A three-week steroid taper for mild checkpoint-inhibitor pneumonitis failed against six weeks

Design

Randomised non-inferiority trial

Population

106 patients with grade 1–2 immune checkpoint inhibitor pneumonitis, median age 72

Primary outcome

Treatment success at 8 weeks (room-air SpO₂ ≥90% without steroid escalation)

Effect

66.7% (3 weeks) vs 85.2% (6 weeks); difference −18.5 points (80% CI −29.0 to −7.9); non-inferiority not shown

This randomised trial assigned 106 patients with grade 1–2 pneumonitis caused by immune checkpoint inhibitors to a 3-week or a 6-week corticosteroid taper. Median age was 72, and 73% had grade 2 disease. Treatment success at 8 weeks meant resting room-air saturation of 90% or more without needing to escalate or extend steroids.

Success was 66.7% with 3 weeks and 85.2% with 6 weeks, a difference of 18.5 percentage points against the shorter course. Non-inferiority was not shown, and a prespecified exploratory analysis favoured 6 weeks. Grade 3 or higher adverse events were more frequent with 6 weeks (24% against 12%), all manageable. Quality of life and overall survival did not differ.

The 6-week taper in current guidance had no randomised evidence behind it; this is the first trial on duration, and it supports that guidance. The trade-off is clear: about one in five more patients succeeds with 6 weeks, at the cost of more adverse events. The trial was modest in size, enrolled mainly older patients and covered mild disease only.

• Use the guideline 6-week corticosteroid taper for grade 2 checkpoint-inhibitor pneumonitis rather than a shortened course.
• Expect more steroid-related adverse events with the longer course; monitor glucose, mood, infection and sleep.
• Consider prophylaxis against Pneumocystis and gastric protection according to steroid dose and duration.
• Reassess saturation and imaging before completing the taper.
• This trial covered mild pneumonitis only; severe pneumonitis needs separate management.

Why it matters

It replaces an expert-opinion duration with trial evidence, and closes off a tempting shortcut.

Don't overread it

The superiority of 6 weeks came from an exploratory analysis in a trial of 106 patients.

The statistics, in plain English

Non-inferiority asks whether the shorter course is no worse than a set margin. The difference of −18.5 percentage points had an 80% confidence interval of −29.0 to −7.9 — the whole range favours 6 weeks — so the shorter course could not be shown to be acceptable.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.