Early Combined Immunosuppression may be Effective and Safe in Older Patients With Crohn's Disease

Post Hoc Analysis of REACT

Siddharth Singh; Larry W. Stitt; Guangyong Zou; Reena Khanna; Parambir S. Dulai; William J. Sandborn; Brian G. Feagan; Vipul Jairath


Aliment Pharmacol Ther. 2019;49(9):1188-119. 

In This Article

Materials and Methods

Data Source

In the REACT trial, between March 2010 and October 2013, 41 community gastroenterology practices from Belgium and Canada were randomly assigned to either early combined immunosuppression (n = 22) or conventional management (n = 19).[13] These practices enrolled up to 60 consecutive adult patients with Crohn's disease, regardless of baseline disease activity or medication exposure, who had been seen within the previous 12 months or who attended the clinic during the study period. A computer-generated minimisation procedure balanced the treatment effects of country and practice size (≤100 Crohn's disease patients treated annually) between the intervention groups. Because of the complex multidrug algorithm that was evaluated, masking of participants, providers or assessors to group assignment was not possible. To minimise the risk of bias related to the open-label nature of the trial, the outcome measures were objective and, to avoid the potential for treatment contamination, individual practices were unaware of the assignment of other practices.[14] The original trial was approved by the Canadian Shield Research Ethics Board, and institutional consent was obtained in Belgium. All participants provided written informed consent.

The original trial was designed to evaluate the impact of a strategy of algorithmic early combined immunosuppression approach vs conventional management in achieving corticosteroid-free clinical remission and risk of composite major adverse outcomes including time to Crohn's disease-related surgery, hospitalisation or Crohn's disease-related complications, over 24 months. The key findings of the trial were: (a) there was no significant difference in the rate of steroid-free clinical remission in the two groups (early combined immunosuppression vs conventional management: 66% vs 62%) at 12 months, which was the primary outcome of the study; (b) the time to occurrence of Crohn's disease-related surgery and serious Crohn's disease-related complications (abscess, fistula, stricture, serious worsening of disease activity or extra-intestinal manifestations) were significantly less in practices randomised to early combined immunosuppression (hazard ratio [HR], 95% CI: 0.69 (0.50-0.97) and 0.73 (0.61-0.87) respectively); and (c) there was no difference in serious drug-related adverse events including infections between the two groups (0.9% vs 1.1%).


For this post hoc analysis, patients were classified as younger (<60 years) or older (≥60 years, population of interest) based on age at the time of enrolment into the trial.

Intervention vs Comparator

In the early combined immunosuppression treatment strategy arm, specific education on the treatment algorithm was provided to the practices (Figure 1). In these practices, patients with failure to achieve clinical remission (defined as Harvey Bradshaw Index score ≤4) after initiation of corticosteroids (4 weeks in Canada, 12 weeks in Belgium) received combination therapy with a tumour necrosis factor-α (TNFα) antagonist and anti-metabolite (thiopurines or methotrexate). Disease activity was reassessed at 12-week intervals, and patients with persistently active disease (Harvey Bradshaw Index score >6) were sequentially escalated in a stepwise algorithm (escalation of TNFα antagonist, switching anti-metabolite, switching TNFα antagonist and finally surgery) until clinical remission was achieved.

Figure 1.

Early combined immunosuppression algorithm in randomised evaluation of an algorithm for Crohn's treatment (REACT)

Practitioners assigned to conventional management were unaware of the early combined immunosuppression algorithm details and managed patients according to usual practice. Patients at conventional management sites were treated according to the usual practice of their physicians.


For this study, primary effectiveness outcomes were achieving corticosteroid-free clinical remission by 24 months and time to composite major adverse outcome of Crohn's disease-related surgery, hospitalisation or development of serious complications (abscess, fistula, stricture, serious worsening of disease activity, extra-intestinal manifestations or serious drug-related complications). Secondary effectiveness outcomes were individual components of the composite outcome—time to disease-related hospitalisation, surgery, disease-related complications (abscess, fistula, stricture), serious worsening of Crohn's disease activity (increased bowel frequency, abdominal pain or rectal bleeding reported by the site investigator that required a treatment intervention) and extra-intestinal manifestations. Safety outcomes of interest were drug-related adverse outcomes and mortality.

Statistical Analysis

For dichotomous outcomes, we used modified Poisson model for correlated binary data.[15] For time-to-event endpoints, we used marginal Cox proportional hazards models in which adjustment was made for practice size and country and for the effect of clustering by practice. To assess potential confounders, the association between demographic and baseline characteristics (sex, smoking status, disease duration, disease location, fistula involvement, previous surgery for Crohn's disease, Harvey Bradshaw Index, corticosteroid use, anti-metabolite, use, TNFα antagonist use, combination anti-metabolite and TNFα antagonist use, and quality of life indices including Short Form-36 version 2.0 [SF-36] and European Quality of Life Index Version 3L [EQ5D]) and outcomes were examined. Those univariable associations found to be significant at the 0.10 level were then forced into the model. Characteristics were then removed in a stepwise manner, retaining those characteristics that were significant at the 0.10 level. For each outcome, we compared whether the magnitude of effect of intervention (early combined immunosuppression vs conventional management) was different between younger and older patients with Crohn's disease, via the interaction between age group and treatment group in each model, adjusting for those baseline characteristics retained in the backward elimination process. Since the original trial was not specifically designed to answer the question of whether age modifies the effect of early combined immunosuppression in patients with Crohn's disease, these results should be considered exploratory. All analyses were conducted using SAS version 9.4 (SAS Institute Inc, Cary, North Carolina, USA).