Anti-therapeutic antibodies and their clinical impact in patients treated with the TNF antagonist adalimumab.

Patients treated with the TNF antagonist adalimumab develop anti-therapeutic antibodies (ATA), the prevalence of which varies depending on the assay used. Most assays are compromised due to the presence of adalimumab in the clinical samples. Our objective was to develop an antibody assay, applicable for clinical testing, which overcomes the limitation of therapeutic interference and to further determine the relationship between ATA development, adalimumab levels and disease activity in patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA) or ankylosing spondylitis (AS). Use of an electrochemiluminescence platform permitted development of fit-for-purpose immunoassays. Serum samples from patients, taken prior to and at 12 and 24weeks of treatment, were retrospectively analysed for levels of adalimumab and ATA. Overall, the antibody prevalence was 43.6% at 12weeks and 41% at 24weeks of treatment. Disruption of immune complexes by acid dissociation, a strategy often adopted for this purpose, only marginally increased the antibody prevalence to 48.7% and 46% at 12 and 24weeks respectively. We found that antibody formation was associated with decreasing levels of circulating adalimumab, but no direct effect on disease activity was evident as assessed using DAS28 for RA patients and BASDAI for PsA and AS patients. However, a negative correlation of free adalimumab trough levels with disease activity scores was observed. Data showed that adalimumab levels can serve as an indicator of ATA development which can then be confirmed by ATA testing. Monitoring of both therapeutic and antibodies should be considered during adalimumab therapy to allow clinicians to personalise treatments for maximal therapeutic outcomes.

Introduction

Tumour necrosis factor alpha (TNF-α) antagonists, which include infliximab and adalimumab, are widely used for treatment of various chronic inflammatory or autoimmune diseases e.g. rheumatoid arthritis (RA), Crohn’s disease, ankylosing spondylitis (AS) and psoriatic arthritis (PsA). However, some patients develop anti-therapeutic antibodies (ATAs) which alter the pharmacokinetics and in some instances neutralize the biological effects of these therapeutics, impacting on clinical outcome. Approximately 10–30% of patients fail to respond to anti-TNF-α therapy and up to 60% of patients who responded initially fail to respond to treatment over time and require either dose-escalation or a switch to an alternative therapeutic to maintain a clinical response [1], [2]. The presence of ATA is thought to be responsible, at least in some patients, for the loss of clinical response.

Some studies have shown that adalimumab-treated patients develop ATA that are associated with lower serum adalimumab trough levels and loss of clinical response [3], [4], [5], [6], depending on the magnitude of the immune response [7]. The reported incidence of anti-adalimumab antibodies varies considerably among studies, from less than 5% to over 80% of patients developing ATAs, sometimes transiently [5], [8], [9], [10]. Such variation can be explained not only by differences in the population studied e.g., disease, therapeutic regimen, concomitant treatment with immunosuppressants and follow-up period but also by the heterogeneity in methodology employed for ATA assessment [8]. Antibodies to adalimumab have mainly been detected using radioimmunoassays [9] or pH-shift-anti-idiotype antigen binding test [7], [10] but the limitation is that these require use of radioisotopes. Bloem et al. [11] compared antibody assays for evaluation of immunogenicity in adalimumab-treated RA patients and concluded that different assays correlated well quantitatively but differed in their discriminatory potential and ability to identify as positive those samples containing low amounts of ATA. This is not unexpected since accurate detection and quantitative measurement of ATA is fraught with technical problems associated not only with differential assay sensitivity for low and high affinity ATA but also with therapeutic and/or target interference. Mitigation of interference is crucial for a thorough assessment of ATA and is particularly relevant for monoclonal antibody therapeutics which persist in the circulation. Formation of immune complexes between circulating therapeutic and ATA compromises ATA detection and strategies for circumventing therapeutic interference are required. An acid dissociation step is often implemented in the immunoassay [10], [12], [13] but this has limitations and cannot be applied universally to eliminate therapeutic interference as acid can degrade the ATAs and/or therapeutic and, in some instances, provide false positive results due to increased target interference [14].

The application of electrochemiluminescence (ECL) technology for ATA testing has gained prominence in recent years [14], [15], based on its increased sensitivity, large dynamic range and greater therapeutic tolerance. We therefore evaluated the utility of ECL-based assays and compared ATA results obtained with or without inclusion of acid treatment, in samples from adalimumab-treated patients in three disease groups - RA, AS and PsA. ATA specificity was confirmed by competitive inhibition and the neutralizing potential of the ATAs assessed in a reporter gene assay [16]. As the half-life of adalimumab is 15–19 days [17], the levels of residual therapeutic were also determined in an ECL assay. Finally, we tried to ascertain if the presence of ATA has any effect on trough levels of adalimumab and on disease progression.

Materials and methods

Patients

Sera from 3 distinct patient cohorts, RA (n = 18), PsA (n = 9) and AS (n = 12), were obtained from the Rheumatology outpatients clinic at the Sapienza University of Rome (Italy). Demographic data are shown in Table 1. Patients were treated with 40 mg adalimumab every other week as monotherapy or in combination with DMARDs therapy. For all patients, sequential samples were collected prior to the next adalimumab injection and retrospectively assessed (baseline - T0, after 12 weeks of therapy - T3 and after 24 weeks of therapy - T6). Therapeutic-naïve patient sera (n = 10) with high titers of rheumatoid factors (RF) and pooled or individual normal healthy sera (nhs, n = 17) were also included in the study. Ethical approval and informed consent was obtained in accordance with the guidelines in the Helsinki Declaration.

Table 1

Demographic data of study patients.

Disease cohort	RA	SpA
Number of patients	18	21
Rheumatoid factor positive	13	N/A
Female	15	11
Male	3	10
Age, yrs. (mean ± standard deviation)	50.3 ± 10.7	47.5 ± 11.9
Disease duration, yrs. (mean ± standard deviation)	10.5 ± 8.9	8.3 ± 6.9
Concomitant therapy n (%)	18 (100)	10 (47.6)
Glucocorticoids n (%)	16 (88.9)	2 (9.5)
Methotrexate n (%)	13 (72.2)	3 (14.3)
Leflunomide n (%)	4 (22.2)	0
Sulphasalazine n (%)	1 (5.6)	8 (63.4)

Disease activity was assessed at baseline and after 12 and 24 weeks of therapy by Disease Activity Score 28 (DAS28) for RA patients and Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) for PsA and AS patients. For clinical assessment, PsA and AS patients were grouped together as SpA since all PsA patients showed axial involvement. For RA patients, low disease activity is DAS28 >2.6 and <3.2 and remission is DAS28 <2.6. For SpA patients, the minimal clinically meaningful BASDAI reduction is 2 and remission is BASDAI <4.

Reagents

Commercially available adalimumab (AbbVie Inc, Illinois, USA), recombinant TNF-α (Xiamen Amoytop Biotech Co., China) and the 3rd WHO International Standard (IS) for TNF-α (coded 12/154) were used. An affinity purified hyperimmune sheep polyclonal, specific for the (Fab′)2 portion of adalimumab, generated at NIBSC served as positive control (PC) in the antibody assays.

Protein labelling

Adalimumab and TNF-α were labelled with EZ-link Sulfo-NHS-LC-Biotin (Thermo Scientific, MA, USA) and with ruthenium-NHS-ester (sulfo-tag NHS Ester, MesoScale Discovery (MSD), Gaithersburg, USA) as per manufacturers’ instructions.

Detection of anti-therapeutic antibodies (ATA)

Dilution series of controls and test sera were incubated with biotinylated adalimumab and ruthenium-conjugated adalimumab (both at 0.5 µg/ml in PBS-0.5% BSA) overnight at room temperature (RT) in polypropylene plates, the mixtures (25 µl per well) transferred to pre-blocked MSD streptavidin-coated plates and incubated for a further 2 h. The plates were washed twice with PBS-0.05% Tween and after addition of read buffer T, the plates were read using a SectorImager 2400 (MSD).

For acid dissociation, test sera were diluted 1:10 with acetic acid 300 mM in polypropylene plates, mixed at RT for 30 min, neutralized with Tris 1 M pH 9.5 and incubated for 2 h at RT with biotinylated adalimumab and ruthenium-conjugated adalimumab, both at 1 µg/ml. The mixtures (50 µl per well) were transferred to pre-blocked MSD streptavidin-coated plates and treated as above.

For assessing therapeutic tolerance, the positive control antibody was pre-incubated with adalimumab and evaluated in the antibody assay. ATA specificity was confirmed by competitive binding in which test sera were incubated with either PBS-0.5% BSA or excess adalimumab or etanercept (50 μg/ml) for 2 h at RT prior to evaluation in the antibody assay.

ATA positivity was defined as the response equal to or higher than the mean + 1.645 standard deviation of the plate negative controls.

Reporter gene neutralization assays

The KJL-Luc cell line (Prof. M. Tovey, ENS Cachan, France), a variant of the human erythroleukemic K562 cell line transfected with a NF-κB regulated reporter gene (firefly luciferase) construct together with a Renilla luciferase gene under the control of a constitutive thymidine kinase promoter was used [16]. The assay was optimized with our affinity purified sheep anti-adalimumab – (Fab′)2 specific - positive control. Since no difference between diluents (assay medium vs 5 or 10% nhs) was observed, assay medium was routinely used for evaluation. PC antibody and test samples were mixed with adalimumab (final concentration 10 ng/ml) for 30 min at 37 °C. The 3rd WHO IS for TNF-α (coded 12/154) was added to the appropriate wells at a final concentration of 1.0 ng/ml and incubated for 30 min at 37 °C, before addition of KJL-Luc cells (2.5 × 104 cells per well) for a further 4 h incubation at 37 °C. A TNF-α dose response curve ranging from 2.0 ng/ml to 0.015 ng/ml was included on each plate. Steadylite reagent was added following manufacturer’s instruction (PerkinElmer, UK) and the plates read in a luminometer (Microbeta2®, PerkinElmer, UK).

Detection of adalimumab

Dilution series of controls and test sera were incubated with biotinylated TNF-α and ruthenium-conjugated TNF-α (both at 0.025 µg/ml in PBS-0.5% BSA) overnight at RT, the mixtures (25 µl per well) transferred to pre-blocked MSD streptavidin-coated plates and further incubated for 2 h at RT. The plates were washed and following addition of read buffer T read in a Sector Imager 2400 (MSD).

Statistical analysis

Statistical analysis was performed using the CombiStats software (European Directorate for the Quality of Medicines and HealthCare, EDQM) based on the titers relative to the positive antibody control. The inter-assay and intra-assay variability was assessed in a 3 plate assay repeated on 3 days by the same operator. The intra-assay variability was assessed by calculating the geometric coefficient of variation (GCV) and expressed as a percentage of the assay mean (%GCV) for each analysed sample. For inter-assay variation, the calculated GCV was expressed as a percentage of the overall mean potency per sample over the 3 assays (%GCV). Clinical data expressed as mean ± standard deviation were analysed using Mann-Whitney and Spearman tests.

Results

Adalimumab measurement

Free residual adalimumab in the clinical samples was measured with an ECL bridging assay that was specifically developed for this purpose. Using both biotin- and ruthenium-conjugated TNF-α at a concentration of 0.025 μg/ml produced an optimal dose-response curve in the assay. The assay was sensitive, with a limit of detection (LoD) of 4 ng/ml and not susceptible to matrix effects as determined by diluting adalimumab in either PBS-0.5%BSA or in pooled normal human sera. Fig. 1a shows representative binding data for a panel of samples (negative or positive for adalimumab).

Fig. 1

Evaluation by ECL assays of adalimumab and ATA in samples from adalimumab-treated AS patients collected at baseline (T0), 12 weeks (T3) and 24 weeks (T6) of therapy are shown (a–c) along with recovery of the positive control antibody specific to adalimumab following an acid dissociation step (d). Representative results showing (a) adalimumab in samples diluted 1:100 (white bars) or 1:1000 (grey bars), the dotted line representing the cut-off point of the particular assay; (b) anti-adalimumab antibodies in samples with (white bars, cut-off shown as plain line) and without acid treatment (grey bars, cut-off shown as dotted line); (c) ATA specificity by pre-incubating samples with either PBS-0.5%BSA (black bars) or in the presence of etanercept (grey bars) or adalimumab (white bars) at 50 μg/ml; (d) positive control antibody incubated with adalimumab (0–125 μg/ml) prior to treatment with acetic acid (plain lines) or PBS (dotted lines).

As expected, none of the normal human sera (individual or pooled, n = 17) contained any adalimumab. However, adalimumab levels varied considerably among patients, ranging from 0 to 30 μg/ml. The median levels and range of adalimumab at 12 and 24 weeks are shown in Table 2. We found that median adalimumab levels were significantly lower at both time points in samples where ATA was detected compared with samples negative for ATA, for example for RA patients at T3 the median was 1.35 μg/ml in presence of ATA and significantly higher at about 8 μg/ml in absence of ATA.

Table 2

Free adalimumab concentration (μg/ml) expressed as median and (range) in samples from adalimumab-treated patients.

Disease	T3		T6
	ATA negative	ATA positive	ATA negative	ATA positive
RAn = 18	8 (4–16)	1.35 (0–4)	8 (2.6–16)	3.65 (0–7)
PsAn = 9	6.6 (5.2–30)	0.6 (0–7)	16.9 (6.8–27)	0.75 (0–10)
ASn = 12	5 (3.5–10)	3.75 (2.2–4.2)	6.2 (1.7–10)	4.8 (2.2–4.8)

Statistical evaluation of the assay showed good reproducibility. The intra-assay %GCV varied between 3.9 and 21.2% and %GCV for inter-assay variation was between 3.5 and 5.1%.

The inclusion of an acid dissociation step in the procedure did not significantly change the detected level of adalimumab for a majority of the samples that were adalimumab and ATA positive. However, adalimumab at low levels was noted for all T3 and/or T6 samples that were previously negative for adalimumab suggesting formation of immune complexes between therapeutic and ATA (data not shown).

Anti-therapeutic antibodies measurement

Anti-adalimumab antibodies were assessed in a bridging assay using the ECL platform. Biotinylated- and ruthenium-conjugated adalimumab at 0.5 µg/ml provided an optimal dose-response curve of the PC antibody. This combination delivered the highest signal to noise ratio and showed no matrix effects when negative controls (normal human sera) were tested or when the PC antibody was diluted in either PBS-0.5%BSA or in normal human sera. The LoD of the assay was 1 ng/ml. Recovery assessed by spiking clinical (n = 12) or healthy sera (n = 4) with PC at either 125 or 31.25 ng/ml was consistent with the nature of the samples and between the accepted criteria of 80–120% for the adalimumab and ATA negative samples. Statistical evaluation of the antibody assay showed very good reproducibility and low variability. The intra-assay variation expressed as %GCV varied between 1.5 and 6.7% and the inter-assay %GCV was between 6.2 and 10.3%.

As circulating adalimumab was present in most samples, at levels up to 30 μg/ml, an acid dissociation step was incorporated to mitigate the therapeutic interference. This required re-optimization of the assay by testing several concentrations of biotinylated- and ruthenium-conjugated adalimumab. Labelled adalimumab at 1 μg/ml was selected for clinical evaluation. The impact of acid dissociation was assessed by spiking the PC with adalimumab (0–125 μg/ml) and testing the mixture after treatment with or without acid (Fig. 1d). As little as 1 μg/ml adalimumab interfered with ATA detection, resulting in an increase in LoD from 1 ng/ml to 50–100 ng/ml. Following acid dissociation, the same level of ATA (50–100 ng/ml) is detected despite addition of 125 μg/ml of adalimumab. As in the case of the therapeutic assessment, dilution series were performed for each sample to generate a binding profile. Representative binding data for samples (negative or positive for ATA), tested in at least 3 independent runs, are shown in Fig. 1b.

Pre-existing antibodies were detected in 1 PsA and 2 RA patients, but not in normal sera. The ATA prevalence at T3, without or with acid treatment in the ECL immunoassay was 33.3% (6/18) vs 38.9% (7/18) for RA, 44.4% (4/9) vs 55.5% (5/9) for PsA and 58.3% vs 58.3% (7/12) for AS patients respectively. At T6, the prevalence of ATA was 22.2% (4/18) vs 27.8% (5/18) in RA, 66.7% vs 66.7% (6/9) in PsA and 50% (6/12) vs 58.3% (7/12) in AS patients. Data on ATA levels and prevalence following acid treatment are summarised in Table 3.

Table 3

ATA prevalence and range in samples from adalimumab-treated patients.

Disease		T0	T3	T6
RAn = 18	ATA prevalence - n (%)	2 (11.1)	7 (38.9)	5 (27.8)
	ATA range, ng/ml	12–75	<10–9000	<10–1000
PsAn = 9	ATA prevalence - n (%)	1 (11.1)	5 (55.5)	6 (66.7)
	ATA range, ng/ml	30	<10–150	<10–900
ASn = 12	ATA prevalence - n (%)	0 (0)	7 (58.3)	7 (58.3)
	ATA range, ng/ml	–	<10–2500	<10–1500

The specificity of the detected antibodies was evaluated by incubating clinical sera with an excess of adalimumab or etanercept (50 μg/ml) prior to testing in the ECL bridging assay. The inhibition seen predominantly with adalimumab but not etanercept showed that the ATAs detected were specific for adalimumab in 97% of the ATA positive samples; the inhibition of the ECL response in presence of excess adalimumab ranged from 62% to 100%. Representative competitive inhibition results are shown in Fig. 1c.

Rheumatoid factors (RF) in RA samples may provide false positive results in the antibody assays as they can bridge the biotin- and the ruthenium-conjugated adalimumab by interacting with the Fc portion of the labelled therapeutic. To estimate the extent of RF interference in the assay, therapeutic naïve RA patients’ sera with a high titer of RF (RF+, n = 10) were tested in parallel with normal human sera (n = 10). We observed no difference of signals between RF+ and normal sera, even after introduction of acid dissociation. Assays were also performed with the PC antibody diluted in either RF+ or normal human sera. A 2 sample t-test performed after log transformation of the data showed that there was no significant difference, at all tested dilutions, between signals obtained among the two types of matrices. Moreover, there was no correlation between the presence of RF in RA patients and positivity in the antibody assay. With AS and PsA samples (RF−, ATA+), we observed in some cases, a minimal (<15%) variation in ECL response when samples were diluted in RF+ instead of normal sera or PBS-0.5%BSA. We therefore inferred that the developed ECL assay is not affected by RF in contrast to the interference noted in the ELISA assay [18] but there remains a possibility of a slight overestimation or underestimation in ATA level.

A reporter gene assay designed to evaluate the neutralizing potential of antibodies against TNF antagonists [16] was optimized using our reagents. Spiking the PC antibody with adalimumab, to simulate the clinical samples setup, resulted in therapeutic interference as evident from a decrease of the signal below the level of the assay control (cells incubated with fixed amounts of TNF-α and adalimumab). (Fig. 2, panels d–f). Since without physical extraction of all circulating adalimumab the effect of residual therapeutic present in clinical samples cannot be avoided in the bioassay, only samples with no detectable adalimumab could meaningfully be assessed for neutralizing antibodies. The results showed that for the ATA-positive, therapeutic-negative patients, the ATAs detected by the ECL platform are neutralizing (Fig. 2, panels a–c).

Fig. 2

Detection of neutralizing antibodies to adalimumab in patient samples at baseline (T0 – closed circles), 12 weeks (T3 – open square) and 24 weeks of treatment - T6 (open triangle) using a reporter gene assay. Panels a to c: ATA positive patient sera with no detectable therapeutic. Assay controls yielded luminescence counts of 70 K (cells incubated with TNF-α alone) and 32 K (cells with TNF-α + adalimumab). The assay confirmed the presence of neutralizing antibodies at T3 and T6 in the assessed patient samples (panels a–c) as well as the presence of pre-existing neutralizing antibodies at T0 for patient PsA-2 (panel b). Panels d to e: ATA positive (panels e and f) or negative (panel d) patient sera with detectable therapeutic. Assay controls yielded a result of 60 K (cells incubated with TNF-α alone) and 20 K (cells with TNF-α + adalimumab). The results indicated that the presence of the therapeutic in the samples interferes with the neutralizing assay as shown by the signals below the control level of 20 K.

Association between adalimumab, ATA and disease activity

Despite the small number of patients in each cohort, we observed a negative correlation between levels of ATA and adalimumab at T3 and T6 (Fig. 3). For example, in ATA negative RA patients, the adalimumab levels varied from 4 to 25 μg/ml. In ATA positive RA patients with ATA levels lower than 10 ng/ml, the levels of adalimumab were estimated between 1.2 and 8 μg/ml; this declined to 0.3–2 μg/ml when ATA levels ranged from 20 to 90 ng/ml. At ATA levels above 300 ng/ml, adalimumab was not detectable.

Fig. 3

Correlation between observed level of ATA (X axis, ng/ml, not detected = 0.001 ng/ml) and observed level of adalimumab (Y axis, μg/ml) in RA (top panel), AS (middle panel) and PsA (bottom panel) patients.

For the RA cohort, the disease activity, as measured by DAS28, was a mean of 5.2 ± 1.3 at baseline which significantly decreased after 12 and 24 weeks of treatment to 3.9 ± 1.4 and 3.7 ± 1.4 (p = 0.009 and p = 0.003 vs baseline) respectively. No significant difference was seen in DAS28 between ATA positive and ATA negative patients at 12 or 24 weeks.

For the SpA patients group, the mean BASDAI at baseline was 57.5 ± 25.4. We observed a significant reduction in disease activity after 12 and 24 weeks of treatment to 42 ± 24 and 37.3 ± 21.8 (p = 0.08 and p = 0.03 vs baseline) respectively. No significant differences were seen between BASDAI at 12 and 24 weeks in ATA positive compared to ATA negative patients.

Interestingly, there is a negative correlation of adalimumab trough levels with DAS28 and BASDAI values (p < 0.0001, r = −0.53 and 0.032, r = −0.27 respectively) as higher levels of adalimumab are associated with lower disease activity scores (Fig. 4).

Fig. 4

Correlation between serum level of adalimumumab (μg/ml) and disease activity scores for RA (Panel a) and SpA (Panel b) patients.

Discussion

In this study, none of the AS patients had pre-existing antibodies against adalimumab, however these were detected in 2/19 RA patients (10.5%) and 1/9 PsA patient (11.1%). Further evaluation showed that these antibodies were: (1) specific, non-neutralizing in a RA patient but disappeared at 12 and 24 weeks; (2) non-specific, non-neutralizing, at low levels in a RA patient, which became specific at 12 weeks; (3) specific, neutralizing in a PsA patient, which increased in magnitude at subsequent time-points. This incidence is consistent with a review of previous studies showing that 5.6% of subjects had pre-existing antibodies. In particular, studies with RA patients concluded that 14.8% of patients had pre-existing antibodies against the therapeutic being administered [19].

Following therapy, 43.6% and 41% of the adalimumab-treated patients overall became antibody positive at 12 and 24 weeks respectively. This increased to 48.7% and 46% when samples were subjected to acid treatment. In our assay, immune complex disruption by acid treatment confirmed ATA in borderline ATA positive samples (n = 3) but since the assay was sensitive even without acid treatment, only five samples originally ATA negative became ATA positive after acid treatment. Therefore, acid dissociation only marginally increased ATA detection although the assay tolerance was enhanced 25 fold and supported approximately a 500 molar excess of adalimumab after incorporation of acid dissociation. Xenogeneic antibody positive controls are often used to define the assay tolerance but the utility of these controls as representatives of ATAs induced in patients is questioned. Regardless of how suitable a positive control is likely to be, it cannot truly reflect the clinical situation as ATAs by their nature are polyclonal and heterogeneous between patients and also within the same patient over the treatment course.

We observed a lower immunogenicity incidence in RA patients (44.4%) compared with SpA patients (71.4%) that could, partially at least, be explained by the fact that all RA patients received concomitant therapy as opposed to only half of SpA patients (10/21). Treatment with methotrexate and/or glucocorticoids concomitantly has been shown to reduce immunogenicity in a dose-dependent manner [6], [20], [21].

The ATA incidence seen here is line with published data as the reported prevalence of anti-adalimumab antibodies varies considerably from less than 5% to over 80% of patients developing ATA, sometimes transiently [8], [22]. Such variation is largely due to differential assay sensitivity and interference from therapeutic or matrix. Most immunogenicity studies have been performed using assays that have poor therapeutic tolerance so ATA detection is often compromised. A pH-shift-anti-idiotype antigen binding test, based on a radioimmunoassay (RIA) which included acid treatment and addition of xenogeneic anti-idiotype for adalimumab removal showed that 70% of RA patients (21/30) developed ATAs within 28 weeks of treatment [10]. Most clinical studies have used predominantly RIA-based methods which although sensitive are not preferred due to radioisotope handling. The ECL platform obviates the use of radioisotope and is claimed to be sensitive, tolerant to therapeutic and minimally impacted by matrix [23]. A recent study comparing ECL and three RIA methods concluded that although the capability of all assays was very similar, they varied in their ability to detect samples with low levels of ATA [11]. The ECL assay used in this study is very sensitive but it is difficult to compare the sensitivity and tolerance of our assay to the published assays [9], [10], [11], as there are no available standardised positive antibody controls to facilitate any comparisons across different studies despite a recent attempt towards this goal [24].

We found that for the antibody-positive, therapeutic-negative patients, the ATAs detected are neutralizing. While neutralization could not be confirmed in all ATA positive samples due to adalimumab interference, a previous study has shown that the humoral response is restricted to the idiotype region of adalimumab and all ATA against adalimumab neutralize adalimumab function [25].

An association was seen between the magnitude of ATA response at 12 and/or 24 weeks of treatment and levels of circulating adalimumab for all patients. Higher ATA levels were associated with lower levels of adalimumab in the samples.

Our data concurs with previous studies in which patients developing ATA during adalimumab treatment show reduced adalimumab levels in comparison with ATA negative patients [7], [26]. There is no agreement as yet on the minimal trough levels of adalimumab required for maintaining patients at low disease level or in remission.

However, in all groups, we observed a negative correlation between trough adalimumab levels and disease activity scores DAS28 and BASDAI at 12 and/or 24 weeks of treatment. No significant difference was found between ATA positive and ATA negative patients when disease activity was evaluated at 12 and 24 weeks of treatment or after an extended period of 12 months. Moreover, ATA positive and ATA negative patients did not significantly differ in terms of EULAR or ASDAS response criteria which classify individual patients in terms of responsiveness to treatment. A possible explanation is that, in our study, the majority of antibody positive samples contain low levels of ATA. While the clinical relevance of such low levels is currently not clear, it is plausible that on further treatment and a longer follow-up including immunogenicity monitoring, a clear picture of the emergence of ATA and its impact on clinical outcome becomes evident [6], [26]. Bartelds et al. showed that in two-thirds of RA patients developing adalimumab antibodies, antibody formation occurs within 28 weeks of treatment but an extended follow-up over 3 years allowed identification of an increasing number of ATA positive patients with impact on disease progression. Association between high ATA levels with clinical non-response was evident [6].

In this study, based on the limited number of samples evaluated, it is clear that ATAs in patients diminish levels of free therapeutic and so if ATA testing is challenging for clinical laboratories engaged in routine follow-up of patients, assessment of levels of adalimumab and their impact on disease activity can be considered as an indirect approach towards monitoring of ATA formation.

Conclusion

Interference from therapeutic is an important confounding factor and needs to be mitigated for proper ATA detection. In the ECL antibody assay described here, acid treatment did not significantly increase the ATA incidence. However, we found that measuring adalimumab in the same samples as used for ATA testing is beneficial as the level of therapeutic is an important parameter governing clinical response and appears to be indicative of ATA development which can then be confirmed during follow-up. Routine monitoring of therapeutic and ATA should be considered during adalimumab therapy so as to allow clinicians to personalise treatments with a view to maximizing individual therapeutic outcomes.

Competing interests

The authors declare that they do not have any financial or non-financial competing interests.

Funding

The work reported in this study is independent research which, in part, was commissioned and funded by the UK Department of Health Policy Research Programme (NIBSC Regulatory Science Research Unit, 044/0069). The views expressed in this publication are those of the authors and not necessarily those of the Department of Health.