What were the key findings of the EVOKE trial?

← Back

EVOKE

Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials

Share Share Copied! Compare

Year of Publication: 2026

Authors: Cummings JL, Atri A, Sano M, ..., Feldman HH; evoke and evoke+ Steering Committees and Investigators

Journal: Lancet

Citation: Lancet 2026;407(10544):2167-2179

Link: https://doi.org/10.1016/S0140-6736(26)00459-9

Bottom Line

In two large phase 3 trials totalling 3808 amyloid-positive early Alzheimer's disease participants, oral semaglutide up to 14 mg once daily did NOT slow CDR-SB progression versus placebo at week 104, despite significant reductions in CSF p-tau and neurodegeneration biomarkers and in hsCRP; both trials were discontinued for negative clinical outcome.

        Major Points
        Twin multicentre phase 3 RCTs (evoke, evoke+) at 566 sites across 40 countries; 3808 amyloid-confirmed early AD participants (72% MCI, 27% mild dementia) randomised 1:1 to oral semaglutide up to 14 mg daily or placebo for up to 156 weeks.
Primary endpoint (change in CDR-SB from baseline to week 104) was not met in either trial: estimated difference -0.08 (95% CI -0.35 to 0.20; p=0.57) in evoke and 0.10 (-0.17 to 0.38; p=0.46) in evoke+.
Confirmatory and supportive secondary endpoints (ADCS-ADL-MCI, ADAS-Cog-13, MoCA, MMSE, ADCOMS, NPI, EQ-5D-5L, composite Z-score, time to CDR global >=1.0) were all null; testing hierarchy stopped after failed primary.
CSF substudy (n=199) showed significant semaglutide-driven reductions of up to ~10% in p-tau181, p-tau217, non-phosphorylated tau181/205, total tau, YKL-40 and neurogranin; plasma GFAP rose slightly; hsCRP decreased. Biomarker signal did not translate into clinical benefit.
Safety was consistent with GLP-1 class effects: TEAEs 91.2% vs 84.8%; AEs leading to discontinuation ~17% vs ~8%; SAEs 20.4% vs 23.8%; treatment-related fatalities 1 (semaglutide) vs 4 (placebo). Bodyweight fell 5.8% (-4.3 kg) with semaglutide vs +0.6% (0.2 kg) with placebo. Falls and fractures were numerically fewer on semaglutide.
Authors interpret the null result as possibly reflecting limited CNS bioavailability of semaglutide, later-stage symptomatic AD population (vs primary prevention in T2D real-world cohorts), and 24-month exposure; earlier-stage or asymptomatic populations may still merit investigation.

      

Design

Study Type: Two parallel multicentre, randomised, double-blind, placebo-controlled, parallel-group phase 3 trials (evoke and evoke+)

Randomization: 1

Blinding: Double-blind (participants, care providers, investigators and outcome assessors masked)

Enrollment Period: May 18, 2021 to Sept 8, 2023

Follow-up Duration: Up to 156 weeks (primary analysis at week 104; 52-week extension)

Centers: 566

Countries: 40 countries (global)

Sample Size: 3808

Analysis: Hybrid hypothetical-treatment policy-composite (HTC) estimand; MMRM for CDR-SB with treatment, visit, APOE-ε4 status, sex, region, baseline AD medication as factors and baseline age as covariate; efficacy in full analysis set (all randomised), safety in all who received ≥1 dose

Inclusion Criteria

Adults aged 55–85 years with mild cognitive impairment (MCI) or mild dementia due to Alzheimer's disease per NIA-AA 2018 criteria
Clinical Dementia Rating (CDR) global score of 0.5 with CDR ≥0.5 in at least one instrumental ADL category, OR CDR global score of 1.0
Repeated Battery for the Assessment of Neuropsychological Status (RBANS) delayed memory index score ≤85
Mini-Mental State Examination (MMSE) score ≥22
Amyloid positivity confirmed by amyloid PET (visual read) or CSF Aβ1-42 or CSF Aβ1-42/Aβ1-40 ratio
If on approved AD symptomatic therapy at screening (donepezil, rivastigmine, galantamine, memantine or aducanumab), stable dose ≥3 months before screening
evoke+ only: participants with significant small vessel pathology (>1 lacunar infarct and/or Age-Related White Matter Changes [ARWMC] score >2, white matter >20 mm) allowed

Exclusion Criteria

Non-Alzheimer's cause of cognitive impairment / non-AD dementia
In the evoke trial: significant small vessel pathology (as defined for evoke+ inclusion)
Advanced dementia (CDR global > 1.0) or scores outside protocol thresholds (RBANS delayed memory index >85 or MMSE <22)
Contraindications to GLP-1 receptor agonist therapy (per protocol; e.g. personal/family history of medullary thyroid carcinoma or MEN2, history of pancreatitis)
Unstable concomitant Alzheimer's disease treatment or planned changes during trial (aside from medically necessary)
Standard trial exclusions per published protocol (see Cummings et al 2025, appendix 1)

Baseline Characteristics

Characteristic	Control	Active
N (pooled placebo)	1904 (evoke 927; evoke+ 977)
Mean Age (years)	evoke 71.7 (SD 7.1); evoke+ 72.5 (SD 7.2)	evoke 71.9 (SD 7.0); evoke+ 72.6 (SD 7.0)
Sex - Female	evoke 53.5%; evoke+ 51.0%	evoke 52.7%; evoke+ 52.8%
Race - White	evoke 82.5%; evoke+ 74.3%	evoke 79.2%; evoke+ 74.3%
Race - Asian	evoke 11.0%; evoke+ 21.0%	evoke 12.6%; evoke+ 20.7%
Ethnicity - Hispanic/Latino	evoke 14.0%; evoke+ 8.7%	evoke 12.3%; evoke+ 12.0%
BMI >=30 kg/m2	evoke 15.0%; evoke+ 15.8%	evoke 15.3%; evoke+ 16.0%
Type 2 diabetes	evoke 12.5%; evoke+ 15.1%	evoke 10.7%; evoke+ 15.9%
APOE-ε4 carrier (hetero)	evoke 48.3%; evoke+ 44.7%	evoke 48.6%; evoke+ 45.4%
APOE-ε4 homozygote	evoke 12.9%; evoke+ 12.5%	evoke 12.6%; evoke+ 11.4%
CDR global 0.5 (MCI)	evoke 73.4%; evoke+ 71.3%	evoke 72.2%; evoke+ 71.6%
CDR global 1.0 (mild dementia)	evoke 25.9%; evoke+ 27.5%	evoke 27.2%; evoke+ 28.0%
Mean CDR-SB (SD)	evoke 3.7 (1.5); evoke+ 3.7 (1.7)	evoke 3.8 (1.6); evoke+ 3.7 (1.5)
Mean ADCS-ADL-MCI (SD)	evoke 39.7 (7.4); evoke+ 39.2 (7.5)	evoke 39.2 (7.3); evoke+ 38.7 (7.5)
Mean MMSE (SD)	evoke 24.0 (3.1); evoke+ 24.1 (3.1)	evoke 24.1 (2.9); evoke+ 24.0 (3.1)
hsCRP mg/L (geometric mean)	evoke 0.9; evoke+ 0.8	evoke 0.8; evoke+ 0.9
Any AD concomitant medication	evoke 56.9%; evoke+ 53.7%	evoke 60.2%; evoke+ 55.9%
Donepezil	evoke 35.6%; evoke+ 35.3%	evoke 39.0%; evoke+ 35.5%
Memantine	evoke 12.7%; evoke+ 12.8%	evoke 15.3%; evoke+ 14.8%
Small vessel pathology (evoke+ only)	2.9%	2.7%
N (pooled semaglutide)		1904 (evoke 928; evoke+ 976)

Arms

Field	Oral semaglutide 14 mg	Control
Intervention	Oral semaglutide titrated (3 mg → 7 mg → 14 mg) to 14 mg once daily with flexible dosing on top of standard-of-care AD treatment	Matching oral placebo once daily on top of standard-of-care AD treatment
Duration	Up to 156 weeks (primary analysis at week 104)	Up to 156 weeks (primary analysis at week 104)

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Change in Clinical Dementia Rating–Sum of Boxes (CDR-SB) from baseline to week 104 (per trial)	Primary	evoke +2.3 (SE 0.1); evoke+ +2.1 (SE 0.1)	evoke +2.3 (SE 0.1); evoke+ +2.2 (SE 0.1)		0.57 (evoke); 0.46 (evoke+)
Change in ADCS-ADL-MCI baseline→wk 104	Secondary				Not tested (hierarchy stopped)
Time to progression to CDR global ≥1.0 in participants with baseline CDR-G 0.5 (per trial)	Secondary			evoke 0.98 (95% CI 0.85–1.14); evoke+ 0.96 (0.83–1.12)	NS
Change in composite Z-score (CDR-SB + ADCS-ADL-MCI + ADAS-Cog-13) at wk 104	Secondary				NS
ADAS-Cog-13, MoCA, ADCOMS, MMSE, NPI (secondary supportive)	Secondary	No differences between treatment groups (appendix 1)
hsCRP at week 104 (ratio to baseline)	Secondary				Significant
EQ-5D-5L index score change baseline→wk 104	Secondary	No difference between groups
Body weight change baseline→wk 104 (pooled)	Secondary	+0.6% (+0.2 kg)	−5.8% (−4.3 kg)
Plasma NfL (ETR wk 104)	Secondary	evoke 1.03 (95% CI 1.00–1.07) NS; evoke+ 1.05 (1.01–1.08) unadjusted p=0.0137
Plasma GFAP (ETR wk 104)	Secondary	evoke 1.05 (95% CI 1.02–1.08) unadjusted p=0.0012; evoke+ 1.07 (1.04–1.10) unadjusted p<0.0001
CSF substudy (n=199) at wk 78 – p-tau181, p-tau217, np-tau181, np-tau205, total tau, YKL-40, neurogranin	Secondary	Semaglutide reduced canonical AD and neurodegeneration/neuroinflammation CSF markers by up to ~10% vs placebo
MACE (non-adjudicated) cumulative incidence at wk 104	Secondary	0.03 (95% CI 0.02–0.04)	0.03 (95% CI 0.03–0.04)
Stroke cumulative incidence at wk 104	Secondary	0.01 (95% CI 0.01–0.02)	0.01 (95% CI 0.01–0.01)
Any TEAE (pooled)	Adverse	Semaglutide 91.2% (1729/1896) vs Placebo 84.8% (1613/1902)
Any TEAE evoke	Adverse	Semaglutide 90.6% vs Placebo 85.5%
Any TEAE evoke+	Adverse	Semaglutide 91.8% vs Placebo 84.1%
Serious AE evoke	Adverse	Semaglutide 19.0% vs Placebo 25.0%
Serious AE evoke+	Adverse	Semaglutide 21.6% vs Placebo 22.7%
AE leading to permanent discontinuation evoke	Adverse	17.2% vs 8.1%
AE leading to permanent discontinuation evoke+	Adverse	16.7% vs 8.6%
Fatal events evoke	Adverse	Semaglutide 3.0% vs Placebo 2.5%
Fatal events evoke+	Adverse	Semaglutide 2.4% vs Placebo 2.9%
Treatment-related fatalities	Adverse	1 (semaglutide; haemorrhagic stroke in participant with hypertension, atherosclerotic coronary disease, HF, carotid stenosis) vs 4 (placebo; congestive HF, metabolic acidosis, intestinal ischaemia, undetermined)
Weight decreased evoke	Adverse	36.3% vs 6.5%
Weight decreased evoke+	Adverse	36.7% vs 8.3%
Decreased appetite evoke	Adverse	31.7% vs 5.7%
Decreased appetite evoke+	Adverse	34.4% vs 6.3%
Nausea evoke	Adverse	25.2% vs 6.9%
Nausea evoke+	Adverse	23.4% vs 6.8%
Diarrhoea evoke	Adverse	14.7% vs 11.2%
Diarrhoea evoke+	Adverse	14.3% vs 8.3%
Vomiting evoke	Adverse	12.3% vs 4.1%
Vomiting evoke+	Adverse	12.3% vs 3.2%
COVID-19 evoke	Adverse	10.5% vs 11.1%
COVID-19 evoke+	Adverse	11.2% vs 11.4%
Falls evoke	Adverse	6.0% vs 9.2%
Falls evoke+	Adverse	7.4% vs 9.0%
Bone/joint injuries or fractures evoke	Adverse	5.0% vs 8.8%
Bone injuries evoke+	Adverse	6.3% vs 7.5%

Subgroup Analysis

Post-hoc subgroup analyses for age, sex, and type 2 diabetes status supported the null primary result; no clinically meaningful heterogeneity of effect on CDR-SB at week 104 (appendix 1 pp 42–43).

Criticisms

Both trials were discontinued for negative clinical outcome; primary endpoint definitively null despite 95% powered design.
Potential functional unblinding from characteristic GLP-1 adverse events (weight loss 36% vs ~7%; nausea, appetite loss) although similar-sounding symptoms occurred in placebo.
Studied a late (already symptomatic) AD population; the biological hypothesis (delaying incident AD) is best tested in primary prevention or preclinical AD populations, not MCI/mild dementia.
Oral formulation has limited CNS bioavailability (CSF geometric mean 0.08 nmol/L; ratio ~1:220 to plasma) and modest CSF biomarker effects (~5–10% vs up to 25% with anti-amyloid mAbs) — CNS exposure may have been insufficient.
Small vessel pathology inclusion criterion in evoke+ was met by only 2.8% of participants, so the additional cerebrovascular-mixed-pathology question was underpowered.
Follow-up limited to 104 weeks (primary) / up to 156 weeks (extension); longer exposure durations in real-world cohorts might explain divergence from observational signals.
Trials sponsored, run and analysed by Novo Nordisk (funder); several authors are Novo Nordisk employees/shareholders.

Funding

Novo Nordisk (sponsor, data collection and analysis; independent Data Monitoring Committee).

Based on: EVOKE (Lancet, 2026)

Authors: Cummings JL, Atri A, Sano M, ..., Feldman HH; evoke and evoke+ Steering Committees and Investigators

Citation: Lancet 2026;407(10544):2167-2179

Content summarized and formatted by NeuroTrials.ai.

EVOKE

(2026)

Objective

To assess whether oral semaglutide up to 14 mg once daily slows cognitive and functional decline in adults with amyloid-confirmed early Alzheimer's disease (mild cognitive impairment or mild dementia).

Study Summary

• Oral semaglutide 14 mg did NOT slow decline on CDR-SB at week 104: estimated difference vs placebo −0·08 (95% CI −0·35 to 0·20; p=0·57) in evoke and 0·10 (−0·17 to 0·38; p=0·46) in evoke+.
• ADCS-ADL-MCI at week 104 showed no benefit (evoke −0·25 [95% CI −1·22 to 0·72]; evoke+ −0·03 [−0·97 to 0·91]); pooled time to CDR-G ≥1·0 also unchanged (HR 0·98 evoke, 0·96 evoke+).
• CSF substudy: semaglutide reduced canonical AD biomarkers (p-tau181, p-tau217, np-tau) and neuroinflammation/neurodegeneration markers (YKL-40, total tau, neurogranin) by up to ≈10%; plasma hsCRP fell and bodyweight dropped 5·8% (−4·3 kg) vs +0·6% with placebo.
• Safety: TEAEs 91·2% vs 84·8% (placebo); AEs leading to discontinuation ~17% vs ~8%; treatment-related fatalities 1 (semaglutide, haemorrhagic stroke) vs 4 (placebo). Both trials were discontinued for lack of efficacy.

Intervention

Oral semaglutide titrated to 14 mg once daily (flexible dosing) added to standard of care, versus matching placebo, for up to 156 weeks.

Inclusion Criteria

Adults 55–85 years with MCI or mild dementia due to Alzheimer's disease (NIA-AA 2018), amyloid positivity by PET or CSF, CDR global 0·5 or 1·0, RBANS delayed memory index ≤85, MMSE ≥22. evoke+ additionally allowed significant small vessel pathology.

Study Design

Arms: Semaglutide 14 mg once daily (flexible dose) vs matching placebo, randomised 1:1, double-blind, added to standard of care for up to 156 weeks.

Patients per Arm: evoke: semaglutide 928, placebo 927; evoke+: semaglutide 976, placebo 977 (pooled N=3808).

Outcome

• Primary (ΔCDR-SB baseline→wk 104): semaglutide vs placebo difference −0·08 (95% CI −0·35 to 0·20; p=0·57) in evoke; 0·10 (−0·17 to 0·38; p=0·46) in evoke+ — not significant.
• Secondary: no difference in ADCS-ADL-MCI, ADAS-Cog-13, MMSE, MoCA, ADCOMS, NPI, EQ-5D-5L; no delay in time to CDR-G ≥1·0 (evoke HR 0·98 [0·85–1·14]; evoke+ 0·96 [0·83–1·12]).
• Biomarkers: CSF p-tau181, p-tau217, np-tau, YKL-40, total tau, neurogranin reduced up to 10% with semaglutide; plasma GFAP rose slightly; hsCRP fell.
• Safety: TEAEs 91·2% vs 84·8%; SAEs 20·4% vs 23·8%; commonest AEs weight decreased 36·5%, decreased appetite 33·1%, nausea 24·3%; falls and fractures numerically lower with semaglutide.

Clinical Question

Does oral semaglutide 14 mg once daily slow cognitive and functional decline over 104 weeks in adults with amyloid-confirmed early Alzheimer's disease (MCI or mild dementia)?

Bottom Line

Major Points

Twin multicentre phase 3 RCTs (evoke, evoke+) at 566 sites across 40 countries; 3808 amyloid-confirmed early AD participants (72% MCI, 27% mild dementia) randomised 1:1 to oral semaglutide up to 14 mg daily or placebo for up to 156 weeks.
Primary endpoint (change in CDR-SB from baseline to week 104) was not met in either trial: estimated difference -0.08 (95% CI -0.35 to 0.20; p=0.57) in evoke and 0.10 (-0.17 to 0.38; p=0.46) in evoke+.
Confirmatory and supportive secondary endpoints (ADCS-ADL-MCI, ADAS-Cog-13, MoCA, MMSE, ADCOMS, NPI, EQ-5D-5L, composite Z-score, time to CDR global >=1.0) were all null; testing hierarchy stopped after failed primary.
CSF substudy (n=199) showed significant semaglutide-driven reductions of up to ~10% in p-tau181, p-tau217, non-phosphorylated tau181/205, total tau, YKL-40 and neurogranin; plasma GFAP rose slightly; hsCRP decreased. Biomarker signal did not translate into clinical benefit.
Safety was consistent with GLP-1 class effects: TEAEs 91.2% vs 84.8%; AEs leading to discontinuation ~17% vs ~8%; SAEs 20.4% vs 23.8%; treatment-related fatalities 1 (semaglutide) vs 4 (placebo). Bodyweight fell 5.8% (-4.3 kg) with semaglutide vs +0.6% (0.2 kg) with placebo. Falls and fractures were numerically fewer on semaglutide.
Authors interpret the null result as possibly reflecting limited CNS bioavailability of semaglutide, later-stage symptomatic AD population (vs primary prevention in T2D real-world cohorts), and 24-month exposure; earlier-stage or asymptomatic populations may still merit investigation.

Study Design

Study Type: Two parallel multicentre, randomised, double-blind, placebo-controlled, parallel-group phase 3 trials (evoke and evoke+)
Randomization: Yes
Blinding: Double-blind (participants, care providers, investigators and outcome assessors masked)
Sample Size: 3808
Follow-up: Up to 156 weeks (primary analysis at week 104; 52-week extension)
Centers: 566
Countries: 40 countries (global)

Primary Outcome

Definition: Change in Clinical Dementia Rating–Sum of Boxes (CDR-SB) from baseline to week 104 (per trial)

Control	Intervention	HR/OR	P-value
evoke +2.3 (SE 0.1); evoke+ +2.1 (SE 0.1)	evoke +2.3 (SE 0.1); evoke+ +2.2 (SE 0.1)	- (evoke −0.35 to 0.20; evoke+ −0.17 to 0.38)	0.57 (evoke); 0.46 (evoke+)

Limitations & Criticisms

Both trials were discontinued for negative clinical outcome; primary endpoint definitively null despite 95% powered design.
Potential functional unblinding from characteristic GLP-1 adverse events (weight loss 36% vs ~7%; nausea, appetite loss) although similar-sounding symptoms occurred in placebo.
Studied a late (already symptomatic) AD population; the biological hypothesis (delaying incident AD) is best tested in primary prevention or preclinical AD populations, not MCI/mild dementia.
Oral formulation has limited CNS bioavailability (CSF geometric mean 0.08 nmol/L; ratio ~1:220 to plasma) and modest CSF biomarker effects (~5–10% vs up to 25% with anti-amyloid mAbs) — CNS exposure may have been insufficient.
Small vessel pathology inclusion criterion in evoke+ was met by only 2.8% of participants, so the additional cerebrovascular-mixed-pathology question was underpowered.
Follow-up limited to 104 weeks (primary) / up to 156 weeks (extension); longer exposure durations in real-world cohorts might explain divergence from observational signals.
Trials sponsored, run and analysed by Novo Nordisk (funder); several authors are Novo Nordisk employees/shareholders.

Citation

Lancet 2026;407(10544):2167-2179

View Original Publication →

EVOKE

Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials

Bottom Line

Major Points

Design

Inclusion Criteria

Exclusion Criteria

Baseline Characteristics

Arms

Outcomes

Subgroup Analysis

Criticisms

Funding

Ask about EVOKE