REACTMust knowCD · Treatment strategy
Early combined immunosuppression for the management of Crohn's disease (REACT): a cluster randomised controlled trial
Khanna R, et al.
Lancet 2015 · PMID 26342731 (PubMed, opens in a new tab) · doi 10.1016/S0140-6736(15)00068-9 (full paper, opens in a new tab)
| Outcome | Early combined | Conventional | Effect (95% CI) |
|---|---|---|---|
| Steroid-free remission at 12 months, practice mean (SD) (primary endpoint) | 66.0% (14.0) | 61.9% (16.9) | Adjusted difference 2.5% (−5.2 to 10.2); P=0.5169 |
| Surgery, admission or serious disease-related complication by 24 months | 27.7% | 35.1% | Absolute difference 7.3%; HR 0.73 (0.62–0.86); P=0.0003 |
The benefit appeared in hard outcomes at 2 years rather than in symptoms at 1 year, but it was a secondary endpoint.
Source: Khanna 2015, abstract (intention to treat; 921 and 806 patients completed 12 months).
Structured abstract
BackgroundCrohn's disease was usually treated step by step, adding stronger drugs after each failure. Whether starting a TNF antagonist plus an antimetabolite early improves outcomes in community practice was unknown.
MethodsOpen-label cluster randomised trial; 41 community gastroenterology practices in Belgium and Canada randomised (22 to early combined immunosuppression, 19 to conventional management). 1,982 patients. Primary endpoint: Practice-level proportion of patients in corticosteroid-free remission (Harvey-Bradshaw Index ≤4) at 12 months.
ResultsPrimary result. 66.0% (SD 14.0) at early-combined-immunosuppression practices vs 61.9% (SD 16.9) at conventional practices; adjusted difference 2.5% (95% CI −5.2 to 10.2); P=0.5169.
Take-homeEarly combined immunosuppression did not improve 1-year symptom remission but lowered 2-year surgery, admissions and complications, supporting early effective therapy.
| Patients | 1,982 |
|---|---|
| Population | Up to 60 consecutive adults with Crohn's disease per practice (1084 patients at early-combined-immunosuppression practices and 898 at conventional practices), followed for 2 years. Broad, real-world inclusion. The practices, not the patients, were randomised. |
| Comparison | Early combined immunosuppression (TNF antagonist plus antimetabolite) algorithm vs conventional step-up management |
| Drugs | TNF antagonist, antimetabolite |
| Classes | Anti-TNF, Immunomodulator |
| Published | Khanna R, et al. Lancet 2015;386:1825–34. |
Key points
- 24-month composite of surgery, hospital admission or serious disease-related complications: 27.7% vs 35.1%; absolute difference 7.3%; HR 0.73 (95% CI 0.62–0.86); P=0.0003.
- No difference in serious drug-related adverse events, and no increase in mortality.
- 921 and 806 patients completed 12 months and formed the intention-to-treat analysis.
- Canadian community practices took part, so the population resembles patients seen in Canadian general GI practice.
Limitations
- The primary endpoint was negative; the reduction in major adverse outcomes was a secondary endpoint that the authors call hypothesis-generating.
- Symptom-based primary outcome (HBI) with no endoscopic endpoint.
- Cluster design and open label, so patient-level treatment depended on how each practice applied the algorithm.
- Academic sponsor (University of Western Ontario) with funding from AbbVie.
Commentary
REACT asked whether an algorithm of early anti-TNF plus antimetabolite, applied across whole community practices, changes outcomes. Symptoms at 1 year were similar, but serious disease outcomes over 2 years were reduced. It was an early signal for the top-down approach that PROFILE later tested directly in newly diagnosed patients.
Journal club
Design
Open-label cluster randomised trial; 41 community gastroenterology practices in Belgium and Canada randomised (22 to early combined immunosuppression, 19 to conventional management). 1,982 patients.
Primary result. 66.0% (SD 14.0) at early-combined-immunosuppression practices vs 61.9% (SD 16.9) at conventional practices; adjusted difference 2.5% (95% CI −5.2 to 10.2); P=0.5169.
Validity and applicability
- The primary endpoint was negative; the reduction in major adverse outcomes was a secondary endpoint that the authors call hypothesis-generating.
- Symptom-based primary outcome (HBI) with no endoscopic endpoint.
- Cluster design and open label, so patient-level treatment depended on how each practice applied the algorithm.
- Academic sponsor (University of Western Ontario) with funding from AbbVie.
Where it sits now
Early combined immunosuppression did not improve 1-year symptom remission but lowered 2-year surgery, admissions and complications, supporting early effective therapy.
- PROFILE (2024): In newly diagnosed active Crohn's disease, starting infliximab plus an immunomodulator straight away gives far better 1-year and 5-year outcomes than step-up care.
- STARDUST (2022): With ustekinumab, an endoscopy-driven treat-to-target algorithm did not improve 48-week endoscopic response over clinically driven care; treat-to-target only helps if escalation has something to add.
- STRIDENT (2022): Symptomatic Crohn's strictures with active inflammation often respond to adalimumab; intensifying therapy did not significantly improve symptoms; secondary results on treatment failure and MRI stricture appearance favoured it but are hypothesis-generating.
- CALM (2017): Escalating therapy on biomarkers plus symptoms, rather than symptoms alone, gives more mucosal healing in early Crohn's disease.
Questions for discussion
- Practices, not patients, were randomised. What are the strengths and pitfalls of a cluster design for testing a treatment strategy?
- The primary endpoint was negative but a secondary endpoint was clearly positive. How much should the secondary result change practice?
- How would you expect the benefit to differ if early combined immunosuppression were started at diagnosis, as in PROFILE?