ELAD
(2026)Objective
To evaluate whether the GLP-1 receptor agonist liraglutide (1.8 mg/day SC) slows decline in cerebral glucose metabolism, cognition and brain volume over 52 weeks in non-diabetic adults with mild-to-moderate Alzheimer's disease
Study Summary
• Cognitive secondary outcome ADAS-Exec z-score declined more slowly with liraglutide vs placebo: difference 0.15 (95% CI 0.03–0.28, unadjusted P=0.01); no benefit on CDR-SoB (−0.06, P=0.81) or ADCS-ADL (−0.58, P=0.65)
• Exploratory MRI: less volume loss with liraglutide in temporal lobe (Δ 696 mm³, P<0.001), total gray matter (Δ 7,274 mm³, P=0.002) and parietal/frontoparietal lobes; hippocampus and entorhinal cortex unchanged
• Safety: GI events more common with liraglutide (25.5% of all events); ≥5% weight loss in 48.1% (vs 13.7% placebo). Serious AEs LESS frequent with liraglutide: 6.9% (7/102) vs 17.6% (18/102) placebo. No new safety signals
Intervention
Liraglutide 1.8 mg subcutaneous once daily (titrated from 0.6 mg) vs matching placebo for 52 weeks
Inclusion Criteria
Age ≥50 years; clinical diagnosis of mild-to-moderate Alzheimer's disease per NINCDS-ADRDA criteria; MMSE ≥15; CDR Global 0.5, 1, or 2
Study Design
Arms: Liraglutide 1.8 mg/day SC (n=102) vs matching placebo SC (n=102); 52-week double-blind treatment with FDG-PET and MRI at baseline and week 52; cognitive batteries at baseline, week 24, week 52
Patients per Arm: Liraglutide 102, Placebo 102 (total 204 randomized; 169 completed; analyzable PET SUV at 52 wk: 72 treatment / 82 placebo)
Outcome
• Secondary cognition (ADAS-Exec z-score change): 0.15 (95% CI 0.03–0.28), unadjusted P=0.01 — POSITIVE (exploratory, not multiplicity-adjusted)
• Secondary CDR-SoB: −0.06 (P=0.81); ADCS-ADL: −0.58 (P=0.65) — no difference
• Exploratory MRI: temporal lobe atrophy attenuated (P<0.001), total gray matter (P=0.002)
• Safety: serious AEs 6.9% vs 17.6%; GI AEs and 5–10% weight loss more frequent with liraglutide
Clinical Question
In non-diabetic adults with mild-to-moderate Alzheimer's disease, does 52 weeks of subcutaneous liraglutide 1.8 mg/day slow the decline in cerebral glucose metabolism (FDG-PET), cognition (ADAS-Exec, CDR-SoB, ADCS-ADL) and brain volume compared with placebo?
Bottom Line
Liraglutide 1.8 mg SC daily for 52 weeks did NOT reduce decline in cerebral glucose metabolism (primary outcome) in mild-to-moderate Alzheimer's disease, but showed an exploratory cognitive benefit on the composite ADAS-Exec (Δ 0.15 z-score; 95% CI 0.03–0.28; unadjusted P=0.01) and less temporal-lobe and gray-matter atrophy. Safety was acceptable, with fewer serious AEs than placebo (6.9% vs 17.6%) and the expected GI/weight-loss profile.
Major Points
- Largest randomized Phase 2b trial of a GLP-1 receptor agonist in Alzheimer's disease to date (n=204, 24 UK sites, 12-month treatment) — the ELAD trial.
- Negative for the imaging primary outcome: composite cerebral glucose metabolic rate (FDG-PET SUV across hippocampus, medial temporal lobe, posterior cingulate) — adjusted Δ −0.17 (95% CI −0.39 to 0.06), P=0.14; spectral-analysis sensitivity confirmed.
- Positive cognitive secondary signal on ADAS-Exec (composite of ADAS-Cog + NTB Executive z-scores): liraglutide arm declined 0.15 SD less than placebo (95% CI 0.03–0.28), unadjusted P=0.01. CDR-SoB and ADCS-ADL not different.
- Exploratory MRI: liraglutide associated with attenuated temporal lobe atrophy (Δ 696 mm³, 95% CI 184–1208, P<0.001), less whole gray matter loss (Δ 7,274 mm³, P=0.002), and trends in parietal and frontoparietal lobes. No hippocampal or entorhinal differences.
- Safety: total AEs slightly higher with liraglutide (541 vs 450) driven by GI events (25.5% of all liraglutide AEs); serious AEs were LESS frequent on liraglutide — 6.9% (7/102) vs 17.6% (18/102) placebo. Weight loss ≥5%: 48.1% liraglutide vs 13.7% placebo.
- Findings are exploratory and not multiplicity-corrected; reduced sample size (vs original power calculation) and 30% drop-off of analyzable scans in the treatment arm limit definitive conclusions. Larger confirmatory Phase 3 trials of GLP-1 analogs in AD are warranted (oral formulation may improve tolerability).
Study Design
- Study Type
- Multicenter, double-blind, placebo-controlled, randomized Phase 2b clinical trial
- Randomization
- Yes
- Blinding
- Double-blind (participants, investigators, outcome assessors)
- Sample Size
- 204
- Follow-up
- 52 weeks (with optional 12-month open-label extension, not reported here)
- Centers
- 24
- Countries
- United Kingdom
Primary Outcome
Definition: Change in cerebral glucose metabolic rate (rCMRglc) on [18F]FDG-PET in a composite cortical ROI (hippocampus + medial temporal lobe + posterior cingulate) from baseline to 52 weeks (SUV method, ANCOVA adjusted for baseline, age, MMSE)
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| Reference (placebo) | Δ vs placebo: −0.17 | - (−0.39 to 0.06) | 0.14 |
Limitations & Criticisms
- Primary outcome (FDG-PET SUV change) was NEGATIVE; the positive cognitive signal on ADAS-Exec was an unadjusted, exploratory secondary endpoint and should be interpreted as hypothesis-generating.
- Achieved sample size and analyzable-scan numbers fell below the original power calculation (only 72 of 102 treatment-arm participants had analyzable week-52 scans), reducing statistical power for the primary endpoint.
- Differential attrition: 70.6% of liraglutide vs 80.4% of placebo had analyzable scans at week 52 — potential informative dropout.
- Population was overwhelmingly White (only 4/204 identified as non-White), limiting generalizability across ethnic groups; UK-only, single tertiary imaging center.
- Study drug discontinued 3 days before week-52 PET (rationale: minimize acute glucose effects), which may have masked any short-term metabolic benefit and differs from a prior 26-week pilot that maintained dosing through scan.
- ELAD did NOT enroll diabetics — efficacy in the AD + T2D population (where signal in prior post-hoc analyses was strongest) was not tested.
- Once-daily subcutaneous injections impose substantial trial burden — future Phase 3 trials of oral semaglutide (e.g., EVOKE/EVOKE+) will be more clinically translatable.
- Enrollment spanned 2014–2021 (including COVID-19 disruption), and follow-up MRI/PET volumes are exploratory — confirmatory data needed.
Citation
Nat Med. 2026 Jan;32(1):353–363. doi:10.1038/s41591-025-04106-7. ClinicalTrials.gov NCT01843075.