- Email: [email protected]
An Affinity Capture Elution (ACE) assay for detection of anti-drug antibody to monoclonal antibody therapeutics in the presence of high levels of drug
Journal of Immunological Methods 327 (2007) 10 – 17 www.elsevier.com/locate/jim
Research paper
An Affinity Capture Elution (ACE) assay for detection of anti-drug antibody to monoclonal antibody therapeutics in the presence of high levels of drug James S. Bourdage ⁎, Carolyn A. Cook, Daphne L. Farrington, Jana S. Chain, Robert J. Konrad Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN 46285, United States Received 13 December 2006; received in revised form 25 May 2007; accepted 5 July 2007 Available online 27 July 2007
Abstract Monoclonal antibody therapeutics typically have relatively long half-lives and can be dosed at high levels. Although formation of anti-drug antibodies (ADA) is relatively rare, detection of these antibodies can be very difficult in the presence of high circulating levels of drug. Typically these ADA are detected by bridging ELISAs which can be very sensitive to even low levels of drug. We describe an ELISA method based on affinity capture of ADA on solid-phase drug followed by removal of excess free drug, release and transfer of bound ADA and subsequent detection using biotinylated drug. The assay is both sensitive and highly tolerant to free drug with detection of 500 ng/ml of ADA readily achieved in the presence of 500 μg/ml of drug. © 2007 Elsevier B.V. All rights reserved. Keywords: Antibody assay; Acid dissociation; Immunogenicity; Therapeutic proteins; ELISA; Antigen interference
1. Introduction Monoclonal antibody therapeutics are finding increased usage for treatment of a wide variety of clinical indications with over 18 approved products. The incidence of anti-drug antibodies (ADA) varies widely from almost 100% for some murine monoclonals to virtually undetectable for some humanized antibodies (Bourdage Abbreviations: ELISA; Enzyme-Linked Immunosorbent Assay; ADA; anti-drug antibody; ACE; affinity capture elution; TMB; tetramethyl benzidine; OD; optical density; HRP; horseradish peroxidase. ⁎ Corresponding author. Eli Lilly Company, 2001 W. Main St., GL55, Greenfield, IN 46140, USA. Tel.: +1 317 277 4139; fax: +1 317 276 5281. E-mail address: [email protected] (J.S. Bourdage). 0022-1759/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2007.07.004
et al., 2005). These antibodies can have significant clinical consequence. Bendtzen et al. reported detection of antiinfliximab antibodies that apparently caused loss of bioavailability, infusion reactions and treatment failure (Bendtzen et al., 2006). Anti-rituximab antibodies caused reduced B-cell depletion and reduced blood levels of rituximab (Looney et al., 2004). There have been cases where neutralizing antibodies to endogenous proteins have caused life-threatening situations (Casadevall et al., 2002; Basser et al., 2002). Humanized monoclonal antibodies typically have half-lives of 10–20 days and can often be administered at relatively large doses leading to blood levels of more than 500 μg/ml (Maloney et al., 1997). Preclinical studies typically use even higher doses to establish safety
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
margins with necessary short washouts prior to evaluation. Bridging or double antigen format assays are often used to evaluate ADA for monoclonals due to their ability to detect multiple classes of antibody (Feldman et al., 2003: Pendley et al., 2003) . However, there is concern for the ability to detect low affinity antibody due to the requirement for monovalent binding to the solid phase and labeled drug during bridge formation. This monovalent binding also causes the assay to be very sensitive to free drug since blocking of only a single arm of the ADA molecule will prevent bridge formation. A double antigen format assay described for infliximab has drug interference at drug levels above 1.4 μg/ml (Baert et al., 2003). This presents a dilemma when developing preclinical and clinical protocol sampling points since antibodies may be undetectable while drug levels are high thus causing a requirement for washouts of several weeks to several months. These potentially long washout periods could in turn cause decreasing ADA titers, thus making detection of ADA difficult or impossible. We have therefore developed a method that first dissociates ADA-free drug complexes with acid treatment followed by neutralization in the presence of solid-phase drug giving the ADA an opportunity to be affinity captured. After washing away excess free drug, ADA are eluted off with acid and subsequently bound to a fresh solid surface. Bound ADA are subsequently detected by addition of biotinylated drug followed by streptavidinHRP and substrate. Results presented here indicate that this affinity capture elution (ACE) assay format is capable of detecting low ng/ml levels of ADA in the presence of a 1000-fold excess of free drug. 2. Materials and methods 2.1. Reagents Wash buffer consists of 0.01 M Tris-buffered saline (TBS) (Fisher Scientific, Fair Lawn, NJ) with 0.05% Tween 20 (Sigma-Aldrich, St. Louis, MO) (TBST). Therapeutic monoclonal antibody (LY) was supplied by Eli Lilly and Co. (Indianapolis, IN). Affinity purified rabbit anti-human IgG (AffiniPure Rabbit Anti-Human IgG (H + L), was obtained from Jackson ImmunoResearch Laboratories, Inc, West Grove, PA. Poly HRP Streptavidin (SA-HRP) was obtained from Pierce (Pierce Biotechnology, Rockford, IL, N200).
11
tion Kits (Pierce, 21435) according to manufacturer's instructions. Briefly, LY was dialyzed overnight against phosphate buffered saline, combined with Sulfo-NHSLC-Biotin in a 20:1 molar ratio and incubated for 1 hour at room temperature. The reaction mixture was dialyzed vs. PBS overnight. Protein content was analyzed using the BCA Protein Assay (Pierce, 23225) and was then diluted with an equal volume of glycerol and stored at − 20 °C. 2.3. ACE assay ELISA plates (Nunc Maxisorp 96 Microplates, Nunc, Rochester, NY) were coated with therapeutic humanized monoclonal antibody (LY) at a concentration of 5 μg/ml in BupH carbonate–bicarbonate buffer (Pierce) by adding 100 μl per well and incubating overnight at 4 °C. Samples were diluted 1:10 in TBS. Aliquots (100 μL) were acidified with 50 μL 300 mM acetic acid and incubated at room temperature for 5 min. LY coated plates were washed three times with TBST and 50 μL 1 M Tris, pH 9.5 was added. Acid-treated samples (100 μL) were added to the buffered, coated plates and allowed to incubate overnight at 4 °C. The following day, plates were washed three times with TBST followed by elution of bound ADA by addition of 65 μL 300 mM acetic acid for 5 minutes at room temperature. Fresh Nunc Maxisorp plates were then loaded with 50 μL of 1 M Tris pH 9.5 buffer. Fifty microliters of the acid eluate was transferred to the buffered Maxisorp plates followed by incubation for 1 h at room temperature to allow binding of eluted ADA to the wells. Plates were washed three times with TBST and blocked with Casein buffer for 1 hour at room temperature. After washing three times with TBST, 100 μL B-LY was added and incubated at room temperature for 1 h to allow binding to platebound ADA. Plates were washed three times with TBST and 100 μL SA-HRP was added and incubated at room temperature for 30 minutes at room temperature. Plates were washed three times with TBST and 100 μl of 3,3′,5,5′ tetramethylbenzidine (TMB) substrate (BioFX, Ewings Mills, MD) substrate was added and incubated for 30 min at room temperature. Color development was stopped by addition of 100 μl of 2 M phosphoric acid, and plates were read at 450 nm in a SpectraMax Plus plate reader. 2.4. Data analysis
2.2. Biotin-LY Conjugate (B-LY) Therapeutic humanized monoclonal antibody (LY) was labeled using EZ-Link Sulfo-NHS-LC-Biotinyla-
Data was transferred to Excel 2003 where averages and standard deviations were calculated. The calculated numbers were transferred to Sigma Plot 8.0 for graphical
12
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
Table 1 Cut point data Sample
Run number 1
2
3
4
5
5B
6
6B
BRH37212 BRH37213 BRH37214 BRH37215 BRH37216 BRH37217 BRH37218 BRH37219 BRH37220 b BRH37221 BRH37222 BRH37223 BRH37224 BRH37225 BRH37226 BRH37227 BRH37228 BRH37229 BRH37230 BRH37231 BRH37232 BRH37233 BRH37234 BRH37235 BRH37236 BRH37241 BRH37242 BRH37243 BRH37244 BRH37245 BRH37246 BRH37247 BRH37248 BRH37249 BRH37250 BRH37251 BRH37252 b BRH37253 BRH37254 BRH37255 BRH37256 BRH37257 BRH37258 BRH37259 BRH37260 BRH37261 BRH37262 BRH37263 BRH37264 BRH37265 BRH37266
0.092 0.183 0.093 0.071 0.196 0.164 0.140 0.151 1.083 0.491 0.204 0.177 a 0.075 0.078 0.247 0.129 0.145 0.068 0.040 0.043 0.112 0.060 0.062 0.046 0.110 0.091 0.049 0.040 0.070 0.047 0.095 0.048 0.082 0.106 0.059 0.132 1.490 0.087 0.065 0.062 0.312 0.057 0.050 0.291 0.082 0.280 a 0.068 0.079 0.307 0.132 a 0.046
0.035 0.243 0.051 0.047 0.182 0.115 0.079 0.080 1.146 0.402 0.122 0.053 0.044 0.040 0.169 a 0.086 0.063 0.094 0.047 0.052 0.108 0.062 0.075 0.057 0.157 0.119 0.060 0.046 0.076 0.043 0.096 0.058 0.080 0.119 0.051 0.141 1.501 0.078 0.064 0.063 0.314 0.065 0.055 0.211 0.074 0.369 0.049 0.069 0.421 0.120 0.059
0.035 0.207 0.053 0.037 0.128 0.087 0.069 0.059 0.867 0.310 0.094 0.040 0.040 0.037 0.169 0.081 a 0.055 0.059 0.027 0.030 0.078 0.041 0.046 0.033 0.123 0.072 0.035 0.029 0.059 0.029 0.077 0.035 0.067 0.107 0.027 0.087 1.355 0.052 0.035 0.036 0.260 0.042 0.028 0.171 0.063 0.235 0.034 0.057 0.389 0.097 0.031
0.038 0.280 0.071 0.042 0.274 a 0.138 0.093 0.118 1.955 0.557 0.169 0.065 0.048 0.039 0.293 0.103 0.108 0.082 0.036 0.044 0.103 0.054 0.064 0.045 0.147 0.105 0.054 0.039 0.080 0.037 0.102 0.052 0.079 0.120 0.050 0.163 1.739 0.074 0.062 0.057 0.401 0.072 0.053 0.283 0.094 0.443 0.054 0.083 0.491 0.139 0.053
0.060 0.208 0.082 0.057 0.281 0.174 0.080 0.074 1.402 0.503 0.142 0.088 a 0.096 0.052 0.199 0.084 0.093 0.081 0.041 0.043 0.134 0.049 0.059 0.050 0.164 0.099 0.048 0.043 0.093 0.042 0.120 0.073 0.086 0.131 0.051 0.143 1.444 0.089 0.059 0.055 0.279 0.067 0.059 0.296 0.101 0.351 0.056 0.117 0.610 0.106 0.055
0.054 0.065 0.055 0.057 0.116 0.056 0.051 0.056 0.094 0.061 0.064 0.059 0.073 0.042 0.087 0.064 0.056 0.046 0.039 0.034 0.067 0.037 0.038 0.048 0.040 0.049 0.033 0.043 0.044 0.037 0.070 0.066 0.044 0.070 0.042 0.072 0.107 0.058 0.040 0.040 0.048 0.042 0.041 0.092 0.053 0.101 0.049 0.074 0.193 0.054 0.048
0.054 0.224 0.099 0.055 0.343 0.196 0.077 0.084 2.078 0.501 0.167 0.096 0.104 0.061 0.270 0.094 0.116 0.096 0.055 0.065 0.158 0.071 0.088 0.059 0.178 0.129 0.051 0.061 0.104 0.052 0.183 0.101 0.104 0.158 0.083 0.185 1.919 0.127 0.095 0.103 0.371 0.096 0.078 0.501 a 0.108 0.493 0.081 0.150 1.074 0.138 0.086
0.053 0.075 0.065 0.053 0.163 0.073 0.049 0.060 0.091 0.069 0.090 0.075 0.124 0.051 0.111 0.063 0.091 0.061 0.039 0.059 0.098 0.058 0.061 0.051 0.063 0.078 0.057 0.060 0.071 0.053 0.095 0.091 0.064 0.083 0.080 0.115 0.146 0.096 0.084 0.089 0.077 0.068 0.079 0.176 0.068 0.160 0.072 0.102 0.319 0.076 0.076
Runs 5B and 6B represent samples run with addition of 500 μg/ml free drug added. a Coefficient of Variation (CV) N25% data omitted from calculations. b Sample ANOVA least square mean is statistical outlier relative to other sample means.
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
13
presentation. Cut point statistical analyses were performed using JMP Statistical Discovery Software (ver. 5.1; SAS Institute, Inc., Cary, NC). 3. Results 3.1. Assay cut point The assay cut point is the level of response above which a sample is defined as “putative” positive and below which is defined as negative (Mire-Sluis et al., 2004). The cut point was determined from the analysis of 51 normal adult human sera (Bioreclamation, Hicksville, NY). Samples were analyzed 3 times each by 2 analysts for a total of 6 determinations. Two of the assays were performed with addition of 500 μg/ml LY. Results are shown in Table 1. Statistical analysis was performed on log base 10 transformed absorbance values. Samples 37220 and 37252 were identified as statistical outliers relative to the overall distribution of sample least square means. These samples also had greater than 90% inhibition with addition of excess free drug and were therefore removed from further analysis. Eight values were identified as having coefficient of variations greater than 25% and were therefore excluded from statistical analysis (Table 1). Tukey's biweight procedure was used to calculate run-specific estimates of population mean and standard deviation of the remaining log-transformed serum data values. Cut point values for each run were then calculated by adding the mean plus 1.645 times the standard deviation and taking the inverse log transformation. Results indicated a combined inter-assay cut point of 0.246. Cut point values were recalculated after exclusion of data for four samples (37221, 37256, 37261 and 37264) that exceeded this initial cut point and had 73–94% inhibition on addition of excess free drug. The inter-assay analysis of the remaining results indicated a cut point value of 0.200 OD units. This lower cut point results in a more conservative approach reducing the possibility of generating false negatives when analyzing clinical samples.
Fig. 1. ACE assay sensitivity. Affinity purified rabbit anti-human IgG was prepared at concentrations of 0.0 to 1000 ng/ml, diluted 1/10 and run in the standard ACE assay format. Results represent mean OD ± SD for 4 replicates run over 2 days.
Fig. 1 indicate a sensitivity of approximately 77 ng/ml when compared to the established assay cut point of 0.200 OD units. 3.3. Effect of acid treatment The effect of the initial acid treatment on sample detection and drug tolerance was determined. Affinity purified rabbit anti-human IgG at concentrations of 125 to 1000 ng/ml or monkey anti-LY antisera at dilutions of 1/1000 to 1/8000 were tested in the presence of 0.5–500 μg/ml added LY. Assays were run using the standard ACE protocol of 300 mM acetic acid followed by neutralization with TRIS buffer or a modified version where the initial acid treatment and subsequent neutralization were both replaced with TBS, pH 7.4. Results shown in Fig. 2 demonstrate first that with these samples, the acid treatment had little or no effect on the assay response when no free drug is present as shown by nearly identical OD's for samples without free. However, the acid treatment indicated a significant increase in assay response, particularly at lower free drug concentrations. Similar results are observed for both the affinity purified rabbit anti-human IgG and the monkey anti-LY antisera.
3.2. Assay sensitivity
3.4. Assay Drug tolerance
Sensitivity was determined using affinity purified rabbit anti-human IgG (Jackson Immunoresearch). Samples were prepared at anti-human IgG concentrations of 0.0, 7.8, 15.6, 31.3, 62.5, 125.0, 250.0, 500.0 and 1000 ng/ml in TBS. These samples were diluted 1/10 into TBS and run using the standard assay format run in duplicate on two different days. Data were reduced using a weighted 5-parameter logistic model algorithm. Results shown in
Assay tolerance to excess drug was established by spiking LY at concentrations of 0, 62.5, 125, 250 and 500 μg/ml into samples containing 500 μg/ml of affinity purified rabbit anti-human IgG and diluting the samples 1/10 into TBS. Each sample was run using the standard assay format in duplicate on two different days. Results shown in Fig. 3 indicate that samples containing 62.5 μg/ ml of LY show a significant decrease of signal over the
14
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
Fig. 2. Effect of acid initial acid treatment on drug tolerance. Affinity purified rabbit anti-human IgG or monkey anti-LY antisera were assayed at various concentrations in the presence of free LY from 0.5–500 μg/ml. Samples were assayed using the standard ACE format using an initial acid treatment followed by neutralization (3a, 3c, without acid) or a modified ACE format where the initial acid treatment and neutralization were both replaced with TBS (3b, 3d, without acid). Each point represents the average of two replicates.
unspiked control. Progressively higher levels of LY cause increasing inhibition of signal. However, even a 1000-fold excess of therapeutic antibody (500 μg/ml) gave a signal of approximately 0.300, well above the established assay cut point of 0.200 OD units, indicating a high level of tolerance to free drug.
The negative control had CVs ranging from 9.0–24.6%. while the low and high controls had CVs ranging from 6.9% to 15.7% and 3.2% to 16.9% respectively, indicating good overall precision. Analysis of plate
3.5. Precision and variability Precision describes the variability of the analytical method when repeated analyses are performed on a given sample. Intra-assay precision describes the variation of a result within a given assay. Inter-assay precision describes the variation of a result between assays. Determination of both intra- and inter- assay precision for the screening assay was evaluated during assay validation as well as effects of sample position on the plate and analyst to analyst variation. Positive monkey antisera (anti-LY) was diluted in normal human serum to prepare low and high positive controls. The negative control was unspiked normal human serum. Each sample was assayed to generate six reportable results per concentration across six assay runs by two analysts over three days (Table 2).
Fig. 3. Assay tolerance to free drug. Affinity purified rabbit antihuman IgG at a concentration of 500 ng/ml was assayed in the presence of free LY at concentrations from 62.5 to 500 μg/ml. Results are shown compared with the assay cut point.
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
15
Table 2 Precision data Run
Analyst
Plate A
Plate B
Plate C
n
Mean
Std
% CV
Front
Back
Front
Back
Front
Back
A B A B A B
0.072 0.063 0.061 0.049 0.070 0.081
0.067 0.056 0.052 0.052 0.062 0.085
0.106 0.064 0.049 0.064 0.066 0.068
0.060 0.070 0.057 0.062 (b) 0.077
0.066 0.072 0.053 0.059 0.057 0.068
(b) 0.067 0.048 0.072 0.056 0.077
5 6 6 6 5 6
0.074 0.065 0.053 0.060 0.062 0.076
0.018 0.006 0.005 0.008 0.006 0.007
24.6 8.7 9.2 14.0 9.5 9.0
1 2 3 4 5 6
A B A B A B
0.123 0.106 0.120 0.107 0.150 0.175
0.144 0.120 0.139 0.117 0.168 0.174
0.175 0.127 0.106 0.117 0.129 0.155
0.155 0.118 0.104 0.129 0.137 0.149
0.126 0.116 0.120 0.120 0.132 0.153
0.178 0.118 0.134 0.119 0.130 0.159
6 6 6 6 6 6
0.150 0.118 0.121 0.118 0.141 0.161
0.024 0.007 0.014 0.007 0.015 0.011
15.7 5.8 11.8 6.0 10.9 6.9
1 2 3 4 5 6
A B A B A B
0.676 0.608 0.661 0.562 0.853 0.792
0.781 0.630 0.697 0.586 0.960 0.818
0.849 0.663 0.648 0.541 0.680 0.849
0.776 0.598 0.579 0.569 0.712 0.782
0.507 0.594 0.712 0.631 0.784 0.785
0.797 0.539 0.748 0.647 0.727 0.811
6 6 6 6 6 6
0.731 0.605 0.674 0.589 0.786 0.806
0.123 0.041 0.059 0.041 0.105 0.025
16.9 6.8 8.7 7.0 13.3 3.2
Negative 1 2 3 4 5 6 Low
High
(b) Replicate %CV greater than 25% — data not included.
location using Students T test shows no significant differences related to sample location on the plates. 4. Discussion Therapeutic monoclonal antibodies can be dosed at high levels and have relatively long half-lives resulting in significant blood levels for extended periods of time. The presence of high concentrations of circulating drug can make detection of anti-drug antibodies extremely difficult, requiring collection of clinical samples after long washouts (Pendley et al., 2003). It also may cause anti-drug antibodies to be non-detectable during multidose trials where elevated drug levels are maintained for extended periods. Many current immunogenicity assays used to assess ADA are bridging assays due to the nature of the drug itself. These assays can have significant interference from even low levels of free drug. Difficulties in detecting anti-drug antibodies in standard bridging assays stem from multiple causes. First, if free drug is present in significant quantities it can complex with anti-drug antibodies directly interfering with the assay. Second, in bridging assays, the anti-drug antibody must form a bridge between solid-phase drug and soluble labeled drug, in each case binding with a single arm of the anti-drug antibody. Free drug in this
type of assay can prevent bridge formation by blocking only one of these reactions, making this format particularly sensitive to interference. The ACE format described here was developed to address these issues and has been shown to maintain detectable assay signals in the presence of a 1000-fold excess of free drug (500 ng/ml anti-drug antibody in the presence of 500 μg/ml free drug). The first step of the assay utilizes an acid dissociation step to break apart potential anti-drug/drug complexes. The acid dissociated complexes are neutralized in the presence of solidphase drug, allowing for the potential capture of the antidrug antibody. After washing away the unbound free drug, the bound antibody is released using a second acid dissociation and subsequent adsorption to a solid-phase well. This plate-bound anti-drug antibody is detected using biotinylated drug, allowing for sensitive detection using poly-HRP labeled streptavidin (Diamondis and Christopoulos, 1991). This assay format has been designed to measure antidrug antibodies for monoclonal antibodies intended as human therapeutics. These antibodies include murine, chimeric, humanized and fully human molecules with the potential for anti-drug antibodies to species differences, allotypes and idiotypes. The affinity purified rabbit anti-human IgG utilized here likely recognizes
16
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
mainly species and allotypic epitopes while the monkey anti-therapeutic recognizes potentially all three epitopes. The anti-idiotypic component of the monkey antisera has been demonstrated by reactivity with the therapeutic molecule even when diluted in normal human sera, which would inhibit most if not all of the anti-species and allotypic antibodies. The demonstration of free drug tolerance with both the rabbit and monkey antibodies with their respective specificity differences indicates the potential power of this method (Fig. 2). Acid dissociation has been previously used in immunoassays for detection of antibodies to insulin (Fineberg et al., 1996; Kansal et al., 1979; Dixon, 1974) and monoclonal antibodies (Patton et al., 2005). There have been reports of acid treatment potentially causing decreased binding of anti-immunoglobulin antibodies depending on the nature of the acid exposure (Moxness et al., 2003). The exposure to acid in the current ACE assay has been minimized to a five minute incubation with 300 mM acetic acid after preliminary experiments that indicated that the maximum dissociation occurs after approximately 1–3 min (data not shown). This keeps the pH from getting below approximately 3.0 for a relatively short period of time. Numerous anti-LY primate antisera were tested with and without the 5 minute acid exposure and subsequent neutralization described here with no significant loss of binding activity in a standard ELISA protocol (data not shown). This indicates that the acid treatment utilized here has a minimal impact on antibody integrity structurally and conformationally. The acid dissociation of potential ADA-LY immune complexes with subsequent neutralization in the presence of solid-phase drug allows the ADA to potentially bind multivalently to the solid-phase drug which should increase the ability to detect lower affinity antibodies. It also takes advantage of the potential for antibodies to form more stable complexes with the solid-phase drug as opposed to solution phase molecules (Patton et al., 2003). Transfer of the captured ADA to a second plate for the subsequent detection of binding with biotinylated antibody again gives the antibody the potential for binding the detection antibody multivalently. Again this gives the opportunity for detection of lower affinity antibodies. The use of biotinylated antibody followed by HRP-streptavidin gives added amplification for the detection step, increasing the potential sensitivity of the method as well as limiting the impact of labeling the drug due to the relatively small size (Mr 244) of the biotin molecule (Diamondis and Christopoulos, 1991). Typically, immunogenicity analyses are performed by first screening for positive samples, running confirmatory tests on putative positives (samples above the cut
point for the assay) and finally titering the confirmed positive samples for quantitation (Mire-Sluis et al., 2004). This assay format can be used for purposed of titering samples although due to reduced responses in the presence of high levels of free drug, the results obtained may not accurately represent the true value that would be obtained for that clinical sample in the absence of free drug. Care must therefore be taken by those interpreting data from this or any drug tolerant immunoassay when assessing clinical responses to protein therapeutics. In summary, the ability of the above described ACE assay to detect low levels of ADA in the presence of high levels of drug enables a robust immunogenicity assay that will provide much more flexibility in design and performance of preclinical and clinical trials. The ACE assay retains the desirable properties of earlier bridging assays such as excellent sensitivity and precision, while removing doubts about the presence of excess free drug in samples causing false negative results. Our group has therefore adopted this format to detect ADA antibodies directed against LY monoclonal therapeutics. Acknowledgements The authors would like to thank Anthony Butterfield and Holly Smith of Lilly for helpful discussion and positive non-human primate antibodies, Mark O'Dell and Ron Bowsher of Linco Diagnostic Services, Inc. for performance of assays and assistance with data interpretation and Wendell Smith of Bowsher Brunelle Smith, LLC for statistical analysis of cut point data. References Baert, F., Norman, M., Vermeire, S., Van Assche, G., D’Haens, G., Carbonez, A., Rutgeerts, P., 2003. Influence of immunogenicity on the long-term efficacy of infliximab in Crohn's disease. NEJM 348, 601. Basser, R.L., O’Flaherty, E., Green, M., Edmonds, M., Nichol, J., Menchaca, D.M., Cohen, B., Begley, C.G., 2002. Development of pancytopenia with neutralizing antibodies to thrombopoietin after multicycle chemotherapy supported by megakaryocyte growth and development factor. Blood 99, 2599. Bendtzen, K., Geborek, T., Svenson, M., Larsson, L., Kapetanovic, M.C., Saxne, T., 2006. Individualized monitoring of drug bioavailability and immunogenicity in rheumatoid arthritis patients treated with the tumor necrosis factor α Inhibitor infliximab. Arthritis Rheum. 54, 3782. Bourdage, J.S., Lee, T., Taylor, J.M., Willey, M.B., Brandt, J.T., Konrad, R.J., 2005. Effect of double antigen bridging immunoassay format on antigen coating concentration dependence and implications for designing immunogenicity assays for monoclonal antibodies. J. Pharm. Biomed. Anal. 39, 685. Casadevall, N., Nataf, J., Viron, B., Kolta, A., Kiladjian, J.J., MartinDupont, P., Michaud, P., Papo, T., Ugo, V., Teyssandier,
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17 I., Varet, B., Mayeaux, P., 2002. Pure red cell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin. N. Engl. J. Med. 346, 469. Diamondis, E., Christopoulos, T.K., 1991. The Biotin–(strept)avidin system: principles and applications in biotechnology. Clin. Chem. 37, 625. Dixon, K., 1974. Measurement of antibodies to insulin in serum. Clin. Chem. 10, 1275. Feldman, E., Kalaycio, M., Weiner, G., Frankel, S., Schulman, P., Schwartzberg, P., Juric, J., Velez-Garcia, E., Seiter, K., Scheinberg, D., Levitt, D., Wedel, N., 2003. Treatment of relapsed or refractory acute myeloid leukemia with humanized anti-CD33 monoclonal antibody HuM195. Leukemia, 17, 314. Fineberg, Naomi S., Fineberg, Edwin S., Anderson, James H., Birkett, Martin A., Gibson, Reid G., Hufferd, Suzan, 1996. Immunologic effects of insulin lispro [Lys (B28), Pro (B29) human insulin] in IDDM and NIDDM patients previously treated with insulin. Diabetes 45, 1750. Kansal, P.C., BAndisode, M.S., Boshell, B.R., 1979. Determination of insulin antibodies. Horm. Metab. Res. 11, 187. Looney, R.J., Anolik, J.H., Campbell, D., Felgar, R.E., Young, F., Arend, L.J., Sloand, J.A., Rosenblatt, J., Sanz, I., 2004. B cell depletion as a novel treatment for systemic lupus erythematosus. Arth. Rheum. 50, 2580.
17
Maloney, D.G., Grillo-López, A.J., White, C.A., Bodkin, D., Schilder, R.J., Neidhart, J.A., Janakiraman, N., Foon, K.A., Liles, T.L., Dallaire, B.K., Wey, K., Royston, I., Davis, T., Levy, R., 1997. IDEC-C2B8 (Rituximab) anti-CD20 monoclonal antibody therapy in patients with relapsed low-grade non-Hodgkin's lymphoma. Blood 90, 2188. Mire-Sluis, A.R., Barrett, Y.C., Devanarayan, V., Koren, E., Liu, H., Maia, M., Parish, T., Scott, G., Shankar, G., Shores, E., Swanson, S.J., Taniguchi, G., Wierda, D., Zuckerman, L.A., 2004. Recommendations for the design and optimization of immunoassays used in the detection of host antibodies against biotechnology products,. J. Immunol. Methods 289, 1. Moxness, M., Foley, J., Stene, M., Finco-Kent, D., Bedian, V., Krasner, A., Kawabata, T., 2003. Development and bioligand Binding Assays to measure total, IgA, IgE, IgG and IgM antibodies in human serum. Ann. N.Y. Acad. Sci. 1005, 265. Patton, P., Mullenix, M., Swanson, S., Koren, E., 2005. An acid dissociation bridging ELISA for detection of antibodies directed against therapeutic proteins in the presence of antigen. J. Immunol. Methods 304, 189. Pendley, C., Schantz, A., Wagner, C., 2003. Immunogenicity of therapeutic monoclonal antibodies. Curr. Opin. Mol. Ther. 5, 179.
Research paper
An Affinity Capture Elution (ACE) assay for detection of anti-drug antibody to monoclonal antibody therapeutics in the presence of high levels of drug James S. Bourdage ⁎, Carolyn A. Cook, Daphne L. Farrington, Jana S. Chain, Robert J. Konrad Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN 46285, United States Received 13 December 2006; received in revised form 25 May 2007; accepted 5 July 2007 Available online 27 July 2007
Abstract Monoclonal antibody therapeutics typically have relatively long half-lives and can be dosed at high levels. Although formation of anti-drug antibodies (ADA) is relatively rare, detection of these antibodies can be very difficult in the presence of high circulating levels of drug. Typically these ADA are detected by bridging ELISAs which can be very sensitive to even low levels of drug. We describe an ELISA method based on affinity capture of ADA on solid-phase drug followed by removal of excess free drug, release and transfer of bound ADA and subsequent detection using biotinylated drug. The assay is both sensitive and highly tolerant to free drug with detection of 500 ng/ml of ADA readily achieved in the presence of 500 μg/ml of drug. © 2007 Elsevier B.V. All rights reserved. Keywords: Antibody assay; Acid dissociation; Immunogenicity; Therapeutic proteins; ELISA; Antigen interference
1. Introduction Monoclonal antibody therapeutics are finding increased usage for treatment of a wide variety of clinical indications with over 18 approved products. The incidence of anti-drug antibodies (ADA) varies widely from almost 100% for some murine monoclonals to virtually undetectable for some humanized antibodies (Bourdage Abbreviations: ELISA; Enzyme-Linked Immunosorbent Assay; ADA; anti-drug antibody; ACE; affinity capture elution; TMB; tetramethyl benzidine; OD; optical density; HRP; horseradish peroxidase. ⁎ Corresponding author. Eli Lilly Company, 2001 W. Main St., GL55, Greenfield, IN 46140, USA. Tel.: +1 317 277 4139; fax: +1 317 276 5281. E-mail address: [email protected] (J.S. Bourdage). 0022-1759/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2007.07.004
et al., 2005). These antibodies can have significant clinical consequence. Bendtzen et al. reported detection of antiinfliximab antibodies that apparently caused loss of bioavailability, infusion reactions and treatment failure (Bendtzen et al., 2006). Anti-rituximab antibodies caused reduced B-cell depletion and reduced blood levels of rituximab (Looney et al., 2004). There have been cases where neutralizing antibodies to endogenous proteins have caused life-threatening situations (Casadevall et al., 2002; Basser et al., 2002). Humanized monoclonal antibodies typically have half-lives of 10–20 days and can often be administered at relatively large doses leading to blood levels of more than 500 μg/ml (Maloney et al., 1997). Preclinical studies typically use even higher doses to establish safety
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
margins with necessary short washouts prior to evaluation. Bridging or double antigen format assays are often used to evaluate ADA for monoclonals due to their ability to detect multiple classes of antibody (Feldman et al., 2003: Pendley et al., 2003) . However, there is concern for the ability to detect low affinity antibody due to the requirement for monovalent binding to the solid phase and labeled drug during bridge formation. This monovalent binding also causes the assay to be very sensitive to free drug since blocking of only a single arm of the ADA molecule will prevent bridge formation. A double antigen format assay described for infliximab has drug interference at drug levels above 1.4 μg/ml (Baert et al., 2003). This presents a dilemma when developing preclinical and clinical protocol sampling points since antibodies may be undetectable while drug levels are high thus causing a requirement for washouts of several weeks to several months. These potentially long washout periods could in turn cause decreasing ADA titers, thus making detection of ADA difficult or impossible. We have therefore developed a method that first dissociates ADA-free drug complexes with acid treatment followed by neutralization in the presence of solid-phase drug giving the ADA an opportunity to be affinity captured. After washing away excess free drug, ADA are eluted off with acid and subsequently bound to a fresh solid surface. Bound ADA are subsequently detected by addition of biotinylated drug followed by streptavidinHRP and substrate. Results presented here indicate that this affinity capture elution (ACE) assay format is capable of detecting low ng/ml levels of ADA in the presence of a 1000-fold excess of free drug. 2. Materials and methods 2.1. Reagents Wash buffer consists of 0.01 M Tris-buffered saline (TBS) (Fisher Scientific, Fair Lawn, NJ) with 0.05% Tween 20 (Sigma-Aldrich, St. Louis, MO) (TBST). Therapeutic monoclonal antibody (LY) was supplied by Eli Lilly and Co. (Indianapolis, IN). Affinity purified rabbit anti-human IgG (AffiniPure Rabbit Anti-Human IgG (H + L), was obtained from Jackson ImmunoResearch Laboratories, Inc, West Grove, PA. Poly HRP Streptavidin (SA-HRP) was obtained from Pierce (Pierce Biotechnology, Rockford, IL, N200).
11
tion Kits (Pierce, 21435) according to manufacturer's instructions. Briefly, LY was dialyzed overnight against phosphate buffered saline, combined with Sulfo-NHSLC-Biotin in a 20:1 molar ratio and incubated for 1 hour at room temperature. The reaction mixture was dialyzed vs. PBS overnight. Protein content was analyzed using the BCA Protein Assay (Pierce, 23225) and was then diluted with an equal volume of glycerol and stored at − 20 °C. 2.3. ACE assay ELISA plates (Nunc Maxisorp 96 Microplates, Nunc, Rochester, NY) were coated with therapeutic humanized monoclonal antibody (LY) at a concentration of 5 μg/ml in BupH carbonate–bicarbonate buffer (Pierce) by adding 100 μl per well and incubating overnight at 4 °C. Samples were diluted 1:10 in TBS. Aliquots (100 μL) were acidified with 50 μL 300 mM acetic acid and incubated at room temperature for 5 min. LY coated plates were washed three times with TBST and 50 μL 1 M Tris, pH 9.5 was added. Acid-treated samples (100 μL) were added to the buffered, coated plates and allowed to incubate overnight at 4 °C. The following day, plates were washed three times with TBST followed by elution of bound ADA by addition of 65 μL 300 mM acetic acid for 5 minutes at room temperature. Fresh Nunc Maxisorp plates were then loaded with 50 μL of 1 M Tris pH 9.5 buffer. Fifty microliters of the acid eluate was transferred to the buffered Maxisorp plates followed by incubation for 1 h at room temperature to allow binding of eluted ADA to the wells. Plates were washed three times with TBST and blocked with Casein buffer for 1 hour at room temperature. After washing three times with TBST, 100 μL B-LY was added and incubated at room temperature for 1 h to allow binding to platebound ADA. Plates were washed three times with TBST and 100 μL SA-HRP was added and incubated at room temperature for 30 minutes at room temperature. Plates were washed three times with TBST and 100 μl of 3,3′,5,5′ tetramethylbenzidine (TMB) substrate (BioFX, Ewings Mills, MD) substrate was added and incubated for 30 min at room temperature. Color development was stopped by addition of 100 μl of 2 M phosphoric acid, and plates were read at 450 nm in a SpectraMax Plus plate reader. 2.4. Data analysis
2.2. Biotin-LY Conjugate (B-LY) Therapeutic humanized monoclonal antibody (LY) was labeled using EZ-Link Sulfo-NHS-LC-Biotinyla-
Data was transferred to Excel 2003 where averages and standard deviations were calculated. The calculated numbers were transferred to Sigma Plot 8.0 for graphical
12
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
Table 1 Cut point data Sample
Run number 1
2
3
4
5
5B
6
6B
BRH37212 BRH37213 BRH37214 BRH37215 BRH37216 BRH37217 BRH37218 BRH37219 BRH37220 b BRH37221 BRH37222 BRH37223 BRH37224 BRH37225 BRH37226 BRH37227 BRH37228 BRH37229 BRH37230 BRH37231 BRH37232 BRH37233 BRH37234 BRH37235 BRH37236 BRH37241 BRH37242 BRH37243 BRH37244 BRH37245 BRH37246 BRH37247 BRH37248 BRH37249 BRH37250 BRH37251 BRH37252 b BRH37253 BRH37254 BRH37255 BRH37256 BRH37257 BRH37258 BRH37259 BRH37260 BRH37261 BRH37262 BRH37263 BRH37264 BRH37265 BRH37266
0.092 0.183 0.093 0.071 0.196 0.164 0.140 0.151 1.083 0.491 0.204 0.177 a 0.075 0.078 0.247 0.129 0.145 0.068 0.040 0.043 0.112 0.060 0.062 0.046 0.110 0.091 0.049 0.040 0.070 0.047 0.095 0.048 0.082 0.106 0.059 0.132 1.490 0.087 0.065 0.062 0.312 0.057 0.050 0.291 0.082 0.280 a 0.068 0.079 0.307 0.132 a 0.046
0.035 0.243 0.051 0.047 0.182 0.115 0.079 0.080 1.146 0.402 0.122 0.053 0.044 0.040 0.169 a 0.086 0.063 0.094 0.047 0.052 0.108 0.062 0.075 0.057 0.157 0.119 0.060 0.046 0.076 0.043 0.096 0.058 0.080 0.119 0.051 0.141 1.501 0.078 0.064 0.063 0.314 0.065 0.055 0.211 0.074 0.369 0.049 0.069 0.421 0.120 0.059
0.035 0.207 0.053 0.037 0.128 0.087 0.069 0.059 0.867 0.310 0.094 0.040 0.040 0.037 0.169 0.081 a 0.055 0.059 0.027 0.030 0.078 0.041 0.046 0.033 0.123 0.072 0.035 0.029 0.059 0.029 0.077 0.035 0.067 0.107 0.027 0.087 1.355 0.052 0.035 0.036 0.260 0.042 0.028 0.171 0.063 0.235 0.034 0.057 0.389 0.097 0.031
0.038 0.280 0.071 0.042 0.274 a 0.138 0.093 0.118 1.955 0.557 0.169 0.065 0.048 0.039 0.293 0.103 0.108 0.082 0.036 0.044 0.103 0.054 0.064 0.045 0.147 0.105 0.054 0.039 0.080 0.037 0.102 0.052 0.079 0.120 0.050 0.163 1.739 0.074 0.062 0.057 0.401 0.072 0.053 0.283 0.094 0.443 0.054 0.083 0.491 0.139 0.053
0.060 0.208 0.082 0.057 0.281 0.174 0.080 0.074 1.402 0.503 0.142 0.088 a 0.096 0.052 0.199 0.084 0.093 0.081 0.041 0.043 0.134 0.049 0.059 0.050 0.164 0.099 0.048 0.043 0.093 0.042 0.120 0.073 0.086 0.131 0.051 0.143 1.444 0.089 0.059 0.055 0.279 0.067 0.059 0.296 0.101 0.351 0.056 0.117 0.610 0.106 0.055
0.054 0.065 0.055 0.057 0.116 0.056 0.051 0.056 0.094 0.061 0.064 0.059 0.073 0.042 0.087 0.064 0.056 0.046 0.039 0.034 0.067 0.037 0.038 0.048 0.040 0.049 0.033 0.043 0.044 0.037 0.070 0.066 0.044 0.070 0.042 0.072 0.107 0.058 0.040 0.040 0.048 0.042 0.041 0.092 0.053 0.101 0.049 0.074 0.193 0.054 0.048
0.054 0.224 0.099 0.055 0.343 0.196 0.077 0.084 2.078 0.501 0.167 0.096 0.104 0.061 0.270 0.094 0.116 0.096 0.055 0.065 0.158 0.071 0.088 0.059 0.178 0.129 0.051 0.061 0.104 0.052 0.183 0.101 0.104 0.158 0.083 0.185 1.919 0.127 0.095 0.103 0.371 0.096 0.078 0.501 a 0.108 0.493 0.081 0.150 1.074 0.138 0.086
0.053 0.075 0.065 0.053 0.163 0.073 0.049 0.060 0.091 0.069 0.090 0.075 0.124 0.051 0.111 0.063 0.091 0.061 0.039 0.059 0.098 0.058 0.061 0.051 0.063 0.078 0.057 0.060 0.071 0.053 0.095 0.091 0.064 0.083 0.080 0.115 0.146 0.096 0.084 0.089 0.077 0.068 0.079 0.176 0.068 0.160 0.072 0.102 0.319 0.076 0.076
Runs 5B and 6B represent samples run with addition of 500 μg/ml free drug added. a Coefficient of Variation (CV) N25% data omitted from calculations. b Sample ANOVA least square mean is statistical outlier relative to other sample means.
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
13
presentation. Cut point statistical analyses were performed using JMP Statistical Discovery Software (ver. 5.1; SAS Institute, Inc., Cary, NC). 3. Results 3.1. Assay cut point The assay cut point is the level of response above which a sample is defined as “putative” positive and below which is defined as negative (Mire-Sluis et al., 2004). The cut point was determined from the analysis of 51 normal adult human sera (Bioreclamation, Hicksville, NY). Samples were analyzed 3 times each by 2 analysts for a total of 6 determinations. Two of the assays were performed with addition of 500 μg/ml LY. Results are shown in Table 1. Statistical analysis was performed on log base 10 transformed absorbance values. Samples 37220 and 37252 were identified as statistical outliers relative to the overall distribution of sample least square means. These samples also had greater than 90% inhibition with addition of excess free drug and were therefore removed from further analysis. Eight values were identified as having coefficient of variations greater than 25% and were therefore excluded from statistical analysis (Table 1). Tukey's biweight procedure was used to calculate run-specific estimates of population mean and standard deviation of the remaining log-transformed serum data values. Cut point values for each run were then calculated by adding the mean plus 1.645 times the standard deviation and taking the inverse log transformation. Results indicated a combined inter-assay cut point of 0.246. Cut point values were recalculated after exclusion of data for four samples (37221, 37256, 37261 and 37264) that exceeded this initial cut point and had 73–94% inhibition on addition of excess free drug. The inter-assay analysis of the remaining results indicated a cut point value of 0.200 OD units. This lower cut point results in a more conservative approach reducing the possibility of generating false negatives when analyzing clinical samples.
Fig. 1. ACE assay sensitivity. Affinity purified rabbit anti-human IgG was prepared at concentrations of 0.0 to 1000 ng/ml, diluted 1/10 and run in the standard ACE assay format. Results represent mean OD ± SD for 4 replicates run over 2 days.
Fig. 1 indicate a sensitivity of approximately 77 ng/ml when compared to the established assay cut point of 0.200 OD units. 3.3. Effect of acid treatment The effect of the initial acid treatment on sample detection and drug tolerance was determined. Affinity purified rabbit anti-human IgG at concentrations of 125 to 1000 ng/ml or monkey anti-LY antisera at dilutions of 1/1000 to 1/8000 were tested in the presence of 0.5–500 μg/ml added LY. Assays were run using the standard ACE protocol of 300 mM acetic acid followed by neutralization with TRIS buffer or a modified version where the initial acid treatment and subsequent neutralization were both replaced with TBS, pH 7.4. Results shown in Fig. 2 demonstrate first that with these samples, the acid treatment had little or no effect on the assay response when no free drug is present as shown by nearly identical OD's for samples without free. However, the acid treatment indicated a significant increase in assay response, particularly at lower free drug concentrations. Similar results are observed for both the affinity purified rabbit anti-human IgG and the monkey anti-LY antisera.
3.2. Assay sensitivity
3.4. Assay Drug tolerance
Sensitivity was determined using affinity purified rabbit anti-human IgG (Jackson Immunoresearch). Samples were prepared at anti-human IgG concentrations of 0.0, 7.8, 15.6, 31.3, 62.5, 125.0, 250.0, 500.0 and 1000 ng/ml in TBS. These samples were diluted 1/10 into TBS and run using the standard assay format run in duplicate on two different days. Data were reduced using a weighted 5-parameter logistic model algorithm. Results shown in
Assay tolerance to excess drug was established by spiking LY at concentrations of 0, 62.5, 125, 250 and 500 μg/ml into samples containing 500 μg/ml of affinity purified rabbit anti-human IgG and diluting the samples 1/10 into TBS. Each sample was run using the standard assay format in duplicate on two different days. Results shown in Fig. 3 indicate that samples containing 62.5 μg/ ml of LY show a significant decrease of signal over the
14
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
Fig. 2. Effect of acid initial acid treatment on drug tolerance. Affinity purified rabbit anti-human IgG or monkey anti-LY antisera were assayed at various concentrations in the presence of free LY from 0.5–500 μg/ml. Samples were assayed using the standard ACE format using an initial acid treatment followed by neutralization (3a, 3c, without acid) or a modified ACE format where the initial acid treatment and neutralization were both replaced with TBS (3b, 3d, without acid). Each point represents the average of two replicates.
unspiked control. Progressively higher levels of LY cause increasing inhibition of signal. However, even a 1000-fold excess of therapeutic antibody (500 μg/ml) gave a signal of approximately 0.300, well above the established assay cut point of 0.200 OD units, indicating a high level of tolerance to free drug.
The negative control had CVs ranging from 9.0–24.6%. while the low and high controls had CVs ranging from 6.9% to 15.7% and 3.2% to 16.9% respectively, indicating good overall precision. Analysis of plate
3.5. Precision and variability Precision describes the variability of the analytical method when repeated analyses are performed on a given sample. Intra-assay precision describes the variation of a result within a given assay. Inter-assay precision describes the variation of a result between assays. Determination of both intra- and inter- assay precision for the screening assay was evaluated during assay validation as well as effects of sample position on the plate and analyst to analyst variation. Positive monkey antisera (anti-LY) was diluted in normal human serum to prepare low and high positive controls. The negative control was unspiked normal human serum. Each sample was assayed to generate six reportable results per concentration across six assay runs by two analysts over three days (Table 2).
Fig. 3. Assay tolerance to free drug. Affinity purified rabbit antihuman IgG at a concentration of 500 ng/ml was assayed in the presence of free LY at concentrations from 62.5 to 500 μg/ml. Results are shown compared with the assay cut point.
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
15
Table 2 Precision data Run
Analyst
Plate A
Plate B
Plate C
n
Mean
Std
% CV
Front
Back
Front
Back
Front
Back
A B A B A B
0.072 0.063 0.061 0.049 0.070 0.081
0.067 0.056 0.052 0.052 0.062 0.085
0.106 0.064 0.049 0.064 0.066 0.068
0.060 0.070 0.057 0.062 (b) 0.077
0.066 0.072 0.053 0.059 0.057 0.068
(b) 0.067 0.048 0.072 0.056 0.077
5 6 6 6 5 6
0.074 0.065 0.053 0.060 0.062 0.076
0.018 0.006 0.005 0.008 0.006 0.007
24.6 8.7 9.2 14.0 9.5 9.0
1 2 3 4 5 6
A B A B A B
0.123 0.106 0.120 0.107 0.150 0.175
0.144 0.120 0.139 0.117 0.168 0.174
0.175 0.127 0.106 0.117 0.129 0.155
0.155 0.118 0.104 0.129 0.137 0.149
0.126 0.116 0.120 0.120 0.132 0.153
0.178 0.118 0.134 0.119 0.130 0.159
6 6 6 6 6 6
0.150 0.118 0.121 0.118 0.141 0.161
0.024 0.007 0.014 0.007 0.015 0.011
15.7 5.8 11.8 6.0 10.9 6.9
1 2 3 4 5 6
A B A B A B
0.676 0.608 0.661 0.562 0.853 0.792
0.781 0.630 0.697 0.586 0.960 0.818
0.849 0.663 0.648 0.541 0.680 0.849
0.776 0.598 0.579 0.569 0.712 0.782
0.507 0.594 0.712 0.631 0.784 0.785
0.797 0.539 0.748 0.647 0.727 0.811
6 6 6 6 6 6
0.731 0.605 0.674 0.589 0.786 0.806
0.123 0.041 0.059 0.041 0.105 0.025
16.9 6.8 8.7 7.0 13.3 3.2
Negative 1 2 3 4 5 6 Low
High
(b) Replicate %CV greater than 25% — data not included.
location using Students T test shows no significant differences related to sample location on the plates. 4. Discussion Therapeutic monoclonal antibodies can be dosed at high levels and have relatively long half-lives resulting in significant blood levels for extended periods of time. The presence of high concentrations of circulating drug can make detection of anti-drug antibodies extremely difficult, requiring collection of clinical samples after long washouts (Pendley et al., 2003). It also may cause anti-drug antibodies to be non-detectable during multidose trials where elevated drug levels are maintained for extended periods. Many current immunogenicity assays used to assess ADA are bridging assays due to the nature of the drug itself. These assays can have significant interference from even low levels of free drug. Difficulties in detecting anti-drug antibodies in standard bridging assays stem from multiple causes. First, if free drug is present in significant quantities it can complex with anti-drug antibodies directly interfering with the assay. Second, in bridging assays, the anti-drug antibody must form a bridge between solid-phase drug and soluble labeled drug, in each case binding with a single arm of the anti-drug antibody. Free drug in this
type of assay can prevent bridge formation by blocking only one of these reactions, making this format particularly sensitive to interference. The ACE format described here was developed to address these issues and has been shown to maintain detectable assay signals in the presence of a 1000-fold excess of free drug (500 ng/ml anti-drug antibody in the presence of 500 μg/ml free drug). The first step of the assay utilizes an acid dissociation step to break apart potential anti-drug/drug complexes. The acid dissociated complexes are neutralized in the presence of solidphase drug, allowing for the potential capture of the antidrug antibody. After washing away the unbound free drug, the bound antibody is released using a second acid dissociation and subsequent adsorption to a solid-phase well. This plate-bound anti-drug antibody is detected using biotinylated drug, allowing for sensitive detection using poly-HRP labeled streptavidin (Diamondis and Christopoulos, 1991). This assay format has been designed to measure antidrug antibodies for monoclonal antibodies intended as human therapeutics. These antibodies include murine, chimeric, humanized and fully human molecules with the potential for anti-drug antibodies to species differences, allotypes and idiotypes. The affinity purified rabbit anti-human IgG utilized here likely recognizes
16
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17
mainly species and allotypic epitopes while the monkey anti-therapeutic recognizes potentially all three epitopes. The anti-idiotypic component of the monkey antisera has been demonstrated by reactivity with the therapeutic molecule even when diluted in normal human sera, which would inhibit most if not all of the anti-species and allotypic antibodies. The demonstration of free drug tolerance with both the rabbit and monkey antibodies with their respective specificity differences indicates the potential power of this method (Fig. 2). Acid dissociation has been previously used in immunoassays for detection of antibodies to insulin (Fineberg et al., 1996; Kansal et al., 1979; Dixon, 1974) and monoclonal antibodies (Patton et al., 2005). There have been reports of acid treatment potentially causing decreased binding of anti-immunoglobulin antibodies depending on the nature of the acid exposure (Moxness et al., 2003). The exposure to acid in the current ACE assay has been minimized to a five minute incubation with 300 mM acetic acid after preliminary experiments that indicated that the maximum dissociation occurs after approximately 1–3 min (data not shown). This keeps the pH from getting below approximately 3.0 for a relatively short period of time. Numerous anti-LY primate antisera were tested with and without the 5 minute acid exposure and subsequent neutralization described here with no significant loss of binding activity in a standard ELISA protocol (data not shown). This indicates that the acid treatment utilized here has a minimal impact on antibody integrity structurally and conformationally. The acid dissociation of potential ADA-LY immune complexes with subsequent neutralization in the presence of solid-phase drug allows the ADA to potentially bind multivalently to the solid-phase drug which should increase the ability to detect lower affinity antibodies. It also takes advantage of the potential for antibodies to form more stable complexes with the solid-phase drug as opposed to solution phase molecules (Patton et al., 2003). Transfer of the captured ADA to a second plate for the subsequent detection of binding with biotinylated antibody again gives the antibody the potential for binding the detection antibody multivalently. Again this gives the opportunity for detection of lower affinity antibodies. The use of biotinylated antibody followed by HRP-streptavidin gives added amplification for the detection step, increasing the potential sensitivity of the method as well as limiting the impact of labeling the drug due to the relatively small size (Mr 244) of the biotin molecule (Diamondis and Christopoulos, 1991). Typically, immunogenicity analyses are performed by first screening for positive samples, running confirmatory tests on putative positives (samples above the cut
point for the assay) and finally titering the confirmed positive samples for quantitation (Mire-Sluis et al., 2004). This assay format can be used for purposed of titering samples although due to reduced responses in the presence of high levels of free drug, the results obtained may not accurately represent the true value that would be obtained for that clinical sample in the absence of free drug. Care must therefore be taken by those interpreting data from this or any drug tolerant immunoassay when assessing clinical responses to protein therapeutics. In summary, the ability of the above described ACE assay to detect low levels of ADA in the presence of high levels of drug enables a robust immunogenicity assay that will provide much more flexibility in design and performance of preclinical and clinical trials. The ACE assay retains the desirable properties of earlier bridging assays such as excellent sensitivity and precision, while removing doubts about the presence of excess free drug in samples causing false negative results. Our group has therefore adopted this format to detect ADA antibodies directed against LY monoclonal therapeutics. Acknowledgements The authors would like to thank Anthony Butterfield and Holly Smith of Lilly for helpful discussion and positive non-human primate antibodies, Mark O'Dell and Ron Bowsher of Linco Diagnostic Services, Inc. for performance of assays and assistance with data interpretation and Wendell Smith of Bowsher Brunelle Smith, LLC for statistical analysis of cut point data. References Baert, F., Norman, M., Vermeire, S., Van Assche, G., D’Haens, G., Carbonez, A., Rutgeerts, P., 2003. Influence of immunogenicity on the long-term efficacy of infliximab in Crohn's disease. NEJM 348, 601. Basser, R.L., O’Flaherty, E., Green, M., Edmonds, M., Nichol, J., Menchaca, D.M., Cohen, B., Begley, C.G., 2002. Development of pancytopenia with neutralizing antibodies to thrombopoietin after multicycle chemotherapy supported by megakaryocyte growth and development factor. Blood 99, 2599. Bendtzen, K., Geborek, T., Svenson, M., Larsson, L., Kapetanovic, M.C., Saxne, T., 2006. Individualized monitoring of drug bioavailability and immunogenicity in rheumatoid arthritis patients treated with the tumor necrosis factor α Inhibitor infliximab. Arthritis Rheum. 54, 3782. Bourdage, J.S., Lee, T., Taylor, J.M., Willey, M.B., Brandt, J.T., Konrad, R.J., 2005. Effect of double antigen bridging immunoassay format on antigen coating concentration dependence and implications for designing immunogenicity assays for monoclonal antibodies. J. Pharm. Biomed. Anal. 39, 685. Casadevall, N., Nataf, J., Viron, B., Kolta, A., Kiladjian, J.J., MartinDupont, P., Michaud, P., Papo, T., Ugo, V., Teyssandier,
J.S. Bourdage et al. / Journal of Immunological Methods 327 (2007) 10–17 I., Varet, B., Mayeaux, P., 2002. Pure red cell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin. N. Engl. J. Med. 346, 469. Diamondis, E., Christopoulos, T.K., 1991. The Biotin–(strept)avidin system: principles and applications in biotechnology. Clin. Chem. 37, 625. Dixon, K., 1974. Measurement of antibodies to insulin in serum. Clin. Chem. 10, 1275. Feldman, E., Kalaycio, M., Weiner, G., Frankel, S., Schulman, P., Schwartzberg, P., Juric, J., Velez-Garcia, E., Seiter, K., Scheinberg, D., Levitt, D., Wedel, N., 2003. Treatment of relapsed or refractory acute myeloid leukemia with humanized anti-CD33 monoclonal antibody HuM195. Leukemia, 17, 314. Fineberg, Naomi S., Fineberg, Edwin S., Anderson, James H., Birkett, Martin A., Gibson, Reid G., Hufferd, Suzan, 1996. Immunologic effects of insulin lispro [Lys (B28), Pro (B29) human insulin] in IDDM and NIDDM patients previously treated with insulin. Diabetes 45, 1750. Kansal, P.C., BAndisode, M.S., Boshell, B.R., 1979. Determination of insulin antibodies. Horm. Metab. Res. 11, 187. Looney, R.J., Anolik, J.H., Campbell, D., Felgar, R.E., Young, F., Arend, L.J., Sloand, J.A., Rosenblatt, J., Sanz, I., 2004. B cell depletion as a novel treatment for systemic lupus erythematosus. Arth. Rheum. 50, 2580.
17
Maloney, D.G., Grillo-López, A.J., White, C.A., Bodkin, D., Schilder, R.J., Neidhart, J.A., Janakiraman, N., Foon, K.A., Liles, T.L., Dallaire, B.K., Wey, K., Royston, I., Davis, T., Levy, R., 1997. IDEC-C2B8 (Rituximab) anti-CD20 monoclonal antibody therapy in patients with relapsed low-grade non-Hodgkin's lymphoma. Blood 90, 2188. Mire-Sluis, A.R., Barrett, Y.C., Devanarayan, V., Koren, E., Liu, H., Maia, M., Parish, T., Scott, G., Shankar, G., Shores, E., Swanson, S.J., Taniguchi, G., Wierda, D., Zuckerman, L.A., 2004. Recommendations for the design and optimization of immunoassays used in the detection of host antibodies against biotechnology products,. J. Immunol. Methods 289, 1. Moxness, M., Foley, J., Stene, M., Finco-Kent, D., Bedian, V., Krasner, A., Kawabata, T., 2003. Development and bioligand Binding Assays to measure total, IgA, IgE, IgG and IgM antibodies in human serum. Ann. N.Y. Acad. Sci. 1005, 265. Patton, P., Mullenix, M., Swanson, S., Koren, E., 2005. An acid dissociation bridging ELISA for detection of antibodies directed against therapeutic proteins in the presence of antigen. J. Immunol. Methods 304, 189. Pendley, C., Schantz, A., Wagner, C., 2003. Immunogenicity of therapeutic monoclonal antibodies. Curr. Opin. Mol. Ther. 5, 179.