Evidence-Based Approach to the Discontinuation of Immunomodulators or Biologics in Inflammatory Bowel Disease

Inflammatory bowel disease (IBD) including ulcerative colitis (UC) and Crohn’s disease (CD) is an inflammatory condition that affects the digestive system with an unknown etiology [1, 2]. Although a variety of novel therapies have been developed, patients still need long-term treatment that imposes a variety of burdens. Immunomodulators (IMs) and anti-tumor necrosis factor (TNF) agents are one of the key drugs for the management in IBD. However, long-term use of IMs increases the rate of infection, lymphoproliferative disorder, and nonmelanoma skin cancer, especially when combined with anti-TNF agents [3, 4]. In addition, repetitive drug administration, hospital visits, and cumulative medical costs may affect patients’ quality of life, leading to treatment fatigue and a decrease in adherence to the treatment [5, 6]. On the other hand, healthcare costs and limited medical resources should be carefully utilized for sustainable treatment [7]. Therefore, an “arbitrary” continuation of these treatments should be avoided and patients have to be informed of the need for treatment continuation/discontinuation via a shared decision-making approach based on accumulated evidence.

Fundamental theories for withdrawal need to be based upon the risk of relapse and the efficacy of re-treatment. Notably, the true disadvantage of withdrawal should not be recognized by the relapse rate after discontinuation itself but by the increase of relapse rate over relapse rate with continued therapy, because a considerable number of patients may still relapse even if the therapy is continued (Fig. 1) [8‒14].

In a single RCT evaluating relapse after discontinuation of IM monotherapy in patients with UC, the relapse rate at 1 year was higher in the placebo group than in the azathioprine continuation group (36% vs. 59%) [15]. Furthermore, a multicenter observational retrospective study in patients with UC showed a two-thirds relapsed in 5 years after discontinuation [16]. A recent systematic review and meta-analysis demonstrated that discontinuation of IM monotherapy in patients with CD was associated with an increased risk of relapse within 24 months [10]. In addition, an observational study showed that in the long-term follow-up of CD patients who discontinued IMs, the cumulative probabilities of relapse were 14.0%, 52.8%, and 62.7% at 1, 3, and 5 years, respectively [17].

Table 1 demonstrates the risk of relapse after discontinuation of each treatment. Some demographic features such as male, younger age, and extensive colitis were associated with risk factors of relapse in UC and CD. The ECCO guideline recommends that IM monotherapy can be re-considered to be modified after 3–4 years if there is no evidence of active inflammation considering the cumulative risk of lymphoproliferative disease and nonmelanoma skin cancers [18]. This suggestion is made based on older studies [19]. In fact, in a recent prospective study of IBD patients with azathioprine of at least 5 years in deep remission, defined as clinical, endoscopic, histological, and biological findings including fecal calprotectin (FC) ≤50 µg/g and C-reactive protein (CRP) ≤10 mg/L, considerable numbers of patients relapse within a median 15 months (UC 58.1% vs. CD 30.8%) [20]. Therefore, whether long-term remission justifies their withdrawal is controversial.

Furthermore, we should take into account that the frequencies of genetic variants in genes coding thiopurine metabolite enzymes including thiopurine S-methyl transferase and nudix hydrolase 15 are different in Asian countries where IMs tend to be used in a lower dose than in Western countries. The clinical significance of thiopurine withdrawal should be elucidated in patients who are on low-dose thiopurines due to nudix hydrolase 15 heterozygous mutation because the impact on the risk-benefit balance might be different in such variants.

Evidence of the efficacy of re-treatment for relapse after discontinuation is scarce. In a previous study, 66 CD patients in sustained remission for more than 42 months who discontinued azathioprine were analyzed. Of the 32 patients who relapsed after discontinuation of azathioprine, 22 of the 23 patients who were re-treated with azathioprine alone achieved remission. The remaining 9 patients did not re-treat with azathioprine because they achieved with a different therapy including 4 cases with surgery. However, this study did not include patients with perianal disease or after surgery [17] Another multicenter retrospective cohort study in the UK following up UC and CD patients after discontinuation of IMs demonstrated that 74% of patients who resumed IMs at relapse regained and maintained remission, mostly in combination with re-induction by systemic steroids. However, it should be noted that a small percentage of patients experienced relapse requiring surgery, especially in CD [21].

A systematic review and meta-analysis revealed that there was no significant difference in clinical relapse between continuing and withdrawing IMs in combination therapy [10]. Furthermore, a recent randomized controlled trial (RCT) SPARE assessed withdrawal from combination therapy in CD comparing three arms: continuing combination therapy, discontinuing infliximab (IFX), and discontinuing IMs. This study indicated that there was no increase in relapse rate by discontinuing IMs compared with continuation of combination therapy (10% in 2 years vs. 14%) [12].

Does the evidence suggest that we can withdraw IMs in all cases? First of all, disease activity should be re-assessed at the time of IM withdrawal [22]. Then, we should consider the role of IMs in combination therapy, not only for the maintenance efficacy but for the pharmacokinetic effect on anti-TNF agents. An RCT comparing azathioprine continuation (2–2.5 mg/kg), dose reduction (1–1.25 mg/kg), and discontinuation from the combination therapy with IFX in IBD patients demonstrated that the increase of clinical relapse rate at week 56 was marginal. However, the trough level of IFX was significantly lower in the discontinuation group with or without anti-drug antibodies, which could result in the difference in poor long-term outcomes [23]. In addition, a recent large retrospective cohort study comparing withdrawal and continuation of thiopurine or methotrexate from combination therapy of IFX or adalimumab (ADA) in IBD demonstrated that there was no statistical difference in loss of response but while the development of anti-drug antibodies was increased by discontinuing IMs, associated with lower drug level of anti-TNF agents [22].

How can we take advantage of the therapeutic drug monitoring of pharmacokinetics? Drobne et al. [24] demonstrated that the higher trough level of IFX (>5 µg/mL) at the time of discontinuing IMs was associated with a lower rate of subsequent relapse in CD patients. On the other hand, Roblin et al. [23] showed that a cut-off value of 6-TGN with <105 pmol/8 × 10⁸ RBC during follow-up was associated with an unfavorable outcome for IFX pharmacokinetics in IBD, including the undetectable level of IFX with or without anti-drug antibody when reducing the dose of IMs. These findings may suggest that therapeutic drug monitoring could provide us with insight into the extent of the contribution of IMs or anti-TNF agents to the current state.

In contrast to IFX, the results of the RCTs in CD patients treated with ADA have not shown the benefit of combination therapy in achieving remission [25]. In addition, the DIAMOND2 trial, an RCT examining the outcome of discontinuation of IMs from combination therapy with ADA in CD, showed no significant differences in corticosteroid-free clinical remission, endoscopic remission, trough concentrations of ADA, or anti-drug antibody expression at week 52, although the sample size in the study was smaller than predetermined [26].

Based on such evidence, withdrawal of IMs from combination therapy seems to be generally acceptable after long-term remission if a trough level of anti-TNF agents is sufficient. Evidence regarding the outcome of the discontinuation of IMs from combination therapy with other biologic agents is lacking. However, in RCTs of ustekinumab and vedolizumab, the development of anti-drug antibodies was very rare, suggesting that combination therapy seems to be less beneficial [27, 28].

The STORI trial was the first prospective, uncontrolled, observational study that stopped IFX in quiescent CD with IMs revealing that 44% of patients relapsed 1 year after withdrawal. Risk factors of relapse in the multivariable analysis were shown as follows: male sex, absence of surgical resection, white blood cell (WBC) >6.0 × 109/L, hemoglobin ≤145 g/L, CRP ≥5.0 mg/L, and FC ≥300 µg/g. Ninety-eight percent of patients who restarted with IFX showed clinical response before the third dose of IFX infusion without a significant decrease in IFX trough levels compared to the time of withdrawal [29]. Furthermore, a long-term follow-up study of the STORI trial revealed that about 20% of patients did not start a biological treatment without a major complication 7 years after IFX withdrawal [30].

A systematic review and meta-analysis of previous observational studies demonstrated that 28% of UC and 36% of CD patients relapsed at 12 months following discontinuation of anti-TNF agents [8]. In addition, a recent large retrospective observational study evaluated >1,000 IBD patients who discontinued IFX or ADA and revealed that 12% of patients relapsed annually, with a cumulative relapse rate of 48% in 5 years after discontinuation [31].

Thereafter, a few RCTs that assessed the effect of withdrawal of anti-TNF agents in IBD have been published. The HAYABUSA study was the first RCT to evaluate IFX discontinuation in quiescent UC. The trial revealed that 80% of patients in the IFX continuation group and 54% of patients in the IFX discontinuation group kept remission 48 weeks after randomization. It demonstrated that neutrophil infiltration at randomization may attribute to relapse at 48 weeks, suggesting the underlying active state of the disease still remained.

Other clinical factors such as concomitant use of IMs, treatment duration prior to discontinuation, a Mayo endoscopic subscore, and IFX trough level were not predictive. On the other hand, Ben-Horin et al. [32] reported that undetectable drug levels of anti-TNF agents at discontinuation were associated with better relapse-free survival during the median follow-up of 12 months, suggesting that patients whose anti-TNF agents mainly contributed to the remission have a higher risk of relapse following withdrawal. This discrepancy in regard to the result of proactive therapeutic drug monitoring of anti-TNF agents might be explained by the difference in the cohort that almost all included patients in the HAYABUSA study had detectable IFX drug levels without anti-drug antibodies. After the withdrawal of IFX in the HAYABUSA study, 12 of 21 (57%) patients in the discontinuation group were re-treated with IFX, resulting in remission of eight cases (66.7%) in 8 weeks after re-treatment without infusion reactions [11].

There have been two RCTs assessing the outcomes of IFX discontinuation in CD. The STOP-IT trial showed a considerably larger risk discrepancy of clinical relapse at week 48 between discontinuation and continuation, despite deep remission defined as combined clinical, biochemical, and endoscopic remission at baseline (relapse-free survival: 51% vs. 100%). Notably, 100% of patients with continued IFX remained in remission for 48 weeks, while 51% of patients relapsed due to IFX withdrawal. Although the study did not reveal any significant risk factors for relapse after IFX withdrawal, concomitant IM at discontinuation demonstrated a tendency to decrease the risk of relapse. The result of the reintroduction of IFX was not assessed in this study [14].

Another RCT, the SPARE trial would give us an insight into which treatment should be withdrawn first from combination therapy by comparing three groups randomized to continuing combination therapy, withdrawing IFX, or withdrawing IM in CD. The study showed that relapse rates in the 2-year follow-up period were 14%, 40%, and 10%, respectively, suggesting that IFX withdrawal from combination therapy had a significantly higher risk of relapse, while IM discontinuation did not worsen the outcome. After discontinuation of IFX, 22 of 23 (96%) patients who relapsed were able to achieve remission with re-treatment with IFX [12].

Although these brand-new pieces of evidence showed favorable responses for re-induction with the same treatment, which are largely consistent with a previous meta-analysis for re-induction therapy (82% of CD and 85% of UC regained response) [8], it should be taken into account that anti-TNF agent withdrawal may increase the development of anti-drug antibodies that could result in infusion reaction and loss of response in the long-term perspective. On the other hand, a recent systematic review reported that the rate of clinical relapse after dose reduction and prolongation of intervals of biologics are similar to that after discontinuation [33]. The FREE study is being conducted as the first prospective, interventional study to assess the efficacy of anti-TNF dosing interval lengthening in adolescents and young adults with IBD. FC and trough levels of anti-TNF agents are monitored every four and 12 weeks, respectively, as outcomes of how they affect clinical relapse at week 58. This study will provide us with consequential evidence of the effect and monitoring strategy after interval lengthening [34].

The European Crohn’s and Colitis Organization topical review (Exit Strategies) recommends intensive follow-up during the first year after withdrawal since most relapses occur within 6–12 months after withdrawal, especially with anti-TNF agents [18]. In fact, the difference in the remission rate between the continued and discontinued IFX group in the HAYABUSA study became significant as early as week 16 [11]. Furthermore, the STOP-IT study showed approximately 60% of patients who discontinued IFX relapsed within 3 months after stopping IFX [14].

A recent prospective study monitored after the withdrawal of IM monotherapy demonstrated that FC showed the highest sensitivity for relapse among other biomarkers including CRP and WBCs [20]. In another prospective cohort study, IBD patients who relapsed after withdrawal of anti-TNF agents showed elevated FC levels at a median of 3 months before relapse [35]. In addition, an observational study in a prospectively enrolled database demonstrated that FC >200 µg/g was strongly predictive of future clinical relapse during post-discontinuation monitoring [36].

Based on these pieces of evidence, it would be advisable to monitor FC every 3 months, especially in the early post-discontinuation period to identify an early sign of relapse. Then, if they are found to be positive, further imaging should be planned. However, a larger prospective study is needed to determine the optimal monitoring strategy after treatment withdrawal.

Interestingly, Pierre et al. [37] demonstrated that different proteins experienced by shotgun proteomics were associated with different phases of relapse in CD patients after IFX withdrawal from combination therapy. For example, acute-phase reactants and complement factors such as mannan-binding lectin serine protease 2 were associated with higher short-term relapse risk (<6 months) while low levels of JCHAIN which incorporated IgA were associated with long-term one (>6 months), suggesting distinct pathophysiological mechanism. This approach could help establish new biomarkers to predict relapse after withdrawal.

Histological examination may be useful to evaluate inflammation and risk of relapse after treatment withdrawal especially in UC, as suggested by the HAYABUSA study, while the current evidence is insufficient [11]. On the contrary, the significance of histological findings of biopsy in CD has not been fully explored due to the difficulty of transmural access. Dysbiosis, defined as the decrease in bacterial biodiversity, may play an important role in the pathogenesis of IBD. A recent study demonstrated that a lower proportion of Firmicutes was shown to be associated with a higher rate of relapse after IFX discontinuation in CD, suggesting microbiome analysis could predict future relapse after withdrawal [38]. Furthermore, the concept of disease modification has been proposed in recent years [39]. Early disease modification might change the natural history of the diseases and improve the remission rate after withdrawal. Further prospective longitudinal studies are needed to address this.

Should we focus just on the relapse rate? The long-term safety (e.g. infection, cancer risk), chemoprevention, and impact on healthcare costs should be also taken into account [4, 40, 41]. In addition, it remained unclear whether discontinuation of treatment truly improves the patient’s quality of life. Finally, the difference in natural history between UC and CD should also be considered. CD is considered to be a more progressive disease in which stricture and fistula could develop due to accumulative damage to the digestive tract, resulting in bowel resection. Thus, the risk of short-term relapse and the efficacy of re-treatment may not be sufficient to justify withdrawal, especially in CD. Further studies with a multidimensional approach are needed to clarify the true impact of treatment withdrawal.

Understanding each piece of evidence and the contribution of the treatment to each patient is imperative. Both the advantages and disadvantages of the treatment should be carefully discussed via a shared decision-making approach.

Yusuke Miyatani has served as a speaker of AbbVie; received research funding from Japan Foundation for Applied Enzymology; and has served as an endowed chair of AbbVie, JIMRO, Zeria Pharmaceutical, Kyorin Pharmaceutical, Mochida Pharmaceutical, Otsuka Holdings, and EA Pharma. Taku Kobayashi has served as a speaker, a consultant, and an advisory board member of AbbVie, Ajinomoto Pharma, Asahi Kasei Medical, Astellas, Alfresa Pharma, Celltrion, Covidien, EA Pharma, Eisai, Eli Lilly, Ferring Pharmaceuticals, Gilead Sciences, Janssen Pharmaceutical, JIMRO, Kyorin Pharmaceutical, Mitsubishi Tanabe Pharma, Mochida Pharmaceutical, Nippon Kayaku, Pfizer, Takeda Pharmaceutical, Thermo Scientific, and Zeria Pharmaceutical; and received research funding from AbbVie, Alfresa Pharma, Asahi Kasei Medical, Activaid, Bristol Myers Squibb, JMDC Inc., EA Pharma, Kyorin Pharmaceutical, Mochida Pharmaceutical, Nippon Kayaku, Otsuka Holdings, Sekisui Medical, Thermo Fisher Scientific, and Zeria Pharmaceutical.

There is no funding resource for the manuscript.

Concept and design; drafting of the manuscript; and critical revision of the manuscript for important intellectual content: Yusuke Miyatani and Taku Kobayashi.

Data sharing is not feasible since this is a review article.