Systematic comparison of drug-tolerant assays for anti-drug antibodies in a cohort of adalimumab-treated rheumatoid arthritis patients.

JIM-11963; No of Pages 10 Journal of Immunological Methods xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of Immunological Methods journal homepage: www.elsevier.com/locate/jim

Systematic comparison of drug-tolerant assays for anti-drug antibodies in a cohort of adalimumab-treated rheumatoid arthritis patients Karien Bloem a,b,c,⁎, Astrid van Leeuwen c, Gerrit Verbeek c, Michael T. Nurmohamed d, Gerrit Jan Wolbink a,b,d, Desiree van der Kleij c, Theo Rispens a,b a

Sanquin Research, Amsterdam, The Netherlands Landsteiner Laboratory, Academic Medical Centre, University of Amsterdam, The Netherlands c Sanquin Diagnostic Services, Amsterdam, The Netherlands d Jan van Breemen Research Institute/Reade, Amsterdam, The Netherlands b

a r t i c l e

i n f o

Article history: Received 21 November 2014 Received in revised form 22 January 2015 Accepted 23 January 2015 Available online xxxx Keywords: Anti-drug antibodies (ADA) Adalimumab Drug interference Drug-tolerant assay Electrochemoluminescence Radioimmunoassay

a b s t r a c t Drug interference complicates assessment of immunogenicity of biologicals and results in an underestimation of anti-drug antibody (ADA) formation. Drug-tolerant assays have the potential to overcome such limitations. However, to which extent drug-tolerant assays provide an unbiased picture of the antibody response to a biological is unknown. In this study, we compared the measurement of ADA to adalimumab in 94 consecutive adalimumab-treated rheumatoid arthritis patients using the traditional antigen binding test (ABT) and four different drug-tolerant assays, the Ph-shift anti-Idiotype Antigen binding test (PIA) and three newly developed assays for this study: an acid-dissociation radioimmunoassay (ARIA), a temperature-shift radioimmunoassay (TRIA) and an electrochemoluminescence-based assay (ECL). Our results indicate that drug-tolerant assays provide a fairly consistent view on the antibody formation: quantitatively, the results from all four assays correlate well (Spearman r N 0.9). However, the percentage of ADA-positive patients ranges from 51 to 66% between assays, with the ARIA identifying the highest number of patients as positive. These differences are largely due to patients making low amounts of ADA; if ADA levels were above ca. 100 AU/ml, a patient was identified as positive in all four assays. Adalimumab concentrations were significantly lower in ADA-positive samples. Taken together, the results indicate that these different drug-tolerant assays provide a similar and reasonably consistent view on ADA responses, which however, breaks down at the lower end of the detectable range, and highlight that ADA is best reported quantitatively. Furthermore, if an even more sensitive drug-tolerant assay could be developed, one would probably find additional positive samples that will predominantly contain very low levels of ADA. © 2015 Elsevier B.V. All rights reserved.

Abbreviations: ADA, anti-drug antibody; ADL, adalimumab; ABT, antigen binding test; PIA, pH-shift-anti-idiotype Antigen binding test; ARIA, Aciddissociation RadioImmunoAssay; TRIA, Temperature-shift RadioImmunoAssay; ECL, electrochemoluminescence. ⁎ Corresponding author at: Department of Immunopathology, Sanquin Research, Plesmanlaan 125, 1066 CX Amsterdam, The Netherlands. Tel.: +31 20 5123158; fax: +31 20 512 3174. E-mail address: [email protected] (K. Bloem).

1. Introduction The treatment of patients with inflammatory diseases has greatly improved since the introduction of therapeutic monoclonal antibodies (biologicals) such as the TNF blockers adalimumab and infliximab (Polman et al., 2006; Elliott et al., 1994; Weinblatt et al., 2003). Efficacy of the treatment is

http://dx.doi.org/10.1016/j.jim.2015.01.007 0022-1759/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: Bloem, K., et al., Systematic comparison of drug-tolerant assays for anti-drug antibodies in a cohort of adalimumab-treated rheumatoid arthritis pati..., J. Immunol. Methods (2015), http://dx.doi.org/10.1016/j.jim.2015.01.007

2

K. Bloem et al. / Journal of Immunological Methods xxx (2015) xxx–xxx

correlated with the serum drug levels (Pouw et al., 2013; Vande Casteele et al., 2014; Ducourau et al., 2014). One explanation for the differences in drug levels is the development of anti-drug antibodies (ADAs), which can be detected in numerous patients. Patients who develop substantial amounts of ADAs may experience loss of efficacy and have a diminished chance to reach sustained remission (Radstake et al., 2009; Bendtzen et al., 2006; Wolbink et al., 2006; Bartelds et al., 2007, 2011; Pascual-Salcedo et al., 2011; Baert et al., 2003; Garces et al., 2013). Therefore, measuring these ADAs is a useful tool for the clinical decision making (van Schouwenburg et al., 2013b; Tatarewicz et al., 2014). However, the detection of ADAs is hampered by the presence of drug, which is referred to as drug interference. The sensitivity for the presence of drug varies for the different assays (Nechansky, 2010; Allez et al., 2010; Hart et al., 2011), as well as for different drugs (Rispens et al., 2013). Differential drug interference in different assay formats is an important reason why for e.g. adalimumab the reported incidence of antibody formation ranges from 1% to over 50% (Vincent et al., 2013). These limitations may be overcome by the use of drug tolerant assays, which could provide more consistent quantification of the amount of ADA formation in biological-treated patients. Different drug tolerant assays are developed (van Schouwenburg et al., 2010, 2013a; Bourdage et al., 2007; Lofgren et al., 2006, 2007; Patton et al., 2005; Schmidt et al., 2009; Sickert et al., 2008; Zhong et al., 2010; Wang et al., 2012; Llinares-Tello et al., 2014). Most of these assays involve a complex dissociation step, for example by diluting complexes in an acidic environment. In theory, this releases all ADAs from the drug, which makes the ADA accessible for detection. In principle, if all ADAs can be liberated from the drug and detection is unbiased, such assays should provide a glimpse of the ‘true’ extent of antibody formation to a therapeutic protein. To what extent different drug-tolerant assays provide similar or discordant information about ADA formation to a biological has not yet been tested. The antigen binding test (ABT), currently used in our laboratory to measure ADAs, correlates nicely with the clinical response in patients (Bartelds et al., 2011); however, drugtolerant assays could give insight into the early antibody formation and potentially can be used for timely identification of patients at risk for developing an excessive antibody response over time (Plasencia et al., 2013; Baert et al., 2014a,b; Krieckaert et al., 2012). Moreover, to properly study the immunological events that underlie immunogenicity an unbiased method for detecting and analyzing antibody responses is crucial. In particular, it has been recognized that subjects can develop tolerance upon continued treatment, i.e., an initial antibody response may disappear in time (van Schouwenburg et al., 2013a; Vennegoor et al., 2013; Vande Casteele et al., 2013). However, if the assay used to measure ADAs will only pick up antibodies in excess of the vast amounts of drug many such events will go by unnoticed. In this study, we compared the detection of ADAs in the standard ABT and in four drug tolerant assays, of which three were newly developed for this study and are far more suitable for implementation in a routine diagnostic setting compared to the previous developed drug tolerant Ph-shift anti-Idiotype Antigen binding test (PIA) (van Schouwenburg et al., 2013a). Besides samples containing different ADA and adalimumab

concentrations prepared in vitro, we also evaluated a clinical cohort of adalimumab treated RA patients. In addition, we examined the anti-adalimumab and adalimumab levels of the samples that were causing differences in the percentage positively detected samples in the four drug tolerant assays and we evaluated the correlation of ADA levels measured in the different drug tolerant assays. 2. Materials & methods 2.1. Patients Sera were obtained from the first 94 consecutive adalimumab-treated RA patients of a prospective observational cohort as previously described (Bartelds et al., 2011). All patients had a disease activity score in 28 joints (DAS28) of ≥3.2 and fulfilled the American College of Rheumatology 1987 revised criteria for RA. Despite earlier treatment with two disease-modifying anti-rheumatic drugs, including methotrexate, all patients had active disease at the start of adalimumab (Abbott, Illinois, USA) treatment. This was according to the Dutch consensus statement on the initiation and continuation of tumor necrosis factor (TNF) blocking therapy in RA. All patients used 40 mg adalimumab every other week by subcutaneous injection. In patients with an inadequate response, as judged by the treating rheumatologist, dose could be increased to 40 mg every week (n = 20). All blood samples were withdrawn just prior to the next adalimumab injection. Blood samples that were withdrawn prior to the start of adalimumab treatment and at 16, 28, and 52 weeks after start of treatment were evaluated. The study was approved by the ethics committee of the Slotervaart Hospital and the Jan van Breemen Research Institute/Reade. 2.2. Measurement of adalimumab concentration Adalimumab levels were measured by ELISA as described previously (Pouw et al., 2013). 2.3. ABT (Antigen Binding Test) The test was essentially carried out as described previously (Bartelds et al., 2007) and further visualized and explained in Fig. 1 and the Results section. In short, antibodies were

125

I-adalimumab F(ab’)2 adalimumab ADA Prot. A Sepharose

Fig. 1. A traditional ADA assay like the ABT is not or only slightly drugtolerant. In the ABT, IgG antibodies from serum are captured by protein A Sepharose, and specific anti-adalimumab antibodies detected by adding radiolabeled adalimumab F(ab′)2 fragments. If an anti-adalimumab antibody was already in complex with adalimumab from serum, it will go by undetected.



captured using protein A sepharose and ADA was detected using 125I labeled F(ab′)2 adalimumab diluted in Freeze buffer (Sanquin). Antibody levels were compared to a standard serum pool containing ADA levels and expressed in arbitrary units (AU/ml). Patients were said to be positive for ADA if at one time point ADA was above the cut-off of 12 AU/ml. All baseline samples were below the cut-off of 12 AU/ml. 2.4. PIA (the Ph-shift anti-idiotype antigen binding) The test was essentially carried out as described previously (van Schouwenburg et al., 2013a) and visualized in Fig. 2. Differences in the drug tolerant assays are discussed in the Results section. Patient serum (30 μl) diluted thirty-fold in PA buffer (phosphate buffered saline/0.3% bovine serum albumin) was mixed with 30 μl of 0.1 M glycine–HCl (pH 2.5). After 30 min incubation at room temperature 30 μl of rabbit antiidiotype Fab directed against adalimumab (67 μg/ml diluted in PA) was added. Then the pH was neutralized by addition of 30 μl of 0.2 M Tris. ADA levels were measured in the ABT. A cutoff of 48 AU/ml was determined as the mean + 3 SD, based on the measurements on all available baseline sera.

3

2.5. ARIA (acid-dissociation radioimmunoassay) In the ARIA (see Fig. 2 and Results section for schematic setup), 1 μl of patient serum, diluted fifty-fold in 30 mM acetic acid adjusted to a pH between 2 and 3 with hydrochloric acid (see below), was incubated for 30 min, followed by 7-fold dilution in 100 mM phosphate buffer (pH 7.6) that contained 50 mM of NaCl, 0.3% bovine serum albumin (BSA) and 0.02% Tween, as well as 5–100 ng adalimumab F(ab′)2-biotin and 20 μg of IVIg F(ab′)2 fragments (Rispens et al., 2012). Subsequently, protein A sepharose was added and the volume was adjusted to 850 μl with PBS/0.3% bovine serum albumin (BSA) and 0.02% Tween (PA buffer) and incubated overnight. After washing with PBS containing 0.02% Tween, ADA with bound adalimumab F(ab′)2-biotin fragments was detected making use of 125I labeled streptavidin. For optimization of the assay several concentrations of adalimumab F(ab′)2-biotin between 5 and 100 ng/test were investigated. To assess the drug tolerance of the assay the 150 kDa fraction of a serum pool of 9 different patients containing high ADA levels and no detectable adalimumab was spiked with 2, 10, and 50 μg/ml adalimumab. The percentage recovery of ADA in the

complex dissociation/competition

detection 125I

bt

bt

1. acid

streptavidin ARIA

2. neutralize / ADL F(ab)2-bt

Prot. A Sepharose

bt

bt

ADL F(ab)2-bt / 37 oC / 16 hr

TRIA

1. acid PIA

2. neutralize / anti-ADL Fab

1. acid

ECL

2. neutralize / ADL-bt / ADLsulfo-tag

ADA

sulfo-tag

bt

bt

adalimumab (ADL)

streptavidin

rabbit anti-ADL Fab

Fig. 2. Drug-tolerant assays to measure anti-adalimumab antibodies. In order to also detect anti-adalimumab in complex with adalimumab, these complexes need to dissociate first. In the ARIA, this is accomplished by acid dissociation. Upon neutralization, biotinylated adalimumab F(ab′)2 is added in amounts that can compete with adalimumab present in the serum sample. Antibodies are subsequently captured by protein A Sepharose and specific anti-adalimumab antibodies detected by adding radiolabeled streptavidin. In the TRIA, biotinylated adalimumab F(ab′)2 competes with adalimumab from serum by incubation for a prolonged time at 37 °C, giving rise to exchange instead of an acidification step which is used in the other assays. Detection is carried out similar to the ARIA. The PIA uses acid dissociation followed by neutralization and addition of rabbit anti-adalimumab Fab fragments to prevent complex re-formation. Detection is then carried out as in the regular ABT. The ECL format also relies on acid dissociation. Here, biotinylated adalimumab and sulfo-tagged adalimumab are added to compete with adalimumab from serum. Part of the complexes that form after neutralization will contain both a molecule of biotinylated adalimumab and a sulfo-tagged adalimumab. These complexes will yield a signal after binding to streptavidin-coated ECL plates.


4


presence of adalimumab improved using higher concentrations of adalimumab F(ab′)2-biotin, but the incremental improvement diminished with higher concentrations. In combination with this, the incremental background signals of sera from healthy donors aggravated with the use of higher concentrations. Therefore, 30 ng/test was chosen as concentration with good ADA recovery together with an acceptable background signal. Furthermore, background signal and acid dissociation in acetic acid and glycine–HCl were tested at pH 2, 2.5, and 3. No improvement in the percentage recovery was observed with lower pHs, while the background signal slightly increased at lower pHs, as well as with using glycine– HCl instead of acetic acid. Therefore, we chose acid dissociation at pH 3 in acetic acid as the most optimal method. The cut-off of 30 AU/ml was determined as the mean + 3 SD from 90 baseline patient sera.

assay by excluding the formation of dimeric complexes, the use of labeled Fab′ fragments was investigated. Indeed, this increased the sensitivity of the assay in ADA samples without adalimumab; however the drug tolerance of the assay was greatly reduced. Therefore, labeled intact adalimumab was used in the assay in combination with human immunoglobulins (GammaQuin, Sanquin, Amsterdam, The Netherlands) to block bridging of labeled adalimumab via the Fc tail by rheumatoid factors. Background signals of baseline serum samples were often, but not always, slightly below the value of the blank and not normally distributed. Therefore, we have chosen a cut-off of 125% of the blank value, which corresponds with 3.9 AU/ml, analogously to Zhong et al. (2010). 99% of the baseline samples are below 3.9 AU/ml.

2.6. TRIA (temperature-shift radioimmunoassay)

Comparisons were carried out using either Dunn's multiple comparison test for differences in adalimumab concentrations between the three groups: positive in all drug tolerant assays (AllPos), positive in 1, 2 or 3 of the drug tolerant assays (123Pos) or negative in all the drug tolerant assays (AllNeg) and for differences in percentage positively detected samples by the five tested assay. Mann–Whitney's test was used for differences in adalimumab concentrations in the ARIA positive and ARIA negative group. Correlations were calculated as Spearman correlation coefficients, all using Graphpad Prism 6.0.

In the TRIA (basic characteristics discussed in Fig. 2 and Results section), 1 μl of patient serum diluted fifty-fold in PBS/0.3% BSA was incubated overnight at 37 degrees in the presence of adalimumab F(ab′)2-biotin and 20 μg IVIg F(ab′)2 fragments (Rispens et al., 2012). The next day protein A sepharose was added and the total volume was adjusted to 850 μl with PA buffer, and incubated at room temperature for a minimum of 5 h. ADA with bound adalimumab F(ab′)2-biotin fragments was detected making use of 125I labeled streptavidin. Analogous to the ARIA, the TRIA was optimized for the concentration of adalimumab F(ab′)2-biotin and 30 ng/test was found to be optimal also for the TRIA. A cut-off of 33 AU/ml was determined as the mean + 3 SD from 90 baseline sera. 2.7. ECL (electrochemoluminescence) The ECL assay (basic characteristics discussed in the Results section and Fig. 2) was essentially carried out as the acid dissociation protocol for bridging immunogenicity assays described in the Guidelines for Assay Development provided by MesoScale Discovery (Gaithersbrug, MD). In short, patient serum was diluted 1/10 in acetic acid and incubated with moderate shaking for 45 min at room temperature. Biotinylated and sulfo-tagged adalimumab were added in combination with Tris and incubated at room temperature with moderate shaking to allow the formation of complexes of ADA with labeled adalimumab. This mixture was incubated on preblocked Streptavidin Gold plates (MesoScale Discovery, Inc., Gaithersbrug, MD) with more vigorous shaking. After washing in PBS/0.05% Tween bound ADA-adalimumab complexes containing a sulfo-tag were detected in read buffer and analyzed with a SECTOR Imager 2400A reader (MesoScale Discovery, Inc., Gaithersbrug, MD, USA). The assay was optimized with different labeling degrees of adalimumab and several concentrations of labeled adalimumab. Increased concentrations of labeled adalimumab improved drug tolerance, but the sensitivity of the assay decreased due to a limited capacity of the Streptavidin Gold plates; therefore 0.67 μg/ml of both types of labeled adalimumab in the final mixtures was chosen as the most optimal concentration. To circumvent false negatives due to the presence of rheumatoid factor and to optimize sensitivity of the

2.8. Statistics

3. Results Three new drug tolerant assays were developed and compared to our currently used diagnostic assay, the ABT, and a previously developed drug tolerant assay (PIA). The drugtolerant assays differ in the way drug-tolerance is induced (acid; temperature), the way re-formation of drug–anti-drug complexes is prevented (competition; blocking of drug), and the technique by which ADAs are revealed, as will be explained below. The five different assay formats are schematically depicted in Figs. 1 and 2. The ABT uses protein A coupled to Sepharose to capture serum IgG antibodies, followed by detection of specific anti-adalimumab antibodies using radiolabeled adalimumab F(ab′)2 fragments. In the acid-dissociation radioimmunoassay (ARIA), complexes are dissociated using acid, followed by neutralization and incubation with biotinylated adalimumab F(ab′)2 fragments that will compete with adalimumab in the sample. Afterwards, IgG is captured using Protein A Sepharose as in the ABT and bound adalimumab F(ab′)2-biotin is quantified using radiolabeled streptavidin. In the temperature-shift radioimmunoassay (TRIA), competition between adalimumab in the sample and biotinylated adalimumab F(ab′)2 is accomplished by overnight incubation at 37 °C. Detection is similar to the ARIA. In the previously developed PIA, acid dissociation is followed by neutralization and addition of rabbit antiadalimumab Fab fragments that will block adalimumab in the sample. Detection can subsequently be carried out essentially following the protocol of the ABT. Finally, we also developed an ADA assay based on electrochemoluminescence (ECL). This platform has gained popularity for pre-marketing ADA testing and can be used for drug-tolerant ADA assays.



3.1. Evaluation of drug tolerance using spike samples

An acid-dissociation step is followed by neutralization and addition of both biotinylated adalimumab and sulfo-tagged adalimumab. Of all the complexes that will form, a fraction will contain both forms of labeled adalimumab, in principle in proportion to the amount of ADA present in the sample. Capture of these complexes using proprietary, streptavidincoated plates is followed by detection of bound complexes that contain a sulfo-tag.

anti-ADL measured (kAU/ml)

Each of the assays was optimized to minimize background while achieving optimal drug-tolerance and sensitivity as detailed in the Materials and methods section. To test the drug tolerance of the different assays we made samples containing different amounts of anti-drug antibodies (made from the monomeric antibody fraction of a serum pool of nine

ABT

3

ADL (g/ml) 0

2

1

2

10 50

0 0.01

0.1 1 anti-ADL (kAU/ml)

10

ARIA

TRIA ADL (g/ml) 0 2

2

10 1

50

3



3

0

ADL (g/ml) 0

2

2 10

1

50 0

0.01

0.1

10

1

0.01

anti-ADL (kAU/ml)

0 1

24

0 1

anti-ADL (kAU/ml)

10

3


2

0.1

1

10

ECL ADL (g/ml)

0.01

0.1

anti-ADL (kAU/ml)

PIA

3


5

ADL (g/ml) 0

2

2

10

1

50 0 0.01

0.1

1

10

anti-ADL (kAU/ml) Fig. 3. Drug interference in different assays to measure antibodies to adalimumab. A standard containing ADAs (but no adalimumab or ADA-adalimumab complexes) pooled from 9 sera is titrated in the presence of several concentrations of adalimumab and evaluated in the different assays. For the PIA, the standard consists of rabbitanti-adalimumab (data from van Schouwenburg et al., 2010).


6


A

B

ABT

A R IA

T R IA

PIA

ECL

C *** 20 15 10 5

N IA AR

AR

IA

Po

s

eg

0

Fig. 4. Evaluation of anti-adalimumab antibodies in 94 RA patients treated with adalimumab using different ADA assays. A) Sera that were evaluated were drawn prior to start of treatment and after 16, 28, and 52 weeks of treatment. B) ADA levels measured in the ARIA of samples that were found positive in all drug-tolerant assays (AllPos) and for samples that were positive in at least one drug-tolerant assay, but not in all four (123Pos). C) Left panel: adalimumab concentrations in samples as defined for B, as well as in samples that were negative in all assays (AllNeg). Right panel: adalimumab concentrations in samples that are either positive or negative in the ARIA, ***: p b 0.001. Pre-treatment samples not included.



ADL treated patients) in combination with different concentrations of drug. For the PIA a similar experiment was performed previously making use of a standard with rabbitanti-drug antibodies in the presence of 24 μg/ml ADL (van Schouwenburg et al., 2010). In Fig. 3, titrations of ADA standard in the presence of different concentrations of adalimumab are shown for the different assays. Whereas in the ABT 2 μg/ml of adalimumab already results in a dramatic decrease in signal, all drug-tolerant assays can tolerate substantial amounts of adalimumab. The ARIA is hardly affected by 10 μg/ml of adalimumab, whereas the other assays yield somewhat lower signals at this adalimumab concentration. All assays are substantially affected by 50 μg/ml of adalimumab, but will still be able to identify a sample containing ca. 200 AU/ml as positive. 3.2. Evaluation of drug tolerance in a cohort of adalimumab-treated RA patients In order to evaluate how the different assays will perform in a real-life setting, we measured ADAs in a panel of 94 adalimumab-treated RA patients. Serum samples drawn before treatment as well as after 16, 28, and 52 weeks after start of treatment were included in the analysis. The results are shown in Fig. 4. Pre-treatment samples are invariably negative, except for 1 sample that tested positive in the ECL assay. All drug tolerant assays detect anti-drug antibodies in quite a large portion of the patients (between 51 and 66%; Table 1), which is significantly more compared to the less drug tolerant ABT (15%). A total of 150 samples were found positive in at least one drug-tolerant assay; with 94 samples being positive in all assays. The ARIA is the most sensitive assay and 142 samples were found positive, whereas in the TRIA, PIA, and ECL assay 116, 102, and 116 samples were positive, respectively. The ARIA detects significantly more samples as positive, while there is no significantly difference between the percentage positively detected samples of the other drug tolerant assays. If ADA levels (as measured in the ARIA) are compared between samples that are positive in all drug-tolerant assays and samples that are tested positive in at least one assay, but not in all four, it is clear that samples containing larger amounts of ADA (i.e., above ca. 100 AU/ml) are consistently identified as positive in all assays, but samples containing smaller amounts (below ca. 100 AU/ml) may be negative in one or more of the assays (Fig. 4B). Thus, although different numbers of positive samples, and not always the same positive samples, were detected using the different assay, these differences are confined to samples that contain small to moderate amounts of ADA. We also measured adalimumab concentrations in samples drawn after start of treatment (van Schouwenburg et al., 2013a). Adalimumab concentrations were highest in samples that were tested ADA negative in all assays (median 8.83 mg/l, Table 1 Number of patients identiﬁed as ADA+ in the different assays during one year of follow-up.

n %

ABT

ARIA

TRIA

PIA

ECL

14/94 14.9

48/94 66.0

54/94 57.4

48/94 51.1

54/94 57.4

7

IQR 6.23–11.4), and significantly lower both in samples that were tested ADA positive in some but not all assays (median 6.41 mg/l, IQR 4.45–8.07, p b 0.001), as well as in samples that were tested ADA positive in all assays (median 4.60 mg/l, IQR 1.28–7.23, p b 0.001) (Fig. 4C). The latter two groups did not differ significantly. ARIA-positive samples contained significantly less adalimumab than ARIA-negative samples (median 5.59 mg/l, IQR 2.66–7.67 vs 8.47 mg/l, IQR 6.09–11.3, p b 0.001) (Fig. 4C). Furthermore, levels of antibodies correlate quite well between any pair of drug-tolerant assays, as can be seen in Fig. 5. For any pair of assays, the correlation coefficient was above 0.9. In Fig. 6, examples are shown of the time-course of ADA formation as measured in the different assays for individual patients. For many patients, the overall picture is similar between the different assays, but with subtle differences, e.g., a transient response may render negative results in some assays after e.g. 52 weeks of treatment, whereas in another assay there are still antibodies detected at this time point. However, since the ARIA is the most sensitive assay, ADA is detected for several patients at multiple time points in the ARIA, whereas the other assays do not find any (not shown). The antibody titers of the samples that are ADA positive in the ARIA, but not in the other assays tested, all have low antibody detection (below 100 AU/ml when measured in the ARIA). 4. Discussion Drug interference complicates the study and assessment of immunogenicity of biologicals and results in an underestimate of antibody formation. Early antibody responses are difficult to study, because the expected levels of ADA are relatively small while drug concentrations will be high. Drug-tolerant assays have the potential to overcome such limitations. However, to which extent drug-tolerant assays provide an unbiased picture of the antibody response to a biological is unknown. The systematic comparison of different drug-tolerant assays using a cohort of adalimumab-treated RA patients presented in this study provides some insight in this matter. Taken together, the results indicate that a similar and consistent view on ADA responses is provided using different drug tolerant assays. Nevertheless, this coherent view on ADA responses breaks down at the lower end of the detectable range. Furthermore, based on these results we assume that if an even more sensitive drug-tolerant assay could be developed, additional positive samples with predominantly low levels of ADA will be discovered. Antibodies are consistently detected in all four drug tolerant assays (i.e., all samples above ca. 100 AU/ml were positive in all drug-tolerant assays; and overall correlation is N0.9 between all drug-tolerant assay formats). On the other hand, samples that are weakly positive in one assay may be negative in another assay. Overall, the ARIA has a slightly higher drug tolerance and/or sensitivity compared to the other drug tolerant assays and picks up the most positive samples. Most samples that are ADA positive in one or more of the assays, but not in all four, are in fact only ADA positive in the ARIA. Since these samples contain on average significantly less adalimumab compared to the samples that were ADA negative in all assays (Fig. 4C), it suggests that patients found positive in


8


1000

TRIA (AU/ml)

TRIA (AU/ml)

1000

100

100

r=0.94 10

100 ARIA (AU/ml)

r=0.91 10 10

1000

ECL (AU/ml)

100

100

10

r=0.91 10

100 ARIA (AU/ml)

r=0.91 1 10

1000

100 PIA (AU/ml)

ECL (AU/ml)

100

10

100

10 r=0.92

r=0.91 1 10

1000

1000

1000

ECL (AU/ml)

1000

1000

1000

PIA (AU/ml)

100 PIA (AU/ml)

100 ARIA (AU/ml)

1000

1 10

100 TRIA (AU/ml)

1000

Fig. 5. Correlation between the different drug tolerant assays. Dashed lines indicate cut-off; the grey dotted line indicates x = y.

the ARIA, in addition to the other assays, may still produce antibodies that negatively affect drug concentrations, even if the levels of ADA appear to be low. A surprisingly high fraction of the RA patients treated with adalimumab that were examined in this study makes antibodies to adalimumab. This is consistent with a previous study using a drug-tolerant assay (van Schouwenburg et al., 2013a), which showed that in many cases the antibodies may not affect clinical efficacy of the adalimumab treatment. The primary effect of antibody formation will be the neutralization of adalimumab (van Schie et al., 2015), and treatment efficacy

will depend largely on the amount of adalimumab that is not blocked by ADAs. In other words, in patients that make low quantities of ADA the treatment efficacy may be essentially unaffected, and many of these patients do not persist in making antibodies. However, in some patients, the ADA response develops in time from low levels at early time points to high levels at later time points. It would be interesting to investigate to which extent early detection of (low amounts of) ADA formation could be predictive of future loss of response due to immunogenicity. We expect that drug-tolerant assays like those examined in this study may be a useful tool to investigate



A RI A

T RI A

30

0

100 50 20 40 time (weeks)

150 100 50 0

60

20 40 time (weeks)

100 50 0

100 50 0

0

20 40 time (weeks)

60

anti-ADL(AU/ml)

0

150 100 50

20 40 time (weeks)

60

30

20 10

0

60

20 40 time (weeks)

60

20 40 time (weeks)

60

30

150 100 50 0

0

10 20 time (weeks)

0 20 40 time (weeks)

200

150

50

30

0

anti-ADL(AU/ml)

150

100

0

30

0

60

200 anti-ADL(AU/ml)

200

10 20 time (weeks)

200

0 0

200 0 0

30

anti-ADL(AU/ml)

150

0

anti-ADL(AU/ml)

10 20 time (weeks)

200 anti-ADL(AU/ml)

anti-ADL(AU/ml)

200

pt 3

200

400

anti-ADL(AU/ml)

10 20 time (weeks)

400

150

anti-ADL(AU/ml)

200

E CL

600 anti-ADL(AU/ml)

400

0 0

pt 2

P IA

600 anti-ADL(AU/ml)

pt 1

anti-ADL(AU/ml)

600

9

20 10 0

0

20 40 time (weeks)

60

0

Fig. 6. Examples of ADA formation for three patients as measured with the different drug-tolerant assays. The patterns of antibody production measured in the four different drug tolerant assays are similar for most patients. Different trends are observed, including patients that transiently develop antibodies, and patients that show increasing antibody production in time.

the factors that differentiate between the apparent development of tolerance in many patients and the (uncontrolled?) expansion of ADA-producing B cells in some patients. The acidic conditions used in most drug-tolerant assays are potentially too harsh for some antibodies to survive without unfolding and/or loss of the capability to bind to the drug. Therefore, we also tested one assay format that uses milder conditions to induce drug-tolerance. In the TRIA, prolonged incubation at 37 °C would theoretically allow competition of labeled drug with drug in the sample for binding to ADA. However, in practice, the ARIA was superior in terms of drug-tolerance compared to the TRIA. Furthermore, we did not observe appreciable loss of reactivity when the standard or monoclonal, patient-derived antibodies (van Schouwenburg et al., 2014) were acidified prior to testing (not shown). Therefore, acid dissociation appears to be a suitable procedure to dissociate ADA–drug complexes. Although at first sight the ECL assay appears to be substantially more sensitive compared to e.g. the ARIA, the ARIA outperforms the ECL assay and identifies a larger fraction of ADA positive samples. This is in part due to the superior drug tolerance of the ARIA at higher concentrations of adalimumab, as is apparent from the spike experiments (Fig. 3). Furthermore, we found that at the lower end, a titration of different samples did not always run parallel in the ECL assay. For this reason, units as determined in the different radioimmunoassays cannot be directly compared to units determined in the ECL assay. This warrants caution when comparing different assay formats, but is hardly a surprising finding. Factors such as affinity, isotype, and

the ability of the ADA to cross-link (e.g. IgG4 antibodies may be unable to cross-link) can differentially affect to which extent a particular assay will render a positive signal, and quantitation of the ADA response will always reflect such limitations. The fact that samples with relatively low levels of ADA appear to systematically yield lower units in the ECL assay as compared to the other assays may furthermore indicate that the standard is not fully representative for these samples. In summary, we compared the measurement of ADA to adalimumab in a cohort of adalimumab-treated RA patients using four different drug-tolerant assays and the standard antigen binding test. Our results indicate that drug-tolerant assays provide a fairly consistent view on the antibody formation, correlate well quantitatively, but differ in the capability to identify samples containing low amounts of ADA as positive. Consequently, when results are simply expressed as the percentage of patients positive for ADA, the results are variable between assays. Acknowledgment K Bloem is supported by the MRC grant: Psoriasis Stratification to Optimise Relevant Therapy (PSORT), MR/L011808/1. References Allez, M., Karmiris, K., Louis, E., Van Asche, G., Ben-Horin, S., Klein, A., Van der Woude, J., Baert, F., Eliakim, R., Katsanos, K., Brynskov, J., Steinwurz, F., Danese, S., Vermeire, S., Teillaud, J.L., Lémann, M., Chowers, Y., 2010. Report


10


of the ECCO pathogenesis workshop on anti-TNF therapy failures in inflammatory bowel diseases: definitions, frequency and pharmacological aspects. J. Crohns Colitis. 4, 355. Baert, F., Noman, 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. N. Engl. J. Med. 348, 601. Baert, F., Drobne, D., Gils, A., Vande Casteele, N., Hauenstein, S., Singh, S., Lockton, S., Rutgeerts, P., Vermeire, S., 2014a. Early trough levels and antibodies to infliximab predict safety and success of reinitiation of infliximab therapy. Clin. Gastroenterol. Hepatol. 12, 1474. Baert, F., Vande Casteele, N., Tops, S., Noman, M., Van Asche, G., Rutgeerts, P., Gils, A., Vermeire, S., Ferrante, M., 2014b. Prior response to infliximab and early serum drug concentrations predict effects of adalimumab in ulcerative colitis. Aliment. Pharmacol. Ther. 40, 1324. Bartelds, G.M., Wijbrandts, C.A., Nurmohamed, M.T., Stapel, S., Lems, W.F., Aarden, L., Dijkmans, B.A., Tak, P.P., Wolbink, G.J., 2007. Clinical response to adalimumab: relationship to anti-adalimumab antibodies and serum adalimumab concentrations in rheumatoid arthritis. Ann. Rheum. Dis. 66, 921. Bartelds, G.M., Krieckaert, C.L., Nurmohamed, M.T., van Schouwenburg, P.A., Lems, W.F., Twisk, J.W., Dijkmans, B.A., Aarden, L., Wolbink, G.J., 2011. Development of antidrug antibodies against adalimumab and association with disease activity and treatment failure during long-term follow-up. JAMA 305, 1460. Bendtzen, K., Geborek, P., 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 alpha inhibitor infliximab. Arthritis Rheum. 54, 3782. Bourdage, J.S., Cook, C.A., Farrington, D.L., Chain, J.S., Konrad, R.J., 2007. An Affinity Capture Elution (ACE) assay for detection of anti-drug antibody to monoclonal antibody therapeutics in the presence of high levels of drug. J. Immunol. Methods 327, 10. Ducourau, E., Ternant, D., Lequerre, T., Fuzibet, P., Le Loët, X., Watier, H., Goupille, P., Paintaud, G., Vittecoq, O., Mulleman, D., 2014. Towards an individualised target concentration of adalimumab in rheumatoid arthritis. Ann. Rheum. Dis. 73, 1428. Elliott, M.J., Maini, R.N., Feldmann, M., Kalden, J.R., Antoni, C., Smolen, J.S., Leeb, B., Breedveld, F.C., Macfarlane, J.D., Bijl, H., et al., 1994. Randomised doubleblind comparison of chimeric monoclonal antibody to tumour necrosis factor alpha (cA2) versus placebo in rheumatoid arthritis. Lancet 344, 1105. Garces, S., Demengeot, J., Benito-Garcia, E., 2013. The immunogenicity of antiTNF therapy in immune-mediated inflammatory diseases: a systematic review of the literature with a meta-analysis. Ann. Rheum. Dis. 72, 1947. Hart, M.H., de Vrieze, H., Wouters, D., Wolbink, G.J., Killestein, J., de Groot, E.R., Aarden, L.A., Rispens, T., 2011. Differential effect of drug interference in immunogenicity assays. J. Immunol. Methods 372, 196. Krieckaert, C., Rispens, T., Wolbink, G., 2012. Immunogenicity of biological therapeutics: from assay to patient. Curr. Opin. Rheumatol. 24, 306. Llinares-Tello, F., Rosas-Gomez de Salazar, J., Senabre-Gallego, J.M., SantosSoler, G., Santos-Ramirez, C., Salas-Heredia, E., Barber-Valles, X., MolinaGarcia, J., 2014. Practical application of acid dissociation in monitoring patients treated with adalimumab. Rheumatol. Int. 34, 1701. Lofgren, J.A., Wala, I., Koren, E., Swanson, S.J., Jing, S., 2006. Detection of neutralizing anti-therapeutic protein antibodies in serum or plasma samples containing high levels of the therapeutic protein. J. Immunol. Methods 308, 101. Lofgren, J.A., Dhandapani, S., Pennucci, J.J., Abbott, C.M., Mytych, D.T., Kaliyaperumal, A., Swanson, S.J., Mullenix, M.C., 2007. Comparing ELISA and surface plasmon resonance for assessing clinical immunogenicity of panitumumab. J. Immunol. 178, 7467. Nechansky, A., 2010. HAHA—nothing to laugh about. Measuring the immunogenicity (human anti-human antibody response) induced by humanized monoclonal antibodies applying ELISA and SPR technology. J. Pharm. Biomed. Anal. 51, 252. Pascual-Salcedo, D., Plasencia, C., Ramiro, S., Nuno, L., Bonilla, G., Nagore, D., Ruiz Del Agua, A., Martinez, A., Aarden, L., Martin-Mola, E., Balsa, A., 2011. Influence of immunogenicity on the efficacy of long-term treatment with infliximab in rheumatoid arthritis. Rheumatology (Oxford) 50, 1445. Patton, A., Mullenix, M.C., Swanson, S.J., 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. Plasencia, C., Pascual-Salcedo, D., Alcocer, P., Bonilla, M.G., Villalba, A., Peiteado, D., Arribas, F., Diez, J., Lopez-Casla, M.T., Martin-Mola, E., Balsa, A., 2013. The timing of serum infliximab loss, or the appearance of antibodies to infliximab (ATI), is related with the clinical activity in ATI-positive patients with rheumatoid arthritis treated with infliximab. Ann. Rheum. Dis. 72, 1888. Polman, C.H., O'Connor, P.W., Havrdova, E., Hutchinson, M., Kappos, L., Miller, D.H., Phillips, J.T., Lublin, F.D., Giovannoni, G., Wajgt, A., Toal, M., Lynn, F.,

Panzara, M.A., Sandrock, A.W., 2006. A randomized, placebo-controlled trial of natalizumab for relapsing multiple sclerosis. N. Engl. J. Med. 354, 899. Pouw, M.F., Krieckaert, C.L., Nurmohamed, M.T., van der Kleij, D., Aarden, L., Rispens, T., Wolbink, G., 2013. Key findings towards optimising adalimumab treatment: the concentration–effect curve. Ann. Rheum. Dis. http://dx.doi.org/10.1136/annrheumdis-2013-204172. Radstake, T.R., Svenson, M., Eijsbouts, A.M., van den Hoogen, F.H., Enevold, C., van Riel, P.L., Bendtzen, K., 2009. Formation of antibodies against infliximab and adalimumab strongly correlates with functional drug levels and clinical responses in rheumatoid arthritis. Ann. Rheum. Dis. 68, 1739. Rispens, T., de Vrieze, H., de Groot, E., Wouters, D., Stapel, S., Wolbink, G.J., Aarden, L.A., 2012. Antibodies to constant domains of therapeutic monoclonal antibodies: anti-hinge antibodies in immunogenicity testing. J. Immunol. Methods 375, 93. Rispens, T., Hart, M.H., Ooijevaar-de Heer, P., van Leeuwen, A., Vennegoor, A., Killestein, J., Wolbink, G.J., van der Kleij, D., 2013. Drug interference in immunogenicity assays depends on valency. J. Pharm. Biomed. Anal. 85, 179. Schmidt, E., Hennig, K., Mengede, C., Zillikens, D., Kromminga, A., 2009. Immunogenicity of rituximab in patients with severe pemphigus. Clin. Immunol. 132, 334. Sickert, D., Kroeger, K., Zickler, C., Chokote, E., Winkler, B., Grenet, J.M., Legay, F., Zaar, A., 2008. Improvement of drug tolerance in immunogenicity testing by acid treatment on Biacore. J. Immunol. Methods 334, 29. Tatarewicz, S.M., Mytych, D.T., Manning, M.S., Swanson, S.J., Moxness, M.S., Chirmule, N., 2014. Strategic characterization of anti-drug antibody responses for the assessment of clinical relevance and impact. Bioanalysis 6, 1509. van Schie, K.A., Hart, M.H., de Groot, E.R., Kruithof, S., Aarden, L.A., Wolbink, G.J., Rispens, T., 2015. The antibody response against human and chimeric antiTNF therapeutic antibodies primarily targets the TNF binding region. Ann. Rheum. Dis. 74, 311. van Schouwenburg, P.A., Bartelds, G.M., Hart, M.H., Aarden, L., Wolbink, G.J., Wouters, D., 2010. A novel method for the detection of antibodies to adalimumab in the presence of drug reveals “hidden” immunogenicity in rheumatoid arthritis patients. J. Immunol. Methods 362, 82. van Schouwenburg, P.A., Krieckaert, C.L., Rispens, T., Aarden, L., Wolbink, G.J., Wouters, D., 2013a. Long-term measurement of anti-adalimumab using pH-shift-anti-idiotype antigen binding test shows predictive value and transient antibody formation. Ann. Rheum. Dis. 72, 1680. van Schouwenburg, P.A., Rispens, T., Wolbink, G.J., 2013b. Immunogenicity of anti-TNF biologic therapies for rheumatoid arthritis. Nat. Rev. Rheumatol. 9, 164. van Schouwenburg, P.A., Kruithof, S., Votsmeier, C., van Schie, K., Hart, M.H., de Jong, R.N., van Buren, E.E., van Ham, M., Aarden, L., Wolbink, G., Wouters, D., Rispens, T., 2014. Functional analysis of the anti-adalimumab response using patient-derived monoclonal antibodies. J. Biol. Chem. 289, 34482. Vande Casteele, N., Gils, A., Singh, S., Ohrmund, L., Hauenstein, S., Rutgeerts, P., Vermeire, S., 2013. Antibody response to infliximab and its impact on pharmacokinetics can be transient. Am. J. Gastroenterol. 108, 962. Vande Casteele, N., Feagan, B.G., Gils, A., Vermeire, S., Khanna, R., Sandborn, W.J., Levesque, B.G., 2014. Therapeutic drug monitoring in inflammatory bowel disease: current state and future perspectives. Curr. Gastroenterol. Rep. 16, 378. Vennegoor, A., Rispens, T., Strijbis, E.M., Seewann, A., Uitdehaag, B.M., Balk, L.J., Barkhof, F., Polman, C.H., Wolbink, G., Killestein, J., 2013. Clinical relevance of serum natalizumab concentration and anti-natalizumab antibodies in multiple sclerosis. Mult. Scler. 19, 593. Vincent, F.B., Morand, E.F., Murphy, K., Mackay, F., Mariette, X., Marcelli, C., 2013. Antidrug antibodies (ADAb) to tumour necrosis factor (TNF)-specific neutralising agents in chronic inflammatory diseases: a real issue, a clinical perspective. Ann. Rheum. Dis. 72, 165. Wang, S.L., Ohrmund, L., Hauenstein, S., Salbato, J., Reddy, R., Monk, P., Lockton, S., Ling, N., Singh, S., 2012. Development and validation of a homogeneous mobility shift assay for the measurement of infliximab and antibodies-toinfliximab levels in patient serum. J. Immunol. Methods 382, 177. Weinblatt, M.E., Keystone, E.C., Furst, D.E., Moreland, L.W., Weisman, M.H., Birbara, C.A., Teoh, L.A., Fischkoff, S.A., Chartash, E.K., 2003. Adalimumab, a fully human anti-tumor necrosis factor alpha monoclonal antibody, for the treatment of rheumatoid arthritis in patients taking concomitant methotrexate: the ARMADA trial. Arthritis Rheum. 48, 35. Wolbink, G.J., Vis, M., Lems, W., Voskuyl, A.E., de Groot, E., Nurmohamed, M.T., Stapel, S., Tak, P.P., Aarden, L., Dijkmans, B., 2006. Development of antiinfliximab antibodies and relationship to clinical response in patients with rheumatoid arthritis. Arthritis Rheum. 54, 711. Zhong, Z.D., Dinnogen, S., Hokom, M., Ray, C., Weinreich, D., Swanson, S.J., Chirmule, N., 2010. Identification and inhibition of drug target interference in immunogenicity assays. J. Immunol. Methods 355, 21.


Golimumab trough levels, antidrug antibodies and clinical response in patients with rheumatoid arthritis treated in daily clinical practice.

High frequency of antidrug antibodies and association of random drug levels with efficacy in certolizumab pegol-treated patients with rheumatoid arthritis: results from the BRAGGSS cohort.

The effect of antidrug antibodies on the sustainable efficacy of biologic therapies in rheumatoid arthritis: practical consequences.

Mycoplasma antibodies in rheumatoid arthritis.

Antibodies to DNA in patients with rheumatoid arthritis and juvenile rheumatoid arthritis.

Rheumatoid arthritis and collagen antibodies.

Anti-keratin antibodies in rheumatoid arthritis.

Anti-neutrophil cytoplasm antibodies in rheumatoid arthritis.

Adaptive immunity in rheumatoid arthritis: anticitrulline and other antibodies in the pathogenesis of rheumatoid arthritis.

A comparison of the malignancy incidence among patients with psoriatic arthritis and patients with rheumatoid arthritis in a large US cohort.

Prevalence of biologics monotherapy in a cohort of patients with Rheumatoid Arthritis in daily clinical practice.

Psychological correlates of fatigue in rheumatoid arthritis: a systematic review.

Comorbidities in patients with psoriatic arthritis: a comparison with rheumatoid arthritis and psoriasis.

Effect of Anti-TNF Antibodies on Clinical Response in Rheumatoid Arthritis Patients: A Meta-Analysis.

Prevalence and clinical prediction of osteoporosis in a contemporary cohort of patients with rheumatoid arthritis.

Performance evaluation of FlowCytomix assays to quantify cytokines in patients with rheumatoid arthritis.

Prevalence of sexual dysfunction among female patients followed in a Brasília Cohort of early rheumatoid arthritis.

Clinical utility of random anti-tumor necrosis factor drug-level testing and measurement of antidrug antibodies on the long-term treatment response in rheumatoid arthritis.

Levamisole, rheumatoid arthritis, and cold lymphocytotoxic antibodies.

Increased Risk of Acute Pancreatitis in Patients with Rheumatoid Arthritis: A Population-Based Cohort Study.

Increased risk of depression in patients with rheumatoid arthritis: a seven-year population-based cohort study.

The pathogenic potential of autoreactive antibodies in rheumatoid arthritis.

Infectious arthritis in patients with rheumatoid arthritis.

Drug treatment of rheumatoid arthritis. A systematic approach.