sign in sign up
<<
Home , Adalimumab, Certolizumab pegol, Talizumab, TNF inhibitor

The effect of certolizumab drug concentration and anti-drug on TNF neutralisation L.C. Berkhout1, E.H. Vogelzang2, M.H. Hart1, F.C. Loeff1, L. Dijk1, N.I.L. Derksen1, R. Wieringa3, W.A van Leeuwen3, C.L.M. Krieckaert2, A. de Vries3, M.T. Nurmohamed2,4, G.J. Wolbink1,2, T. Rispens1

1Department of Immunopathology, Sanquin Research, Amsterdam, The Netherlands, and Landsteiner Laboratory, Amsterdam UMC, Academic Medical Centre, University of Amsterdam, The Netherlands; 2Amsterdam and immunology Center | Reade, Amsterdam, The Netherlands; 3Biologics Lab, Sanquin Diagnostic Services, Amsterdam, The Netherlands; 4Amsterdam Rheumatology and immunology Center | Amsterdam UMC, VU University Medical Center, Amsterdam, The Netherlands.

Abstract Objective Tumour necrosis factor (TNF) inhibitors like certolizumab, elicit an immunogenic response leading to the formation of anti-drug antibodies (ADAs). We sought to mechanistically investigate the relationship between certolizumab concentrations, ADAs, and the effective TNF neutralising capacity in sera of (RA) patients. TNF neutralising capacity of certolizumab was compared to the neutralising capacity of .

Methods Serum samples were collected from 40 consecutive certolizumab-treated RA patients at baseline and 4, 16, 28 and 52 weeks after treatment initiation [Dutch Trial Register NTR (Nederlands Trial Register) Trial NL2824 no. 2965]. Certolizumab concentration and ADA titre were measured with a certolizumab bridging enzyme-linked immunosorbent assay (ELISA) and a drug-tolerant radioimmunoassay (RIA), respectively. TNF neutralisation by certolizumab and adalimumab, in presence or absence of ADAs, was analysed with the TNF-sensitive WEHI bioassay.

Results Despite a high incidence of ADAs during one year of follow-up (65%; 26/40 patients), certolizumab levels of >10 μg/ml were measured in most patients. The capacity for TNF neutralisation highly correlated with certolizumab serum concentration, whereas no association with ADAs was observed. Similar results were obtained for adalimumab. The relative in vitro neutralising potency was higher for certolizumab compared to adalimumab.

Conclusion Anti-certolizumab antibodies were detected in a large proportion of patients, but in most cases where ADAs were detected, certolizumab was also present in high concentrations, directly correlating with in vitro neutralising capacity. These results indicate that measurement of certolizumab drug levels, rather than ADAs, have direct clinical signifcance.

Key words certolizumab, anti-drug antibodies, TNF, rheumatoid arthritis, adalimumab

Clinical and Experimental Rheumatology 2020; 38:Clinical 306-313. and Experimental Rheumatology 2020 TNF neutralisation by certolizumab / L.C. Berkhout et al.

Lea C. Berkhout, Msc Introduction viously, we have shown that the anti- Erik H. Vogelzang, MD, LLM Biological disease-modifying anti- body response to a range of therapeutic Margreet H. Hart, Bsc rheumatic drugs (bDMARDs) have antibodies, including certolizumab and Floris C. Loeff, Msc been developed to inhibit the activity of adalimumab, is highly restricted to Lisanne Dijk, Msc Ninotska I.L. Derksen, Msc infammatory cytokines such as tumour the binding site, thereby pre- Roeland Wieringa, PhD necrosis factor (TNF). These TNF in- dominantly neutralising (13, 14). In a W. Astrid van Leeuwen hibitors have proven to be a successful number of studies, increasing serum Charlotte L.M. Krieckaert, MD, PhD treatment option for patients with rheu- concentrations of TNF inhibitors, in- Annick de Vries, PhD matoid arthritis (RA) and other infam- cluding certolizumab, were associated Michael T. Nurmohamed, MD, Prof. matory disorders (1, 2). Therapeutic with better clinical outcome (4, 7, 15- Gerrit J. Wolbink, MD, PhD antibodies such as certolizumab have 18). Furthermore, the amount of TNF Theo Rispens, PhD been shown to elicit an immunogenic inhibition will depend on the strength Please address correspondence to: response leading to the formation of of binding between TNF inhibitor and Theo Rispens, Department of Immunopathology, anti-drug antibodies (ADAs). How- ADAs on the one hand and TNF inhibi- Sanquin Research, ever, the reported incidence, and levels tor and TNF on the other hand. In other Plesmanlaan 125, of anti-certolizumab antibodies varies words, the balance between TNF inhib- 1066 CX, Amsterdam, The Netherlands. highly between different studies (~5- itor concentration, ADA titre and TNF E-mail: [email protected] 37% of the patients) (2-4). In a recent concentration plays a role in determin- Received on February 19, 2019; accepted study by Jani et al. detection of ADAs ing the TNF neutralising effcacy of the in revised form on May 29, 2019. in certolizumab-treated RA patients TNF inhibitor. © Copyright CLINICAL AND was not associated with the 12 months In the present study we describe the in- EXPERIMENTAL RHEUMATOLOGY 2020. European League Against Rheumatism cidence of anti-certolizumab antibod- (EULAR) response (4). In contrast, in ies, as well as the relationship between certolizumab-treated Crohn’s disease serum certolizumab concentrations and patients persistent ADAs were corre- the TNF neutralising capacity in pres- lated with reduced effcacy, while tran- ence and absence of ADAs. We com- sient ADAs were not (5). These mixed pared the certolizumab neutralising results are seemingly in contrast with capacity with the neutralising capacity many studies demonstrating clear cor- of adalimumab, since these drugs are relations between ADA formation to structurally different and have a differ- adalimumab or infiximab and a lower ent binding strength for TNF (19). likelihood of minimal disease activity or clinical remission (6-10). Materials and methods Reprints will not be available from the The detection of ADAs in patients var- Details about the methodology can be author. ies widely between studies, depending found in the Supplementary fle. Brief- Funding: this work was supported by on duration of follow-up, concomitant ly, certolizumab concentration and ZonMw, the Netherlands Organisation for , the type of TNF inhibi- anti-certolizumab titre were Health Research and Development, in the tor and the type of assay that is used measured with a rabbit anti-certolizum- program 2Treat (grant no. 436001001). The certolizumab and adalimumab cohort to detect ADAs; a golden standard for ab bridging ELISA and a one-tiered or studies were partially supported by UCB the quantifcation of ADAs is missing two-tiered certolizumab RIA, respec- (IIS-2014-101540) and AbbVie (11). In particular, drug concentration tively, in 40 consecutive RA patients (ACA-NETH-05-012), respectively. may profoundly affect the detection of starting certolizumab treatment [Dutch UCB and AbbVie had no involvement in ADAs, depending on how drug-tolerant Trial Register NTR (Nederlands Trial the study design; in the collection, analysis, the ADA assay is. Importantly, ADAs Register) Trial NL2824 no. 2965]. The and interpretation of data. and drug levels will affect each other study was approved by the medical eth- Competing interests: mutually. ics committee of the Slotervaart Hospi- M.T. Nurmohamed has received consultancy ADAs are generally expected to only tal and Reade Medical Research Ethics fees from Abbott,Roche, Pfzer, MSD, UCB, SOBI and BMS, and payment for lectures affect treatment effcacy by lower- Committee, Amsterdam, the Nether- from Abbott, Roche and Pfzer. ing exposure to free active drug, via lands (CCMO NL35209.048.11). All G.J. Wolbink has received a research neutralisation and/or enhanced clear- patients gave written informed consent. grant from Pfzer (paid to the institution) ance. Hence, ADAs might only infu- To determine the TNF neutralising ac- and honoraria for lectures from UCB, ence clinical response when they affect tivity of certolizumab in patient sera, Pfzer, AbbVie, and BMS. (PK) to a noticeable in presence or absence of ADAs, the T. Rispens has received honoraria for degree (12). When enough free active TNF-sensitive WEHI bioassay was lectures from Pfzer, AbbVie and Regeneron, and a research grant from Genmab. drug is left to bind to its target, despite used. The TNF neutralising capacity of The other co-authors have declared no the presence of ADAs, ADAs are un- certolizumab was compared with the competing interests. likely to impair clinical response. Pre- neutralising capacity of adalimumab. A

Clinical and Experimental Rheumatology 2020 307 TNF neutralisation by certolizumab / L.C. Berkhout et al.

Fig. 1. Development of bridging ELISA for certolizumab. A cross-linking assay in which for both capture and detection polyclonal rabbit anti-idiotype antibodies were used, resulted in a sensitive assay to quantify certolizumab serum concentrations. A: Schematic overview of the certolizumab concentration assay. Certolizumab in serum is captured by polyclonal rabbit-anti certolizumab antibodies. Subsequently, biotinylated F(ab’)2 fragments of the same polyclonal rabbit anti-certolizumab antibodies are added for detection of certolizumab. B: Representative calibration curve of the certolizumab concentration assay. C: Inhibition of certolizumab specifc signal by increasing concentration of TNF. The concentration assay does not detect certolizumab bound to its target, assuring that only functional certolizumab is measured in this assay.

Table I. Demographics, previous and con- tres (20–830 AU/ml anti-certolizumab antibodies formed in a patient, this comitant therapies, and disease status at antibodies and 30–113 AU/ml or 30– fraction - that will no longer be active - baseline 1380 AU/ml anti-adalimumab antibod- will also not be measured. This assures Patients (n=40) ies, detected with the drug-sensitive that only functional, free certolizumab RIA or drug-tolerant ARIA, respec- is measured in this assay. Demographics tively). This selection was irrespective Age, median (IQR) (years) 56 (47-63) of any other (clinical) parameter. Patients Female, no. (%) 30 (75) BMI, median (IQR) 24.4 (22.0-27.8) Over one year, 147 serum samples Results from 40 consecutive RA patients who DMARD therapy Development of a rabbit started treatment with certolizumab Previous Biologic, no. (%) 20 (50) Previous DMARDs, median (IQR) 1 (0-2) anti-certolizumab bridging ELISA were included. The median age was 56 DMARDs use, no. (%) 34 (85) In order to measure certolizumab lev- years and 75% was female. The major- MTX use, no. (%) 28 (70) els in certolizumab-treated RA patients, ity of patients was concomitantly treat- MTX dose, median (IQR) 22.5 (15-25) we designed a bridging ELISA using ed with MTX (70%) and half of the (mg/week) polyclonal rabbit anti-idiotype antibod- patients was previously treated with Prednisone use, no. (%) 11 (28) Prednisone dose, median (IQR) 5 (5-10) ies for both capture and detection (Fig. a different biologic. Patients’ baseline (mg/day) 1A). This approach resulted in a high- characteristics are shown in Table I. 18 ly sensitive assay with a quantifable patients completed the study follow- Disease Status Disease duration, median (IQR) 8 (3-16) range between 0.1 and 120 μg/ml (Fig. up of one year. Twenty-two patients (years) 1B). The same principle was previously dropped out of the study: twelve due ACPA positive, no. (%) 30 (75) demonstrated to be valid by using rab- to treatment failure, three due to treat- IgM-RF positive, no. (%) 23 (58) bit anti- antibodies and na- ment failure and side effects, fve due Erosive, no. (%) 16 (40) Fab (20). to adverse events (including recurrent DAS28, median (IQR) 4.6 (3.2-5.3) ESR, median (IQR) (mm/hour) 20 (9-37) An advantage of this certolizumab respiratory infections, allergic reac- CRP, median (IQR) (mg/litre) 6 (2-17) bridging ELISA format is that it allows tions and pustulosis) and two Tender joint count, median (IQR) 6 (4-9) specifc detection of certolizumab. To due to withdrawal of their informed Swollen joint count, median (IQR) 3 (2-6) test whether this assay only detects consent. This high drop-out rate could IQR: inter quartile range; no: number; BMI: functional certolizumab, we conducted probably be explained by the fact that body mass index; DMARD: disease-modifying an inhibition experiment with TNF. Al- approximately half of the patients anti-rheumatic drug; MTX: ; ACPA: most complete inhibition of assay re- were not TNF inhibitor naïve, and anti-citrullinated protein antibody; IgM-RF: im- munoglobulin M rheumatoid factor; DAS28: Dis- sponse was observed when excess TNF thus previously failed treatment with ease Activity Score in 28-joints; ESR: erythrocyte was titrated into a sample containing 5 other TNF inhibitors. We observed 6 sedimentation rate; CRP: C-reactive protein. ng/ml certolizumab (Fig. 1C). In other cases with a drug hypersensitivity re- words, if the TNF binding site of cer- action. However, there was no signif- selection of samples was made, to rep- tolizumab is blocked with TNF it is no cant difference between patients with resent a range of serum certolizumab longer detected in this assay. Therefore, or without detectable ADAs (respec- (4.8–60 μg/ml) or adalimumab concen- we anticipate that if the TNF binding tively 2/26 [7.7%] and 4/14 [28.6%], trations (0.52–32.5 μg/ml) and ADA ti- site would be blocked by anti-idiotype p=0.214).

308 Clinical and Experimental Rheumatology 2020 TNF neutralisation by certolizumab / L.C. Berkhout et al.

Fig. 2. Longitudinal certolizumab and anti-certolizumab antibody concentration. A: Longitudinal certolizumab trough serum concentrations (μg/ml) were measured in 40 consecutive certolizumab-treated RA patients during one year follow-up. Red triangles indicate median certolizumab concentration of all patients at the individual time points. B: Longitudinal anti-certolizumab antibody titres (AU/ml) measured with a drug-tolerant one-step RIA. Numbers above the graph indicate anti-certolizum- ab antibody positive samples / total sample number. A-B: Solid lines connect measurements at different time points within a patient. Certolizumab concentrations and anti-certolizumab antibodies in orange, green and blue are from the same individual, respectively. C: Correlation between certolizumab concentration and anti-certolizumab antibody titre (Spearman’s ρ=-0.688, p<0.0001 (n=108)). Each dot represents the combined certolizumab trough level and ADA titre from a single individual at one time point. Baseline samples were excluded. A-C: Dashed lines indicate LLOQ of 0.1 μg/ml certolizumab or LOD of 20 AU/ml anti-certolizumab antibody.

Longitudinal certolizumab and anti- ment (green and orange line, Fig. 2B), 2A and B). Another patient had unde- certolizumab antibody concentrations in most patients with detectable anti- tectable certolizumab at week 28, but An initial loading dose resulted in body formation, ADAs could be de- ADAs could only be detected at week certolizumab trough levels reaching tected from week 16 onwards. Overall, 52. In the majority of patients, levels maximum concentration at week four the incidence of ADA positivity during of >10 μg/ml certolizumab were de- after treatment initiation. Substantial one year of follow-up was 65% (26/40 tected despite the simultaneous detec- variation in certolizumab trough levels patients). ADA titres of the ADA posi- tion of ADAs (Spearman’s ρ=-0.688, was observed between patients, with tive patients were 166 and 1120 AU/ p<0.0001; Fig. 2C). Therefore, we hy- concentrations of 45.0 (3.00–136) μg/ ml (n=2) at week four and 44 (21-517) pothesised that in the majority of ADA- ml, 29.5 (4.00–67.0) μg/ml, 24.5 (0.1– AU/ml, 112 (22-3120) AU/ml and 160 positive patients, certolizumab can still 78.0) μg/ml and 20.4 (0.6–59) μg/ml (22-830) AU/ml (median (IQR)), at exert its TNF neutralising function. (median (inter quartile range (IQR)) week 16, 28 and 52, respectively. Of all at week 4, 16, 28 and 52, respectively. patients who completed the one-year TNF neutralisation by certolizumab Despite inter-patient variation in cer- follow-up, two patients had an appar- correlates with drug concentration tolizumab concentration, longitudinal ent transient ADA response. To further examine the relationship be- intra-patient variation was minimal; The two patients with detectable ADAs tween anti-TNF, ADAs, and neutralisa- in 58% of the patients certolizumab at week four also had the lowest cer- tion capacity, we tested TNF neutralis- concentrations remained stable in time tolizumab concentrations at that point ing activity of patient sera with a TNF- within a factor three (Fig. 2A). (green and orange line, Fig. 2A and B). sensitive WEHI bioassay. This assay Next, we investigated anti-certolizum- Moreover, the patient with the highest quantifes killing of WEHI cells by bi- ab antibody formation with the drug- antibody concentrations overall had un- ologically active TNF (Fig. 3A). Kill- tolerant one-step RIA (21). Although in detectable certolizumab at week 28 and ing of WEHI cells can be rescued by two patients (5%) ADAs could already discontinued treatment at that time, due neutralisation of TNF by certolizumab be detected after four weeks of treat- to skin abnormalities (blue line, Fig. or adalimumab in serum. However,

Clinical and Experimental Rheumatology 2020 309 TNF neutralisation by certolizumab / L.C. Berkhout et al.

Fig. 3. TNF neutralisation by certolizumab and adalimumab serum samples, in presence or absence of ADAs. A: Schematic representation of the TNF-sensitive WEHI bioassay, used to analyse TNF neutralising activity. The intensity of the purple color, measured at 595 nm in a microtitre plate reader, is used as a readout for cell viability. TNF neutralisation by certolizumab or adalimumab in serum samples rescues WEHI cells (right upper panel), whereas presence of ADAs abolishes this protective effect (right lower panel). B: Serum samples were serially diluted and incubated with 100 pg/ml TNF to determine TNF neutralisation. Decreasing drug concentration resulted in re- duced TNF neutralisation, as represented by decreasing OD. A representative certolizumab serum sample is shown (Serum #2), and a baseline certolizumab sample was included as negative control (Serum #1). Data expressed as mean ± SD of duplicate measurements. C-D: Correlation between TNF neutralisation, as expressed by EC50 values, and (C) certolizumab concentration (Pearson r=0.909, p<0.0001 (n=12)) or (D) anti-certolizumab antibody titre (Pearson r=-0.471, p=0.122 (n=12)). E: Bioactivity of a titration of TNF incubated in presence or absence of 5 ng/ml certolizumab or adalimumab. Data expressed as mean ± SD of duplicate measurements. Graph depicts representative data of three independent experiments. F-H: Correlation between TNF neutralisation, as expressed by EC50 values, and (F) adalimumab concentration (Pearson r=0.900, p<0.0001 (n=21) or (G) anti-adalimumab antibody titre, measured with the drug-tolerant ARIA (Pearson r=-0.434, p=0.049 (n=21)) (H) or measured with the drug-sensitive RIA (Pearson r=-0.500, p=0.021 (n=21)). D, G-H: Red dots represent samples derived from patients that did not use methotrexate, black dots represent samples derived from patients concomitantly treated with methotrexate. Stratifcation by methotrexate did not change the results. C-D, F-H: Grey lines indicate log-log linear ft, weight by 1/Y2. the presence of ADAs can abolish this serially diluted and 100 pg/ml TNF determined (as described in Materials protective effect. A selection of patient was added to determine the neutralis- and methods), with higher EC50 values samples was made (see Materials and ing capacity. Subsequently, these sam- representing larger TNF neutralising methods), to represent a range of serum ples were added to the WEHI cells and capacity. certolizumab concentrations and ADA after 24 hours TNF neutralisation was Healthy donor serum and baseline pa- titres, irrespective of any other (clini- determined with the MTT-reduction tient samples were tested as negative cal) parameter. These samples were method. The EC50 of all samples was controls and were indeed not able to

310 Clinical and Experimental Rheumatology 2020 TNF neutralisation by certolizumab / L.C. Berkhout et al. neutralise TNF (EC50<40; serum #1 not anti-certolizumab antibodies. The bioassay, although we observed higher in Fig. 3B). TNF could be dose-de- same applied for adalimumab. EC50 values for certolizumab than pendently neutralised by certolizumab, for adalimumab. However, the direct as shown by a representative certoli- Discussion comparison of EC50 values should be zumab serum sample titration in Figure The present study describes the rela- done with caution, since certolizumab 3B (serum #2). Next, we analysed TNF tionship between certolizumab concen- serum concentrations are much higher neutralising activity in the abovemen- trations and TNF neutralising capac- compared to adalimumab concentra- tioned selected serum samples, with ity, in presence and absence of anti- tions. The high dosing regimen of cer- different certolizumab concentrations certolizumab antibodies, using sera tolizumab is remarkable, as we dem- and ADA titres. The TNF neutralising derived from certolizumab-treated RA onstrated effcient TNF neutralisation activity is expressed by an EC50 value. patients. We showed that certolizumab by certolizumab. for ex- TNF neutralisation was highly corre- trough concentrations were highest four ample, binds TNF with higher affnity lated with certolizumab serum concen- weeks after initiation of treatment, and compared to adalimumab (19), and has tration (Pearson r=0.909, p<0.0001; in 58% of the patients certolizumab a lower dosing regimen compared to Fig. 3C). By contrast, no correlation concentrations remained stable in time adalimumab. In striking contrast, cer- was observed between anti-certolizum- within a factor three. The minor drop in tolizumab binds TNF with even higher ab antibodies and TNF neutralisation certolizumab concentration after week affnity compared to golimumab, while (Pearson r=-0.471, p=0.122; Fig. 3D). four was not associated with the forma- the dosing regimen of certolizumab is Despite the presence of ADAs, certoli- tion of anti-certolizumab antibodies as the highest. Nonetheless, dose-fnding zumab could still exert its TNF neutral- shown in this study, but could rather be studies demonstrated a dose-dependent ising activity. explained by the initial loading dose of improvement in clinical response with Since certolizumab is a Fab domain certolizumab. increasing certolizumab doses (22–24). and thus structurally different from We demonstrated an inverse correlation Recently, it was shown that higher IgG1 antibodies, we compared TNF between anti-certolizumab antibod- certolizumab plasma levels were as- neutralisation by certolizumab to the ies and drug concentrations (Fig. 2C). sociated with larger DAS28 improve- TNF neutralising activity of adali- Only in a minority of patients (35%) ment from baseline (4, 18), although a mumab. To investigate whether there no anti-certolizumab antibodies could plateau effect in ΔDAS with 20 μg/ml is any difference in TNF neutralisation be detected during one year of follow- certolizumab has been described (17). by certolizumab and adalimumab it- up. However, in most samples in which Since certolizumab effciently neutral- self, we frst analysed TNF bioactivity anti-certolizumab antibodies were de- ises TNF, it is noteworthy that high in the presence or absence of 5 ng/ml tected, certolizumab serum concentra- certolizumab doses are required for certolizumab or adalimumab (Fig. 3E). tions remained well above 10 μg/ml, clinical response. There appears to be Although both drugs neutralised TNF, which is within the therapeutic range a discrepancy between in vitro effcacy the relative neutralising potency of according to the certolizumab concen- of TNF neutralisation and in vivo clini- certolizumab was higher compared to tration-effect curve (17), irrespective of cally effective certolizumab concentra- adalimumab. Next, we analysed TNF the presence of detectable ADAs. tions. One might wonder whether local neutralisation by serum samples from We demonstrated that both certolizum- certolizumab concentrations, at the site adalimumab-treated RA patients (6). ab and adalimumab neutralised TNF, al- of infammation, are lower compared Like certolizumab, adalimumab serum though the relative in vitro neutralising to systemic concentrations. However, levels were highly correlated with TNF potency was higher for certolizumab as certolizumab effectively entered the neutralising activity (Pearson r=0.900, compared to the neutralising potency of infamed tissue (25, 26), a lack of distri- p<0.0001; Fig. 3F). A weak correlation adalimumab. Certolizumab binds TNF bution to the local site of infammation was observed between the anti-adali- with higher affnity compared to adali- does not explain this controversy. Fur- mumab antibody titre, measured with mumab, and can therefore more eff- thermore, differences in Fc-Rn-mediat- a drug-tolerant ARIA, and TNF neu- ciently bind and neutralise TNF (19). ed recycling and Fc-mediated clearance tralisation (Pearson r=-0.434, p=0.049; Of note, certolizumab is a monovalent between certolizumab, adalimumab Fig. 3G). Anti-adalimumab antibodies Fab fragment, whereas adalimumab is and golimumab might contribute to measured with a drug-sensitive RIA as- a bivalent antibody. However, when the difference in dosing regimen, given say showed a slightly stronger correla- the concentrations would have been that certolizumab is a Fab domain while tion with TNF neutralisation (Pearson corrected for equal amounts of bind- adalimumab and golimumab are IgG1 r=-0.500, p=0.021; Fig. 3H). However, ing sites, the difference in neutralising antibodies. These structural differences this correlation was still weak com- effcacy would have been even more might also result in different Fc-mediat- pared to the correlation between adali- pronounced. ed effector functions, including apopto- mumab serum levels and TNF neutrali- For both certolizumab and adalimum- sis and complement activation (27-29). sation. Together, we showed that TNF ab, there was a strong correlation be- Finally, it is possible that high certoli- neutralisation by certolizumab relates tween drug levels and TNF neutralis- zumab dosing contributes to the induc- directly to drug concentration, but ing capacity as measured in the WEHI tion of tolerance, thereby suppressing

Clinical and Experimental Rheumatology 2020 311 TNF neutralisation by certolizumab / L.C. Berkhout et al. immunogenicity. This would suggest adherence, could not be performed. association of random drug levels with ef- that after the initial certolizumab load- Instead, we used TNF neutralising ca- fcacy in -treated patients with rheumatoid arthritis: results from the ing dose, lower doses of certolizumab pacity as a functional outcome meas- BRAGGSS cohort. Ann Rheum Dis 2016; might be suffcient to maintain clinical urement. For a direct assessment of an 76: 208-13. effectiveness. However, the main ef- association between certolizumab con- 5. SANDBORN WJ, WOLF DC, KOSUTIC G et fect of antibody formation is expected centrations and clinical response, larger al.: Effects of transient and persistent anti- drug antibodies to certolizumab pegol: lon- to be lowering the effective drug expo- studies should be performed. gitudinal data from a 7-year study in Crohn’s sure. In a concentration-effect curve the Of note, anti-certolizumab antibody Disease. Infamm Bowel Dis 2017; 23: 1047- impact of immunogenicity on effective titres were also determined using a 56. drug levels c.q. drug exposure is taken two-tiered screening and confrmation 6. BARTELDS GM, KRIECKAERT CL, NURMO- HAMED MT et al.: Development of antidrug into account, and this analysis still sug- approach, conform the FDA guidelines antibodies against adalimumab and associa- gests added beneft of certolizumab (31), but this appeared to have little tion with disease activity and treatment fail- concentrations up to almost 20 μg/ml consequences for the overall number of ure during long-term follow-up. JAMA 2011; 305: 1460-8. for RA (17). samples identifed as positive (Supple- 7. CHEN D, CHEN Y, TSAI W et al.: Signifcant In this study, antibodies were detected mentary Fig. S1). associations of antidrug antibody levels with in 65% of the patients using a previ- In conclusion, we demonstrated that serum drug trough levels and therapeutic re- ously developed ADA assay (21). This although a large proportion of patients sponse of adalimumab and treat- ment in rheumatoid arthritis. Ann Rheum Dis is much higher compared to three previ- has detectable anti-certolizumab anti- 2015;74: e16. ously published certolizumab trials (2- bodies high concentrations of certoli- 8. WOLBINK GJ, VIS M, LEMS W et al.: Devel- 4). Two of these studies did not report zumab were found in most patients. We opment of antiinfiximab antibodies and rela- which assay had been used, making it showed that drug concentrations, but tionship to clinical response in patients with rheumatoid arthritis. Arthritis Rheum 2006; diffcult to compare results from one not the presence of anti-certolizumab 54: 711-5. study to another. The high percentage antibodies, was highly correlated with 9. PASCUAL-SALCEDO D, PLASENCIA C, RAMI- of antibody positive patients in our co- the capacity to neutralise TNF. So, se- RO S et al.: Infuence of immunogenicity on hort, could potentially be explained by rum drug concentrations refect clinical the effcacy of long-term treatment with in- fiximab in rheumatoid arthritis. Rheumatol- the drug-tolerant assay we used (21). effectiveness, which indicates that drug ogy 2011; 50: 1445-52. Jani et al. used the same drug-tolerant measurements can be used for the iden- 10. RADSTAKE TRDJ, SVENSON M, EIJSBOUTS one-step RIA for the detection of anti- tifcation of under treatment. Overall, AM et al.: Formation of antibodies against infiximab and adalimumab strongly corre- certolizumab antibodies, as used in this we advocate not to measure ADAs in lates with functional drug levels and clinical study (4). The difference in the percent- a clinical setting, unless certolizumab responses in rheumatoid arthritis. Ann Rheum age (37% in the study by Jani et al. vs. concentrations are very low. Dis 2009; 68: 1739-45. 65% in our study) of patients in whom 11. VAN SCHOUWENBURG PA, RISPENS T, WOLBINK GJ: Immunogenicity of anti-TNF ADAs are detected might, partly, be Acknowledgements biologic therapies for rheumatoid arthritis. due to differences in sampling strat- The authors would like to thank Karin Nat Rev Rheumatol 2013;9:164–72. egy. In this study samples were taken at van Schie for assistance with the WEHI 12. BLOEM K, HERNÁNDEZ-BREIJO B, MARTÍ- trough, as opposed to random sampling assay. NEZ-FEITO A et al.: Immunogenicity of thera- peutic antibodies: monitoring antidrug anti- in Jani et al. Even with a drug-tolerant bodies in a clinical context. Ther Drug Monit assay, drug levels will still infuence de- References 2017; 39: 327-32. tection of ADAs to certain degree. Fur- 1. KEYSTONE EC, KAVANAUGH AF, SHARP JT 13. VAN SCHIE KA, HART MH, DE GROOT ER et thermore, in the study by Jani et al. rel- et al.: Radiographic, clinical, and functional al.: The antibody response against outcomes of treatment with adalimumab (a and chimeric anti-TNF therapeutic antibod- atively more bDMARD naïve patients human anti- monoclo- ies primarily targets the TNF binding region. started certolizumab compared to our nal antibody) in patients with active rheuma- Ann Rheum Dis 2015; 74: 311-4. study (92% in the study by Jani et al. toid arthritis receiving concomitant metho- 14. VAN SCHIE K, KRUITHOF S, VAN SCHOU- trexate therapy: a randomized, placebo-con- WENBURG P et al.: Neutralizing capacity of vs. 50% in our study). It might be that in trolled. Arthritis Rheum 2004; 50: 1400-11. monoclonal and polyclonal anti-natalizumab our study some patients failed their pre- 2. SMOLEN J, LANDEWÉ R, MEASE P et al.: antibodies: The immune response to anti- vious bDMARD, due to immunogenic- Effcacy and safety of certolizumab pegol body therapeutics preferentially targets the ity. Therefore, these patients might be plus methotrexate in active rheumatoid ar- antigen-binding site. J Clin Immunol thritis: the RAPID 2 study. A randomised 2016; 3: 125-6. prone to develop ADAs against other controlled trial. Ann Rheum Dis 2009; 68: 15. KNEEPKENS EL, KRIECKAERT CLM, VAN DER bDMARDs (11, 30), further explaining 797-804. KLEIJ D et al.: Lower etanercept levels are the difference in the percentage of ADA 3. KEYSTONE E, HEIJDE DV, MASON D et al.: associated with high disease activity in an- positive patients. Certolizumab pegol plus methotrexate is kylosing spondylitis patients at 24 weeks of signifcantly more effective than placebo follow-up. Ann Rheum Dis 2015; 74: 1825-9. The small sample size is a limitation plus methotrexate in active rheumatoid ar- 16. POUW MF, KRIECKAERT CL, NURMOHAMED of the present study. Consequently, a thritis: fndings of a ffty-two-week, phase MT et al.: Key fndings towards optimising direct analysis of the relationships be- III, multicenter, randomized, double-blind, adalimumab treatment: The concentration- tween certolizumab concentrations, an- placebo-controlled, parallel-group study. effect curve. Ann Rheum Dis 2015; 74: 513-8. Arthritis Rheum 2008; 58: 3319-29. 17. WOLBINK G, GOUPILLE P, SANDBORN W et ti-certolizumab antibodies and clinical 4. JANI M, ISAACS JD, MORGAN AW et al.: al.: Association between plasma certolizum- response i.e. EULAR response or drug High frequency of antidrug antibodies and ab pegol concentration and improvement in

312 Clinical and Experimental Rheumatology 2020 TNF neutralisation by certolizumab / L.C. Berkhout et al.

disease activity in rheumatoid arthritis and 22. KEYSTONE E, CHOY E, KALDEN J et al.: pegol in patients with rheumatoid arthritis Crohn’s disease [abstract]. Arthritis Rheu- CDP870, a novel, PEGylated humanised and spondyloarthritis. RMD Open 2016; 2. matol 2016; 68. TNF-inhibitor is effective in treating the 27. ARORA T, PADAKI R, LIU L et al.: Differences 18. GEHIN JE, SYVERSEN SW, WARREN DJ et al.: signs and symptoms of rheumatoid arthritis. in binding and effector functions between Certolizumab pegol serum levels ≥20 mg/l Arthritis Rheum 2001; 44: 2946. classes of TNF antagonists. Cytokine 2009; are associated with improvement in das28 in 23. CHOY EHS, HAZLEMAN B, SMITH M et al.: 45: 124-31. rheumatoid arthritis patients. data from the Effcacy of a novel PEGylated humanized 28. ALBANESI M, DAËRON M: The interactions nor-dmard study [abstract]. Ann Rheum Dis anti-TNF fragment (CDP870) in patients of therapeutic antibodies with Fc receptors. 2018; 77: A953. with rheumatoid arthritis: a phase II double- Immunol Lett 2012; 143: 20-7. 19. VAN SCHIE KA, OOIJEVAAR-DE HEER P, DIJK blinded, randomized, dose-escalating trial. 29. ARMUZZI A, LIONETTI P, BLANDIZZI C et L et al.: Therapeutic TNF inhibitors can dif- Rheumatology 2002; 41: 1133-7. al.: Anti-TNF agents as therapeutic choice ferentially stabilize trimeric TNF by inhibit- 24. GOEL N, STEPHENS S: Certolizumab pegol. in immune-mediated infammatory diseases: ing monomer exchange. Sci Rep 2016; 6. MAbs 2010; 2: 137-47. focus on adalimumab. Int J Immunopathol 20. RISPENS T, VAN LEEUWEN A, VENNEGOOR A 25. PALFRAMAN R, AIREY M, MOORE A et al.: Pharmacol 2014;27: 11-32. et al.: Measurement of serum levels of natal- Use of biofuorescence imaging to com- 30. BARTELDS GM, DE GROOT E, NURMO- izumab, an immunoglobulin G4 therapeutic pare the distribution of certolizumab pegol, HAMED MT et al.: Surprising negative asso- . Anal Biochem 2011; adalimumab, and infiximab in the infamed ciation between IgG1 allotype disparity and 411: 271-6. paws of mice with collagen-induced arthritis. anti-adalimumab formation: a cohort study. 21. RISPENS T, HART MH, OOIJEVAAR-DE HEER J Immunol Methods 2009; 348: 36-41. Arthritis Res Ther 2010; 12: R221. P et al.: Drug interference in immunogenicity 26. CARRON P, LAMBERT B, VAN PRAET L et 31. Assay Development and Validation for Im- assays depends on valency. J Pharm Biomed al.: Scintigraphic detection of TNF-driven munogenicity Testing of Therapeutic Protein Anal 2013; 85: 179-85. infammation by radiolabelled certolizumab Products Guidance for Industry.

Clinical and Experimental Rheumatology 2020 313