ECCO Guidelines on Therapeutics in Crohn's Disease: Medical Treatment

1. Introduction

Crohn’s disease [CD] is a chronic inflammatory bowel disease [IBD] that can result in progressive bowel damage and disability.¹ CD can affect individuals of any age, from children to the elderly,^2,3 and may cause significant morbidity and impact on quality of life. Up to one-third of patients present with complicated behaviour [strictures, fistula, or abscesses] at diagnosis.⁴ Most patients over time will develop a complication, with roughly 50% of patients requiring surgery within 10 years of diagnosis.^5–7 As the precise aetiology of CD remains unknown, a curative therapy is not yet available.⁸ Several agents are available for the medical treatment of CD. Medical agents include mesalazine [5-ASA], locally active steroids [such as budesonide], systemic steroids, thiopurines such as azathioprine [AZA] and mercaptopurine [MP], methotrexate [MTX], and biologic therapies (such as anti-tumour necrosis factor [TNF], anti-integrins, and anti-interleukin [IL] 12/23].

The European Crohn’s and Colitis Organisation [ECCO] produces and regularly updates several guidelines aimed at providing evidence-based guidance on critical aspects of IBD care to all health care professionals who manage patients with IBD. To provide high-quality evidence-based recommendations on medical treatment in CD, ECCO decided to develop these guidelines by adopting the GRADE [Grading of Recommendations Assessment, Development, and Evaluation] approach.⁹ GRADE is a systematic process for developing guidelines which addresses how to frame the health care questions, summarise the evidence, formulate the recommendations, and grade their strength and the quality of the associated evidence. GRADE increases transparency at all levels of this process and makes explicit the three considerations that lead to a particular recommendation: the quality of the evidence, the balance of benefits and harms, and the patients’ values and preferences. Therefore ECCO reviewed the available high-quality evidence on the medical management of CD and developed evidence-based recommendations on the medical treatment of adult patients with CD. These guidelines do not cover specific situations, such as postoperative management of adult patients with CD, which was already covered in the latest ECCO Guidelines on Crohn’s disease.¹⁰

2. Methods

Based on the GRADE workflow, the Guidelines Committee of ECCO [GuiCom] selected a panel of 48 experts supported by a team of methodologists and librarians. Selection was based on IBD expertise, scientific background, and knowledge of the GRADE methodology. All panellists received adequate training in GRADE before starting the process. Additionally, four patients with CD representing the European Federation of Crohn’s and Colitis Associations [EFCCA] were invited to participate in all face-to-face meetings and to provide their experiences and state their preferences.

Three domains for medical treatment of CD were identified:

• 1] induction therapy;

• 2] maintenance therapy;

• 3] therapy of fistulising perianal disease.

All panellists were assigned to one of three working groups coordinated by one to two working group leaders under the supervision of two Guideline coordinators. The panellists first formulated a series of specific questions using the PICO format [Population, Intervention, Comparator, Outcomes] which were deemed to be clinically important for the medical treatment of CD. The outcomes of all PICO questions were subsequently graded as ‘not important’, ‘important’, or ‘critical’ during a face-to-face kick-off meeting in Vienna, using a Delphi consensus process.

A team of professional librarians performed a comprehensive literature search on EMBASE, PubMed/Medline, and Cochrane Central databases using specific search strings for each PICO question [Supplementary Files 1, 2, and 3, available as Supplementary data at ECCO-JCC online]. Two independent working group members [one assigned to the PICO and another one from the same group as a second reviewer] assessed the relevance of each abstract to PICO and included or excluded all the relevant papers for the final data extraction and analysis. Subsequently, the working group members assigned to each PICO question systematically reviewed and summarised the evidence on every outcome, to compile a Summary of Findings [SoF] table for each question. The GRADE method follows a hierarchical approach to synthesise evidence; recent high-quality systematic reviews and meta-analyses of clinical trials were preferentially used to create the recommendations. When these were not available, individual randomised clinical trials [RCTs] followed by observational studies were reviewed; results of individual studies were pooled using random-effects meta-analysis as appropriate and when needed. To define disease activity and severity [mild-to-moderate and moderate-to-severe], we accepted the definitions used by the investigators of the studies selected as an evidence basis for our work.

The quality of evidence was classified into the following four categories in accordance with the GRADE approach: ‘high’ [meaning that further research is unlikely to change our confidence in the effect estimates], ‘moderate’ [further research may change our confidence in the effect estimates], ‘low’ [further research likely to change our confidence in the effect estimates], and ‘very low’ [meaning that any estimate of effect is very uncertain].⁹ For each PICO question, the quality of evidence was equal to the lowest quality of evidence among those outcomes graded as ‘critical’. The strength of each recommendation was graded as either ‘strong’ [meaning the desirable effects of an intervention clearly outweigh the undesirable effects, or vice versa] or as ‘weak’ [meaning the balance is less certain], considering also the quality of evidence, values or preferences, and resource use. Whenever the chosen outcomes were not reported in the clinical trials, this was indicated in the corresponding SoF table. To support the recommendations, we either extracted summary effect estimates from the preselected systematic reviews or our group of methodologists directly performed the comparisons. All recommendations were subject to online voting by the panel members, the ECCO National Representatives [two for each country affiliated with ECCO], and 13 additional reviewers from a list of ECCO members who applied to the open call but were not selected to be part of the Working Groups [see Acknowledgements section]. The final version of all statements/recommendations was discussed among panel members during a final consensus meeting in Vienna and put to a vote; final recommendations were approved if at least 80% of the panellists agreed with the statement and its associated strength grading. The list of statements, the supporting text and material, and the draft of the manuscript were critically reviewed by two external Guideline Committee members and by the ECCO Governing Board members, who also approved the final version of these Guidelines.

The literature search strategies, the relevant definitions of patient populations and outcomes, a detailed description of the process, and the SoF tables summarising the evidence can be found in the Supplementary Material, available as Supplementary data at ECCO-JCC online.

3. General approach to the management of Crohn’s Disease

As CD is a lifelong disease, therapy aims to induce remission in the short term and maintain remission in the long term. The recognition that chronic and untreated inflammation [even if asymptomatic] ultimately results in poor outcomes^11–14 has led to a recent paradigm shift in medical treatment and disease monitoring; it is nowadays recognised that early intervention and intensive monitoring may prevent complications.^15,16 Stratifying patients according to their prognostic risk factors and individualising therapy are crucial steps to optimise patient management, although high-quality evidence is not currently available to support this approach. Many factors affect the choice of medical therapy. These include disease location, disease activity and severity, previous response to therapy, and presence of complications [i.e., perianal or fistulising disease]. In addition, the individual risk factors for progression and complications, the individual patient’s characteristics, and the costs and benefit/risk ratio of each drug should be considered. As there is often a disconnect between clinical symptoms and underlying inflammation, it is of crucial importance to monitor disease and therapy at regular intervals based on objective and measurable markers [endoscopy, C-reactive protein [CRP], calprotectin, imaging].¹⁷ This approach will provide the clinician with the possibility to adjust therapy if needed, thereby maximising the probability of achieving tight control of the disease and inflammation, which is believed to be essential to prevent disease progression.^16–18 In addition to drug therapy, the management of CD should also involve a series of general health care maintenance measures. Patients should be encouraged to stop smoking, nutritional deficiencies should be corrected, therapy-related side effects [i.e., cancer and infections] should be monitored, and appropriate guidance or surveillance for vaccinations, osteoporosis, and sun protection should be implemented, as detailed in previous ECCO guidelines, topical reviews, or both.^17,19–23

4. Medical management of Crohn’s disease

Section 1 - Induction of Remission

Mild-to-moderate disease

5-ASA compounds

We performed a meta-analysis of seven eligible RCTs that compared the use of oral 5-ASA [five trials]^24–28 or sulphasalazine [two trials]^29,30 with placebo in patients with active CD [Supplementary Material, SoF Table 1, available as Supplementary data at ECCO-JCC online]. The dosage of 5-ASA administered ranged from 1 g to 3.2 g/day; patients with mild-to-moderate disease with ileal, ileo-colonic, or colonic disease were included. Overall, there was no significant effect for induction of clinical remission (relative risk [RR]: 1.28; 95% confidence interval [CI]: 0.97–1.69) [Supplementary Figure 1, available as Supplementary data at ECCO-JCC online]. A recent Cochrane review also found no significant overall effect.³¹ Both 5-ASA and sulphasalazine appeared to be well tolerated in our meta-analysis, as there was no significant increase in withdrawals due to adverse effects [AEs] when compared with placebo [RR: 1.13; 95% CI: 0.73–1.84] [Supplementary Figure 2, available as Supplementary data at ECCO-JCC online].

Among the five trials of 5-ASA alone there was also no benefit over placebo for inducing clinical remission [RR: 1.27; 95% CI: 0.79–2.03] [Supplementary Figure 3, available as Supplementary data at ECCO-JCC online]. No significant increase in withdrawal due to AEs was observed in trials that compared 5-ASA alone versus placebo [RR: 1.0; 95% CI:0.58–1.71] [Supplementary Figure 4, available as Supplementary data at ECCO-JCC online]. One published network meta-analysis noted a small statistically significant effect on clinical remission among the study arms that evaluated 5-ASA at daily doses of >2.4 g/day.³² However, another network meta-analysis was unable to confirm any such dose effect.³³ A pooled analysis of three double-blind placebo-controlled trials of a slow-release preparation of 5-ASA reported a significantly greater reduction in the Crohn’s Disease Activity Index [CDAI] with 5-ASA versus placebo.³⁴ However, the effect size [an 18-point greater reduction in CDAI score comparing 5-ASA and placebo] was not clinically significant.

Two older trials compared sulphasalazine with placebo for induction of clinical remission. A pooled analysis showed a small effect of borderline statistical significance [RR: 1.38; 95% CI: 1.00–1.89] [Supplementary Figure 5, available as Supplementary data at ECCO-JCC online]. This was not accompanied by any significant increase in withdrawals for AEs [RR: 1.88; 95% CI: 0.65–5.47] [Supplementary Figure 6, available as Supplementary data at ECCO-JCC online]. Subgroup analyses in both trials suggested that the efficacy of sulphasalazine was limited to patients with colonic CD.^29,30

The use of topical 5-ASA [enema or suppository] for the treatment of CD has not been studied in RCTs.

Budesonide

A Cochrane systematic review and meta-analysis³⁵ included three RCTs^36–38 that compared budesonide at a dose of 9 mg/day with placebo [Supplementary Material, SoF Table 2, available as Supplementary data at ECCO-JCC online]. Two of these trials^37,38 evaluated clinical response [defined as decrease in CDAI score ≥100 or total CDAI score ≤150] at 8 weeks. Clinical remission [CDAI score ≤150] at 8 weeks was reported in all three RCTs. Budesonide was superior to placebo for inducing clinical response [RR: 1.46; 95% CI: 1.03–2.07] and clinical remission [RR: 1.93; 95% CI: 1.37–2.73] in patients with mildly active CD in the small and/or large intestine limited to the ascending colon. As compared with conventional steroids [e.g., prednisolone], which are usually associated with many systemic side effects, budesonide presented high topical anti-inflammatory activity and low systemic absorption and bioavailability, and therefore had a better safety profile. Budesonide was shown to be safe [AEs; RR: 0.98; 95% CI: 0.77–1.25] in the reviewed meta-analysis.³⁵

A Cochrane systematic review and meta-analysis from 2015 reviewed two RCTs^39,40 that compared budesonide at a dose of 9 mg/day with mesalazine up to 4.5 g/day. More recently, a Japanese trial⁴¹ also evaluated budesonide versus mesalazine in patients with active CD [Supplementary Material, SoF Table 3, available as Supplementary data at ECCO-JCC online]. All trials evaluated clinical response [decrease in CDAI ≥ 100 or total CDAI ≤ 150] and clinical remission [CDAI ≤ 150] at 8 weeks. Budesonide was not superior to mesalazine for inducing clinical remission [RR: 1.30; 95% CI: 0.98–1.72] in patients with mildly active CD in the small and/or large intestine [Supplementary Figure 7, available as Supplementary data at ECCO-JCC online]. Nevertheless, clinical response was more frequently seen in patients receiving budesonide [RR: 1.22; 95% CI: 1.03–1.45] than in patients receiving mesalazine [Supplementary Figure 8, available as Supplementary data at ECCO-JCC online]. The safety profile of both compounds was comparable, with similar AE [RR: 0.91; 95% CI: 0.79–1.05] and serious AE [RR: 0.94; 95% CI: 0.24–3.75] rates in both intervention groups [Supplementary Figures 9 and 10, available as Supplementary data at ECCO-JCC online].

Antibiotics

Numerous studies have studied the efficacy of antibiotic treatment on luminal CD. Metronidazole, ciprofloxacin, and anti-mycobacterial regimens have been extensively studied. Overall, none has demonstrated efficacy to consistently induce clinical remission or mucosal healing compared with placebo.^42–44 In addition, side effects limit the use of these therapies; recently, the European Medicines Agency has imposed restrictions on the use of ciprofloxacin due to disabling or potentially permanent events [EMA/668915/2018]. Therefore a recommendation was not made specifically on antibiotics to treat luminal CD, although they remain indicated for the treatment of septic complications.

Moderate-to-severe disease

Systemic corticosteroids

Two RCTs reported on the efficacy of systemic corticosteroids [oral prednisolone or oral methylprednisolone] compared with placebo for the treatment of moderately-to-severely active CD^29,30[Supplementary Material, SoF Table 4, available as Supplementary data at ECCO-JCC online]. Oral methylprednisolone was administered at a dose of 48 mg/day and reduced on a weekly basis to 32 mg, 24 mg, 20 mg, 16 mg, and 12 mg.²⁹ Doses of oral prednisolone ranged from 0.50 to 0.75 mg/kg with a maximum daily dose of 60 mg.³⁰ Prednisolone is usually tapered at 5 mg/week over an 8-to 12-week period. Data from these studies have been synthesised in a Cochrane systematic review.⁴⁵

One trial involving 105 patients reported on induction of clinical response.²⁹Clinical response was more common in patients receiving methylprednisolone as compared with placebo [93.6% vs 53.4%; RR: 1.75; 95% CI: 1.36–2.25]. Corticosteroids were reported to be twice as effective in inducing clinical remission than placebo in the two studies involving 267 patients [RR: 1.99; 95% CI: 1.51–2.64].⁴⁵ Data on the proportion of patients experiencing AEs from the use of systemic corticosteroids was available from one trial involving 162 patients treated with oral prednisolone.^30,46 The frequency of AEs was 5-fold higher in patients receiving corticosteroids compared with placebo [31.8% vs 6.5%; RR: 4.89; 95% CI: 1.98–12.07]. Steroid-related AEs included Cushing syndrome, acne, infection [increased risk of abdominal and pelvic abscesses in patients with CD], ecchymoses, hypertension, diabetes mellitus, osteoporosis, cataracts, glaucoma, and growth failure in children. Imprecision was serious for the outcomes considered, due to sparse data, which yielded a moderate quality of evidence overall.

Immunosuppressants

Thiopurines

Several studies have reported on the use of thiopurines compared with placebo for induction of remission and response in CD^30,47–53 [Supplementary Material, SoF Table 5, available as Supplementary data at ECCO-JCC online]. Five trials evaluated the use of thiopurines for induction of clinical remission [12–17 weeks] in comparison with placebo^30,47,48,51 [using CDAI or Harvey-Bradshaw index]. Overall, 380 patients were analysed. The active comparator was AZA in four of these trials,^30,47,51 and the active drug was MP in the remaining trial.⁵⁴ The trials were heterogeneous in terms of study design, follow-up time, definition of active disease, and definition of remission. Except for Summers et al.,³⁰ most of the trials allowed for the use of concomitant steroids. The pooled analysis was performed on an intention-to-treat basis and revealed no differences for induction of remission between thiopurines and placebo; 48% [95/197] in the active intervention compared with 37% [68/183] in the placebo group achieved remission [RR: 1.23; 95% CI: 0.97–1.55].

Three trials reported on clinical response, albeit not with standardised measures of disease activity.^49,52,53 In these trials, different types of physician global assessment of disease improvement [clinical response] were used.^49,52,53 Overall, 42.8% of the patients receiving thiopurines, as compared with 26.9% of those receiving placebo, showed clinical improvement. The RR of obtaining clinical response was 1.87 [95% CI: 0.44–7.96]. Heterogeneity was serious [I² = 69%] and imprecision very serious due to sparse data and wide confidence intervals; thus the quality of evidence was very low for this outcome [Supplementary Figure 11, available as Supplementary data at ECCO-JCC online].

Only one trial reported on AEs during induction.⁵¹ The pooled RR of any AEs was not significantly different between placebo and thiopurines [86% vs 69%; RR: 0.81; 95% CI: 0.64–1.02]. Serious AEs were reported in two trials^30,51 including 125 patients; 13.5% of those receiving AZA versus 3.8% of those receiving placebo developed serious AEs [pooled RR: 2.57; 95% CI: 0.92–7.13]. The quality of evidence was deemed low due to a very low number of events [n = 19] and wide confidence intervals.

One study reported on a validated quality of life measure [Inflammatory Bowel Disease Questionnaire: IBDQ].⁵¹ The greatest difference between groups was observed at Week 4 [43% for AZA and 30% for placebo]. Regarding biochemical improvement, only some of the trials reported on changes at the end of the induction period; no dichotomous data were available that allowed for a pooled analysis calculation. Overall, most trials reported no differences in biomarkers of inflammation such as erythrocyte sedimentation rate [ESR], CRP, or orosomucoid in those receiving thiopurines as compared with placebo.^48,52–54 Reinisch et al.⁵¹ reported a similar proportion of elevated faecal calprotectin at baseline and at Weeks 4 and 12 for the thiopurines and placebo groups. Candy et al.⁴⁷ reported a slight increase of ESR in the group receiving placebo and prednisolone versus a statistically significant decrease in ESR in those receiving AZA and prednisone.

Methotrexate

Only one relevant placebo-controlled RCT was retrieved. In this study,⁵⁵ 141 steroid-dependent patients with active CD were randomised to either 25 mg/week of intramuscular MTX or placebo for 16 weeks, with a concomitant daily dose of prednisolone [20 mg at initiation] that was reduced over a 3-month period [Supplementary Material, SoF Table 6, available as Supplementary data at ECCO-JCC online]. After 16 weeks, a significantly larger proportion of patients treated with MTX were in clinical remission than placebo [RR: 2.06; 95% CI: 1.09–3.89]. The rate of treatment discontinuation for AEs [mainly liver enzyme elevations and nausea] was significantly higher in comparison with placebo [RR: 8.00; 95% CI: 1.09–58.51]. However, this study is strongly limited by imprecision and some confounding factors, such as the concomitant use of steroids. No studies were found that compared MTX monotherapy versus placebo for the induction of remission of moderate-to-severe CD. No agreement was achieved in the Consensus regarding the use of MTX for inducing CD remission, and therefore the Consensus decided to make no recommendation.

Three small and heterogeneous studies compared the efficacy of MTX and thiopurines for induction of remission in CD^56–60 [Supplementary Material, SoF Table 7, available as Supplementary data at ECCO-JCC online]. These studies used different dosages and routes of administration. Two studies used oral MTX at doses of 16 mg/week⁵⁹ and 12.5 mg/week,⁵⁶ and one used intravenous [IV] MTX at 25 mg/week.⁶⁰ All patients were steroid-dependent and received systemic steroids at inclusion. None of the individual studies or the pooled analysis demonstrated a significant difference in the likelihood to achieve remission [RR: 0.87; 95% CI: 0.70–1.09] [Supplementary Figure 12, available as Supplementary data at ECCO-JCC online]. Although the risk of AEs is higher with MTX, the data are very sparse and the quality of evidence is very low for both outcomes. Accordingly, no recommendation can be made.

Based on the current evidence, agreement on a recommendation for the use of MTX for inducing clinical remission in patients with CD could not be reached. However, MTX may be considered as an option for steroid-dependent patients with moderate-to-severe disease when alternative options [including surgery] cannot be used. The need to stop therapy in patients planning a pregnancy must be considered.⁶¹

Monoclonal antibodies

Monoclonal antibodies directed against TNF-α are fast-acting and potent anti-inflammatory agents. Anti-TNF therapies approved for the treatment of CD include infliximab, adalimumab, and certolizumab pegol [the latter is not approved in the European Union for CD, but is commercially available in Switzerland and Russia]. Infliximab is a chimeric mouse-human immunoglobulin [Ig] G1 monoclonal antibody administered intravenously at a dose of 5 mg/kg at 0, 2, and 6 weeks during induction and every 8 weeks thereafter. Adalimumab is a fully humanised IgG1 monoclonal antibody given subcutaneously [SC] at a dose of 160 mg, and then 80 mg 2 weeks after induction, followed by 40 mg SC every 2 weeks. Certolizumab pegol is a PEGylated Fab fragment against TNF-α, self-administered SC at a dose of 400 mg at Weeks 0, 2, and 4, followed by 400 mg every 4 weeks thereafter.

Data on anti-TNF agents versus placebo [infliximab, adalimumab, and certolizumab pegol] from several meta-analyses of RCTs^62–64 support their efficacy for induction of clinical remission [RR: 1.6; 95% CI: 1.17–2.36] and clinical response [RR: 1.43; 95% CI: 1.17–1.73] [Supplementary Material, SoF Table 8, available as Supplementary data at ECCO-JCC online] in patients who did not achieve adequate response or were intolerant to corticosteroids and/or immunosuppressants. Limited endoscopic data were available for the induction period; two studies showed a non-significant trend towards enhanced mucosal healing [RR: 3.25; 95% CI: 0.53–19.8].^65,66 However, the evidence was downgraded due to imprecision. Data on clinical remission were highly heterogeneous [I² = 63%], and data on endoscopic improvement were affected by high imprecision due to the low number of patients included in the meta-analysis [n = 35]. Data on patient-reported outcomes [PRO] response and remission, biochemical and radiological improvement, and quality of life are insufficient. There was no difference in terms of AEs [RR: 0.99; 95% CI: 0.90–1.08].

The choice of anti-TNF agent depends on patient preference, availability, cost, and accessibility. However, in a 2015 network meta-analysis, pairwise comparison revealed that infliximab with AZA [OR: 3.1; 95% CI: 1.4–7.7] and adalimumab monotherapy [OR: 2.1; 95% CI: 1–4.6] were superior to certolizumab pegol for induction of remission.⁶⁷

The timing of introduction of biologic agents is a matter of debate. It has been suggested that patients presenting with poor prognostic factors [e.g. fistulising perianal disease, extensive disease, deep ulcerations, complicated phenotype] would benefit from the early introduction of anti-TNF to achieve a reduced risk of surgery, hospitalisation, or development of disease-related complications.¹⁵ Furthermore, anti-TNF agents might be more effective if introduced earlier [in the first 2 years] in disease course,^68–72 although these results are based on post-hoc analyses from clinical trials.

Only one RCT [the DIAMOND trial]⁷³ studied the use of combination therapy of adalimumab with thiopurine as compared with adalimumab monotherapy for the induction of clinical remission in patients naïve to both therapies [Supplementary Material, SoF Table 9, available as Supplementary data at ECCO-JCC online]. In this trial, combination therapy was not superior to adalimumab monotherapy for inducing clinical remission [RR: 0.95; 95% CI: 0.78–1.15]. However, combination therapy was associated with endoscopic improvement at Week 26 [RR: 1.32; 95% CI: 1.06–1.65], although this benefit was lost at the end of 1 year. There was no increase in AEs leading to discontinuation associated with combination therapy [RR: 1.03; 95% CI: 0.60–1.78]. Of note, the dose of AZA used in this trial was lower than the usual dose used in CD patients [25–100 mg/day instead of 2–2.5 mg/kg/day].

The SONIC [Study Of Biologic and Immunomodulator Naive Patients In Crohn’s Disease] RCT⁷⁰ compared the efficacy of infliximab combined with AZA over infliximab monotherapy in patients naïve to both therapies, who failed to respond to steroids or 5-ASA [Supplementary Material, SoF Table 10, available as Supplementary data at ECCO-JCC online]. Combination therapy resulted in higher rates of clinical remission at Week 26 as compared with infliximab monotherapy [RR: 1.64; 95% CI: 1.07–2.53]. Combination therapy was also more likely to result in mucosal healing at this timepoint [RR: 1.82; 95% CI: 1.01–3.26]. There was no difference in AEs for those receiving combination therapy. Rather, there were significantly lower rates of serious AEs in those receiving combination therapy [RR: 0.56; 95% CI: 0.32–0.97].

A commonly encountered scenario in clinical practice is patients who have failed or have had an inadequate response to thiopurines and in whom anti-TNF therapy is planned. No RCT has directly compared whether in such cases thiopurine maintenance in combination with the anti-TNF would carry additional benefits in terms of efficacy. A post-hoc analysis of RCTs has shown no added benefit of the continued use of immunomodulator therapy after starting anti-TNF therapy in this setting.⁷⁴ However, immunogenicity should be considered and, in the absence of direct evidence, an individualised approach should be considered.⁷⁴

Ustekinumab is an IgG1 monoclonal antibody that binds to the p40 subunit shared by the pro-inflammatory interleukins 12 and 23.⁷⁵ In CD, induction should be given IV using a weight-based regimen of approximately 6 mg/kg.^75,76 One systematic review and meta-analysis pooled the results from RCTs in which ustekinumab was compared with placebo for induction of remission in patients with moderate-to-severe active luminal CD⁷⁷[Supplementary Material, SoF Table 11, available as Supplementary data at ECCO-JCC online]. Four trials^76,78–80 involving 1947 patients treated with different ustekinumab intravenous doses or equivalent placebo reported induction of clinical response and induction of clinical remission at Week 6. Data were extracted and a meta-analysis was performed, yielding an RR of obtaining clinical response of 1.56 [95% CI: 1.38–1.77] versus placebo [Supplementary Figure 13, available as Supplementary data at ECCO-JCC online]. The quality of evidence was high. The RR of obtaining clinical remission was 1.76 [95% CI: 1.40–2.22] [Supplementary Figure 14, available as Supplementary data at ECCO-JCC online]. The quality of evidence was high. An endoscopic substudy involving 252 CD patients revealed that 47.7% of patients receiving ustekinumab achieved endoscopic improvement at 8 weeks as compared with 29.9% of those receiving placebo [RR: 1.60; 95% CI: 1.13–2.26]. The quality of evidence was moderate. Four trials^76,78–80 reported on AEs [2024 patients] or serious AEs [1947 patients] after induction. The pooled RR of any AEs was not significantly different between ustekinumab and placebo [62.0% vs 63.9%; RR: 0.96; 95% CI: 0.90–1.03] [Supplementary Figure 15, available as Supplementary data at ECCO-JCC online]. Similarly, the pooled RR of any serious AEs was not significantly different between ustekinumab and placebo [5.2% vs 6.4%; RR: 0.79; 95% CI: 0.54–1.15] [Supplementary Figure 16, available as Supplementary data at ECCO-JCC online]; the quality of evidence was high. The rate of antibody drug formation seems to be low [under 5%].⁸¹

Vedolizumab is a monoclonal IgG1 antibody that acts by blocking the α4β7 integrin resulting in gut-selective anti-inflammatory activity.⁸² It is administered intravenously at a fixed dose of 300 mg at 0, 2, and 6 weeks for induction, and every 8 weeks thereafter. Patients who do not achieve response at Week 6 can benefit from an additional administration at Week 10.⁸³ Three randomised trials involving 969 patients treated with vedolizumab or placebo reported on induction of clinical response, induction of clinical remission, and serious AEs in adult patients with moderate-to-severe active CD^82,84,85[Supplementary Material, SoF Table 12, available as Supplementary data at ECCO-JCC online]. Patients in these studies were followed up for 6 to 10 weeks. Clinical remission was more common in patients receiving vedolizumab compared with placebo [RR: 2.01; 95% CI: 1.50–2.71] [Supplementary Figure 17, available as Supplementary data at ECCO-JCC online]. Likewise, clinical response was also more common in patients receiving vedolizumab compared with placebo [40.8% vs 25.7%; RR: 1.55; 95% CI: 1.14–2.11] [Supplementary Figure 18, available as Supplementary data at ECCO-JCC online]. The quality of evidence for these outcomes was high. Rates of serious AEs with vedolizumab were not significantly different with placebo [RR: 0.94; 95% CI: 0.61–1.45] [Supplementary Figure 19, available as Supplementary data at ECCO-JCC online]. The quality of evidence for this outcome was moderate due to serious imprecision arising from sparse data.

One systematic review and meta-analysis performed an indirect comparison of ustekinumab and vedolizumab for induction of remission in patients with moderate-to-severe active luminal CD who were non-responsive or intolerant to previous anti-TNF agents.⁸⁶

Four trials^76,79,82,85 involving a total of 1249 patients treated with ustekinumab or vedolizumab reported on induction of clinical response and clinical remission [Supplementary Material, SoF Table 13, available as Supplementary data at ECCO-JCC online]. The pooled RR of clinical response [35.8% vs 33.1%; RR:1.14; 95% CI: 0.65–1.99] and clinical remission [16.3% vs. 13.3%; RR: 1.16; 95% CI: 0.54–2.48] were not significantly different between ustekinumab and vedolizumab, but the quality of evidence was very low for both outcomes.

Four trials^76,79,82,85 involving a total of 1541 patients treated with ustekinumab or vedolizumab reported on AEs or serious AEs after induction. The pooled RR of any AEs was not significantly different between ustekinumab and vedolizumab [64.2% vs 56.2%; RR: 1.00; 95% CI: 0.82–1.23]. Finally, the pooled RR of any serious AEs was not significantly different between ustekinumab and vedolizumab [7.1% vs 7.7%; RR: 0.95; 95% CI: 0.43–2.12]; the quality of evidence was very low. However, surgery should always be considered as an option in refractory patients.

Key Points for Clinical Practice

Budesonide is effective for the induction of remission in patients with mild-to-moderate CD, defined as a CDAI between 150 and 220, and/or presence of mild lesions at endoscopy, or a Simple Endoscopic Score-CD [SES-CD] ≤6, or a Crohn’s Disease Endoscopic Index of Severity [CDEIS] ≤8 with ileal and/or right colon involvement; 5-ASA compounds and sulphasalazine have no therapeutic effect. There is a knowledge gap on how to treat mild-to-moderate CD localised in different parts of the gastrointestinal tract other than the ileum and right colon, or in patients with extensive disease. Therefore the decision is left to the clinician, who should consider the patient’s individual characteristics, prognostic factors, and cost/benefit ratios of therapies.

Although systemic steroids are effective in inducing remission in moderate-to-severe CD, they are limited by important side effects. Additionally, long-term use of corticosteroids does not prevent disease relapse.^30,87 Therefore we suggest that the presence of corticosteroid dependency or excess [the inability to wean steroids below the equivalent of prednisolone 10 mg/day or budesonide 3 mg/day within 3 months of starting steroids, a relapse within 3 months of stopping steroids, or the need for more than a single course of corticosteroids in 1 year] should all warrant a steroid-sparing strategy. Thiopurines alone are not effective in inducing remission. However, since thiopurines have a slow onset of action [8–12 weeks]⁴³ and are effective for maintaining remission in steroid-dependent CD patients [see Maintenance, 6.2.1., Recommendation 2.2.], they are frequently combined with steroids at the commencement of therapy. In patients with steroid dependency, a combination of steroids and MTX has limited efficacy in inducing remission at Week 16 and is associated with a high risk of AEs. Therefore, this option may be considered only where other medical treatments and surgery are not indicated or are associated with some increased individual risks.⁸⁸

For patients with moderate-to-severe CD [usually defined as a CDAI >220 and/or CDEIS >8 or SES-CD >6] with inadequate response or intolerance to conventional therapy [steroids and/or thiopurines], we recommend the use of monoclonal antibodies. These include anti-TNF agents [such as infliximab, adalimumab, and certolizumab pegol], ustekinumab, or vedolizumab. All these agents are effective both in biologic-naïve and -experienced CD populations. The choice depends on patient characteristics and preferences, costs, and local availability.⁸ For the induction of remission, in treatment-naïve patients, the combination of infliximab with thiopurines is more effective than infliximab alone for achieving steroid-free remission.⁷⁰ For adalimumab, no benefit of combination therapy over adalimumab alone was observed in the only RCT performed to date.⁷³ The SONIC trial⁷⁰ demonstrated the superiority of either infliximab alone or the combination of infliximab and AZA over AZA monotherapy or even in combination with steroids; this option should be considered and even preferred in steroid-dependent patients. The REACT [Early Combined Immunosuppression for the Management of Crohn’s Disease] trial showed that the early use of monoclonal antibodies [adalimumab] combined with immunosuppressants in patients at high risk of complications, as compared with a more conventional stepwise management, was associated with significantly lower rates of complications and need for hospitalisation and/or surgery in patients with early CD.¹⁵ A prospective cohort study demonstrated that concomitant immunomodulator use is associated with lower immunogenicity to anti-TNF.⁸⁹ In clinical practice, the potential added efficacy benefit and lower immunogenicity of combination therapy needs to be balanced against a potential increase in AEs in the long term.^90,91 Combination therapy does not seem to be associated with safety concerns, at least in the short term. However, a large nationwide cohort study showed that combination therapy is associated with higher risk for lymphoma and serious infection, as compared with anti-TNF monotherapy. ^90,91 Therefore the decision is left to the clinician, who should consider patient characteristics, costs, risks, and local regulations. Importantly, risk needs to be individualised as specific patient groups, such as the elderly, maybe at higher risk for infections or lymphoma and young males maybe at higher risk for specific complications, such as hepatosplenic T cell lymphoma.^92,93

In patients who fail anti-TNF therapy, ustekinumab or vedolizumab are indicated. There is currently no direct evidence on the comparison between vedolizumab versus anti-TNF and ustekinumab versus anti-TNF in patients treated either with vedolizumab or ustekinumab as a first biologic. No RCTS have specifically assessed the efficacy and safety of these agents when used in combination therapy as compared with monotherapy; however, overall immunogenicity rates seem to be low. Besides, in the originator trials, no difference in efficacy was observed in those patients treated concomitantly with immunomodulator. However, in patients with moderate-to-severe CD with limited disease extent or refractory to at least one monoclonal antibody, surgery should always be considered as an alternative option.

While RCTs evaluate the efficacy of a drug for induction of remission and thereafter for maintaining remission using validated indices of clinical activity, the clinician usually bases his or her choice of first-line therapy not only on symptoms but also on the perceived disease severity [the impact of disease in the individual patient, the cumulative complications and surgical resections, risk factors for complications, the inflammatory burden of disease, and disease course].¹ Therefore, appropriate studies that address the early use of biologics over a stepwise approach, focusing on the prevention of complications and disease-modification outcomes, and that validate risk factors for disease progression [age, extensive disease, upper tract involvement] should be performed. Such studies were identified by this Consensus as very important research gaps.

Section 2 - Maintenance of Remission

5-ASA compounds

Oral 5-ASA compounds have been extensively studied for the maintenance of medically induced remission of CD [Supplementary Material, SoF Table 14, available as Supplementary data at ECCO-JCC online]. No statistically significant benefit has been demonstrated [RR: 1.03; 95% CI: 0.92–1.16] [Supplementary Figure 20, available as Supplementary data at ECCO-JCC online]. Overall, 11 placebo-controlled trials that assessed doses between 1 and 4 g/day were identified. Treatment durations ranged from 4 months to 36 months, with most trials evaluating a 12-month duration of therapy.⁹⁴ There were no significant differences in the proportion of patients experiencing an AE, or withdrawing due to AEs or serious AEs [RR: 1.93; 95% CI: 0.18–21.1]. The safety data were very sparse [three events] and considerably limited this conclusion [Supplementary Figure 21, available as Supplementary data at ECCO-JCC online].

Immunosuppressants

Thiopurines

The effect of maintenance treatment with AZA or MP administered to patients with CD who are steroid-dependent has been investigated in one meta-analysis⁹⁵ [Supplementary Material, SoF Table 15, available as Supplementary data at ECCO-JCC online]. This meta-analysis included data from six trials published between 1971 and 2013.^{30,47,53,96–99} A total of 489 patients treated with AZA [1.0 to 2.5 mg/kg/day] were included and followed for 6 to 18 months. Clinical remission was defined according to different criteria [CDAI in three, disease activity score [DAS] two, in others one]. AZA was superior to placebo for the maintenance of remission in steroid-dependent patients [RR: 1.19; 95% CI: 1.05–1.34].

Safety outcomes were reported in four trials published between 1978 and 2013,^30,96–98 including a total of 556 patients followed for 6 to 18 months. The overall risk of inducing serious AEs during maintenance treatment with thiopurines was significantly higher than with placebo [RR: 2.45; 95% CI: 1.22–4.90]. The rate of serious AEs reported in patients treated with thiopurines versus placebo was 9.0% [22/245] versus 2.9% [9/311]. Pancreatitis, leukopenia, nausea, allergic reaction, and infections were the most frequent serious AEs.

It has been hypothesised that the early introduction of thiopurines could modify disease course. Two studies have evaluated the efficacy of early use of thiopurines: the AZTEC [AZathioprine for Treatment of Early Crohn’s disease in adults]⁹⁸ and the RAPID [Résultat de l’Adjonction Précoce d’ImmunoDépresseurs]¹⁰⁰ trials [Supplementary Material, SoF Table 16, available as Supplementary data at ECCO-JCC online]. The latter has been excluded from our SoF table because it was not conducted against placebo or no treatment. In the AZTEC study, adult patients with a recent [<8 weeks] diagnosis of uncomplicated CD were randomised to receive either AZA or placebo up to Week 76. Only corticosteroids were allowed to treat active disease in this study population. The results were not statistically significant for any of the critical outcomes evaluated. After 76 weeks of treatment, clinical remission did not differ between the two groups [RR: 1.27; 95% CI: 0.94–1.72]; 30 patients treated with AZA [44.1%] and 23 given placebo [36.5%] were in sustained corticosteroid-free remission [RR: 1.21; 95% CI: 0.79–1.84]. The rates of relapse [defined as CDAI score >175] and corticosteroid requirements were similar between groups. Serious AEs occurred in 14 patients [20.6%] in the AZA group and 7 [11.1%] in the placebo group [RR: 1.85; 95% CI: 0.8–4.29].

Methotrexate

Data on the use of parenterally administered MTX are derived from one double-blind, placebo-controlled RCT⁵⁵ where patients were administered weekly intramuscular injections of 15 mg MTX, or placebo of identical appearance, for 40 weeks [Supplementary Material, SoF Table 17, available as Supplementary data at ECCO-JCC online]. Patients with previously active CD, who had entered remission after 16 to 24 weeks of treatment with 25 mg MTX given intramuscularly once weekly, were randomly assigned to receive either MTX at a dose of 15 mg intramuscularly once weekly or placebo, for 40 weeks. No other treatments for CD were permitted. After 40 weeks, the proportion of patients who remained in remission was higher in the MTX group than in the placebo group [65% vs 39 %; RR: 1.67; 95% CI: 1.05–2.67]. Fewer than 50% of the patients in the MTX group had relapsed by the end of the study.

There were no differences in severe AEs in the MTX group [n = 40] as compared with the placebo group [n = 36] over the 40-week observational period [one patient had cervical dysplasia and the other had a viral respiratory tract infection]. Nausea and vomiting occurred more frequently among patients in the MTX group [40% vs 25% in the placebo group]. Although none of the symptoms was severe, one patient discontinued treatment because of these symptoms. No patient had leukopenia of sufficient severity to require withholding treatment or withdrawal from the study. The overall incidence of AEs was similar in both groups.

Monoclonal antibodies

Two systematic reviews analysed the effect of maintenance treatment with anti-TNFs [infliximab, adalimumab, and certolizumab pegol] administered to patients with CD who had achieved disease remission with the same anti-TNF drug^62,63 [Supplementary Material, SoF Table 18, available as Supplementary data at ECCO-JCC online]. Five landmark trials published between 2002 and 2007^71,101–104 were pooled in the meta-analysis from Stidham et al.⁶²; one study was on infliximab, two on adalimumab, and two on certolizumab pegol. A total of 1771 patients were included and followed for 24 to 30 weeks. Four of the five studies included primary responders only, and one study included all subjects. Clinical remission was defined as a CDAI score <150. The overall likelihood of maintaining remission with anti-TNFs versus placebo was 1.78 [95% CI: 1.51–2.09]. The following values were achieved with infliximab: 1.86 [95% CI: 1.21–2.86]; with adalimumab: 2.06 [95% CI: 1.59–2.82]; and with certolizumab pegol: 1.62 [95% CI: 1.30–2.02]. A network meta-analysis⁶² found no statistically significant differences between the three agents.

There are no pooled data available on serious AEs of all anti-TNFs as against placebo. In a network analysis performed in the framework of a Cochrane collaboration,¹⁰⁵ the dose-adjusted odds ratios [Ors] [95% CI] for SAEs for adalimumab, infliximab, and certolizumab pegol were 1.01 [0.64–1.59], 1.13 [0.79–1.62], and 1.57 [0.96–2.57], respectively. Thus monotherapy with anti-TNFs is considered safe as compared with placebo for the maintenance of remission in CD patients, although the relatively small sample size and short follow-up of RCTs do not allow the detection of AEs that may appear in larger and longer observational studies.

Vedolizumab monotherapy, given IV at 300 mg every 8 weeks, was superior to placebo in maintaining clinical remission in patients with moderate-to-severe CD who achieved remission with vedolizumab [RR: 1.81; 95% CI: 1.26–2.59] [Supplementary Material, SoF Table 19, available as Supplementary data at ECCO-JCC online]. At Week 52, 60/154 patients [39.0%] receiving vedolizumab every 8 weeks and 56/154 patients [36.4%] receiving vedolizumab every 4 weeks were in clinical remission as compared with 33/153 patients [21.6%] receiving placebo [p <0.001 and p = 0.004, respectively]. Moreover, vedolizumab was effective at maintaining steroid-free clinical remission [RR: 2.00; 95% CI: 1.11–3.61] and showed a similar incidence of AEs compared with placebo through week 54 [RR: 1.21; 95% CI: 0.73–2.00].⁸² Longer-term data beyond 52 weeks are required to correctly evaluate the safety profile.

One RCT reported outcomes for the maintenance of remission with ustekinumab in CD patients [Supplementary Material, SoF Table 20, available as Supplementary data at ECCO-JCC online].⁷⁹ Patients responding to ustekinumab in the induction period were re-randomised to receive ustekinumab every 8 or 12 weeks or placebo. Over a 44-week period, 51% of the patients receiving SC ustekinumab were in clinical remission as compared with 35.9% of those receiving placebo [RR: 1.42; 95% CI: 1.10–1.84]. A subgroup analysis demonstrated that at Week 44, clinical remission was achieved by 53.1% of patients receiving ustekinumab every 8 weeks and by 48.8% of patients receiving ustekinumab every 12 weeks, as compared with 35.9% in the placebo group. The difference between treatment every 8 weeks and placebo was 17.2% [95% CI: 5.3–29.2] and was 13% between treatment every 12 weeks and placebo [95% CI: 1.1–24.9]. Therefore there was no difference between ustekinumab administered every 8 or 12 weeks. At 44 weeks, corticosteroid-free remission was achieved in 29.8% of patients receiving placebo versus 44.7% of patients receiving ustekinumab [RR: 1.50; 95% CI: 1.12–2.02]. The pooled RR of any AEs was not significantly different between patients who were given placebo and those administered ustekinumab [15.0% vs 11.0%; RR: 0.73; 95% CI: 0.43–1.25].

There are limited data on endoscopic remission, as this was assessed in a subgroup analysis of 70 patients [46 receiving ustekinumab vs 24 receiving placebo] at 44 weeks. There was no statistically significant difference in endoscopic remission between patients in the placebo group as compared with patients in the treatment arm [RR: 2.61; 95% CI: 0.32–21.08].

There were no deaths during the 44 weeks of maintenance. Common AEs were headache, nausea, and arthralgia, with no significant difference in occurrence between the ustekinumab and placebo groups. There was an identical occurrence of non-melanoma skin cancers in the maintenance groups [n = 4 patients in placebo and ustekinumab groups]. Longer-term data beyond 52 weeks are required to correctly evaluate the safety profile.

There are no randomised head-to-head trials comparing vedolizumab or ustekinumab with anti-TNF agents for the maintenance of clinical remission in patients with moderate-to-severe CD who have achieved response or remission with the same agent. A network meta-analysis¹⁰⁶ included nine RCTs [all trials used the CDAI to define clinical remission] with varying follow-up times. The certolizumab pegol trial had a follow-up time of only 26 weeks. All comparisons were indirect [through the placebo, the ‘common comparator’]. Therefore, the quality of evidence was very low. No specific agent was safer than the others in the maintenance phase. Based on efficacy data, there is no evidence to support switching to vedolizumab or ustekinumab in patients who responded to induction treatment with any anti-TNF, or vice versa. There is a clear need to identify biomarkers that could guide therapeutic choices, and to conduct appropriately sized head-to-head trials that could allow for the identification of patient subgroups who would benefit from a given biologic over the other.

Maintenance strategies

Data from two RCTs with a total of 395 patients with CD were used to support this recommendation [Supplementary Material, SoF Table 21, available as Supplementary data at ECCO-JCC online]. These two RCTs showed no advantage of therapeutic drug monitoring [TDM] over clinically based anti-TNF dosing for any of our critical outcomes, namely clinical remission [one study; 62.6% vs 54.9%; RR: 1.14; 95% CI: 0.89–1.47], steroid-free clinical remission [one study; 30.5% vs 40.0%; RR: 0.76; 95% CI: 0.46–1.26], endoscopic remission [one study; 51.2% vs 52.5%; RR: 0.98; 95% CI: 0.68–1.40], biochemical remission [one study; 62.6% vs 54.9%; RR: 1.14; 95% CI: 0.89–1.47], or serious AEs [one study; 34.1% vs 27.5%; RR: 1.24; 95% CI: 0.68–2.23].^107,108

In the TAXIT trial, a total of 273 IBD patients with stable response to maintenance infliximab therapy were randomised either to concentration- or clinically-based infliximab dosing. Both groups were dose-optimised or dose-reduced to achieve a baseline trough level between 3 and 7 μg/mL. This dose-optimisation phase of the study showed that in patients in clinical remission, a trough level <3 μg/mL or >7 μg/mL was observed in 29% and 27% of patients, respectively. No differences in clinical or biochemical remission at 1 year were observed between clinically- [66%] and proactive TDM- [69%] based groups.¹⁰⁷ Nevertheless, the group who received proactive monitoring had fewer relapses during follow-up [7% vs 17%; p = 0.018].

In the TAILORIX trial, 122 biologically naïve patients with CD, treated with an induction combination therapy with infliximab and immunosuppressant, were randomised after 14 weeks to the following three groups: dose intensification based on clinical features, biomarkers, and trough levels of infliximab, with optimisation steps of 2.5 mg/kg [TDM1]; or of 5 mg/kg [TDM2]; or dose intensification based on clinical features alone [control group].¹⁰⁸ The infliximab dose was adapted to maintain a trough level >3 μg/mL. There was no difference in sustained steroid-free clinical remission with mucosal healing [CDAI <150 from Weeks 22 to 54] in the three randomisation arms [33% in TDM1; 27% in TDM2; 40% in control; p = 0.50].

Both studies have important limitations in their study designs,^107,108 which collectively have lowered the strength of our recommendation. The outcomes in both studies were clinical remission but other important issues, such as costs and immunogenicity, also need to be considered. The prospective cohort study PANTS [Personalised Anti-TNF Therapy in Crohn’s Disease Study] showed that anti-TNF failure is highly dependent on low drug concentrations and immunogenicity, and that dose intensification, especially during the induction period, may improve outcomes and treatment success.⁸⁹ Therefore, the Consensus believes that large, well-powered prospective RCTs with adequate stratification of patients are still required.

Reactive TDM refers to the practice of measuring anti-TNF trough level drug concentration and/or antidrug antibodies [ADA] in patients on anti-TNF therapy with active disease, to elucidate the mechanism of loss of response [LOR] and to guide clinical decision making. Reactive TDM was compared with empirical IFX optimisation [based on clinical judgment alone] in only one randomised, controlled, single-blind, multicentre study in a cohort of 69 patients with CD with secondary IFX failure.¹⁰⁹ Patients were randomised to IFX dose intensification [5 mg/kg every 4 weeks; n = 36] or interventions based on serum IFX and IFX ADA levels using the proposed algorithm [n = 33]. There was no difference in regaining clinical response between the TDM-based group [19/33, 57.6%] and the symptom-based group [19/36, 52.8%] [RR: 1.09; 95% CI: 0.71–1.67; p = 0.81] [Supplementary Material, SoF Table 22, available as Supplementary data at ECCO-JCC online].

However, numerous studies have shown a positive association between adequate drug concentration and various clinical outcomes from clinical response to mucosal healing. Based on these observational data, recent clinical practice guidelines and a group of 25 international experts supported the use of reactive TDM, despite recognising the very low quality of evidence.^110,111 Supporting evidence comes from case-control observational studies.^112,113 In a retrospective study of 312 patients with endoscopically active IBD treated with IFX who underwent dose escalation, TDM-based [n = 149] and clinical decision-based [n = 163] cohorts were compared for endoscopic remission and CRP at a median of 6 months after adjustment. Post-adjustment, endoscopic remission was observed in 63% of patients in the TDM cohort as compared with 48% in the non-TDM cohort [p = 0.05]; clinical response was observed in 69% versus 57% [p = 0.01], and there fewer hospitalisations in the TDM group [22% TDM vs 35% non-TDM; p = 0.025].¹¹² In another study, a modified version of the Steenholdt optimisation algorithm,¹¹³ using a cut-off of 3 μg/ml, was applied to a prospective cohort. Clinical response at 12 weeks was compared between this group and a retrospective control group in which dosing decisions were made independently of TDM results. There was no significant difference in clinical outcomes,¹¹³ but the TDM approach was significantly more cost-effective [>10% decrease in costs]. Therefore, the existing limited evidence does not support an association between a reactive TDM strategy and superior clinical outcomes but does suggest a cost savings benefit even after considering biosimilar use.¹¹⁴

We conducted our own meta-analysis to compare the two strategies [i.e., cessation vs continuation of treatment] in 215 CD patients in long-term remission on thiopurine maintenance therapy [Supplementary Material, SoF Table 23, available as Supplementary data at ECCO-JCC online]. Data from four trials were included.^{96,97,115–117} Patients included received AZA from 6 to 42 months before being randomised to continue or stop AZA¹¹⁵ or to continue AZA or placebo.^96,97,116 All studies had a follow-up time of 12 to 18 months. Our results revealed that the RR of clinical relapse is 2.39 [95% CI: 1.38–4.13] [Supplementary Figure 22, available as Supplementary data at ECCO-JCC online]. Our meta-analysis effect estimate for serious AEs was RR 0.32 [95% CI: 0.04–2.92]. Although the data showed a trend towards fewer serious AEs occurring with discontinuation of treatment, the results were not statistically significant [Supplementary Figure 23, available as Supplementary data at ECCO-JCC online]. However, regular assessment for the long-term risks/benefits should be performed considering the long-term safety data from the population base. To summarise, the evidence for the prevention of clinical relapse is in favour of continuation of treatment, as significantly more relapses occurred when the treatment was discontinued; the risk of SAEs was not different between AZA and placebo/no treatment. Data from studies that compared patients receiving AZA versus placebo/no treatment for more than 18 months are lacking, and this represents an important research gap. Data from observational population studies suggest caution and regular monitoring, especially for the risk of non-melanoma skin cancer and lymphoma in patients exposed to long-term treatment with thiopurines.²⁰ The limited follow-up time and the number of patients included in the studies of the meta-analysis are unable to capture AEs and serious AEs that may occur in the long term.

We also reviewed the literature to compare the approach of using long-term, low-dose thiopurines versus drug discontinuation. After an exhaustive literature search, we did not find evidence comparing the two treatment strategies. Only one trial was identified where dose reduction of thiopurines was compared with discontinuing thiopurines in the setting of combination therapy in patients with IBD. The information was incomplete as it was not possible to separate data from ulcerative colitis and CD patients.¹¹⁸ Therefore, no specific recommendation was made.

A Cochrane review¹¹⁷ based on two RCTs^118,119 revealed the same relapse rate among patients who continued combination therapy with AZA [27/56; 48%] or infliximab monotherapy [27/55; 49%] [RR: 1.02; 95% CI: 0.68–1.52] [Supplementary Material, SoF Table 24, available as Supplementary data at ECCO-JCC online]. The same meta-analysis¹¹⁷ analysed the rates of AEs for infliximab versus combination therapy [RR:1.11; 95% CI: 0.44–2.81; very low-quality evidence] or serious AEs [RR: 1.00; 95% CI: 0.21–4.66; very low-quality evidence]. These results are rather uncertain due to an unclear risk of bias. Common AEs in the combination therapy studies included infections, elevated liver values, arthralgia, and infusion reactions. For some infrequent AEs, longer follow-up studies [>12 months] are necessary to correctly evaluate the safety profile. A higher risk of lymphoma exists when anti-TNF agents are combined with conventional immunosuppression. However, the absolute rates remain very low [3.23; 95% CI 1.5–6.9] and were estimated as 1.9 per 10 000 patient-years in one meta-analysis consisting of almost 9000 patients included in the SEER database.¹²⁰

On the basis of a meta-analysis of nine studies on adalimumab by Chalhoub et al.,¹²¹ the data included were re-analysed because the intervention and control groups had to be reversed to match the relevant PICO question. The result of this recalculation did not reveal any differences in maintenance of clinical remission [RR: 1.01; 95% CI: 0.91–1.13] between combination therapy and monotherapy [Supplementary Material, SoF Table 25 and Supplementary Figure 24, available as Supplementary data at ECCO-JCC online]. Whereas this meta-analysis was limited to 1 year of follow-up [Week 56] in the sensitivity analysis, studies with a longer follow-up [>52 weeks] showed similar results. There are no quality data available for steroid-free clinical remission. However in the ADHERE cohort, which is an open-label extension study that prospectively follows up the cohort of patients originally enrolled in the CHARM study on adalimumab,⁷¹ the rates of steroid-free remission were similar in patients with or without concomitant immunosuppression at baseline after 3 years of follow-up.¹²² The meta-analysis by Chalhoub et al.¹²¹ which was re-calculated did not show any differences in serious AEs between monotherapy with adalimumab and combination therapy [RR: 0.88; 95% CI: 0.62–1.26] [Supplementary Figure 25, available as Supplementary data at ECCO-JCC online].

Currently, no randomised controlled study data regarding the withdrawal of anti-TNF therapy in CD patients with inactive disease are available.¹²³ This is true for anti-TNF therapy as monotherapy or when used in a combination therapy regimen. Several observational studies investigated disease course in CD patients following withdrawal of anti-TNF therapy. A prospective study followed 115 patients with CD on combination therapy for at least 1 year, who discontinued anti-TNF after being in steroid-free clinical remission for at least 6 months. The relapse rates at 12 and 24 months were 43.9% ± 5.0% and 52.2% ± 5.2%, respectively.⁷⁰ A systematic review and meta-analysis included 23 observational cohort studies of 920 CD patients and found an overall relapse rate of 44% [95% CI: 36–51%; follow-up range: 6–125 months].¹²⁴ Furthermore, the relapse rate was 38% [95% CI: 13–63%; 126 patients] at 6 months after discontinuation, 40% [95% CI: 33–48%; 813 patients] at 12 months, and 49% [95% CI: 31–68%; 228 patients; range of follow-up 28–125 months] at >25 months. The meta-analysis included studies in children and patients with perianal disease.

Following the aforementioned meta-analysis, 10 observational cohort studies reported relapse rates in accordance with the findings of the meta-analysis.^125–134 Two of these studies represent extensions of studies included in the meta-analysis.^131,132 One study investigated the risk of relapse in patients treated with a combination of anti-TNF and an immunomodulator, who discontinued either of the two drugs.¹³⁴ The study found no difference in relapse rates with regards to the withdrawn drug; that is, 17/55 patients [30.9%] on biologic therapy withdrawal relapsed compared with 4/20 patients [20%] in which the immunomodulator was withdrawn [p = 0.401].

In conclusion, observational studies report that up to half of patients will experience a relapse within the following 12 months after withdrawal. However, in the absence of controlled studies, the evidence surrounding withdrawal of anti-TNF therapy in patients with long-term remission remains scarce and inconclusive. Hence, no recommendation regarding anti-TNF therapy can be made. The management decision therefore lies with the clinician, who should carefully consider the patient’s profile, values, and preferences, and any resource implications.¹³⁵

Key Points for Clinical Practice

Immunosuppressants and biologic agents are the most effective therapies to maintain medically-induced remission in moderate-to-severe CD patients. Aminosalicylates and steroids are not recommended in this setting due to lack of efficacy and long-term risk of serious AEs [steroids]. For patients with mild disease, no data are available which suggest any specific treatment strategy; no therapy and tight monitoring may be considered in this patient population in the maintenance phase.

Our literature search and data analysis showed that immunosuppressants, such as thiopurines and MTX, are recommended to maintain remission in steroid-dependent patients. As discussed in the previous section, the role of adding MTX or thiopurines to steroids for the induction of remission is limited. However, after steroids are stopped, maintenance with thiopurines or MTX [administered parenterally] can be an appropriate strategy. The choice between the two drug classes depends on careful consideration of patient’s individual characteristics and preferences, safety profile, and drug availability. There is low-quality evidence supporting the continuation of thiopurines for long-term remission, as studies that directly compared long-term treatment with AZA, versus no treatment or placebo, did not have follow-up times >18 months. Clinicians should balance the increased risk of relapse of thiopurine discontinuation with the increased risk of AEs. Many observational studies have now reported an increased risk of lymphoma and skin cancer for patients treated with thiopurines.^136,137 Therefore, regular monitoring should be provided to patients continuing thiopurines in the long term. Given the increased risk of AEs due to thiopurines, monoclonal antibodies can also be considered in this particular group of patients.

For CD patients where medically-induced remission has been achieved by a biologic agent-based strategy, the use of the same agent is recommended to maintain remission. There is high-quality evidence in favour of this approach for anti-TNF agents, vedolizumab, and ustekinumab. There is no evidence to support switching to a different monoclonal antibody after treatment induction with a monoclonal antibody that was successful. Longer-term data beyond 52 weeks are required to correctly evaluate the safety profile of monoclonal antibodies, as the relatively small sample size and short follow-up of RCTs does not allow for detection of some AEs, particularly very rare AEs, which may appear in larger and longer observational studies.

The combination of an anti-TNF agent and thiopurines is effective and safe both for induction and for maintenance. The risk of lymphoma with infliximab and thiopurines remains very low, but should be considered¹⁹ and adequately addressed with the same screening and prevention and regular monitoring recommended for thiopurine therapy.¹⁹ Therefore, when remission is achieved with combination therapy with anti-TNF agents, maintenance with the same biologic agent in monotherapy can be suggested.

Routine strategies to monitor and optimise biologic therapy in CD by a TDM approach are not supported by the available controlled evidence, although we recognise the limitations. There is no clear clinical benefit in favour of a proactive or reactive TDM approach, from the current data. However, some recent data suggest that a reactive TDM approach can result in cost savings also in the era of biosimilars,¹¹⁴ potentially justifying the use of such an approach where TDM is available. TDM can at least be used to guide dose optimisation.

There is currently no evidence to give any recommendation regarding dose reduction of thiopurines during maintenance and there is no evidence on the benefits of withdrawing or continuing biologic agents in patients with stable long-term remission, due to the lack of controlled studies. As stated in our Consensus, the decision is left to the clinicians and should be individualised and discussed with the patient, carefully considering the risk of relapse, disease progression and development of complications, and the risks of potential side effects. The long-term management of patients in remission is therefore an important research gap.

Section 3 - Perianal Fistulising Disease

Therapeutic management of complex perianal fistulising disease

Infliximab was the first agent shown to be effective in an RCT for inducing closure of perianal fistulae and for maintaining this response over 1 year. Complete response [defined as the absence of any draining fistulae at two consecutive visits at least 4 weeks apart] was observed in 4/31 placebo patients [12.9%] versus 29/63 infliximab patients [46%] [RR: 3.57; 95% CI: 1.38–9.25¹³⁸] [Supplementary Material, SoF Table 26, available as Supplementary data at ECCO-JCC online]. Subsequently, the ACCENT II trial evaluated the efficacy of infliximab [5 mg/kg every 8 weeks] in a maintenance trial in 195 patients who had a response [defined as a reduction of 50% of draining fistulae in two visits at least 4 weeks apart] at Week 14 after open-label induction treatment with infliximab. A complete response was maintained until Week 54 in 19 of 99 placebo patients [19.2%] versus 33 of 96 infliximab patients [34.4%] [RR: 1.79; 95% CI: 1.10–2.92].¹³⁹ A meta-analysis of the existing data revealed that infliximab was found to be effective in inducing fistula healing [RR: 3.57; 95% CI:1.38–9.25] and in maintaining clinical fistula healing [RR: 1.79; 95% CI:1.10–2.92] with no significant risk of serious AEs as compared with placebo [RR: 1.31; 95% CI: 0.11–15.25] [Supplementary Figure 26, available as Supplementary data at ECCO-JCC online]. A combined evaluation of both RCTs for safety revealed a risk of serious AEs of 18.9% [33/175 patients] in placebo groups versus 11.9% [24/201] in infliximab patients. These data from RCTs have been confirmed in several uncontrolled studies.¹⁴⁰

In clinical practice, infliximab is often used in combination with immunosuppressants, antibiotics, and surgical treatment.^141–144 Some retrospective data suggest that fistula healing is more likely in patients with higher infliximab trough levels, which suggests the need for personalised dosing in this setting.^145,146

Fistula healing in the subgroup of patients with enterocutaneous and/or perianal fistulae at baseline [n = 117] was a secondary endpoint of the CHARM double-blind, placebo-controlled, randomised trial.¹⁴⁷ A subsequent post-hoc analysis that focused specifically on the efficacy of adalimumab over time in this subgroup confirmed the superiority of adalimumab over placebo [RR: 2.57; 95% CI: 1.13–5.84] for fistula healing after 56 weeks¹⁴⁷ [Supplementary Material, SoF Table 27, available as Supplementary data at ECCO-JCC online]. Data from CHARM combined with data from the open-label extension study ADHERE showed that there was no significant increase in serious AEs for patients treated with adalimumab [RR: 1.21; 95% CI: 0.43–3.38].^71,148 Evidence was also sought for maintenance of fistula healing beyond 56 weeks, resolution of perianal sepsis, stoma-free survival, and quality of life; however, data were insufficient. Although we strongly recommend infliximab as first-line biologic therapy in complex perianal CD [Recommendation 3.1], adalimumab may have a role in patients with previous infliximab failure due to immunogenicity [either primary non-responders or secondary loss-of-responders]. The open-label CHOICE trial indeed demonstrated that complete fistula healing [mainly perianal fistula] could be achieved in 39% of patients [34/88] who initiated adalimumab after infliximab failure.¹⁴⁹ This finding has also been reported in a limited case series.¹⁵⁰

We identified a single study⁷⁴ [a pooled analysis of individual data from the intervention arms only of studies] that compared anti-TNF versus placebo. Only a pooled effect estimate was provided [i.e., OR of complete fistula closure in those on immunomodulators vs those not on immunomodulators was 1.10; 95% CI: 0.68–1.78] without further information on numbers of patients by compared group. Therefore, event rates and absolute risk differences could not be calculated. Furthermore, a retrospective study revealed a hazard ratio of 2.58 [95% CI: 1.16–5.6] for fistula healing in favour of the intervention arm [combination infliximab and immunomodulator] in patients with CD naïve to immunosuppressive therapy.¹⁵¹ There is thus insufficient evidence to support a decision for or against the use of immunomodulators in this context. Further research is necessary to reduce uncertainty and may be warranted, considering the anticipated costs and side effects of combination therapy. In particular, we note the evidence in luminal CD, where addition of immunomodulators reduces immunogenicity of long-term anti-TNF therapy. We therefore recommend further research that should focus on the additional treatment effect of combination therapy and the impact on immunogenicity to anti-TNF agents.

No randomised trial has directly assessed the role of ustekinumab in fistula healing. A post-hoc analysis of 238 patients who entered the phase 2 CERTIFI and phase 3 UNITI 1/2 studies with fistulae at baseline has been reported¹⁵² [Supplementary Material, SoF Table 28, available as Supplementary data at ECCO-JCC online]. This study included only patients with perianal fistulae and did not differentiate between simple and complex fistulae. The analysis showed a measurable but statistically insignificant effect of ustekinumab for induction of remission [RR: 1.77; 95% CI:0.93–3.37] but no difference in comparison with placebo was found for maintenance of remission.¹⁵³ We also sought evidence for the effect of ustekinumab on longer-term maintenance of fistula remission, serious AEs, resolution of perianal sepsis, stoma-free survival, and quality of life; however, data were insufficient. Further research is therefore warranted to determine if ustekinumab is beneficial to patients with perianal fistulae.

A post-hoc analysis of 45 patients, who entered the GEMINI 2 study with complex perianal fistulae at baseline, demonstrated a trend in favour of vedolizumab compared with placebo for fistula healing [RR: 2.23; 95% CI: 0.57–8.72] although this result was not statistically significant^153,154 [Supplementary Material, SoF Table 29, available as Supplementary data at ECCO-JCC online]. The interpretation of this study was limited by sparse data [only 13 patients met the endpoint across treatment arms] and specification of fistulae type [perianal in only 74% of patients]. Evidence was sought also for long-term maintenance of clinical fistula healing, serious AEs, quality of life, resolution of perianal sepsis, and stoma-free survival; however, data were insufficient. The only RCT [NCT02630966]¹⁵⁵ that compared two different induction schedules of vedolizumab [300 mg at Weeks 0, 2, 6, 10, and 14 vs 300 mg at Weeks 0, 2, 6, and 14] was prematurely stopped due to slow recruitment and therefore is inconclusive. However, significant differences were observed between the two study groups. The efficacy of vedolizumab for fistula healing remains an important research gap.

Antibiotics are widely used in the treatment of perianal CD, but most published studies are uncontrolled.¹⁴¹ To our knowledge, there is only one RCT that compared placebo with antibiotics in perianal fistulae [Supplementary Material, SoF Table 30, available as Supplementary data at ECCO-JCC online]. Remission at Week 10 was observed in 1/8 [12.5%] placebo patients versus 3/17 [17.6%] patients treated with antibiotics [RR: 1.41; 95% CI: 0.17–11.54].¹⁵⁶Complete healing was observed in 3/10 [30%] patients treated with ciprofloxacin and 0/8 patients treated with metronidazole. Uncontrolled data or data from studies on combination therapy with anti-TNF suggest that ciprofloxacin can improve the efficacy of anti-TNF in the short term with good safety. However, this combination does not impact on longer-term healing rates.^42,157 Despite the lack of evidence to support their role as monotherapy in closing perianal fistulae, antibiotics remain indicated and recommended to treat and control perianal sepsis.

The effect of AZA on fistula healing in complex perianal CD has been numerically reported in RCTs in 18 patients only.^49,52,53,158 A meta-analysis on this limited group of patients demonstrated that AZA is not superior to placebo for fistula healing [RR: 2.00; 95% CI: 0.67–5.93].⁹⁵ A fourth study⁵⁰ reported complete fistula closure in 9/29 [31%] fistulae during MP therapy, in contrast to 1/17 [6%] in placebo-treated fistulae [Supplementary Material, SoF Table 31, available as Supplementary data at ECCO-JCC online]. Nevertheless, these data could not be incorporated in the pooled analysis, as data were reported as number of fistulae closing rather than number of patients who had complete fistulae closing. With the availability of effective anti-TNF agents, the group felt that it would be inappropriate to recommend any further randomised, placebo-controlled, double-blind trial studying the efficacy of AZA in complex perianal fistulae.

Key Points for Clinical Practice

This section contains recommendations on the medical treatment of perianal disease. However, the management of complex perianal disease should be considered together with the concomitant treatment of luminal disease.

For the medical treatment of perianal fistulae, no evidence supports the use of monotherapy with antibiotics or thiopurines. The highest-quality evidence supports the use of infliximab as first choice. In patients refractory or intolerant to infliximab, there is low-quality evidence to support the use of adalimumab. The current evidence is too limited to support the use of ustekinumab and vedolizumab in clinical practice. However, ustekinumab or vedolizumab may be considered in patients where anti-TNFs are ineffective or contraindicated and there are no treatment options, especially when concomitant luminal disease is present. There is insufficient evidence on the use of combination therapy [specifically infliximab] combined with thiopurines. However, this can be considered when chosen as a therapy for concomitant luminal disease or for anti-immunogenicity purposes.

Although there is no randomised study that compared the combination of surgical treatment and infliximab with infliximab alone, joint management and approach by IBD clinicians and surgeons is considered the standard of care for treatment of complex perianal disease. This is important, since control of sepsis and prevention of perianal infections is necessary before starting any treatment that affects the immune system response. Any immunosuppressive treatment must be stopped in case of onset of septic complications in patients with IBD.

5. Conclusion

These recommendations summarise the current evidence on the medical management of patients with CD. Several research gaps have been identified during the revision and analysis of data, which should be addressed by further research. Where evidence is lacking or is very weak and evidence-based recommendations cannot be given, ECCO provides alternative tools, such as Topical Reviews^{21,92,135,159–161} or Position Papers.¹⁶² We state that Guidelines aim to guide the clinicians’ decisions with the best evidence available, but it is up to every clinician to adapt these Guidelines to local regulations and to the patient’s individual characteristics and needs. ECCO will also aim to disseminate these guidelines by educational activities [i.e., educational platforms, ECCO Workshop, e-learning, and e-Guide] and to support any initiative to integrate ECCO Guidelines into clinical practice; the ECCO e-Guide will primarily serve as a resource to examine how the Guideline recommendations can be implemented into daily clinical practice and patient care pathways.¹⁶³ These guidelines will be regularly updated according to the Guideline Committee outline for the update of Guidelines in the future, using the GRADE approach and considering the most recent evidence emerging from clinical research in the field.

Funding

This project was initiated, funded, and supported by the European Crohn’s and Colitis Organisation.

Conflict of Interest

ECCO has diligently maintained a disclosure policy of potential conflicts of interests [CoI]. The conflict of interest declaration is based on a form used by the International Committee of Medical Journal Editors [ICMJE]. The CoI disclosures are not only stored at the ECCO Office and the editorial office of JCC, but are also open to public scrutiny on the ECCO website [https://www.ecco-ibd.eu/about-ecco/ecco-disclosures.html], providing a comprehensive overview of potential conflicts of interest of the authors.

Acknowledgments

We gratefully thank: Paul Freudenberger for the literature search and full text retrieval; Torsten Karge for the support on informatics and on the web Guidelines platform; and Julia Gabriel, Dauren Ramankulov, Karine David, and the ECCO Office for logistical and coordination support. We gratefully thank the EFCCA patient representatives Ciara Drohan, Evelyn Groß, Sanna Lönnfors, and Marko Perovic, who proactively collaborated in the development of these Guidelines. We would like to acknowledge the ECCO National Representatives who acted as external reviewers and reviewed and provided suggestions on the recommendations and supporting text to this document: Filiz Akyuz, Raja Atreya, Manuel Barreiro De Acosta, Dominik Bettenworth, Clas-Göran Björkesten, Ante Bogut, Emma Calabrese, Mirjana Cvetkovic, Pieter Dewint, Srdjan Djuranovic, David Drobne, Dana Duricová, Jérôme Filippi, Christoph Högenauer, Ioannis Kaimakliotis, Gediminas Kiudelis, Maria Klopocka, Ioannis Koutroubakis, Zeljko Krznaric, Hendrik Laja, Alexander Moschen, Gregor Novak, Alexander Potapov, Ilus Tuire, Svetlana Turcan, Willemijn van Dop, Fiona van Schaik, Ana Isabel Vieira, Stephanie Viennot, and Signe Wildt. We would also like to thank the additional reviewers of these Guidelines: Giorgos Bamias, Yago Gonzalez Lama, Marietta Iacucci, Anna Valeryevna Kagramanova, Jost Langhorst, Gaetano Luglio, Annick Moens, Nurulamin Noor, Iago Rodríguez-Lago, Gerhard Rogler, Simone Saibeni, Carsten Schmidt, Tony Tham, and Andres Yarur.

Author Contributions

JT, GF, MA, OZ coordinated the project; SB, TL, and MG-L advised on GRADE methodology, trained the working group members, and performed the analysis of data; GD, TK, JG, TR, AS, and JW coordinated the working groups; all the authors listed contributed to the identification of relevant data and data interpretation, and drafted and discussed the final recommendations; all the authors participated in the final Consensus; GF, JT, SB, GD, TK, JG, and TR drafted this manuscript; all authors, the ECCO Guideline Committee [GuiCom], and the ECCO Governing Board approved the final version of the manuscript.

References