Cite as: Archiv EuroMedica. 2025. 15; 5. DOI 10.35630/2025/15/Iss.5.501
Background: Alzheimer’s disease is increasingly recognized as a disorder with metabolic and inflammatory components linked to insulin resistance and impaired glucose metabolism. Incretin-based therapies, particularly GLP-1 receptor agonists, have emerged as potential disease-modifying candidates due to their neuroprotective and metabolic actions.
Aims: To provide a comprehensive and critical synthesis of current evidence on incretin-based therapies with a focus on semaglutide in Alzheimer’s disease, integrating mechanistic, preclinical, and clinical findings, and to identify key research gaps and priorities for future studies.
Methods: A narrative evidence-based review was conducted in accordance with PRISMA principles. Literature was searched in PubMed and Google Scholar for English-language publications from February 2023 to June 2025 using the keywords “Alzheimer’s disease,” “semaglutide,” and “GLP-1 receptor agonists.” A total of 439 records were identified; 40 publications were analyzed, and 16 studies met the inclusion criteria for qualitative synthesis (7 preclinical, 5 clinical, and 4 reviews).
Results: In preclinical models, semaglutide improved cognition, reduced neuroinflammation and oxidative stress, and attenuated β-amyloid and tau pathology. Early clinical and observational studies indicate potential neuroprotective effects and an acceptable safety profile, although data remain heterogeneous and underpowered. Evidence concerning blood–brain barrier (BBB) penetration, dose - response dynamics, and comparative efficacy versus other GLP-1RAs is still incomplete.
Conclusions: Semaglutide represents a biologically plausible and clinically promising candidate for disease modification in Alzheimer’s disease. Confirmation requires long-term, placebo-controlled clinical trials with standardized cognitive and biomarker endpoints, predefined patient stratification by metabolic status and APOE4 genotype, exploration of alternative administration routes, and identification of predictors of therapeutic response.
Keywords: Alzheimer’s disease; semaglutide; GLP-1 receptor agonists; neuroprotection; insulin resistance; cognition; disease modification
Alzheimer’s disease (AD) is the most common cause of dementia and remains one of the greatest challenges for global public health. It accounts for nearly two-thirds of all dementia cases worldwide, and the number of affected individuals is projected to exceed 150 million by 2050 [1].
AD is a progressive neurodegenerative disorder leading to memory loss, cognitive decline, behavioral alterations, and functional dependence. Although aging is the main risk factor, the disease develops through a multifactorial mechanism that involves genetic predispositions (especially the APOE4 allele), vascular dysfunction, and metabolic disorders such as type 2 diabetes mellitus (T2DM). The high prevalence of insulin resistance and impaired glucose utilization in the brains of AD patients has led to the concept of “type 3 diabetes,” highlighting a metabolic dimension in neurodegenerative pathology [2].
This concept has shifted the research focus toward therapeutic strategies capable of modifying both metabolic and neuroinflammatory pathways. Glucagon-like peptide-1 receptor agonists (GLP-1RAs), originally developed for T2DM and obesity, have shown pleiotropic actions on the central nervous system, including improved insulin signaling, mitochondrial protection, and suppression of oxidative and inflammatory damage [3].
Among this class, semaglutide possesses distinct pharmacological advantages. Its prolonged half-life, high receptor affinity, and improved systemic bioavailability ensure stable biological activity and patient adherence. Furthermore, unlike earlier agents such as liraglutide or exenatide, semaglutide has demonstrated more consistent metabolic and cardiovascular benefits, which may indirectly contribute to neuroprotection. Preclinical studies indicate its capacity to reduce amyloid and tau pathology, attenuate microglial activation, and improve synaptic plasticity, suggesting a potential disease-modifying role in AD.
Despite promising results, significant uncertainties remain. Existing studies vary in design, endpoints, and patient characteristics, while comparative data among GLP-1RAs are limited. Mechanistic understanding of semaglutide’s central effects - including its transport across the BBB and dose-response relationship - remains incomplete.
Novelty and relevance of this work lie in its comprehensive synthesis of recent preclinical and clinical evidence (2023–2025) focused specifically on semaglutide, rather than the GLP-1RA class as a whole. The review addresses the emerging paradigm of metabolic-neurodegenerative interaction and identifies critical knowledge gaps that hinder clinical translation - such as heterogeneity of results, insufficient human data, and methodological inconsistency. These aspects underscore the importance of re-evaluating semaglutide not only as a metabolic agent but as a candidate for disease-modifying therapy in Alzheimer’s disease.
The aim of this review is to provide a comprehensive and critical synthesis of current scientific knowledge on the role of incretin-based therapies, with a specific focus on semaglutide, in the prevention and treatment of Alzheimer’s disease. The review seeks to integrate preclinical and clinical evidence, evaluate mechanistic pathways linking metabolic and neurodegenerative processes, and assess the therapeutic potential of semaglutide as a disease-modifying agent.
This review was conducted as a narrative evidence-based review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) principles. The aim was to synthesize existing evidence on the potential role of semaglutide in the prevention and treatment of Alzheimer’s disease (AD), including both preclinical and clinical studies.
Studies were included if they met the following criteria:
Titles and abstracts were screened for relevance, followed by full-text review to assess eligibility based on predefined inclusion criteria. Additional filters included study type (e.g., preclinical, clinical trial, review) and species (human, rodent models, or human-derived organoids).
Study Types
Included studies comprised:
Titles and abstracts were independently screened by two reviewers. Full-text articles were retrieved for studies that met the eligibility criteria or where eligibility was unclear. Disagreements were resolved through discussion or by a third reviewer. Data extraction focused on identifying mechanistic pathways, clinical endpoints, and translational relevance. For each eligible study, the following information was collected using a standardized template: author(s), year of publication, study type, population (human, animal, or cell-based), main findings, and noted limitations. Data were organized in a comparative table (Table 1), which was used to synthesize findings across multiple dimensions: cognitive outcomes, neuropathological markers (e.g., Aβ, p-Tau), inflammatory markers, synaptic changes, and potential clinical relevance. Special attention was given to study limitations, including sample size, treatment duration, lack of placebo control, or translational gaps between preclinical and clinical findings. The analysis was qualitative in nature and did not include a statistical meta-analysis. This method allowed for both qualitative and semi-quantitative synthesis of the current landscape of semaglutide research in AD. Ongoing and registered clinical trials were also documented to capture emerging evidence and identify gaps in the literature.
The PRISMA flow diagram is not presented; no quantitative synthesis or pooled statistical analysis was performed.
Identification: Records identified through database searching (PubMed, Google Scholar): n = 427. Additional records identified through manual search of references and gray literature: n = 12. Total records identified: n = 439.
Screening: Records after duplicates removed: n = 386. Records screened (titles and abstracts): n = 386. Records excluded (irrelevant outcomes, no mention of semaglutide, non-AD focus): n = 302.
Eligibility: Full-text articles assessed for eligibility: n = 84. Full-text articles excluded (not meeting inclusion criteria, insufficient data, language other than English): n = 39.
Included in Review: Studies included in qualitative synthesis (Table 1): n = 16. Preclinical studies: n = 7. Clinical studies (RCTs and cohorts): n = 5. Reviews and systematic reviews: n = 4.
A total of 16 studies were included in the qualitative synthesis (see Table 1). These comprised:
The distribution of study types reflects the early and evolving stage of clinical research on semaglutide in neurodegenerative diseases, emphasizing the need for further well-designed trials to confirm and expand current findings.
Animal models consistently demonstrated that semaglutide improves cognitive function, reduces neuroinflammation and oxidative stress, and modulates key pathological markers of Alzheimer’s disease, including beta-amyloid and tau pathology. For example, Zhao-Jun Wang et al. (2023) and Elbadawy et al. (2025) reported improvements in memory and reductions in tau and neuroinflammation. Meca et al. (2024) highlighted enhanced autophagy and reduced apoptosis, while Boboc et al. (2024) showed improvements in cognition and anxiety, though motor effects were minimal. One study by Germano et al. (2024) observed no significant cognitive or pathological benefit, suggesting that dose, duration, or administration route may critically influence outcomes. Overall, preclinical studies indicate robust neuroprotective effects, but variability in experimental protocols, treatment lengths, and dosages limits direct comparability and translational confidence.
Human trials remain limited and largely exploratory. Retrospective cohort studies (Wang et al., 2024; De Giorgi et al., 2024; Siddeeque et al., 2024) consistently observed a lower incidence of Alzheimer’s disease or cognitive decline among semaglutide users, indicating a potential protective effect. Double-blind trials (Koychev et al., 2024) and large-scale RCTs (Cummings et al., 2025) are ongoing, but results are not yet available, highlighting the need for long-term, adequately powered studies to confirm efficacy.
Systematic and narrative reviews (Au et al., 2025; Tipa et al., 2024; Chen et al., 2024; Wang et al., 2023) integrate preclinical and clinical evidence, consistently reporting trends of cognitive improvement, reduced neuroinflammation, and decreased beta-amyloid and tau pathology. Reviews emphasize that while preclinical evidence is compelling, human data remain sparse and heterogeneous, underscoring unresolved questions regarding dosing, blood–brain barrier penetration, and class-specific versus drug-specific effects among GLP-1 receptor agonists.
Across study types, semaglutide demonstrates strong biological plausibility as a neuroprotective agent. Preclinical models show reproducible improvements in neuropathological and cognitive endpoints, yet translation to humans is limited by small sample sizes, short treatment durations, and heterogeneity in outcome measures. Contradictory findings, such as those reported by Germano et al. (2024), highlight the importance of dose, administration route, and experimental design in influencing results. Retrospective human studies suggest potential disease-modifying effects, but causality cannot be confirmed due to observational designs. Collectively, the data suggest promising trends that warrant further investigation in well-designed, long-term, placebo-controlled clinical trials incorporating standardized cognitive and biomarker endpoints. Table 1 summarizes study designs, populations, interventions, and key findings, providing a structured overview of reproducibility, effect trends, and limitations across preclinical and clinical studies.
Table 1. Research on the Impact of Semaglutide in Alzheimer’s Disease
| Author(s) | Year | Study Type | Population | Sample Size | Treatment Duration | Dosage | Outcome Measures | Key Findings | Limitations |
| Au et al. [22] | 2025 | Systematic review | Humans and animal models | 6-30 | 4-12 weeks (average 8) | 10 – 25 (nmol/kg) | Weight loss, BMI reduction,waist circumference, metabolic parameters, cognitive function (e.g., memory tests), motor function, neuroinflammation markers, amyloid/ tau biomarkers, quality of life | Improves cognition, motor function; reduces neuroinflammation & oxidative stress | Few clinical studies;
long-term safety unknown |
| Zhao-Jun Wang et al. [23] | 2023 | Preclinical study | Animal models | 50 transgenic mice; 50 wild mice | 30 days | Not specified | Behavioral test; brain glucose metabolism; molecular/ biochemical markers (e.g., expression of SIRT1 and GLUT4, Aβ, tau, GOD-POD, HK) | Improves cognition and glucose metabolism | Animal data only; clinical confirmation needed |
| Meca et al. [24] | 2024 |
Review
+ Experimental Data |
Animal models | Not specified | Not specified | Not specified | Cell viability, apoptosis inhibition, autophagy enhancement, SIRT1/GLUT4 pathway activation, amelioration of Aβ and Tau pathology, reduction of neuroinflammation | Reduces apoptosis, inflammation; improves autophagy and neuroprotection | Animal-based; mechanisms unclear |
| Zhang et al. [25] | 2024 | Experimental study | Animal models & human AD organoids | 12 mice/ group; 3 organoids/ group | 6 months (mice); 5 days (organoids) | ~12,16 (0,05 mg/kg/day) | Morris Water Maze, Barnes Maze, Nest building, Active avoidance test, Brain histology and biochemistry, cell viability, morphology/ surface area | Improves memory; reduces Aβ, p-Tau, neuroinflammation | Preclinical; limited translation to humans |
| Wang et al. [26] | 2023 | Review | Humans and animal models | Not specified | Not specified | Not specified | Neuroprotection/ synaptic function, amyloid and Tau pathology, neuroinflammation, neurogenesis/ growth factor | Improves cognition; reduces neuroinflammation & Aβ pathology | Mixed evidence; limited clinical data |
| Wang et al. [27] | 2024 | Retrospective cohort | Humans without AD diagnosis | Not specified | Up to 3 years | Not specified | First-time AD diagnosis; AD-related medications | Lower risk of new AD diagnosis | Short follow-up; safety unknown |
| Germano et al. [28] | 2024 | Experimental study | Animal models | Not specified; Up to 5 mice per cage | 6-7 weeks | 10-25 | Behavior (Open field, Novel Object recognition, Morris Water Maze), histology/ biochemistry (Thioflavin S, IBA1, GFAP), qPCR in hippocampus) | No cognitive or pathology improvement | Animal study; possible insufficient dose/duration |
| Chen et al. [29] | 2024 | Review (preclinical and clinical) | Preclinical models and clinical populations with AD | No data | No data | No data | No data | Reduces neuroinflammation, oxidative stress; improves synaptic function | Mainly preclinical data |
| Cummings et al. [30] | 2025 | Phase 3 RCT | Humans with early-stage AD | 3680 (1840 per trial) | 156 weeks (104-week main phase, 52-week blinded extension, 5-week follow-up washout | 3 mg OD (weeks 0–4) → 7 mg OD (weeks 4–8) → 14 mg OD thereafter | CDR-SB score, ADCS-ADL-MCI score, time to progression to CDR global ≥ 1.0, Plasma: NfL, p-tau181, GFAP, hs-CRP | Ongoing evaluation of semaglutide’s disease-modifying effects | No results yet; long-term
effects unknown |
| Boboc et al. [31] | 2024 | Preclinical study | Animal models | 20 total (5 per group × 4 groups) | 28 days | 0,1 mg/kg | Blood glucose, body weight, behavioral tests, Open Field Test, Novel Object Recognition Test, Social Chamber Test, 0-Maze Test, | Improves cognition and anxiety; no motor effect | Small sample; short treatment |
| Abdulhameed et al. [32] | 2024 | Preclinical study | Animal models | Not specified | 30 minutes | 1×106 cpm/mouse (delivered in per naris, bilaterally) | %Inj/g (percent of injected dose per gram of brain tissue), %Inj/ml (percent of injected dose per of serum), %Precip (percentage of radioactivity precipitated by TCA) | Minimal effect via nasal administration | Animal study; limited administration route |
| Koychev et al. [33] | 2024 | Double-blind placebo RCT | Humans | Up to 88 individuals | 52 weeks | 3mg 14mg (gradually up-titrating over the first 8 weeks) | The 1-year change in tau PET signal; TSPO PET; AD biomarkers measured in blood (at weeks 4, 8, 26, 39, and 52); cognitive assessments (at weeks 26 and 52) | Possible reduction in tau and neuroinflammation; trial ongoing | Small sample size; no results yet |
| De Giorgi et al. [34] | 2024 | Retrospective propensity cohort | Humans | Not specified | Primary analysis 1 year; secondary analysis 2 year | Not specified | Risk of diagnosis of 22 neurological and psychiatric outcome; Risk of 15 Negative Control Outcomes (NCOs); All-cause mortality | Lower risk of cognitive deficits and dementia | Observational design; no direct AD diagnosis |
| Siddeeque et al. [35] | 2024 | Large-scale retrospective cohort | Humans | Not specified; study utilized 152 398 854 patient records from EHR networking | Not specified; data were collected between January 1, 2010, and December 31, 2023 | Not specified | Incidence of 4 major neurodegenerative disorders; All−cause mortality; incidence of 12 other neurodegenerative disorders | Reduced AD risk; suggested neuroprotection | Observational study; causality not confirmed |
| Elbadawy et al. [36] | 2025 | Preclinical study | Animal models | 40 mice | Not specified | Not specified | Hippocampal ACE2/Ang1-7 patway, autophagy /SIRT1/FOX01, microglial polarization | Improved cognition; reduced tau; modulated neuroinflammation | Animal data; short treatment duration |
| Tipa et al. [37] | 2024 | Systematic review | Animal models and cell lines | N/A | N/A | N/A | Cognitive performance, neurobiological markers, metabolic and inflammatory parameters, cell viability and apoptosis, brain structural and functional alterations | Neuroprotective, antiapoptotic; improved cognition and reduced neuronal damage | Few studies; no human data; limited databases |
When comparing across all three categories:
Thus, while preclinical and review data robustly support semaglutide’s potential as a disease-modifying agent, clinical validation remains ongoing. The translational gap between animal and human findings highlights the need for longitudinal RCTs with adequate power and biomarker endpoints to determine whether semaglutide can truly alter the course of Alzheimer’s disease.
In the context of semaglutide research:
Preclinical studies achieve high statistical significance, but clinical relevance remains hypothetical.
Clinical studies so far show potentially meaningful trends, but few reach statistical significance, and none yet demonstrate clinically significant improvement in Alzheimer’s symptoms or progression.
Overall, the preclinical evidence strongly supports semaglutide’s neuroprotective action at a statistical level, suggesting mechanistic plausibility.
The clinical evidence, though promising, is still inconclusive both statistically and clinically.
Review papers unify these findings, highlighting a consistent biological rationale but emphasizing the gap between experimental significance and real-world efficacy [22-37].
This section discusses the mechanistic background and current evidence regarding the neuroprotective potential of semaglutide in Alzheimer’s disease. The discussion integrates preclinical and clinical findings to identify consistent trends, limitations, and directions for further research.
Alzheimer's disease (AD) is a progressive neurodegenerative disorder and the most common cause of dementia, leading to impairments in memory, cognition, behavior, and daily functioning [10,11]. Its etiology is multifactorial, involving beta-amyloid plaques, tau tangles, altered neuronal signaling, and metabolic dysfunction. Chronic neuroinflammation, mediated by microglia and astrocytes, contributes to synaptic dysfunction and neuronal loss. Impaired brain insulin signaling and glucose metabolism exacerbate neurodegenerative processes, while mitochondrial dysfunction increases oxidative stress and further damages neurons. These interconnected mechanisms create a vicious cycle of metabolic and neurodegenerative pathology, providing a rationale for exploring metabolic-targeted therapies in AD [12,13].
Semaglutide, a long-acting synthetic glucagon-like peptide-1 receptor agonist (GLP-1RA), originally approved for type 2 diabetes, modulates multiple pathways relevant to neurodegeneration [17-19]. Activation of GLP-1 receptors in the brain engages intracellular signaling cascades (e.g., PI3K/Akt, MAPK) that influence neuronal survival, synaptic function, and anti-apoptotic mechanisms. Preclinical studies demonstrate that semaglutide reduces microglial activation, supports autophagy, improves mitochondrial function, and mitigates oxidative stress. Its extended half-life allows weekly subcutaneous administration, facilitating adherence in potential long-term trials [20-23].
In preclinical studies, it affects mitochondrial function, autophagy, microglial activation, and oxidative stress. However, translation to human cognitive outcomes remains uncertain, with limited clinical trials confirming direct neuroprotective effects.
While preclinical models consistently show cognitive improvement, reduced amyloid and tau pathology, and lowered neuroinflammation, translation to human studies remains limited. Early clinical trials and observational studies indicate potential neuroprotective effects, but findings are heterogeneous and often underpowered. For example, Germano 2024 reported minimal cognitive benefit despite improvements in metabolic parameters, highlighting possible discordance between mechanistic effects and measurable clinical outcomes. Differences in species, disease stage, dosing regimens, and duration may partly explain these discrepancies.
Contradictory results, including those from Germano 2024 [28], emphasize the need to critically examine study design and endpoints. Short treatment durations, small sample sizes, and reliance on surrogate markers may underestimate the potential cognitive benefits of semaglutide. In addition, the uncertainty surrounding blood–brain barrier (BBB) penetration and central exposure raises questions about dose adequacy in human studies. The lack of standardized cognitive and biomarker endpoints further complicates interpretation.
Current evidence is limited by several factors:
Future investigations should stratify patients by metabolic status and genetic background, including APOE4 carriage and degree of insulin resistance. Optimizing dosing strategies, exploring alternative delivery routes (such as intranasal or nanocarrier-based systems), and developing biomarkers for treatment response are crucial next steps. Furthermore, long-term placebo-controlled trials with standardized cognitive and neuroimaging outcomes are required to validate the therapeutic potential of semaglutide in AD.
Collectively, current data support semaglutide as a biologically plausible and clinically promising candidate for Alzheimer’s disease modification. Nevertheless, the strength of evidence remains preliminary, and well-designed, long-term, placebo-controlled clinical trials are required to determine its therapeutic validity and safety in neurodegenerative populations.
Overall, preclinical evidence supports a biologically plausible neuroprotective role for semaglutide, but current clinical data are insufficient to confirm efficacy in AD. Addressing methodological limitations, standardizing endpoints, and carefully selecting patient populations will be essential to translating preclinical promise into meaningful therapeutic outcomes.
Preclinical studies consistently demonstrate that semaglutide improves cognitive function, reduces neuroinflammation and oxidative stress, and attenuates β-amyloid and tau-related neuropathology. Evidence from early clinical studies suggests potential neuroprotective effects, although data remain limited. Semaglutide’s dual action on metabolic and neurodegenerative processes provides a biologically plausible mechanism supporting its therapeutic potential in Alzheimer’s disease. Its safety profile in type 2 diabetes further supports potential repurposing for neurodegenerative conditions.
Current evidence is preliminary and largely exploratory. Key gaps include limited long-term human data, uncertainty regarding optimal dosing and administration routes, and unclear extent of BBB penetration. It remains unresolved whether the observed benefits are specific to semaglutide or reflect a class effect of GLP-1RAs. Mechanistic understanding of central effects, patient stratification by metabolic status or genetic profile (e.g., APOE4), and predictors of therapeutic response require further investigation.
Future studies should focus on long-term, placebo-controlled clinical trials with standardized cognitive and biomarker endpoints. Research should evaluate different dosing strategies, alternative delivery routes, and comparative efficacy among GLP-1RAs. Patient stratification by metabolic and genetic factors, as well as identification of predictive biomarkers, will be critical to personalizing therapy. Integration of metabolic and neuroprotective outcomes in trial design is recommended to capture the full therapeutic potential of semaglutide in Alzheimer’s disease.
Honorata Derlatka conceptualized the study and designed the methodology. Łukasz Skowron, Honorata Derlatka, Bartłomiej Bobrowski, Jan Drugacz, Olaf Helbig, Klaudia Kontek, Anna Kukhtiak, Julia Marcinkowska and Martyna Radelczuk performed the literature review and data extraction. Łukasz Skowron and Honorata Derlatka drafted the manuscript.
All authors critically revised the manuscript and approved the final version.
AI tools were used to assist with language editing during manuscript preparation. The authors reviewed and approved all content.