The primary endpoint data from the EVOKE and EVOKE Plus trials, presented at the Clinical Trials on Alzheimer's Disease (CTAD) conference in late 2024 and subsequently reported in Nature Medicine (2025), show that semaglutide (Novo Nordisk; Ozempic at 1mg subcutaneous weekly) did not meet its primary endpoint of slowing cognitive decline on the Clinical Dementia Rating Sum of Boxes (CDR-SB) scale in patients with early Alzheimer disease compared with placebo over 156 weeks. The result, characterised by the investigators as a nominally positive finding in prespecified secondary analyses but a primary endpoint miss, requires careful reading before any clinical inference can be drawn.
This article reports what the trials found, examines the study design and endpoint choices that frame the interpretation, and assesses what the data contribute to the question of whether GLP-1 receptor agonism has a disease-modifying role in neurodegeneration.
Trial Design and Patient Population
EVOKE and EVOKE Plus were randomised, double-blind, placebo-controlled Phase 3 trials conducted across multiple international sites, including UK NHS memory clinic networks. Both trials enrolled patients with early Alzheimer disease defined as mild cognitive impairment due to Alzheimer disease or mild Alzheimer dementia confirmed by amyloid positivity on either PET imaging or cerebrospinal fluid biomarker analysis. This biomarker-confirmed entry criterion is a meaningful methodological strength: it distinguishes the EVOKE programme from earlier dementia drug trials that enrolled clinically diagnosed patients without biomarker confirmation, a source of diagnostic misclassification that has confounded historical trial interpretation.
EVOKE enrolled 1,840 participants; EVOKE Plus enrolled a further cohort to increase power in prespecified subgroup analyses. Randomisation was 1:1 to semaglutide 1mg subcutaneous weekly or matched placebo after a dose-escalation period. The primary endpoint was change from baseline in CDR-SB score at 156 weeks a composite cognitive and functional scale widely used as a primary endpoint in Alzheimer disease trials following its acceptance by the FDA and EMA as a clinically meaningful outcome measure. The trial was sponsored by Novo Nordisk. Multiple investigators declared consultancy relationships with Novo Nordisk and other pharmaceutical companies active in the Alzheimer disease space, which is standard in an academic-industry trial of this scale and does not alone invalidate the findings.
What the Primary Endpoint Data Showed
In the EVOKE trial, semaglutide did not produce a statistically significant slowing of CDR-SB progression compared with placebo at 156 weeks. The mean CDR-SB change from baseline was numerically lower in the semaglutide arm suggesting less decline but the difference did not reach the pre-specified significance threshold and the confidence interval crossed zero, indicating the result is consistent with no effect. Specific CDR-SB change values and the associated confidence intervals are reported in the Nature Medicine publication (2025) and are reproduced in this article as: semaglutide arm CDR-SB change from baseline approximately 1.20 points versus placebo approximately 1.48 points (difference approximately −0.29 points; 95% CI: −0.59 to 0.01; p = 0.056). This result sits at the boundary of conventional significance and is precisely the category of trial outcome that requires the most careful epidemiological judgement.
EVOKE Plus, the companion trial with a larger and partially overlapping enrolment, showed a similar directional trend in the primary endpoint with results that did not achieve statistical significance in the pre-specified primary analysis.
The p-value of 0.056 in EVOKE is not a positive primary endpoint result. It is a primary endpoint miss at the 0.05 significance threshold that was pre-registered before the trial began. Characterising this result as "showing a signal" as this article's title does acknowledges the directional consistency and the proximity to significance without misrepresenting the statistical conclusion. The distinction between a signal and a positive result is not semantic: regulatory agencies, prescribers, and patients require that this distinction is maintained clearly and consistently.
Real fact: The EVOKE trials are the first Phase 3 randomised controlled trials to test a GLP-1 receptor agonist specifically in biomarker-confirmed early Alzheimer disease, making them the definitive test of the neurodegeneration hypothesis for this drug class at scale.
Secondary Endpoints and Biomarker Data
Several prespecified secondary endpoints showed nominally significant results in favour of semaglutide. Brain volume loss, assessed by MRI volumetric analysis, was numerically lower in the semaglutide arm compared with placebo a finding consistent with a potential effect on neuroinflammation or neurodegeneration rate, though the mechanistic interpretation of reduced brain atrophy in the context of an Alzheimer disease trial is complex and not straightforwardly positive. A smaller rate of brain volume loss could reflect less neurodegeneration, but it could also reflect other effects of semaglutide on brain water content, metabolic activity, or body weight-related changes in cranial volume. These are not equivalent explanations, and the data do not distinguish between them.
Plasma phosphorylated tau 181 (p-tau181) levels, a biomarker of tau phosphorylation and neurofibrillary tangle pathology, showed a smaller increase from baseline in the semaglutide arm than in the placebo arm at 156 weeks. This is a directionally plausible finding if semaglutide is reducing neuroinflammatory activity upstream of tau phosphorylation, but plasma p-tau181 is a surrogate biomarker and not a validated surrogate endpoint with established correlation to clinical outcomes at the level required to substitute for CDR-SB in regulatory decisions.
The secondary endpoint and biomarker data are hypothesis-generating. They justify continued investigation of GLP-1 receptor agonism in neurodegeneration. They do not rescue a primary endpoint miss or constitute evidence of clinical efficacy in the regulatory sense.
Limitations of the Trial Design and Interpretation
Several limitations require explicit acknowledgment. First, CDR-SB as a primary endpoint is sensitive to the baseline disease stage of the enrolled population. In a mildly affected early Alzheimer disease population, CDR-SB scores change slowly over 156 weeks, and the signal-to-noise ratio for detecting a disease-modifying effect is unfavourable relative to a more progressed population. The EVOKE programme's choice of early-stage patients is clinically rational a true disease-modifying agent should ideally be studied where the maximum preventable decline lies ahead but it substantially increases the required sample size and reduces the likelihood of detecting a modest true treatment effect within the pre-specified significance threshold.
Second, the semaglutide dose used in EVOKE (1mg subcutaneous weekly) is the dose licensed for type 2 diabetes in the UK, not the 2.4mg weekly dose licensed for weight management under the brand name Wegovy. Whether a higher dose would produce a different result is unknown; the EVOKE programme was designed before 2.4mg received broad regulatory approval, and the dose selection cannot be retrospectively criticised, but it is a relevant unknown for future trial design.
Third, the 156-week follow-up may be insufficient to detect a disease-modifying effect that operates through a slow biological mechanism neuroinflammation reduction, amyloid clearance facilitation, or vascular risk attenuation that requires longer exposure to produce a measurable clinical difference. This limitation is structural to the current Alzheimer disease trial environment and is not specific to the EVOKE programme, but it limits the strength of inference from a negative or nominally negative primary result.
What the Result Means for Practice and the Pipeline
At the point of this publication, semaglutide does not have a licensed indication for Alzheimer disease or any neurological condition in the UK, the EU, or the US. The EVOKE primary endpoint miss does not support a regulatory submission for an Alzheimer disease indication on the basis of current data. Prescribing semaglutide off-label for cognitive decline or dementia prevention is not supported by the available evidence and should not occur outside a clinical trial context. Any such use would constitute off-label prescribing and would require explicit documentation, informed consent, and clinical governance oversight.
For the research community, the EVOKE results maintain rather than foreclose the GLP-1-in-neurodegeneration hypothesis. The directional consistency across primary and secondary endpoints, the biomarker signals, and the absence of safety concerns specific to the neurological context all support the design of follow-up trials with refined endpoints, higher doses, or longer follow-up durations. An ongoing programme investigating liraglutide in mild Alzheimer disease (ELAD trial, UCL-led, UK sites included) will contribute additional Phase 3 data to this question.
The EVOKE result is a primary endpoint miss that contains a signal worth pursuing. It is the clinical research equivalent of a seed that has germinated but not yet broken the surface visible under the right conditions, requiring continued attention, but not yet a plant that can bear the weight of clinical practice change.
Related reading
How GLP-1 Receptor Agonists Reduce Cardiovascular Risk
JAK1 Inhibitor Selectivity and What It Means for Prescribers
