A notable proportion of patients with inflammatory bowel disease (IBD) are switched from infliximab (IFX) to adalimumab (ADL). We investigated if immunogenicity of IFX influenced immunogenicity and clinical outcomes of later ADL therapy.
Single-center cohort study including all patients with IBD assessed for antibodies (Abs) against IFX or ADL.
Anti-IFX Abs were evaluated in 187 patients treated with IFX as first line anti-TNF agent. Approximately, half (49%) were positive. Detected anti-IFX Abs had functional capacity as judged by a median IFX concentration below limit of detection (interquartile range, 0.0–0.0 μg/mL) versus 3.8 μg/mL (IQR, 1.3–7.9) in anti-IFX Ab-negative patients, P < 0.0001; but did not cross-react with ADL. Anti-ADL Abs were assessed in 57 ADL-treated patients. Twelve (21%) tested positive. Patients with previous anti-IFX Ab development were significantly more prone to develop anti-ADL Abs (33%) than those without (0%): odds ratio estimated 11, P = 0.04. The anti-ADL Abs were also functional because ADL was undetectable in all anti-ADL Ab-positive patients versus median 8.3 μg/mL (IQR 5.0–11.0) in anti-ADL-negative patients, P < 0.0001. The presence of anti-ADL Abs increased the risk of secondary ADL treatment failure with OR 28 (3–248), P < 0.001. ADL trough levels, irrespectively of anti-ADL Ab status, associated with efficacy of ADL maintenance therapy: AUCROC 0.77 (0.62–0.93), P < 0.01.
Switchers with anti-IFX Abs are prone to develop de novo anti-ADL Abs, which may result in therapeutic failure. Assessment of ADL immunogenicity in anti-IFX Ab-positive switchers is required to ensure optimal interventions at inadequate treatment responses and to avoid inappropriate ADL intensification regimens.
Anti-tumor necrosis factor–α (TNF) therapy is effective in case of insufficient disease control on conventional treatment in patients with inflammatory bowel disease (IBD) and has markedly improved clinical outcomes, life quality, and work productivity in those with a moderate-to-severe disease phenotype.1,2 Unfortunately, a notable proportion experiences relapse of active disease, despite ongoing anti-TNF therapy with infliximab (IFX).3,4 International guidelines suggest intensifying the IFX regimen in this event, but only about half the patients regain clinical effect with diminishing effectiveness over time.5,–8 In case of inadequate response to IFX intensification, it is generally recommended to switch to a second TNF-inhibitor, such as adalimumab (ADL).5,–8 Switching between TNF-inhibitors is also used in case of side effects, such as infusion reactions to IFX.9,–11
There is considerable variation in the response to ADL after switching from IFX, and factors influencing outcome of ADL therapy in this setting are largely unknown.12,–14 Immunological recognition of ADL, a fully human IgG1 monoclonal antibody (Ab), may induce anti-ADL Abs. Anti-ADL Abs have been reported in up to 17% of patients with IBD and seem to develop less frequently than antibodies against IFX presumably due to the chimeric structure of IFX.15,–18 Anti-ADL Abs primarily target the TNF-binding sites of ADL.19 Depending on avidity, these antibodies may therefore interfere with binding of the drug to TNF and, in addition, increase drug clearance through formation of immune complexes. Analogous to IFX, anti-ADL Abs have been associated with low serum trough levels of ADL and secondary treatment failure.15,–17,20 Observations in patients with rheumatoid arthritis suggest that anti-ADL Abs develop more frequently in IFX switchers with previous anti-IFX Abs than in anti-TNF-naive patients.21 Therefore, we investigated if patients with IBD positive for anti-IFX Abs are more prone to develop anti-ADL Abs than IFX-to-ADL switchers without anti-IFX Abs. Furthermore, if anti-ADL Abs were functional and reduced circulating ADL levels and influenced on clinical efficacy.
Methods
Study Design and Patients
This was a retrospective cohort study of patients with IBD treated with IFX or ADL at a single tertiary IBD center until July, 2013 (Department of Gastroenterology, Herlev Hospital, Denmark). The study population comprised all patients having been assessed for antibodies against IFX or ADL. Patients were identified by review of files of anti-TNF-treated patients with IBD during the designated time period. The study was approved by the Danish Health and Medical Authority and the Danish Data Protection Agency.
Endpoints
The objective was to investigate if patients with IBD with previous anti-IFX Ab development had increased risk of developing anti-ADL Abs in case of switching anti-TNF agent. In addition, to assess if anti-ADL Abs were functional and influenced the clinical efficacy.
Evaluations
Variables potentially associated with the development of anti-ADL Abs were predefined and consisted of patient and disease characteristics, demographics, and characteristics of the anti-TNF treatment regimen. Clinical outcome of ADL treatment was classified according to the treating physician's global evaluation. Primary nonresponse was complete lack of effect of ADL induction and consequently followed by discontinuation of ADL therapy. Secondary ADL treatment failure was defined as initial favorable response to ADL induction and maintenance therapy and later followed by loss of clinical effect with symptoms of active disease despite dose optimization and finally resulting in discontinuation of ADL. Maintained ADL response was sustained favorable clinical effect on ADL maintenance therapy at time of study.
Serum Analyses
Samples
All serum analyses had been done at the discretion of the treating physician. Included samples were obtained as trough levels except for a few cases collected in relation to an allergic reaction to IFX (explorative exclusion of these samples did not change the findings of this study). In case of more than 1 available test, the most recent one was used. Serum was collected after centrifugation of 10 mL venous blood and sent at room temperature for analysis at Biomonitor A/S (Copenhagen, Denmark) under blinded conditions. A minority of samples were stored at −80°C until analysis.
Anti-TNF Drug Bioavailability and Immunogenicity
Functional bioactive concentrations of anti-TNF drugs were measured as the levels of IFX or ADL providing the same TNF binding as those of the IgG-fraction of the serum samples and was tested by clinically validated fluid-phase radioimmunoassays (RIA) as previously detailed (limit of detection for IFX 0.15 μg/mL and limit of detection for ADL 0.60 μg/mL).22,–24 Anti-drug Ab assays were carried out as previously described using fluid-phase RIA and antihuman λ light-chain Abs to distinguish between free drug and drug in complex with any class of λ-containing human immunoglobulin.22,–24 Anti-IFX Ab concentrations were expressed as assay-specific arbitrary laboratory units (U) per milliliter (limit of detection 10 U/mL). Anti-ADL Abs were reported as positive when activity was >2× activity in pooled normal human serum. Cross binding between detected anti-IFX Abs and ADL was tested by coincubation of 125I-IFX and sample as described above with and without IFX at 25 μg/mL and ADL at 50 μg/mL, respectively. The limit for positive cross binding of anti-IFX Abs to ADL was set to 20% of the displaced 125I-IFX obtained by the addition of IFX.11
Statistics
Descriptive statistics were given as percentages for discrete variables and as medians with interquartile ranges (IQR) for continuous variables. Univariate analysis of discrete variables was done by Fisher's exact test or χ2 test, as appropriate, and by Mann–Whitney U test (unpaired) or Wilcoxon signed-rank test (paired) for continuous variables. Receiver operating characteristic analysis was used to determine ADL cutoff values associated with clinical outcomes. Statistical analyses were done in SPSS version 20 (IBM, Somer, NY) and GraphPad Prism (version 5; GraphPad Software, San Diego, CA). Two sided P values <0.05 were significant.
Results
Study Population
The study population comprised all anti-TNF-treated patients with IBD assessed for antibodies against IFX or ADL (Fig. 1). A total of 187 patients had been evaluated for anti-IFX Abs after median 6 IFX infusions (IQR, 4–14). Approximately, half (49%) were positive, and anti-IFX Abs generally eliminated IFX detection (Fig. 2). Anti-IFX Abs did not cross-react with ADL in any of the 29 patients evaluated. Anti-ADL Abs had been assessed in 57 ADL-treated patients after median 9 months of ADL therapy (IQR, 5–22). Twelve patients (21%) were positive for anti-ADL Abs (Fig. 1).
Study population. The entire IBD cohort treated with IFX or ADL is shown. The study population is marked by bold cells. *Anti-IFX Abs not measured in 227 patients, and invalid (nontrough) testings in 61 patients. §Cross reactivity testing was done in a positive anti-IFX Ab sample obtained at a different time points than test results otherwise used in the study in 7 patients. #Anti-ADL Abs not measured in 33 patients, and invalid (nontrough) testings in 3 patients. †Anti-ADL Abs not measured. ¤Anti-ADL Abs not measured in 16 patients and invalid (nontrough) testings in 3 patients.
Drug neutralizing capacity of anti-IFX antibodies. IFX trough levels in patients positive (+) for anti-IFX antibodies (Abs) (75 of 92 patients assessed) or negative (−) for anti-IFX Abs (85 of 95 patients assessed). Horizontal lines denote medians and IQR. *P < 0.0001.
Risk Factors of Anti-ADL Ab Development
Patients with previous anti-IFX Ab development were significantly more prone to develop anti-ADL Abs (10 of 30 patients: 33%) than those without (0 of 10 patients: 0%): OR estimated 11, P = 0.04 (Fig. 1). No other variables associated with anti-ADL Ab development (Table 1).
Influence of Anti-ADL Abs on ADL Bioavailability
As shown in Figure 3, anti-ADL Abs eliminated ADL detection in all patients. In contrast, anti-ADL Ab-negative patients generally had relatively high ADL levels at time of testing (median 8.3 μg/mL; IQR, 5.0–11.0; P < 0.0001), and only a single of these patients had undetectable ADL.
Drug neutralizing capacity of anti-ADL Abs. ADL trough levels in patients positive (+) for anti-ADL Abs (9 of 12 assessed) or negative (−) for anti-ADL Abs (40 of 45 patients assessed). Horizontal lines denote medians and IQR. *P < 0.0001.
Clinical Outcomes
Anti-ADL Abs
Of the 57 patients evaluated for anti-ADL Abs, 4 patients (7%) had primary nonresponse and discontinued ADL after the induction series. Twenty patients (35%) developed secondary loss of response to ADL and consequently discontinued ADL therapy again, whereas 29 patients (51%) had a maintained response on ADL and either received maintenance therapy at time of study (n = 20) or discontinued ADL due to remission (n = 9). The remaining 4 patients (7%) had discontinued ADL because of recurrent infections or lack of compliance. Classification of clinical outcomes was supported by biochemical parameters (Fig. 4). Development of anti-ADL Abs substantially and highly significantly increased the risk of secondary ADL treatment failure: OR 28 (3–248), P < 0.001 (Fig. 5A and Table 2).
Biochemical markers of disease activity in patients with loss of response (n = 20), and maintained response (n = 29) to ADL therapy at time of ADL initiation and at discontinuation/follow-up, respectively. Samples from 3 patients with loss of response and 6 with maintained response were unavailable at ADL initiation and at discontinuation/follow-up from 1 patient with loss of response and 1 with maintained response. *P < 0.05; **P < 0.01; ***P < 0.001.
ADL bioavailability and immunogenicity in patients stratified for clinical outcomes of ADL maintenance therapy. A, Anti-ADL Abs. Dark columns denote Ab positive patients. B, ADL serum trough levels. Horizontal lines denote medians and IQR. C, Receiver operating characteristic (ROC) curve of ADL trough level including line of identity. *P < 0.01; **P < 0.001.
Accuracy of Monitoring of ADL Bioavailability and Immunogenicity for Evaluation of Secondary ADL Treatment Failure
 |
|
Accuracy of Monitoring of ADL Bioavailability and Immunogenicity for Evaluation of Secondary ADL Treatment Failure
|
|
ADL Bioavailability
ADL levels were significantly lower in patients with secondary ADL treatment failure (median, 0.0 μg/mL; IQR, 0.0–8.6) than in patients with maintained effect of ADL (median, 9.2 μg/mL; IQR, 6.1–11.6; P < 0.01) (Fig. 5B). Receiver operating characteristic analysis was used to determine ADL bioavailability cutoff values associated with secondary treatment failure: AUCROC 0.77 (0.62–0.93); P < 0.01 (Fig. 5C). Test characteristics are detailed in Table 2.
Discussion
Although elective switching of anti-TNF therapy from IFX to ADL in patients with IBD in remission is controversial, change of therapy from IFX to ADL is commonly used in case of insufficient efficacy or side effects.5,–8,25 The clinical outcome of ADL therapy in IFX switchers is unpredictable, and little is known about factors influencing efficacy.9,10,13,14,18,26 In line with previous reportings, we observed that functionally active antibodies against IFX are highly drug-specific and do not cross-react with ADL.11,22,23,27 Therefore, switching from IFX to ADL may at first glance be considered unproblematic even in the presence of anti-IFX Abs. This study shows, however, that switchers with previous anti-IFX Ab development more often form antibodies against ADL than those without.
Anti-ADL Abs detected in this study seemed to be functional because they associated to diminished ADL detection and with secondary ADL treatment failure with a high positive predictive value of 0.91. Others have reported similarly.16,17,28 The fact that patients who previously formed antibodies against IFX are more prone to develop antibodies against ADL, and thus have a higher risk of therapeutic failure, provides novel insights of clinical relevance in case of changing anti-TNF agent. Accordingly, we recommend switchers previously tested positive for anti-IFX Abs to be monitored closely for development of antibodies to the new drug. This information should support rational decisions on clinical interventions in case of inadequate responses based on underlying mechanisms.29 Indeed, routine intensification of the ADL regimen at treatment failure, as is generally recommended in guidelines, is inappropriate in the presence of anti-ADL Abs because the drug is neutralized and rapidly cleared resulting in inadequate TNF-inhibition, lack of clinical effect, and reduced cost-effectiveness. In this situation, ADL therapy should be discontinued and alternative strategies explored, e.g., optimization of conventional immunosuppressive agents, switching to a third TNF-inhibitor, or switching out of class to another type of biologic agent if available. The optimal strategy needs to be investigated in future studies.
ADL bioavailability, irrespectively of anti-ADL Ab status, was found to be significantly lower among patients with secondary ADL treatment failure as compared with patients with sustained clinical effect of ADL maintenance therapy. Clinically relevant ADL cutoff values for discrimination of clinical response types were determined to allow for maximal sensitivity (<15 μg/mL), maximal specificity (<0.4 μg/mL), or equal sensitivity and specificity (<6.9 μg/mL). Others have found comparable associations.16,17,28 Taken together, observations on ADL bioavailability and immunogenicity indicate that these values may be used clinically to establish underlying mechanisms for ADL treatment failure and to guide interventions in individual patients.30 Hence, we propose that secondary ADL treatment failure may, analogously to secondary failure of IFX therapy, be caused by subtherapeutic drug levels due to changes in pharmacokinetics afforded by either immune or nonimmune mechanisms, or in case of therapeutic drug levels, by pharmacodynamic issues or noninflammatory conditions resembling relapse of active disease.29,–32
Study limitations include a limited number of patients and the consequences of a retrospective cohort study design, including evaluations of anti-TNF drug bioavailability and immunogenicity at the discretion of the physician and without standardized time points of assessments about treatment regimen. The study design may also have contributed to the relatively high frequency of anti-ADL Ab development in our cohort because a notable proportion of IFX-to-ADL switchers had previously developed anti-IFX Abs. However, testings were done in comparable numbers of patients with secondary ADL treatment failure or sustained ADL response, and this suggests that assessments were not likely biased by clinical conditions. Another reason could be that the RIA assays used here for detection of drug bioavailability and immunogenicity have proven more sensitive than most other assays.33 Along this line, the cutoff values for ADL bioavailability associated with clinical outcomes cannot directly be compared with other types of assays, and these values should preferably be validated in prospective cohorts.33 Classification of clinical outcome of ADL therapy was based on the treating physician's global evaluation and supported by biochemical parameters. Although clinically validated disease activity scoring systems are generally preferred, the clinician's evaluation of response type and decision regarding continuation or change in the therapeutic regimen reflects the clinical setting and is also commonly used.16,27 This study did not report on association between IFX bioavailability and immunogenicity and clinical outcomes because this has already been reported in our cohort in separate studies.11,24,34
In conclusion, we show that patients, who have previously developed antibodies against IFX, are highly prone to develop antibodies also against ADL after switching resulting in therapeutic failure. This should raise the alertness of testing for anti-ADL Abs in anti-IFX Ab positive switchers to encircle mechanisms for inadequate efficacy instead of routinely intensifying the ADL regimen at treatment failure, and thereby aid to ensure optimal interventions and avoid inappropriate drug use.
Acknowledgments
The authors would like to thank Niels Fogh Andersen, MD, DMSc, from Department of Clinical Biochemistry, Herlev Hospital, Denmark for assistance in retrieving biochemical test results.
Author contributions: Study design: All authors; Collection of data: M. T. Frederiksen, C. Steenholdt; Analysis and interpretation of data: All authors; Drafting of manuscript: M. T. Frederiksen, C. Steenholdt; Revision of the manuscript and approval of final manuscript: All authors.
References
Author notes
Reprints: Casper Steenholdt, MD, PhD, Department of Gastroenterology, Herlev Hospital, Herlev Ringvej 75, DK-2730 Herlev, Denmark (e-mail: steenholdt@brygge.dk).
J. Brynskov has served as advisory board member for Abbvie. O. Ø. Thomsen has served as a speaker and consultant for UCB and Zealand Pharma, speaker for MSA and primary investigator for Amgen, Biogen, Novo-Nordisk, and Pfizer. K. Bendtzen has served as a speaker for Pfizer, Roche, Novo-Nordisk, Bristol-Meyers Squibb, and Biomonitor and owns stocks in Novo-Nordisk and Biomonitor. C. Steenholdt has served as speaker for MSD and Abbvie and as a consultant for MSD and Takeda Pharmaceutical Company Limited. M. T. Frederiksen and M. A. Ainsworth has no conflicts of interest to disclose.
