Перекресток между диабетом и болезнью Альцгеймера: Молекулярная перспектива.

Abbreviations

angiotensin‐converting enzyme

Alzheimer's disease

protein kinase B

apolipoprotein E

amyloid precursor protein

amyloid‐related imaging abnormalities

amyloid‐related imaging abnormalities–oedema

amyloid‐related imaging abnormalities–haemorrhage

adenosine triphosphate

amyloid‐beta

amyloid‐beta 40

amyloid‐beta 42

β‐site APP cleaving enzyme (1)

blood–brain barrier

brain‐derived neurotrophic factor

complement component 1q

complement component 3

cyclin‐dependent kinase 5

Clinical Dementia Rating–Sum of Boxes

central nervous system

cyclooxygenase‐2

complement receptor 3

damage‐associated molecular patterns

diabetes mellitus

dipeptidyl peptidase‐4

endothelin‐converting enzyme

alpha‐enolase

extracellular signal‐regulated kinase (2)

fructose bisphosphate aldolase

glyceraldehyde 3‐phosphate dehydrogenase

gastrointestinal

glucose‐dependent insulinotropic polypeptide

glucagon‐like peptide‐1

glucagon‐like peptide‐1 receptor agonists

glycogen synthase kinase 3‐beta

4‐hydroxynonenal

insulin‐degrading enzyme

International Diabetes Federation

insulin‐like growth factor‐1/Insulin‐like growth factor‐1 receptor

inhibitor of nuclear factor kappa‐B kinase

interleukin‐1 beta

interleukin‐6

interleukin‐18

inducible nitric oxide synthase

insulin receptor

insulin receptor substrate (1)

c‐Jun N‐terminal kinase

lipopolysaccharide

macrophage‐colony stimulating factor

mitogen‐activated protein kinase

microtubule‐associated proteins

macrophage colony‐stimulating factor

mild cognitive impairment

malondialdehyde

malate dehydrogenase

mammalian target of rapamycin

nuclear factor kappa‐light‐chain‐enhancer of activated B cells

neurofilaments

NOD‐like receptor protein 3

N ‐methyl‐D‐aspartate receptor

O ‐linked β‐ N ‐acetylglucosamine

pathogen‐associated molecular patterns

phosphoinositide 3‐kinase

phosphatidylinositol 4,5‐bisphosphate

phosphatidylinositol 3,4,5‐trisphosphate

protein kinase C

peroxisome proliferator‐activated receptor gamma

presenilin 1

presenilin 2

receptor for advanced glycation end‐products

randomized controlled trial

reactive nitrogen species

reactive oxygen species

ribosomal protein S6 kinase beta‐1

stress‐activated protein kinase

sodium‐glucose cotransporter‐2

sulphonylurea receptor 1

type 1 diabetes mellitus

type 2 diabetes mellitus

toll‐like receptor

tumour necrosis factor alpha

triggering receptor expressed on myeloid cells 2

uridine diphosphate N ‐acetylglucosamine

World Health Organization

Introduction

The global prevalence of type 2 diabetes mellitus (T2DM) and Alzheimer's disease (AD) is increasing significantly in an age‐dependent manner. Epidemiological studies suggest that there may be a strong pathophysiological link between T2DM and dementia. Research findings indicate that impaired insulin signalling, neuroinflammation, oxidative stress and mitochondrial dysfunction may represent shared pathological mechanisms underlying both AD and T2DM (Figure  1 ). Furthermore, AD has been proposed as a metabolic disease caused by insulin resistance in the brain, supporting the hypothesis that AD is ‘type 3 diabetes,’ a term first introduced by Steen et al. [ 1 ] and critically appraised by Kandimalla et al. [ 2 ] In this review article, we discuss the possible pathophysiological mechanisms linking AD and T2DM, as well as antidiabetic agents with therapeutic potential in AD associated with T2DM.

Integrated mechanisms linking type 2 diabetes mellitus (T2DM) to Alzheimer's disease (AD). Hyperglycaemia and systemic metabolic alterations induce brain insulin resistance and neuroinflammation, which promote amyloid‐β (Aβ) accumulation and tau pathology. Aβ and tau, in turn, further aggravate insulin resistance and inflammation, forming self‐amplifying vicious cycles that drive synaptic dysfunction, neuronal loss and cognitive decline. Abbreviations: Aβ, amyloid‐β; AD, Alzheimer's disease; AGE, advanced glycation end product; Akt, protein kinase B; APP, amyloid precursor protein; ATF6, activating transcription factor 6; ATP, adenosine triphosphate; BACE1, β‐site APP‐cleaving enzyme 1; BBB, blood–brain barrier; CDK5, cyclin‐dependent kinase 5; COX‐2, cyclooxygenase‐2; ER, endoplasmic reticulum; ERK, extracellular signal‐regulated kinase; FFA, free fatty acid; GLUT4, glucose transporter type 4; GSK‐3β, glycogen synthase kinase‐3β; IKKβ, IκB kinase β; IL‐1β, interleukin‐1β; IL‐6, interleukin‐6; IL‐33, interleukin‐33; iNOS, inducible nitric oxide synthase; IR, insulin receptor; IRE1α, inositol‐requiring enzyme 1α; IRS, insulin receptor substrate; JNK, c‐Jun N‐terminal kinase; LTP, long‐term potentiation; mPTP, mitochondrial permeability transition pore; mTOR, mammalian target of rapamycin; mtROS, mitochondrial reactive oxygen species; NF‐κB, nuclear factor κB; NLRP3, NOD‐like receptor protein 3; NO, nitric oxide; PERK, protein kinase R–like endoplasmic reticulum kinase; PI3K, phosphoinositide 3‐kinase; RAGE, receptor for advanced glycation end products; RNS, reactive nitrogen species; ROS, reactive oxygen species; T2DM, type 2 diabetes mellitus; TNF‐α, tumour necrosis factor‐α. This figure was created using generative AI tools (ChatGPT, OpenAI; Claude, Anthropic; accessed July 2026) and subsequently edited, corrected and finalized by the authors, who take full responsibility for its content.

Neurodegeneration
emerges with the loss of neuronal function caused by various factors, such as changes in the distribution, pattern and number of specific types of neurons, eventually leading to irreversible central nervous system (CNS) dysfunction [ 3 , 4 , 5 ].

Neurodegenerative disorders
are generally characterized by extrapyramidal and pyramidal movement disorders, as well as cognitive and behavioural disorders [ 6 , 7 ]. Accumulation of specific proteins and anatomic vulnerability contribute to the pathogenesis of neurodegenerative disease. Moreover, other mechanisms are associated with progressive neuronal dysfunction, neuronal cell death, proteasomal and autophagosomal/lysosomal systems, oxidative stress and neuroinflammation [ 6 , 7 ]. AD, the leading cause of dementia worldwide, is an increasingly prevalent neurodegenerative disorder. Emerging evidence indicates that insulin resistance and metabolic dysfunction may play important roles in AD pathogenesis, highlighting a potential mechanistic link between AD and T2DM [ 3 , 5 , 6 , 7 , 8 ].

Dementia
is a progressive cognitive impairment that affects daily living activities and can cause dependency, disability and mortality. According to WHO estimates, approximately 57 million people were living with dementia in 2021, with more than 60% residing in low‐ and middle‐income countries. Each year, nearly 10 million new cases are reported worldwide [ 9 ]. There are several causes of dementia, of which AD is the main one. Approximately 50%–70% of dementia occurs as a result of AD, and because it is related to the normal brain aging process, its prevalence doubles every 5 years over age 65 [ 10 , 11 , 12 ].

AD
is one of the most common CNS disorders, and it is a chronic, progressive neurodegenerative condition that leads to impairment of daily functioning [ 13 ]. AD causes brain atrophy, with loss of brain mass and neuronal connections [ 14 ].

ad may be familial (inherited), manifesting at an early age, or sporadic, manifesting at an older age, with progression influenced by health conditions, environmental factors and lifestyle [ 4 , 5 , 12 , 15 , 16 ]. In addition to environmental factors, several genetic factors contribute to sporadic AD, most notably the apolipoprotein E (apoE) gene, which plays a crucial role in its development [ 12 , 17 ]. On the other hand, mutations in the amyloid precursor protein (APP), presenilin 1 (PSEN‐1) and presenilin 2 (PSEN‐2) genes cause familial AD, but this form is rare [ 12 , 15 ]. These presenilin mutations primarily enhance amyloidogenic processing and Aβ42 accumulation through altered γ‐secretase activity, rather than acting through the metabolic disturbances that characterize T2DM‐associated (sporadic) AD [
18 ]; nevertheless, emerging evidence indicates that some presenilin mutations may also influence neuronal insulin signalling [ 19 ]. While inherited AD accounts for at most 5% of patients, 90% have the sporadic form [ 5 , 20 ].

Pathophysiology

The most widely accepted cause for AD is the extracellular deposition of Aβ, among other reasons like the formation of intracellular neurofibrillary tangles by tau, mutations in the apoE gene, neuronal damage caused by oxidative stress, and many other inflammatory conditions [ 21 , 22 , 23 ].

Although the pathology of AD remains unclear, there are some hypotheses to elucidate the mechanism of this condition.

Aβ Hypothesis in AD

Amyloid
is a type of protein with a β‐sheet‐rich structure [ 7 , 24 ]. Aβ is normally present in the brain in specific physiological amounts and with unknown function [ 25 , 26 ]. In 2010, Morley et al. performed a study to discover its function and revealed a significant physiological role for Aβ. According to their research, the physiological amount of this peptide increases acetylcholine release in the hippocampus and helps improve memory and learning, which is paradoxical in the context of AD [ 25 , 26 ].

Contrary to these beneficial functions, aggregation of amyloid‐like filaments in the cytoplasm of neurons and glia or extracellular deposition of amyloids in the parenchyma of the brain or the blood vessel as a plaque can be seen in neurodegenerative diseases [ 7 , 27 ]. Aβ, a hallmark of neurodegenerative diseases especially AD, is the proteolytic product of a transmembrane protein called amyloid precursor protein (APP) [ 7 , 28 ].

The amyloid hypothesis
claims that in physiological conditions, APP, which is found in the neuronal and glial cells of the brain, is cleaved by α‐secretase and γ‐secretase and extracellular nontoxic soluble fragments and intracellular domains of APP are formed [ 14 , 29 , 30 , 31 ]. This soluble fragment has many roles such as supporting synaptic plasticity, regulating neuron excitability and protecting neurons against oxidative and metabolic stress [ 14 , 31 ].

In the nonamyloidogenic pathway, APP is first cleaved by α‐secretase within the Aβ domain, thereby preventing Aβ formation, and the resulting fragment is subsequently processed by γ‐secretase to generate nontoxic soluble peptides. These soluble products support synaptic plasticity, regulate neuronal excitability and protect neurons against oxidative and metabolic stress [ 14 , 29 , 30 , 31 ].

In the amyloidogenic pathway, APP is first cleaved by β‐secretase (BACE), producing a membrane‐bound C99 fragment, which is subsequently cleaved by γ ‐secretase to generate Aβ peptides, particularly Aβ40 and Aβ42. Among these isoforms, Aβ42 is more prone to aggregation and is considered more pathogenic because of its hydrophobic properties [ 14 , 31 ]. Following their generation, Aβ monomers aggregate sequentially into soluble oligomers, protofibrils and insoluble fibrils, which ultimately accumulate as extracellular amyloid plaques. Among these Aβ assemblies, soluble oligomers are considered the most synaptotoxic species. These soluble oligomers disrupt synaptic transmission, impair long‐term potentiation, alter calcium homeostasis and interfere with neuronal insulin signalling pathways, thereby contributing to synaptic dysfunction and early cognitive decline in AD [ 14 , 31 ]. While soluble oligomers are primarily associated with early neuronal dysfunction and memory impairment, fibrillar Aβ deposits and plaques contribute to chronic neuroinflammation by activating microglia and astrocytes, leading to oxidative stress and neuronal death [ 14 , 31 ]. Therefore, different forms of Aβ contribute to distinct stages of AD progression, with soluble oligomers mediating early synaptic toxicity and plaque deposition promoting later neurodegenerative changes [ 14 , 31 ].

Recent clinical trials of monoclonal antibodies targeting Aβ have demonstrated statistically significant but clinically modest reductions in cognitive decline in early‐stage AD. Lecanemab slowed decline by 27% on CDR‐SB over 18 months [ 32 ], and donanemab by 35% over 76 weeks [ 33 ]. However, aducanumab failed to show consistent clinical benefit across its two Phase 3 trials [ 34 ], leading to its discontinuation in 2024 [ 35 ]. Importantly, all anti‐Aβ antibodies carry a significant risk of amyloid‐related imaging abnormalities (ARIA), including cerebral oedema (ARIA‐E) and microhemorrhages (ARIA‐H), which require careful clinical monitoring [ 32 , 33 ].

In familial AD cases, genetic mutations of amyloid precursor protein (APP) gene result in the overproduction of Aβ, especially the 42 amino acid Aβ isoform (Aβ 42 ), which has amyloidogenic properties [ 29 ]. Thus, an increased Aβ 42 /Aβ 40 ratio is seen in the plasma of patients with familial AD caused by an APP mutation [ 36 , 37 ]. In addition to APP mutation cases, PSEN‐1 and PSEN‐2 mutations may cause an increase in the Aβ 42 /Aβ 40 ratio [ 38 ].

Tau

Tau protein is one of the main microtubule‐associated proteins (MAPs) in the CNS. These proteins interact with tubulin and promote microtubule polymerization and stabilization in cells [ 7 , 14 , 39 ]. The function of tau protein changes with the degree of phosphorylation [ 14 ].

Tauopathies
are disorders characterized by the accumulation of hyperphosphorylated tau proteins that bind to one another and form helical and straight filaments [ 14 , 40 ]. These filaments form insoluble neurofibrillary tangles, which cause destabilization of microtubules in cognitively important brain regions [ 7 , 14 , 41 ]. Moreover, Aβ plaque deposition can further promote tau hyperphosphorylation, thereby contributing to neuroinflammation [ 14 ]. According to these two hypotheses, Aβ peptide and tau act synergistically and cause AD by increasing each other's toxic effects [ 42 ]. According to these hypotheses, the accumulation of Aβ peptide and tau provides neuronal loss and progression of AD by inducing apoptosis [ 43 , 44 ].

Apolipoprotein E

apoE is a type of protein produced mainly in the liver and is responsible for lipid transportation and metabolism [ 21 , 45 ]. The commonly known isoforms of apoE with only one or two different amino acids are apoE2, apoE3 and apoE4 [ 21 ]. Amino acid substitutions affect the overall charge and structure of apoE, thus altering binding to both lipoprotein particles and cellular receptors [ 46 ]. ApoE is highly expressed in the brain, primarily produced by astrocytes and microglia [ 46 ].

The polymorphism of apoE (on chromosome 19) is one of the major risk factors for severe senile AD [ 21 , 45 , 47 ]. Particularly, patients with apoE4 are at high risk compared to noncarriers of apoE4 [ 21 , 48 , 49 ]. Conversely, patients with apoE2 are likely to be protected against AD [ 21 ]. A study on patients with an APP mutation and apoE2 revealed that these patients did not develop dementia even though they had mutated APP [ 21 ]. In addition, another study showed that apoE4 has more affinity for Aβ than apoE3, suggesting that apoE4 might be acting as a pathogenic chaperone for Aβ [ 21 ]. Moreover, in the presence of apoE4, the aggregation of Aβ is much higher than in the presence of apoE3 [ 21 ].

In addition to these three crucial hypotheses, some other potentially modifiable risk factors for AD have been identified, including diabetes mellitus, midlife hypertension, midlife obesity, physical and mental inactivity, depression, smoking and low education level [ 50 , 51 ].

Genetic Risk Factors

Inheritance is especially important in young patients with AD. AD could manifest before the age of 65 if there is an autosomal dominant pattern in the family [ 26 ]. The mutations mostly occur in amyloid precursor protein (APP) (on chromosome 21q21), PSEN‐1 (on chromosome 14q24), PSEN‐2 (on chromosome 1q24) and apoE4 [ 5 , 21 ]. Furthermore, their impact on AD progression is mostly related to their ability to augment Aβ aggregation [ 5 , 21 ].

APP

The APP is produced in different cells through secretory and endosomal‐lysosomal pathways [ 21 ]. Even though its real role is unclear, there is some evidence that it is likely to be involved in cell‐substrate adhesion, synaptic plasticity, neuronal cell survival and cell movement regulation [ 21 ]. In addition to APP's physiological roles, it is known to cause inherited AD if it undergoes some specific mutations at certain sites, such as those close to the γ‐secretase cleavage site [ 21 ].

Presenilins

Presenilin 1 and presenilin 2 are catalytic components of the γ‐secretase complex; mutations in these genes increase Aβ42 production and are the most common cause of early‐onset familial AD [ 21 , 52 ].

Environmental Factors

Environmental exposures, including heavy metals, pesticides, head trauma and metabolic disturbances such as hypercholesterolemia, have also been identified as modifiable risk factors for AD [ 5 , 21 ]. These factors may contribute to chronic metabolic dysfunction and insulin resistance, which are increasingly recognized as important mechanisms linking environmental risk factors to AD pathogenesis. Impaired insulin signalling promotes neuroinflammation, oxidative stress, and Aβ and tau pathology, thereby increasing the risk of cognitive decline and AD [ 53 ].

Age

Except for inherited types of AD, aging is one of the most powerful factors in the development of sporadic AD, because the capacity for self‐repair declines [ 5 ]. At the cellular level, aging is associated with elevated oxidative stress (caused by reduced activity of antioxidants like glutathione reductase), attenuated mitochondrial function, inadequacy in DNA repair and low tissue regeneration ability [ 5 ].

Diabetes Mellitus

Diabetes mellitus (DM) was described by the World Health Organization (WHO) in 2014 as a chronic and complex metabolic disorder characterized by elevated blood glucose levels [ 54 , 55 ]. Several symptoms can be seen, such as polyuria, polydipsia and polyphagia; in severe cases, diabetic ketoacidosis may also develop [ 56 , 57 ]. Furthermore, some complications can occur at the late stage of diabetes, such as nephropathy, retinopathy, neuropathy, diabetic heart diseases, as well as structural and functional alterations in the CNS [ 55 , 57 , 58 , 59 ].

There are mainly two types of diabetes: type 1 and T2DM [ 57 , 60 ].

Type 1 diabetes mellitus (T1DM) is also called ‘insulin‐dependent diabetes’ or ‘juvenile diabetes.’ T1DM is mostly defined as impaired insulin production by pancreatic β cells, and it constitutes 5%–10% of diabetes cases, with increasing prevalence [ 54 , 55 , 57 , 60 , 61 ]. According to previous studies performed on patients with T1DM, there is a nonnegligible risk of developing cognitive problems, such as impaired information processing and psychomotor retardation [ 55 , 62 ].

T2DM is defined as ‘noninsulin‐dependent diabetes’ or ‘adult‐onset diabetes.’ This type of diabetes occurs in patients with insulin resistance [ 54 , 55 , 57 , 60 , 61 ]. In patients with T2DM, normal or raised insulin levels and higher blood glucose levels are observed [ 60 ]. Although the function of β cells is generally normal, insulin levels are not enough to compensate for insulin resistance [ 60 ]. Risk factors for T2DM mainly include obesity, aging and insufficient physical activity [ 55 , 57 , 60 ]. According to the International Diabetes Federation (IDF) Diabetes Atlas (10th Edition), an estimated 537 million adults aged 20–79 years were living with diabetes worldwide in 2021, corresponding to a global prevalence of 10.5%; this number is projected to rise to 643 million by 2030 and 783 million by 2045. Approximately 90% of these cases are type 2 diabetes [ 57 , 63 ]. The consumption of a high‐sugar, high‐calorie diet, high blood pressure, deficient physical activity, genetic susceptibility and lifestyle can be major risk factors for diabetes [ 57 ].

Moreover, the clinical picture in patients with T2DM becomes increasingly complex with age. As people age, they become more vulnerable to chronic and complex conditions, such as neurodegenerative diseases (including AD) and metabolic disorders (including T2DM) [ 55 , 64 ]. According to recent studies, the interrelated mechanisms between AD and diabetes include hyperglycaemia, vascular abnormalities, altered amyloidosis and impaired insulin signalling [ 55 ]. Although these interrelated mechanisms exist, the relationship between AD and T2DM remains unclear.

Possible Interrelated Mechanisms Between AD and Diabetes

In 1999, the Rotterdam Study was the first population‐based study to reveal an association between T2DM and an elevated risk of dementia, followed by many others worldwide [ 65 , 66 , 67 , 68 ].

There are several suggested mechanisms, such as vascular factors, abnormalities in insulin signalling and inflammation, that may explain how T2DM alters the risk of dementia.

Glucose Metabolism

It is a well‐known fact that diabetes is related to the high glucose level in the periphery, but it is not clear how this increment has an impact on the brain [ 55 , 69 ]. One hypothesis linking glucose level and cognitive disorders is related to the brain hypometabolism and its relationship to AD [ 55 ]. The brain is known to be the most active organ of the body, so its need for energy (glucose) is continuous [ 70 ]. When the brain encounters an interruption in glucose supply, such as a hypometabolism condition, it undergoes metabolic stress and could cause some significant neuronal damage, including in AD‐related brain regions [ 55 , 70 ]. This suggests that type 2 diabetes could be a significant contributor to AD, and hypometabolism should be considered a risk factor [ 55 , 70 ].

Insulin Signalling

Insulin resistance has been suggested as a potential factor associated with the cognitive dysfunction observed in both AD and T2DM [ 55 , 71 ]. Because brain glucose level is mostly regulated independently of insulin, the brain is considered an insulin‐insensitive organ [ 55 , 72 ]. Insulin is not only a hormone that regulates blood glucose levels, but is also crucial for neuronal growth, neuronal repair and regulation of brain functions [ 73 , 74 ]. Moreover, it promotes neuronal plasticity, dendritic spine formation, synapse formation, neuronal stem cell formation and neuroprotection, and it supports cognitive functions such as learning and memory [ 73 , 75 ]. Insulin can cross the blood–brain barrier (BBB) through a receptor‐mediated process that can be saturated [ 55 , 76 ]. Insulin receptors (IRs) are expressed in neurons and glial cells; the majority are found in the hypothalamus, olfactory bulb and AD‐related regions such as the hippocampus and cerebral cortex [ 73 , 77 , 78 , 79 ]. Furthermore, these receptors and insulin‐like growth factor‐1 receptors (IGF‐1Rs) are mostly found in neurons but less commonly in glial cells [ 73 ].

Under physiological conditions, activation of insulin receptor substrate (IRS) occurs when insulin binds to IR at the cell surface and this binding activates several downstream effectors like phosphoinositide 3‐kinase (PI3K) and Akt, which inhibit some downstream molecules via negative feedback mechanisms such as glycogen synthase kinase 3β (GSK‐3β), mammalian target of rapamycin (mTOR)/S6 kinase 1 (S6K1) and protein kinase C (PKC) molecules to modulate cellular signalling cascades [ 73 ]. PI3K promotes the conversion of phosphatidylinositol 4,5‐bisphosphate (PIP2) to phosphatidylinositol 3,4,5‐trisphosphate (PIP3) [ 2 , 80 ]. PI3K promotes the recruitment of protein kinase B (PKB), which is also known as Akt, to the plasma membrane, where it is then phosphorylated and activated by specific protein kinases [ 2 , 81 ]. In addition, PKB acts on several downstream targets, including the GSK‐3β phosphorylation pathway, which links IR signalling at the cell surface to the enzymes of cellular glycogen metabolism [ 2 , 73 ].

In addition to this negative feedback mechanism, this cascade is also regulated via feed‐forward inhibition by ERK 2 (extracellular signal‐regulated kinase 2), IKK (inhibitor kappa B kinase) and JNK (c‐Jun N‐terminal kinase) [ 73 ]. These kinases phosphorylate IR substrate‐1 (IRS‐1) on serine residues and prevent further transmission of IR‐mediated downstream signalling [ 73 ].

In recent studies, a reduction in the level of insulin and gene expression of IRs in patients with AD was observed [ 73 , 82 , 83 ]. Unfortunately, this is not a precise indicator of the relationship between AD and insulin because the insulin pathway already changes with ageing [ 55 , 84 ].

Impaired CNS Insulin Signalling and Aβ

Recent studies revealed a relation between impaired insulin signalling and two major components of AD [ 55 , 73 , 85 , 86 ].

Insulin resistance may promote Aβ deposition, which in turn drives neurodegeneration through synaptotoxicity and other downstream actions, and may therefore serve as a valuable indicator of AD pathology [ 73 , 87 ]. While insulin resistance promotes Aβ deposition, Aβ impairs insulin signalling by reducing IR autophosphorylation, which prevents proper activation of downstream pathways such as IRS‐PI3K‐Akt [ 73 , 88 ]. In addition, Aβ oligomers have been shown to significantly reduce IR levels and IR activity at the cell surface of dendrites of hippocampal neurons [ 73 , 89 ]. Aβ oligomers also act on downstream signalling from IRS‐1 and PI3K, where they can induce IRS‐1 serine phosphorylation and promote tau phosphorylation and inflammatory processes [ 73 , 90 ]. Clearance and proteolytic degradation of Aβ are conducted by endothelin‐converting enzyme (ECE 1 and ECE 2) and angiotensin‐converting enzyme (ACE), but mainly by the metallopeptidase insulin‐degrading enzyme (IDE) [ 73 , 91 ]. IDE levels tend to decrease in AD patients, which in turn leads to hyperinsulinemia that causes insulin resistance and contributes to T2DM pathology [ 55 , 73 , 85 , 86 , 92 ]. In addition to IR, there is another receptor that connects AD and impaired insulin signalling called N‐Methyl‐D‐aspartate receptors (NMDARs) [ 73 , 93 ]. NMDARs have a crucial role in glutamatergic neurotransmission in postsynaptic neurons, and it allows depolarization of recipient neurons and the generation of action potential in the neurotransmission process [ 73 ]. However, excessive activation of NMDARs can cause synaptic dysfunction and neuronal injury, thereby promoting the progression of AD and other cognitive disorders [ 73 , 93 ]. Increasing evidence indicates that IR signalling and NMDAR signalling are functionally interconnected and this crosstalk is essential for maintaining normal synaptic activity and cognitive function [ 73 ]. Soluble Aβ oligomers aberrantly activate.

NMDARs lead to excessive intracellular Ca +2 influx and activation of stress‐response kinases, including JNK, IKK and ERK. Activation of these kinases induces serine phosphorylation of IRS‐1, thereby impairing its interaction with the IR and attenuating downstream PI3K/Akt signalling [ 94 ]. Consequently, neuronal insulin signalling is disrupted, contributing to impaired synaptic plasticity, memory deficits and the development of brain insulin resistance. Supporting this mechanism, the NMDAR antagonist memantine prevents Aβ oligomer‐induced inhibition of IR signalling and reduces excitotoxicity, intracellular calcium overload and oxidative stress, thereby helping preserve neuronal function [ 73 , 89 , 95 , 96 ].

Impaired CNS Insulin Signalling and Tau Phosphorylation

Insufficient insulin signalling can contribute to tau hyperphosphorylation through several converging routes. Under physiological conditions, insulin‐stimulated PI3K activity generates phosphatidylinositol 3,4,5‐trisphosphate (PIP3), which recruits Akt to the plasma membrane and promotes its phosphorylation and activation; activated Akt then phosphorylates and inhibits glycogen synthase kinase 3β (GSK‐3β), a major tau kinase [ 2 , 73 , 97 , 98 , 99 ]. In states of brain insulin resistance, reduced Akt activity relieves this inhibition, leading to GSK‐3β hyperactivity and consequent tau hyperphosphorylation, a pathological hallmark of AD [ 73 , 98 ]. In parallel, cyclin‐dependent kinase 5 (CDK5), another major proline‐directed tau kinase, undergoes aberrant activation in ad : calcium‐dependent activation of calpain cleaves the physiological CDK5 regulator p35 into the more stable, mislocalized p25 fragment and the resulting CDK5/p25 complex hyperphosphorylates tau at AD‐relevant Ser/Thr‐Pro sites [ 100 , 101 ]. Because GSK‐3β and CDK5 act on partially overlapping tau residues, dysregulation of either kinase amplifies tau pathology. Hyperglycaemia and impaired glucose metabolism can further aggravate this process by promoting nonenzymatic glycation of tau and Aβ, which facilitates their aggregation [ 55 , 102 ].

Recent evidence suggests that the relationship between insulin resistance and tau pathology is bidirectional [ 103 , 104 ]. While impaired insulin signalling promotes tau hyperphosphorylation, hyperphosphorylated tau may also contribute to impaired neuronal insulin signalling. Although the underlying mechanisms are not yet fully understood, hyperphosphorylated tau has been associated with defects in IR signalling and downstream pathways [ 104 ]. Therefore, insulin resistance and tau hyperphosphorylation are likely to reinforce each other, forming a vicious cycle that accelerates synaptic dysfunction, neurodegeneration and cognitive decline in AD [ 103 ].

Inflammation

Inflammation is a biological process observed in both AD and T2DM, and generally reflects cellular and tissue responses to stressors such as trauma, injury or infection [ 105 , 106 ]. Inflammation occurs as a consequence of pathological conditions such as cancer, diabetes and AD, as well as general aging [ 107 , 108 ].

Because the brain is an extremely sensitive organ, inflammation may cause both acute and chronic brain disorders [ 106 ]. To resist inflammation, neurons mount neuroprotective responses by regulating the secretion of neurotrophic factors, proteases, cytokines and the clearance of cellular debris [ 106 , 109 ]. In summary, the progression of inflammation depends on complex local immune responses, which can influence the neuronal microenvironment [ 106 , 109 ]. Inflammation may be protective or detrimental in the context of brain‐associated injuries [ 106 , 110 ].

Neuroinflammation

Neuroinflammation is a crucial factor in brain‐related diseases and is mounted as a response by all cell types of the CNS, such as neurons, macroglia and microglia [ 105 , 106 , 111 ]. These pathological neuroinflammatory pathways are initiated by environmental factors, age, genetics and health status through microglial activation [ 105 ]. The toll‐like receptor (TLR) signalling pathway is one example of a neuroinflammatory pathway [ 112 ]. Lipopolysaccharide (LPS), which induces a systemic inflammatory response via TLR signalling, is an endotoxin located on the outer membrane of Gram‐negative bacteria [ 112 ]. It binds to the TLR on the surface of microglia and activates diverse signal transduction pathways, such as the phosphoinositide 3‐kinase/protein kinase B (PI3K/Akt), mitogen‐activated protein kinase (MAPK) and mammalian target of rapamycin (mTOR) pathways, all of which activate nuclear factor kappa B (NF‐κB) [ 113 , 114 ]. NF‐κB activation leads to the production of inflammatory cytokines, inducible enzymes (inducible nitric oxide synthase [iNOS] and cyclooxygenase‐2 [COX‐2]) and chemokines, which constitute the main mechanism of neuroinflammation [ 113 , 114 ].

Neuroinflammation associated with T2DM has been proposed as a potential contributor to AD pathogenesis, possibly through its impact on metabolic regulation [ 73 , 115 ]. Recent studies revealed that hyperinsulinemia increases inflammatory responses in the CNS, and the degree of chronic inflammation in AD correlates with cognitive decline and brain atrophy [ 73 , 92 , 116 , 117 , 118 ]. Patients with high insulin levels in the periphery tend to have higher levels of interleukin‐1β (IL‐1β), interleukin‐6 (IL‐6) and tumour necrosis factor α (TNF‐α) in cerebral tissues, similar to patients with AD, who have higher levels of these cytokines in areas close to Aβ plaques [ 73 , 92 , 117 , 119 , 120 , 121 , 122 ]. Cytokines such as TNF‐α, IL‐1β and IL‐6 can cross the BBB [ 73 , 116 ]. Furthermore, increased production of these inflammatory cytokines, together with peripheral insulin resistance driven by Aβ‐activated microglia, may block intracellular insulin actions through IRS‐1 serine phosphorylation via c‐Jun N‐terminal kinase (JNK), IκB kinase (IKK) and extracellular signal‐regulated kinase 2 (ERK2) [ 73 , 117 , 118 , 119 , 123 ].

The receptor for advanced glycation end products (RAGE), which is expressed in various brain cell types, such as endothelial cells, neurons, microglia and astrocytes, appears to be another player in vascular inflammation. RAGE mediates the transport of Aβ from the periphery to the brain across the BBB and causes synaptotoxicity, neurovascular stress and cerebrovascular dysfunction, increasing TNF‐α and IL‐6 levels [ 120 , 124 ]. Aβ binds to RAGE and activates the NF‐κB pathway, which induces the expression of TNF‐α, IL‐1β, IL‐6 and macrophage colony‐stimulating factor (M‐CSF) in neurons [ 120 ]. M‐CSF activates microglia and induces BACE1 (β‐site APP‐cleaving enzyme 1) expression and, in turn, the amyloidogenic process, contributing to the deposition of Aβ [ 125 ]. BACE1 contains an NF‐κB binding site, and RAGE‐ligand interaction promotes activation of this site, thereby increasing RAGE expression [ 125 , 126 ].

Chronic Inflammation and the Brain

Constant exposure to inflammation leads to a chronic inflammatory response, which is detrimental for survival and neuronal health in the brain [ 74 , 105 ]. This potent chronic inflammatory response is involved in the pathology of neurodegenerative diseases, especially in AD and causes the activation of brain microglia, promotes the release of pro‐inflammatory cytokines (such as growth factor signalling inhibitors, nitric oxide and oxidative stressors) and increases reactive oxygen species [ 74 , 105 , 127 , 128 ].

One of the major links between AD development and chronic inflammation is the role of microglia. Microglial cells produce ApoE, a lipid‐transport protein that can influence inflammatory signalling pathways [ 74 , 128 , 129 ]. Another feature of ApoE is its ability to block receptors with roles similar to IRs, such as in glucose regulation, energy usage, synaptic plasticity and protection against inflammation [ 74 , 130 ]. In particular, the ApoE4 allele has a higher affinity for IR‐like receptors and even the IR itself than other apoE isoforms, which is associated with an increased risk of AD [ 74 , 128 , 131 ]. It may also elevate TNF‐α plasma levels and reduce the activity of Akt/PKB [ 132 ]. All these findings support the link between AD and the decreased growth factor activity and an increased inflammatory response. Another important factor that accelerates the pathology of AD is the TREM2 gene, which encodes an immune‐related phagocytic transmembrane receptor that is expressed in brain microglia cells. Triggering receptor expressed on myeloid cells 2 (TREM2) is a major regulator for microglia activation in inflammation response [ 74 , 133 , 134 ]. Moreover, it has an important role in Aβ clearance; it could interact with ApoE during this process, boosting its response [ 26 , 74 , 134 , 135 ]. A mutation in the TREM2 gene enhances the inflammatory response and an increase in pro‐inflammatory cytokines [ 136 , 137 ].

Another mutation that plays a role in chronic inflammation occurs on an APP‐related gene [ 74 , 138 , 139 , 140 ]. Because the mutations in APP are permanent, misfolded Aβ production is continuous, and because aggregates are persistent in AD, this could cause chronic inflammation, which activates inflammation‐related components like the toll‐like receptor [ 74 , 138 , 139 , 140 ].

The presence of viral infection in the brain, high blood pressure, stroke and haemorrhage could lead to chronic neuroinflammation as well [ 74 , 141 , 142 ]. Beyond cytokine‐driven activation, microglia contribute to AD pathology through aberrant reactivation of the developmental complement‐dependent synaptic pruning program. During normal brain development, the classical complement components C1q and C3 tag weak or supernumerary synapses, which are then phagocytosed by microglia via the complement receptor CR3 [ 143 , 144 ]. In AD mouse models, this pathway is pathologically re‐engaged: C1q immunoreactivity is elevated at synapses well before overt plaque deposition, soluble Aβ oligomers increase the colocalization of C1q with postsynaptic markers and drive C3 deposition, and phagocytic microglia are recruited to synaptic sites [ 145 ]. Consistent with a causal role, genetic or pharmacological inhibition of C1q, C3, or CR3 attenuates early synapse loss and protects hippocampal long‐term potentiation in these models [ 145 ]. Because region‐specific synapse loss correlates more strongly with cognitive decline in AD than plaque or tangle counts, this complement–microglia axis represents a mechanistically defined link between neuroinflammation and the early cognitive impairment of AD.

NOD‐Like Receptor Protein 3 Inflammasome

The NOD‐like receptor protein 3 (NLRP3) inflammasome is one of the most representative members of the NLRP family of inflammasomes. It plays a pivotal role in the regulation of the innate immune system by activating caspase‐1 and inflammatory cytokines interleukin‐1β (IL‐1β) and interleukin‐18 (IL‐18) and functioning as a ‘molecular sensor’ for detecting intracellular and extracellular stimuli including pathogen‐associated molecular patterns (PAMPs) and damage‐associated molecular patterns (DAMPs) [ 133 , 146 ]. While tightly regulated NLRP3 activation is essential for host defence, aberrant or chronic activation of this inflammasome has been implicated in the pathogenesis of numerous inflammatory, autoimmune, metabolic and neurodegenerative diseases [ 134 , 135 ]. Furthermore, chronic inflammation is a shared pathological feature between T2DM and neurodegenerative disorders, with the NLRP3 inflammasome emerging as a key molecular mediator linking metabolic dysfunction to neuroinflammation [ 138 , 139 ]. The metabolic stressors of T2DM such as hyperglycaemia, insulin resistance, oxidative stress, mitochondrial stress, endoplasmic reticulum stress, elevated levels of advanced glycation end products promote NLRP3 activation in peripheral immune cells and metabolically active tissues [ 140 , 141 , 142 ]. Moreover, aberrant activation of the NLRP3 inflammasome has also been implicated in CNS pathology. Accumulating evidence suggests that NLRP3 activation contributes to sustained microglial activation, disruption of BBB integrity, promotion of Aβ aggregation and tau hyperphosphorylation, all of which represent key pathological hallmarks of AD [ 147 , 148 ].

Recent studies suggest that antidiabetic therapies, particularly sodium–glucose cotransporter‐2 (SGLT‐2) inhibitors, may exert pleiotropic effects beyond glycaemic control by modulating inflammatory pathways, including inhibition of the NLRP3 inflammasome [ 149 ]. This inhibitory effect is proposed to occur through several mechanisms, such as reduced glucose uptake, attenuation of oxidative stress, modulation of mitochondrial function and alterations in the gut microbiome [ 150 ]. Preclinical studies conducted in animal models have demonstrated a protective role of SGLT‐2 inhibitors via suppression of NLRP3 inflammasome activation in various pathological conditions, including obesity, myocardial infarction, diabetic kidney disease and depression [ 151 , 152 , 153 , 154 ]. These anti‐inflammatory and neuroprotective properties may partially underlie the observed benefits of SGLT‐2 inhibitors on cognitive function and the reduced risk of neurodegenerative disorders in patients with T2DM [ 155 , 156 , 157 ]. Furthermore, these findings suggest that the NLRP3 inflammasome serves as a critical molecular mediator linking T2DM‐associated metabolic inflammation to neurodegeneration and that its modulation by antidiabetic drug classes such as SGLT‐2 inhibitors may be a promising therapeutic approach for mitigating the progression of both metabolic and neurodegenerative diseases.

Mitochondrial Dysfunction

Mitochondria are generally known to be organelles that are responsible for adenosine triphosphate (ATP) production, but they also generate ROS and antioxidants [ 73 , 158  ]. Furthermore, they maintain cellular processes like Ca +2 homeostasis and apoptosis [ 73 , 159 ]. Additionally, mitochondrial transport and movement play an important role in neuronal polarity, synaptic function and fusion and fission cycles of mitochondria, and Aβ production could disturb these vital roles [ 5 , 160 , 161 ].

Nearly 90% of ATP requirement for normal neuronal functioning is provided by neuronal mitochondria, and structural or functional mitochondrial deterioration may lead to neurodegeneration [ 73 , 162  ]. Mitochondrial dysfunction is generally related to morphological abnormalities, altered dynamics, bioenergetics or disrupted mitochondrial transport [ 5 ]. Studies suggest that mitochondrial dysfunction may be an early contributor to the pathophysiological processes involved in chronic diseases such as AD and T2DM [ 73 ]. Moreover, it has been shown that Aβ oligomers tend to be located close to mitochondria, and by reducing cytochrome oxidase activity, they block the electron transfer chain and cause ROS (reactive oxygen species) and RNS (reactive nitrogen species) production [ 73 , 163 ]. On the other hand, impaired mitochondrial oxidative capacity of the cells is thought to be one of the main underlying causes of T2DM as well [ 73 , 164  ]. However, it is still unclear whether mitochondrial dysfunction is a cause or a consequence of insulin resistance [ 73 , 165 ].

Oxidative Stress

Oxidative stress, mitochondrial dysfunction and insufficient antioxidant defence mechanisms are other common features of T2DM and AD [ 73 , 166 , 167 ]. Oxidative stress arises from the increased ratio of ROS/RNS or from the imbalance between ROS/RNS production and antioxidant defence levels [ 73 ]. The imbalance between ROS and RNS causes oxidative stress, which has a deleterious effect on cells [ 73 ]. In that condition, free radicals are produced constantly in cells by metabolism, and to avoid apoptosis and cellular damage, antioxidants are produced by activating enzymes [ 2 , 168 ]. The brain is highly vulnerable to high levels of ROS and RNS because of its weak antioxidant mechanism and its high lipid and amino acid content, which are favourable components for oxidative stress [ 73 ]. High levels of oxidative stress markers, including oxidized lipids, proteins and DNA, are found in the brains of AD patients [ 73 , 167 ].

Redox proteomic studies on AD brain tissue have shown that enzymes involved in metabolic pathways such as glycolysis and the Krebs cycle are commonly oxidized [ 73 ]. These enzymes are malate dehydrogenase (MDH), alpha‐enolase (ENO1), fructose bisphosphate aldolase (FBA/AC), glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) and ATP synthase [ 73 ]. Oxidation of these proteins indicates a decrease in cerebral glucose metabolism in AD and mild cognitive impairment (MCI), suggesting impaired neuronal glucose homeostasis and a contribution to neuronal dysfunction [ 73 , 169 ].

JNK

JNK proteins, which are stress‐activated protein kinases (SAPK), are part of the mitogen‐activated protein kinase (MAPK) family [ 73 ]. In the periphery and in the brain, JNK tends to act differently [ 73 ]. While it is generally active and essential for the regulation of brain maturation, repair, cognitive functions, axon integrity and dendritic formation in the brain, in the periphery, it needs a triggering stress condition, such as oxidative stress or UV radiation [ 73 , 170 ]. JNK seems to play a role in neuroinflammation and neuronal destruction, and from this perspective, it may be a major signalling pathway in AD [ 73 ]. Some studies provide evidence that JNK signalling may be an essential pathway in AD [ 73 , 171 ].

It is known that activation of JNK could lead to translocation of mitochondria, resulting in an alteration in bioenergetics and decreased ATP production in neurons [ 73 ]. The same scenario is valid during the aging process, and this increased JNK activity also shifts the balance to anaerobic glycolysis [ 73 ]. In terms of diabetes, JNK has important roles, such as the regulation of pancreatic β‐cell death and negative regulation of insulin [ 73 ]. Evidence suggests that JNK activation can cause insulin resistance in peripheral tissue and can alter glucose metabolism [ 73 ].

Another observed association between AD and T2DM is the elevation of lipid peroxidation products, such as malondialdehyde (MDA) and 4‐hydroxynonenal (HNE) [ 73 , 159 , 172 ]. Moreover, HNE could alter insulin signalling by altering membrane‐associated proteins, disabling pancreatic β‐cells and blocking cell‐insulin interaction [ 73 , 173 ]. In addition, HNE could disable neurons by altering membrane‐associated glucose transporters and Aβ oligomer formation‐related enzymes [ 73 , 173 ].

O ‐Linked β‐ N ‐Acetylglucosamine (O‐GlcNAc)

O‐GlcNAcylation, a glucose‐responsive post‐translational protein modification, plays a critical role in essential biological processes, including protein maturation, transcriptional regulation and cellular processing of proteins [ 174 , 175 ]. Increasing evidence indicates that many effector molecules within the insulin signalling pathway are subject to O ‐GlcNAcylation, often in a reciprocal relationship with phosphorylation at specific serine residues [ 176 , 177 ]. This modification involves the transfer of a single β‐ N ‐acetylglucosamine moiety from uridine diphosphate N ‐acetylglucosamine (UDP‐GlcNAc) to serine or threonine residues of target proteins [ 165 ]. A finely tuned balance between O ‐GlcNAcylation and phosphorylation under physiological conditions is therefore essential for proper insulin signalling, and disruption of this homeostatic interplay may contribute to metabolic dysfunction and disease [ 178 ]. Beyond its role in metabolic regulation, O ‐GlcNAcylation has been implicated in neurodegenerative disorders, including AD. Several AD‐ and neurodegeneration‐associated proteins such as APP, α‐synuclein, neurofilaments (NFs), tau and synapsin‐I are also known targets of O ‐GlcNAc modification [ 179 , 180 , 181 ]. Moreover, alterations in O ‐GlcNAcylation levels have been directly linked to key aspects of AD pathogenesis, including abnormal tau phosphorylation and γ‐secretase–mediated Aβ production [ 182 , 183 ]. Collectively, these findings suggest that therapeutic strategies aimed at modulating O ‐GlcNAcylation may represent a promising approach to alleviate AD‐related pathology and symptoms.

Gut–Brain Axis and the Microbiome

The human gastrointestinal (GI) tract harbours the largest and most complex microbial community in the body, comprising approximately 10 14 microorganisms that play a critical role in maintaining host health [ 173 , 184 ]. The functions of intestinal microbiota include preservation of intestinal barrier integrity, inhibition of adhesion of pathogens to intestinal surfaces, synthesis of essential vitamins such as vitamin K, as well as proper maturation and regulation of the immune system [ 185 , 186 ]. The microbiota‐gut–brain axis represents a bidirectional communication network linking the gut and the CNS through neural, immune, endocrine and metabolic pathways [ 187 , 188 , 189 ]. Disruption of gut microbial homeostasis, commonly referred to as dysbiosis, and alterations in microbial composition have been implicated in the development and progression of a wide range of human diseases, including inflammatory bowel disease, T2DM, obesity, autism, cardiovascular disease, depression, Parkinson's disease, multiple sclerosis and AD [ 190 , 191 , 192 , 193 , 194 , 195 , 196 , 197 , 198 , 199 ]. Growing evidence suggests that gut microbiota alterations may play a modulatory role in the pathophysiological interactions observed between T2DM and AD [ 200 ]. In particular, microbiota‐derived metabolites such as short‐chain fatty acids, tryptophan metabolites and secondary bile acids have been shown to influence both peripheral glucose homeostasis and CNS processes relevant to AD pathology, including synaptic plasticity, Aβ accumulation and tau phosphorylation [ 201 ]. Collectively, these findings suggest that targeting the gut microbiome may represent a promising therapeutic strategy for mitigating metabolic dysfunction in T2DM and modifying disease progression in AD.

Clinical Relevance and Translational Perspectives for Antidiabetic Therapies and Alzheimer Disease

Considering the accumulating evidence indicating an association between T2DM and an elevated risk of dementia, the question arises whether antidiabetic treatments could improve cognitive dysfunction and/or reduce the risk of dementia. Preclinical studies and some observational cohorts have suggested that drugs such as metformin, thiazolidinediones, incretins, SGLT‐2 inhibitors and intranasal insulin may exert beneficial effects in the brain [ 1 , 82 , 202 ]. However, most dedicated randomized controlled trials (RCTs) in AD have so far been disappointing. The pioglitazone TOMMORROW trial did not delay the onset of mild cognitive impairment in at‐risk individuals [ 203 ], and the phase 3 EVOKE and EVOKE+ trials of oral semaglutide showed no significant slowing of disease progression in early AD despite improvements in biomarkers [ 204 ]. Similarly, a large multicenter RCT of intranasal insulin demonstrated no significant cognitive or functional benefit in the intention‐to‐treat cohort at 12 months [ 205 ], although a subsequent phase 2A/B trial combining intranasal insulin with empagliflozin reported improvements in cognition and neuroinflammatory markers [ 206 ]. Overall, current clinical evidence does not support a clear disease‐modifying benefit of antidiabetic drugs in established AD, and further well‐designed studies are needed to clarify their therapeutic potential, particularly for prevention. The potential protective effects of antidiabetic drugs reported in the literature and their mechanisms of action are summarized in Table  1 .

TABLE 1 Antidiabetic agents with potential effects in Alzheimer's disease: mechanisms, effects, and level of evidence. Drug class Drug(s) Mechanism Study type and model Key outcome(s) Direction of effect Evidence level and clinical relevance References Biguanides Metformin Insulin sensitizer; reduces hepatic glucose production Preclinical: transgenic‐mouse SR. Human: large observational cohorts/meta‐analyses Memory/cognition; incident dementia risk Preclinical: mixed. Observational: ↓ dementia risk (no RCT; some conflicting reports) Moderate (observational); suggestive, unconfirmed by RCT [ 207 , 208 , 209 ] Thiazolidinediones (glitazones) Pioglitazone; rosiglitazone PPARγ agonists; ↑ peripheral insulin sensitivity Human: Phase 3 RCT (TOMMORROW) + meta‐analysis (Cao). Preclinical: transgenic mice; rat febrile‐seizure model (not AD) MCI onset; cognition, memory, tau, Aβ, inflammation Preclinical: ↓ tau, ↓ Aβ, ↓ proinflammatory markers, ↑ cognition & memory. Human Phase 3 RCT: no delay of MCI onset (negative) High (Ph3 RCT) but negative; not supported for prevention [ 203 , 210 , 211 , 212 ] α‐Glucosidase inhibitors Acarbose; miglitol; voglibose Inhibit intestinal α‐glucosidase; ↓ postprandial glucose Preclinical (animal: 3xTg‐AD and aging mouse models) Cognitive decline; metabolic and inflammatory markers ↓ Western‐diet‐induced cognitive impairment, ↓ inflammation (acarbose); inconclusive (voglibose) Low (animal only); preclinical [ 208 , 213 ] Sulfonylureas Glibenclamide ↑ insulin secretion via SUR1 on ATP‐sensitive K + channels Preclinical (rat model of AD) HPA‐axis; affective (depression/anxiety) behaviour; not memory‐focused ↓ hypothalamic–pituitary–adrenal (HPA) axis hyperactivity, ↓ affective symptoms (modest) Low (animal only); preclinical [ 208 , 214 ] GLP‐1 receptor agonists (incretins) Exenatide; liraglutide; semaglutide; dulaglutide; others Activate GLP‐1 receptors; glucose‐dependent insulin secretion; potential neuroprotection Preclinical (streptozotocin‐induced rat model). Human: Ph2b RCT (liraglutide, ELAD), Ph3 RCTs (semaglutide, EVOKE/EVOKE+); observational cohorts Memory, tau, amyloid, TNF‐α; cognitive decline, brain atrophy, progression (CDR‐SB), biomarkers Preclinical: ↑ memory, ↓ tau, ↓ amyloid, ↓ TNF‐α. Liraglutide Ph2b: primary endpoint not met (2° signals ↓ decline/atrophy). Semaglutide Ph3: no slowing of progression. Observational: ↓ dementia risk High (Ph2b/3 RCTs) but mixed/negative for progression; not established [ 204 , 215 , 216 , 217 , 218 ] DPP‐4 inhibitors (incretins) Sitagliptin; vildagliptin; linagliptin; others Inhibit DPP‐4; ↑ incretins (GLP‐1, GIP) Preclinical: in vitro (SK‐N‐MC human neuronal cells; linagliptin); in vivo (3xTg‐AD mice) Aβ cytotoxicity, tau, oxidative stress, mitochondrial dysfunction, incretin levels In vitro: ↓ Aβ cytotoxicity, ↓ tau hyperphosphorylation, ↓ oxidative stress, ↓ mitochondrial dysfunction. in vivo: ↑ incretin levels, ↓ Aβ, ↓ tau, ↓ neuroinflammation Low (in vitro + animal model); preclinical [ 208 , 219 , 220 ] SGLT‐2 inhibitors (gliflozins) Canagliflozin; dapagliflozin; empagliflozin; others Inhibit renal SGLT‐2; anti‐inflammatory, antioxidant, NLRP3‐modulating Preclinical (diabetic mice). Human: large observational cohorts Memory/cognition, oxidative stress, BDNF; dementia risk Preclinical: ↑ memory/cognition, ↓ oxidative stress markers, ↑ BDNF. Observational: ↓ dementia risk (no AD RCT) Moderate (observational); suggestive, unconfirmed by RCT [ 208 , 218 , 221 ] Insulin Intranasal; short‐acting (lispro); long‐acting (detemir) Activates central insulin receptors; supports neuronal survival, synaptic integrity, glucose metabolism; intranasal bypasses BBB Preclinical (aged rats). Human: RCT [ 205 ], Ph2A/B [ 206 ] Amyloid, tau, cognition, function, neuroinflammatory markers Preclinical: ↓ amyloid, ↓ tau. RCT (Craft): no cognitive/functional benefit. Ph2A/B (Erichsen): ↑ cognition, ↓ neuroinflammatory markers High (RCTs) but mixed; clinically relevant [ 73 , 210 , 215 , 222 , 223 ]
Note: Evidence level: High = human Phase 2b/3 randomized controlled trial; Moderate = human meta‐analysis or large observational cohort; Low = animal model; Very low = in vitro. Arrows indicate the direction of the reported effect (↑ increase, ↓ decrease). Abbreviations: Aβ, amyloid‐β; AD, Alzheimer's disease; BBB, blood–brain barrier; BDNF, brain‐derived neurotrophic factor; CDR‐SB, Clinical Dementia Rating–Sum of Boxes; DPP‐4, dipeptidyl peptidase‐4; GIP, glucose‐dependent insulinotropic polypeptide; GLP‐1, glucagon‐like peptide‐1; HPA, hypothalamic–pituitary–adrenal; MCI, mild cognitive impairment; NLRP3, NOD‐like receptor protein 3; PPARγ, peroxisome proliferator‐activated receptor γ; RCT, randomized controlled trial; SGLT‐2, sodium–glucose cotransporter‐2; SR, systematic review; SUR1, sulphonylurea receptor 1; TNF‐α, tumour necrosis factor‐α.

Antidiabetic agents with potential effects in Alzheimer's disease: mechanisms, effects, and level of evidence.

Note: Evidence level: High = human Phase 2b/3 randomized controlled trial; Moderate = human meta‐analysis or large observational cohort; Low = animal model; Very low = in vitro. Arrows indicate the direction of the reported effect (↑ increase, ↓ decrease).

Abbreviations: Aβ, amyloid‐β; AD, Alzheimer's disease; BBB, blood–brain barrier; BDNF, brain‐derived neurotrophic factor; CDR‐SB, Clinical Dementia Rating–Sum of Boxes; DPP‐4, dipeptidyl peptidase‐4; GIP, glucose‐dependent insulinotropic polypeptide; GLP‐1, glucagon‐like peptide‐1; HPA, hypothalamic–pituitary–adrenal; MCI, mild cognitive impairment; NLRP3, NOD‐like receptor protein 3; PPARγ, peroxisome proliferator‐activated receptor γ; RCT, randomized controlled trial; SGLT‐2, sodium–glucose cotransporter‐2; SR, systematic review; SUR1, sulphonylurea receptor 1; TNF‐α, tumour necrosis factor‐α.

This review aimed to summarize the connection between diabetes and AD. Furthermore, the current status of antidiabetic treatments for the prevention of AD, as well as their potential to increase the risk of cognitive dysfunction, is summarised.

Conflicts of Interest

The authors declare no conflicts of interest.