Antidrug Antibodies to Tumor Necrosis Factor α Inhibitors in Patients With Noninfectious Uveitis

This cohort study assesses the frequency of antidrug antibodies (ADAs) in patients with noninfectious uveitis receiving tumor necrosis factor α inhibitors and the association of ADAs with drug levels and clinical response in these patients.

Key Points

Question

What is the frequency of antidrug antibodies (ADAs) in patients with noninfectious uveitis receiving tumor necrosis factor α inhibitors (TNFis), and what is the clinical relevance of ADAs for TNFi therapy in patients with noninfectious uveitis?

Findings

In this cross-sectional study of 54 patients with noninfectious uveitis, the frequency of ADAs was 35.7% among 42 patients receiving adalimumab and the presence of ADAs was associated with lower drug levels. A threshold adalimumab drug level above 2.7 μg/mL or antibody level less than 15.2 arbitrary units/mL was associated with a favorable clinical response.

Meaning

These findings suggest that drug monitoring may be warranted among patients judged to be inadequately responding to TNFi treatment.

Abstract

Importance

Tumor necrosis factor inhibitors (TNFis) can induce antidrug antibody (ADA) formation and loss of therapeutic response. However, the utility of ADA testing and the association between ADAs and treatment response in patients with noninfectious uveitis (NIU) is not well understood.

Objective

To assess the frequency of ADAs and their association with drug levels and clinical response in patients with NIU treated with adalimumab or infliximab.

Design, Setting, and Participants

This retrospective cross-sectional study included patients diagnosed with NIU who received adalimumab or infliximab and underwent testing for serum drug level and ADAs at the National Eye Institute from September 2017 to July 2021.

Exposures

Serum drug level testing with reflex testing for ADA levels was performed.

Main Outcomes and Measures

The main outcome was the association between drug levels and ADAs, clinical response, and concurrent antimetabolite use in patients treated with TNFis for NIU.

Results

Of 54 patients included in the study, 42 received adalimumab (mean [SD] age, 43.6 [19.6] years; 25 [59.5%] female) and 12 received infliximab (mean [SD] age, 42.7 [20.4] years; 7 [58.3%] male). In the adalimumab group, mean (SD) drug level was 9.72 (6.82) μg/mL, mean (SD) ADA level was 84.2 (172.9) arbitrary units/mL, and ADA frequency was 35.7% (15 of 42 patients). Mean drug level was lower in those with ADAs compared with those without ADAs (mean [SD], 2.8 [2.6] μg/mL vs 13.6 [5.2] μg/mL; difference: 10.8 μg/mL; 95% CI, 8.3-13.2 μg/mL; P < .001). There was a higher mean drug level with concurrent antimetabolite use compared with monotherapy (mean [SD], 11.0 [7.3] μg/mL vs 6.8 [4.5] μg/mL; difference: –4.2 μg/mL; 95% CI, –8.7 to 0.2 μg/mL; P = .06). Multivariable modeling showed that a 1−arbitrary unit increase in ADAs was associated with a –0.02 μg/mL (95% CI, –0.01 to –0.34 μg/mL) difference in mean drug level (P < .001). Favorable clinical response was associated with a threshold drug level above 2.7 μg/mL or an antibody level below 15.2 μg/mL. The mean (SD) drug level in the infliximab group was 27.02 (18.15) μg/mL, and no ADAs were detected.

Conclusions and Relevance

In this study, 35.7% of adalimumab-treated patients with NIU had ADAs. The presence of ADAs was associated with lower drug levels, and higher ADA levels were associated with increased risk of TNFi treatment failure. Although limited by the retrospective design, our results suggest that therapeutic drug monitoring may be considered among patients experiencing therapy failure to help exclude ADAs as a potential cause of treatment failure.

Introduction

The tumor necrosis factor α inhibitors (TNFis) infliximab and adalimumab have been shown to be effective in treating noninfectious uveitis (NIU).^1,2,3 TNFis are typically used in cases refractory to conventional immunomodulatory therapy, in more severe presentations, or as a primary treatment for certain etiologies such as Behçet disease.⁴

Despite the effectiveness of TNFis, few studies to date have investigated antidrug antibodies (ADAs) in patients with NIU, which can result in lower circulating drug levels and potential loss of medication efficacy.^1,5,6,7 Thus, the goal of this study was to investigate the frequency of ADAs and their association with circulating drug levels, clinical response, and concurrent treatment with antimetabolites in patients receiving adalimumab or infliximab for NIU.

Methods

This single-site, retrospective cross-sectional study was conducted at the National Eye Institute in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Patients with a diagnosis of NIU treated with either adalimumab or infliximab who underwent ADA testing from September 2017 to July 2021 were included. This study was approved by the National Institutes of Health (NIH) Intramural Institutional Review Board. This was a retrospective medical record review; thus, informed consent was not required.

Demographic data including age and sex (determined by self-report) were obtained by medical record review. Diagnosis, disease duration, duration of therapy, frequency of drug administration, concurrent oral steroid treatment, coexisting antimetabolite treatment, clinical response to therapy, reason for laboratory examination (routine vs therapy failure), drug level at time of laboratory examination, and ADA levels were documented.

Anti-adalimumab and anti-infliximab antibody levels were tested in serum samples using an enzyme-linked immunosorbent assay–based platform available through Mayo Clinic Laboratories. The test provides the drug level, and reflexes to ADA testing drug levels are below a specific threshold (8.0 μg/mL for adalimumab and 1.0 μg/mL for infliximab). The limits of detection for ADA levels were less than 10 arbitrary units (AU)/mL for the adalimumab assay and less than 20 AU/mL for the infliximab assay. Testing was done in response to failure of therapy (reactive testing) or on a routine basis (proactive testing). To minimize variability in drug level due to timing of TNFi administration, only patients who had laboratory examination done at the trough period of their treatment cycle were included. Trough was defined as a level obtained within 48 hours of a scheduled dose in patients receiving adalimumab and within 1 week of the scheduled infusion in patients receiving infliximab.

Clinical response was categorized into complete response, partial response, and nonresponse and was based on medical record review of clinical examination findings and multimodal imaging. A complete response was defined as resolution of anterior chamber or vitreous cell, vitreous haze, and/or angiographic leakage after initiation of TNFi therapy. Partial responders had improvement in disease without complete resolution. Nonresponders had no improvement or worsening of disease activity after initiation of TNFi therapy.

Statistical Analysis

Statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc). Comparisons of drug and ADA levels between groups were performed using the nonparametric Kruskal-Wallis test. The threshold P value for significance was adjusted to ≤.017 (Bonferroni correction) to account for multiple pairwise comparisons among groups. Two-sided P values were used but not adjusted for multiple comparisons given the exploratory nature of the study. For group comparisons, the 95% CI for the difference between groups was computed by using standard methods for 2-sample t tests (P values for comparisons were based the nonparametric Kruskal-Wallis test). Multivariable analyses for patients receiving adalimumab were done using parametric analysis of variance models to assess the association of ADA with mean drug level. A binary logistic regression model was used to generate receiver operating characteristic (ROC) curves and optimum J values that corresponded to threshold mean drug and ADA levels associated with a partial and/or complete response to adalimumab therapy. Comparisons of area under the ROC curve were done by using the SAS roccontrast statement in Proc Logistic.

Results

Of 54 patients included in the analysis, 42 received adalimumab and 12 received infliximab (Table 1). In the adalimumab cohort, 25 patients (59.5%) were female and 17 were male (40.5%). The mean (SD) age was 43.6 (19.6) years (range, 10-85 years). The mean (SD) time from diagnosis to laboratory examination was 7.0 (4.4) years (range, 1.2-20.3 years). The mean (SD) time from therapy initiation to laboratory examination was 2.5 (2.5) years (range, 0.2-9.8 years). For the 14 patients who had reactive testing for suspicion of therapy failure, mean (SD) time of therapy initiation to laboratory examination was shorter compared with that in those who were routinely tested (1.7 [1.6] years vs 3.0 [2.8] years; difference: –1.3 years; 95% CI, –2.9 to 0.4 years). A total of 33 patients (79%) had bilateral disease. Furthermore, 29 (69.0%) and 13 (31.0%) patients were treated with weekly and biweekly doses of adalimumab, respectively. Specific disease etiologies among included patients are listed in Table 2.

Table 1. Demographic Information for the Study Sample, Prevalence of Antidrug Antibodies, and Mean Drug and Antibody Levels.

Characteristic	Adalimumab (n = 42)	Infliximab (n = 12)	Total (N = 54)
Age, mean (SD), y	43.6 (19.6)	42.7 (20.4)	43.5 (19.6)
Duration of therapy, mean (SD), y	2.6 (2.5)	1.7 (1.4)	2.3 (2.3)
Time from diagnosis to laboratory examination, mean (SD), y	7.0 (4.4)	7.7 (4.8)	7.2 (4.5)
Bilateral disease, No. (%)	33 (78.6)	11 (91.7)	44 (81.5)
Sex, No. (%)
Female	25 (59.5)	5 (41.7)	30 (55.6)
Male	17 (40.5)	7 (58.3)	24 (44.4)
Antidrug antibodies present, No. (%)	15 (35.7)	0	15 (27.8)
Drug level, mean (SD), μg/mL	9.72 (6.82)	27.02 (18.15)	NA
Antibody level, mean (SD), AU/mL	84.2 (172.9)	NA	NA

Abbreviations: AU, arbitrary units; NA, not applicable.

Table 2. Diagnoses in Patients in the Adalimumab and Infliximab Cohorts.

Diagnosis	Patients, No. (%)
	Adalimumab (n = 42)	Infliximab (n = 12)
Ampiginous choroidopathy	1 (2.38)	0
Anterior scleritis	2 (4.76)	0
Behçet disease	1 (2.38)	0
Birdshot chorioretinopathy	7 (16.67)	1 (8.33)
Cogan syndrome–associated anterior or intermediate uveitis	1 (2.38)	0
Idiopathic uveitis	15 (35.7)	8 (66.67)
Anterior	1 (2.38)	0
Anterior and intermediate	4 (9.52)	2 (16.67)
Intermediate	5 (11.9)	1 (8.33)
Posterior and panuveitis	5 (11.9)	5 (41.67)
Juvenile idiopathic arthritis–associated uveitis	4 (9.52)	0
Lyme disease–associated uveitis	0	1 (8.33)
Posterior scleritis	1 (2.38)	0
Punctate inner choroidopathy	4 (9.52)	0
Serpiginous choroidopathy	0	1 (8.33)
Sympathetic ophthalmia	1 (2.38)	0
Sarcoid-associated uveitis	2 (4.76)	1 (8.33)
Vogt-Koyanagi-Harada disease	3 (7.14)	0

In the infliximab cohort (n = 12; 11 patients routinely tested, 1 reactively tested), 5 patients (41.7%) were female and 7 (58.3%) were male; the mean (SD) age was 42.7 (20.4) years (range, 15-73 years). The mean (SD) time from diagnosis to laboratory examination was 7.7 (4.8) years (range, 0.9-17.0 years), and mean (SD) time of therapy initiation to laboratory examination was 1.7 (1.4) years (range, 0.4-5.4 years). Bilateral disease was present in 11 patients (91.7%) (Table 1). The majority had a diagnosis of idiopathic uveitis. The mean (SD) drug level was 27.02 (18.15) μg/mL. Of 12 patients, 9 (75.0%) were complete responders, 2 (16.7%) were partial responders, and 1 (8.3%) was a nonresponder. No patients receiving infliximab were receiving concurrent oral steroids, and none developed ADAs.

In the adalimumab cohort, the mean (SD) drug level was 9.72 (6.82) μg/mL and mean (SD) antibody level was 84.2 (172.9) AU/mL; ADAs were present in 15 of 42 patients (35.7%) (Table 1). Notably, mean drug level was lower in patients with ADAs than in those without ADAs (mean [SD], 2.8 [2.6] vs 13.6 [5.2] μg/mL; difference: 10.8 μg/mL; 95% CI, 8.3-13.2 μg/mL; P < .001). When stratified by clinical response, mean drug level remained lower in patients with ADAs present compared with those without ADAs (Table 3); this was most apparent in complete responders (mean [SD], 4.1 [2.8] vs 13.4 [5.1] μg/mL; difference: 9.3 μg/mL; 95% CI, 4.3-14.4 μg/mL; P = .002) and nonresponders (mean [SD], 2.2 [2.9] vs 12.2 [5.5] μg/mL; difference: 10.0 μg/mL; 95% CI, 4.2-15.8 μg/mL; P = .006).

Table 3. Adalimumab Level Based on Presence or Absence of Antidrug Antibodies Stratified by Clinical Response.

Clinical response	Adalimumab level, mean (SD), μg/mL				Difference (95% CI)	P value
	Antibodies not present	P value	Antibodies present	P value
Complete response	13.4 (5.1)	.26	4.1 (2.8)	.34	9.3 (4.3-14.4)	.002
Partial response	17.2 (5.0)		2.1 (0.8)		15.1 (3.0-27.1)	.05
Nonresponse	12.2 (5.5)		2.2 (2.9)		10.0 (4.2-15.8)	.006

To investigate the association of clinical response and mean drug level, the adalimumab cohort was stratified based on clinical response (Table 4). A total of 23 of 42 patients (54.8%) were complete responders, 6 of 42 (14.3%) were partial responders, and 13 of 42 (30.9%) were nonresponders. The mean (SD) drug level was 11.4 (6.1) μg/mL (95% CI, 8.7-14.0 μg/mL) in complete responders, 9.7 (8.9) μg/ML (95% CI, 0.4-19.0 μg/mL) in partial responders, and 6.8 (6.6) μg/mL (95% CI, 2.8-10.8 μg/mL) in nonresponders (P = .13). Furthermore, ADAs were present in a higher proportion of partial responders (3 of 6 [50.0%]) and nonresponders (7 of 13 [53.8%]) compared with complete responders (5 of 23 [21.7%]).

Table 4. Mean Drug and Antibody Levels in All Adalimumab-Treated Patients Stratified by Clinical Response.

	Clinical response			Difference, mean (95% CI)
	Complete response	Partial response	Nonresponse	Complete vs partial	P valuea	Complete vs nonresponse	P valuea	Partial vs nonresponse	P valuea
Patients, No./total No. (%)	23/42 (54.8)	6/42 (14.3)	13/42 (30.9)	NA	NA	NA	NA	NA	NA
Adalimumab level, mean (SD) [95% CI], μg/mL	11.4 (6.1) [8.7-14.0]	9.7 (8.9) [0.4-19.0]	6.8 (6.6) [2.8-10.8]	1.7 (–4.6 to 8.0)	.69	4.6 (0.1 to 9.0)	.04	2.9 (–10.5 to 4.8)	.38
Antibody level, mean (SD), AU/mL	16.4 (49.1)	168.4 (223.4)	165.5 (239.6)	–152.0 (–386.2 to 82.2)	.15	–149.1 (–294.8 to –3.3)	.02	2.9 (–247.6 to 241.7)	.85

Abbreviations: AU, arbitrary units; NA, not applicable.

^a P values are based on Wilcoxon 2-sample test, a nonparametric alternative to the 2-sample t test.

Ten patients receiving adalimumab (23.8%) were receiving oral prednisone at the time of their laboratory examination. Of these 10 patients, 6 were complete responders who were tested on a routine basis (mean [SD] dose, 6.7 [2.6] mg/d; range, 2.5-10.0 mg/d), 3 were nonresponders (mean [SD] dose, 16.7 [11.6] mg/d; range, 10-30 mg/d), and 1 was a partial responder (5-mg/d dose). Mean drug level was similar between those receiving oral steroids and those not receiving oral steroids (mean [SD], 10.3 [6.8] μg/mL vs 9.6 [6.9] μg/mL; difference: –0.70 μg/mL; 95% CI, –5.76 to 4.35; P = .95).

We subsequently examined the differences in mean drug level and mean antibody level between patients using adalimumab weekly compared with every 2 weeks. Patients receiving weekly adalimumab (13 of 42 [31.0%]) had a mean (SD) drug level of 9.60 (7.85) μg/mL and a mean (SD) antibody level of 35.95 (66.69) AU/mL compared with 9.78 (6.46) μg/mL and 105.9 (200.7) AU/mL, respectively, for those on a biweekly schedule (difference in mean drug level, –0.18 μg/mL [95% CI, –4.8 to 4.5 μg/mL]; P = .82; difference in mean antibody level, –70.0 AU/mL [95% CI, –154.2 to 14.3 AU/mL]; P = .41).

There were 14 patients (13 nonresponders, 1 partial responder) in the adalimumab cohort who had drug levels checked based on suspicion of therapy failure. The mean (SD) duration of adalimumab therapy in patients who experienced treatment failure was 1.7 (1.6) years; 8 of 14 (57.1%) had ADA, 6 of 14 (42.9%) received weekly treatment, and 11 of 14 (78.6%) were receiving a concurrent antimetabolite. Interestingly, 5 of 6 patients (83.3%) who experienced therapy failure without ADA formation were receiving a concurrent antimetabolite, with drug levels ranging from 11.1 to 18.0 μg/mL (mean [SD], 14.2 [3.1] μg/mL).

The association of concurrent antimetabolite use with mean drug level and mean antibody level in the adalimumab cohort was analyzed (eTable in Supplement 1). Of 42 patients, 29 (69.0%) were receiving concurrent treatment with either mycophenolate mofetil (MMF) (15 [51.7%]), methotrexate (MTX) (13 [44.8%]), or azathioprine (1 [3.5%]). When comparing mean drug level and mean antibody level in patients with concurrent antimetabolite use (n = 29) with those receiving adalimumab monotherapy (n = 13), mean (SD) drug level was 11.0 (7.3) μg/mL in the antimetabolite group compared with 6.8 (4.5) μg/mL in the monotherapy group (difference: 4.2 μg/mL; 95% CI, –8.7 to 0.20 μg/mL; P = .06). Mean (SD) antibody level was lower in the antimetabolite group compared with the monotherapy group (61.3 [157.0] vs 135.5 [201.3] AU/mL; difference, 74.2 AU/mL; 95% CI, –41.4 to 189.9 AU/mL; P = .09).

The concurrent antimetabolite group was then stratified into those using MMF or MTX and compared with the adalimumab monotherapy group. The rate of ADA formation was 20.0% (3 of 15 patients) in the MMF group, 38.5% (5 of 13) in the MTX group, and 53.8% (7 of 13) in the adalimumab monotherapy group (eTable in Supplement 1). Mean drug level was higher in the MMF group compared with the adalimumab monotherapy group (mean [SD], 12.6 [7.4] vs 6.8 [4.5] μg/mL; difference: –5.8 μg/mL; 95% CI, –10.5 to –1.1 μg/mL; P = .03). Mean antibody level was lower in the MMF group compared with the adalimumab monotherapy group (mean [SD], 10.4 [30.6] vs 135.5 [201.3] AU/mL; difference: 125.1 AU/mL; 95% CI, 2.8 to 247.4 AU/mL; P = .03). In the MTX group, there was no difference between the MTX and monotherapy group in mean (SD) drug level (9.0 [7.3] vs 6.8 [4.5] μg/mL) or in mean (SD) antibody level (124.7 [220.2] vs 135.5 [201.3] AU/mL).

Multivariable models showed that the presence of ADAs in patients receiving adalimumab was associated with a mean difference in mean drug level of –11.0 μg/mL (95% CI, –14.0 to –8.06 μg/mL; P < .001). Furthermore, for each 1-AU/mL increase in ADA level, there was a mean difference of –0.02 μg/mL (95% CI, –0.01 to –0.34 μg/mL; P < .001) in circulating drug level. When adjusting for the presence of ADAs, multivariable modeling showed no significant difference in drug levels between the response groups, with a mean difference in drug level of 1.11 μg/mL (95% CI, –2.21 to 4.42 μg/mL; P = .50) between complete responders and nonresponders and 2.32 μg/mL (95% CI, –2.19 to 6.84 μg/mL; P = .30) between partial responders and nonresponders, thus suggesting an association of ADA with decreased mean drug level.

Analysis of an ROC curve (Figure) in the adalimumab group demonstrated that a drug level above the threshold of 2.7 μg/mL was associated with any positive (partial or complete) response to therapy, with a sensitivity of 90% (95% CI, 73%-100%) and specificity of 46% (95% CI, 26%-64%) and an area under the curve (AUC) of 0.69. A drug level above 3.3 μg/mL was associated with a complete therapeutic response, with 96% (95% CI, 86%-100%) sensitivity and 47% (95% CI, 27%-68%) specificity and an AUC of 0.67. Conversely, a threshold antibody level less than 15.2 AU/mL was associated with any positive response, with a sensitivity of 79% (95% CI, 57%-100%) and specificity of 54% (95% CI, 36%-72%) and an AUC of 0.67, and the same antibody value was associated with a complete response, with 86.9% (95% CI, 72%-100%) sensitivity and 53% (95% CI, 32%-73%) specificity and an AUC of 0.69.

Figure. Receiver Operating Characteristic Curves for Drug and Antibody Levels for Complete or Partial Clinical Response.

AUC indicates area under the curve.

Discussion

Our results suggest that the presence of ADAs is associated with lower mean drug level and that higher ADA levels may be associated with increased risk of TNFi treatment failure. The rate of ADA formation is an important therapeutic consideration when using TNFis in patients with NIU. Data regarding ADAs is mostly derived from rheumatologic disease, and to date, there remains a paucity of data regarding the frequency and implications of ADAs following TNFi treatment for NIU. An enduring ADA response depends on the formation of IgG antibodies that primarily target the antigen binding sites of monoclonal antibodies and are thought to be neutralizing.⁸ Antidrug antibodies in rheumatological diseases are reported to develop anywhere from 2 weeks to years after therapy initiation.⁹

One study¹⁰ examining the incidence of ADA formation in patients with juvenile idiopathic arthritis after adalimumab initiation showed ADA formation as early as 2 months and a peak frequency of 37% at 6 months. The authors described female gender, higher disease activity, and concurrent leflunomide use as associated with higher rates of ADA formation, while MTX use was associated with lower rates of ADA formation. Additionally, a meta-analysis of 14 651 patients with systemic autoimmune disease (rheumatoid arthritis, spondylarthritis, and inflammatory bowel disease) showed a cumulative ADA incidence of 12.7% across all TNFis.⁹ In general, clinical factors that may be associated with ADA formation include higher disease activity, longer disease duration, concomitant infection, intramuscular route of administration, missed doses, and lower doses; specific human leukocyte antigen alleles have also been implicated.^{8,9,11,12,13,14} In our study, the adalimumab group demonstrated an ADA formation rate of 35.7%, which was in line with the limited data on NIU in which reported rates of ADA formation for adalimumab ranged from 2.7% and 5% in the VISUAL I¹ and VISUAL II² trials, respectively, to 12.5% and 45% in smaller prospective studies and retrospective case series.^5,6,7 While the absence of anti-infliximab antibodies in our study is unusual, these results were likely due to our small sample size.

Although data remain limited for NIU, ADA formation has been associated with decreased circulating drug levels and loss of clinical response to TNFi treatment.^5,6,7 In our study, there were higher rates of ADAs in patients receiving adalimumab tested due to therapy failure compared with those tested for routine reasons. In the proactive group, ADAs were noted in 7 patients, with 5 of 7 showing a complete response to therapy and 2 of 7 showing a partial response. In the reactive group, 5 of 6 patients who experienced therapy failure with no ADAs had mean drug levels ranging from 11.1 to 18.0 μg/mL, suggesting that treatment failure may be due to other reasons such as the specific cytokine pathways affected in these patients. Additionally, clinical factors affecting TNFi pharmacokinetics, including higher baseline TNF-α levels, low albumin levels, high baseline C-reactive protein level, increased body mass index, and male sex, have all been implicated as having an association with diminished or variable efficacy.¹⁵ These non–ADA-mediated factors may also explain the similar mean drug level but lower mean antibody level among patients receiving adalimumab weekly compared with those receiving adalimumab biweekly, since those with altered TNFi pharmacokinetics may require more frequent doses.

In the adalimumab cohort, presence of ADA was associated with a reduction in mean drug level compared with no ADAs. When clinical response was compared between those with ADAs and those without ADAs, the presence of ADAs was associated with a lower mean drug level in complete responders and nonresponders, suggesting that ADAs may be associated with reduced mean drug level even in those with a response to adalimumab.

Multivariate modeling quantified the effects of ADAs. ADA formation was associated with a mean drug level reduction of 11.0 μg/mL. Additionally, for each increase of 1 AU in antibody level there was a mean decrease of 0.02 μg/mL in mean drug level, suggesting an antibody level–dependent inverse association between ADAs and drug levels. This association was in accord with reports from the rheumatology literature in which anti-adalimumab antibodies in patients with rheumatoid arthritis were associated with a lower drug level than in those without antibodies (1.2 mg/L vs 11.0 mg/L), and presence of ADAs was associated with a 68% reduction in drug response rate in a meta-analysis.^16,17

Concurrent antimetabolite use, especially MTX, has been associated with lower rates of ADA formation in patients with systemic rheumatologic disease.^{9,17,18,19,20} However, the data to support concomitant antimetabolite use to reduce ADA formation in patients with NIU has been limited to a few studies. Skrabl-Baumgartner et al⁵ showed that patients with juvenile idiopathic arthritis treated with adalimumab who did not form ADAs were more likely to be receiving a concurrent antimetabolite; in contrast, Cordero-Coma et al⁶ did not find an association of a concomitant immunomodulatory therapy agent with lack of ADA formation. When patients with concurrent antimetabolite use were analyzed in this study, there was a mild increase in mean drug level compared with the monotherapy group.

There are no established guidelines for the role of therapeutic drug monitoring (TDM) in patients receiving TNFis for NIU, although several studies showed that it may be beneficial. Sejournet et al,²¹ in their study on TDM in patients with chronic NIU, reported an ADA frequency of 13.6% among patients treated with adalimumab. Their results also showed that ADA formation and low drug levels together with clinical evidence of therapy failure were associated with increased adalimumab dose frequency in 13 of 31 patients (42%), an increased dose in 1 of 31 patients (3.2%), and intraclass or interclass switching in 10 of 31 patients (32%), suggesting that TDM may be useful in determining the need for treatment. Additionally, TDM for rheumatologic disease and inflammatory bowel disease is associated with lower cost of treatment per year and facilitation of treatment optimization.^22,23,24,25 The American Gastroenterological Association recommends reactive TDM for patients with active inflammatory bowel disease receiving maintenance TNFi therapy, with a target trough level of 7.5 μg/mL or greater for adalimumab and 5.0 μg/mL or greater for infliximab.²⁶ In active disease when drug levels are below this threshold, dose escalation can be considered; if drug levels are above the threshold, class switching can be considered. In our study, ROC analysis found that a threshold drug level value for adalimumab of 2.7 μg/mL was associated with a partial or complete response to therapy and 3.3 μg/mL with a complete therapeutic response.

Limitations

This study is limited due to its retrospective design and sample size. The retrospective nature of the study meant that there were variable intervals (even with routine testing) of TNFi initiation to ADA testing, as well as some selection bias with the inclusion of patients experiencing therapy failure. Additionally, ADA prevalence could have been underreported due to therapy failure causing drug cessation before testing. Despite these limitations, this study provides one of the largest sample sizes in analyzing the association of ADAs with TNFi therapy in patients with NIU.

Conclusions

This study highlights the importance of recognizing the formation of ADAs and its association with TNFi therapy in patients with NIU. The presence of ADAs was associated with lower drug levels; however, their presence was not always associated with therapy failure. Results from multivariable models suggest that drug level is associated not only with the presence of ADAs but also with the quantity of ADAs. As a result, patients with sufficiently high ADA levels may be expected to have essentially no effective circulating drug and loss of response to TNFi therapy. These findings support consideration of drug and antibody level testing in patients receiving adalimumab who are experiencing therapy failure. Failure due to ADA formation is suggested if serum drug level is low or undetectable and ADA levels are elevated, in which case intraclass or interclass switching can be considered. Further prospective studies are needed to clarify the role of TDM and the association of ADAs with TNFi treatment in patients with NIU.

Supplement 1.

eTable. Mean adalimumab drug level based on presence or absence of concurrent antimetabolite therapy

Supplement 2.

Data Sharing Statement