← Back
NeuroTrials.ai
Neurology Clinical Trial Database

Clarity AD

Lecanemab in Early Alzheimer's Disease

Year of Publication: 2023

Authors: Christopher H. van Dyck, Chad J. Swanson, Paul Aisen, ..., Takeshi Iwatsubo

Journal: New England Journal of Medicine

Citation: N Engl J Med 2023;388:9-21

Link: https://www.nejm.org/doi/full/10.1056/NEJMoa2212948


Clinical Question

Does lecanemab slow cognitive and functional decline compared with placebo in patients with early Alzheimer's disease?

Bottom Line

Lecanemab reduced markers of amyloid in early Alzheimer's disease and resulted in moderately less decline on measures of cognition and function than placebo at 18 months (27% relative reduction in CDR-SB worsening), but was associated with adverse events including ARIA-E (12.6%) and infusion-related reactions (26.4%).

Major Points

  • Primary endpoint met: CDR-SB change was 1.21 with lecanemab vs 1.66 with placebo (difference −0.45, P<0.001), representing 27% less decline
  • Robust amyloid reduction: mean amyloid level dropped to 22.99 centiloids in lecanemab group, below the 30 centiloid positivity threshold
  • All key secondary endpoints favored lecanemab: ADAS-cog14, ADCOMS, and ADCS-MCI-ADL (all P<0.001)
  • Treatment separation visible by 3 months on clinical measures
  • ARIA-E occurred in 12.6% (2.8% symptomatic), mostly mild-moderate (91%), asymptomatic (78%), occurring early (71% within 3 months), and resolving within 4 months (81%)
  • ARIA-E rates higher in APOE ε4 homozygotes (32.6%) than heterozygotes (10.9%) or noncarriers (5.4%)
  • Infusion-related reactions in 26.4%, largely mild-moderate (96%), mostly with first dose (75%)
  • Biomarker improvements across CSF (Aβ, tau, p-tau181, neurogranin) and plasma (Aβ42/40, p-tau181, GFAP) markers, except NfL

Design

Study Type: Multicenter, randomized, double-blind, placebo-controlled, parallel-group phase 3 trial

Randomization: 1

Blinding: Double-blind; clinical assessment raters were unaware of safety assessments and trial-group assignments; independent medical monitoring team blinded to assignments reviewed ARIA

Enrollment Period: March 2019 through March 2021

Follow-up Duration: 18 months

Centers: 235

Countries: United States, Canada, Japan, South Korea, Australia, Singapore, Netherlands, Sweden, Spain, France, Germany, United Kingdom, Italy, Poland

Sample Size: 1795

Analysis: Modified intention-to-treat analysis using mixed model for repeated measures (MMRM); model included baseline CDR-SB score as covariate, with trial group, visit, stratification variables (clinical subgroup, AD medication use, ApoE ε4 status, geographic region), baseline-by-visit interaction, and trial group-by-visit interaction as fixed effects


Inclusion Criteria

  • Age 50 to 90 years
  • Mild cognitive impairment due to Alzheimer's disease or mild Alzheimer's disease dementia (NIA-AA criteria)
  • Amyloid positivity confirmed by PET or CSF Aβ1-42 measurement
  • Objective memory impairment (≥1 SD below age-adjusted mean on Wechsler Memory Scale IV–Logical Memory II)
  • Global CDR score of 0.5 to 1.0

Exclusion Criteria

  • Not explicitly detailed in the publication

Arms

FieldControlLecanemab 10 mg/kg
InterventionPlacebo intravenous infusion every 2 weeksLecanemab 10 mg/kg intravenous infusion every 2 weeks
Duration18 months18 months

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Change from baseline at 18 months in CDR-SB score (range 0-18, higher = worse impairment)Primary1.661.210.45%<0.001
Amyloid burden on PET (centiloids) - change at 18 monthsSecondary+3.64−55.48<0.001
ADAS-cog14 score change at 18 monthsSecondary5.584.14<0.001
ADCOMS change at 18 monthsSecondary0.2140.164<0.001
ADCS-MCI-ADL score change at 18 monthsSecondary−5.5−3.5<0.001
Any adverse eventAdverse735 (81.9%)798 (88.9%)
Serious adverse eventAdverse101 (11.3%)126 (14.0%)
DeathAdverse7 (0.8%)6 (0.7%)
Discontinuation due to AEAdverse26 (2.9%)62 (6.9%)
Infusion-related reactionAdverse66 (7.4%)237 (26.4%)
ARIA-EAdverse15 (1.7%)113 (12.6%)
Symptomatic ARIA-EAdverse025 (2.8%)
ARIA-E - ApoE ε4 noncarrierAdverse1/286 (0.3%)15/278 (5.4%)
ARIA-E - ApoE ε4 heterozygoteAdverse9/478 (1.9%)52/479 (10.9%)
ARIA-E - ApoE ε4 homozygoteAdverse5/133 (3.8%)46/141 (32.6%)
ARIA-H (microhemorrhages or hemosiderin deposits)Adverse81 (9.0%)155 (17.3%)
MacrohemorrhageAdverse1 (0.1%)5 (0.6%)
HeadacheAdverse73 (8.1%)100 (11.1%)
FallAdverse86 (9.6%)93 (10.4%)

Subgroup Analysis

Results were consistent across key randomization strata (clinical subgroup, AD medication use, ApoE ε4 status, region). Exploratory analysis in ApoE ε4 homozygotes (15% of population) numerically favored lecanemab for ADAS-cog14 and ADCS-MCI-ADL but not for CDR-SB or ADCOMS. Prespecified exploratory analysis of time to worsening of global CDR score showed HR 0.69 favoring lecanemab (multiplicity-unadjusted).


Criticisms

  • Only 18 months of data; longer trials needed to confirm durability of effect and long-term safety
  • Clinical meaningfulness of 0.45 CDR-SB point difference is debated; no established threshold for clinically meaningful difference
  • ARIA-E risk is substantial, especially in ApoE ε4 homozygotes (32.6%); requires MRI monitoring
  • Infusion-related reactions common (26.4%), may impact treatment adherence
  • Trial conducted during COVID-19 pandemic with some missed doses and delayed assessments
  • Dropout rate of 17.2%; sensitivity analyses suggest robustness but missing data is a limitation
  • Modified intention-to-treat analysis without imputation of missing values
  • ARIA may have caused functional unblinding; sensitivity analyses attempted to address this
  • Underrepresentation of Black participants (2.3-2.7%) limits generalizability
  • Biomarker NfL did not show improvement, raising questions about neurodegeneration effect

Funding

Eisai (regulatory sponsor), with partial funding by Biogen

Based on: Clarity AD (New England Journal of Medicine, 2023)

Authors: Christopher H. van Dyck, Chad J. Swanson, Paul Aisen, ..., Takeshi Iwatsubo

Citation: N Engl J Med 2023;388:9-21

Content summarized and formatted by NeuroTrials.ai.