GLP-1 and the Degenerating Brain: Exploring Mechanistic Insights and Therapeutic Potential

Understanding the expression and distribution of GLP-1 and its receptor within the CNS is crucial for elucidating their neuroprotective roles and therapeutic potential in neurodegenerative diseases. GLP-1 is synthesized both peripherally and centrally. While enteroendocrine L cells in the gut are the primary peripheral source, GLP-1 is also produced by preproglucagon-expressing neurons located in the nucleus tractus solitarius (NTS) of the brainstem. The central production of GLP-1 allows it to exert direct effects within the brain, bypassing the limitations posed by the blood–brain barrier and rapid degradation of peripheral GLP-1 by dipeptidyl peptidase-4 (DPP-4), which reduces its half-life to less than two minutes [4]. Peripheral GLP-1 influences the brain through two primary pathways: neural and humoral (Figure 1). The neural pathway involves GLP-1 interacting with vagal afferents in the intestinal or hepatic portal area, transmitting signals to the NTS. The humoral pathway, which is a less likely mode of action due to rapid degradation of GLP-1, involves direct entry into the brain via the blood–brain barrier. Consequently, the neural pathway is considered the predominant route for gut-derived GLP-1 to affect central functions [5]. Although GLP-1RAs are frequently described as exerting "central" effects, their pharmacokinetic and BBB profiles differ markedly, shaping the extent of true central exposure. Peptide-based agonists such as exenatide (4.2 kDa) and lixisenatide (4.8 kDa) exhibit measurable but limited BBB permeability, with cerebrospinal fluid (CSF) levels reaching only 0.02% (range: 0.002–0.07%) of plasma concentrations for liraglutide [6], though both liraglutide and lixisenatide demonstrate elevated brain levels 30 min following injection [7]. Liraglutide (3.8 kDa, albumin-bound, t½ ≈ 13 h) shows minimal CSF penetration (6.5 pmol/L CSF vs. 31 nmol/L plasma, representing approximately 0.02%) [6,8], yet a prolonged systemic half-life of 13 h due to albumin binding [9] may sustain indirect neurotrophic and anti-inflammatory signaling. Semaglutide (4.1 kDa, t½ ≈ 7 days [approximately 165–184 h]) likewise demonstrates minimal direct CNS entry but robust peripheral vascular and metabolic modulation, which may confer secondary neuroprotection [10]. Notably, rodents exhibit higher relative CNS uptake than humans due to differences in endothelial transport mechanisms [11], and effective preclinical doses (25–250 nmol/kg) [7] far exceed typical human-equivalent exposures. These interspecies pharmacokinetic disparities must be acknowledged when interpreting "central" mechanisms or extrapolating animal efficacy to clinical settings.

Moreover, GLP-1RAs encompass pharmacologically distinct subclasses rather than a homogeneous entity. Short-acting compounds such as exenatide and lixisenatide yield rapid peak concentrations with transient receptor activation and relatively greater short-term BBB penetration, whereas long-acting analogues, including liraglutide, dulaglutide, and semaglutide, provide sustained systemic exposure but reduced peak brain concentrations due to their albumin binding and extended half-lives [12,13]. Extended-release platforms like PT302, which encapsulates exenatide in biodegradable PLGA microspheres of 20 μm diameter, maintain stable plasma levels from day 10–28 following a single subcutaneous dose and achieve CSF exendin-4 concentrations of 18.3–30 pg/mL compared to <6.9 pg/mL for immediate-release exendin-4 [14] in rodent models. These kinetic distinctions likely underlie differences in preclinical efficacy profiles, where short-acting agents often outperform in acute models, correlating with their superior BBB penetration, while long-acting molecules exert more consistent anti-inflammatory and metabolic benefits in chronic paradigms [10]. Recognizing formulation-specific dynamics is essential for accurate interpretation of mechanistic and translational findings.

Unlike the limited areas that produce GLP-1, GLP-1R is widely distributed throughout the CNS, with expression patterns that correlate with key neurophysiological functions. The receptor's presence in specific brain regions underscores its potential impact on neuroprotection, cognition, appetite regulation and emotional processing. GLP-1R-expressing cells were observed throughout the rostro-caudal extent of the mouse brain [15]. The largest numbers of GLP-1R cells were observed in regions associated with autonomic and behavioural control of energy balance, such as the paraventricular nucleus (PVN), arcuate nucleus (ARC), dorsomedial hypothalamic nucleus (DMH) and central amygdala. A schematic representation of the distribution of GLP-1R neurons throughout the mouse brain is given in Figure 2. The hypothalamus plays a central role in regulating energy balance, appetite, and glucose homeostasis. GLP-1R expression is notably high in hypothalamic nuclei such as the arcuate nucleus, paraventricular nucleus and dorsomedial nucleus [16].

Comparative studies have highlighted both similarities and differences in GLP-1R distribution across species, which have significant implications for translational research. In rodents, GLP-1R expression is prominent in the hypothalamus, aligning with the strong influence of GLP-1 on feeding behaviour in these animals. Non-human primates also show substantial hypothalamic expression but with notable differences in cortical regions. Based on transcriptomic mapping from the Human Protein Atlas (Brain Atlas), GLP-1R mRNA is detected across multiple brain regions including the cerebral cortex, hippocampal formation, amygdala, basal ganglia, thalamus, cerebellum, and white matter at normalized transcripts per million (nTPM) ≥ 1.0. The Human Protein Atlas also indicates that the highest GLP-1R mRNA levels occur in the hypothalamus (nTPM~8.0), followed by the pons and medulla oblongata (nTPM~3.0) [17]. However, recent mapping of the human brain revealed that, while GLP-1R is present in the hippocampus across species, humans exhibit the highest expression in the frontal cortex rather than the hypothalamus [18] (Figure 3). Additionally, GLP-1Rs are absent in the human cerebellum, contrasting with their presence in the cerebellum in animal models. Humans also display GLP-1R expression throughout the cerebral cortex, except for the orbitofrontal cortex, whereas animal models show limited cortical expression. These species-specific differences underscore the importance of cautious interpretation when extrapolating animal data to humans. They highlight the need for human-specific studies to accurately assess the therapeutic potential of GLP-1R agonists in neurodegenerative diseases [18]. Some data indicate that GLP-1R expression in the mouse brain is more regionally concentrated, for example, the mouse brain exhibits region-enriched GLP-1R expression in the hindbrain nuclei and hypothalamus [15,16]. In contrast, the human brain displays a more uniform GLP-1R distribution with low regional specificity and no single dominant site (except the hypothalamus as discussed above, which may draw some similarities to mice) [17]. GLP-1R in mice was observed in regions associated with autonomic and behavioural control of energy balance, such as the paraventricular nucleus (PVN), arcuate nucleus (ARC), dorsomedial hypothalamic nucleus (DMH) (all parts of the hypothalamus) and central amygdala [16].

These interspecies discrepancies extend beyond regional expression to functional receptor pharmacology. For example, rodents display high GLP-1R expression in thyroid C-cells and alveolar tissues, sites with minimal or absent expression in primates and humans [19,20], highlighting why rodent toxicology findings such as C-cell hyperplasia do not translate clinically, as humans and cynomolgus monkeys had low GLP-1 receptor expression in thyroid C-cells, and GLP-1 receptor agonists did not activate adenylate cyclase or generate calcitonin release in primates [19]. Moreover, receptor coupling efficiency and downstream signaling bias differ subtly between species; while specific data on Gαs-biased signaling strength differences between murine and human GLP-1Rs remains limited in the literature, β-arrestin recruitment and receptor desensitization pathways play critical roles in GLP-1R function, with reduced β-arrestin recruitment associated with prolonged cAMP signaling and enhanced metabolic efficacy [21,22], and compounds that reduce β-arrestin recruitment and retain GLP-1R at the plasma membrane produce greater long-term insulin release with faster agonist dissociation rates [23]. Differences in blood–brain barrier permeability, receptor turnover, and neural network integration further limit direct extrapolation from animal data. Consequently, translational studies should prioritize humanized or non-human primate models and early-phase trials employing PET-based receptor occupancy mapping to verify central engagement of GLP-1RAs in humans. While 68Ga-NODAGA-exendin-4 PET showed significant uptake in the pituitary area of obese subjects, no significant uptake was found in other parts of the brain [24], and PET imaging using GLP-1R binding peptides has been employed for monitoring biodistribution and receptor occupancy in various tissues including the pancreas [25].

Activation of GLP-1Rs in these areas suppresses appetite and enhances satiety, contributing to metabolic homeostasis. This high expression suggests that GLP-1Rs in the hypothalamus are a critical mediator of feeding behaviour and energy expenditure [26]. Activation of GLP-1Rs in these areas suppresses appetite and enhances satiety, contributing to metabolic homeostasis. The relatively high expression of GLP-1Rs in the hypothalamus supports their critical role in controlling feeding behaviour. In human studies, using exenatide (GLP-1R agonist) in individuals with type 2 diabetes and obesity has been shown to decrease food intake, underscoring the receptor's importance in appetite regulation [27].

GLP-1Rs are expressed in the hippocampus and various regions of the cerebral cortex, including the frontal, prefrontal and parietal cortices. In the hippocampus, GLP-1R activation enhances synaptic plasticity, learning and memory formation [28]. The frontal cortex, associated with executive functions and decision-making, exhibits the highest levels of GLP-1R expression in humans [18]. GLP-1R activation in the hippocampus enhances synaptic plasticity, learning and memory formation [29]. In the cerebral cortex, GLP-1Rs are expressed in various regions, including the frontal, prefrontal and parietal cortices (Figure 2). The frontal cortex, associated with executive functions and decision-making, exhibits the highest levels of GLP-1R expression in cortical areas of the brain according to some reports [18], but stronger and more comprehensive analyses do not support a unique "highest" region in humans [17]. These findings indicate that GLP-1R signalling in these regions may play a significant role in cognitive processes and could be targeted for treating cognitive deficits in neurodegenerative diseases.

Additional regions expressing GLP-1Rs include the amygdala, the bed nucleus of the stria terminalis, ventral midbrain, periaqueductal grey matter and the lateral septum. In these areas, GLP-1Rs interact with neurotransmitter systems such as dopamine, serotonin and glutamate pathways, influencing mood regulation and reward-related behaviours. For example, GLP-1R activation in the lateral septum modulates dopamine release, affecting reward processing and potentially impacting addictive behaviours [30]. GLP-1R has been detected in nucleus accumbens, ventral tegmental area, amygdala, and prefrontal cortex. In these areas, GLP-1Rs interact with neurotransmitter systems such as dopamine, serotonin and glutamate pathways, influencing mood and emotion regulation, reward-related behaviours and stress responses [31]. In a recent randomized controlled trial, Hendershot et al. (2025) demonstrated that low-dose Semaglutide significantly reduced alcohol consumption and craving in adults with alcohol use disorder [32]. In addition, GLP 1 receptor agonists (GLP 1RAs) influence the dopamine transporter (DAT) function and dopamine release within mesolimbic pathways, which might contribute to behaviours against addiction [33]. These findings have sparked interest in potential neuropsychiatric effects of GLP 1 signalling such as its possible use in alleviating symptoms of depression or anxiety or improving cognitive deficits especially in neurodegenerative disorders. However, these ideas remain largely speculative. Evidence for mood-enhancing or anxiolytic effects of GLP 1RAs in humans is limited and inconclusive. Carefully designed in vivo studies and robust clinical trials are essential to confirm any potential psychiatric benefits.

The regional distribution of GLP-1R in the human brain aligns with areas affected by neurodegenerative diseases. The high expression in the hippocampus and frontal cortex suggests that GLP-1R agonists could have therapeutic effects on cognitive functions impaired in conditions like Alzheimer's disease (AD) [28]. GLP-1R activation may enhance neurogenesis, promote synaptic plasticity and inhibit apoptotic pathways, contributing to neuroprotection [34]. Furthermore, the presence of GLP-1Rs in regions involved in mood and reward processing indicates potential benefits for neuropsychiatric symptoms associated with neurodegenerative diseases. By influencing neurotransmitter systems, GLP-1R agonists might ameliorate depression, anxiety and motivational deficits commonly observed in these conditions [35]. While the widespread distribution of GLP-1Rs offers promising therapeutic avenues, several challenges persist. The differential expression patterns between humans and animal models (Table 1) raise questions about the translatability of preclinical findings. For instance, the lower hypothalamic expression of GLP-1Rs in humans compared to rodents suggests that appetite suppression observed in animal studies may not directly correlate with human outcomes [35]. Understanding why certain brain regions exhibit higher GLP-1R expression is essential for targeted therapy development. Factors such as regional differences in receptor regulation, neuronal subtype specificity and varying roles of GLP-1R signalling pathways could contribute to these patterns. Elucidating these mechanisms may enhance the effectiveness of GLP-1R-based interventions. Limitations in current data include a lack of longitudinal studies assessing the long-term effects of GLP-1R activation in humans and insufficient exploration of how GLP-1R's interactions with other cellular pathways influence neuroprotection. Addressing these gaps through advanced imaging techniques, molecular studies and clinical trials will be critical for translating basic research into effective treatments.

Upon GLP-1 binding, the GLP-1R-coupled Gα subunit is activated by GTP, which then stimulates adenylate cyclase, converting adenosine triphosphate (ATP) into cyclic adenosine monophosphate (cAMP). Increased cAMP levels activate protein kinase A (PKA) through phosphorylation. PKA activation promotes cell growth and survival by enhancing neurotransmitter release and synaptic plasticity [37]. Activation of GLP-1R has been shown to protect cortical neurons from oxidative DNA damage by stimulating cyclic AMP response element-binding protein (CREB), which induces apurinic/apyrimidinic endonuclease 1 (APE1) expression for DNA repair via the base excision repair (BER) pathway. APE1 expression was found to be reduced by phosphatidylinositol-3 kinase (PI3K) inhibition but not by mitogen-activated protein kinase (MEK) suppression, and administration of exendin-4 (a GLP-1 analogue) was observed to enhance DNA repair in ischemic stroke rat brains [38]. Insulin rapidly induces tyrosine phosphorylation and catalytic activity of PI3K. Evidence suggests that PKB/Akt is of major importance in mediating the effects of PI3K in neuronal survival. In vivo, activated PKB/Akt in the hippocampus is associated with neuronal protection against hypoxic stress and nitric oxide toxicity [39]. PI3K activates protein kinase B (Akt) via phosphorylation, supporting synapse formation and protection. Akt also increases the production of myelin proteins, such as myelin protein zero (MPZ) and peripheral myelin protein 22 (PMP22), contributing to remyelination and Schwann cell health [40] (Figure 4).

Exendin-4 (Ex-4), a GLP-1RA, has demonstrated neurotrophic capabilities by enhancing the survival and proliferation of Schwann cells and promoting neurite outgrowth from dorsal root ganglion neurons. In studies using immortalized adult rat Fischer Schwann cells (IFRS1), Ex-4 enhanced cell migration towards neurites and upregulated the expression of MPZ and PMP22, essential components of the myelin sheath, and the effects were mediated through the activation of the PI3K/Akt pathway [41].

Neuroinflammation plays a pivotal role in the progression of demyelinating diseases. GLP-1R activation exerts anti-inflammatory effects that can mitigate neuronal damage. Administration of Ex-4 in animal models reduced demyelination, microglial activation and clinical symptoms of multiple sclerosis (MS) [42]. Ex-4 suppressed mRNA expression of pro-inflammatory cytokines such as interleukin-1β (IL-1β), IL-6, IL-17, and tumor necrosis factor-alpha (TNF-α), which are implicated in excessive innate immune responses and MS progression [42]. Similarly, Dulaglutide, another GLP-1RA, improved clinical symptoms by reducing lymphocyte infiltration and demyelination when administered subcutaneously. It also suppressed encephalitogenic Th1/Th17 cells in the CNS, further attenuating the inflammatory response [43]. These anti-inflammatory actions contribute to preserving myelin integrity and neuronal function. By modulating immune responses, GLP-1R activation offers a multifaceted approach to preventing demyelination and promoting neural health.

Ex-4 administration in spinal cord injury (SCI) models promoted autophagy and suppressed apoptosis, facilitating neuronal survival and regeneration. Higher Ex-4 doses showed stronger neuroprotective effects, evidenced by increased LC3-II and Beclin 1 expression and decreased caspase-3 levels, key markers of autophagic activity [44]. Additionally, a lack of Beclin 1 in cultured neurons and transgenic mice leads to the deposition of Aβ peptides, while its overexpression helps to reduce their accumulation. This highlights the positive role of autophagy in AD by lowering the levels of Aβ, which are a defining feature of the disease's pathology [45]. In neurons, under stress conditions or specific stimuli, Akt signaling can promote autophagy by stimulating the production of phosphatidylinositol-3-phosphate (PtdIns3P), which aids autophagosome formation by recruiting autophagy-related proteins. Beclin 1, as part of the Class III PI3K complex, contributes to PtdIns3P production, further supporting autophagosome formation [46] (Figure 4). Concurrently, Ex-4 reduced apoptosis by decreasing caspase-3 expression, a crucial executor of apoptosis. By balancing autophagy and apoptosis, GLP-1R activation supports the clearance of damaged cellular components while preventing unnecessary cell death.

Microglia are the primary immune cells of the CNS, essential for maintaining homeostasis and responding to injury or pathological stimuli. While transient microglial activation supports tissue repair, chronic activation drives sustained neuroinflammation, contributing to neurodegenerative processes. GLP-1RAs have emerged as powerful modulators of microglial behaviour, shifting them from pro-inflammatory to anti-inflammatory phenotypes. In models of AD, GLP-1RAs such as liraglutide have been shown to significantly reduce activated microglia numbers in the hippocampus and cortex by approximately 50%, accompanied by decreased pro-inflammatory cytokine levels like IL-1β and TNF-α [47]. Furthermore, dual GLP-1/GIP receptor agonists exert superior anti-inflammatory effects by dampening microglial activation and reducing neurotoxic mediators [48]. In addition to attenuating neuroinflammation, GLP-1RAs enhance microglial phagocytic capacity, promoting clearance of amyloid-β and other pathological aggregates, thereby contributing to neuroprotection [49]. These findings underscore the potential of GLP-1RAs as promising therapeutics in neurodegenerative conditions such as AD and PD, where microglia-driven inflammation plays a pivotal pathogenic role [1].

GLP-1 receptor activation has emerged as a powerful neuroprotective mechanism by attenuating oxidative stress and promoting neuronal resilience. Under diabetic and neurodegenerative conditions, excessive reactive oxygen species (ROS) production—partly due to elevated levels of advanced glycation end products (AGEs)—leads to oxidative damage, mitochondrial dysfunction, and neuronal death. GLP-1RAs activate the PI3K/Akt pathway, which in turn stimulates the phosphorylation of the cAMP response element-binding protein (CREB). CREB upregulates the expression of antioxidant enzymes and survival factors, reducing oxidative stress and improving neuronal function [50]. Furthermore, GLP-1RAs downregulate pro-apoptotic proteins such as Bax and cleaved caspase-3 while enhancing anti-apoptotic proteins like Bcl-2 through the combined action of cAMP-PKA, PI3K/Akt, and CREB signaling pathways (as indicated by broken lines in Figure 4), thereby promoting neuronal survival and protecting against apoptosis [51]. This integrated signaling cascade not only protects neurons from apoptosis but also facilitates neuronal differentiation and neurite outgrowth, further contributing to neuronal repair and plasticity [52]. These effects make GLP-1RAs promising therapeutic candidates for neurological diseases characterized by oxidative stress and neuronal apoptosis, such as Alzheimer's and Parkinson's diseases.

AD is the most common neurodegenerative disease, pathologically characterized by amyloid-β (Aβ) plaque deposition and neurofibrillary tangles. A study investigating the relationship between GLP-1 and AD reported that decreased plasma GLP-1 levels were observed in older individuals, and even lower levels were found in patients with AD [53]. In addition, elevated plasma GLP-1 levels were linked to improved cognitive function, as assessed by Mini-Mental State Examination scores, while demonstrating an inverse correlation with plasma pTau181 levels, a biomarker associated with the progression of AD [53]. Mechanistically, GLP-1 has been shown to enhance mitochondrial activity and promote the cleavage of amyloid precursor protein (APP), reducing the production of Aβ [53,54]. Ex-4 attenuates neuroinflammation by shifting microglia from a pro-inflammatory M1 phenotype to an anti-inflammatory M2 state in both in vitro and in vivo settings via restoration of Arf and Rho GAP adapter protein 3 (ARAP3) expression; collectively, this points to the PI3K/ARAP3/RhoA signaling axis as pivotal for Ex-4's anti-inflammatory effects [55]. In an Alzheimer's disease mouse model, liraglutide suppresses microglial activation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome, thereby reducing pro-inflammatory cytokine production and Aβ plaque burden [56]. Semaglutide, a GLP-1 receptor agonist, diminishes seizure severity and cognitive deficits by inhibiting the NLRP3 inflammasome pathway triggered by lipopolysaccharide (LPS) and nigericin, while also lowering lactate dehydrogenase (LDH) release from microglia [57].

PD is a progressive neurodegenerative disorder that causes both motor and non-motor symptoms. Numerous studies have investigated the neuroprotective effects of the newly emerging GLP-1 agonists. A systematic review and meta-analysis illustrated that treatment with GLP-1 agonists in preclinical rodent models of PD showed a substantial reduction in motor symptoms and dopaminergic neurotransmission, suggesting neuroprotective benefits in these models [73]. Another study examined the neuroprotective effects of the GLP-1R agonist liraglutide in an Mitochondrial permeability transition pore (MPTP)-induced PD model, where MPTP exposure led to mitochondrial dysfunction, impaired autophagy, and increased cell apoptosis [74]. Liraglutide counteracted these effects by activating proliferator-activated receptor-γ coactivator 1α (PGC-1α), improving mitochondrial quality control, and reducing neurotoxicity. However, downregulating PGC-1α eliminated the neuroprotective effects of liraglutide, suggesting that PGC-1α plays a key role in regulating mitochondrial biogenesis, function, autophagy, and apoptosis [74]. Furthermore, a study investigated the relationship between insulin-resistant neurons and PD on iPSC (induced pluripotent stem cells) models of synucleinopathy, shedding light on a potential pathway that can be targeted to treat PD [75]. The neurons of the iPSC models showed abnormal insulin response and signalling, which resulted in the blockage of neuroprotective pathways and the activation of stress pathways. As a result, α-synuclein buildup, compromised protein synthesis and neuronal death was observed. However, Exenatide, a GLP-1 RA, reversed the effects of insulin resistant neurons by activating the neuroprotective Protein Kinase B (Akt) signalling, inhibiting MAPK (mitogen-activated protein kinase) pathways involved in stress response and enhancing the activity of mitochondria and lysosomes, in addition to decreasing oxidative stress, α-synuclein aggregates, and inflammatory cytokine IL-6 (interleukin 6) [75]. Another study mentioned that, despite the morphological evidence that identifies the presence of GLP-1 receptors in the substantia nigra pars compacta (SNc), the effect of GLP-1 on the firing patterns of dopamine-producing neurons remained unclear until extracellular in vivo recordings were conducted to show that GLP-1 is involved in controlling the firing activity of neurons. This was evident through the reduced firing rate of dopaminergic neurons located in the SNc and GLP-1's ability to stimulate channels involved in the excitation of neurons [76]. Another study reported that GLP-1 agonists played a significant role in improving motor symptoms associated with PD in phase II clinical trials [67]. In another clinical trial that involved PD patients, Liraglutide treatment resulted in notable enhancements in patients' overall quality of life, daily activities and non-motor symptoms. This was evident through the non-motor symptom scale scores, PD Questionnaire results and Parkinson's Anxiety Scale scores. However, there was no change in cognitive performance during the duration of the clinical trial [77]. Another study demonstrated a clear association between abnormal gut–brain communication and neurodegenerative diseases; PD, in particular, is characterised by an imbalance in gut microbiota and neuroinflammation, further supporting the role of GLP-1 in neuroprotection [78].

Huntington's disease (HD) is a progressive and fatal neurodegenerative disease caused by CAG repeat expansion in the coding region of huntingtin (HTT) protein. Huntington's disease features mutant HTT (mHTT)-driven autophagy defects and oxidative stress, and epidemiology shows higher type-2 diabetes prevalence in HD; in neurons, mHTT overexpression impaired insulin signaling and triggered apoptosis. leading to progressive motor, cognitive, and psychiatric symptoms [79]. Liraglutide (GLP-1 analogue) restored insulin sensitivity, improved cell viability, upregulated antioxidant pathway genes to reduce oxidative stress, and activated AMPK-dependent autophagy to lessen HTT aggregate accumulation—suggesting GLP-1 may mitigate mHTT neurotoxicity [80]. In a 3-nitropropionic acid (3-NP) rat model of HD, liraglutide improved motor function. It boosted protective brain pathways (miR-130a, BDNF/TrkB, PI3K–Akt–CREB, β-catenin), while reducing stress and cell death signals (sortilin, p75NTR, oxidative damage, Bax, caspase-3), suggesting strong antioxidant, anti-apoptotic, and neurotrophic effects [81]. Exendin-4 improved peripheral glucose regulation reduced cellular pathology in brain and pancreas and was associated with fewer mHTT aggregates in islet and brain cells. These effects translated into better motor performance and significantly longer survival in the Huntington's disease mouse model [82]. These findings align with broader reviews showing that incretin hormones, including GLP-1, can cross the blood–brain barrier, regulate mitochondrial function, and mitigate oxidative stress, reinforcing their promise as therapeutic agents in HD and other neurodegenerative disorders.

Substantial preclinical evidence supports the neuroprotective potential of GLP-1RAs in models of neurodegenerative disorders. In cultured SH-SY5Y human neuroblastoma cells, liraglutide pretreatment significantly reduced oxidative stress and promoted neuronal survival and differentiation, largely through activation of the PI3K-Akt or Akt-STAT3 signaling pathway [83,84].

The key challenges in exploring GLP-1RAs for neurodegenerative diseases (AD) begin with findings from animal studies. While some transgenic mouse models have shown encouraging results, including reduced Aβ accumulation and improved cognitive performance, other models have failed to demonstrate these effects [121]. This variability highlights the complexity of neurodegenerative diseases (AD and PD) and suggests that therapeutic responses may be highly dependent on disease stage and genetic background [122]. Translating these findings into human clinical studies brings further difficulties. Although GLP-1RAs are known to cross the blood–brain barrier [123], the efficiency of central delivery and the appropriate therapeutic dose needed to achieve neuroprotection in humans remain unclear [124]. Human trials so far did not significantly alter core AD biomarkers (Aβ and tau) nor show improvements in cognition, although metabolic improvements—such as better glucose control and weight reduction—have been observed [66].

A further challenge lies in the biological plausibility and dose–response relationships observed clinically. While preclinical studies show robust neuroprotective effects via anti-inflammatory, anti-apoptotic, and mitochondrial pathways, the extent to which therapeutic doses of GLP-1RAs achieve sufficient central nervous system (CNS) penetration in humans remains uncertain. Liraglutide and semaglutide are large, hydrophilic peptides (3751.2 Da and 4113.6 Da, respectively) with limited blood–brain barrier permeability, and studies using fluorescently labeled compounds demonstrate that semaglutide does not cross the regular blood–brain barrier but rather accesses discrete brain regions through circumventricular organs and sites adjacent to ventricles [125]. Cerebrospinal fluid (CSF) concentrations of liraglutide are minimal even after months of treatment in humans, suggesting very limited passage across the blood–CSF barrier [7,11]. Without direct evidence of receptor occupancy or signaling in human brain tissue, it is difficult to confirm whether observed clinical benefits arise from central GLP-1R engagement or from systemic metabolic and vascular effects.

Attrition and adherence issues are another recurrent concern. Gastrointestinal side effects, including nausea and weight loss, are highly prevalent with GLP-1RAs. Systematic analyses indicate that 40–70% of patients experience gastrointestinal adverse events, with some studies reporting rates up to 85% [126,127,128,129,130,131,132,133,134]. Nausea affects up to 50% of patients, vomiting and diarrhea are common, and these symptoms lead to treatment discontinuation in up to 12% of patients compared to approximately 2% with placebo [135], an attrition pattern that may disproportionately exclude frail elderly participants. Consequently, trial populations often become progressively younger and healthier over time, introducing selection bias that inflates apparent cognitive benefit. Parallel limitations exist in observational cohorts, where confounding by indication, socioeconomic status, or concomitant medication use (e.g., metformin, SGLT2 inhibitors) may account for part of the observed risk reduction.

Safety considerations are also paramount. Although GLP-1RAs are generally safe and well tolerated in individuals with type 2 diabetes, older adults with AD may be more prone to side effects due to age and comorbidities. Common adverse effects include gastrointestinal issues such as nausea, vomiting, diarrhea, and delayed gastric emptying, which are often dose-related and tend to decrease over time [136]. Long-acting formulations typically lead to fewer nausea and vomiting complaints but are associated with higher rates of diarrhoea. In rare cases, more serious side effects have been reported, including cutaneous reactions and kidney injury [137]. Additional potential adverse events include hepatic, immunologic, endocrine, metabolic, hematologic, neurological, cardiovascular complications and angioedema [136]. Concerns about possible associations with pancreatitis and certain neoplasms have also been raised, although current clinical evidence is inconclusive. Regulatory authorities such as the FDA and the European Medicines Agency (EMA) currently view these risks as manageable with ongoing monitoring [138]. Injection site reactions and antibody formation occur more frequently with exendin-4-based GLP-1RAs [139]. Given that many of these side effects are already common in elderly individuals, cautious use and vigilant monitoring are essential when considering GLP-1RAs for patients with neurodegenerative diseases.

Despite growing enthusiasm surrounding GLP-1RAs as potential neuroprotective agents, the clinical landscape remains fragmented and methodologically heterogeneous. Most published studies to date are exploratory in nature, characterized by modest sample sizes, short follow-up durations, and considerable variation in patient selection criteria. For instance, the ELAD trial enrolled 204 participants with mild Alzheimer's disease, and approximately 102 participants per group completed the 52-week treatment phase, limiting statistical power to detect subtle cognitive changes [114]. The trial failed to meet its primary endpoint of changes in cerebral glucose metabolic rate, though secondary outcomes showed some promising signals including reduced brain volume loss. Similarly, early studies investigating exenatide in Parkinson's disease included 45 patients in an initial open-label pilot trial [140], 62 patients in a Phase 2 trial [141], and 194 participants in the recent Phase 3 Exenatide-PD3 trial, which ultimately found no evidence to support exenatide as a disease-modifying treatment [142]. Such small cohorts raise concerns about both Type I and Type II errors, and the generalizability of results to real-world populations remains uncertain. The divergent endpoints and analytical frameworks across studies further complicate interpretation. Some trials use global cognitive batteries (e.g., ADAS-Cog, CDR-SB), whereas others employ domain-specific or exploratory composite scores, leading to inconsistent effect estimates and difficulties in meta-analysis. Biomarker endpoints also vary, some focusing on amyloid and tau changes, others on FDG-PET metabolic flux or MRI volumetrics [66]. The lack of standardized cognitive and imaging metrics, coupled with differing trial durations (ranging from 6 months to 2 years), undermines cross-trial comparability and reproducibility. Further, many trials employ per-protocol rather than intention-to-treat analyses, thereby excluding participants who discontinue due to adverse events, potentially biasing outcomes toward efficacy.

Another critical limitation lies in population heterogeneity. Most trials preferentially recruit highly selected patient groups, often younger, with fewer comorbidities, and largely of European ancestry, thereby underrepresenting the ethnically and metabolically diverse populations most affected by neurodegenerative disease. Cognitive reserve, glycemic control, vascular burden, and APOE genotype can each modulate disease progression and drug responsiveness; yet few studies are adequately powered to explore these interactions. Moreover, coexisting diabetes, a common inclusion criterion in earlier GLP-1RA studies, may itself confound neuroprotective effects through independent metabolic and vascular improvements. As a result, it remains uncertain whether observed cognitive benefits are driven by direct central GLP-1R activation or by secondary systemic effects such as improved insulin sensitivity and cerebrovascular perfusion.

Finally, the literature is affected by publication bias and incomplete reporting of null findings. Many smaller investigator-led studies are published following positive interim analyses, while negative or inconclusive data are less widely disseminated. These bias skews meta-analyses toward overestimation of effect size and underestimation of variability. Transparency and preregistration of outcomes, particularly for ongoing large-scale efforts such as EVOKE and EVOKE+ (evaluating semaglutide in over 1800 participants with early Alzheimer's disease) [110], will be crucial to resolving these uncertainties. Moving forward, the field must prioritize adequately powered, multicentric, biomarker-driven trials with standardized endpoints, diverse patient recruitment, and open data sharing to establish the true therapeutic potential of GLP-1RAs in neurodegenerative disease.

Advances in GLP-1 biology and neurodegeneration now invite a precision "neuroincretin" approach: tailoring GLP-1RA therapy to individual patient biology. GLP-1RAs exert broad anti-inflammatory, metabolic and neurotrophic effects [147], but the heterogeneity of Alzheimer's, Parkinson's and Huntington's disease demands personalized strategies. Emerging multimodal biomarkers, genetic risk (e.g., APOE4, LRRK2, TREM2), metabolic status (insulin resistance, diabetes), imaging (amyloid/tau PET, MRI), and fluid omics (CSF proteomics, metabolomics), could be integrated into companion-diagnostic frameworks as in oncology. For example, patients with evidence of brain insulin resistance or heightened neuroinflammation might be prioritized for GLP-1RA therapy, whereas others might benefit more from alternative approaches. By aligning GLP-1RA modality and dosing to a patient's molecular profile, a concept we term precision neuroincretin medicine, clinicians could maximize efficacy and minimize off-target effects.

High-throughput omics technologies will drive this precision paradigm by uncovering GLP-1–related signatures. For instance, integrated genomics, transcriptomics, proteomics and metabolomics can identify pathways or biomarkers predictive of GLP-1RA response. Single-cell RNA sequencing in postmortem or induced-pluripotent-stem-cell (iPSC) models of AD, PD and HD can map which neuronal and glial subpopulations express GLP-1R or downstream effectors, and how GLP-1RAs reprogram their states. Recent multi-omic microglia atlases reveal diverse "disease-associated microglia" subtypes (DAM, MGnD, WAM, LDAM) with distinct transcriptional profiles [148,149,150,151]. Future studies could apply single-cell profiling before and after GLP-1RA treatment to see which microglial or astrocyte states are shifted. Likewise, spatial transcriptomics will place these cellular changes in anatomical context, correlating GLP-1RA distribution (by imaging mass cytometry or labeled analogs) with regional gene expression changes. In practice, scRNA-seq and spatial maps from patient brain samples or organoid models could identify GLP-1RA-sensitive pathways (e.g., inflammatory signaling, autophagy genes) as candidate biomarkers. These multimodal data layers should reveal "responder" endophenotypes, turning high-dimensional profiles into actionable patient stratification. Indeed, emerging technologies including single-cell omics and spatial transcriptomics are already being applied to align biomarker profiles with disease states and improve patient stratification. By mining such data, researchers may identify CSF or blood biomarkers (e.g., cytokine panels, metabolite signatures, miRNA changes) that track target engagement and therapeutic effect of GLP-1RAs.

Artificial intelligence will be essential to parse these complex datasets and match patients to therapies. Machine learning classifiers trained on clinical and multi-omic data could predict which individuals are most likely to benefit from GLP-1RA treatment. For example, models could combine polygenic risk scores (APOE, TREM2, HTT repeat length), metabolic measures (HbA1c, insulin levels), and cognitive scores to estimate response probability. Advanced approaches like digital twins, computational models of an individual's physiology, could simulate how a given GLP-1RA regimen alters neuronal and immune signaling. Indeed, a recent review highlights that AI-guided personalization and digital twins are promising strategies in GLP-1 nanomedicine [152]. In practice, an AI system might analyze a patient's MRI/PET scans, fluid biomarkers and genomics to recommend a tailored GLP-1RA (or combination of incretin agonists) and dosing schedule. Early clinical trials could use adaptive designs with real-time data monitoring (wearables, cognitive tests) feeding back into AI algorithms to optimize dose. Overall, integrating AI-driven analytics with biomarker profiling could create "closed-loop" precision therapies: algorithms continually refine patient selection and dosing as new data accrue.

Improving CNS delivery of GLP-1RAs remains a key challenge. Nanoparticle platforms are already being engineered for this purpose. Polymeric nanoparticles functionalized with blood–brain barrier (BBB) targeting ligands (such as Angiopep-2) can encapsulate GLP-1RAs and ferry them across the BBB [153,154,155]. Angiopep-2 is a 19-amino acid peptide that binds the low-density lipoprotein receptor-related protein 1 (LRP1), which is widely expressed on brain endothelial cells and activates receptor-mediated transcytosis across the BBB [156]. Such nanocarriers prolong peptide half-life, enable sustained release, and can be tuned to target receptors on neurons or glia. Future nanotherapeutics might co-deliver GLP-1RA with imaging agents ("theranostics") or sensors, allowing in vivo tracking of biodistribution

Intranasal delivery offers another innovative route. By targeting the olfactory/trigeminal pathways, intranasal administration bypasses systemic degradation and minimizes gastrointestinal side effects. Recent rodent studies found that intranasal GLP-1RAs achieve remarkably high brain concentrations: for example, intranasal dulaglutide rapidly accumulates in the hippocampus and neocortex, regions most vulnerable in Alzheimer's disease, much more efficiently than peripheral delivery. In CD-1 mice and APP/PS1 Alzheimer's models, intranasal dulaglutide, exenatide, and the dual agonist DA4-JC showed widespread brain distribution with highest uptake in hippocampus and neocortex within 30 min [157]. These findings suggest that nose-to-brain administration of GLP-1RAs could greatly enhance CNS penetration and cognitive outcomes, even if some amyloid pathology is present. Novel intranasal formulations (e.g., cell-penetrating peptide-conjugated GLP-1 NPs) have already shown rapid cognitive benefits in animal models within minutes [158]. In the future, biocompatible "exosome-like" nanovesicles may be loaded with GLP-1 analogs for chronic nasal administration. Other cutting-edge delivery approaches are on the horizon. Implanted osmotic pumps or nanofluidic reservoirs could continuously infuse GLP-1RA into cerebrospinal fluid to maintain stable brain levels. Exosome or virus-based gene therapies delivering GLP-1 or its agonists directly to the brain are also conceivable. Ongoing clinical trials (e.g., EVOKE semaglutide in early AD) will inform how much CNS exposure is needed; positive outcomes may spur adoption of these advanced platforms.