EMERGE / ENGAGE
(2022)Objective
Aducanumab - To evaluate the efficacy and safety of aducanumab (anti-amyloid beta monoclonal antibody) in patients with early Alzheimer's disease (MCI due to AD or mild AD dementia)
Study Summary
• ENGAGE did not meet its primary or any secondary endpoint (CDR-SB difference 0.03; P=.833), likely due to lower cumulative high-dose exposure from differential enrollment timing relative to protocol amendments
• Both trials showed dose- and time-dependent amyloid PET reduction (71% in EMERGE, 59% in ENGAGE at high dose) and reductions in plasma p-tau181, confirming target engagement
• ARIA-E was the most common adverse event (35–36% in high-dose groups), predominantly in ApoE ε4 carriers (42–43%), with most events resolving within 16 weeks
Intervention
Aducanumab (human anti-amyloid beta monoclonal antibody) IV infusion every 4 weeks: low dose (3 or 6 mg/kg target) vs high dose (10 mg/kg target) vs placebo, over 76 weeks (20 doses)
Inclusion Criteria
Age 50–85, clinical criteria for MCI due to AD or mild AD dementia, amyloid pathology confirmed by PET, MMSE 24–30, CDR global 0.5
Study Design
Arms: Low-dose aducanumab (3 mg/kg ApoE ε4+ or 6 mg/kg ApoE ε4−), High-dose aducanumab (10 mg/kg target), Placebo
Patients per Arm: EMERGE: Placebo 548, Low dose 543, High dose 547; ENGAGE: Placebo 545, Low dose 547, High dose 555
Outcome
• ENGAGE primary endpoint not met: no significant difference on CDR-SB or any secondary endpoint
• Discordant results between the two identically designed trials attributed to differential high-dose exposure due to protocol amendment timing; both trials were terminated early following a futility analysis based on violated assumptions
Bottom Line
Results were discordant between two identically designed trials. EMERGE met its primary endpoint, with high-dose aducanumab reducing CDR-SB decline by 22% vs placebo and showing consistent benefit across all secondary endpoints. ENGAGE did not meet its primary or secondary endpoints. Both trials confirmed dose-dependent amyloid plaque reduction and downstream biomarker changes. The discordance is attributed to differential high-dose exposure resulting from protocol amendment timing. ARIA-E was the most common adverse event (35–36% at high dose).
Major Points
- EMERGE and ENGAGE were two identically designed phase 3 RCTs of aducanumab in early AD, enrolling 1638 and 1647 patients respectively across 348 sites in 20 countries
- Both trials were terminated early based on a prespecified futility analysis after ~50% of patients had opportunity to complete week 78; the futility analysis was based on two violated assumptions (similar treatment effect between studies and constancy of effect over time)
- EMERGE primary endpoint met: high-dose aducanumab reduced CDR-SB decline by 22% vs placebo (−0.39; 95% CI −0.69 to −0.09; P=.012)
- EMERGE secondary endpoints all met: MMSE −18% (P=.049), ADAS-Cog13 −27% (P=.010), ADCS-ADL-MCI −40% (P<.001)
- ENGAGE primary endpoint not met: CDR-SB difference 0.03 vs placebo (95% CI −0.26 to 0.33; P=.833); no secondary endpoints were significant
- Both trials showed significant dose- and time-dependent reduction in amyloid PET SUVR: EMERGE high dose −0.278 (71% reduction on centiloid scale), ENGAGE high dose −0.232 (59% reduction); the 16.5% lower reduction in ENGAGE may partly explain the clinical discordance
- Dose-dependent reductions in plasma p-tau181 and CSF p-tau/t-tau observed in both studies, confirming downstream target engagement
- Two protocol amendments (PV3 and PV4) allowed more patients to achieve the 10 mg/kg target dose; due to enrollment differences, these amendments benefited more EMERGE than ENGAGE patients (29% vs 22% received full 14 doses of 10 mg/kg)
- ARIA-E occurred in 35% (EMERGE) and 36% (ENGAGE) of high-dose patients; incidence was higher in ApoE ε4 carriers (42–43%) and homozygous carriers (65%); 98% resolved on study, most within 12–16 weeks
- 16 deaths occurred across both studies (5 placebo, 3 low dose, 8 high dose); none attributed to study treatment
- Low-dose results were generally intermediate between high dose and placebo in both trials but not statistically significant
Study Design
- Study Type
- Two identically designed randomized, double-blind, placebo-controlled, global phase 3 trials
- Randomization
- Yes
- Blinding
- Double-blind. Participants, families, and blinded study team were unaware of treatment assignment. Two independent efficacy raters were blinded to ARIA and medical information. Unblinded pharmacy staff managed treatment preparation. Treating physicians managed ARIA but could not serve as raters and had no access to post-baseline efficacy data.
- Sample Size
- 3285
- Follow-up
- 78 weeks (76 weeks of dosing + week 78 assessment); mean follow-up ~74–75 weeks
- Centers
- 348
- Countries
- USA, Canada, UK, Netherlands, France, Sweden, Japan, and 13 other countries (20 total)
Primary Outcome
Definition: Change from baseline to week 78 in CDR-SB (Clinical Dementia Rating Sum of Boxes; range 0–18, higher = greater impairment)
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| - | - | - | - |
Limitations & Criticisms
- Discordant results between two identically designed trials fundamentally undermine confidence in efficacy conclusions; a single positive trial (EMERGE) is insufficient to establish clinical benefit
- Both trials were terminated early based on a futility analysis whose two core assumptions were violated — this raises questions about data completeness and the validity of post-hoc explanations for the discordance
- The explanation for discordance (differential dosing due to protocol amendments) is post-hoc; 29% vs 22% receiving full 14 doses of 10 mg/kg is a modest difference to explain fully negative vs fully positive results
- The 22% reduction in CDR-SB decline (0.39 points on an 18-point scale) in EMERGE is of uncertain clinical meaningfulness for individual patients
- High rates of ARIA-E (35–43% at high dose, 65% in ApoE ε4 homozygotes) raise significant safety concerns and could lead to functional unblinding despite efforts to maintain blinding
- The trial population lacked racial/ethnic diversity (≤1% Black/African American), limiting generalizability
- CSF and tau PET biomarker substudies had small sample sizes from non-random subsets of participants
- Low-dose arms were not statistically significant in either trial on any endpoint, weakening the dose-response argument for clinical outcomes despite biomarker dose-response
- Increased lateral ventricular volume in all aducanumab groups relative to placebo raises questions about brain volume loss, though authors attribute this to non-neurodegenerative factors
- The sponsor (Biogen) played a role in study design, conduct, data analysis, and manuscript preparation, introducing potential conflict of interest
Citation
J Prev Alz Dis 2022;2(9):197-210