Diabetes mellitus (DM) is a chronic metabolic disorder characterized by hyperglycemia, primarily classified into type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM) based on etiological differences (1, 2). T1DM results from the autoimmune destruction of pancreatic islet cells, leading to absolute insulin deficiency (3). In contrast, T2DM is caused by relative insulin deficiency and/or insulin resistance in target tissues, resulting in persistent hyperglycemia and subsequent multi-organ damage (4, 5). T2DM accounts for 90%–95% of DM patients, with its prevalence increasing annually. Complications such as diabetic nephropathy and diabetic foot severely impact human health (6).

Dementia is a neurodegenerative disease that progressively damages nerve cells and impairs brain function, leading to deterioration of cognitive abilities (7, 8). As one of the most pressing global public health challenges, dementia imposes a substantial economic burden worldwide (9). Alzheimer’s disease (AD) is the most common form of dementia, accounting for approximately 60%–70% of cases (10). Pathologically, AD is characterized by extracellular amyloid plaques composed of amyloid-beta (Aβ) peptide and intracellular neurofibrillary tangles (NFTs) formed by hyperphosphorylated tau protein (11). Currently, approximately 55 million people worldwide are living with AD, with the number doubling every 5 years (12, 13). It is estimated that by 2050, the global prevalence of AD will reach 152 million, equivalent to one new case every 3 s (12). In the United States, approximately 5.3 million individuals have AD, of whom 5.1 million are 65 years or older (14). In 2019, the number of AD and other dementia cases in China was 13,143,950, accounting for 25.5% of the global total (15), and the number of deaths attributed to these conditions was 320,715, representing 19.8% of global deaths (15). Cognitive impairment in patients with AD is the primary driver of loss of self-care ability, which imposes a substantial socioeconomic burden on societies worldwide. The global socioeconomic cost of dementia was estimated at $957.56 billion in 2015, and projections indicate that this figure will rise to $2.54 trillion by 2030 and further to $9.12 trillion by 2050 (16).

Approximately one-third of AD cases are attributed to modifiable and treatable risk factors, highlighting the importance of early prevention (17). Among these factors, DM has emerged as a key candidate risk factor for AD. Accumulating evidence suggests a complex bidirectional association between DM and AD: DM may induce AD, while pathological products of AD may promote DM progression (18, 19). Both diseases share common pathogenic mechanisms, including insulin resistance, inflammation, and mitochondrial dysfunction. Additionally, DM-related metabolic disturbances may contribute to amyloid formation in AD (19–22).

This bidirectional association, however, remains controversial: some studies have failed to detect a significant association between DM and AD, and this may be due to confounding by comorbidities such as hypertension and stroke. Therefore, this meta-analysis aimed to systematically investigate the association between DM and the risk of AD by synthesizing evidence from cohort and case–control studies.

This meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (23–25) and was registered on PROSPERO (CRD420251159844).

This rigorous meta-analysis synthesized evidence from 11 studies involving over 3.3 million participants, confirming a significant positive association between DM and AD risk across diverse populations. The association remained consistent regardless of sample size, indicating that sample size had little impact on the results. The subgroup analysis by ethnicity showed that DM was associated with increased AD risk in both Asian and non-Asian populations, with a slightly stronger effect in Asians. This suggests that ethnicity may moderately influence the association but is not a decisive factor, which may be attributed to differences in genetic background, lifestyle, and environmental factors among ethnic groups—for example, the high-carbohydrate diet in Asian populations may lead to greater blood glucose fluctuations, increasing DM risk and subsequently affecting AD development. Additionally, genetic susceptibility in Asians may make them more vulnerable to the adverse effects of DM on cognitive function.

The APOE ϵ4 allele is the most well-established genetic risk factor for sporadic AD, closely linked to core AD biomarkers (Aβ and tau protein) (37, 38). After adjusting for the APOE ϵ4 allele, the association between DM and AD risk became non-significant. This may be due to the limited availability of APOE ϵ4 genotyping data in only a subset of included studies, which restricted the robustness of the adjusted analysis. The APOE ϵ4 allele promotes AD pathogenesis by regulating Aβ metabolism and clearance, exacerbating neuroinflammation, and inducing tau hyperphosphorylation (39). Pathological processes associated with DM, such as advanced glycation end product (AGE) formation induced by hyperglycemia, oxidative stress, and systemic inflammation, can upregulate brain APOE expression and promote its pathological conformational changes (40). Therefore, we hypothesize that DM may not directly cause AD but acts as an important disease-modifying factor or accelerator. In APOE ϵ4 carriers, DM amplifies pre-existing pathological vulnerability, thereby increasing AD risk; even in the absence of APOE ϵ4, DM-related metabolic disturbances (e.g., insulin resistance, oxidative stress) may independently contribute to AD pathogenesis, although the effect may be weaker (consistent with the significant association observed without APOE ϵ4 adjustment: HR = 1.42, 95% CI (1.23–1.64), P < 0.00001). Conversely, in the presence of APOE ϵ4, DM and related metabolic dysregulations (e.g., brain insulin resistance) may synergize with APOE ϵ4 to accelerate Aβ deposition, tau pathology, and neuronal dysfunction, leading to a statistically significant epidemiological association. This explanation is consistent with our findings (significant association without APOE ϵ4 adjustment vs. non-significant association with adjustment) and is supported by previous studies (33). Future research should collect APOE ϵ4 genotyping data in large-scale prospective cohorts and conduct comprehensive adjusted analyses to more reliably evaluate the moderating or mediating role of APOE ϵ4 in the DM–AD association.

Our results are consistent with previous studies (34–36). Insulin resistance and oxidative stress induced by DM can lead to neuroinflammation and Aβ deposition, thereby increasing AD risk (41). This consistency further supports the notion that DM promotes AD onset and progression through multiple pathophysiological pathways. Additionally, Mendelian randomization studies have shown that DM is associated with increased AD risk through genetic predisposition, providing genetic evidence for a causal relationship between DM and AD (42).

DM may increase the risk of AD through multiple mechanisms, with dysfunction of the insulin signaling pathway representing a key contributor (43). In patients with DM, insulin resistance disrupts insulin signaling cascades and impairs neuronal glucose uptake and utilization, leading to energy metabolism dysfunction. This subsequently deprives neurons of adequate energy supply and compromises their physiological function (44). Dysregulated insulin signaling also interferes with the synthesis, release, and metabolism of neurotransmitters, inducing neurotransmitter imbalances that disrupt neuronal communication and precipitate cognitive impairment (45, 46). Oxidative stress constitutes another critical mechanism underlying the promotion of AD pathogenesis by DM. Hyperglycemic conditions trigger excessive generation of reactive oxygen species (ROS), which overwhelms the body’s antioxidant defense systems and initiates a state of oxidative stress (47). Oxidative stress induces lipid peroxidation of neuronal cell membranes and oxidative damage to proteins and nucleic acids, disrupting neuronal structure and function and accelerating neurodegenerative processes. Additionally, it promotes Aβ aggregation and abnormal tau hyperphosphorylation, leading to the formation of neurofibrillary tangles—hallmarks pathological of AD (47, 48).

APOE ϵ4 is one of the strongest genetic risk factors for AD (49). Carriers of APOE ϵ4 allele exhibit an increased propensity for Aβ deposition and neurofibrillary tangle formation, thereby elevating their risk of developing AD (50). Chronic low-grade inflammation is a common feature in patients with DM, characterized by elevated circulating levels of proinflammatory cytokines including tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and C-reactive protein (CRP) (51). These proinflammatory mediators can cross the blood–brain barrier (BBB), infiltrate the central nervous system (CNS), and activate microglia and astrocytes, triggering neuroinflammation (52, 53). Neuroinflammation plays a pivotal role in the pathogenesis of AD (54); excessive microglial activation leads to the overproduction and release of proinflammatory cytokines, which further exacerbate neuronal damage and cognitive dysfunction (54, 55). Additionally, neuroinflammation may perturb insulin signal transduction, forming a vicious cycle that exacerbates the pathological progression of both DM and AD (56).

Under the hyperglycemic conditions associated with DM, non-enzymatic glycation reactions occur between glucose and biomacromolecules such as proteins and lipids, leading to the formation of advanced glycation end products (AGEs) (57). AGEs accumulate in the brain parenchyma, where they promote Aβ aggregation and tau hyperphosphorylation, further amplify oxidative stress and neuroinflammation, and induce neuronal injury (58). Notably, metabolic perturbations in DM also disrupt the homeostatic balance between glutamate and gamma-aminobutyric acid (GABA), a key contributor to the development of cognitive impairments (59).

Given the confirmed association between DM and AD, exploring potential interventions to mitigate this risk is clinically relevant. Accumulating evidence suggested that antidiabetic agents may exert potential efficacy on AD. Preclinical and clinical studies have demonstrated that these agents can enhance neuronal survival and ameliorate cognitive dysfunction in AD patients by attenuating Aβ aggregation and abnormal tau hyperphosphorylation, reducing proinflammatory markers, mitigating oxidative stress, and upregulating neurotrophic factor expression (60). Glucagon-like peptide-1 receptor agonists (GLP-1RAs), beyond their well-established role in glycemic control, exert robust neuroprotective properties that slow the progression of neurodegenerative diseases. Specifically, GLP-1RAs effectively inhibit Aβ accumulation and alleviate excessive tau hyperphosphorylation and aggregation during disease progression (61). In vivo studies have further revealed the anti-inflammatory effects of GLP-1RAs, which modulate neuroinflammatory responses to reduce neuronal damage and cognitive decline in AD (62, 63). Liraglutide, a representative GLP-1RA, may alleviate cognitive impairment in AD by reducing tau hyperphosphorylation (64). A database analysis indicated that GLP-1RAs and rosiglitazone may reduce the risk of AD in patients with T2DM (65). A recent observational study reported that both sodium-glucose cotransporter-2 inhibitors (SGLT2is) and GLP-1RAs were associated with a reduced risk of AD and related dementias in patients with T2DM, with no significant difference in efficacy observed between the two drug classes (66). GLP-1RA administration was consistently linked to a lower risk of dementia, particularly in the T2DM population (67). A pooled analysis of data from Danish national health registries and cardiovascular outcome clinical trials showed that individuals treated with semaglutide or liraglutide had a dementia risk of less than half that of non-users (67). Another study identified a significant association between semaglutide use and a reduced risk of first-time AD diagnosis (68). Despite these promising observational and pooled analysis findings, however, the 3-year EVOKE and EVOKE+ trials found no significant differences in all primary and secondary cognitive and functional outcomes between the semaglutide and placebo groups (69).

To date, the effects of antidiabetic agents on AD remain under active investigation, and additional well-designed clinical trials are required to validate their therapeutic efficacy and safety profiles in AD patients. Collectively, the association between DM and AD is mediated by a complex interplay of multiple pathophysiological mechanisms, including insulin resistance and signaling dysfunction, chronic neuroinflammation, AGE accumulation, and neurotransmitter imbalance, with antidiabetic agents exhibiting potential modulatory effects on these pathological processes. Future research should focus on further elucidating these underlying mechanisms to develop more effective preventive and therapeutic strategies for AD, particularly in individuals with comorbid DM.

This meta-analysis provides strong evidence that DM is an independent risk factor for AD, offering important implications for clinical practice and future research. Clinicians should closely monitor cognitive function in DM patients, detect early cognitive impairment, and implement timely interventions to delay AD onset and progression. However, due to methodological limitations, the results should be interpreted with caution. Future large-scale, high-quality prospective cohort studies are needed to fully elucidate the relationship between DM and AD risk.