Favorable Outcomes and Anti-TNF Durability After Addition of an Immunomodulator for Anti-Drug Antibodies in Pediatric IBD Patients

Abstract

Background

Anti-drug antibodies (ADAs) to anti-tumor necrosis factor alpha (anti-TNF) drugs are associated with increased drug clearance and loss of response. We aimed to assess the effectiveness of starting an immunomodulator (IM) drug in patients with newly detected ADAs on anti-TNF monotherapy.

Methods

We reviewed the medical records of pediatric patients with inflammatory bowel disease on infliximab or adalimumab monotherapy with first-time detection of significant ADAs between 2014 and 2018. Patients who started an IM within 3 months of ADA detection were compared with those who did not (No-IM). Outcomes included steroid-free clinical and biochemical remission on the same anti-TNF , anti-TNF durability, and ADA reversal.

Results

We identified 89 patients with ADAs: 30 IM patients and 59 No-IM patients. The initial anti-TNF was stopped shortly after ADA detection in 36% of the No-IM patients vs none of the IM patients, driving longer survival on the initial anti-TNF in the IM group (P = 0.005). At 12 months, steroid-free clinical and biochemical remission on the same anti-TNF occurred in 53.9% of the IM group vs 26.8% in the No-IM group (P = 0.025). Drug levels rose higher (P = 0.003) and ADA levels fell farther (P = 0.037) in the IM group than in the No-IM group. Baseline ADA level predicted ADA reversal in the No-IM patients with an area under the receiver operating characteristic of 0.79 (P = 0.006). An ADA level <329 ng/mL had a 76.2% sensitivity and an 83.3% specificity for ADA reversal without IM.

Conclusions

Pediatric patients with inflammatory bowel disease on anti-TNF monotherapy who started an IM for significant ADA levels exhibited longer anti-TNF durability and a higher likelihood of steroid-free clinical and biochemical remission on the same anti-TNF. Patients not treated with an IM were unlikely to reverse ADAs >329 ng/mL.

INTRODUCTION

Over the past 20 years, tumor necrosis factor alpha antagonists (anti-TNFs), such as infliximab (IFX) and adalimumab (ADL), have revolutionized medical treatment paradigms for inflammatory bowel diseases (IBD).^1-3 In pediatric IBD patients, anti-TNFs induce clinical remission, restore growth, improve bone mineralization, and prevent penetrating Crohn disease complications.^2-5

Although many patients experience an excellent response to anti-TNFs initially, a frustratingly high percentage later exhibit loss of response (LoR). An LoR is often associated with the development of anti-drug antibodies (ADAs), which accelerate drug clearance and lead to low trough levels (or vice versa).^6-8 The reported incidence of the development of ADAs to anti-TNFs, also known as immunogenicity, is as high as 61%.⁹ In addition to LoR, immunogenicity is also associated with an increased risk of infusion reactions.⁹

Recognition of immunogenicity to anti-TNFs may trigger early switching to another agent. This may become problematic, as the decision to switch agents may prevent the reutilization of that same agent or anti-TNF class, thus limiting the therapeutic options left in the armamentarium for a patient in the future. Dose intensification strategies often reverse immunogenicity; however, dose escalation alone is often ineffective in patients with high-titer ADA.^{10, 11}

The addition of an immunomodulator (IM) to anti-TNF monotherapy is a proposed alternative approach to reverse immunogenicity. Limited data from small studies with adult-onset IBD suggest that the addition of a thiopurine or methotrexate can reverse ADAs and increase drug levels, resulting in the restoration of clinical response.^{12, 13} However, it is not known whether the addition of an IM in the setting of ADA improves clinical outcomes, reduces the level (concentration) of immunogenicity, or prolongs drug durability compared with the decision not to add an IM (No-IM).

The aim of this study was to compare clinical outcomes and anti-TNF durability between patients on anti-TNFs with new significant ADAs who were started on an IM with patients in a No-IM group. An additional aim was to identify an ADA concentration threshold in No-IM patients above which ADA reversal was unlikely.

METHODS

Study Design

We performed a single-center retrospective cohort study of pediatric IBD patients receiving anti-TNF monotherapy (IFX or ADL) in whom ADA was detected for the first time between May 2014 and May 2018. This study was approved by the institutional review board at Cincinnati Children’s Hospital Medical Center.

We compared patients who were started on an IM after high ADA detection with patients who were not (No-IM). The IM group was defined as patients who started an IM within the first 3 months of detection of likely significant ADAs (see ADA thresholds below in anti-TNF Assay section) and remained on both treatments (an anti-TNF and an IM) for at least 1 month. The No-IM group was defined as patients who did not start an IM within the first year of ADA detection.

Patients were excluded if they were on combination therapy when the ADA level was detected, if they were not on the anti-TNF therapy for which the ADA level was detected, if they started an IM between 3 and 12 months after the ADA was detected, or if an ADA level was >3000 ng/mL).

Data Extraction

Patients were identified by extracting ADA levels from the electronic health records. A chart review documentation form with predesignated criteria for enrollment and data collection was utilized. Chart review was performed electronically, and data were extracted into Research Electronic Data Capture tools hosted at Cincinnati Children’s Hospital Medical Center.^{14, 15} Baseline and management data collected included sex, diagnosis, physician global assessment (PGA) at a corresponding clinic visit, Paris disease classification, steroid exposure, current and previous anti-TNF exposure, initial anti-TNF dose, and clinical management after ADA detection including drug discontinuation, dose intensification, and whether IM was started.

Anti-TNF Assay and Proactive Therapeutic Drug Monitoring Management

We measured IFX levels, ADL levels, and ADAs using the electrochemiluminescence-based immunoassay (ECLIA) method by Esoterix (LabCorp, Calabasas Hills, CA). Although these assays detect antibodies to IFX (ATI) ≥22 ng/mL and antibodies to ADL ≥25 ng/mL, for the purposes of this study significantly elevated ADAs were designated as ≥200 ng/mL for IFX or ≥ 100 ng/mL for ADL. And although there are no published guidelines that exist for these cutoffs, ADAs below these cutoffs have minimal influence on the mean drug level per the assay manufacturer.^{16, 17} In addition, a small prospective study showed that ATI levels <200 ng/mL were more likely to achieve mucosal healing with IFX dose-intensification than levels >200 ng/mL.¹¹ These cutoffs of an IFX level ≥200 ng/mL and an ADL level ≥100 ng/mL also reflect the thresholds of immunogenicity reversal as recommended by our internal medical center’s proactive therapeutic drug monitoring guidelines since 2014.⁸

Outcomes

Outcomes were assessed between the IM and No-IM groups. Primary outcomes included a composite measure of steroid-free clinical remission and biochemical remission (SFCBR; C-reactive protein level <0.5 mg/dL) for patients who remained on the same anti-TNF agent (SFCBR-SD) and SFCBR regardless of anti-TNF maintenance at 6 months and 12 months. Clinical remission was defined by a quiescent score on the PGA, which was measured by the previously published ImproveCareNow Physician Global Assessment, a 4-point categorical scale that divides clinical disease status into quiescent, mild, moderate, or severe disease.¹⁸ A secondary outcome was anti-TNF durability, which was defined as overall survival on the initial anti-TNF agent for which ADAs were detected until the day of chart abstraction (up to 4 years). A subanalysis was performed among those whose providers had an intent to keep them on the initial anti-TNF (at least 1 anti-TNF dose after ADA detection) and who were dose-optimized if the anti-TNF drug concentration was <5 µg/mL. Finally, we assessed the change in anti-TNF drug levels and change in ADA levels within 1 year between the IM and No-IM groups with the repeat level occurring >3 months from initial ADA discovery.

Statistical Analysis

Data were assessed for normality and baseline descriptive statistics were assessed between the IM and No-IM groups. Baseline and outcome categorical variables were compared using the × ² test or Fisher exact test, and continuous variables were compared with the Mann-Whitney U test. A Kaplan-Meier survival curve with a log-rank test was used to evaluate the durability of the initial anti-TNF agent. To assess potential confounders of the association of IM management with outcomes, univariable logistic regression was performed for each baseline variable that differed between groups with a type 1 error rate <0.1. Then, to determine the independent association of IM management with outcomes, variables associated with outcomes on univariable analysis with a type 1 error rate <0.1 were entered into a multivariable logistic regression model. The logistic regression model was limited to 3 variables to achieve a medium effect size with our available sample size.¹⁹

Changes between baseline and follow-up anti-TNF drug and ADA levels were analyzed by the Wilcoxon matched-pairs signed rank test. The Kruskall-Wallis test was used to compare baseline ADA levels between patients who did and did not ultimately reverse ADA without the addition of an IM. A receiver operating characteristic (ROC) curve analysis with a Youden J statistic was calculated to estimate the optimal ADA level cutoff point under which ADAs could be reversed without the addition of an IM.

Calculations were performed with Graphpad Prism version 8.2.1 for Windows (Graphpad Software, San Diego, CA). Logistic regression was performed with IBM SPSS Statistics version 20.0 (IBM, Armonk, NY). A P value <0.05 was determined to be statistically significant.

Ethical Considerations

The institutional review board determined that consent or assent would be waived because of the nature of the retrospective study.

RESULTS

Baseline Characteristics and Initial Post-ADA Management

We identified 320 pediatric IBD patients with detectable ADA levels between May 2014 and May 2018. Two hundred thirty-one patients were excluded, including 193 with low-level ADAs (ATI <200 ng/mL or antibodies to ADL <100 ng/mL), 6 who were not on the drug for which they had a high ADA level, 5 who had an ADA level >3000 ng/mL, and 10 who started an IM >3 months after ADA detection (Supplementary Fig. 1).

Eighty-nine patients met the inclusion criteria, including 30 IM and 59 No-IM patients. Of the 89 patients included in the final analysis, the median age at the time of ADA detection was 16.1 years (interquartile range [IQR] = 13.2-17.6), with 70 (79%) diagnosed with Crohn disease and 66 of the 89 (74%) receiving maintenance IFX. Interestingly, 61 of the 89 (69%) were PGA-quiescent at their previous clinic visit. Comparison of baseline characteristics and initial post-ADA management between the IM and No-IM patients is detailed in Table 1. Baseline characteristics between the IM and No-IM group, including drug and ADA levels, were similar, with the exception that patients who were prescribed an IM had longer exposure to the anti-TNF agent. Of the 30 IM patients, 28 (93%) received methotrexate (median dose 8.8 mg/m² (IQR = 7.8-14.0), 89% (25/28) with oral administration. The other 2 patients were started on 6-mercaptopurine.

TABLE 1.

Baseline Characteristics and Management at the Time of High ADA

	No-IM	IM	P
	n = 59	n = 30
Age, y	16.4 (13.6-17.7)	15.1 (12.2-17.3)	0.333
Sex			0.071
Female	31 (52.5)	9 (30.0)
Male	28 (47.5)	21 (70.0)
Diagnosis			0.939
CD	47 (79.7)	23 (76.7)
UC	10 (16.9)	6 (20.0)
IBD-U	2 (3.4)	1 (3.3)
CD location			0.267
L1 ileocecal	7 (14.9)	3 (12.5)
L2 colonic	8 (17.0)	1 (4.2)
L3 ileocolonic	32 (68.1)	20 (83.3)
UC extent			1.000
E1/E2 proctitis/left-sided	0 (0.0)	0 (0.0)
E3/E4 extensive/pancolitis	10 (100.0)	6 (100.0)
PGA at time of ADA			0.696
Quiescent	39 (66.1)	22 (73.3)
Mild	13 (22.0)	7 (23.3)
Moderate	5 (8.5)	1 (3.3)
Severe	1 (1.7)	0 (0.0)
Time since diagnosis, y	2 (1.0-5.9)	2.6 (1.0-3.7)	0.799
Time since anti-TNF start (wks)	29.7 (14.6-59.4)	50.4 (27.6-101.1)	0.038
Laboratory
Albumin (g/dL)	3.6 (3.4-3.9)	3.7 (3.5-3.9)	0.560
CRP (mg/dL)	0.48 (0.29-2.11)	0.29 (0.29-0.85)	0.279
Hgb (g/dL)	12.8 (12.0-14.2)	13.6 (13-14.35)	0.110
Anti-TNF drug level (µg/mL)	1.0 (0.4-3.9)	0.65 (0.4-3.48)	0.508
ADA level (ng/mL)	352 (259-759)	492 (368-835)	0.104
Drug
ADL	18 (30.5)	5 (16.7)	0.204
IFX	41 (69.5)	25 (83.8)
On high-dosage anti-TNF*	31 (52.5)	17 (56.7)	0.823
Exposure to >1 anti-TNF	10 (16.9)	5 (16.7)	1.000
On corticosteroids	9 (15.3)	1 (3.3)	0.155
Management after high ADA detection
Immunomodulator started
6-mercaptopurine	0 (0.0)	2 (6.7)
Methotrexate	0 (0.0)	28 (93.3)
Initial anti-TNF management after ADA detection			< 0.001
Anti-TNF continued; dose optimized	28 (47.5)	25 (83.3)
Anti-TNF continued; no change in dose	12 (20.3)	5 (16.7)
Dose previously optimized	9	1
Therapeutic drug level †	3	4
Anti-TNF discontinued	19 (32.2)	0 (0.0)
Cycle to another anti-TNF drug	14
Biologic class switch	2
Switch to nonbiologic therapy	3

Data reported as n (%) or median (IQR).

*IFX >5 mg/kg every 8 weeks or ADL >40 mg/kg every 2 weeks.

^†Therapeutic drug level >5 µg/mL.

CD, Crohn disease; CRP,C-reactive protein; E, extent; Hgb, hemoglobin; UC, ulcerative colitis; IBD-U, IBD-unclassified; L, location.

After ADA detection, anti-TNF dose intensification occurred in the majority of patients (25 [83%] of the IM patients vs 28 [48%] of the No-IM patients). Among the 5 patients (17%) in the IM group and 12 patients (20%) in the No-IM group who were not dose-intensified subsequent to threshold ADA detection, all were either recently dose-intensified or had an optimized drug level (≥ 5 µg/mL) at the time of ADA detection. Nineteen patients (32%) in the no-IM group were stopped on their initial anti-TNF drug regimen without additional doses after detection of the threshold ADA level. Fourteen (74%) of these patients cycled within class to another anti-TNF agent and 5 (26%) switched out of class to a treatment with a different mechanism of action. By definition, all patients in the IM group continued their anti-TNF agent. Supplementary Table 1 outlines the indications for discontinuing the initial anti-TNF agent shortly after ADA detection without further optimization.

Sixteen patients within the entire cohort had an initial threshold ADA level ≥1000 ng/mL. No patients with an ADA level >1000 ng/mL in the No-IM group attempted to continue on the initial anti-TNF drug. In contrast, all 5 patients in the IM group with ADA levels >1000 ng/mL remained on the initial anti-TNF drug for at least 400 days. Within the No-IM group, the patients who stopped their anti-TNF agent shortly after threshold ADA detection had significantly higher ADA levels (median 1335 ng/mL; IQR = 355-1735) than those who continued the initial anti-TNF (312 ng/mL; IQR = 235-502; P < 0.001). In line with the ADA level differences, within the No-IM group, the patients who stopped their anti-TNF agent shortly after threshold ADA detection had significantly lower anti-TNF drug levels (median 0.4 µg/mL; IQR = 0.4-0.6) than those who continued the initial anti-TNF (2.1 µg/ml; IQR =0.4-6.5; P = 0.003). Only 4 of 19 (21%) No-IM patients who immediately stopped their initial anti-TNF had detectable drug levels, compared to 29 of 40 patients (73%) who continued their initial anti-TNF treatment (P < 0.001).

Anti-TNF Durability

Patients in the No-IM group were more likely to discontinue the initial anti-TNF related to LoR or adverse events over time than those in the IM group (P = 0.007) (Fig. 1). Whereas patients in the No-IM group mostly discontinued anti-TNF within the first 6 months, discontinuation in the IM group happened more gradually. Fifty-four percent (32/59) of patients in the No-IM group remained on the initial anti-TNF agent at 6 months, and 93% (28/30) of patients in the IM group remained on the initial anti-TNF. At 12 months, 53% (31/59) of patients in the No-IM group remained on the initial anti-TNF compared with 87% (26/30) of those in the No-IM group.

FIGURE 1.

Kaplan-Meier plot of survival on initial anti-TNF after ADA detection.

The difference in anti-TNF durability was driven by the substantial number of patients in the No-IM group whose providers discontinued the initial anti-TNF in response to the initial threshold ADA level and represents the outcomes related to treatment decisions made in real-life practice. To compare anti-TNF durability between those in each group whose providers intended to keep them on the initial anti-TNF, we performed a secondary analysis that included only patients who had received at least 1 anti-TNF dose after ADA detection and were dose-optimized if the anti-TNF drug concentration was less than 5 µg/mL (30 IM and 40 No-IM patients). Of note, these subgroups were much less similar with regard to baseline drug and ADA levels, as compared with the entire IM and No-IM groups. The subgroup of No-IM patients whose providers attempted to keep them on the initial anti-TNF had significantly lower ADA levels (median = 312 ng/mL [IQR = 235-502] vs 491 ng/mL [IQR = 368-834]; P < 0.001) and a corresponding trend toward higher drug levels (median = 2.1 µg/mL [IQR = 0.6-6.5] vs 0.7 µg/mL [IQR = 0.4-3.5]; P = 0.097) than the subgroup of IM patients. In this secondary Kaplan-Meier analysis with fewer patients, numerically, a greater percentage of IM vs No-IM patients remained on initial anti-TNF at 6 months (93% vs 80%) and 12 months (77% vs 65%), but the curves converged at 17 months and the differences were not statistically significant (P = 0.73) (Supplementary Figure 2).

Clinical Outcomes After ADA Detection

At 6 months after ADA detection, 63% of the IM patients were in SFCBR on the same anti-TNF agent vs 30.9% of the No-IM patients. (risk difference [RD] = 11.5%; 95% confidence interval [CI], 10.4-38.6; P = 0.008; Fig. 2A). On univariable analysis, the addition of IM was significantly associated with the likelihood of SFCBR-SD (odds ratio = 3.8; 95% CI, 1.44-10.01; P = 0.007). Longer duration of anti-TNF treatment before ADA development was also associated with SFCBR-SD at 6 months (β-coefficient = 1.02; 95% CI, 1.01-1.03; P = 0.005). On multivariable logistic regression, both IM use and anti-TNF duration before ADA detection remained independent predictors of SFCBR in patients who were on the same anti-TNF at 6 months (Table 2).

FIGURE 2.

SFCBR at 6 and 12 months after ADA detection (A) on or (B) regardless of remaining on same anti-TNF drug.

TABLE 2.

Logistic Regression Analysis of SFCBR-SD at 6 and 12 Months

	Univariable Logistic Regression			Multivariable Logistic Regression
	OR	95% CI	P	aOR	95% CI	P
SFCBR-SD at 6 months
IM (IM vs No-IM)	3.80	1.44-10.01	0.007	3.81	1.32-10.99	0.013
Anti-TNF duration	1.02	1.01-1.03	0.005	1.02	1.004-1.030	0.009
ADA level	0.99	0.99-1.00	0.079	0.99	0.99-1.00	0.048
Male sex	1.11	0.46-2.69	0.825	-	-	-
SFCBR-SD at 12 months
IM (IM vs No-IM)	3.19	1.21-8.43	0.019	3.03	1.13-8.12	0.027
Anti-TNF duration	1.01	0.99-1.02	0.114	1.01	0.99-1.01	0.153
ADA level	0.99	0.99-1.00	0.241	-	-	-
Male sex	1.73	0.70-4.32	0.237	-	-	-

aOR indicates adjusted odds ratio; OR, odds ratio.

SFCBR-SD at 12 months was also attained more frequently in IM patients (53.8%) than in No-IM patients (26.8%; P = 0.025; Fig. 2A). At 12 months, IM was the only predictor for SFCBR-SD by univariable or multivariable analysis (Table 2).

When assessed without regard for patients remaining on the same anti-TNF agent, the incidences of SFCBR at 6 and at 12 months were similar between the groups (Fig. 2B). Therefore, patients in the IM group had comparable favorable outcomes to those in the No-IM group despite fewer patients in the IM group changing drugs.

Effect on Anti-TNF Drug and ADA Levels

Changes in anti-TNF drug and ADA levels were assessed among patients who had at least 1 additional level measured between 3 and 12 months after ADA detection. Patients in the IM group had a median increase in their anti-TNF drug level of 7.2 µg/mL (IQR = 1.0-15.6) compared with a median change of 1.6 µg/mL (IQR = –0.9 to 4.6) in the No-IM patients (P = 0.003) (Fig. 3A). In line with this finding, we also observed that patients in the IM group exhibited a larger median relative decrease (87.0% decrease; IQR = 44.5%-92.5%) in ADA levels between baseline and follow-up than did the No-IM patients (60% decrease; IQR = 15.5%-75%; P = 0.037; Fig. 3B).

FIGURE 3.

Change in (A) drug and (B) ADA levels among patients with at least 1 follow-up level between 3 and 12 months after initial detection of high ADA. Error bars indicate median and IQR. A, Graph of initial ADA levels stratified by ADA reversal between 3 and 12 months in No-IM and IM groups. B, ROC curve showing performance characteristics of baseline ADA level for predicting ADA reversal in No-IM patients. Error bars indicate median/IQR.

Likelihood of Immunogenicity Reversal

Fig. 4A compares the initial ADA levels of patients who remained on the same anti-TNF (and had at least 1 additional ADA level measured between 3 and 12 months after ADA detection) between those who did and did not reverse ADAs (ADAs <200 ng/mL for IFX or <100 ng/mL for antibodies to ADL) within the first year, stratified by IM group. Higher ADA levels were reversed in the IM patients (median initial ADA level = 488 ng/mL; IQR = 313-947) than in the No-IM patients (271 ng/mL; IQR = 217-364; P = 0.004). The maximum baseline ADA level that was ultimately reversed in the IM group was 2178 ng/mL compared to a maximum of 855 ng/mL reversed in the No-IM group. Furthermore, we observed a relationship between initial ADA level and subsequent ADA reversal in the No-IM group but did not see this relationship in the IM group. Specifically, within the No-IM group, patients who failed to reverse ADA levels had a higher initial median ADA level (525 ng/mL; IQR = 344-748) than patients who were able to reverse ADA levels (271 ng/mL; IQR = 217-364; P = 0.029). An ROC analysis of No-IM patients showed that initial ADA levels <329 ng/mL predicted a higher likelihood of ADA reversal (Fig. 4B). Among No-IM patients treated with IFX, 329 ng/mL (sensitivity 71%, specificity 100%) remained the level under which ADA reversal was more likely (area under the ROC [AUROC] = 0.86; 95% CI, 0.69-1.00; P = 0.013). Of note, the ROC curve for ADL showed similar performance characteristics, but this was not statistically significant (Supplementary Table 2).

FIGURE 4.

A, Graph of initial ADA levels stratified by ADA reversal between 3 and 12 months in No-IM and IM groups. B, ROC curve showing performance characteristics of baseline ADA level for predicting ADA reversal in No-IM patients. Error bars indicate median/IQR. C, Graph of initial anti-TNF drug levels stratified by ADA reversal between 3 and 12 months in No-IM and IM groups. D, ROC curve showing performance characteristics of baseline anti-TNF level for predicting ADA reversal in No-IM patients. Error bars indicate median/IQR. E, Kaplan-Meier survival analysis on initial anti-TNF agent among No-IM patients who remained on monotherapy and had repeat levels drawn, stratified by initial ADA and anti-TNF level.

Fig. 4C compares the initial anti-TNF drug levels between patients who remained on the same anti-TNF (and had at least 1 additional level measured between 3 and 12 months after ADA detection) between those who did and did not reverse ADAs (ADAs <200 ng/mL for IFX or <100 ng/mL for ADL) within the first year, stratified by IM group. Similar to what we observed for ADA levels, we found a relationship between initial drug level and subsequent ADA reversal in the No-IM group but did not see this relationship in the IM group. Among No-IM patients, those who reversed their ADA levels had a higher median baseline anti-TNF drug level (4.7 µg/mL; IQR = 1.5-9.4) than those who did not reverse their ADAs (0.7 µg/mL; IQR = 0.5-2.3; P = 0.005). Moreover, among the no-IM group, an anti-TNF level >2.8 µg/mL was more likely to result in ADA reversal (AUROC = 0.792; P = 0.006; Fig. 4D). Furthermore, whereas only 1 of 7 (14%) No-IM patients with undetectable anti-TNF levels were able to reverse their ADAs, 10 of 14 (71%) IM patients with undetectable anti-TNF levels were able to reverse their ADAs (P = 0.024). Among No-IM patients treated with IFX, 2.9 µg/mL (sensitivity 64%, specificity 83%) was the drug level above which ADA reversal was numerically more likely, but this was not statistically significant (AUROC = 0.70; 95% CI, 0.41-0.99; P = 0.161). In contrast, the ROC curve for ADL showed similar performance characteristics and showed that patients with an ADL level >2.6 µg/mL (sensitivity 71%, specificity 100%) were more likely to reverse their ADAs (AUROC = 0.87; 95% CI, 0.65-1.00; P = 0.027; Supplementary Table 2).

Combining the test criteria of an ADA level <329 ng/mL or an anti-TNF level >2.8 µg/mL improved the overall performance characteristics to a sensitivity of 90% (95% CI, 71%-97%), a specificity of 75% (95% CI, 47%-91%), positive predictive value 86% (95% CI, 67%-95%), and negative predictive value 82% (95% CI, 52%-95%) (P < 0.001). A subsequent Kaplan-Meier survival subanalysis of these No-IM patients who remained on anti-TNF monotherapy and had subsequent levels measured showed longer anti-TNF durability among patients with an initial ADA level <329 ng/mL or an anti-TNF level >2.8 µg/mL (P = 0.002; Fig. 4E).

DISCUSSION

In this study, we showed that pediatric patients on anti-TNF monotherapy who develop ADA and are started on an IM are more likely to remain on the same anti-TNF drug and subsequently be in SFCBR-SD. Initiation of IMs also led to higher increases in anti-TNF drug levels and larger reductions of ADA concentrations. Furthermore, we identified an ADA level threshold of 329 ng/mL above which patients were highly unlikely to reverse their ADAs without addition of an IM. Moreover, patients who had an anti-TNF level <2.8 µg/mL at the time of ADA detection who remained on monotherapy were less likely to reverse their ADAs. This is the first study to compare the addition of an IM drug in the setting of new ADA with management without IM and the first to examine management of ADAs detected by a commonly used ECLIA.

This study provides evidence that initial monotherapy with an anti-TNF agent followed by dose intensification and secondary IM initiation in response to immunogenicity may be an alternative anti-TNF treatment approach to initial combination therapy. In adults, early combination therapy with azathioprine has been shown to be more efficacious and to improve anti-TNF durability compared with anti-TNF monotherapy.^20-22 Similarly, in pediatric patients it has been shown that early combination therapy is more durable than monotherapy but that durability is better with methotrexate than with thiopurines.²³ However, combination therapy in the pediatric population is more controversial; malignancies such as hemophagocytic lymphohistiocytosis and hepatosplenic T-cell lymphoma are more common among youth (young men) treated with thiopurines.^{24, 25} In addition, monotherapy may sometimes be preferred over combination therapy because of methotrexate intolerance or patient and/or family preferences for limiting the number of drugs to which a child is exposed.

Several small adult case series and cohort studies previously reported that ADAs could be reversed and LoR could be restored.^{12, 13, 26} Although those studies addressed ADAs in the setting of LoR, there is some evidence that ADA reversal may be more beneficial if it is attempted during proactive therapeutic drug monitoring regardless of LoR, such as in this current study.²⁷ Those studies also reported on patients managed with the addition of an IM and provided initial evidence that IM addition at the time of ADA development was associated with ADA reversal and recapturing of response. The current study builds on these findings by direct comparison of the outcomes of those patients who started IMs in the face of new ADAs with the outcomes of those who did not.

In the pediatric IBD literature, there is 1 other study that has evaluated the management of anti-TNF therapy after ADA detection.²⁸ In contrast to our study, patients in that study were generally managed with either anti-TNF dose-switching or a change in drug therapy (IM was only added in a single patient), and the drug and ADA levels were measured using a homogeneous mobility shift assay. Those authors similarly observed a threshold ADA level above which no patients reversed their ADA levels (10 U/mL). Our study builds on this work by providing new data regarding the utility of adding an IM to reverse ADA levels, as measured by a commonly used sensitive ECLIA, in pediatric patients.

We acknowledge several limitations of our study. As with any retrospective observational study, there is a risk of confounding. However, baseline characteristics between the groups were quite similar except for longer prior anti-TNF duration, more men, and higher ADA levels in the IM group, and we applied multivariable logistic regression to control for these variables. We also recognize that the durability analysis is at risk for confounding by indication—providers who did not start an IM may have intended to stop the initial anti-TNF earlier and those providers starting an IM intended to continue anti-TNF therapy. Although we attempted to overcome this limitation with a secondary durability analysis that only included patients who were intended be optimized and to be continued on the initial anti-TNF therapy, narrowing the inclusion criteria led to highly unbalanced groups with regard to baseline drug and ADA levels. Therefore, it is difficult to draw firm conclusions from this this secondary analysis and we limit our conclusions regarding durability to the observation that in our real-world practice, IM patients remained on initial anti-TNF longer with favorable outcomes. Clinical remission was defined based on PGA because it is recorded at the point of care at each visit and is the key metric of the quality improvement network in which we participate. Although we were not able to include endoscopic measures in our outcome, we combined the PGA with an objective biochemical assessment, C-reactive protein, to render our outcome more stringent. Although our study had a relative small sample size, our sample size was larger than if not commensurate with previous pediatric studies and most adult studies, and we included more patients on IM than any prior adult IBD study assessing immunogenicity reversal.^26-28

CONCLUSIONS

In conclusion, combining an IM drug with anti-TNF monotherapy in addition to anti-TNF dose intensification in pediatric patients with new significant ADA levels is effective for maintaining response to the original anti-TNF and for reversing ADA levels. Dose optimization alone is unlikely to reverse ADA levels >329 ng/mL, particularly if combined with an anti-TNF level <2.8 µg/mL. Reserving the addition of an IM to proactively monitored patients on anti-TNF monotherapy who develop ADAs may be an alternative strategy to initial combination therapy for all patients for maintaining drug durability and response.

Supported by: This project was supported in part by the National Institutes of Health grant number P30 DK078392 (clinical component of the Digestive Diseases Research Core Center in Cincinnati). RJC was supported by National Institute of Diabetes and Digestive and Kidney Diseases grant number T32DK007727. Research Electronic Data Capture tools hosted at Cincinnati Children’s Hospital Medical Center were supported in part by the National Institutes of Health (NIH/NCATS grant number UL1 TR000445).

Conflicts of interest: The authors have no specific conflicts of interest for this article to declare.