Materials and methods

Patient inclusion and exclusion criteria

Patients with RA, fulfilling the 1987 American College of Rheumatology criteria for RA and about to start treatment with infliximab were enrolled consecutively in this open, prospective observational study. All patients were cared for at the department of rheumatology at the Radboud University Nijmegen Medical Centre and Sint Maartenskliniek, Nijmegen, and all gave informed consent before inclusion; the local ethics committee approved this study. Patients were excluded from this study if they used disease-modifying antirheumatic drugs other than methotrexate (MTX), had another coinciding autoimmune disease, used high doses of glucocorticoids (>10 mg daily) or had severe infections. Treatment of the patients was in accordance with the consensus statement on the initiation and continuation of TNFα blocking therapy in RA.19

Before starting infliximab/adalimumab treatment all patients had evidence of active disease as indicated by a 28-joint count Disease Activity Score (DAS28,20) of >3.2 taking into account that monotherapy with at least two disease-modifying antirheumatic drugs, including MTX, 25 mg weekly, had failed. All patients were given intravenous infusions of 3 mg/kg infliximab at baseline, 2, 6 and 14 weeks. At week 14 the treating rheumatologist was allowed to increase the dosage of infliximab to 5 mg/kg in patients for whom treatment had produced an inadequate response. Thereafter, dosing intervals were kept stable at 8-week intervals. The patients with RA receiving adalimumab originated either from a clinical trial (Abbott DE018, n = 20) or our outpatient clinics (n = 14). All patients started with a dosing regimen of 40 mg biweekly subcutaneously (SC). Only for those patients not included in the clinical trial could the dosage be increased to 40 mg SC weekly when response to treatment was judged inadequate by the rheumatologist. During the study, concomitant medication was monitored, and non-steroidal anti-inflammatory drugs and MTX were continued at constant dosage. Fourteen of 20 adalimumab-treated patients with RA in the clinical trial were receiving adalimumab monotherapy.

Functional serum levels of infliximab/adalimumab

One hour before each infusion or SC injection, 10 ml serum was collected and stored at −80°C. All sera were tested under blinded conditions at BioMonitor A/S, Copenhagen, Denmark (http://www.biomonitor.dk, accessed 20 June 2009). Functional infliximab/adalimumab levels, measured as TNFα binding capacity of serum IgG, were obtained by a standardised procedure described previously.13 16 Briefly, 1% patient serum was incubated with ¹²⁵I-TNFα (Perkin Elmer, Boston, Massachusetts, USA), and after separation of the tracer using rabbit anti-human Fcγ antibody (Dako, Glostrup, Denmark), the pellet activity was measured using a Wallac1470 automatic gamma counter. S-infliximab/adalimumab were assessed as the levels of infliximab (Schering-Plough, Farum, Denmark) and adalimumab (Abbott, Gentofte, Denmark), respectively, which in a pool of normal serum bound equivalent amounts of radiolabelled TNFα; the inter- and intra-assay variations were <20 and <10%, respectively.

Antibodies to infliximab/adalimumab

The fact that infliximab and adalimumab are IgG constructs consisting only of κ light chains makes it possible to use anti-human λ light-chain antibody to distinguish between free infliximab/adalimumab and infliximab/adalimumab in complex with any class of λ-containing human immunoglobulin.13 16 The assays were carried out in a blinded fashion at BioMonitor ApS using 1% serum with an added 4000 cpm/100 μl of ¹²⁵I-infliximab and ¹²⁵I-adalimumab, respectively. After incubation, free and immunoglobulin-bound (any isotype) tracer were separated by affinity chromatography using small columns of matrix-bound anti-human immunoglobulin λ-chain antibody (Dako) (fig 1). The data are given as percentage bound cpm/total cpm added; the backgrounds were 3%, and the inter- and intra-assay variations were <20 and <10%, respectively. Interference by rheumatoid factor was tested by cross titrations with five known RF-positive sera without anti-infliximab or anti-adalimumab antibodies. None of these influenced the antibody measurements.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Fluid-phase radioimmunoassay measuring antibodies (Abs) to infliximab (all isotypes).13 16 Serum containing anti-infliximab Abs is incubated overnight with purified ¹²⁵I-infliximab. Free and immunoglobulin-bound tracer are separated by affinity chromatography using Sepharose-bound anti-human λ-chain Ab. Adapted from Bendtzen K. Immunogenicity of anti-TNF antibodies. In: van de Weert M, Møller EH, eds. Biotechnology: pharmaceutical aspects. Vol. VIII. Immunogenicity of biopharmaceuticals. Springer; 2008.

Statistical analysis

Descriptive statistics were provided using mean (SD). The disease activity between groups (independent samples) and between study start and study end (paired samples) was compared using a Student t test or paired t test, respectively. Non-normally distributed data were analysed using the Mann–Whitney U test (independent samples) and the Wilcoxon signed-ranks test (paired samples). Associations between δ DAS and S-drug levels and/or presence/absence of antibodies were quantified by Pearson’s correlation, and the corresponding p values were calculated.

Results

Characteristics of patient groups

A total of 69 patients entered the study, of whom 35 were treated with infliximab and 34 with adalimumab. At baseline no differences were apparent between the patient groups (table 1).

The majority of patients were women with mean age of disease onset of 57 years. About 71% were rheumatoid factor positive, which is similar to other recently published reports on TNFα-neutralising therapies.11 12 14 Infliximab in all cases was given at a dose of 3 mg/kg at baseline and after 2, 6 and 14 weeks. After this initiation regimen, infliximab was administered at 8-week intervals, which was continued throughout the study in all but one case that was increased to 5 mg/kg because of moderate response. The 8-week interval was shortened to 6 weeks in six cases: one of them was a moderate responder at 12 weeks (responder at 6 months), two were moderate responders at 3 and 6 months and three were non-responders throughout the whole study. Ten of 35 patients receiving infliximab used low-dose prednisolone varying between 5 mg and 10 mg daily. Although, there was a trend for more use of prednisolone in the non-responders (35%) and moderate responders (33%) compared with responders (20%), the mean levels of prednisolone per patient were comparable (data not shown).

Although all patients with RA who started treatment with adalimumab began with 40 mg every other week, only the patients not included in the clinical trial were allowed to increase the dose. From those 14 patients, five who were all moderate responders at 3 months, increased the dosage to 40 mg weekly upon which two of them evolved to good responders at 6 months. From the 20 patients included in the DE018 trial, six received MTX in a dose varying from 10 mg to 30 mg once weekly, the dose of which was kept stable during the trial. Nine out of 31 patients receiving adalimumab used low-dose prednisolone varying between 5 mg and 10 mg daily. A comparable trend for use of prednisolone was seen as in the patients receiving infliximab, being higher in the non-responders (50%) and moderate responders (43%) than in the responders (13%). Again, the mean levels of prednisolone between the groups were comparable (data not shown).

Clinical response and association with functional drug levels and antibodies against infliximab and adalimumab

Infliximab or adalimumab as well as antibodies against infliximab or adalimumab were undetectable in all patients at the start of treatment.

At 6 months, 43% of the infliximab-treated patients fulfilled the EULAR criteria for good responders, whereas 17% and 40% of the patients were moderate or non-responders (table 2).

Figure 2A and B show that all patients having a good response receiving infliximab had detectable drug levels after 3 (R = 0.54, p = 0.03) and 6 months. In the patients who were classified as moderate responders, only three patients had detectable levels of infliximab at 6 months, two of whom turned out to be good responders at 12 months (figs 2C and D) having a DAS28 of 3.2 and 3.1, respectively. In contrast, none of the patients who were classified as non-responders had detectable infliximab levels at 6 months, strongly suggesting that low trough levels of infliximab at 3 and 6 months predict treatment failure (figs 2E and F). This phenomenon is underlined by two patients who were responders at months 3, having high levels of S-infliximab but became non-responder at 6 months where infliximab levels were not detectable any more (figs 2E and F), all of which coincided with the formation of antibodies against infliximab at 3 months.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Association between clinical response and trough level of infliximab. Course of disease activity (A, C and E) at baseline, after 3 and 6 months from patients after initiation of infliximab and correlation with trough levels of infliximab (B, D and F). The patients indicated in grey (C and D) represent two patients who were moderate responders at 6 months despite having high levels of S-infliximab. These patients were responders after 12 months (data not shown). The two patients indicated in grey (E and F) indicate patients who were moderate responders at 3 months despite high levels of S-infliximab; however, these patients developed antibodies and then became non-responders.

The formation of antibodies to infliximab was strongly associated with both response (R = 0.63, p = 0.008) to treatment and levels of soluble infliximab (R = 0.74, p = 0.01). For example, 14/15 (93%) patients who were good responders at 6 months had no detectable levels of anti-infliximab antibodies at that time, whereas 50% of the moderate responders had low levels (figs 3A and B). In contrast, all patients who were classified as non-responders at 6 months had detectable antibody formation either at low concentrations (7%), at moderate (36%) or high (57%) levels. With regard to the association between soluble levels of infliximab and anti-infliximab antibodies, none of the patients with detectable infliximab levels at any time point had circulating antibodies to the drug (fig 4). In contrast, in all but three patients with antibodies (regardless of level) soluble levels of infliximab were undetectable. In a total of six patients with RA classified as non-responders at 3 months, soluble infliximab levels were still detectable even though low levels of antibodies were measured. In all these cases, the levels of trough infliximab dropped below the level of detection and the anti-infliximab antibody levels rose to medium or high, suggesting that the absence of detectable levels of infliximab is accompanied by the formation of antibodies to the drug. The mean (SD) weekly dosage of MTX was 20.3 (3.5) mg, 20.8 (3.8) mg and 20.0 (3.4) mg for the responders, moderate responders and non-responders, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Association between clinical response and trough levels of adalimumab. Relationship between response to tumour necrosis factor α neutralising therapy in patients using adalimumab and trough levels of adalimumab.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Occurrence of antibodies against infliximab and adalimumab in patients with rheumatoid arthritis (RA). (A) The percentage (as a whole group) of patients with RA who develop anti-infliximab antibodies at 3 and 6 months, respectively; (B) those patients developing antibodies against adalimumab. Frequency of antibodies against infliximab (A) and adalimumab (B) after 3 and 6 months of treatment and stratified for non-responder, moderate responder and good responders to treatment. The white bars represent the responders whilst the grey and black bars represent the moderate and non-responders, respectively.

Of the patients who started treatment with adalimumab, 17 (50%) patients fulfilled the criteria for good response, whereas 7 (21%) and 10 (29%) were classified as moderate and non-responders at 3 months, respectively (table 2). As observed in patients receiving infliximab, the levels of soluble adalimumab were also associated with response to treatment (R = 0.64, p = 0.01) (fig 3). For instance, all responders and moderate responders had detectable levels of soluble adalimumab, which were higher in responders than in moderate responders (table 2). In contrast, non-response was clearly associated with undetectable (n = 9) or very low serum levels (n = 1) of adalimumab. In line with that seen in patients treated with infliximab, antibody formation against adalimumab also correlated with response to treatment (fig 4B). None of the responders had antibodies to adalimumab at 3 and 6 months (and thus circulating levels of drug). In sharp contrast, all non-responders had detectable antibodies against adalimumab both at 3 and 6 months. Interestingly and in contrast with the moderate responders treated with infliximab, where 20% and 30% already had circulating antibodies at 3 and 6 months, respectively, none of the adalimumab patients had detectable antibody levels. Of the patients who had a response to treatment at 3 months, seven (41%) originated from the DE018 trial as did five (71%) who were considered as moderate responders and eight (80%) non-responders. From those patients 42%, 40% and 13% concomitantly used MTX in the responder, moderate and non-responders, respectively.

Discussion

Response failure due to generation of antibodies against biopharmaceutical agents is increasingly appreciated as one of the mechanisms of drug failure.21 Hitherto, dose- or dose frequency adjustments of anti-TNFα neutralising therapies owing to insufficient responses are a frequent necessity in patients with RA.22 23 For example, secondary response failure has been reported in patients receiving infliximab treatment.12 13 22 Furthermore, many patients with secondary response failure to one anti-TNFα drug benefit from a switch to other anti-TNFα drugs, suggesting a role of neutralising antibodies directed specifically against the protein used for treatment.24 25 26 Several studies have shown that secondary response failure is associated with low trough levels of anti-TNFα biological agents and development of antibodies against these drugs.11 12 13 14 27 Indeed, trough level assessments of anti-TNFα drugs and measurements of antibodies against these biological agents are among the measures that best predict the response of individual patients with RA.12 14 16 28 29 In the most recent report from Bartelds et al an anti-adalimumab antibody frequency of 34% in non-responders and 5% in good responders was found, an observation that closely resembles our results.11

Here we demonstrate that response to anti-TNFα antibodies, both infliximab and adalimumab, is closely related to the formation of antibodies to the constructs and, in turn, the trough levels of neutralising anti-anti-TNFα antibodies. Interestingly, patients treated with adalimumab demonstrated a 10% higher response rate than their counterparts treated with infliximab, which mirrored the percentage of patients who formed antibodies to the drug. The latter finding corroborate findings from two recent studies, which found comparable responses.30 31 32

An interesting observation from this study is that the moderate responders still have circulating levels of drug and do not have detectable levels of antibodies, which is especially remarkable in the patients treated with adalimumab. Several mechanisms may explain this phenomenon. At first, very small amounts of anti-TNF antibodies might be already present that cannot be detected by our assays. This is unlikely since the presence of very low concentrations of antibodies correlated well with measurable drug levels and clinical response in the non-responders. Finally, other mechanisms may cause the moderate response in these patients that are unconnected with response to the treatment itself. Perhaps, the patients with a very high baseline DAS28 represent a different RA phenotype that is simply not responsive or only moderately responsive to TNFα neutralising therapy.

We have shown that response to two anti-TNFα therapies closely follows the trough levels of the drugs and the presence of antibodies directed against the drugs. Additionally, we have provided a rationale for the recent observation that response to treatment with adalimumab is greater than that seen after treatment with infliximab. Further studies that focus on the underlying pathways leading to antibody formation are warranted since it would be helpful to predict response to anti-TNF therapies.