Liraglutide in mild to moderate Alzheimer's disease: a phase 2b clinical trial
Edison P, Hooper N, Femminella GD, Frangou E, Holmes C, Robinson L, Love S, Ritchie CW, +2 more
Nature Medicine (2026) · n=204
The ELAD Phase 2b in 204 non-diabetic adults with mild-to-moderate Alzheimer's disease missed its primary endpoint of cerebral glucose metabolic rate (p=0.14) — but secondary endpoints reported nearly 50% reduction in brain atrophy and 18% slower ADAS-Exec decline, the secondary-positive signal that handed the question to the larger semaglutide EVOKE Phase 3.
This is the ELAD ("Evaluating Liraglutide in Alzheimer's Disease") Phase 2b investigator-initiated trial published in Nature Medicine in January 2026 and the most-discussed peptide-and-Alzheimer's clinical-trial readout of the 2026 cycle. The trial was led by Paul Edison's group at Imperial College London across 24 UK clinical sites (NCT01843075). The design tested whether liraglutide — the once-daily GLP-1 receptor agonist FDA-approved as Victoza for type 2 diabetes (2010) and Saxenda for chronic weight management (2014) — could engage Alzheimer's-disease-relevant brain-metabolism and structural endpoints in a non-diabetic mild-to-moderate AD population.
Investigators randomized 204 non-diabetic adults with mild-to-moderate Alzheimer's-disease syndrome 1:1 to once-daily subcutaneous liraglutide (escalated to 1.8 mg/day) or matching placebo for 52 weeks, with double-blind treatment assignment. All participants underwent baseline and end-of-treatment 18F-FDG PET imaging for cerebral glucose metabolic rate, volumetric MRI for brain-atrophy assessment, and a detailed cognitive-and-functional neuropsychometric battery covering ADAS-Cog, ADAS-Exec, Clinical Dementia Rating-Sum of Boxes (CDR-SoB), and ADCS-Activities of Daily Living. The prespecified primary endpoint was change in cerebral glucose metabolic rate by 18F-FDG PET — selected on the hypothesis that GLP-1-receptor agonism would engage central brain-energetics pathways and produce measurable improvement in regional glucose utilization before downstream cognitive-functional changes manifested.
The primary endpoint was not met. Between-group difference in cerebral glucose metabolic rate was −0.17 (95% CI −0.39 to 0.06; p = 0.14) — directionally favoring liraglutide but not statistically significant. The secondary-endpoint signals are what drove the publication and the post-trial discussion. Brain atrophy on volumetric MRI was reduced by nearly 50% in the liraglutide arm versus placebo. The ADAS-Exec executive-function subscale showed an 18% slower decline (0.15; 95% CI 0.03 to 0.28; unadjusted p = 0.01). CDR-SoB and ADCS-ADL secondary endpoints did not separate (p = 0.81 and p = 0.65 respectively). Safety was acceptable across the trial period, consistent with the broader liraglutide safety database. The authors framed the result as supportive of further GLP-1-class investigation in Alzheimer's disease, with the structural-and-executive-function secondary signals warranting confirmatory work; that confirmatory program was handed to semaglutide via the EVOKE/EVOKE+ Phase 3 pair, which announced primary-negative topline results on 24 November 2025.
Cross-link to the liraglutide peptide page, to the Alzheimer's-and-peptides dossier where ELAD is the load-bearing 2026 entry, and to the failed-peptide-trials archive for the broader primary-negative / secondary-positive framing context.
The primary-endpoint result is negative on the prespecified test, and the secondary-endpoint signals (brain atrophy, ADAS-Exec) carry the multiple-comparisons buffer that secondary endpoints in a primary-negative trial conventionally carry — the unadjusted p = 0.01 on ADAS-Exec is not the same as a primary endpoint at 0.01. The 18F-FDG PET primary was chosen on a brain-energetics-engagement rationale rather than a cognitive-functional rationale; the regulatory standard for Alzheimer's-disease-modifying claims is CDR-SoB or comparable cognitive-functional change, and CDR-SoB did not separate in this trial. The 52-week treatment duration is moderate by Alzheimer's-trial standards; the structural-atrophy signal in particular has been an area where short-trial effects have not always reproduced at longer durations and larger sample sizes. The sample size of 204 is appropriate for Phase 2b proof-of-concept on imaging and cognitive endpoints but underpowered for definitive cognitive-functional efficacy claims. The trial was investigator-initiated, with Novo Nordisk supplying liraglutide as drug rather than running the program; industry independence is high relative to a sponsor-driven program. Liraglutide remains FDA-approved as Victoza and Saxenda but the AD indication did not advance to Phase 3 because development moved to the more potent and longer-acting semaglutide. The 2025 Phase 3 EVOKE/EVOKE+ readout — primary-negative on CDR-SoB on 24 November 2025 — is the most-consequential follow-on signal, and is best read as the larger trial answering the question this Phase 2b raised more cleanly than ELAD itself answered it. The combined reading: GLP-1 agonism at clinically-used monotherapy doses does not slow Alzheimer's-disease progression on regulatory-grade cognitive primaries; the secondary structural and executive-function signals from ELAD have not translated to clinical-primary endpoints in the Phase 3 program. The neuroprotection hypothesis remains biologically plausible and supported by the observational pharmacoepidemiology in diabetic populations (Nørgaard et al. 2022) but has not survived the prospective interventional confirmation at clinically-relevant doses and durations.
07·Member discussion
No member discussion yet.
Member-only conversation lives here — cycle notes, practitioner commentary, pattern-matching. Be the first paying member to start the thread.