Edaravone dexborneol in acute ischaemic stroke: an updated GRADE assessed systematic review and meta-analysis with trial sequential analysis

Abstract Background Despite three decades of neuroprotective failures in acute ischaemic stroke (AIS), edaravone dexborneol (EDB) achieved regulatory approval in China by demonstrating superiority over edaravone monotherapy. The only prior meta-analysis evaluated a single functional endpoint, excluded four subsequent studies including the pivotal TASTE-2 trial, and performed no comparator stratified analyses leaving the totality of evidence uncharacterised. Methods A PROSPERO registered systematic review and meta-analysis searched PubMed, Embase, Scopus, clinical trials.gov and Cochrane through May 2026. Eligible studies were randomised controlled trials and comparative observational studies of EDB in adult AIS patients. Co-primary outcomes were 90-day excellent functional recovery (mRS 0–1) and functional independence (mRS 0–2). Random effects meta-analysis, GRADE certainty rating, and Trial Sequential Analysis (TSA) were applied throughout. Results Eleven studies enrolling 4706 patients seven RCTs and four comparative observational studies were included. EDB significantly improved mRS 0–1 (RR 1.14, 95% CI 1.08–1.20; p < 0.00001; I 2 = 0%; 9 studies; n = 4,474; GRADE: HIGH) and mRS 0–2 (RR 1.08, 95% CI 1.03–1.13; p = 0.001; I 2 = 1%; 8 studies; n = 4089; GRADE: HIGH). TSA showed that accrued information exceeded the required information size for both endpoints (555.4% for mRS 0–1; 505.3% for mRS 0–2), and the O’Brien–Fleming benefit boundary was crossed for mRS 0–1 (Z = 4.70), confirming evidentiary sufficiency for this endpoint. For mRS 0–2, the required information size was reached within the earliest contributing trial, so the sequential boundary could be assessed only at the first information fraction, where it was not crossed (Z = 1.56); TSA-based sufficiency is therefore not claimed for mRS 0–2. Ninety day mortality was non-significantly reduced (RR 0.87, 95% CI 0.67–1.13; GRADE: MODERATE), with only 32.4% of the required information size accrued. Adverse event rates were equivalent (RR 1.01); symptomatic intracranial haemorrhage was numerically lower with EDB (RR 0.73, 95% CI 0.49–1.07). All seven RCTs carried a low risk of bias, indicating a consistently high methodological quality across the included evidence base. Conclusion EDB produces a TSA-confirmed improvement in 90-day excellent functional recovery (mRS 0–1). For functional independence (mRS 0–2), the pooled effect was significant with GRADE HIGH certainty, but TSA could not confirm evidentiary sufficiency because the sequential boundary was assessable only at the earliest information fraction, where it was not crossed, so further corroboration is warranted; a reassuring though not fully conclusive safety profile was observed; certainty for mortality and several secondary and safety outcomes remains lower and is detailed below. Definitive inference on mortality requires blinded, adequately powered, multinational trials.