Skip to main content

Trial Summaries & Critiques

EMPHASIS

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Efficacy and safety of minocycline in patients with acute ischaemic stroke (EMPHASIS): a multicentre, double-blind, randomised controlled trial
  • Acronym: EMPHASIS
  • Year: 2026
  • Journal published in: The Lancet
  • Citation: Lu Y, Guan L, Wu J, Zhang M, Huang R, Li J, et al; EMPHASIS investigators. Efficacy and safety of minocycline in patients with acute ischaemic stroke (EMPHASIS): a multicentre, double-blind, randomised controlled trial. Lancet. 2026 Jan 30:S0140-6736(25)01862-8.

Context & Rationale

  • Background
    • Even in the reperfusion era, a large proportion of patients with acute ischaemic stroke (AIS) remain disabled; adjunctive strategies targeting secondary injury pathways remain a major unmet need.
    • Post-ischaemic injury evolves over hours–days via neuroimmune activation, matrix metalloproteinase (MMP) activity, blood–brain barrier dysfunction, oedema, and secondary neuronal/glial injury.
    • Minocycline is a lipophilic tetracycline with pleiotropic anti-inflammatory and MMP-inhibitory actions; early human studies and meta-analyses suggested possible functional benefit, but evidence was limited by small sample sizes and heterogeneous designs.
  • Research Question/Hypothesis
    • In adults with imaging-confirmed AIS treated within 72 hours of onset, does a 4.5-day course of enteral minocycline improve 90-day functional outcomes compared with placebo, without excess intracranial haemorrhage or other serious harms?
  • Why This Matters
    • Minocycline is inexpensive, widely available, and potentially scalable in health systems with limited access to reperfusion therapies.
    • A positive signal in a late window (hours–days) would support “cerebro-/cytoprotection” approaches targeting glia/vasculature and subacute injury biology, rather than hyperacute neuronal salvage alone.

Design & Methods

  • Research Question: Among adults with AIS within 72 hours, does minocycline plus routine care increase the proportion achieving mRS 0–1 at 90 days versus placebo plus routine care?
  • Study Type: Multicentre, double-blind, randomised, placebo-controlled, parallel-group trial (58 hospitals in China); investigator-initiated; no interim efficacy analyses.
  • Population:
    • Setting: In-hospital acute stroke care across 58 centres.
    • Key inclusion: age 18–80 years; imaging-confirmed AIS; randomisation ≤72 h from onset; NIHSS 4–25 and NIHSS 1a ≤1; first-ever stroke or pre-stroke mRS ≤1.
    • Key exclusions: tetracycline allergy; pregnancy/breastfeeding; severe renal or hepatic insufficiency; active bacterial infection; bleeding tendency or recent surgery; other protocol-defined exclusions.
  • Intervention:
    • Study drug: minocycline 200 mg loading dose orally within 30 min of randomisation, then 100 mg every 12 h for 4 days (total treatment duration 4.5 days); administration via feeding tube if dysphagic.
    • Co-interventions: routine AIS management permitted, including intravenous thrombolysis and/or thrombectomy if clinically indicated.
    • Key prohibited co-interventions: other tetracyclines; edaravone–dexborneol; selected anti-inflammatory/immune-modulating therapies (eg, colchicine, tocilizumab, immunosuppressants); systemic retinoids (vitamin A derivatives).
  • Comparison:
    • Study drug: matching placebo capsules on the same schedule (200 mg loading-equivalent then 100 mg-equivalent every 12 h for 4 days).
    • Background care: routine AIS management as above, with the same protocol-defined prohibited co-interventions.
  • Blinding: Double-blind (participants, clinical teams, investigators, and outcome assessors); identical-appearing study drug; 90-day mRS adjudicated by an independent blinded panel.
  • Statistics: A total of 1672 patients were required to detect an 8% absolute increase in mRS 0–1 at 90 days (from 60% to 68%) with 90% power at a two-sided 5% significance level, allowing for 10% loss to follow-up; primary analysis in a modified intention-to-treat population (randomised and received ≥1 dose) with complete-case analysis for the primary endpoint; prespecified sensitivity analyses included multiple imputation and worst-case imputation for missing primary outcomes.
  • Follow-Up Period: 90 ± 7 days (with additional assessments at 24 h and day 6 or discharge).

Key Results

This trial was not stopped early. No interim efficacy analyses were performed.

Outcome Minocycline Placebo Effect p value / 95% CI Notes
mRS 0–1 at 90 days (primary) 447/850 (52.6%) 403/851 (47.4%) RR 1.11 95% CI 1.03–1.20; p=0.0061 Complete-case mITT (mRS missing: 12 vs 11)
Ordinal mRS shift at 90 days 0: 22.9%; 1: 29.6%; 2: 19.8%; 3: 16.5%; 4: 6.8%; 5: 2.7%; 6: 1.6% 0: 20.6%; 1: 26.8%; 2: 20.4%; 3: 20.9%; 4: 6.0%; 5: 2.9%; 6: 2.4% cOR 1.19 95% CI 1.03–1.38; p=0.018 cOR >1 indicates shift towards better (lower) mRS
mRS 0–2 at 90 days 615/850 (72.4%) 577/851 (67.8%) RR 1.07 95% CI 1.02–1.12; p=0.0056 Complete-case mITT
mRS 0–3 at 90 days 755/850 (88.8%) 755/851 (88.7%) RR 1.00 95% CI 0.97–1.03; p=0.94 Complete-case mITT
NIHSS change baseline → 24 h 0 (0 to 0) 0 (0 to 0) β −0.07 95% CI −0.20 to 0.06; p=0.32 NIHSS missing: 15 vs 8
NIHSS change baseline → day 6 −2 (−3 to 0) −1 (−3 to 0) β −0.28 95% CI −0.50 to −0.05; p=0.015 NIHSS missing: 21 vs 18
Early neurological deterioration at 24 h 52/848 (6.1%) 54/850 (6.3%) RR 0.97 95% CI 0.64–1.47; p=0.89 Defined by NIHSS worsening; missing: 14 vs 12
Early neurological deterioration at day 6 56/842 (6.7%) 66/840 (7.9%) RR 0.85 95% CI 0.59–1.21; p=0.36 Missing: 20 vs 22
hs-CRP change baseline → day 6 (mg/L) 0 (−1.20 to 1.97) 0.20 (−0.94 to 3.73) β −2.72 95% CI −5.64 to 0.19; p=0.067 hs-CRP missing: 172 vs 184
Stroke recurrence at 90 days 51/862 (5.9%) 47/862 (5.5%) HR 1.09 95% CI 0.73–1.62; p=0.68 Time-to-event analysis; safety population
Composite vascular events at 90 days 59/862 (6.8%) 52/862 (6.0%) HR 1.14 95% CI 0.79–1.65; p=0.49 Composite endpoint; safety population
Symptomatic intracranial haemorrhage at day 6 3/859 (0.3%) 0/861 (0%) Not reported Not reported Heidelberg-based definition; rare events
Any bleeding event at 90 days 63/862 (7.3%) 69/862 (8.0%) HR 0.90 95% CI 0.64–1.27; p=0.56 Safety population
All-cause death at 90 days 14/862 (1.6%) 20/862 (2.3%) HR 0.69 95% CI 0.35–1.36; p=0.28 Safety population
Serious adverse events (during trial follow-up) 40/862 (4.6%) 51/862 (5.9%) Not reported p=0.24 Effect estimate not reported; trial not powered for rare harms
  • Minocycline increased the proportion of patients achieving functional independence (mRS 0–1) at 90 days (52.6% vs 47.4%; RR 1.11; 95% CI 1.03–1.20; p=0.0061) and favoured an ordinal mRS shift (cOR 1.19; 95% CI 1.03–1.38; p=0.018).
  • Early neurological recovery signals were modest and temporally delayed (NIHSS improvement at day 6 but not at 24 h), and inflammatory marker reduction (hs-CRP) did not reach conventional statistical significance.
  • Major safety outcomes were broadly similar between groups (bleeding, mortality), with very low rates of symptomatic intracranial haemorrhage.

Internal Validity

  • Randomisation and Allocation: Centralised computer-generated randomisation stratified by study site with a fixed block size of four; allocation implemented via unique drug bottle numbers.
  • Drop out or exclusions: 90-day mRS available for 850/862 (98.6%) in the minocycline group and 851/862 (98.7%) in the placebo group; 23/1724 (1.3%) did not complete 90-day follow-up (4 withdrew consent; 19 lost to follow-up).
  • Post-randomisation exclusions: Per-protocol analysis excluded 48/862 (5.6%) vs 50/862 (5.8%) for major protocol deviations; primary analysis was a complete-case approach within a modified intention-to-treat population.
  • Performance/Detection Bias: Double-blind design; 90-day mRS adjudicated by an independent blinded panel; many secondary outcomes were objective (NIHSS, clinical events).
  • Protocol Adherence: All randomised participants received at least one dose; treatment discontinuation 12 vs 13; prohibited concomitant medication exposure 16 vs 13.
  • Baseline Characteristics: Median age 65.0 (57.0–72.0) vs 65.0 (57.0–71.0); baseline NIHSS 5.0 (4.0–7.0) in both groups; pre-stroke mRS 0 in 739/862 (85.7%) vs 744/862 (86.3%).
  • Heterogeneity: 58-centre design; primary models adjusted for pooled study centre (mixed-effects); directionally similar estimates were reported in sensitivity analyses.
  • Timing: Median onset-to-treatment time 41.9 h (26.1–52.8) vs 40.5 h (26.5–52.0) (missing for 96 vs 95); treatment initiated ≤24 h in 179/862 (20.8%) vs 175/862 (20.3%), >24–48 h in 375/862 (43.5%) vs 396/862 (45.9%), and >48–72 h in 308/862 (35.7%) vs 291/862 (33.8%).
  • Dose: Loading 200 mg within 30 min of randomisation followed by 100 mg every 12 h for 4 days (enteral route; feeding tube permitted).
  • Separation of the Variable of Interest: Reperfusion therapy use 114/862 (13.2%) vs 130/862 (15.1%); major protocol deviations similar (48 vs 50) and treatment discontinuation similar (12 vs 13).
  • Outcome Assessment: NIHSS missing at 24 h for 15 vs 8 and at day 6 for 21 vs 18; hs-CRP at day 6 missing for 172 vs 184.
  • Statistical Rigor: Achieved enrolment exceeded target (1724 vs 1672); no interim efficacy analyses; prespecified missing-data sensitivity analyses reported (multiple imputation; worst-case imputation).

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong, supported by robust blinding, balanced baseline characteristics, high follow-up completeness, and prespecified analyses; the principal vulnerabilities relate to the complete-case approach for the primary endpoint and the modest event rates for safety outcomes.

External Validity

  • Population Representativeness: Adults aged 18–80 with predominantly mild-to-moderate AIS (median NIHSS 5) and pre-stroke independence; conducted exclusively in China.
  • Applicability: Enteral minocycline is inexpensive and logistically simple, potentially attractive in systems without ready access to reperfusion; applicability to very severe stroke (NIHSS >25), very minor stroke (NIHSS <4), older patients (>80), and different ethnic/genetic backgrounds remains uncertain.
  • Practice Context: Low use of reperfusion therapies (13–15%) suggests findings primarily reflect outcomes in patients not receiving thrombolysis/thrombectomy; translation to high-thrombectomy populations is not established.

Conclusion on External Validity: Generalisability is moderate for patients resembling the enrolled cohort (mild–moderate AIS, largely non-reperfused, age ≤80), but is limited for very severe/minor strokes, older populations, and non-Chinese health systems and case-mix.

Strengths & Limitations

  • Strengths: Large multicentre sample (n=1724); double-blind, placebo-controlled design; prespecified statistical plan and sensitivity analyses; high 90-day outcome completeness; central blinded adjudication of mRS; pragmatic background care with reperfusion therapies permitted.
  • Limitations: Single-country population and age cap at 80 years; predominantly mild-to-moderate strokes (median NIHSS 5) with low reperfusion rates; primary endpoint analysed as complete-case; substantial missingness for some biomarkers (hs-CRP) and secondary outcomes were not consistently positive (eg, mRS 0–3, vascular events); quality-of-life endpoint (EQ-5D) was prespecified but not reported.

Interpretation & Why It Matters

  • Clinical signal
    • Enteral minocycline was associated with a statistically significant improvement in 90-day functional outcome (mRS 0–1 and ordinal shift), with broadly similar safety event rates.
    • The magnitude and pattern (delayed NIHSS improvement) is compatible with an anti-inflammatory/vasculoprotective mechanism rather than immediate reperfusion-like salvage.
  • Implementation considerations
    • Oral/enteral administration and 72-hour enrolment window increase practical feasibility, but also mean results apply to a late-treated population.
    • In critically ill or dysphagic patients, enteral absorption and drug delivery logistics may differ; intravenous formulations (tested in earlier studies) may be relevant for future protocols.
  • Position within acute stroke care
    • Findings support renewed evaluation of adjunct “cerebro-/cytoprotection” alongside reperfusion and best medical therapy.
    • Replication and mechanistic validation (biomarkers/imaging) are key before routine adoption.

Controversies & Other Evidence

  • Randomisation and missing-data sensitivity: Fixed block randomisation can increase predictability at the site level; the primary analysis excluded 23 randomised participants with missing 90-day mRS (complete-case), and sensitivity analyses highlighted dependence on missing-data assumptions.1
  • Mechanism and outcome coherence: The observed 90-day functional benefit was accompanied by modest early neurological changes and no clearly significant reduction in hs-CRP, raising questions about the most relevant biological pathway and the optimal biomarkers for effect confirmation.1
  • Late treatment window as a “cerebro-/cytoprotection” test case: EMPHASIS deliberately targeted a subacute therapeutic window (≤72 h) consistent with delayed neuroimmune and vascular injury biology; this frames minocycline as a potential cerebroprotective adjunct rather than classic hyperacute neuroprotection.23
  • Prior evidence base: Earlier small trials and meta-analyses suggested possible benefit but were limited by sample size, design heterogeneity, and imprecision; EMPHASIS provides the largest blinded, placebo-controlled dataset to date, but the effect size and generalisability warrant independent confirmation.4
  • Guideline position: Major contemporary AIS guidelines focus on reperfusion, physiological optimisation, and complication prevention; they do not yet provide an implementation pathway for routine minocycline neuro-/cerebroprotection in AIS care.5

Summary

  • EMPHASIS randomised 1724 adults with imaging-confirmed AIS within 72 h (median NIHSS 5) to enteral minocycline vs placebo for 4.5 days.
  • Primary outcome favoured minocycline: mRS 0–1 at 90 days 52.6% (447/850) vs 47.4% (403/851); RR 1.11; 95% CI 1.03–1.20; p=0.0061.
  • Ordinal functional recovery also favoured minocycline (cOR 1.19; 95% CI 1.03–1.38; p=0.018), but broader disability-free survival thresholds (mRS 0–3) and vascular events were unchanged.
  • Safety outcomes were broadly similar, with very low symptomatic intracranial haemorrhage and no statistically significant differences in bleeding or mortality.
  • The trial supports late-window cerebro-/cytoprotection as a plausible strategy, but replication and mechanistic corroboration are needed before routine adoption.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Minocycline is an antibiotic; if used off-label for cerebroprotection, antimicrobial stewardship considerations and monitoring for tetracycline-class adverse effects are relevant alongside stroke-specific safety surveillance.

Overall Takeaway

EMPHASIS is the largest blinded, placebo-controlled evaluation of minocycline for AIS to date and reports a statistically significant improvement in 90-day functional outcomes with broadly similar safety event rates. Its late enrolment window (≤72 h) tests a cerebro-/cytoprotection paradigm aligned with subacute neuroimmune and vascular injury biology. While promising for a low-cost, scalable adjunct, the modest effect size, reliance on complete-case primary analysis, and single-country setting mean replication and mechanistic corroboration are required before routine adoption.

Overall Summary

  • Enteral minocycline for 4.5 days (started ≤72 h) improved 90-day functional outcome in AIS versus placebo.
  • Secondary outcomes showed a modest delayed NIHSS improvement; vascular event rates were similar.
  • Safety signals were broadly neutral, with very low symptomatic intracranial haemorrhage.

Bibliography