Evaluation of the Cross-reactivity of Antidrug Antibodies to CT-P13 and Infliximab Reference Product (Remicade): An Analysis Using Immunoassays Tagged with Both Agents.

BACKGROUND: During two pivotal clinical trials of the infliximab biosimilar CT-P13 (PLANETAS and PLANETRA), antidrug antibodies (ADAs) and neutralising antibodies (NAbs) were detected in the sera of patients treated with CT-P13 and the reference product (RP; Remicade).
OBJECTIVE: The aim was to assess the comparability of Remicade- and CT-P13-tagged immunoassays for the detection of ADAs and NAbs using data from these trials, in order to determine the cross-reactivity of CT-P13 and RP ADAs.
METHODS: Sera from patients with rheumatoid arthritis and ankylosing spondylitis were analysed using an electrochemiluminescence (ECL) bridging assay or Gyros immunoassay, tagged with Remicade or CT-P13 at screening, weeks 14, 30 and 54, and the end of study visit. NAb titre was compared at screening and weeks 14 and 30. The proportion of cross-reactive samples was determined and an inter-rater agreement analysis performed to assess the concordance of results between assays.
RESULTS: In PLANETAS, 93.1% (94/101) of RP ADA-positive samples and 93.0% (93/100) of RP NAb-positive samples cross-reacted with CT-P13; 99.0% (103/104) of CT-P13 ADA-positive and 98.0% (98/100) of CT-P13 NAb-positive samples cross-reacted with the RP. In PLANETRA, 94.7% (426/450) of RP ADA-positive samples and 94.3% (415/440) of RP NAb-positive samples cross-reacted with CT-P13, and 96.6% (458/474) of CT-P13 ADA-positive and 96.4% (452/469) of CT-P13 NAb-positive samples cross-reacted with the RP. In both studies, there was strong agreement in outcome between assays at all post-screening time points (PLANETAS: Cohen's κ 0.89-0.98 for ADA, 0.86-0.98 for NAb; PLANETRA: 0.92-0.94 for both ADA and NAb, all p < 0.001). Significant concordance between assays was observed for NAb titre at weeks 14 and 30 (PLANETAS: Spearman's ρ 0.73 and 0.74, respectively; PLANETRA: 0.61 and 0.72, respectively; all p < 0.001).
CONCLUSIONS: This study has demonstrated that ADAs and NAbs against CT-P13 and RP are cross-reactive, indicating that CT-P13 and RP share immunodominant epitopes.

Key Points

Background

The introduction of biological agents that target the immune response has had a dramatic effect on the treatment of immune-mediated inflammatory diseases (IMIDs), such as rheumatoid arthritis (RA), spondyloarthritis (SpA) and inflammatory bowel disease (IBD). For example, it has been consistently shown that agents targeting tumour necrosis factor (TNF) are effective and superior to placebo in treating IMIDs [1-4]. However, some patients may experience a loss of treatment efficacy over time. One of the primary reasons for this phenomenon is the development of antidrug antibodies (ADAs), which have been noted for all TNF inhibitors to varying degrees [5]. ADAs that inhibit the activity of the drug by binding to or near the active site are referred to as neutralising antibodies (NAbs), and it is thought that the majority (>90%) of ADAs generated against anti-TNF antibodies (adalimumab, certolizumab, golimumab and infliximab) are of this type [6]. Non-neutralising ADAs are also thought to contribute to a lack or loss of response by forming immune complexes with the drug, thereby reducing bioavailability [7]. There are safety concerns surrounding the development of ADAs, as they have been associated with an increased incidence of infusion-related reactions, injection site reactions, and hypersensitivity [5, 8]. The concomitant use of immunosuppressants such as methotrexate or azathioprine/mercaptopurine may delay or prevent the formation of ADAs [5, 9].

The European Medicines Agency (EMA) and the US Food and Drug Administration (FDA) have issued guidance on the assessment of immunogenicity of biological agents [10, 11]. Both stress the importance of developing valid and sensitive assays that are optimised for the ADAs being measured. The approval of biosimilars for the treatment of IMIDs has further highlighted the importance of using appropriate immunoassays for ADA testing. Differences in the manufacturing processes for biosimilars mean that they are never identical to their originator biological agents, or reference products (RPs), and they therefore may have different immunogenicity profiles. Thus, a comparative immunogenicity assessment of a biosimilar and its RP is a necessary step in the biosimilar approval process [12, 13]. A one-or two-assay approach may be taken for this assessment. In a one-assay approach, an assay tagged with the biosimilar is used to detect ADAs against the biosimilar and RP. Although information on the immunogenicity of the RP may be lost, this method minimises the potential impact of assay bias and ensures that all antibodies developed against the biosimilar are detected [10]. In a two-assay approach, an assay tagged with the RP is used to detect RP ADAs and a separate assay tagged with the biosimilar is used to detect biosimilar ADAs. This provides a comprehensive analysis reflecting the true immunogenicity for both drugs, but requires extensive validation as well as more resources and time [14].

CT-P13 (Remsima®; Inflectra®) was the first biosimilar of infliximab RP (Remicade®) to be approved by the EMA and the FDA. In the pivotal randomised controlled trials that led to its approval (PLANETAS and PLANETRA), ADAs and NAbs were assessed using electrochemiluminescence (ECL) and automated Gyros™ immunoassays, respectively. Each assay used EU-approved Remicade as the ‘tag’ to capture any antibodies present in the sera of CT-P13- or RP-treated patients [15-18]. In addition, in the PLANETRA study, ADA and NAb detection assays with CT-P13 as the tag were used in parallel with the Remicade-tagged assays. Comparison of immunogenicity results up to week 30 of PLANETRA indicated analytical agreement between the two assays. This suggests that RP ADAs have the ability to recognise and bind to similar antigenic sites on CT-P13, i.e. antibodies against the RP are cross-reactive with the biosimilar, and vice versa [17]. ADAs were detected in a proportion of patients throughout these trials [16, 18]. A higher proportion of patients with RA developed ADAs compared with patients with ankylosing spondylitis (AS). However, these proportions were comparable between those treated with CT-P13 and RP in each patient population [16, 18]. Furthermore, in the majority of patients (80–90%) in whom ADAs developed, these antibodies persisted to the end of the entire study period [19, 20].

The aim of the current study was to fully evaluate the comparability of Remicade- and CT-P13-tagged immunoassays for the detection of ADAs and NAbs by analysing all available data from the pivotal clinical studies of CT-P13, and to thus assess the cross-reactivity of ADAs against both CT-P13 and its RP.

Methods

Immunoassay Validation

The immunoassays used in the PLANETAS and PLANETRA trials were fully validated in accordance with criteria defined in the regulatory guidance and industry recommendations [11, 21]. To demonstrate that CT-P13 and infliximab RP were comparable in the immunoassays, a cross-inhibition test was performed. Here, immunocompetition controls [high positive control (HPC), 1000 ng/mL; low positive control (LPC), 150 ng/mL; and pooled negative control (PNC)] were analysed with and without spikes of CT-P13 (50 μg) or infliximab RP (50 μg), using CT-P13- or Remicade-tagged reagents.

Patient Population

Patient disease characteristics, interventions and study visits have previously been described in detail [15-18]. Briefly, PLANETAS included patients aged 18–75 years diagnosed with AS according to the 1984 modified New York classification criteria [22] for at least 3 months prior to screening for eligibility for the trial, with a Bath Ankylosing Spondylitis Disease Activity Index score of ≥4 and Visual Analogue Score for spinal pain of ≥4. PLANETRA included patients aged 18–75 years diagnosed with RA according to the revised 1987 American College of Rheumatology classification criteria [23] for at least 12 months prior to screening who had not responded adequately to methotrexate for at least 3 months. Patients were randomised 1:1 to receive CT-P13 or infliximab RP at a dose of 5 mg/kg (PLANETAS) or 3 mg/kg (PLANETRA) via 2-h intravenous infusion at weeks 0, 2 and 6 and every 8 weeks thereafter, up to week 54. Patients in PLANETRA received concomitant methotrexate and folic acid. Patients in both studies were permitted to receive low-dose oral glucocorticoid and nonsteroidal anti-inflammatory drugs.

Immunogenicity Evaluation

Samples for immunogenicity testing were drawn prior to drug administration to avoid potential drug interference. Sera from patients enrolled in PLANETAS/PLANETRA were analysed for levels of ADAs at the time of screening (before administration of study drug), at weeks 14, 30 and 54 and at the end of study (EOS) visit at week 62. Therefore, the maximum number of samples analysed per patient was five. The level of ADAs was assessed in 5% patient serum using a validated ECL bridging immunoassay tagged with either EU-approved Remicade (assay ADA-A) or CT-P13 (assay ADA-B). This involved a screening assay that categorised samples as positive or negative depending on the presence or absence of ADAs, followed by a confirmatory competitive/inhibition assay (Fig. 1). Relative sensitivity of the assay was 75 ng/mL in 100% human serum, using either rabbit anti-CT-P13 or anti-Remicade antibodies as surrogate positive controls.

Fig. 1

Schematic of the bridging ECL immunoassay used to detect ADAs. In the screening assay, serum samples that potentially contain ADAs are acidified and then neutralised with labelled drug (biotinylated and SULFO-TAG™-labelled Remicade in assay ADA-A and biotinylated and SULFO-TAG™-labelled CT-P13 in assay ADA-B). ADAs present in the sample form immune complexes with the labelled drug. Samples are then loaded onto a streptavidin-coated electrode plate. During incubation, the biotin-containing antibody bridge binds to streptavidin on the plate. A reaction buffer is added, and SULFO-TAG™ labels near the electrode are electrochemically stimulated to emit light. The amount of luminescence is proportional to the amount of ADA in the sample. In the confirmatory assay, unlabelled drug is incubated with the sample and labelled drug. Unlabelled drug competes with the labelled drug to bind to ADAs in the sample. A reduction in the luminescence signal occurs if ADAs are present, thereby confirming the specificity of ADA and the positivity of the sample. Adapted from [30]. ADA antidrug antibody, ECL electrochemiluminescence, RP reference product

The neutralising activity of detected ADAs was assessed using a validated automated microfluidic Gyros™ immunoassay (Gyros AB, Uppsala, Sweden) tagged with either EU-approved Remicade (assay NAb-A) or CT-P13 (assay NAb-B). The assay is described in detail in Fig. 2. Relative sensitivity of the assay in 100% human serum, using rabbit anti-CT-P13 and rabbit anti-Remicade antibodies as surrogate positive controls, was 250 ng/mL and 631 ng/mL, respectively.

Fig. 2

Schematic of automated Gyros™ immunoassay used to detect NAbs. The automated microfluidic Gyros immunoassay utilises capillary action and centrifugal force to load TNF-specific biotinylated capture antibodies and streptavidin-coated beads onto miniaturised affinity columns. Full-length recombinant TNF proteins are then added to the columns. Samples that have been pre-incubated with Alexa fluorophore (fluorescence)-labelled drug (EU-approved Remicade in assay NAb-A and CT-P13 in assay NAb-B) are added to the columns. If the sample contains no NAbs, the Alexa-labelled drug binds to the immobilised TNF, is retained during subsequent wash steps, and generates a fluorescent signal. In samples containing RP or CT-P13 NAbs, the NAb binds to the Alexa-labelled drug during pre-incubation, thereby preventing it from binding to the TNF proteins in the column, resulting in a reduction in the fluorescence signal. The greater the reduction in fluorescence, the greater the amount of NAb in the sample. Adapted from [31]. EU European Union, NAb neutralising antibody, RP reference product, TNF tumour necrosis factor

Assay ADA-B and assay NAb-B replicated the original assays (assay ADA-A and assay NAb-A, respectively) in all respects, with the exception of the CT-P13 tag in place of the Remicade tag. Assay ADA-B and assay NAb-B were applied to verify the original results for the presence of ADAs and NAbs up to week 54/EOS. NAb titre data were also compared up to week 30.

Statistical Analysis

The analyses were conducted in the safety populations of PLANETAS and PLANETRA, which included all patients who received at least one full or partial dose of study treatment during any dosing period.

Assay Concordance

In order to assess the concordance of results obtained from assay ADA-A versus assay ADA-B and assay NAb-A versus assay NAb-B, an inter-rater agreement analysis was performed based on Cohen’s κ coefficient up to the EOS visit and per treatment group. The 95% confidence interval (CI) and p value were calculated for κ, with p < 0.05 indicating a statistically significant agreement between the outcomes obtained from the differently tagged assays.

The comparison of NAb titre results was performed at screening, week 14 and week 30. Statistical analysis was performed using Spearman’s rank correlation coefficient (ρ) based on actual titre values.

Cross-reactivity of ADA/NAbs

To assess the cross-reactivity of RP ADAs and NAbs, the proportion of positive samples (assessed with assay ADA-A and assay NAb-A, respectively) from RP-treated patients that also gave a positive result using the CT-P13-tagged immunoassays was determined. Similarly, to assess the cross-reactivity of CT-P13 ADAs and NAbs, the proportion of positive samples (assessed with assay ADA-B and assay NAb-B, respectively) from CT-P13-treated patients that were also positive using the Remicade-tagged immunoassays was calculated.

Results

Cross-inhibition Test

The cross-inhibition test assesses the ability of the study drug (CT-P13 or RP) to compete with CT-P13- and Remicade-tagged reagents and bind to immunocompetition controls. This is determined by calculating the reduction in signal generated by CT-P13 and RP in the assay compared with that produced with buffer (i.e. no study drug) using ADA controls of varying concentrations. Results from the cross-inhibition test demonstrate that CT-P13 and RP were comparable in their ability to inhibit the signal across all controls (Table 1).

Table 1

Cross-inhibition of CT-P13 and infliximab RP

	% signal inhibitiona
	CT-P13 ADA			RP ADA
	HPC	LPC	PNC	HPC	LPC	PNC
CT-P13	97.2	87.7	55.3	97.0	87.1	48.5
Infliximab RP	97.1	88.9	49.0	97.1	89.0	47.1

ADA antidrug antibody, HPC high positive control (1000 ng/mL), LPC low positive control (150 ng/mL), PNC pooled negative control, RP reference product

a% signal inhibition = (1 − [mean signal with drug/mean signal with buffer] × 100)

PLANETAS

The PLANETAS safety population (all patients who received at least one full or partial dose of study treatment during any dosing period) included all 250 patients with AS who were randomised to treatment in the study (n = 128 and n = 122 in the CT-P13 and RP treatment groups, respectively). The mean [standard deviation (SD)] total number of infusions received up to and including week 54 was 8.4 (1.7) and 8.5 (1.5) in the CT-P13 and RP groups, respectively. The mean (SD) total dose administered up to and including week 54 was also similar between treatment groups [3186.7 (969.1) mg and 3258.0 (861.5) mg, respectively]. Baseline patient characteristics in the safety population are shown in online resource 1 (see the electronic supplementary material).

Cross-reactivity of ADAs and NAbs

Of the 101 post-screening serum samples that tested positive for RP ADA in the RP-treated group, 94 (93.1%) were cross-reactive with CT-P13. A similar proportion of RP NAb-positive samples [93/100 (93.0%)] cross-reacted with the biosimilar (Fig. 3a). Almost all sera negative for RP ADA [345/346 (99.7%)] and NAb [345/347 (99.4%)] were also negative for CT-P13 ADA and NAb, respectively.

Fig. 3

All positive and cross-reactive ADA/NAb samples from a PLANETAS, b PLANETRA and c pooled analysis of PLANETAS and PLANETRA. *Number of samples found to be positive for ADAs/NAbs using the assay tagged with the treatment drug, i.e. assay ADA-A/assay NAb-A in RP-treated patients and assay ADA-B/assay NAb-B in CT-P13-treated patients. **Number of samples found to be positive using both CT-P13- and Remicade-tagged assays. ADA antidrug antibody, NAb neutralising antibody, RP reference product

In the CT-P13 group, 104 samples tested positive for CT-P13 ADA. Of these, 103 (99.0%) were cross-reactive with RP. Similarly, 98 of the 100 samples (98.0%) containing CT-P13 NAb were also found to cross-react with RP (Fig. 3a). Of the 365 samples that were negative for CT-P13 ADA, 356 (97.5%) were negative for RP ADA. A similar proportion of CT-P13 NAb-negative samples were negative for RP NAb [356/367 (97.0%)]. The number of patients that developed cross-reactive ADAs and NAbs per study week is presented in Table 2.

Table 2

Number (%) of patients who developed cross-reactive ADAs or NAbs per study weeka

Visit	PLANETAS		PLANETRA
	ADAs	NAbs	ADAs	NAbs
CT-P13 treatment group
Screeningb	–	–	–	–
Week 14	10/10 (100.0)	9/9 (100.0)	68/72 (94.4)c	67/71 (94.4)
Week 30	30/31 (96.8)	29/30 (96.7)	120/126 (95.2)	117/123 (95.1)
Week 54	25/25 (100.0)	25/25 (100.0)	120/125 (96.0)	120/125 (96.0)
EOS	38/38 (100.0)	35/36 (97.2)	150/151 (99.3)	148/150 (98.7)
RP treatment group
Screeningb	–	–	–	–
Week 14	13/13 (100.0)	13/13 (100.0)	66/69 (95.7)d	63/66 (95.5)
Week 30	24/25 (96.0)	23/24 (95.8)	118/122 (96.7)	117/122 (95.9)
Week 54	26/28 (92.9)	26/28 (92.9)	100/108 (92.6)	96/104 (92.3)
EOS	31/35 (88.6)	31/35 (88.6)	142/151 (94.0)	139/148 (93.9)

ADA antidrug antibody, EOS end of study, NAb neutralising antibody, RP reference product, TNF tumour necrosis factor

aAs a proportion of the total number of patients who developed ADAs or NAbs (cross-reactive and non-cross-reactive), respectively

bResults are not included for the screening visits because ADAs detected at this time point were non-specific, as patients were anti-TNF agent-naïve

cFour patients had developed CT-P13 ADAs that were not detected by Remicade-tagged assay

dThree patients had developed RP ADAs that were not detected by CT-P13-tagged assay

Agreement of ADA Detection Assays

There was significant agreement in ADA detection between assay ADA-A and assay ADA-B at all post-screening visits [Cohen’s κ (95% CI): week 14 (n = 239), 0.98 (0.93–1.00); week 30 (n = 228), 0.95 (0.91–1.00); week 54 (n = 214), 0.97 (0.94–1.00); EOS (n = 235), 0.89 (0.83–0.95); all p < 0.001], with similar results observed for both treatment arms (Fig. 4a; online resource 2). Agreement of at least 86% was observed for both the positive and negative ADA subgroups at these time points (Table 3; online resource 2). Weaker agreement was demonstrated for samples taken at screening.

Fig. 4

Agreement of immunogenicity testing between Remicade- and CT-P13-tagged immunoassays for the detection of a ADAs and b NAbs in PLANETAS. *Assay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag. **Assay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag. ADA antidrug antibody, CI confidence interval, EOS end of study, EU European Union, NAb neutralising antibody, NE not estimable, RP reference product

Table 3

Percentage agreement of immunogenicity testing between immunoassays in PLANETAS (safety population)

Visit	Assay ADA-A vs. assay ADA-Ba		Assay NAb-A vs. assay NAb-Bb
	PPAc (%)	NPAd (%)	PPAc (%)	NPAd (%)
All patients (N = 250)
Screening	66.67	99.18	0.00	99.59
Week 14	95.83	100.00	95.65	100.00
Week 30	94.74	99.42	94.55	98.84
Week 54	96.23	100.00	96.23	100.00
EOS	87.34	99.36	84.62	98.73
CT-P13 treatment group (N = 128)
Screening	50.00	99.20	0.00	99.21
Week 14	90.91	100.00	90.00	100.00
Week 30	93.75	98.82	93.55	98.82
Week 54	100.00	100.00	100.00	100.00
EOS	86.36	100.00	81.40	98.73
RP treatment group (N = 122)
Screening	100.00	99.16	0.00	100.00
Week 14	100.00	100.00	100.00	100.00
Week 30	96.00	100.00	95.83	98.85
Week 54	92.86	100.00	92.86	100.00
EOS	88.57	98.72	88.57	98.72

ADA antidrug antibody, EOS end of study, EU European Union, NAb neutralising antibody, NPA negative percentage agreement, PPA positive percentage agreement, RP reference product

aAssay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag

bAssay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag

cPPA = (number of patients with positive outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with positive outcome for Remicade-tagged immunoassay) × 100

dNPA = (number of patients with negative outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with negative outcome for Remicade-tagged immunoassay) × 100

Agreement of NAb Detection Assays

The NAb detection assays showed strong agreement for the presence of NAb at all post-screening time points [Cohen’s κ (95% CI): week 14 (n = 239), 0.98 (0.93–1.00); week 30 (n = 228), 0.94 (0.89–0.99); week 54 (n = 214), 0.97 (0.94–1.00); EOS (n = 235), 0.86 (0.79–0.93); all p < 0.001]. Comparable results were observed for patients treated with CT-P13 and RP (Fig. 4b; online resource 3). Agreement between assays ranged from 81.40–100.00 for the positive and negative NAb subgroups at these time points (Table 3; online resource 3). Samples taken at screening demonstrated weaker agreement.

Significant concordance was observed between NAb titres in the total patient population at week 14 [Spearman’s ρ (95% CI) 0.73 (0.44–0.88), p < 0.001, n = 22] and at week 30 [0.74 (0.58–0.84), p < 0.001, n = 52]. Similar results were observed in both treatment groups [CT-P13 group: week 14, 0.82 (0.35–0.96), p = 0.004; week 30, 0.66 (0.38–0.83), p < 0.001; RP group: week 14, 0.76 (0.37–0.93), p = 0.001; week 30, 0.83 (0.64–0.92), p < 0.001]. Correlation at screening could not be assessed as there were no samples that gave a positive result for NAb using both detection assays at this time point.

PLANETRA

The safety population in PLANETRA (all patients who received at least one full or partial dose of study treatment during any dosing period) included 602 of 606 randomised patients with RA (n = 302 and n = 300 treated with CT-P13 or RP, respectively). The mean (SD) total number of doses received up to and including week 54 was 8.0 (2.1) and 7.9 (2.1) in the CT-P13 and RP groups, respectively. The mean (SD) total dose administered up to and including week 54 was also similar between treatment groups [1712.4 (608.3) mg and 1672.8 (595.1) mg, respectively]. Baseline patient characteristics are shown in online resource 4. Results for ADA and NAb detection up to week 30 using assay ADA-A and assay NAb-A, respectively, have been published previously [17].

Of the 450 samples from patients in the RP-treated group that tested positive for RP ADA, 426 (94.7%) were cross-reactive with CT-P13. A similar proportion of RP NAb-positive samples [415/440 (94.3%)] cross-reacted with the biosimilar (Fig. 3b). Most of the RP ADA- [550/560 (98.2%)] and NAb-negative [556/568 (97.9%)] sera were also negative for CT-P13 ADAs and NAbs, respectively.

In the CT-P13 group, 474 samples tested positive for CT-P13 ADAs. Of these, 458 (96.6%) were cross-reactive with RP. Similarly, 452 of the 469 samples (96.4%) containing CT-P13 NAbs were also found to cross-react (Fig. 3b). Of the 555 samples that were negative for CT-P13 ADA, 540 (97.3%) were negative for RP ADA. A similar proportion of CT-P13 NAb-negative samples were negative for RP NAb [544/560 (97.1%)]. The number of patients that developed cross-reactive ADAs and NAbs per study week is presented in Table 2.

There was significant and strong agreement in ADA detection between assay ADA-A and assay ADA-B at all study visits [Cohen’s κ (95% CI): screening (n = 599), 0.88 (0.76–1.00); week 14 (n = 543), 0.94 (0.91–0.97); week 30 (n = 505), 0.94 (0.92–0.97); week 54 (n = 455), 0.92 (0.88–0.95); EOS (n = 539), 0.92 (0.89–0.96); all p < 0.001], with similar results observed for both treatment arms (Fig. 5a; online resource 5). Agreement of at least 92% was observed for both the positive and negative ADA subgroups at the post-screening time points (Table 4; online resource 5).

Fig. 5

Agreement of immunogenicity testing between Remicade- and CT-P13-tagged immunoassays for the detection of a ADAs and b NAbs in PLANETRA. *Assay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag. **Assay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag. ADA antidrug antibody, CI confidence interval, EOS end of study, EU European Union, NAb neutralising antibody, RP reference product

Table 4

Percentage agreement of immunogenicity testing between immunoassays in PLANETRA (safety population)

Visit	Assay ADA-A vs. assay ADA-Ba		Assay NAb-A vs. assay NAb-Bb
	PPAc (%)	NPAd (%)	PPAc (%)	NPAd (%)
All patients (N = 602)
Screening	100.00	99.32	60.00	99.66
Week 14	97.10	98.01	96.30	98.02
Week 30	97.54	96.93	96.30	96.95
Week 54	94.83	96.86	95.15	96.49
EOS	94.50	98.69	94.72	97.87
CT-P13 treatment group (N = 302)
Screening	100.00	98.97	33.33	99.66
Week 14	98.55	98.02	97.10	98.02
Week 30	98.36	95.38	96.69	95.42
Week 54	96.77	95.58	97.56	95.61
EOS	94.94	99.10	95.48	98.25
RP treatment group (N = 300)
Screening	100.00	99.66	100.00	99.66
Week 14	95.65	98.01	95.45	98.02
Week 30	96.72	98.47	95.90	98.47
Week 54	92.59	98.18	92.31	97.37
EOS	94.04	98.31	93.92	97.52

Results for ADA and NAb detection up to week 30 using assay ADA-A and assay NAb-A, respectively, have been published previously [17]

ADA antidrug antibody, EOS end of study, EU European Union, NAb neutralising antibody, NPA negative percentage agreement, PPA positive percentage agreement, RP reference product

aAssay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag

bAssay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag

cPPA = (number of patients with positive outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with positive outcome for Remicade-tagged immunoassay) × 100

dNPA = (number of patients with negative outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with negative outcome for Remicade-tagged immunoassay) × 100

The NAb detection assays showed high concordance for the presence of NAb at all post-screening time points [Cohen’s κ (95% CI): week 14 (n = 543), 0.94 (0.90–0.97); week 30 (n = 505), 0.93 (0.90–0.96); week 54 (n = 455), 0.92 (0.88–0.95); EOS (n = 539), 0.92 (0.89–0.95); all p < 0.001]. Comparable results were observed for patients treated with CT-P13 and those treated with RP (Fig. 5b; online resource 6). Agreement of at least 92% was observed for the positive and negative NAb subgroups at these time points (Table 4; online resource 6). Samples taken at screening demonstrated weaker agreement.

A statistically significant degree of concordance was observed between the NAb titre obtained using assay NAb-A and assay NAb-B in the total patient population at week 14 [Spearman’s ρ (95% CI) 0.61 (0.49–0.71), p < 0.001, n = 131] and at week 30 [0.72 (0.65–0.78), p < 0.001, n = 233]. Similar results were observed in both treatment groups [CT-P13 group: week 14, 0.64 (0.47–0.76); week 30, 0.69 (0.59–0.78); RP group: week 14, 0.59 (0.40–0.73); week 30, 0.74 (0.64–0.81); all p < 0.001]. Correlation at screening could not be determined due to small sample sizes.

Pooled PLANETAS and PLANETRA Study Data

Of the 551 samples from patients treated with RP that tested positive for RP ADAs, 520 (94.4%) were cross-reactive with CT-P13. A similar proportion of RP NAb-positive samples [508/540 (94.1%)] cross-reacted with the biosimilar (Fig. 3c). Most of the RP ADA- [895/906 (98.8%)] and NAb-negative [901/915 (98.5%)] sera were also negative for CT-P13 ADA and NAb, respectively.

For patients treated with CT-P13, 578 samples tested positive for CT-P13 ADAs. Of these, 561 (97.1%) were cross-reactive with RP. Similarly, 550 of the 569 samples (96.7%) containing CT-P13 NAb were also found to cross-react (Fig. 3c). Of the 920 samples that were negative for CT-P13 ADA, 896 (97.4%) were negative for RP ADA. A similar proportion of CT-P13 NAb-negative samples were negative for RP NAb [900/927 (97.1%)].

There was significant and strong agreement in ADA detection between assay ADA-A and assay ADA-B at all study visits [Cohen’s κ (95% CI): screening (n = 846), 0.83 (0.70–0.95); week 14 (n = 782), 0.95 (0.92–0.98); week 30 (n = 733), 0.95 (0.93–0.97); week 54 (n = 669), 0.94 (0.91–0.96); EOS (n = 774), 0.92 (0.89–0.95); all p < 0.001], with similar results observed for both treatment arms (all p < 0.001; Fig. 6a; online resource 7). Agreement of at least 92% was observed for both the positive and negative ADA subgroups at the post-screening time points (Table 5; online resource 7).

Fig. 6

Agreement of immunogenicity testing between Remicade- and CT-P13-tagged immunoassays for the detection of a ADAs and b NAbs in PLANETAS and PLANETRA pooled analysis. *Assay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag. **Assay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag. ADA antidrug antibody, CI confidence interval, EOS end of study, EU European Union, NAb neutralising antibody, RP reference product

Table 5

Percentage agreement of immunogenicity testing between immunoassays in pooled analysis of PLANETAS and PLANETRA (safety populations)

Visit	Assay ADA-A vs. assay ADA-Ba		Assay NAb-A vs. assay NAb-Bb
	PPAc (%)	NPAd (%)	PPAc (%)	NPAd (%)
All patients (N = 852)
Screening	94.44	99.28	42.86	99.64
Week 14	96.91	98.71	96.20	98.71
Week 30	97.01	97.92	95.97	97.70
Week 54	95.09	98.18	95.36	97.94
EOS	93.04	98.96	92.65	98.21
CT-P13 treatment group (N = 430)
Screening	90.91	99.04	25.00	99.53
Week 14	97.50	98.72	96.20	98.72
Week 30	97.40	96.74	96.05	96.76
Week 54	97.32	97.46	97.97	97.47
EOS	93.07	99.47	92.42	98.45
RP treatment group (N = 422)
Screening	100.00	99.51	66.67	99.76
Week 14	96.34	98.69	96.20	98.70
Week 30	96.60	99.08	95.89	98.62
Week 54	92.65	98.93	92.42	98.43
EOS	93.01	98.47	92.90	97.99

ADA antidrug antibody, EOS end of study, EU European Union, NAb neutralising antibody, NPA negative percentage agreement, PPA positive percentage agreement, RP reference product

aAssay ADA-A = ADA detection immunoassay with EU-approved Remicade tag; assay ADA-B = ADA detection immunoassay with CT-P13 tag

bAssay NAb-A = NAb detection immunoassay with EU-approved Remicade tag; assay NAb-B = NAb detection immunoassay with CT-P13 tag

cPPA = (number of patients with positive outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with positive outcome for Remicade-tagged immunoassay) × 100

dNPA = (number of patients with negative outcome for both Remicade- and CT-P13-tagged assays)/(number of patients with negative outcome for Remicade-tagged immunoassay) × 100

The NAb detection assays showed strong agreement for the presence of NAb at all post-screening time points [Cohen’s κ (95% CI): week 14 (n = 782), 0.94 (0.92–0.97); week 30 (n = 733), 0.94 (0.91–0.96); week 54 (n = 669), 0.94 (0.91–0.96); EOS (n = 774), 0.91 (0.88–0.94); all p < 0.001]. Weaker agreement was demonstrated for samples taken at sampling [0.46 (0.12–0.80), p < 0.001]. Comparable results were observed for patients treated with CT-P13 and with RP at all post-screening time points (all p < 0.001; Fig. 6b; online resource 8). Agreement between assays ranged from 92.42 to 98.72 for the positive and negative NAb subgroups at the post-screening time points (Table 5; online resource 8).

Significant concordance was observed between NAb titres in the total patient population at week 14 [Spearman’s ρ (95% CI) 0.63 (0.52–0.72), p < 0.001, n = 153] and at week 30 [0.72 (0.66–0.77), p < 0.001, n = 285]. Similar results were observed in CT-P13 and RP treatment groups at week 14 [0.68 (0.53–0.78) vs. 0.59 (0.42–0.72), respectively, both p < 0.001] and week 30 [0.70 (0.60–0.77) vs. 0.74 (0.65–0.80), respectively, both p < 0.001]. Correlation at screening could not be determined due to small sample sizes.

Discussion

The pivotal PLANETAS and PLANETRA studies demonstrated that CT-P13 is equivalent to RP in terms of efficacy and pharmacokinetics (PK) and has comparable pharmacodynamics, safety and immunogenicity [15-20]. The immunoassays used during the main comparative immunogenicity testing of CT-P13 versus RP utilised EU-approved Remicade as the tag to detect ADAs to both RP and CT-P13. When a one-assay approach is used, the regulatory view is that the biosimilar should be used as both the capture and detector tag, as this ensures optimal detection of biosimilar ADA, making it the more conservative approach [10]. However, this study has clearly demonstrated that there is little difference between the Remicade-tagged and CT-P13-tagged immunoassays used for ADA and NAb detection, with similar results obtained from both assays at all post-screening time points. This was true for both treatment groups, suggesting a lack of bias in either CT-P13 or RP antibody detection. The use of Remicade-tagged assays for samples from both CT-P13- and RP-treated patients in the PLANETAS and PLANETRA studies was therefore not inappropriate. Furthermore, a subsequent comparative review of ADA assays revealed that the ECL immunoassay tagged with EU-approved Remicade used in the trials was sensitive and reliable for the detection of CT-P13 ADAs [24].

Weaker agreement between assays was observed for samples taken at screening in both treatment groups in both studies. This finding may be due to the over-sensitivity of the Cohen’s κ coefficient with the extremely low numbers of patients who were initially positive for ADAs or NAbs in these studies. As patients in PLANETAS and PLANETRA were naïve to biological treatment prior to study commencement, the positive results obtained from screening samples were not a result of the presence of ADAs or NAbs, but reflect background noise due to non-specific binding of molecules that can occur in some subjects. As ADA or NAb status at screening is not related to treatment with RP or CT-P13 during the trial, little emphasis should be placed on the concordance of data at this time point, with data from weeks 14 to 54 providing a more relevant assessment of immunogenicity of the drugs. Results of the statistical analysis support significant concordance for all treatment groups at all study visits after infusion of the test product. A limitation to the present study is that the concordance of ADA titres between immunoassays was not analysed. Furthermore, information on the kinetics of development of ADAs and NAbs has not been presented as this was outside the scope of the current study, which aimed to evaluate the concordance of results obtained from immunoassays using differently tagged reagents.

The present study provides further support for the cross-reactivity of RP ADAs in patients with rheumatic diseases, as recently reported elsewhere [25]. A comparison of three bridging enzyme-linked immunosorbent assays (ELISAs) that used Remicade, Remsima or Inflectra to detect ADAs in RP-treated patients with RA or SpA (n = 250) revealed highly comparable results between assays [25]. All patients that tested positive for RP ADAs were also found to be positive using the Remsima and Inflectra assays, giving positive and negative percentage agreements of 100% for all comparisons. Significant correlation was observed between ADA titres assessed by the three assays (ρ ≥ 0.99, p < 0.001), and there was no significant difference in bias, or between indications or concomitant immunosuppressant use. The study used a commercial ELISA for RP ADA detection (Promonitor-ANTI-IFX kit, Progenika-Grifols, Spain) and concluded that this could be used to monitor ADAs in biosimilar-treated patients [25]; however, this cannot be generalised to other biosimilars. Cross-reactivity of RP ADAs has also been demonstrated in patients with IBD treated with RP. Gils et al. [26] developed three bridging ELISAs to compare cross-reactivity of RP ADAs with Remsima and Inflectra in 36 serum samples from patients with IBD. The study showed excellent correlation of ADA titres between assays (ρ ≥ 0.975, p < 0.0001), although small but significant differences were observed between the titres obtained by the three ELISAs [26]. In another study in patients with IBD, all sera positive for RP ADAs were cross-reactive with Remsima, with strong correlation of ADA titres demonstrated for all experiments (ρ ≥ 0.92, p < 0.001) [27]. In addition, the study demonstrated that RP ADAs inhibited the TNF-binding capacity of RP and Remsima to a similar extent and found a higher background signal for Remsima in the negative controls compared with the RP [27]. The exact cause of the background signal was undetermined, but was not related to glycosylation patterns or the immunoglobulin G molecule itself. No such background signal was evident in our study or in any of the other cross-reactivity studies mentioned here [25, 26].

A limitation to the studies mentioned above is that they only demonstrate cross-reactivity of RP ADAs with CT-P13 and not the cross-reactivity of CT-P13 ADAs with RP. PLANETRA and PLANETAS offered the opportunity to address this issue, as head-to-head data from both RP- and CT-P13-treated patients were available. Here, we have shown that CT-P13 ADAs are equally cross-reactive with RP, and we have also demonstrated the cross-reactivity of NAbs against RP and CT-P13. Evidence to date has indicated that most ADAs against the RP are neutralising [6]. This study has shown that ADAs against CT-P13 were also largely neutralising and that the prevalence of ADAs and NAbs, as well as NAb titres, was similar in both treatment groups. Our data also suggest that in a small number of patients, CT-P13-induced ADAs may be directed towards epitopes not present in the originator, and that RP-treated patients may develop ADAs that are not able to recognise the biosimilar. In this circumstance, using a one-assay approach for comparative immunogenicity testing would be a disadvantage, as these differences between products would not be detected. However, as this only occurs in a small number of patients, the clinical relevance is debatable. It is likely that such differences represent stochastic effects and not a systematic difference. Furthermore, without titre information, it is not possible to determine whether these discordant assay results are due to ADAs against a novel antigenic epitope in the biosimilar, low ADA titres that are close to the assay cut point, or, alternatively, a result of background noise. One potential cause for this background signal could be the presence of naturally occurring autoantibodies against TNF that are known to exist in some healthy individuals and those with inflammatory diseases [28, 29].

Determining the impact of ADAs versus NAbs on efficacy and tolerance in PLANETAS and PLANETRA was outside the scope of this study, but is a matter of considerable interest for future research. However, the clinical impact of ADAs was assessed throughout the PLANETAS and PLANETRA studies. In PLANETRA, peak serum drug concentrations (C max) and clinical responses were lower in the ADA-positive versus ADA-negative patients at week 30 [17]. Similarly, in PLANETAS, lower C max and clinical response rates were evident in ADA-positive patients at week 54. There was also a trend for steady-state PK parameters to be lower at higher ADA titres; however, as the study was not powered for this analysis, no statistical inference was made [16]. In open-label extensions of these trials in which patients were maintained on CT-P13 or switched from RP to the biosimilar, ADA-positive patients had lower clinical responses, higher C-reactive protein levels and erythrocyte sedimentation rates, and a higher incidence of infusion-related reactions compared with ADA-negative patients. These ADA effects, however, were highly similar between treatment groups in both the main study and during the extension [19, 20].

Conclusions

A comprehensive evaluation of the immunogenicity data from the pivotal clinical trials of CT-P13 has demonstrated that ADAs against CT-P13 and its RP recognise and bind the RP and CT-P13, respectively, to a similar degree, indicating that the two biological drugs share immunodominant epitopes. This cross-reactivity suggests that there should be no concerns about using commercialised kits that are utilised for either CT-P13 or its RP for comparative immunogenicity assessment.

Below is the link to the electronic supplementary material.

Supplementary material 1 (PDF 224 kb)

This study demonstrated that antibodies against the infliximab biosimilar CT-P13 and the infliximab reference product (RP; Remicade) recognise and bind RP and CT-P13, respectively.

The cross-reactivity of CT-P13 and its RP indicate that the two products share immunodominant epitopes.

The findings of this study are consistent with the previously reported high biosimilarity of CT-P13 to its RP and suggest that commercialised kits for the detection of antidrug antibodies against infliximab can be utilised for either CT-P13 or its RP.