recommendations for the design, optimization, and qualification of cell-based assays used for the...

Journal of Immunological Methods 321 (2007) 1–18www.elsevier.com/locate/jim

Review

Recommendations for the design, optimization, and qualification ofcell-based assays used for the detection of neutralizing antibody

responses elicited to biological therapeutics

Shalini Gupta a,⁎, Stephen R. Indelicato b, Vijay Jethwa c, Thomas Kawabata d,Marian Kelley e, Anthony R. Mire-Sluis f, Susan M. Richards g, Bonita Rup h,

Elizabeth Shores i,1, Steven J. Swanson a, Eric Wakshull j

a Clinical Immunology, Amgen Inc., Thousand Oaks, CA 91320, USAb PharmSci Quality-Quality Systems Improvement Office, Schering Plough Research Institute, Kenilworth, NJ, 07033, USA

c Bioanalytical QC RTP, Biogen Idec, Research Triangle Park, NC 27709, USAd Safety Sciences, Pfizer Global Research and Development, Groton, CT 06340, USA

e Clinical Pharmacology & Experimental Medicine, Centocor R&D Inc., Radnor, PA 19087, USAf Corporate Operations, Amgen Inc., Thousand Oaks, CA 91320, USA

g Immunology, Cell & Protein Therapeutics R&D, Genzyme Corp. Framingham, MA 01701, USAh Bioanalytical R&D, Wyeth Research, Andover, MA 01810, USA

i Division of Therapeutic Protein, Office of Biotechnology Product, Center for Drug Evaluation and Research, FDA, MD 20892, USAj Clinical Science & Technology, Biogen Idec, Cambridge, MA 02142, USA

Received 4 April 2006; received in revised form 29 November 2006; accepted 6 December 2006Available online 12 January 2007

Abstract

The administration of biological therapeutics can evoke some level of immune response to the drug product in the receivingsubjects. An immune response comprised of neutralizing antibodies can lead to loss of efficacy or potentially more serious clinicalsequelae. Therefore, it is important to monitor the immunogenicity of biological therapeutics throughout the drug productdevelopment cycle. Immunoassays are typically used to screen for the presence and development of anti-drug product antibodies.However, in-vitro cell-based assays prove extremely useful for the characterization of immunoassay-positive samples to determine ifthe detected antibodies have neutralizing properties. This document provides scientific recommendations based on the experience ofthe authors for the development of cell-based assays for the detection of neutralizing antibodies in non-clinical and clinical studies.© 2006 Elsevier B.V. All rights reserved.

Keywords: Neutralizing antibody bioassay; Cell-based assay; Serum-based bioassay; Immunogenicity assay; NAb assay

Abbreviations: BrdU, bromodeoxyuridine; CAT, chloramphenicol acetyl transferase; CV, coefficient of variance; ECL, electrochemilumines-cence; EIA, enzyme immunoassay; ELISA, enzyme-linked immunosorbent assay; EPO, erythropoietin; FACS, flow activated cell sorting; GFP,Green Fluorescent Protein; HSA, human serum albumin; Ig, immunoglobulin; IL-1, interleukin-1; KIRA, kinase induced receptor activation; MAb,monoclonal antibody; MGDF, megakaryocyte growth and development factor; mRNA, messenger ribonucleic acid; MTT, [3-(4,5-dimethylthiazol-2-yl-2,5-diphenyltetrazolium bromide]; NAb, neutralizing antibody; RIA, radioimmunoassay; PRCA, pure red cell aplasia; rHuEPO, recombinanthuman EPO; SPA, scintillation proximity assay; TNF, tumor necrosis factor.⁎ Corresponding author. Tel.: +1 805 447 1117; fax: +1 805 499 5842.E-mail address: [email protected] (S. Gupta).

1 The views expressed herein reflect the opinions of the author and not necessarily those of the FDA.

0022-1759/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.jim.2006.12.004

mailto:[email protected]

http://dx.doi.org/10.1016/j.jim.2006.12.004

2 S. Gupta et al. / Journal of Immunological Methods 321 (2007) 1–18

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22. General principles for NAb assay design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1. Principle of a cell-based NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2. Impact of drug's mode of action on NAb assay design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3. Risk-based approach to cell-based NAb testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.4. Utility of matrix interference assays in NAb testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

3. Assay design: considerations and critical components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43.1. Selection of the cell line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43.2. Selection of the assay endpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.3. Positive control antibody . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4. Assay development: assay formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.1. Direct NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.2. Indirect NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74.3. Qualitative NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84.4. Quasi-quantitative NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84.5. Matrix interference assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

4.5.1. Alternative stimulus assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94.5.2. Sample-induced response assay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.5.3. Immunodepletion assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.5.4. Immunocompetition assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

5. Assay optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125.1. Selection of drug product concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

5.1.1. Direct NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125.1.2. Indirect NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

5.2. Selection of sample matrix dilution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135.3. Assay response range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135.4. Assay robustness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135.5. Assay incubation conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

6. Assay qualification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136.1. Characteristics of the drug product standard curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146.2. Characteristics of the positive control antibody curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146.3. Assay cut points for the NAb assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

6.3.1. Blank donor sera . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146.3.2. Ratio of antibody spiked and blank donor sera (post to pre ratio) . . . . . . . . . . . . . . . . . . . 15

6.4. Assay cut point for the matrix interference assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156.4.1. Alternative stimulus assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156.4.2. Sample-induced response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

6.5. Assay sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.6. Assay standardization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1. Introduction

The development of antibodies to biological therapeu-tics upon their administration in study subjects is not anuncommon occurrence (Porter, 2000). On a case-by-casebasis, the antibody response may be inconsequential ormay have a dramatic effect on the efficacy and/or safety ofthe drug product if the antibodies alter drug pharmaco-kinetics or are of a neutralizing nature. Neutralizing anti-

bodies (NAbs) block the biological activity of thetherapeutic molecule by either binding directly toepitope(s) that lie within the active site of the therapeuticmolecule or by blocking its active site by steric hindrancedue to binding to epitope(s) thatmay lie in close proximityto the active site. While, in certain cases NAb presencemay not result in a clinical effect, at sufficient NAb levelsin other cases, a decrease in efficacy may be observedwhich may require administration of higher doses of the

2 Note: These recommendations are based on the experience of theauthors. They reflect scientific concepts that should assist assaydevelopers form a rationale for the development of their specificassay. They are not intended for use in lieu of published FDAregulations and guidance, or direct discussions with the variousregulatory agencies.

3S. Gupta et al. / Journal of Immunological Methods 321 (2007) 1–18

drug product in order to achieve similar efficacy. In certaincases non-neutralizing antibodiesmay also reduce clinicalresponse (Hjelm Skog et al., 2001) therefore it is impor-tant to correlate immunogenicity assay data with clinicaleffects even though the detected antibodies appear to benon-neutralizing in vitro.

Neutralizing antibodies to interferon products, wherethe endogenous protein exists in multiple forms, have beenshown to reduce or abrogate efficacy of the drug product,but have no other apparent adverse effects (Steis et al.,1988). However, NAb formation has a serious outcomewhen the antibodies not only neutralize the administereddrug product but cross-react with and neutralize the biolog-ical activity of a homologous, non-redundant, endogenousprotein. Examples of this phenomenon include the develop-ment of severe thrombocytopenia in healthy volunteersadministered with pegylated MGDF that induced a NAbresponse which cross-reacted with endogenous thrombo-poietin (Li et al., 2001) and the recent occurrence of purered cell aplasia resulting from NAbs against administerederythropoietin (Cassadevall et al., 2002).

Determination of NAb development in non-clinicalstudies is also useful since NAbs prevent the drug productfrom being active, making the presence or absence of anytoxicological findings difficult to interpret. In situationswhere immunogenicity of the drug product poses a patientsafety risk, evaluation of NAb incidence is useful both forthe non-clinical and clinical evaluation of product safety.Therefore, it is important to develop reliable NAb assays ofappropriate sensitivity, specificity, and robustness that candetect and distinguish NAbs from non-neutralizing anti-bodies in non-clinical or clinical samples collected forantibody testing. In certain cases, detected NAbs mayrequire quantitation; however, due to the lack ofappropriate reference standards for immunogenicityassays, any effort to quantitate anti-drug product antibodiesyields a quasi-quantitative result (Mire-Sluis et al. 2004).

Typically, immunogenicity testing for biologicaltherapeutics uses a tiered assay approach (Swanson,2003; 2004; Wadhwa et al., 2003). Samples withdrawnprior to dosing (pre dose) and appropriate post-treatmentsamples (post dose) designated for antibody testing areinitially analyzed using an immunoassay for thepresence of anti-drug product antibodies. The recom-mendations for the design and optimization of theseimmunoassays have been published recently (Mire-Sluis et al., 2004). Samples yielding a positive result inthe immunoassay are subsequently tested for anti-drugneutralizing activity using an in-vitro cell-based assay ora non-cell-based competitive ligand binding assay.

The objective of this manuscript is to provide recom-mendations on the development, optimization, and quali-

fication of cell-based assays for assessing the neutralizingcapacity of anti-drug product antibodies by using a fixedconcentration of drug product in the NAb assay. A back-ground on NAb assay formats, a risk-based approachto assay design, and the choice of potential assayendpoints are presented. In addition, specific con-siderations for parameters for assay optimization andqualification are discussed. Since cell-based assays areextremely sensitive to changes in the cell culturemedium, introduction of test species serum into theassay may cause interference leading to erroneousresults; therefore, a description of matrix interferenceassay formats is also included. It is important to notethat these recommendations may be applicable to amajority of cell-based NAb assays; however, certainsituations may require modified approaches to assaydesign, optimization, and qualification. Recommenda-tions on the validation parameters of cell-based NAbassays will be described in a future manuscript.2

2. General principles for NAb assay design

2.1. Principle of a cell-based NAb assay

An in-vitro cell-based assay provides the mostappropriate biological model for the assessment ofanti-drug product neutralizing antibodies since it mostclosely mimics the mechanism by which neutralizingantibodies may exert their effect in vivo. A cell-basedNAb assay can be defined as an in-vitro assay utilizingcultured cells that interact with or respond to the drugeither directly or indirectly in a measurable manner inthe presence of test species matrix for the detection ofanti-drug product neutralizing antibodies.

The detection of NAbs is based on the principle thatany sample containing NAbs would reduce or abolish thebiological activity associated with a known concentra-tion of drug product used in a cell-based NAb assay.

2.2. Impact of drug's mode of action on NAb assaydesign

The mechanisms by which most biological therapeu-tics exert their desired effect in vivo could be broadlyclassified by their ability to function as agonists or

Table 1Characteristics of drug products that influence the risk of inducing animmune response

Lower risk productcharacteristics

Higher risk product characteristics

No endogenous counterpart Endogenous counterpartRedundant endogenouscounterpart

Non redundant endogenouscounterpart

Immunosuppressed patients Autoimmune patientsSingle dosing Chronic dosing‘Humanized’ product ‘Foreign’ productIntravenous administration SubcutaneousHSA containing HSA free a

Highly pure ImpureNo aggregates Aggregatesa The risk described here refers to the formation of immunogenic

product/HSA adducts. In some cases, the presence of HSA absorbed orprevented interactions of other components with protein products thatcould have led to an immunogenic potential.


antagonists. Biological therapeutics with agonistic prop-erties include cytokines, growth factors, hormones,agonistic monoclonal antibodies, etc. that exert theireffect by directly binding to receptors on the target cellsurface and inducing a measurable response. Biologicaltherapeutics with antagonistic properties include antago-nistic MAbs and soluble receptor based therapeutics thatact by blocking a ligand that binds to the target receptorexpressed on the surface of responsive cells. In simpleterms, agonists induce a response in a direct manner whileantagonists influence a cell's response in an indirectmanner. This distinction plays an important role in cell-based NAb assay design and the choice of critical assaycomponents that comprise the NAb assay system. In thispaper, cell-based NAb assays for therapeutics withagonistic or antagonistic properties have been describedas direct and indirect NAb assays, respectively (Sections4.1 and 4.2). The critical components of a direct NAbassay include: (a) a drug-responsive cell line, (b) the drugproduct, (c) a positive control NAb, and (d) test speciesserum that represents the matrix of the biological samplesthat require testing. The critical components of an indirectNAb assay include (a) a cell line that responds to theligand that is blocked by the drug product, (b) ligand,(c) drug product, (d) positive control NAb and (e) testspecies serum.

2.3. Risk-based approach to cell-based NAb testing

The overall cell-based NAb assay testing schemerequires a risk-based analysis of potential harm topatients should a NAb response occur. Listings of drugproduct characteristics that contribute towards itsimmunogenicity resulting in high or low risk to thepatient are provided in Table 1. For instance, high-riskproducts may be defined as those that have a non-redundant endogenous counterpart. For such products, itis recommended that a sensitive NAb assay that closelymimics the biological activity of the endogenous proteinbe implemented. In addition, it would be necessary todetermine whether antibodies to the drug product cross-react with the endogenous protein. If so, then NAbassays for both the drug product and the endogenousprotein would require implementation. The ability toperform this assessment is dependent upon the avail-ability of sufficient quantities of the purified nativeendogenous protein. Alternatively, a recombinant formof the endogenous protein may be used. For low-riskproducts such as those with no endogenous counterpart(e.g. MAb therapeutics, bacterially derived enzymes),the requirements for NAb assay sensitivity may be moreflexible.

2.4. Utility of matrix interference assays in NAb testing

The introduction of test species serum or plasma canhave a profound effect on the performance of a cell-based assay. Known or unknown factors present in testmatrices could interfere with the performance of cell-based NAb assays by (a) affecting the cells in theirbasal state or in their ability to respond to the drug, or(b) affecting the potency of the drug in a stimulatory orinhibitory fashion. The impact of test species serum onboth these critical NAb assay components could result ina loss of specificity of the NAb assay system. Test matrixoriginating from different subjects and disease popula-tions could introduce additional interference which mayrequire characterization to discern between a true Nabpositive result and non-specific matrix interference. Theuse of a matrix interference assay in the NAb testingstrategy is recommended to assist in differentiatingbetween drug-specific NAbs and other non-specificinhibitory factor(s). Details on matrix interference assayformats are provided in Section 4.5.

3. Assay design: considerations and criticalcomponents

3.1. Selection of the cell line

A good starting point for choosing the cell line for theNAb assay is often the one used for the biological potencyassay. The major advantage of this approach is thatoptimization of the cell line maintenance, cultureconditions, culture media, etc, has already been achieved.However, the adaptability of the potency assay to thepresence of test species serum and response to the drug


product to yield a sufficiently sensitive NAb assayrequires consideration. Ideally, the cell line should yielda functional endpoint upon treatment with drug product.The selected endpoint should represent some aspect of thedrug product's mechanism of action. Optimally, the assayshould be simple to perform and the biological endpointshould be tolerant to test species serum and work over awide range of drug product concentrations. Cell linestransfected with target receptors can also be used and mayoffer the advantage of higher receptor density and thusenhanced signaling. Both natural and engineered cell linesshould be well characterized to ensure responsiveness tothe drug product during continuous culture.

Since NAb assays may be needed for non-clinical,clinical, and post-marketing surveillance studies, theselected line (if several are available to choose from)should be able to tolerate serum from different speciesplanned to be used in the drug development program.The specificity of the cell line's response to the targetligand or drug product in the presence of animal andhuman serum should be evaluated during assaydevelopment to minimize major re-development effortsto support the different types of studies.

The conditions used for routine maintenance of thecell line should be given careful consideration andshould be controlled to obtain consistent results. As anexample, cell passage number can have a major impacton assay variability. As shown in Fig. 1, a decline inassay precision was observed in the behavior of an assaycontrol as cells with higher passage numbers were usedin the assay. It is recommended that a cell bank beestablished early on in the assay development process toensure continuous availability of a consistent cell sourceat the culture age required for optimal assay perfor-

Fig. 1. Effect of cell passage number on precision of a positive control inclneutralizing antibody titers were determined from the weekly analysis of thDirect NAb assay format. Cells were maintained in routine culture starting witpassage 30 through passage 45.

mance. This requires the establishment of appropriateacceptance criteria for cells to allow their use in assays.

3.2. Selection of the assay endpoint

In general, assay endpoints could utilize (i) early or(ii) late biological responses that are triggered in the cellsupon treatment with the drug product or by the drug-inhibitable ligand in NAb assays. Examples of NAb as-say endpoints using early biological responses include(i) binding of the drug product to its target on the cellsurface, (ii) internalization of the drug product, (iii) phos-phorylation of intracellular substrates, (iv) protein traffick-ing, etc. Assay endpoints based on late biological eventsdetect the ability of NAbs to interfere with drug product-induced downstream events such as cell proliferation, celldeath, apoptosis, cell differentiation, or the cellular induc-tion of measurable products such as cytokines or enzymes.Since these assays measure terminal cellular responses, theassay duration may be long and it can take several days toobtain final results. Assay endpoints based on nuclearevents such as changes in mRNA expression or reportergene expression may fall into the early or late categorybased upon the kinetics of drug product-induced expressionof the mRNA or reporter gene. The most commonlyemployed assay endpoints that may be utilized for NAbassays are shown in Table 2.

3.3. Positive control antibody

The positive control antibody for a NAb assay mustneutralize the biological activity of the drug product invitro. The ideal positive control would be antisera fromhumans that mount a NAb response to the administered

uded routinely in an assay. The coefficient of variation (%CV) of thee replicates of an assay positive control (pooled Rhesus antisera) in ah a thaw from a working cell bank at passage 27 and assayed starting at

Table 2Assay endpoints utilized by cell-based NAb assays

Assay endpoint Assay platform NAb action Relevant assay considerations References

Cell surfaceinteractions

Fluorescenceactivated cellsorting (FACS)

• Interferes with binding of drug to targetsite on the cell, or

• Requires labeling of drugwith a fluorescent tag.Important to determine theimpact of labeling of drug onits biological activity

Griffith et al. (1999)

• Affects internalization of drug productinto the cell

Phosphorylationof intracellularsubstrates

KIRA, ELISA • Affects drug-inducedphosphorylation of targetreceptor/substrate, or

• Need cell line that naturallyexpresses or is stablytransfected with the targetreceptor or substrate

(Sadick et al., 1999;Steele-Perkins et al., 1988;Kaplan et al., 1991,Sadick et al., 1997)

• Affects ligand-inducedphosphorylation by blocking drug

• Need phosphorylation site-specific antibodies

Intracellulartrafficking

Fluorescenceimaging

• Affects movement of signalingmolecules (eg. transcription factors, etc.)from one cellular compartment to another

• Need cell line engineeredwith GFP-tagged reportermolecule. GFP fluorescencemay be very susceptible totest species serum

(Ashby et al., 2004;Almholt et al., 2004;Moon et al., 2001)

• Need fluorescently taggedantibodies to the signalingmolecule that redistributeswithin the cell

Cellproliferation

Radioactive(tritiated thymidineuptake)

• Affects cell proliferation inducedby drug, or

• Need cell line that is drug-dependent for growth

(Wei et al., 2004;Mossman, 1983;Byth et al., 2001;Shevach, 1997;Gearing and Bird, 1987;Kitamura et al., 1989)

Non-radioactive(BrdU ELISA,MTT, WST,Alamar bluedye, etc)

• Affects ligand-induced cellproliferation by blocking drug, or

• Response to other stimuliuseful for developing analternative stimulus matrixinterference assay

• Affects anti-proliferative effect of drug • Need cell line that isdependent upon ligand forgrowth• Generally utilizes a factor-independent cell line

Mitchen et al. (1995)

Proteinexpression

ELISA, EIA, RIA,ECL, SPA

• Influences protein expressionby drug, or

• The protein can bemeasured in the cellsupernatant or in cell lysateusing an ELISA

(Musco et al., 1998;Zhu et al., 1999)

• Influences ligand-induced proteinexpression by blocking drug

mRNAexpression

Branched DNAtechnology

• Affects mRNA expression of targetprotein, or

• The mRNA is measured incell lysate which tends to beviscous and may requireoptimization to reduce assayvariability

Sakamoto et al. (2003)

• Affects ligand-induced mRNAexpression of target protein

Reporter geneexpression

Luciferase, β-galactosidase,chloramphenicoltransacetylase(CAT)

• Affects reporter gene expressionby drug product, or

• Need to construct reportergene cell line that canrespond in presence of testspecies serum

(Wu et al., 2003;Littlejohn et al., 2003)

• Affects ligand-induced reporter geneexpression by blocking drug


drug product; however, such antisera are unlikely to beavailable during the assay development phase. Conse-quently, positive controls are generally antisera obtainedfrom hyper-immunized animals. Polyclonal antibodiescan be appropriately purified and spiked back into asuitable matrix at a specific concentration for use as

positive controls. If polyclonal NAbs are not available,monoclonal NAbs can be produced and used as apositive control, either individually or as a “cocktail.”

Evaluation of the utility of the positive controlantibody used for the screening immunoassay forsuitability to act as a NAb assay positive control is

Table 3Types of NAb assay formats and relevant assay controls

Assayformat

Drugproduct'smechanismof action

Examples Assaycomponents

Assaycontrols(in assaymatrix)

Direct Acts directlyon the cells toinduce abiologicalresponse

Therapeuticcytokines,growthfactors,peptides,agonisticMAbs

Cells, drugproduct, testspeciesserum,positivecontrolantibody

(i) Cellsalone(ii) Cells+drugproduct(iii) Cells+drugproduct+positivecontrolantibody

Indirect Acts byinterfering orenhancing thebinding of aligand to areceptor onthe cell surface

TherapeuticMAbs,solublereceptors,etc.

Cells, ligand,drug product,test speciesserum,positivecontrolantibody

(i) Cells alone(ii) Cells+ligand(iii) Cells+ligand+drugproduct(iv) Cells+ligand+drugproduct+positivecontrolantibody


recommended provided sufficient quantities are avail-able. All pertinent documentation such as source,purification protocols, concentration (if measurable)should be procured. Since pools of antibodies from thesame immunized animal may be different, lot-to-lotqualification is essential. Attempts should be made todetermine the stability of the positive control antibodyupon long-term storage and consideration may be givento preparing smaller aliquots to avoid repeated freeze/thaw cycles.

4. Assay development: assay formats

As mentioned in Section 2.2, the choice of NAbassay format depends upon the mode of action of thedrug. Similarly, the format of any matrix interferenceassay is also guided by the influence of the drug producton the cellular response. The consideration of whetherthe sample will be tested in a qualitative or quasi-quantitative mode also determines the choice of assayformat that is most appropriate.

4.1. Direct NAb assay

A direct NAb assay format is used for drug productsthat directly exert a biological effect on the cell (e.g. drugproduct-induced proliferation). This assay format iscomposed of the cell line possessing the appropriatereceptor, the drug product, and the sample matrix (testspecies serum) containing NAbs (Table 3). For example,if a drug product induces cell proliferation, this assay isbased on the principle that if a sample contains detectablelevels of NAb, it would inhibit the response resulting in alowered assay signal (Kelley et al., 2005). In the directNAb assay format, there is potential for interactionsbetween the matrix/NAb and the drug product (i.e. drugproduct binding by antibody), the drug product and thecell line (i.e. drug product/receptor association) and thecell line with the matrix (matrix effects).

Assay controls for direct NAb assays are shown inTable 3. These basic assay controls would allowmonitoring of the interactions between the variouscritical components of the assay. If the drug productcauses an increase in signal, tracking the fold-increase insignal obtained with the control that contains cells anddrug product proves useful in monitoring cell culturequality and/or drug product activity. The control thatcontains cells, drug product, and positive controlantibody would allow monitoring of the reactivity ofthe positive control antibody and the assay overall. Thecontrol that only contains cells is a good indicator ofbasal cell activity in the assay matrix. Aberrant values

may provide an indication of either changes in the assaymatrix that could be having an adverse effect on the cellsor that the cell population may have undergone a driftthat may affect the assay background and/or the abilityof the cells to respond to the drug product.

4.2. Indirect NAb assay

NAb assays for drug products including MAbtherapeutics, soluble receptors, receptor antagonists,etc., often fall into the indirect NAb assay formatcategory where the antigen for the therapeutic antibodyor the ligand for the soluble receptor is required for theassay. An indirect NAb assay for an anti-cytokine MAbwould have to be designed to contain the serum matrix,the cytokine-responsive cell line, the cytokine and thedrug product, all of which have the potential to interactwith each other (Table 3). If the cytokine induces theexpression of a target mRNA in the cell line, the drugproduct would lower the mRNA expression; however, asample containing detectable levels of NAb wouldinhibit the drug thereby allowing the cytokine to inducethe target mRNA, resulting in an enhanced assay signal.In certain cases, it may be possible that a NAb to a


soluble receptor type drug product may interact with thereceptor expressed on the cell surface and produce anagonistic response. An agonistic NAb in this case wouldfunction like a ligand and may escape detection in theNAb assay. If agonistic NAbs are suspected to bepresent in the sample, it may be important to considerincubating the cells only with the serum sample to assessits ability to induce the assay signal in the absence ofadded ligand. Indirect NAb assay formats become evenmore complex if additional biological components arerequired, for example the requirement of a secondaryanalyte to assist in receptor binding of the antigen and/ordrug product, or to synergize the cellular response beingmeasured.

Assay controls that could be utilized in indirect NAbassays are shown in Table 3. If the drug's ability to blockthe action of a ligand on the cell line results in a loweredsignal, tracking the reactivity of the control that containscells, ligand, and drug product relative to the control thatcontains only the cells and ligand would provide anindication of drug product and/or ligand activitybetween assays. As mentioned in the previous section,the control that contains only cells constitutes the assaybackground and is important to monitor.

4.3. Qualitative NAb assay

In this approach, the sample is tested at a singledilution in the assay, generally at the lowest dilution

Fig. 2. A single dilution NAb testing approach for the detection of anti-erythromediated PRCA. Test serum samples were incubated with 1 ng/mL recombina30 minutes. One hundred microliters of each mixture were added to 2×105 stathe assay endpoint (Wei et al., 2004). Assay controls were prepared in the assa(N); (ii) cells+1 ng/mL rHuEPO (M) and (iii) cells+1 ng/mL rHuEPO+900assay cut point. The response from all 7 samples fell below the assay cut poinrHuEPO thereby meeting the criteria for drug product inhibitory activity. Thinterference assay (data not shown) and were not observed to cause any inhibfor drug-specific NAb positive status (Fig. 4).

where the assay matrix interference is minimal, andallows for a higher throughput of sample analysis. Thisapproach is useful for screening samples for neutralizingactivity and typically yields a Positive/Negative resultand is often used for non-clinical studies. Clinicalsamples may also be tested using this approach. In adirect NAb assay, the sample is allowed to incubate withthe drug product before addition to the cells. For anindirect NAb assay, the sample is allowed to incubatewith the drug product followed by incubation with theligand before addition to the cells. The final concentra-tion of the serum, drug, and ligand (if required) in theassay is equivalent to those determined during assayoptimization (Section 5). A sample is considered to haveneutralizing activity in the final validated assay if itinhibits drug product activity above or below the assay'sestablished threshold (Fig. 2).

4.4. Quasi-quantitative NAb assay

Relative quantitation of NAb present in a serumsample can also be determined by performing a dilutionseries and establishing a titer value. This approach canbe useful when one needs to assess changes in NAb overtime. Also titer values allow for further characterizationof potential impact of antibody responses, such aswhether there is a titer threshold above which anassociation with changes in surrogate markers or clinicalparameters could be made. This approach is also useful

poietin neutralizing antibodies in samples from patients with antibody-nt human EPO (rHuEPO) and incubated at room temperature for at leastged cells to conduct a direct NAb assay that utilized cell proliferation asy matrix (5% pooled human/15% rat serum) and included (i) cells aloneng/mL positive control neutralizing antibody (P). The line indicates thet indicating inhibition of the biological activity associated with 1 ng/mLese samples were subsequently tested in an alternative stimulus matrixition below the interference assay endpoint, thereby meeting the criteria


for conducting comparisons of NAb development andincidence between studies. It must be stressed that theantigen–antibody interaction is a dynamic processdependent on the concentration of antibody and drugproduct. In addition, a test serum sample most likelyconsists of a mixture of antibodies, including bindingantibodies, which can vary in class and affinity.Consequently, similar to immunoassays for antibodies,the quasi-quantitative approach provides for a relativeassessment of the amount of NAb present.

Usually, the first step involves an initial dilution ofthe test serum sample with medium plus appropriatetissue culture supplements, however not containing testspecies serum. Subsequently a 2-fold or other appropri-ately selected dilution series is prepared in mediumcontaining pooled test species serum in order tomaintain an equivalent serum matrix across thedilutions. Each dilution of the test sample is incubatedwith a fixed concentration of the drug product (andligand, if applicable) before addition to the cells. Serumsample without drug spike is also included as a controlto determine the background of the sample. If decreasedactivity is detected in the test sample, its neutralizingactivity may ultimately be reported as a titer value,which corresponds to the logarithm or the reciprocal ofthe last dilution before it tests negative (Fig. 3).

A panel of 20–50 normal human donor (or testspecies) serum samples should be analyzed at thevarious dilutions to determine background interferencedue to serum matrix. The mean +/−1.645 SD can be setas a provisional cut point for each dilution. The assayreadout from the patient dilution series is compared to

Fig. 3. A multiple dilution NAb testing approach for characterization of ansubject with antibody-mediated PRCA. Serum sample #1 shown in Fig. 2 walegend for Fig. 2. The sample was tested at 1:50, 1:100, 1:200, 1:400, 1:800, 1assay matrix that was held constant at 5% pooled human/15% rat serum. Tdiamonds). The concentrations of the positive control antibody (in ng/mL) ar1:3200 dilution, the test sample yielded a response that lay above the assay cu

these cut points and the highest dilution that hasneutralizing activity before reaching normal serumassay reactivity is considered the titer. Alternatively, ifthe data is reported as percent neutralization, a generalpercent inhibition of assay response attributable tonormal human serum matrix (based on the panel ofdonor sera) can be established as the assay thresholdusing appropriate statistical approaches and the patientdilution series compared to this assay threshold. Thehighest dilution of the patient's sample that continues toyield inhibition greater than the assay threshold wouldbe considered the titer.

4.5. Matrix interference assays

A matrix interference assay may prove useful indiscriminating a true NAb response from other interferingfactors that may be present in the serum sample that couldmimic a NAb effect in the absence of NAbs. If it isplanned to incorporate a matrix interference assay into thetesting strategy, its development should occur in parallelwith the primary NAb assay (Fig. 4). The following arefour general strategies that could be considered whendeveloping amatrix interference assay. The recommenda-tions for some relevant assay controls for these 4approaches are included in Table 4. Additional novelstrategies and accompanying assay controls for matrixinterference assays may also be possible.

4.5.1. Alternative stimulus assayThis type of matrix interference assay may be

developed with a cell line that elicits the same response

ti-erythropoietin neutralizing antibodies in a sample obtained from as tested at multiple dilutions in the same NAb assay as described in the:1600, 1:3200, and 1:6400 (closed squares). The dilutions were made inhe positive control antibody curve was included in the assay (closede shown within parenthesis. The line represents the assay cut point. Att point, testing negative. A titer of 1:3200 was assigned to the sample.

Fig. 4. Flow chart depicting use of NAb and matrix interference assays to ascertain drug product-specific neutralizing antibodies.


as the drug product in the presence of another stimulant(e.g., another cytokine, growth factor, hormone, etc.).This approach may be employed when using pleiotropiccell lines, which respond to multiple cytokines, e.g., TF-

Table 4Types of matrix interference assays for detection of drug-specific neutralizin

Matrix interferenceassay type

Matrix interference assay format NAbformacompwith

Alternative stimulus Effect of sample on response induced byanother stimulus that the cell line responds toin the same manner as the drug product

Direc

Sample-inducedresponse

Effect of sample on cells in the absence ofadded drug product or ligand

Indire

Immunodepletion Treatment of the sample with animmunoadsorbent followed by retest of theunbound fraction in the NAb assay

Direcindire

Immunocompetition Treatment of the sample with excess drugproduct followed by retest in the NAb assay

Direcindire

aWei et al. 2004.

1 cells (Kitamura et al., 1989). Therefore, whileselecting a cell line during assay development, effortsshould be made to gather as much information aspossible about the ability of the cell line to yield a

g antibodies

assaytatible

Drug productsapplicable to

Assay controls

t Cytokines(e.g. EPOa),growth factors,agonistic Mabs

(i) Cells alone(ii) Cells+alternative stimulus

ct MAbs, solublereceptors, etc.

(i) Cells alone

t orct

All types ofbiologicaltherapeutics

(i) Assay controls(ii) Positive control antibody spikedinto pooled serum andimmunodepleted by immunoadsorbent resin(iii) Positive control antibody spiked intopooled serum and treated withunconjugated resin

t orct

All types ofbiologicaltherapeutics

(i) Cells alone(ii) Cells+drug± ligand(iii) Cells+drug± ligand+positivecontrol antibody(iv) Cells+excess drug± ligand+positivecontrol antibody


similar response as the drug product when treated withanother stimulant. The concentration of the otherstimulant used in the matrix interference assay shouldbe carefully selected and is recommended to be such thatit yields a similar response to that observed with theselected concentration of the drug product in the NAbassay. A sample containing drug product-specific NAbwould be expected only to inhibit the response inducedby the drug product and not the response induced by thealternative stimulus. This type of matrix interferenceassay generally complements direct NAb assays and hasbeen used for the detection of anti-EPO neutralizingantibodies (Wei et al., 2004). However, with NAb assaysthat use drug product-induced proliferation as an assayendpoint, samples from certain subjects may containcytotstatic agents at sufficiently high levels that mayresult in an inhibitory response in the alternativestimulus matrix interference assay, giving rise to thepossibility of false negative results (Fig. 4). In such asituation, it is advisable to use the immunodepletionmatrix interference assay described below (Section4.5.3) which could help discern if the drug productinhibition observed in the NAb assay is attributable to anantibody moiety or not.

4.5.2. Sample-induced response assayThis approach is generally complementary to indirect

NAb assays. In this approach, samples that testinhibitory in the NAb assay may be tested for theireffect on the cells in the absence of added ligand anddrug. Using the example of a MAb drug product thatblocks the action of a cytokine which stimulates theexpression of a target mRNA in the cell line, a samplecontaining detectable levels of NAb would induce anassay signal that corresponds with cytokine-inducedmRNA expression. However, if the sample has elevatedlevels of cytokine due to the disease state of the serum, afalse positive result may be obtained in the NAb assay.Testing the sample for its effect on the cell line in theabsence of added drug or ligand allows the determina-tion if the response observed in the NAb assay isspecific to the drug or not. In cases where there isevidence that the NAb may be agonistic, this matrixinterference assay approach may yield false negativesand it would be advisable to use the immunodepletionmatrix interference assay approach to confirm thepresence of NAbs in the test sample.

4.5.3. Immunodepletion assayThis assay type involves an immunodepletion step in

which a sample that tests inhibitory in the NAb assaywould require a pre-treatment with a solid phase capable

of binding to the heavy and/or light chains ofimmunoglobulin molecules (e.g., Protein G or ProteinL conjugated agarose or sepharose beads). Test samplesare tested with or without treatment with conjugatedresin in the primary NAb assay. The ability todemonstrate that such a pre-treatment removes neutral-izing activity from an inhibitory sample providesverification that the neutralizing activity was due toantibodies. Immunodepletion assays may be useful ifthe other two types of matrix interference assaysdescribed above are not feasible, or yield an unaccept-able rate of false positives or negatives due to seruminterference in the primary NAb assay.

Careful consideration should be given to the choiceof conjugated resin keeping in mind the source of thepositive control antibody used for the assay. Protein Lpossesses better affinity for the human kappa light chainfraction of all human immunoglobulins (Igs) and theirisotypes (De Chateau et al., 1993). Protein A and ProteinG bind well to all human IgG isotypes but have pooraffinity for other Ig subclasses and isotypes (Bjorck andKronvall, 1984; Hakoda et al., 1994; Soundarajan et al.,2005) and therefore may not remove other classes ofimmunoglobulins which may be potentially contributingto a specific neutralizing response. Experiments con-ducted during assay development should demonstratethe effectiveness of the resin for removing the positivecontrol antibody spiked into pooled or individual donorsera.

4.5.4. Immunocompetition assayCertain NAb assays may be coupled to a matrix

interference assay in which the sample is retested in theNAb assay after being spiked with an excess amount ofthe drug product. If the sample contains drug product-specific NAbs, they will be bound to the exogenouslyadded drug product and unavailable to inhibit the assayendpoint in the NAb assay. For these assays, it isimportant to establish the amount of drug to beintroduced into the sample to block the NAbs in areliable and reproducible manner. The extent of NAbinhibition necessary to ascertain the true presence ofNAbs in a sample would also need to be established.

The key point to consider for this approach is theamount of drug product that is utilized in the NAb assay.If the amount of drug product used in the NAb assaycorresponds to the concentration that yields maximum(∼ 100%) response in the dose response curve, thismatrix interference assay approach may work. However,if the concentration of drug product lies within the lowerregion of the dose response curve (which is recom-mended for NAb assays), this approach may give


confounding results, since the addition of exogenousdrug in addition to blocking the NAbs may exert its owneffect on the cells and/or ligand, thereby yielding datathat may be difficult to interpret.

5. Assay optimization

Following development of the prototype assay, it isimportant to follow a rigorous assay optimization phaseto arrive at a final robust assay. To facilitate assayoptimization, one might consider employing a factorialdesign approach. Such a mathematical approach allowsone to evaluate assay performance and assay robustnessand to optimize assay parameters under diverse condi-tions with minimal resources. Some considerations thatare useful for cell-based assays in general are included inTable 5. Considerations specific for cell-based NAbassays are described below.

5.1. Selection of drug product concentration

5.1.1. Direct NAb assayThe sensitivity of a direct NAb assay depends upon

the concentration of drug product that will be used in theassay for neutralization by test samples. As a general

Table 5General assay optimization considerationsfor cell-based assays

Parameter Considerations Suggestions for optimization

Cellpreparation

• Cell bank Create a cell bank with cells th• Cell number Perform a 2-fold serial dilution

background and (ii) maximal re• Viability Monitor viability using a micro

as they recover from freeze-thaw• Morphology Perform regular visual inspecti

advisable to initiate a new batc• Density Establish a cell culture density

product• Confluence at timeof assay

This feature is pertinent to adhability to respond to stimuli. Ugive high background values ingive an expectedresponse to drug product

• Passage number Monitor cell response to drug pEstablish the range of passage

• Cell pre-treatment Evaluate usefulness of synchroWashingconditions

• Adherent cells Monitor tightness of cell adherecell loss and to control assay v

• Non-adherent cells Use gentle centrifugation to w96-wellassay plates

• Clear Useful for plating/growing cel• Black Useful for fluorescence-based• White Useful for luminescence-based

Assay plateconfiguration

• Placement of samplesand controls

Avoid edge wells, include mul

rule, the amount of drug product used in the assayshould be the lowest possible concentration that elicits≥50% response and lies on the linear portion of the doseresponse curve. This may be evaluated by testing theeffect of the drug at several concentrations in the assaymatrix, preferably, so that both upper and lower plateauscan be observed.

5.1.2. Indirect NAb assayThe sensitivity of an indirect NAb assay depends

upon the concentration of ligand that will be used in theassay, which further facilitates the selection of the drugproduct concentration that will be used for evaluation ofneutralization by test samples. Similar to the considera-tions employed for the selection of drug productconcentration in a direct NAb assay, the amount ofligand used in the assay may be the lowest possibleconcentration that elicits ≥50% response. This may beevaluated by testing the effect of the ligand at severalconcentrations in the assay matrix. The lowest concen-tration of the ligand that yields ≥50% of the totalresponse may be a suitable one to select. Next, oneshould test the ability of the drug product to inhibit theselected ligand concentration in the assay matrix usingseveral, ideally using a concentration range, so that both

at have not been subjected to extensive passagingtitration series to determine the cell number that yields (i) lowsponse to drug productscope and standard cell staining techniques; check performance of cells, select viability state that yields maximal response to drug producton of cells, if unexpected change in morphology is noticed it ish of cultured cellsduring routine cell maintenance that yields optimum results to drug

erent cells that prefer a certain level of confluence to maintain theirnder-confluence may yield a low response while over-confluence maythe assay. Establish the confluence level that yields cells which

roduct periodically during the span of cell culture for a particular batch.numbers that give optimum results to the drug productnizing cell culture (serum-starving) to obtain optimal resultsnce to the plate if washing is involved. Use gentle washing to avoid anyariabilityashls in a colorimetric assayassaysassaystiple replicates (≥3) of samples and controls to control assay precision


upper and lower plateaus are achieved. The concentra-tion of the drug product that alters ligand function by atleast 50% may be a suitable concentration to select.

5.2. Selection of sample matrix dilution

Determination of the appropriate dilution of assaymatrix is an important part of NAb assay optimizationbecause this will dictate the minimal test sample dilutionand therefore the assay sensitivity. The considerationsfor defining the minimum dilution used for the antibodyscreening immunoassay described by Mire-Sluis et al.(2004) may also be applied to cell-based NAb assays.The effect of sample matrix on the ability of the cells torespond to drug product or ligand should be evaluated atmultiple dilutions preferably by different pools of thetest species serum. The dilution of sample matrix thathas a minimal effect on the cellular response should beselected and further evaluated using the positive controlantibody spiked into sample matrix. A NAb assayshould be able to detect the antibody and to differentiateit in the presence of assay matrix components thatmay be expected to be present. These might includecomplement, coagulation factors, soluble receptors,lipids, concomitant medications, the endogenous ho-mologous counterpart, as well as the administered drugproduct.

Factors such as soluble receptors, that either bindthe drug product, or act directly on the cells, caninterfere in the assay, leading to false positive results.Alternatively, certain factors may alter the responseof the cells in such a manner that may mask the pres-ence of a neutralizing antibody in the sample. Matrixinterference assays can help to determine the presenceof such interfering substances (Section 4.5). Assess-ment of matrix effects during assay development canbe accomplished using pooled sera. Ideally, the assaymatrix should be defined using multiple individualsera due to heterogeneity of the various possible inter-fering substances. Diseased state sera may contain ad-ditional factors that may be encountered during analysisof study samples and whenever possible should beevaluated.

If lipemic, hemolyzed, incompletely clotted andpreferably disease state sera from naïve subjects areavailable they should be screened in the assay both withand without the addition of antibody. If the unspikedsera give a response in the assay and/or the serainterferes in the detection of the antibody, the analyticalprocedure should state that samples compromised insuch a manner may not yield reliable results and mayhave to be excluded from testing.

5.3. Assay response range

The observed signal to background ratio of thecellular response to the drug product and ligand in directand indirect NAb assays, respectively, and the variabil-ity associated with the ratio, can provide an earlyindication of the assay's performance. Assays demon-strating ≤ 5-fold change in signal upon treatment withthe selected drug product or ligand concentration inpooled serum might not give the range of assay responsenecessary to observe an inhibitory effect due to NAbs.The evaluation of the Z′ factor for the drug product/ligand-induced cellular response may be useful forassessing assay performance quality. The Z′ factor is acoefficient that takes into account the dynamic range ofthe assay signal and the data variation associated withthe signal measurements (Zhang et al., 1999).

5.4. Assay robustness

The robustness of the NAb assay should be gaugedduring assay development. The variability of the drugproduct dose response curve, its range or shape from dayto day and/or the behavior of assay controls under minordeliberate changes in certain assay conditions (e.g.incubation time) may provide some early information ondeveloping strategies for the final assay format.

5.5. Assay incubation conditions

For NAb assays it is important that the positivecontrol antibody or the sample be allowed to incubatewith the drug product for an established period of time atan appropriate temperature to allow binding to occur,before addition of the mixture to the cells. A range ofincubation times and temperature may be studied todetermine the conditions that favor these binding eventsas well as those that are optimum for maximal cellresponse to the drug product or ligand. In general, anincubation of 30 min to an hour at room temperature or37 °C is usually sufficient to allow drug–antibodybinding to occur; however, this would need to beevaluated for each assay.

6. Assay qualification

Assay development and optimization may be fol-lowed or integrated with a qualification phase thatallows a better understanding of assay performanceand determines the ability of the assay to stand the testof validation. The experiments performed duringthis qualification phase allow the emergence of the


acceptance criteria to be included in the validationprotocol. The information gathered during the assayqualification phase can be extremely useful for draftingthe NAb analytical procedure that will ultimatelyundergo validation. The types of experiments that canbe performed during this phase are described below.

6.1. Characteristics of the drug product standard curve

The primary NAb assay format relies on the ability ofthe cells to respond to the drug product. Hence the drugproduct dose response relationship should be wellcharacterized in the selected assay matrix. The mathe-matical fit that best describes the dose responserelationship may be explored. Usually a 4-parameterlogistic fit aptly describes the dose response relationshipfor cell-based assays. A 5-parameter logistic fitgenerally provides a better fit for dose–response curveswith asymmetrical asymptotes and/or inconsistentvariances across the dose range. An estimate of theasymptotes, slope, linear region and ED50 functions maybe obtained using the selected mathematical model. Theestimates for accuracy and precision of the individualdrug product concentrations that comprise the standardcurve standards may be derived. The estimates obtainedfor the drug product concentration selected for routineuse in the NAb assay (control consisting of cells+drugproduct ± ligand; Table 3) allows the setting of thedesirable acceptance criteria for this control in routineuse.

6.2. Characteristics of the positive control antibodycurve

The positive control antibody should be titrated tocharacterize the drug product inhibitory profile of theantibody. A dose-dependent inhibition of the selecteddrug product concentration should be evident. Thepositive control antibody curves should be prepared inneat serum at selected concentrations and tested for theirability to inhibit the drug product at the selected samplematrix concentration. The repeatability of the positivecontrol antibody curves could be assessed in pooled and/or individual donor sera by determining the (i) ED50,(ii) ED90, etc. Precision estimates should be made on thecritical portions of the curves depending on the assayformat. Separately prepared quality control samplesconsisting of antibody concentrations that are notincluded in the titration curve may be included. TheED50 provides an estimate of the concentration ofantibody required to cause a 50% inhibition of theselected drug product in pooled serum and may provide

an early indication of assay sensitivity in that particularassay matrix. The precision of the various antibodyconcentrations is usually a useful indicator of assayperformance in the assay matrix.

6.3. Assay cut points for the NAb assay

A NAb assay cut point may be defined as the level ofresponse of the assay above or below which a sample isdefined to be negative or positive for neutralizingactivity towards the drug product. Establishment of astatistically derived assay cut point for the NAb assayallows the emergence of a criterion to be used forterming samples “positive” or “negative” for the pre-sence of neutralizing activity.

6.3.1. Blank donor seraA limited number of drug naïve individual healthy

donor or target untreated diseased state sera (n=10 to20, if possible) should be screened in the assay toestablish a preliminary threshold response value thatserves as the assay cut point. The cut point value may beobtained using appropriate statistical approaches: para-metric, non-parametric, or a robust alternative. It isimportant to point out that the assay cut point for cell-based NAb assays must take into account the effect ofthe expected NAbs on the assay endpoint. If the drugproduct were a cytokine that induces proliferation, aNAb would result in reduced proliferation or reducedassay signal in the NAb assay. For such an assay, usingparametric or non-parametric approaches, lower boundof a one-sided 95% prediction interval may serve as theassay cut point for the tested population. If the drugproduct were a MAb therapeutic that blocks ligand-induced response, a NAb would result in increasedresponse or an enhanced assay signal. For such anassay, using parametric or non-parametric approaches,the upper bound of a one-sided 95% prediction intervalmay serve as the assay cut point for the tested popu-lation. An outlier test may be performed to removeaberrant values to ensure that the assay cut pointdoes not yield a high number of false positives or falsenegatives.

The screening of multiple donor sera in the NAbassay in this manner also allows an early assessment ofthe robustness of the assay response in the presence ofsera from different sources, thereby qualifying theappropriateness of the selected serum matrix. If highvariability of response is observed within the testeddonors, it is recommended to evaluate a more diluteassay matrix, within limits so that assay sensitivity isnot compromised.


6.3.2. Ratio of antibody spiked and blank donor sera(post to pre ratio)

If the assay cut point developed with blank donorsera shows too much inter-donor variability that cannotbe overcome with further dilution of the assay matrix, analternative approach might be acceptable. The concen-tration of positive control antibody that shows aconsistent 1.5 to 2 fold inhibition of the drug productin the experiments conducted to qualify the antibody(Section 6.2) may be spiked into 10 to 20 individualdonor sera. These spiked sera (surrogate post doses)may be tested in the NAb assay along with theirunspiked counterparts (surrogate pre doses) and theresults obtained may be used to compute post/pre ratiosfor this donor population. The lower bound of a one-sided 95% prediction interval for these post/pre ratiosmay be used as the assay cut point. If false negatives arestill obtained, the lowest post/pre ratio obtained from thetested population may be used as the assay cut point.The underlying goal should be to obtain an assay thathas the lowest possible rate of false negatives. For high-risk products, false negatives should not be acceptableand if the assay continues to yield false negatives, effortsshould be made to evaluate the critical components ofthe assay and if needed a new assay with a different cellline may need to be designed.

6.4. Assay cut point for the matrix interference assay

The matrix interference assay cut point may bedefined as the level of response of the assay above orbelow which a sample may be deemed to have non-specific activity that could artificially mimic a NAbresponse.

6.4.1. Alternative stimulus assayThe ability of the cells to respond to another stimulus

that yields a similar response to that obtained by treatingthe cells with the drug product should be wellcharacterized in the selected assay matrix. Similar tothe primary NAb assay where the drug productconcentration is held constant, the concentration of theother stimulant should be held constant in the matrixinterference assay format. For the assay cut point, 10 to20 blank donor sera may be tested in the assay with theselected concentration of the other stimulant to derivethe mean response level of the tested population. Thisapproach to assessing if any matrix interference hasoccurred complements direct NAb assays used forcytokine and growth factor-based drug products that(i) generally require factor-dependent cell lines (Table 4),and (ii) the presence of a NAb is associated with a drop

in the assay signal. Therefore, using parametric or non-parametric approaches described in Section 6.3.1, thelower bound of a one-sided 95% prediction interval,may serve as the matrix interference assay cut point forthe tested population. If a sample yields an assayresponse that lies below the matrix interference assaycut point it would be considered inhibitory to the otherstimulant, thereby indicating that the sample containsnon-specific inhibitory activity (Fig. 4). A sample thatcontains drug-specific NAbs and is relatively free ofnon-specific serum factors that could affect the assayendpoint, would be expected to yield an assay responsevalue that lies above the matrix interference assay cutpoint.

6.4.2. Sample-induced responseA limited number of blank naïve individual healthy

donor or target naïve diseased state sera (n=10 to 20, ifpossible) should be screened with the cells, in the assaymatrix. The addition of drug or ligand is not required inthis approach. Similar to the approach used for derivingthe NAb assay cut point with blank donor sera (Section6.3.1) the matrix interference assay cut point value forthis type may be obtained using appropriate statisticalapproaches: parametric, non-parametric, or a robustalternative. Since this matrix interference assay is moresuited for indirect NAb assays in which the presence ofNAb correlates with an enhanced assay signal, usingparametric or non-parametric approaches as described inSection 6.3.1, the upper bound of a one-sided 95%prediction interval may serve as the assay cut point forthe tested population. If a sample contains non-specificactivity that mimics a NAb response in the assay, itwould yield a response value that lies above the matrixinterference assay cut point. However, if a samplecontains drug-specific NAb and lacks any confound-ing non-specific factors, it should yield a response valuethat lies below the matrix interference assay cut point(Fig. 4). However, as mentioned earlier, agonistic NAbselicited to soluble receptor therapeutics may yield falsenegative results in these circumstances.

It has been observed with drug products that requireindirect NAb assays (e.g., IL-1- or TNF-blockingtherapies), occasionally the naïve untreated diseasestate serum may have high levels of ligand (IL-1 orTNF, respectively) that may yield high response valuesin the corresponding matrix interference assay where thesample is incubated with the cell (sample-inducedresponse matrix interference assay, Table 4). In such asituation, high inter-donor variability may be observedwhile deriving the assay cut point. An alternativeapproach that compares the behavior of the sample in

Table 6Summary of key activities undertaken during assay design, optimization and qualification of cell-based NAb assays

Assay design Assay optimization Assay qualification Assay validation parameters

• Choose a cell line • Select drug productconcentration

• Repeatability of the drug product standardcurve

• Assay cut points

• Select an assay endpoint • Select assay matrix • Repeatability of the positive control antibody • Sensitivity• Obtain drug product supply • Determine assay

response range• Preliminary assay cut points for the NAb andmatrix interference assays

• Precision

• Obtain supply of positive controlantibody

• Set assay controls • Assay sensitivity • Robustness

• Obtain adequate supply of test speciesserum

• Gauge assayrobustness

• Interference

• Test if positive control antibody hasneutralizing properties

• Set assay incubationconditions

• Stability of the positivecontrol antibody

• Choose NAb assay format • Optimize washing/lysis conditions

• Ruggedness

• Choose matrix interference assayformat

• Optimize assay plateconfiguration


the NAb and the matrix interference assays could beadopted. This can be accomplished by deriving ratiosof the response obtained in the NAb assay and thematrix interference assay (NAb/Matrix interferenceassay). However, this approach requires parallel test-ing of the samples in the NAb and matrix interferenceassays. Statistical approaches similar to those usedto derive the NAb assay cut point described in Section6.3.1 with blank donor serum may be utilized to obtainthe NAb/Matrix interference assay cut point. Typi-cally, if a sample contains clinically relevant levelsof NAb, the response obtained in the NAb assayshould exceed the response observed in the interfer-ence assay.

6.5. Assay sensitivity

Upon establishment of the assay cut point, thepositive control antibody curves prepared in pooled ordisease state (if available) neat serum and tested at thesample matrix concentration should be examined toidentify the lowest antibody concentration that yields apositive result. This concentration serves as the initialassessment of the assay's sensitivity. An assessment ofthe assay sensitivity should also be made in pooled andindividual naïve donor serum.

The reliability of the assay to detect the concentrationof the positive control antibody established above at theassay's level of sensitivity should be assessed. This maybe accomplished by spiking that particular concentrationof the antibody into a limited number of individual donorsera. If, due to serum interference, all spiked samples donot test positive, attempts should bemade to determine theconcentration of antibody that consistently tests positive

in the assay. This concentration can serve as a moredefinitive assessment of the assay's sensitivity.

6.6. Assay standardization

Since reference material or standards are not readilyavailable for most immunogenicity assays, it is verydifficult to compare antibody results among laborato-ries. Assay formats can vary and may significantly differin sensitivity, specificity and precision from one lab tothe next. This is particularly relevant for cell-basedassays where different cell lines, passage numbers, andserum matrixes can influence assay results. Proficiencytesting or Alternative Assessment procedures, asproposed by the Clinical and Laboratory StandardsInstitute (www.nccls.org), can be considered as animportant part of quality management. Split-sampleprocedures can be implemented to obtain comparativeresults. This approach calls for the same samples to beanalyzed by participating laboratories to evaluate inter-laboratory agreement and testing errors.

7. Conclusions

Cell-based NAb assays provide a valuable tool forthe detection of anti-drug product neutralizing anti-bodies elicited to biological therapeutics. These assaystend to be complex; therefore this manuscript focusedon providing information that would aid the assaydeveloper during assay development, optimization, andqualification. The key activities that may be undertakenduring these 3 phases have been summarized in Table 6.Assay validation ensues upon the completion of assayqualification. Some validation parameters that have

http://www.nccls.org


been discussed among the authors while working on thismanuscript are included in Table 6.

Acknowledgements

The Ligand Binding Assay Bioanalytical FocusGroup of AAPS sponsored this work. The authorswould like to thank members of Amgen's ClinicalImmunology group for contributing data and criticalreview of the manuscript, and Dr. Viswanath Devanara-yan (Merck) for helpful discussion on statisticalapproaches.

References

Almholt, K., Tullin, S., Skyggebjerg, O., Scudder, K., Thastrup, O.,Terry, R., 2004. Changes in intracellular cAMP reported by aredistribution assay using a cAMP-dependent protein kinase-greenfluorescent protein chimera. Cell. Signal. 16, 907.

Ashby, M.C., Ibaraki, K., Henley, J.M., 2004. It's green outside:tracking cell curface proteins with pH sensitivie GFP. TrendsNeurosci. 27, 257.

Bjorck, L., Kronvall, G., 1984. Purification and some properties ofStreptococcal protein G, a novel IgG-binding reagent. J. Immunol.133, 969.

Byth, H.-A., Mchunu, B.I., Dubery, I.A., 2001. Assessment of asimple, non-toxic Alamar blue cell survival assay to monitortomato cell viability. Phytochem. Anal. 12, 340.

Cassadevall, N., Nataf, J., Viron, B., Kolta, A., Kiladjian, J.-J.,Martin-Dupont, P., Michaud, P., Papo, T., Ugo, V., Teyssandier,I., Varet, B., Mayeux, P., 2002. Pure red-cell aplasia and anti-erythropoietin antibodies in patients treated with recombinanterythropoietin. N. Engl. J. Med. 346, 469.

De Chateau, M., Nilson, B.H., Erntell, M., Myhre, E., Magnusson, C.G.,Akerstrom, B., Bjorck, L., 1993. Scand. J. Immunol. 37, 399.

Gearing, A.J., Bird, C.B., 1987. In: Clemens, M.J., et al. (Ed.),Lymphokines and Interferons, a Practical Approach. IRL Press.

Griffith, T.S., Rauch, C.T., Smolak, P.J., Waugh, J.Y., Boiani, N., et al.,1999. Functional analysis of TRAIL receptors using monoclonalantibodies. J. Immunol. 162, 2597.

Hakoda, M., Hayashimoto, S., Yamanaka, H., Terai, C., Kamatani, N.,Kashiwazaki, S., 1994. Molecular basis for the interaction betweenhuman IgM and Staphylococcal protein A. Clin. Immunol.Immunopathol. 72, 394.

Hjelm Skog, A.L., Wadhwa, M., Hassan, M., Gharizadeh, B., Bird, C.,Ragnhammar, P., Thorpe, R., Mellstedt, H., 2001. Alteration ofinterleukin 2 (IL-2) pharmacokinetics and function by IL-2 antibodiesinduced after treatment of colorectal carcinoma patients with acombination of monoclonal antibody 17–1A, granulocyte macro-phage colony stimulating factor, and IL-2. Clin. Cancer Res. 7, 1163.

Kaplan, D.R., Hempstead, B.L., Martin-Zanca, D., Chao, M.V.,Parada, L.F., 1991. The trk proto-oncogene product: a signaltransducing receptor for nerve growth factor. Science 252, 554.

Kelley, M., Cooper, C., Matticoli, A., Greway, A., 2005. The detectionof anti-erythropoietin antibodies in human serum and plasma PartII. Validation of a semi-quantitative 3H-thymidine uptake assay forneutralizing antibodies. J. Immunol. Methods 300, 179.

Kitamura, T., Tange, T., Terasawa, T., Chiba, A., Kuwaki, T.,Miyagawa, K., Piao, Y.F., Miyazono, K., Urabe, A., Takaku, F.,

1989. Establishment and characterization of a unique human cellline that proliferates dependently on GM-CSF, IL-3 or erythro-poietin. J. Cell. Physiol. 140, 323.

Li, J., Yang, C., Yuping, X., Bertino, A., Glaspy, J., Roberts, M., et al.,2001. Thrombocytopenia caused by the development of antibodiesto thrombopoietin. Blood 98, 3241.

Littlejohn, A.F., Tucker, S.J., Mohamed, A.A., McKay, S., Helms, M.J., Vandenabeele, P., MacEwan, D.J., 2003.Modulation of caspasesfor tumor necrosis factor-stimulated c-Jun N-terminal kinaseactivation but not nuclear factor-kappaB signaling. Biochem.Pharmacol. 65, 91.

Mire-Sluis, A.R., Barrett, Y.C., Devanarayan, D., Koren, E., Liu, H.,Maia, M., Parish, T., Scott, G., Shankar, G., Shores, E., Swanson,S.J., Taniguchu, G., Wierda, D., Zuckerman, L.A., 2004.Recommendations for the design and optimization of immunoas-says used in the detection of host antibodies against biotechnologyproducts. J. Immunol. Methods 289, 1.

Mitchen, J., Bletzinger, D., Rago, R., Wilding, G., 1995. Use of a DNAmicrofluorometric assay tomeasure proliferative response ofmink lungcells to purified TGF-beta and to TGF-beta activity found in prostatecell conditioned medium. In Vitro Cell. Dev. Biol., Anim. 31, 692.

Moon, K.-Y., Han, B.-S., Lee, J., Kim, Y.S., 2001. A cell-based assaysystem for monitoring NF-kB activity in human HaCaT transfec-tants cells. Anal. Biochem. 292, 17.

Mossman, T., 1983. Rapid colorimetric assay for cellular growthand survival: Application to proliferation and cytotoxicity assay.J. Immunol. Methods 65, 55.

Musco, M.L., Cui, S., Small, D., Nodelman, M., Sugarman, B., Grace,M., 1998. A comparison of flow cytometry and laser scanningcytometry for the intracellular evaluation of adenoviral infectivityand p53 protein expression in gene therapy. Cytometry 33, 290.

Porter, S., 2000. Human immune response to recombinant humanproteins. J. Pharm. Sci. 90, 1.

Sadick, M.D., Galloway, A., Shelton, D., Hale, V., Week, S., Anicetti,V., Wong, W.L.T., 1997. Analysis of neurotrophin/receptorinteractions with a gD-flag-modified quantitative kinase receptoractivation (gD.KIRA) enzyme linked immunosorbent assay. Exp.Cell. Res. 234, 354.

Sadick, M.D., Intintoli, A., Quarmby, V., McCoy, A., Canova-Davis, E.,Ling,V., 1999.Kinase receptor activation (KIRA) a rapid and accuratealternative to end-point bioassays. J. Pharm. Biomed. Anal. 19, 883.

Sakamoto, S., Yokoyama, M., Prakash, K., Tsuruha, J., Masamoto, S.,Getzenberg, R.H., Kakehi, Y., 2003. Development of quantitativedetection assays for CYR61 as a new marker for benign prostatichyperplasia. J. Biomol. Screen. 8, 701.

Shevach, E., 1997. Labeling cells in microtiter plates for determinationof [3H]thymidine uptake. Current Protocols Immunol. A.3D.1.

Soundarajan, S., Kikuchi, Y., Joseph, K., Kaplan, A.P., 2005.Functional assessment of pathogenic IgG subclasses in chronicautoimmune urticaria. J. Allergy Clin. Immunol. 115, 815.

Steele-Perkins, G., Turner, J., Edman, J.C., Hari, J., Pierce, S.B.,Stover, C., Rutter, W.J., Roth, R.A., 1988. Expression andcharacterization of a functional human insulin-like growth factor1 receptor. J. Biol. Chem. 263, 11486.

Steis, R.G., Smith, J.W., Urba, W.J., Clark, J.W., Itri, L.M., Evans, L.M.,Schoenberger, C., Longo, D.L., 1988. Resistance to recombinantinterferon alfa-2a in heavy-cell leukemia associatedwith neutralizinganti-interferon antibodies. N. Engl. J. Med. 318, 1409.

Swanson, S.J., 2003. New technologies for the detection of antibodiesto therapeutic proteins. In: Brown, F., Mire-Sluis, A.R. (Eds.),Immunogenicity of Therapeutic Biological Products, Dev. Biol.,vol. 112, p. 127.


Swanson, S.J., 2004. Evaluation of methods to detect and characterizeantibodies against recombinant human erythropoietin. NephronClin. Pract. 96, 88.

Wadhwa, M., Bird, C., Dilger, P., Gaines-Das, R., Thorpe, R., 2003.Strategies for detection, measurement and characterization of un-wanted antibodies induced by biological therapeuetics. J. Immunol.Methods 278, 1.

Wei, X., Swanson, S.J., Gupta, S., 2004. Development andvalidation of a cell-based bioassay for the detection ofneutralizing antibodies against recombinant human erythropoi-etin in clinical studies. J. Immunol. Methods 293, 115.

Wu, K., Yang, C., Davoli, M.A., D'Amico, M., Li, A., Cveklova, K.,Kozmik, Z., Lisanti, M.P., Russell, R.G., Cvekl, A., Pestell, R.G.,

2003. DACH1 inhibits transforming growth factor-beta signalingthrough binding Smad4. J. Biol. Chem. 278, 51673.

Zhu, J., Grace, M., Casale, J., Chang, A.T.-I., Musco, M.L., Bordens,R., Greenberg, R., Schaefer, E., Indelicato, S., 1999. Character-ization of replication-competent adenovirus isolates from large-scale production of a recombinant adenoviral vector. Hum. GeneTher. 10, 113.

Zhang, J.-H., Chung, T.D.Y., Oldenburg, K.R., 1999. A simplestatistical parameter for use in evaluation and validation of highthroughput screening assays. J. Biomol. Screen. 4, 67.

recommendations for the design, optimization, and qualification of cell-based assays used for the...

Documents