← Back
NeuroTrials.ai
Neurology Clinical Trial Database

SAkuraSky

Trial of Satralizumab in Neuromyelitis Optica Spectrum Disorder

Share Share Copied! Compare

Year of Publication: 2019

Authors: Takashi Yamamura, Ingo Kleiter, Kazuo Fujihara, ..., Jerome De Seze

Journal: New England Journal of Medicine

Citation: N Engl J Med 2019;381:2114-24

Link: https://doi.org/10.1056/NEJMoa1901747

Clinical Question

Does satralizumab, a humanized monoclonal antibody targeting the interleukin-6 receptor, reduce the risk of relapse when added to stable immunosuppressant treatment in patients with NMOSD?

Bottom Line

Satralizumab added to baseline immunosuppressant treatment reduced the risk of protocol-defined relapse by 62% compared to placebo (HR 0.38; P=0.02), with relapse occurring in 20% vs 43% of patients. The benefit was most pronounced in AQP4-IgG-seropositive patients (HR 0.21; relapse in 11% vs 43%), while no clear benefit was demonstrated in seronegative patients (HR 0.66). However, key secondary endpoints of pain (VAS) and fatigue (FACIT-F) were not significant, causing the testing hierarchy to fail. The safety profile was favorable, with rates of serious adverse events and infections similar between groups and no deaths or anaphylaxis. The small sample size (n=83) and event-driven design with differential treatment duration are important limitations.

Major Points

  • SAkuraSky was a phase 3, randomized, double-blind, placebo-controlled trial of satralizumab added to immunosuppressants in 83 NMOSD patients across 34 sites in 11 countries
  • Primary endpoint met: protocol-defined relapse in 8/41 (20%) satralizumab vs 18/42 (43%) placebo (HR 0.38; 95% CI 0.16–0.88; P=0.02); multiple imputation sensitivity analyses yielded HRs of 0.34–0.44 (P=0.01–0.04)
  • Relapse-free rates: at 48 weeks 89% vs 66%; at 96 weeks 78% vs 59%; at 144 weeks 74% vs 49% (satralizumab vs placebo)
  • Marked AQP4-IgG serologic effect: seropositive (n=55) HR 0.21 (95% CI 0.06–0.75); seronegative (n=28) HR 0.66 (95% CI 0.20–2.24, crossing 1.0)
  • Annualized relapse rate: 0.11 (satralizumab) vs 0.32 (placebo); rate ratio 0.34 (95% CI 0.15–0.77)
  • Key secondary endpoints not significant: VAS pain change difference 4.08 (95% CI −8.44 to 16.61; P=0.52); FACIT-F difference −3.10 (95% CI −8.38 to 2.18) — hierarchy failed
  • Satralizumab was engineered with pH-dependent antigen binding (recycling antibody technology) for prolonged half-life, allowing subcutaneous dosing every 4 weeks
  • Median treatment duration was substantially longer in satralizumab group (107.4 weeks vs 32.5 weeks), reflecting event-driven design and longer time to relapse/withdrawal in satralizumab arm
  • Safety: 90% satralizumab vs 95% placebo had any adverse event; SAEs 17% vs 21%; serious infections 5% vs 7%; injection-related reactions 12% vs 5%; no deaths or anaphylaxis in either group
  • Approximately two-thirds of patients were AQP4-IgG-seropositive; baseline immunosuppressants included glucocorticoids (~45%), azathioprine (~35%), and mycophenolate mofetil (~15%)

Design

Study Type: Phase 3, randomized, double-blind, placebo-controlled, parallel-assignment, event-driven, multicenter trial with open-label extension

Randomization: 1

Blinding: Double-blind. Randomization stratified by baseline annualized relapse rate (1 vs >1) and geographic region (Asia vs Europe or other). Relapses adjudicated by independent clinical endpoint committee unaware of group assignments.

Enrollment Period: Not explicitly stated; ClinicalTrials.gov NCT02028884

Follow-up Duration: Event-driven design ending after 26 protocol-defined relapses; median treatment duration 107.4 weeks (satralizumab) and 32.5 weeks (placebo)

Centers: 34

Countries: Japan, Germany, UK, USA, Poland, Italy, Taiwan, Spain, France, and 2 others (11 total)

Sample Size: 83

Analysis: Intention-to-treat. Primary: two-sided log-rank test stratified by baseline ARR and region; Kaplan-Meier for time-to-event; Cox proportional-hazards model for HR. Four post-hoc multiple imputation analyses for censored data. Key secondary: ANCOVA with random hot-deck multiple imputation. Serial gatekeeping for multiplicity control (primary → VAS pain → FACIT-F). Annualized relapse rate by Poisson regression. Remaining secondary endpoints: mixed-effects model repeated measures; point estimates with unadjusted CIs only. Prespecified AQP4-IgG serologic subgroup analysis.

Inclusion Criteria

  • Age 12–74 years (adolescents and adults)
  • AQP4-IgG-seropositive NMO (by published criteria) or AQP4-IgG-seropositive NMOSD with longitudinally extensive myelitis (≥3 vertebral segments) or recurrent/simultaneous bilateral optic neuritis
  • AQP4-IgG-seronegative NMO (limited to ~30% of adult population)
  • ≥2 relapses in prior 2 years, with ≥1 in prior 12 months
  • EDSS score 0–6.5 at screening
  • Stable dose of permitted baseline immunosuppressant for ≥8 weeks: azathioprine (max 3 mg/kg/d), mycophenolate mofetil (max 3000 mg/d), or oral glucocorticoids (max 15 mg prednisolone equivalent/d)

Exclusion Criteria

  • Previous treatment with any IL-6 pathway-targeting agent, alemtuzumab, total-body irradiation, or bone marrow transplantation at any time
  • Use of eculizumab, belimumab, or MS disease-modifying treatment within 6 months before baseline
  • Use of anti-CD20 agents (including rituximab) during trial or within 6 months before baseline
  • Use of anti-CD4 agents, cladribine, or mitoxantrone within 2 years before baseline

Arms

FieldControlSatralizumab 120 mg
InterventionMatching placebo SC at weeks 0, 2, 4 and every 4 weeks thereafter, added to stable baseline immunosuppressant treatmentSatralizumab 120 mg SC at weeks 0, 2, 4 and every 4 weeks thereafter, added to stable baseline immunosuppressant treatment. pH-dependent antigen-binding (recycling antibody) design for prolonged half-life.
DurationEvent-driven; median 32.5 weeks (range 0–180)Event-driven; median 107.4 weeks (range 2–224)

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Time to first protocol-defined relapse (new/worsening neurologic symptoms with specified EDSS or functional system score changes, persisting >24 hours, adjudicated by blinded endpoint committee)Primary18/42 (43%) had relapse8/41 (20%) had relapse0.380.02
VAS pain score change from baseline to week 24 (key secondary; range 0–100)Secondary−3.73 (SE 4.12)+0.35 (SE 4.52)0.52 (not significant; hierarchy failed)
FACIT-F score change from baseline to week 24 (key secondary; range 0–52)Secondary+3.12 (SE 1.79)+0.02 (SE 2.00)Not presented (hierarchy failed at VAS pain)
Annualized relapse rateSecondary
Relapse-free rate at 48 weeksSecondary
Relapse-free rate at 96 weeksSecondary
Protocol-defined relapse in AQP4-IgG-seropositive subgroup (n=55)Secondary0.21 (95% CI 0.06–0.75)
Protocol-defined relapse in AQP4-IgG-seronegative subgroup (n=28)Secondary0.66 (95% CI 0.20–2.24)Not significant
Any adverse eventAdverse95% (40/42)90% (37/41)
Serious adverse eventAdverse21% (9/42); 20.2/100 pt-yr17% (7/41); 11.5/100 pt-yr
InfectionAdverse62% (26/42); 149.6/100 pt-yr68% (28/41); 132.5/100 pt-yr
Serious infectionAdverse7% (3/42); 5.0/100 pt-yr5% (2/41); 2.6/100 pt-yr
Injection-related reactionAdverse5% (2/42); 3.4/100 pt-yr12% (5/41); 21.7/100 pt-yr
DeathAdverse00
Anaphylactic reactionAdverse00
Discontinuation due to AE (double-blind)Adverse12% (5/42)7% (3/41)

Subgroup Analysis

Prespecified AQP4-IgG serologic subgroup analysis showed a striking difference: seropositive patients had HR 0.21 (95% CI 0.06–0.75), while seronegative patients had HR 0.66 (95% CI 0.20–2.24, CI crossing 1.0). This suggests satralizumab's benefit is primarily in AQP4-IgG-seropositive disease, consistent with the IL-6–driven AQP4-IgG production mechanism. However, the seronegative subgroup was small (n=28), limiting interpretation. Other subgroup analyses (region, baseline ARR, age) are shown in supplementary figures.

Criticisms

  • Very small sample size (n=83) limits statistical power, particularly for subgroup analyses and secondary endpoints
  • Event-driven design created substantially different treatment durations between arms (median 107 vs 33 weeks), complicating safety comparisons and potentially introducing informative censoring
  • Key secondary endpoints (VAS pain and FACIT-F) were not significant, causing the testing hierarchy to fail; no confirmed benefit beyond relapse prevention
  • No active comparator; comparison to other emerging NMOSD therapies (eculizumab, inebilizumab) is indirect only
  • Differential censoring rates (80% satralizumab vs 57% placebo had data censored) and high early censoring raises concerns about potential bias despite sensitivity analyses
  • AQP4-IgG-seronegative subgroup showed no clear benefit (HR 0.66; CI crossing 1.0); the trial may be enriching for a specific disease phenotype
  • Rituximab use was excluded, yet rituximab is the most commonly used immunosuppressant in NMOSD practice, limiting generalizability
  • The trial could not determine if there was a difference between groups at any given time point (no between-group comparisons at fixed timepoints for primary endpoint)
  • Adolescent patients were included (age 12–17) but the trial was not separately powered for this subgroup
  • Sponsored by Chugai Pharmaceutical (Roche Group); all analyses conducted by sponsor

Funding

Chugai Pharmaceutical (member of the Roche Group). Chugai designed the trial, provided trial drug and placebo, paid for professional medical writing, and analyzed the data.

Based on: SAkuraSky (New England Journal of Medicine, 2019)

Authors: Takashi Yamamura, Ingo Kleiter, Kazuo Fujihara, ..., Jerome De Seze

Citation: N Engl J Med 2019;381:2114-24

Content summarized and formatted by NeuroTrials.ai.