RIGHT-2 ICH
(2019)Objective
Prespecified subgroup analysis of RIGHT-2 evaluating the effect of prehospital transdermal GTN 5mg (vs sham) in 145 participants with a confirmed final diagnosis of intracerebral hemorrhage.
Study Summary
• GTN was associated with significantly increased in-hospital mortality: 45.9% vs 29.6%, aOR 2.26 (95% CI 1.03–4.95)
• GTN associated with larger hematoma (aOR 1.95, 95% CI 1.07–3.58) and more mass effect (aOR 2.42, 95% CI 1.26–4.68) on admission CT
• Global outcome across 5 domains was significantly worse with GTN: Wei-Lachin difference 0.18 (95% CI 0.01–0.35)
Intervention
Transdermal glyceryl trinitrate (GTN) 5mg patch applied in the ambulance within 4 hours of stroke onset, then daily for up to 3 additional days in hospital (total up to 4 days), vs sham dressing
Inclusion Criteria
Adults with presumed stroke within 4 hours, FAST score 2-3, systolic BP ≥120 mmHg; this analysis restricted to those with a confirmed final hospital diagnosis of ICH
Study Design
Arms: Glyceryl trinitrate 5mg transdermal patch for up to 4 days (n=74) vs Sham dressing (n=71)
Patients per Arm: 74 GTN, 71 sham (145 total; 13% of 1149 RIGHT-2 participants had ICH)
Outcome
• In-hospital mortality: GTN 45.9% vs sham 29.6%, aOR 2.26 (95% CI 1.03–4.95)
• 90-day mortality: GTN 47.9% vs sham 32.4%, aOR 1.50 (95% CI 0.86–2.62, NS)
• Discharge disposition worse with GTN: aOR 2.31 (95% CI 1.17–4.57)
• Quality of life (EQ-5D) worse with GTN: difference −0.13 (95% CI −0.23 to −0.03)
Bottom Line
Ultra-acute prehospital GTN is harmful in ICH and should not be used. Despite a technically neutral primary endpoint, GTN consistently worsened clinical outcomes, neuroimaging findings, and mortality. The harm is most pronounced in patients randomized within 1 hour of onset, likely reflecting inhibition of vasoconstriction and platelet plugging during active bleeding.
Major Points
- Primary mRS outcome was technically neutral (aOR 1.87, p=0.058) but all 4 sensitivity analyses were statistically significant and favored sham, suggesting harm with GTN in ICH
- GTN significantly increased in-hospital mortality (45.9% vs 29.6%, aOR 2.26, p<0.05) — more than 40% of deaths occurred by day 4
- GTN was associated with larger hematoma and more mass effect on CT at hospital admission (within 55 min of patch placement)
- Global analysis across 5 outcomes (mRS, Barthel, cognition, QoL, mood) was significantly worse with GTN (Wei-Lachin difference 0.18, 95% CI 0.01–0.35)
- Time-dependent interaction: GTN dramatically worsened outcome when randomized within 1 hour (OR 8.39, 95% CI 2.22–31.69); trend toward benefit beyond 2 hours — likely reflecting inhibition of early hemostasis
- Combined with MR ASAP ICH signal, this strongly argues against prehospital vasodilators in ICH outside of RCTs
Study Design
- Study Type
- Prespecified subgroup analysis of a multicenter, randomized, sham-controlled, participant- and outcome-blinded (PROBE), phase 3 trial (RIGHT-2)
- Randomization
- Yes
- Blinding
- Participant-blinded (sham dressing); outcome-blinded (central telephone mRS assessment at 90 days by trained assessors masked to treatment allocation); SAEs adjudicated by expert panel blinded to treatment
- Sample Size
- 145
- Follow-up
- 90 days
- Centers
- 12
- Countries
- United Kingdom
Primary Outcome
Definition: 7-level modified Rankin Scale (mRS) score at 90 days, ordinal shift analysis (adjusted common odds ratio for poor outcome with GTN vs sham)
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| Median 5 (IQR 3–6), n=71 | Median 5 (IQR 4–6), n=74 | 1.87 (0.98–3.57) | 0.058 (technically nonsignificant; all 4 sensitivity analyses significant, all favoring sham) |
Limitations & Criticisms
- Small subgroup (n=145 ICH out of 1149 RIGHT-2 patients) — underpowered for definitive conclusions; this is a prespecified subgroup, not a standalone RCT
- Primary outcome technically neutral (p=0.058) — results cannot be interpreted as definitively significant despite strongly consistent trends
- Baseline imbalances: premorbid mRS >2 (16% GTN vs 7% sham), antiplatelet therapy (17% vs 38%), prior stroke (16% vs 21%) — may confound results despite randomization
- Prehospital stroke diagnosis is imprecise — all patients had suspected stroke by FAST/BP criteria; ICH diagnosis confirmed retrospectively by imaging
- 100% adherence to first dose, but only 49% received all 4 patches in both arms — treatment fidelity limited for full-course analysis
- The time-by-treatment interaction (harm within 1h, possible benefit after 2h) is hypothesis-generating from a subgroup of a subgroup — requires prospective validation
Citation
Stroke. 2019;50:3064-3071