HAYABUSAUC · Positioning and de-escalation
Discontinuation of infliximab in patients with ulcerative colitis in remission (HAYABUSA): a multicentre, open-label, randomised controlled trial
Kobayashi T, et al.
Lancet Gastroenterol Hepatol 2021 · PMID 33887262 (PubMed, opens in a new tab) · doi 10.1016/S2468-1253(21)00062-5 (full paper, opens in a new tab)
| Endpoint | Continue | Stop | Difference |
|---|---|---|---|
| Remission, Week 48 (primary endpoint) | 80.4% (37/46), 95% CI 66.1–90.6 | 54.3% (25/46), 95% CI 39–69.1 | Difference 26.1% (95% CI 7.7–44.5); P=0.0076. Adjusted 27.3% (8.0–44.1); P=0.0059 |
Of 12 patients retreated with infliximab after relapse, 8 regained remission within 8 weeks.
Source: Kobayashi 2021, abstract (full analysis set: 92 of 95 randomised). Open-label.
Structured abstract
BackgroundLong-term anti-TNF therapy raises concerns about infection, malignancy and cost, but no randomised trial had tested stopping infliximab in UC remission.
MethodsMulticentre, open-label, randomised controlled trial (24 specialist centres in Japan). 95 patients. Primary endpoint: Remission at week 48.
ResultsPrimary result. 80.4% (37/46; 95% CI 66.1–90.6) with continued infliximab vs 54.3% (25/46; 39.0–69.1) after stopping; difference 26.1% (95% CI 7.7–44.5; P=0.0076), 27.3% (8.0–44.1; P=0.0059) after adjustment.
Take-homeStopping infliximab in UC remission lowered remission at week 48 (54.3% vs 80.4% when continued), although most patients respond to retreatment; continue anti-TNF therapy unless there is a strong reason to stop.
| Patients | 95 |
|---|---|
| Population | Patients with UC on infliximab 5 mg/kg every 8 weeks (started at least 14 weeks earlier), in steroid-free remission for more than 6 months with a Mayo endoscopic subscore of 0 or 1. 92 in the full analysis set. |
| Comparison | Continue infliximab (n=48) vs stop infliximab (n=47); randomisation stratified by immunomodulator use and endoscopic subscore |
| Drugs | infliximab |
| Classes | Anti-TNF |
| Published | Kobayashi T, et al. Lancet Gastroenterol Hepatol 2021;6:429–37. |
Key points
- Of 12 patients retreated with infliximab after relapse, 8 (66.7%) regained remission within 8 weeks, with no infusion reactions.
- Adverse events: 17% (8/48) continuing vs 13% (6/47) stopping; P=0.59.
- Later evidence: BIOSTOP (172 Norwegian patients with UC in deep clinical, biochemical and endoscopic remission) found endoscopic remission at 2 years in 80% (66/82) after anti-TNF withdrawal vs 96% (78/81) with maintenance; non-inferiority not shown. Flares 52% (45/86) vs 12% (10/86); anti-TNF retreatment worked in 74% (20/27) (PMID 42557857).
- In BIOSTOP, 48% of the withdrawal group stayed in steroid-free remission, and no predictors of relapse were found.
Limitations
- Open-label; small; Japanese population.
- Mayo endoscopic subscore of 1 was allowed, so not all patients had complete mucosal healing.
- Many patients were not on an immunomodulator (59% and 65% were, as summarised in the EXIT trial report), which may affect relapse after withdrawal.
- Partly industry funded (Mitsubishi Tanabe Pharma).
Commentary
HAYABUSA is the first randomised trial of anti-TNF withdrawal in ulcerative colitis. Its message, confirmed in the larger Norwegian BIOSTOP trial in patients in deep remission, is that stopping anti-TNF therapy in UC leads to many more flares than continuing it.
For patients who want to stop (for example to reduce infusions or long-term drug exposure), discuss the trade-off: about half will flare within 1 to 2 years, and roughly 7 in 10 regain remission with retreatment. Monitor closely with calprotectin after stopping. For Crohn’s disease, see EXIT, SPARE and STORI.
Journal club
Design
Multicentre, open-label, randomised controlled trial (24 specialist centres in Japan). 95 patients.
Primary result. 80.4% (37/46; 95% CI 66.1–90.6) with continued infliximab vs 54.3% (25/46; 39.0–69.1) after stopping; difference 26.1% (95% CI 7.7–44.5; P=0.0076), 27.3% (8.0–44.1; P=0.0059) after adjustment.
Validity and applicability
- Open-label; small; Japanese population.
- Mayo endoscopic subscore of 1 was allowed, so not all patients had complete mucosal healing.
- Many patients were not on an immunomodulator (59% and 65% were, as summarised in the EXIT trial report), which may affect relapse after withdrawal.
- Partly industry funded (Mitsubishi Tanabe Pharma).
Where it sits now
Stopping infliximab in UC remission lowered remission at week 48 (54.3% vs 80.4% when continued), although most patients respond to retreatment; continue anti-TNF therapy unless there is a strong reason to stop.
- EXIT (2025): In carefully selected patients in deep remission who stay on an immunomodulator, stopping the anti-TNF did not significantly reduce 1-year sustained remission, but calprotectin rose more often; withdrawal is an option to discuss, not a default.
- SPARE (2023): If you de-escalate combination therapy in stable Crohn's disease, stop the immunosuppressant, not infliximab; relapse is much less likely.
- STORI (2012): About half of patients relapse within a year of stopping infliximab from combination therapy, but most regain remission on re-treatment, and biomarkers identify a lower-risk group.
Questions for discussion
- A Mayo endoscopic subscore of 1 was allowed. Would you expect fewer relapses if only patients with a subscore of 0 had been included?
- Many patients were not on an immunomodulator. How does that compare with EXIT, and could it explain the different results?
- A patient in remission wants to stop infliximab. How would you present the relapse risk and the chance of regaining remission on retreatment?