (AAA)• Peptide mapping via restriction enzymatic digestion• Determination of sulfide linkages• Circular Dichroism (CD) confirmation of structure• Chromatography: HPLC (chiral, reverse phase, size exclusion, ion-exchange,

etc.)• Spectroscopy: MS, MALDI-TOF, EI, CI, FAB, ESI, NMR, FT-NIR, etc.• Electrophoresis (reduced and non-reduced): SDS PAGE, Capillary

Electrochromatography• Signal Transduction Fingerprinting• Micro-array technology• Immunochemical Analysis: ELISA, RIA, radial immunodiffusion, flow

cytometry, etc.• Microheterogeneity analyses: glycosylation profiles, IgG sub-types, etc.• Posttranslational modifications: assess glycosylation, sulfation,

phosphorylation, or formylation• Products derivatized with other agents (eg, other proteins, toxoids, drugs,

radionuclides, or chemicals): degree of derivatization or conjugation, amountof unmodified product, removal of free materials, and stability of modifiedproduct

Potency • Bioassay equivalence• Bioequivalence

Safety • Sterility• Bioburden• Endotoxin (LAL)• Residual host DNA• Host cell proteins• Impurities/degradation products: assays to detect product-related proteins

including deamidated, oxidized, cleaved, and aggregated forms or othervariants (eg, amino acid substitutions, adducts, derivatives)

948 Robert L. Zeid

and use of a bioassay, there are a vast number swayed one way or another by bioassay dataalone against a comparator product unlessof points to consider in their use as part of

any characterization profile. Choosing the there is reasonable sensitivity and signal-to-noise ratios (expressed as RSD). This is allassay, characterizing it, system suitability

criteria, and validation are just a few of the evaluated on a case-by-case basis.One of the most important features of anymultitude of variables. Once the assay is in

use, there are a number of other considera- bioassay is its sensitivity to detect subtlechanges in SAR. Similar to determining thetions for how it will be used in the context

of the development program (see Table 7). sensitivity of a drug assay, assess the sensitiv-ity of a bioassay to various concentrations ofBioassays typically have a lot of variabil-

ity and this limits the precision and accuracy. forced degradation samples of product/drugsubstance. How does a 20% deamidated pro-Significant ‘signal-to-noise’ ratio problems

can be seen as false positives or nonspecific tein perform in the method? How about 30%aggregated product? Does the unconjugatedbinding. For some animal-based bioassays,

it is not unusual to have relative standard protein compete with conjugated protein? Ifso, at what concentrations is interference ob-deviation (RSD) values of 15% to 20% or

higher. That is why it is helpful to have large served? And so on. An experienced bioassaymethod development person can flesh out adata sets using various batches of reference

material. This will help in the out-of-specifi- protocol rather handily. Once an acceptablesensitivity level has been established, it cancation (OOS) data investigations and outlier

analysis—even though most bioassays rou- be used to qualify links to manufacturing andstability.tinely incorporate a number of positive and

negative controls with every test sample. Sometimes, a sensitive yet “noisy” assaycan be accommodated by increased sampleNote that bioassays with high RSD may not

be very valuable in demonstrating compara- size—incorporating the assay variability intothe calculation for sample number (eg, vari-bility to another product. So, do not be easily

TABLE 7Bioassay Equivalence: Points to Consider

Compendial assays vs. noncompendial or proprietary assays: Need to assess if compendialmethods (if applicable) are sensitive enough for distinguishing slight variations in criticalproduct parameters; if compendial methods exist, do crossover studies for comparison

Establishing critical product parameters: Need to assess/correlate with functional activity/stability in bioassays (eg, antigen content vs. immunogenicity)

Relative Standard Deviation (RSD): What is noise-to-signal ratio for bioassay? > 25% RSD?Need to use statistical modeling to ensure adequate sample size for assessing widevariability (see Juran’s Handbook of Quality, Chapter 25-Acceptance Sampling)

Use “gold standards” in assays: Compare product to compendial reference standards or lotsof innovator product. Establish what physico-chemical differences relate to bioassayactivity

Process Development: Using neural nets to (1) help establish critical process parameterswith impact on product quality and (2) develop an “edge of failure” for any given processand establishing ranges within those critical areas. Do not depend on early stage materialfor market needs

Statistical Process Control (SPC): Sampling: continued analysis of product quality againstlong-term manufacturing history

Combining Tests into a Panel: Complex moieties may require several tests combined into aconglomerate panel to assess potency (eg, RP-HPLC, SDS PAGE, and bioassay)

Express Potency as Quantifiable Units: Need to express potency in some quantifiablemeasure-units [of activity]/mg of compound (anhydrous weight)

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 949

ables analysis or sampling scheme) (42)— is experienced in cell line characterizationtesting since they will most likely also havebut if the assay is not sensitive enough, then

testing more samples is futile. In that situa- the analytical prowess to perform some ofthe downstream product characterizationtion, it may be resolved by using a panel of

tests and combining the data into a single set testing as well. Be sure the search includesacademic centers with pharmaceutical re-of release criteria (43). That can be tricky if

one of the test methods is troublesome (eg, search and development centers. AlthoughCalifornia and Massachusetts have extensivelots of OOS investigations) because then the

entire method or lot is suspect until the biotechnology expertise, there may be simi-lar expertise closer (and more economically)method is fixed or an errant data determina-

tion. Last, but not least, potency needs to be such as in Colorado, Maryland, Nebraska,Iowa, and Kansas. Internationally, there areexpressed in quantifiable units per standard

unit of measure (eg, greater than 60% cyclic centers in Japan, Israel, Belgium, and Eng-land, to name a few. Once the methods areadenosine monophosphate (cAMP) stimula-

tion [as compared to control]/mg [of pure established and validated, it may be cost-effective to bring the analytical testing in-product]).

A key consideration in the use of bioas- house, given the enormity of testing that willneed to be performed for routine scale-upsays is that they should be extensively linked

(eg, comparability testing) for impact of and process development, in addition to allthe IPC, bulk product, finished product, andCMC changes to:stability testing.

1. Analysis of the finished product,2. Drug substance,

Structure-Activity Relationship Database3. Surrogate endpoints, and

and Quantitative SAR (QSAR)4. Stability.

A central tenet in demonstrating therapeuticequivalence is establishing structure-activityGiven the cost and time-consuming nature

of a lot of bioassays, this may not be feasible relationships databases throughout develop-ment for links to safety and efficacy. Al-until some later stages of development. They

should be routinely including these in the though SAR databases are not specificallymentioned in the drug or biologic CMC guid-comparability testing program once the final

formulation is established, but as early on ances, they derive the same concept—per-forming extensive testing throughout devel-in the manufacturing process as possible. If

using the bioassay for intermediates or IPC opment with an eye for impact on identity,purity, potency, and safety. The result is atesting, be sure to include those aspects in

the validation report or do that later. Try to product/process database—created and man-aged throughout development—that eventu-avoid a situation where the data look good,

but the validation report cannot support the ally relates process control to structuralintegrity throughout manufacturing and sta-method for the samples that were tested.

A final consideration for analytical testing bility. Any firm—innovator or generic—drug or biotech—needs to establish andis that once the methods are developed, a

decision about where they will be performed maintain an SAR database if it wants to sci-entifically justify manufacturing changesmust be made: in-house or contracted out.

Using an experienced contract analytical lab- later. For a multisource biotech initiative, itis even more imperative to gather and estab-oratory has a number of advantages. Be wary

of labs that have not been doing this for a lish an SAR database early on.This used to be problematic since therenumber of years or have limited experience

on the part of the personnel who will be was such limited SAR data in the public do-main, but that has all changed with the In-performing these tasks. It may be helpful to

look for a contract analytical laboratory that ternet. Today, there are tremendous data sets

950 Robert L. Zeid

and SAR activity known for a number of do this with enough physico-chemical dataand perhaps some supporting surrogate end-major biologic classes. With enough perse-

verance, culling the publications will enable point data. If there were significant differ-ences then a PK/PD-safety study might beyou to establish a cutting-edge SAR data-

base. in order.Once the SAR data are compiled, there

needs to be a readily retrievable, electronicPHARMACOKINETICS/

mechanism in-house that allows the growingPHARMACODYNAMICS OVERVIEW

database to be used across a variety of areas.Searching the Internet will reveal a number Current guidance considers PK/PD studies a

requisite for either multisource biotech orof database information systems geared tocreating quantitative structure-activity rela- comparability protocols, based on a premise

that:tionships. These software programs allowtwo-dimensional and three-dimensional rela-tional plots to identify compounds versus ac- 1. Current analytical methods are not defini-

tive enough to show comparability or thattivity. Some software capabilities includemultilinear regression analysis, genetic anal- 2. Some agents with substantial in vitro ac-

tivity may have limited efficacy in vivoysis, principle component analysis (PCA),and partial least squares to aid in constructing due to unfavorable pharmacokinetic prop-

erties.SAR databases.What if there is very little SAR data on

the compound of interest? Create it. Take Although some may disagree with the ration-ale, the regulatory consensus reflected in thethe innovator compound through a series of

forced degradation experiments, much the comparability guidelines is that PK/PD anal-yses should be routinely done; they are con-same as is done for an in-house project, and

see what 25% deamidation, 10% aggrega- sidered complementary to analytical testingsince some aspects of analytical SAR maytion, 15% impurities, and so forth act like

in the characterization profile testing. What not detect subtle differences that can only beobserved in vivo. It is not clear if this PK/kind of activity do peptide-digested frag-

ments have in the bioassays? Obviously, keep PD recommendation extends to all biotechproducts or only for those where character-in mind any additional considerations for

method validation to ensure that the analyti- ization requires complex analytical methods.Regardless, one of the difficulties in per-cal data are accurate.

With only SAR data, can therapeutic forming a PK/PD study for a biotherapeuticis to make sure that the study design modelequivalence be demonstrated solely on the

basis of analytical equivalence? Maybe. It is appropriate. A growing concern over theclassic drug model of PK/PD study designreally depends on the nature of characteriza-

tion data already in hand and the nature of is that it is not definitive enough to discerndifferences in biotherapeutic applications,observed changes/differences from an inno-

vator/reference compound. It is possible for which may have significantly delayed onsetor duration. Different biotech classes maySAR data to support some minor changes,

but each case is unique. Current guidance require different models depending upon thedose-response relationship; some studiesdocuments on comparability testing do not

rule it out completely; it must be evaluated may require a surrogate endpoint markersuch as for immunomodulatory drugs im-against the manufacturer’s existing SAR da-

tabase for the proposed old versus new prod- pacting T cell lymphocyte counts or activity.Some proteins or monoclonal antibodies mayuct characteristics. This is clearly a case

where the innovator’s database with exten- require extensive assessment of immunoge-nicity or neutralizing antibodies (40,41).sive safety and efficacy links will have great

advantages. It is possible a comparator could Other proteins may induce tolerance. Still,

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 951

other proteins may induce cross-reactivity to 1. The shape of the dose-response curve,2. Duration of response to develop and per-normal tissues. Given the unique require-

ments to assess PK/PD relationships, surro- sist in relation to PK, and3. The concentration as well as dose to sepa-gate endpoints, and safety issues, some PK/

PD studies may resemble an amalgamation rate PK and PD components.of safety/tolerability with PK/PD assessments(surrogate endpoints) mixed in. Obviously

These downstream effects of biologics obvi-when it comes to designing a PK/PD study

ously complicate use of the classic drugmodel for a particular biotherapeutic, cling-

model for PK/PD determinations of biothera-ing to the classic drug model may be highly

peutics.inappropriate. So, depending upon the

Another aspect of examining the dose-choices for biotherapeutic model, surrogate

response curve is whether the PK/PD re-endpoints, assay sensitivity, relationship of

sponse is linear or whether there is a maximalresponse to clinical outcome, and countless

effect (EMAX or sigmoidal EMAX). Within theother features, a poorly planned PK/PD study

PK/PD data, what are the curves for AUC,may provide an expensive morass of data

TMAX, CMAX, and T1/2 and how well do theythat is counterproductive to streamlined de-

correlate with any safety measurements orvelopment.

surrogate marker responses? One can thenPerhaps the best place to start is finding

link the dose-response curve to best fit as athe best model to fit to the observed duration

time-dependent feature, concentration-de-of the pharmacokinetic-pharmacodynamic

pendent feature, or a combination.phase, which may be captured from preclini-

A key aspect in biotherapeutics is receptorcal studies or published clinical trials. With

kinetics. The PD response reflects the typesregards to the PK/PD relationship, there are

of receptors, receptor populations and sub-at least three scenarios for the duration of

types, affinity for these binding sites, and alleach component: PD < PK, PD = PK, or

the multitudes of factors that disease mayPD > PK phase. In instances where the dura-

have upon expression of a desired effect.tion of the PD effect is shorter than the PK

That is why certain in vitro or ex vivo studiesprofile these products are usually weeded out

may be critical in assessing comparability ofby Phase II or Phase III studies. In other

therapeutic effect at the target site—despiteproduct profiles the duration of PD is compa-

significant differences in PK parameters. Arable to PK, reflecting competitive receptor

theoretical maximal biologic response mightoccupancy, which is often seen with classi-

consist of the following: 100% of the injectedcally synthesized drugs and some rDNA-

active moiety reaching the precise target re-derived products. Where the duration of PD

ceptor population fully intact with full ago-is significantly longer than PK, however, this

nist activity and a maximally efficient ex-suggests either alteration of an existing bio-

pression of the receptor activation. Morelogic response or induction of a new re-

often, it is a reflection of:sponse. Often with biological systems, theremay be a ‘cascade’ effect allowing activationof a small receptor population to escalate 1. Dilution or attenuation of the dose by

plasma volume,downstream into a systemic response viamodulation of a key regulatory pathway or 2. Nonspecific binding proteins that remove

the active moiety from general circulation,a long-lived therapeutic effect. The possiblemechanisms for this effect (where PD > PK) 3. Variable affinity expressed across a het-

erogeneous pool of receptor sites,could be an immune response, homeostaticperturbation, altered signaling network, or 4. Variable levels of agonist activity (eg, full

or partial agonist),priming. Some key considerations for evalu-ating the PK/PD relationship with biothera- 5. Dependence on intermediate messengers

for expression of the biological effect,peutics is to evaluate:

952 Robert L. Zeid

6. Competitive antagonism and/or noncom- sessment of clinical efficacy. Consider it astandard part of any BLA filing for now,petitive inhibition,

7. Antagonism by functional or physiologi- which will also necessitate an IND to per-form any clinical evaluations outside the ex-cal mechanisms, and

8. Degradation/removal rates. emptions noted in 21 CFR 312.2(b). If ana-lytical equivalence and bioequivalence are

In addition to these complex assessments still not sufficient to meet regulatory stan-for single-dose studies, repeated administra- dards, there is one alternative to repeatingtion therapies with biologics often need to full pivotal efficacy studies: surrogate end-assess immunogenicity and/or the formation points.of neo-antigens that could antagonize the Surrogate endpoints are valuable determi-therapeutic effect with repeated administra- nants when primary efficacy endpoints oftions (eg, antigenicity of met-hGH) and in- morbidity or mortality take too long to assessduce tolerance. Immunogenicity can arise in chronic disorders or may become unethicalfrom coeluted process impurities (eg, cell delays in life-threatening conditions such assubstrates or media fractions) or product- criteria for interim analyses and “stoppingrelated features (eg, fragments, aggregates, rules” in a blinded pivotal efficacy study.chemical modifications, etc.), which are fur- Surrogate endpoints may have had some in-ther influenced by comorbid disease, preexist- auspicious roots from the early days of “dataing antibodies, or concomitant medications. dredging” for primary and secondary end-

What about situations when pharmacoki- points in integrated efficacy summariesnetics profiles differ? All is not necessarily (IES)—a veritable “Hail Mary” of statisti-lost. Differences in total AUC, CMAX, or TMAX cians to find something statistically (clini-may not be as critical as bioequivalence for cally?) significant in the morass of data. Buta surrogate marker or ex vivo assessment at surrogate endpoints quickly evolved intothe proposed site of action. Any differences something much more tangible—a quantifi-will need to be assessed for impact on effi- able early indication of a positive clinicalcacy or safety; some significant changes may prognosis, a surrogate marker of efficacy. Itrequire a clinical bridging study. quickly won prominence as part of any ‘fast-

Even when all the PK/PD study dust has track’ program and more importantly, strucksettled, some might argue that it is still not a real balance of moving new therapies forenough. Bioequivalence is not a true measure life-threatening conditions into the clinic/of clinical efficacy. Regardless of having medicine chest based on objective criteria.shown both analytical equivalence and bio- Now they are a standard part of any efficacyequivalence, one could still argue that valu- study. Obviously the real value of a surrogateable efficacy assessments—which can only endpoint is how well it correlates to the even-be collected from a pivotal study—would tual clinical outcome, which has become astill be needed. Here again, it would have to science unto itself.be evaluated on a case-by-case basis of the Surrogate endpoint determinations/equiv-existing SAR data against the proposed clini- alence is a moving target—an evolving sci-cal indication. Time and experience might ence—that often reflects the latest clinicalmitigate the need for additional clinical data. prognostic indicators. Thus, methods used 20Some clinical studies required 20 years ago years ago may be outdated by virtue of tech-might not need to be repeated today even if nology and clinical experience/knowledge ofthe situations are identical. the disease condition. Consider rheumatoid

arthritis. Until early 1990, inflammation andSURROGATE ENDPOINTS joint pain (essentially symptomatic relief)

EQUIVALENCE OVERVIEW were the predominant measures of treatmentsuccess. This therapy reduced joint swelling,There will likely be many situations where

regulatory authorities will require some as- joint stiffness, and joint pain which increased

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 953

range of motion and quality of life. In spite novator—and thus sparing extraordinaryexpense while achieving the same goal: dem-of these treatments, however, a number of

patients would progress from Stage II to onstration of efficacy (as assessed by today’sstandards).Stage IV. The X-ray data indicated that de-

spite symptomatic relief, occult joint erosion/ Put another way, the surrogate endpointsof yesterday may or may not be the bestdegeneration continued and eventually wors-

ened the patient’s state. Today’s therapies fo- fit, the most economical, or even the mostrelevant diagnostic/prognostic tool. Just ascus on maintenance of the joint integrity by

selectively inhibiting mediators of joint ero- the lethal dose testing (eg, LD50) has beenreplaced by more elegant toxicological meth-sion (eg, Celebrex and the category of

cyclo-oxygenase [COX]-2 inhibitors) while ods, current technology offers more opportu-nities to compare a multisource product tosymptomatic relief is a separately managed

affair. Thus, surrogate endpoints for arthritic- an established clinical standard and chart anew course of evaluating efficacy at a mini-related conditions today would encompass

X-ray or other imaging for the joint. Since mum of cost. A brief summary of surrogateendpoints is outlined below (see Table 8).new diagnostic tools and correlations are oc-

curring all the time, it is imperative to assessthe most recent technologies applied to the

SUMMARYclinical indication.

There may be situations where the innova- Demonstrating therapeutic equivalence forbiotech-derived products is enormously com-tor product was approved and no surrogate

endpoints were established at the time of the plex—filled with scientific, medical, legal,and regulatory issues, and a plethora of man-filing. Search the literature to see if any sur-

rogate endpoint analyses were performed ufacturing and process development mile-stones. Each situation is unique and yet, therepostapproval either by the innovator or by

an independent clinical group. If so, this may are aspects that are highly conserved and canbe extrapolated to other situations. The puz-be a good place to start. Are there any surro-

gate endpoint studies conducted with similar zle is in knowing which is which. The topicdoes not lend itself to easy summaries orproducts or from that class of compounds?

If there is nothing in the public domain, it simple messages and it would be a disserviceto such weighty issues to try. Even trying tomay require establishing them, but it may

not be as bad as it sounds, at least not com- separate the factions into innovator versusgeneric companies seems too simplisticpared to the time and cost of a full-blown

Phase III pivotal efficacy study. A possible given the enormous integration most compa-nies have: innovators and generic firms understrategy may stem from an ancient mathe-

matical principle: If A = B, and B = C, then one conglomerate subsidiary roof. These areissues that concern us all.A = C.

Hypothetically, a multisource biotech One immutable aspect keeps surfacing,however, multisource biotech is possible andproduct to be used as a chemotherapy adjunct

might be tested today by comparing stem cell the feasibility is growing all the time. Somemay dismiss the idea entirely, but their num-stimulation against an innovator reference

standard. A multisource interferon might be ber is dwindling. Put another way, it is impos-sible to say today what will be possible nextassessed against a battery of sophisticated

in vitro/ex vivo prognostication tests used year or even in 10 years. No one truly knows.We are witnessing a dramatic evolution ofroutinely in assessing cancer patients’ ther-

apy, that is, as the levels of measuring thera- science and technology. Those attuned to the‘Information Age’ will be able to accomplishpeutic benefit become more sophisticated, it

is possible to leap upon that current technol- things that were unheard of only 10 years ago.If there is a meta-message to this topic, itogy and exploit it in a precise fashion—com-

paring the multisource product against an in- is that demonstrating therapeutic equivalence

954 Robert L. Zeid

TABLE 8Surrogate Endpoint Equivalence

Vaccines: Geometric Mean Titer (GMT) and seroconversion (vs. demonstrated prophylaxisfrom long-term studies)

Fibrinolytics: in vitro clot lysis, post-MI (myocardial infarction) patency, ventricular functiontesting (ECG), etc. (vs. survival rates)

Somatotropins: hyphosectomized rat model and comparison to external reference standards(vs. long-term growth confirmation)

Cystic Fibrosis Rx: pulmonary function testing (PFT), days in ICU, days on intravenousantibiotics (vs. reduced number of exacerbations and long-term survival)

HIV/AIDS: reduction in CD4 count, viral burden, reduced rate of opportunistic infections (vs.survival rates)

Arthritis: radiological imaging of joint damage, reduction of inflammatory mediators,improved joint mobility, etc.

Chemotherapeutics: immunological mediators, carcinogenic antigen levels, lymph nodeinvolvement (vs. long-term survival)

for biotech-derived products will be a more efforts have delivered a realm of possibilitiesto medicine and society that were unimagin-circumspect and holistic approach than what

has been applied toward the classic drug able 20 years ago. Some of us would not bestanding in this place and time discussingmodel. Considerations for microheteroge-

neity, SAR data, safety, impurity profiles, and these ideas were it not for some whose think-ing ‘outside-the-box’ has come to epitomizeclinical equivalence will all be weighed in

by regulatory authorities in their determina- biotech, an industry that defines the cutting-edge of science and medicine. The biotechtion of sameness or therapeutic equivalence.

These concepts, already integrated into a industry, constantly searching for faster waysto develop revolutionary products, collabo-number of guidance documents, must come

to fruition throughout development for any rated closely with regulatory agencies to ulti-mately develop the ‘specified biotech’ para-sponsor to use in support of product/process

changes. This learning curve must also find digm facilitating critical medicines fasterthan ever to the clinic and medicine cabinet.its way into decision trees for regulatory re-

viewers to use in product evaluation. That is no small matter considering thefledgling state of the biotech industry at theIn the case of multisource biologics, the

sponsor will be pushing the envelope to sup- time.Throughout all of these successes, thereport product interchangeability and thus,

may have to provide an extraordinarily well- have been tremendous obstacles: technicalchallenges of forging new medicines fromdeveloped SAR database supporting those

claims. The amount of work involved will cutting-edge science, inequitable patent andintellectual property protection, competitionprobably be the same as for any new drug

development—just exerted in different from abroad, spotty funding from WallStreet, education of the public and regulators,areas. There will not be many shortcuts to

these new opportunities. And no doubt, there and spectacular clinical failures. Biotech sur-vived because it was smarter than the averagewill be many that miss the mark. To avoid

becoming the “Donner Party” of the industry, bear. Biotech found political allies to helpredress the inequities in patent law and intel-find a good guide and study the trail of those

who have gone before. lectual property. Biotech found funding out-side mainstream sources and forged newIn homage to the trailblazing, innovation,

and genius emanating from the biotech in- ground in licensing and marketing arrange-ments. Biotech marketed itself to the publicdustry and the pharmaceutical industry, their

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 955

in new and better ways—a glimpse of poten- generations to come. We owe homage tothose who partnered together in bringing ittial things to come. Biotech even survived

some of its own management! It would have to fruition. But now, as another part of theindustry wants to trod that same trail, somekilled a lesser industry.

Sometimes we forget what brilliance, cun- are working to selectively undo or qualifythose achievements. Can we as a societyning, and chutzpah it has taken to accomplish

these things in spite of those odds. Some- allow that to happen? Is there a going back?If so, can we afford to do that? Can we permittimes society just does not get it—all the

extraordinary effort required to create a ‘mir- a select portion of the pharmaceutical indus-try to open, and then close, a tremendous gateacle drug’—one after another. Once upon

a time, medical practice was constrained to of opportunity? Is the scientific objectivity ofthis specified biotech premise only applica-leaches, bloodletting, and arsenicals. In 50

years, will we look back on today’s chemo- ble to an innovator? Is it only good for alimited time? No, and I think most in thetherapy with the same shudder of barbarism?

If we do, how do you think we will be deliv- pharmaceutical industry, and society, wouldagree. What is done is done and we are allered from that?

The faster and faster pace of innovation, better off for it.But the extrapolation of specified biotechrather than amazing the general public, has

sadly desensitized many to this bounty. Not to multisource biotech is a great leap thatsome say we are not ready to take. It requiresonly have the successes lessened the incredi-

ble impact these new products bode for us more time and study. Some contend the spec-ified biotech paradigm is not the same asindividually and as a society, but ironically

they may have raised public expectations to FDA saying it will accept a multisource bio-tech process. The law—a 16-year-old law—a realm of unrealistic promise—a drug with

no risk? Maybe it also reflects how familiar specifically prohibits that. Change the lawand you will have your multisource biotechthis terrain has become for many. What bio-

tech delivers may still be spectacular, but we process. Congress will compel FDA to createa process. This is a valid point. Specifieddo not see the industry in the special way we

used to. Biotech is growing up. And maybe biotech does not equal multisource biotech.But if specified biotech can flourish, thensome of the distinctions that protected a

fledgling industry seem a little out of place, the possibility of multisource biotech is astone’s throw away. Obviously, that worriesodd now, considering its size and experience.

For nearly a century, biologics have been a lot of people. Because changing a 16-year-old law, in light of what we know today andregulated a world apart from classically syn-

thesized drugs. For the last 20 years, biotech will learn tomorrow, is not going to be thatdifficult, certainly not with the rest of soci-has fought hard to overcome that mindset

using science and cold, hard data. In morph- ety—or the rest of the world—looking overyour shoulder.ing from the “product = process” dogma, bi-

otech aspired for FDA reviews and approvals Outside the United States, there are noconstraints from the semantics of Hatch–more like mainstream pharmaceuticals (eg,

elimination of establishment license require- Waxman and the idea of multisource biotechhas already taken hold with its promise ofments for specified biotech). It got its wish

with ‘specified biotech’ and leveled the play- cost-containment for socialized medicineformularies. Many countries are moving for-ing field for a great many.

Some people in biotech and FDA were ward, embracing the concept and strugglingwith how to implement it. Ironically, manyvisionary and acted on things that most of

us did not even know existed. We should all United States innovator firms may seize thisopportunity to do overseas what they saytake special note to mark this extraordinary

evolutionary step in unifying drugs and bio- cannot be done by a comparator to them onhome soil. That type of action reflects a purelogics evaluation. It will benefit society for

956 Robert L. Zeid

economic decision devoid of any dissonance to knowing a unique domain—a discretephase—and escorting these products alongfrom other markets. But how long can this

lack of international harmonization bode in their journey, improving upon the success ofmaking it to their final destination: a person,favor of a multisource biotech process solely

outside the United States? Some think, not a patient. That is our job, our responsibility.With the ever shifting fiefdoms that arisefor long.

Regardless of whether the United States from mergers and new relationships, theseare marked by changes in the landscape withlaw is changed soon or not—regardless of

any dissonant action on the part of United new roads. New technologies allow us totravel faster. It is our challenge to learn theseStates industry abroad—the international

winds of change are stirring. Will the age- new avenues and back roads, to embracethese new technologies for our advantage.old ‘drug gap’ reappear but cast in a new

light: cost, not availability? Such cost differ- Our travels may bring us alongside othersdoing the same, sharing a common path.entials may only exacerbate an already grow-

ing problem of the ‘gray’ market and ‘black’ Some sojourns may involve truly uniqueproducts and uncharted territory. Other jour-market pharmaceuticals that stem from im-

ports and smuggling from cheaper countries neys may be with products that bear a resem-blance to the first, but are not the same. Eachabroad. Will there be a special curtailment on

Internet sales/shipments of pharmaceuticals journey is unique.Maybe in the end, we can all lay claim tofrom abroad? Or can these be handled like

any other personal use exclusion to import a special relationship with that product, butnone of us can ever lay claim to owning thelaw? Will this disparity spawn a whole new

line of charter bus trips for taking patients road or justifying blocked passage of otherrightful products on their way to a final desti-across the border into Canada and Mexico

for cheaper medicines? It is such a conun- nation. Because when you deny passage toa partner in this industry—access that is nodrum, and one which is quickly evolving.

The exclusion of biologics from Title I longer in your control—on the basis of poli-tics or money, it is capricious, shallow, andof the Hatch-Waxman Amendments may be

current law, but the scientific rationale that greedy. More importantly, the impact is feltfar beyond your competition. It has impactsupported it—that 90-year sojourn and the

ship it sailed in on—was crashed upon the on real people—people who can barely af-ford the generic version of most medicines,shoals when the specified biotech paradigm

moved forward. As more and more shining let alone the innovator. That population isgrowing.examples of this cognitive dissonance come

to light, it will eventually eclipse all the peo- So another aspect, ethics, gets hurled intothis hubris. As biotech flourishes and pro-ple objecting and pointing to that wreck—

holding up select bits of tattered dogma in duces more products for mainstream medi-cine, price will become an even greater deter-support of maintaining an outdated status

quo. The public perception of biotech will minant of who gets what medicine. Rationinghealth care resources among low income/shift. They may come to realize an avarice

that they had never seen before. At that point, fixed income families—among formularies,health care providers, and third-party pay-biotech and mainstream pharmaceuticals will

truly become one—a single indistinguishable ers—has already reached crisis levels and isonly going to get worse. If something is notblight on the solemnity and ethics of medicine

and what this industry really stands for. done proactively, biotech could well becomethe symbol embodying elitism in medicineThe pharmaceutical industry is but one of

many partners in an extraordinary health care and the catalyst for drastic reform.The bottom line is that staving off genericsystem—each of us shepparding ideas from

conception to fruition—from bench top to competition (or not) of increasing numbersof biotech-derived therapies has vast implica-medicine chest. Each partner can lay claim

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 957

Kurt Nielsen (TEVA), Sally Seaver (Seaver Associates),tions for containing health care costs andTom Speace (Gensia Sicor), Michael Spino (Apotex),managed care. The remedy may come asAnthony Ridgway (HPB), Michael Kunitani (Chiron),

price controls or some other noxious solu- Pierrette Zorzi-Morre (AFSSaPS), Loren Miller (PPD),tion. It has been done before for other indus- and David Rosen (McDermott, Will, & Emery).tries and with enough people clamoring, itmay resemble more of the French Revolution

REFERENCES:than a civilized discussion about containinghealth care costs from biotech-derived prod- 1. Public Health Service Act (42 U.S.C. 262).ucts. A lot of gains from these past years 2. Federal Food, Drug, and Cosmetic Act (21 U.S.C.

355).may be lost. There is really only one thing3. Drug Price Competition Act and Patent Term Resto-left to do.

ration Act of 1984.Lead the way. Biotech has done it before.4. Food and Drug Administration Modernization Act

Now, do it again. Take the high road and (FDAMA) of 1997.partner with FDA and ICH to forge scientifi- 5. CBER. Points to Consider in the Production and

Testing of New Drugs and Biologicals Produced bycally sound monographs (SAR databases) forRecombinant DNA Technology. Rockville, MD: U.S.various product classes. By creating high cal-Food and Drug Administration; April 1985.iber compendial standards for the product

6. CBER. Supplement to the Points to Consider in theclasses, the biotech industry will have the Production and Testing of New Drugs and Biologi-highest impact on the process. Partner with cals Produced by Recombinant DNA Technology:

Nucleic Acid Characterization and Genetic Stability.other factions of the industry and CongressRockville, MD: U.S. Food and Drug Administration;to change the laws that allow for a unifiedApril 1992.approach to multisource biotech. By doing

7. CBER. Points to Consider in the Characterizationso, biotech will have tremendous input into of Cell Lines Used to Produce Biologicals. Rock-the process—not be held captive to its results ville, MD: U.S. Food and Drug Administration; May

1993.made outside its control. In doing this, the8. CBER. Guidance on Alternatives to Lot Release forbiotech industry will spare itself and others

Licensed Biological Products. Rockville, MD: U.S.protracted litigation that may have little moreFood and Drug Administration; July 1993.

result than bolstering a lot of attorneys’ re- 9. CBER. Elimination of Establishment License Appli-tirement accounts. cation and Product License—Final Rule. U.S. Food

and Drug Administration. Federal Register (Vol 64;Do this—not for the money or the pres-202; 56441–56454), October 20, 1999.tige—but because it is the right thing to do.

10. CBER. Guidance for the Demonstration of Compa-By doing it, the biotech industry will distin-rability of Human Biological Products, Including

guish itself from the rest of the industry as Therapeutic Biotechnology-Derived Products. Rock-innovative and collaborative—ready and ville, MD. U.S. Food and Drug Administration; April

1996.willing to take on any competition for the11. CBER. Guidance for Industry: The Sourcing andgreater good. On the cusp of this new millen-

Processing of Gelatin to Reduce the Potential Risknium we are all looking for leadership to takePosed by Bovine Spongiform Encephalopathy (BSE)

us in new directions. Is that not what biotech in FDA-Regulated Products for Human Use. Rock-has always been about? ville, MD: U.S. Food and Drug Administration;

October 1997.12. CBER. Guidance for Industry: Revised Precaution-

Acknowledgment—The author would like to thank the ary Measures to Reduce the Possible Risk of Trans-mission of Creutzfeldt-Jakob Disease (CJD) andfollowing people for their invaluable insight and com-

ments surrounding this topic: Donald Hare (FDA), Yuan- New Variant Creutzfeldt-Jakob Disease (nvCJD) byBlood and Blood Products. Rockville, MD: U.S.Yuan Chiu (FDA), Kathy Zoon (FDA), Gordon Johnston

(Lachman Consultants), Kenneth Seamon (Immunex), Food and Drug Administration; August 1999.13. CBER. Guidance for Industry: Possible Dioxin/PCBRoger Williams (FDA), Doug Sporn (FDA), Bruce Dow-

ney (Barr Labs), Robert Bell (Barr Labs), George Bar- Contamination of Drug and Biological Products.Rockville, MD: U.S. Food and Drug Administration;rett (TEVA), Carol Ben-Maimon (TEVA), Beth Brannon

(Geneva), Bill Haddad (MIR Pharmaceuticals), Greg August 1999.14. International Conference on Harmonization (ICH).Davis (Eli Lilly), Bruce Mackler (Heller Ehrman), Su-

zette Kox (Ratiopharm), Erich Kohler (Ratiopharm), Final Guideline on Quality of Biotechnological

958 Robert L. Zeid

Products: Analysis of the Expression Construct in Human Use.” Rockville, MD: U.S. Food and DrugAdministration; May 1999.Cells Used for Production of rDNA-Derived Protein

Products. Geneva, Switzerland: ICH Secretariat; 25. CBER. Draft Guidance for Industry: Current GoodManufacturing Practice for Blood and Blood Com-February 1996.

15. International Conference on Harmonization (ICH). ponents: (1) Quarantine and Disposition of PriorCollections from Donors with Repeatedly ReactiveDraft Guidance on Specifications: Test Procedures

and Acceptance Criteria for Biotechnological/Bio- Screening Tests for Hepatitis C Virus (HCV); (2)Supplemental Testing, and the Notification of Con-logical Products. Geneva, Switzerland: ICH Secre-

tariat; August 1998. signees and Transfusion Recipients of donor TestResults for Antibody to HCV (Anti-HCV). Rockville,16. International Conference on Harmonization (ICH).

Guidance on Quality of Biotechnological/Biological MD: U.S. Food and Drug Administration; June 1999.26. CBER. Guidance For the Submission of Chemistry,Products: Derivation and Characterization of Cell

Substrates Used for Production of Biotechnological/ Manufacturing and Controls Information and Estab-lishment Description for Autologous Somatic CellBiological Products. Geneva, Switzerland: ICH Sec-

retariat; September 1998. Therapy Products. Rockville, MD: U.S. Food andDrug Administration; January 1997.17. International Conference on Harmonization (ICH).

Guidance on Viral Safety Evaluation of Biotechnol- 27. CBER. Guidance for Industry for the Submission ofChemistry, Manufacturing, and Controls Informa-ogy Products Derived From Cell Lines of Human

or Animal Origin. Geneva, Switzerland: ICH Secre- tion for Synthetic Peptide Substances. Rockville,MD: U.S. Food and Drug Administration; Novembertariat; September 1998.

18. International Conference on Harmonization (ICH). 1994.28. CBER. Guidance for Industry for the Submission ofGuidance on Specifications: Test Procedures and

Acceptance Criteria for Biotechnological/Biological Chemistry, Manufacturing, and Controls Informa-tion for a Therapeutic rDNA-derived Product of aProducts. Geneva, Switzerland: ICH Secretariat;

August 1999. Monoclonal Antibody Product for In Vivo Use. Rock-ville, MD: U.S. Food and Drug Administration; Au-19. CBER. Final Rule: Changes to an Approved Appli-

cation for Specified Biotechnology and Specified gust 1996.29. CBER. Points to Consider in the Manufacture andSynthetic Biological Products and Biological Prod-

ucts. U.S. Food and Drug Administration. Federal Testing of Monoclonal Antibody Products for HumanUse. Rockville, MD: U.S. Food and Drug Adminis-Register. (Vol 62; 39889–39903), July 24, 1997.

20. CBER. Guidance for Industry: Content and Format tration; February 1997.30. CBER. Guidance for Industry: Interpreting Same-of Chemistry, Manufacturing and Controls Informa-

tion and Establishment Description Information for ness of Monoclonal Antibody Products Under theOrphan Drug Regulations. Rockville, MD: U.S.a Vaccine or Related Product. Rockville, MD: U.S.

Food and Drug Administration; January 1999. Food and Drug Administration; July 1999.31. FDA. Guidance for Industry: Fast Track Drug De-21. CBER. Guidance for Industry: For the Submission

of Chemistry, Manufacturing and Controls and velopment Programs—Designation, Development,and Application Review. Rockville, MD: U.S. FoodEstablishment Description Information for Human

Plasma-Derived Biological Products, Animal and Drug Administration; November 1998.32. Zeid RL. Generic biologics: Notes from the pathPlasma or Serum-Derived Products. Rockville, MD:

U.S. Food and Drug Administration; February 1999. less traveled. BioPharm. 1999;12(3):24–32.33. Zeid RL. Generic biologics: Glimpses through the22. CBER. Guidance for Industry: Content and Format

of Chemistry, Manufacturing and Controls Informa- mist. RAPS. February 1999.34. Zeid RL. Generic biologics: Could the impossibletion and Establishment Description Information for a

Biological In Vitro Diagnostic Product. Rock- become reality? RAPS. March 1999.35. Zeid RL. Demonstrating Therapeutic Equivalenceville, MD: U.S. Food and Drug Administration; March

1999. for Biotech-derived Products. Presentation at Inter-national Business Communications (IBC) USA Con-23. CBER. Guidance for Industry On the Content and

Format of Chemistry, Manufacturing and Controls ference: Rx & Biotech Generics, Georgetown Uni-versity, Washington, D.C.; September 21–23, 1999.Information and Establishment Description Infor-

mation for an Allergenic Extract or Allergen Patch 36. Statement of Policy for Biotechnology Products.Federal Register. (51) 23309, June 26, 1996.Test. Rockville, MD: U.S. Food and Drug Adminis-

tration; April 1999. 37. Approved Drug Products with Therapeutic Equiva-lence Evaluations. (‘Orange Book’), XVI ed., 1996,24. CBER. Guidance for Industry For the Submission

of Chemistry, Manufacturing and Controls and CCH (Chicago).38. CDER. Draft Guidance for Industry—ApplicationsEstablishment Description Information for Human

Blood and Blood Components Intended for Transfu- Covered by Section 505(b)(2). Rockville, MD: U.S.Food and Drug Administration; October 1999.sion or for Further Manufacture and For the Com-

pletion of the Form FDA 356h “Application to Mar- 39. BIO (Biotechnology Industry Organization) Presen-tation by Dr. Alan Goldhammer during Generic Re-ket a New Drug, Biologic or an Antibiotic Drug for

Demonstration of Therapeutic Equivalence for Multisource Biotech Pharmaceuticals 959

combinant Drugs/Biologics Session. 1998 Annual Godfrey, eds. New York: McGraw Hill; March1999.RAPS Meeting.

40. Stein KE. Preclinical and Clinical Development of 43. Zabrecky JR, Brown, EK, Compton, BJ, KretschmerMW, Fowler E, and Bernardy JD. CombiningBiological Therapeutics: Focus on Pharmacokinetics

and Pharmacodynamics. Annapolis, MD; October ELISA, RP-HPLC, and SDS-PAGE to Define thePotency of a Complex Biologic. BioPharm. 1998;18, 1999.

41. FDA. Biological Response Modifiers Advisory 11(10), 34–44.44. Kittle JD, Pimentel BJ. Testing the Genetic StabilityCommittee Meeting Minutes: Bethesda, MD, July

15, 1999. of Recombinant DNA Cell Banks. BioPharm. 1997;10(10) 48–51.42. Chapter 25: Acceptance Sampling in Juran’s Quality

Handbook, 5th ed., Joseph M. Juran & A. Blanton