Influence of HLADQA1*05 Genotype in Adults With Inflammatory Bowel Disease and Anti-TNF Treatment With Proactive Therapeutic Drug Monitoring: A Retrospective Cohort Study

Abstract

Background

Carriers of the human leucocyte antigen variant HLADQA1*05 (rs2097432) are at risk of developing antibodies against infliximab and adalimumab with reduced tumor necrosis factor (TNF) antagonist persistence. The impact of proactive therapeutic drug monitoring (PTDM) on this association has been barely assessed.

Methods

We conducted a retrospective single-center cohort study including patients with inflammatory bowel disease starting anti-TNF therapy between January 2017 and March 2021. Proactive therapeutic drug monitoring was defined as periodic drug level measurement (≥2 determinations during the first year of treatment and ≥1/annual determination during the following years), regardless of clinical condition, followed by dose optimization. Variables associated with treatment persistence were assessed with multivariable Cox regression analysis.

Results

A total of 112 patients were included, 52 (46.4%) HLA-DQA1*05 carriers, with a median follow-up of 73.9 (interquartile range, 35.4-133.1) weeks. Combination therapy with thiopurines was more frequent among HLA-DQA1*05 noncarriers (28 [46.7%] vs 12 [23.1%]; P = .01). Clinical remission rates at week 14 (77.9% vs 73.9%; P = .69) and 56 (73.2% vs 68.4%; P = .64) were similar between HLA-DQA1*05 noncarriers and carriers. Drug persistence was higher among HLA-DQA1*05 carriers (hazard ratio [HR], 0.32; 95% confidence interval, 0.14-0.71; P = .01). Multivariable Cox regression analysis identified systemic steroids at anti-TNF initiation (HR, 4; 95% confidence interval, 1.7-9.7) as a risk factor and HLA-DQA1*05 carriers (HR, 0.31; 95% confidence interval, 0.12–0.81) as a protective factor of treatment cessation.

Conclusion

In adult patients with PTDM, a positive HLA-DQA1*05 genotype does not associate a higher risk of treatment cessation nor worse clinical outcomes.

Open in new tabDownload slide

Introduction

Tumor necrosis factor (TNF) antagonists constitute a hallmark on the treatment of inflammatory bowel disease, although primary or secondary loss of response (LOR) remain a major problem leading to drug withdrawal.¹ Immunogenicity counts among the leading causes of loss of response; up to 65.3% and 38% of patients receiving infliximab and adalimumab, respectively, may develop specific antidrug antibodies.²

Different strategies aim at limiting this risk. The combination of TNF antagonists with immunomodulators reduces the immunogenicity to both adalimumab and infliximab³ and achieves higher remission rates than TNF antagonist monotherapy.^4,5 Therapeutic drug monitoring using a proactive strategy (PTDM), where drug levels are measured regardless of symptoms and inflammatory biomarkers to ensure optimal drug levels, has also been proposed. Specifically designed randomized controlled trials showed PTDM was associated with lower risk of flares during maintenance⁶ but failed to achieve differences in clinical, endoscopic, and corticosteroid-free remission in patients with active Crohn’s disease (CD).⁷ Nevertheless, PTDM is increasing, as observational studies suggest it could have a positive effect with lower drug discontinuation rates,⁸ lower relapse rates⁹ and increased rates of mucosal healing.¹⁰ PTDM has shown increased clinical remission rates compared with reactive therapeutic drug monitoring in paediatric patients.¹¹

A recently published genome-wide association study including over 1200 patients with Crohn’s disease from the PANTS cohort identified that the carriers of a genetic variation in the class II human leucocyte antigen (HLA) gene region (HLADQA1*05, rs2097432) present an increased risk of developing antidrug antibodies (ADAs) and a reduced TNF antagonists’ persistence.¹² These results were confirmed by Wilson et al in a retrospective study assessing subjects with CD and ulcerative colitis (UC) receiving infliximab.¹³

The change these findings might introduce on the current available strategies to reduce immunogenicity is yet to be determined. Although co‐immunosuppression reduced immunogenicity and treatment cessation both in variant and nonvariant carriers,^12,13 the improvement was clinically significant among HLADQA1*05 carriers, suggesting combination therapy could be tailored by the HLA status. Recently, a single-center study assessing the impact of PTDM in a pediatric and young adult cohort starting infliximab suggested that failure to achieve target drug concentrations during the induction, rather than the HLA status, was associated with antidrug formation.¹⁴ Therefore, the aim of this study was to assess the influence of the HLADQA1*05 carriers in a cohort of patients receiving TNF antagonists under PTDM.

Methods

We performed a single-center, retrospective cohort study assessing the impact of HLADQA1*05 genotype in IBD patients receiving TNF antagonists under PTDM. The study was approved by the institutional review board (PI-21E0087). All participants provided written informed consent prior to inclusion in the study. This was an industry-independent study designed according to the STROBE recommendations.¹⁵

Patients

All patients 14 years of age or older with a diagnosis of UC or CD according to ECCO guidelines¹⁶ who were started on TNF antagonists (infliximab or adalimumab) between January 2017 and March 2020 were considered eligible. Patients could receive the following concomitant medications: immunosuppressors (thiopurines or methotrexate) and oral mesalazine in ulcerative colitis. Patients were identified using the hospital pharmacy logs. The following exclusion criteria were stated: patients lacking proactive drug monitoring, subjects started on TNF antagonists in other centers or those prescribed for any indication different than IBD or by a nongastroenterologist, and those declining to participate or to assess their HLA status.

HLADQA1*05 Assessment

The HLADQA1*05 screening was performed at one of the follow-up visits at the IBD outpatient clinic between September 2020 and August 2021. Patients were assigned to 2 different groups: HLADQA1*05 noncarriers and HLADQA1*05 carriers (of 1 or 2 allele copies) according to the result. Determination of HLADQA1*05 was performed with the polymerase chain reaction-sequence specific oligonucleotide probe (PCR-SSO) method using the line probe assay (LiPA) typing kit INNO-LiPA HLA-DQA1 multiplex (Fujirebio Europe N.V., Belgium). It is based on the principle of reverse hybridation of the polymerase chain reaction (PCR) products of exon 1 and exon 2 + 3 of HLADQA1*05 gene on allele-specific probes fixed on a membrane.

Follow-up

All participants were attended at a specialized IBD outpatient clinic for a minimum of 3 months. Follow-up was considered complete if clinical data were available until treatment cessation, patient death, or the period June to August 2021. Patients’ visits were scheduled at week 0, 6 to 8, and 12 to 16 after induction, whereas the subsequent visits were scheduled every 4 to 6 months. If relapse or primary nonresponse were suspected, intervals were shortened at the physician’s discretion. Complete blood counts, C reactive protein (CRP), and calprotectin values were usually checked in each visit.

Outcomes and Definitions

The primary outcome was treatment persistence at the end of follow-up, which was defined as continuation of the TNF antagonist. Secondary outcomes measured at week 14 and 54 were treatment persistence with clinical remission (defined as a Harvey-Bradshaw index of <5 in CD¹⁷ or a partial Mayo Clinic score of 2 or lower, and no subscore higher than 1 in UC¹⁸), steroid-free clinical remission (including oral prednisone, beclomethasone, or budesonide), and treatment persistence with clinical and biochemical remission (defined as a fecal calprotectin <250 mg/kg). Patients who ceased treatment for any reason before these assessments were deemed treatment failures. Adverse events were graded following the Common Terminology Criteria for Adverse Events (CTCAE).¹⁹ Mild infusion reactions with infliximab (transient symptoms not requiring specific treatment) were managed by decreasing the infusion rate.

Proactive drug monitoring was defined as the assessment of trough concentrations to optimize dosing during induction therapy and intermittently thereafter regardless of symptoms or inflammatory markers, with ≥2 determinations in the first year of treatment and ≥1 in the following years. During induction, the following target concentrations were employed: 25 to 30 mcg/mL (week 2) and 20 mcg/mL (week 6) for infliximab, and >10 mcg/mL (week 2 and 4) for adalimumab. Checks were performed systematically at week 6 (infliximab) and at week 4 (adalimumab). Induction concentrations at week 2 were performed on a case-by-case basis. They were also routinely performed following any dose modification. Timing of the remaining checks varied depending on the number of previous dose modifications, previous concentrations achieved, disease extension, severity, and inflammatory markers. During the maintenance, the targets were 5 to 10 mcg/mL for infliximab and 8 to 12 mcg/mL for adalimumab. In case of associated perianal disease, extensive small bowel disease or active disease, higher concentrations (10-20 mcg/mL) were sought. Antidrug antibodies were only checked in patients with undetectable TNF antagonist concentrations using drug-sensitive assays.

Dose modifications were not decided using a predefined algorithm but discussed on a one-by-one basis. Depending on the trough concentration, intervals/dosing were reduced or increased to approach the target levels. Treatment intensification was defined as any increase from conventional dosing, namely 5 mg/kg every 8 weeks for infliximab and 40 mg every other week for adalimumab. They were further divided in major (those reaching >9 mg/kg or an interval of 4 weeks or fewer for infliximab, an 80 mg dose or a weekly interval for adalimumab and those requiring a reinduction in both groups) and minor (those not fulfilling the criteria stated previously) intensifications.

Data Retrieval

Data on baseline demographics, diagnosis, and clinical and analytical assessments during follow-up data were retrospectively obtained from electronic medical records and introduced into a specifically designed database.

Statistical Methods

Statistical analysis was performed using Stata (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX). Continuous variables were presented as mean and standard deviation or median with interquartile range as warranted. Categorical variables were presented as numbers and percentages. The Pearson χ² test and Fisher exact test were used to assess differences between categorical values as warranted, and the Student t test was employed in continuous variables. Unadjusted Kaplan-Meier survival curves were used to demonstrate the extent of treatment persistence, and comparisons were performed using the log-rank test. Patients were censored if lost during follow-up or if they completed the study period. Multivariable Cox proportional hazards regression was used to assess possible risk factors of treatment discontinuation; results were reported using hazard ratios (HRs) with 95% confidence intervals (CIs). The following items were initially assessed using univariable regression analysis: age at diagnosis, interval between diagnosis and TNF antagonist therapy, baseline calprotectin and CRP, body mass index, sex, smoking status, type of IBD, type of TNF antagonist, concomitant steroids or immunomodulators while starting TNF antagonists, induction trough levels at induction (week + 6 for infliximab, week + 4 for adalimumab) divided in quartiles, and HLA status (as a dichotomous variable). Predictors with a significance level ≤0.10 were included in a multivariable model using bidirectional elimination. A P value <.05 was considered statistically significant.

Results

A total of 126 subjects were started on TNF antagonists during the study period. Fourteen were excluded; 8 patients with a median follow-up of 340 (196-1246) days lacked a proactive drug monitoring; and 6 did not allow having their HLADQA1*05 status checked. Thus, a total of 112 patients with a median follow-up of 517.5 (248-932) days were included in the final analysis. The HLADQA1*05 status was screened after ceasing anti-TNF treatment in 29 (25.9%) patients. In the remaining 83 (74.1%) patients, it was requested in the last 2 outpatient visits in 84.3% of them 118 (90-137) days before completing follow-up.

Participants

Among the 112 included patients, 52 (46.4%) were HLADQA1*05 carriers. Baseline characteristics of the study population are shown in Table 1. Thirty-one patients (27.7%) were on steroid treatment when TNF antagonists were initiated. Most patients (100 of 112) were started on TNF antagonists due to endoscopically and/or radiologically confirmed disease activity, although 38% of them were on clinical remission at TNF antagonists first administration. Thiopurine and methotrexate cotherapy was less frequent among HLADQA1*05 carriers (23.1%) than among noncarriers (46.7%; P = .01), although HLADQA1*05 carriers had a higher proportion of previous immunomodulator therapy before starting TNF antagonists (25% vs 11.7%; P = .07). Nine (8%) patients had received a TNF antagonist before inclusion; only 3 (2 noncarriers) had ceased treatment due to secondary loss of response, whereas the rest had suspended the TNF antagonist due to adverse events, intolerance, or patient’s decision. Nine (15%) noncarriers had a previous history of IBD-related surgery: 3 (5%) intestinal resections and 6 (10%) perianal interventions. Among the 10 (19.2%) carriers with previous surgery, 6 (11.5%) of them had undergone intestinal resections.

Proactive Drug Monitoring

During the first year of treatment, both groups underwent a very similar number of trough concentrations checks (3 [2-3.5] in noncarriers and 3 [2-4] in carriers). The overall median annual number of checks during follow-up was also equivalent: 3.3 (2.1-4.8) vs 3.3 (2.2-4.3), respectively. Overall, 497 checks were performed: 247 in HLADQA1*05 noncarriers and 250 in HLADQA1*05 carriers. The proportion of checks leading to dose modifications was similar in both groups (31.6% vs 28.4%; P = .44). Although most dose modifications (86.6%) increased dosing (either increasing the dose or reducing the interval between treatments), 13.4% of changes reduced it. Dose modifications were frequently performed in patients in clinical remission (52.6% in HLADQA1*05 noncarriers vs 63.4% in HLADQA1*05 carriers; P = .18). Trough concentrations at week +6 in infliximab treated patients were 17 (7-20) mcg/mL in HLADQA1*05 noncarriers and 17.5 (13-26) mcg/mL in HLADQA1*05 carriers (P = .27), whereas adalimumab concentrations at week +4 were also comparable (13 [10-18] mcg/mL and 12 [10-14] mcg/mL; P = .43).

There were no differences in the proportion of patients with a major intensification at the end of follow-up 45 of 60 (75%) among noncarriers and 36 of 52 (69.2%) among HLADQA1*05 carriers (P = .67). Most intensifications were performed shortening dosing intervals; at the end of follow-up, 64.6% of patients receiving infliximab were on a 4-week interval, and 46.8% of ADA patients received treatment weekly. The last available trough concentration during follow-up was slightly lower among HLADQA1*05 carriers, both in subjects treated with adalimumab (10.5 [8-15] vs 15 [11-18.5]; P = .14) and infliximab (8.5 [6-22.5] vs 12.5 [7-18.5]; P = .04).

Six (5.4%) patients with Crohn’s disease, 3 HLADQA1*05 carriers, and 3 noncarriers presented undetectable TNF antagonists’ concentrations and could be checked for ADA—2 subjects in each cohort testing positive. Four were on clinical remission. Reinduction was attempted in all cases. One symptomatic patient underwent reinduction in the last available follow-up visit, so the outcome was not available. In the remaining patients, acceptable levels could be recovered, and the other symptomatic patient achieved clinical remission. One patient ceased infliximab posteriorly due to difficult intravenous access.

Clinical End Points

The proportion of patients reaching clinical remission at week 14 was 73.9% among noncarriers and 77.9% among HLADQA1*05 carriers (P = .69). Steroid-free clinical remission and clinical/biochemical clinical remission were also quite similar, as shown in Figure 1. The assessment performed at week 54 showed no differences between HLADQA1*05 noncarriers and carriers regarding clinical remission (68.4% vs 73.2%; P = .64), steroid-free clinical remission (68.4% vs 73.2%; P = .64), and clinical and biochemical remission (65.8% vs 61%; P = .66).

Overall, adverse events were observed in 3 (5%) HLADQA1*05 noncarriers (1 local infection after hip replacement surgery, 1 severe infusion reaction requiring intravenous administration of steroids, and 1 paradoxical psoriasis) and 8 (15.4%) HLADQA1*05 carriers (2 cases of paradoxical psoriasis, 3 mild infections, 1 severe pneumoni,a and 2 mild, self-limited, infusion reactions; P = .11).

Treatment Persistence

Median follow-up was 517.5 (248-932) days. Only 1 patient was lost after a follow-up of 398 days. Among the 60 HLADQA1*05 noncarriers, 23 (38.3%) ceased treatment for the following reasons: 10 (16.7%) due to loss of response; 5 (8.3%) to primary treatment failure; 3 (5%) to adverse events (infusion reaction, severe infection, and paradoxical psoriasis); 3 (5%) decided to cease treatment despite their practitioner’s recommendation during the COVID-19 pandemic; 1 (1.7%) patient on combination treatment stopped the TNF antagonist maintaining the thiopurine; and 1 (1.7%) patient stopped infliximab due to difficult venous access. Treatment was suspended in 8 (15.4%) HLADQA1*05 carriers: 5 (9.6%) due to loss of response and 3 (5.8%) due to adverse events (2 cases of paradoxical psoriasis and 1 severe respiratory infection). Additionally, HLADQA1*05 carriers were more likely to experience medication persistence than noncarriers (P = .003; Figure 2). The 12-month persistence was 93.6% among HLA-DQA1*05 carriers and 79.8% among noncarriers. Figure 3 presents the treatment persistence curves by type of TNF-antagonist.

Factors Associated With Treatment Cessation

Results of univariable and multivariable Cox regression analysis are summarized in Table 2. Univariable analysis identified steroid treatment at TNF antagonist initiation, type of TNF antagonist, baseline calprotectin values >250 mg/kg, HLADQA1*05 status and age at diagnosis as potential risk factors of treatment persistence. Multivariable Cox regression analysis kept systemic steroids at anti-TNF initiation (HR, 4.01; 95% CI, 1.66-9.68) as a risk factor and a positive HLADQA1*05 genotype (HR, 0.31; 95% CI, 0.12-0.81) as a protective factor of treatment cessation. If the 3 patients stopping treatment on their own were not considered as events in the regression analysis, systemic steroids at anti-TNF initiation would maintain a similar HR (3.82; 95% CI, 1.69-8.64), but a positive HLADQA1*05 genotype would lose the statistical significance (HR, 0.43; 95% CI, 0.19-1.01; Suppl Table 1).

Discussion

In this retrospective study including 112 patients started on TNF antagonists under PTDM, HLADQA1*05 carriers and noncarriers presented similar clinical and biochemical remission rates at week 14 and 54. Although a significantly higher drug persistence was observed among HLADQA1*05 carriers, this was probably related to the 8.3% of noncarriers ceasing treatment due to reasons other than treatment failure or adverse events, as suggested by the multivariable analysis results.

These results agree with the available studies assessing the impact of the HLADQA1*05 allele in patients undergoing PTDM. A post hoc analysis of the Precision IFX trial did not observe an increased risk of ADA formation or lower drug persistence in HLADQA1*05 carriers.¹⁴ In this cohort, the only risk factor of ADA formation was not reaching a trough level of 17 µg/mL at the third infliximab dose (HR, 5.0; 95% CI, 1.8-13.7).¹⁴ Another study post hoc analysis of a prospective multicenter pediatric pharmacokinetic study including 78 patients identified a neutrophil CD64 ratio >6 and a starting dose <7.5 mg/kg as the only predictors of ADA formation.²⁰ The Precision trial instituted a strict PTDM, where all infliximab doses and intervals were calculated using a model including various variables (eg, weight, albumin, inflammatory markers, infliximab drug level, and ADA level). Interestingly, our cohort attained similar results, eliminating the impact of the HLADQA1*05 allele on drug persistence, despite using a much looser PTDM, with a median of 3 drug level checks in the first year of treatment.

Predicting ADA development and, more importantly, secondary loss of response, is crucial in patients with TNF antagonists in order to personalize treatment in high-risk patients.²¹ Thus, the identification of a genetic risk factor of ADA development such as the HLADQA1*05 allele was saluted as a game changer. Furthermore, the first studies reporting the increased risk observed in HLADQA1*05 carriers also reported it was ameliorated by the concomitant use of immunomodulators.^12,13 However, the use of immunomodulators has shown an increased risk of infections in UC and malignancy in CD,²² and combination therapy further increases the risks of serious and opportunistic infection.²³ Moreover, combination therapy may associate a higher risk of developing lymphomas²⁴ and nonmelanoma skin cancer.²⁵

Based on our results and the data published by Spencer et al,¹⁴ we consider that TNF antagonist monotherapy under PTDM could be an effective approach in HLADQA1*05 carriers. Although not considering the HLA status, previous studies comparing TNF antagonist monotherapy and combination therapy support our hypothesis. A post hoc analysis of the SONIC study found no difference regarding corticosteroid-free remission at week 26 between combination therapy and infliximab monotherapy in the highest quartile drug levels.²⁶ Specifically assessing PTDM, a retrospective cohort study showed that infliximab durability did not differ between monotherapy under PTDM and combination therapy,²⁷ and a post hoc analysis of the PAILOT trial showed the addition of immunomodulators added no benefits to monotherapy under PTDM.²⁸ Moreover, in contrast to combination therapy, high TNF antagonists drug levels do not harbor an increased risk of infections over lower drug levels.^29,30

However, some questions regarding proactive drug monitoring remain unsolved, the most important is the optimal therapeutic range for both adalimumab and infliximab.³¹ Overall, higher TNF antagonist levels seem to be related to better outcomes, although an optimal therapeutic range has not been established yet. In patients with CD, trough levels over 7 µg/mL at week 14 have been associated with better outcomes.^32,33 For ulcerative colitis, a post hoc analysis showed infliximab concentrations ≥6.7 μg/mL at week 14,³⁴ and a threshold of 12 µg/mL for adalimumab has been proposed.³⁵ In our center, we favor a dynamic approach, where target drug concentrations are variable, aiming for a higher concentration in case of perianal disease, extensive small bowel disease, or in case of any clinical, analytical, or radiological evidence of active disease. Although we lacked data on ADA and a predetermined algorithm to accurately guide dose adjustment, nearly 75% of our patients completed a major intensification during follow-up, mostly using interval reductions. This fact led to high trough levels compared with previously proposed thresholds.

To our knowledge, this is the first study assessing the effect of proactive drug monitoring of anti-TNF therapy on the immunogenicity associated with HLADQA1*05 in adults in a real-life scenario with an extended follow-up. There are some limitations in our study that should be mentioned. First, this is a retrospective, single-center study with a small sample size, including patients receiving adalimumab and infliximab. Secondly, the lack of standardized clinical follow-up did not allow us to assess radiological or endoscopic outcomes, although C reactive protein and calprotectin were routinely checked in all outpatient visits. Thirdly, drug-tolerant assays are not available in our clinical practice, so no meaningful information on antidrug antibodies was attained. Lastly, the 3 HLADQA1*05 noncarriers who decided to cease treatment during the COVID-19 pandemic biased the permanence analysis, although this fact did not impact the clinical and biochemical remission rates.

In summary, in patients undergoing PTDM we observed no differences in clinical and biochemical remission or treatment persistence between HLADQA1*05 carriers and noncarriers. We believe these results could support proactive drug monitoring as a tool to overcome the reduced treatment survival observed in HLADQA1*05 carriers, although these data remain to be proven on prospective, specifically designed studies.

Funding

This project has received no funding.

Conflicts of Interest

Dr. Jesús Barrio has served as a speaker, as consultant, or has received research or education funding from MSD, Abbvie, Takeda, Janssen, and Ferring.

References